Hi,
This patchset based off-of 3.7-rc3, introduces the Linux kernel port to
ARC700 processor family (750D and 770D) from Synopsys.
ARC700 is highly configurable and power efficient 32-bit RISC core with MMU.
It is embedded in SoCs deployed in TV Set Top boxes, Digital Media Players,
all the way to Network-on-Chips.
More information is available @
http://www.synopsys.com/IP/ProcessorIP/ARCProcessors/Pages/default.aspx
This series is part #1 of the entire port - just enough to build the kernel
and bring up a Busybox shell on console (on ARCAngel4 FPGA ref platform)
The console driver has already been accepted/merged into tty-next tree for 3.8
The part #2 series containing additional features ptrace/SMP/perf/OProfile..
will follow as we move along the review cycle. It has been withheld just to
make it easier for reviewers.
The entire series however is available at
git://github.com/foss-for-synopsys-dwc-arc-processors/linux.git arc-3.7-rc3
The GNU tool-chain, based off of gcc 4.4 + uClibc 0.9.30.3 is also available
from github. Please refer to Readme.md in "toolchain" repository also in same
hierarchy as kernel above.
We are already working on a buildroot based distro, to be made available soon.
More info about the kernel port can also be found in the ELCE-Barclone '12
talk on same topic.
https://raw.github.com/vineetgarc/publish/master/ELCE-2012-ARC-Linux.pdf
All comments/criticisms are welcome !
P.S. Despite 3.7 based, we are still short of the UAPI split for arch headers.
That is also being worked upon with help from David. That will show up in v2
of the patchseries.
Thx,
-Vineet
Vineet Gupta (31):
ARC: Generic Headers
ARC: irqflags
ARC: atomic/bitops/cmpxchg/barriers
asm-generic headers: uaccess.h to conditionally define segment_eq()
ARC: uaccess friends
asm-generic headers: Allow yet more arch overrides in checksum.h
ARC: checksum/byteorder/swab routines
ARC: Fundamental ARCH data-types/defines
ARC: spinlock/rwlock/mutex primitives
ARC: string library
ARC: Low level IRQ/Trap/Exception(non-MMU) Handling
ARC: Interrupt Handling
ARC: Non-MMU Exception Handling
ARC: syscall support
ARC: Process/scheduling/clock/Timers/Delay Management
ARC: Signal handling
ARC: Cache Flush Management
ARC: Page Table Management
ARC: MMU Context Management
ARC: MMU Exception Handling
ARC: TLB flush Handling
ARC: Page Fault handling (incl uaccess fixup)
ARC: I/O and DMA Mappings
ARC: startup #1: low-level, setup_arch(), /proc/cpuinfo, mem init
ARC: [plat-arcfpga] Hooking up platform to ARC UART
ARC: Build system: Makefiles, Kconfig, Linker script
ARC: Last bits (stubs) to get to a running kernel with UART
ARC: split ret_from_fork, simplify kernel_thread()
ARC: switch to generic kernel_thread()
ARC: switch to generic kernel_execve() and sys_execve()
ARC: [plat-arcfpga] defconfig
arch/arc/Kbuild | 2 +
arch/arc/Kconfig | 338 ++++++++++
arch/arc/Kconfig.debug | 34 +
arch/arc/Makefile | 115 ++++
arch/arc/boot/Makefile | 27 +
arch/arc/configs/fpga_defconfig | 607 ++++++++++++++++++
arch/arc/include/asm/Kbuild | 58 ++
arch/arc/include/asm/arcregs.h | 330 ++++++++++
arch/arc/include/asm/asm-offsets.h | 9 +
arch/arc/include/asm/atomic.h | 232 +++++++
arch/arc/include/asm/barrier.h | 42 ++
arch/arc/include/asm/bitops.h | 507 +++++++++++++++
arch/arc/include/asm/bug.h | 37 ++
arch/arc/include/asm/byteorder.h | 18 +
arch/arc/include/asm/cache.h | 75 +++
arch/arc/include/asm/cachectl.h | 30 +
arch/arc/include/asm/cacheflush.h | 67 ++
arch/arc/include/asm/checksum.h | 101 +++
arch/arc/include/asm/cmpxchg.h | 143 +++++
arch/arc/include/asm/delay.h | 68 ++
arch/arc/include/asm/dma-mapping.h | 176 ++++++
arch/arc/include/asm/dma.h | 14 +
arch/arc/include/asm/elf.h | 97 +++
arch/arc/include/asm/entry.h | 495 +++++++++++++++
arch/arc/include/asm/exec.h | 15 +
arch/arc/include/asm/io.h | 26 +
arch/arc/include/asm/irq.h | 26 +
arch/arc/include/asm/irqflags.h | 143 +++++
arch/arc/include/asm/kdebug.h | 19 +
arch/arc/include/asm/linkage.h | 30 +
arch/arc/include/asm/mmu.h | 23 +
arch/arc/include/asm/mmu_context.h | 209 ++++++
arch/arc/include/asm/module.h | 17 +
arch/arc/include/asm/mutex.h | 9 +
arch/arc/include/asm/page.h | 134 ++++
arch/arc/include/asm/pgalloc.h | 134 ++++
arch/arc/include/asm/pgtable.h | 401 ++++++++++++
arch/arc/include/asm/processor.h | 146 +++++
arch/arc/include/asm/ptrace.h | 133 ++++
arch/arc/include/asm/sections.h | 17 +
arch/arc/include/asm/segment.h | 24 +
arch/arc/include/asm/setup.h | 22 +
arch/arc/include/asm/sigcontext.h | 23 +
arch/arc/include/asm/signal.h | 27 +
arch/arc/include/asm/smp.h | 34 +
arch/arc/include/asm/spinlock.h | 144 +++++
arch/arc/include/asm/spinlock_types.h | 35 +
arch/arc/include/asm/string.h | 40 ++
arch/arc/include/asm/swab.h | 99 +++
arch/arc/include/asm/switch_to.h | 41 ++
arch/arc/include/asm/syscall.h | 72 +++
arch/arc/include/asm/syscalls.h | 30 +
arch/arc/include/asm/thread_info.h | 120 ++++
arch/arc/include/asm/timex.h | 18 +
arch/arc/include/asm/tlb-mmu1.h | 104 +++
arch/arc/include/asm/tlb.h | 58 ++
arch/arc/include/asm/tlbflush.h | 28 +
arch/arc/include/asm/uaccess.h | 605 ++++++++++++++++++
arch/arc/include/asm/unistd.h | 46 ++
arch/arc/include/asm/vmlinux.lds.h | 12 +
arch/arc/kernel/Makefile | 16 +
arch/arc/kernel/arcksyms.c | 56 ++
arch/arc/kernel/asm-offsets.c | 46 ++
arch/arc/kernel/ctx_sw.c | 91 +++
arch/arc/kernel/ctx_sw_asm.S | 58 ++
arch/arc/kernel/entry.S | 635 +++++++++++++++++++
arch/arc/kernel/fpu.c | 55 ++
arch/arc/kernel/head.S | 78 +++
arch/arc/kernel/irq.c | 107 ++++
arch/arc/kernel/process.c | 259 ++++++++
arch/arc/kernel/ptrace.c | 26 +
arch/arc/kernel/reset.c | 33 +
arch/arc/kernel/setup.c | 166 +++++
arch/arc/kernel/signal.c | 360 +++++++++++
arch/arc/kernel/stacktrace.c | 43 ++
arch/arc/kernel/sys.c | 18 +
arch/arc/kernel/time.c | 237 +++++++
arch/arc/kernel/traps.c | 125 ++++
arch/arc/kernel/troubleshoot.c | 17 +
arch/arc/kernel/vmlinux.lds.S | 117 ++++
arch/arc/lib/Makefile | 9 +
arch/arc/lib/memcmp.S | 124 ++++
arch/arc/lib/memcpy-700.S | 66 ++
arch/arc/lib/memset.S | 59 ++
arch/arc/lib/strchr-700.S | 123 ++++
arch/arc/lib/strcmp.S | 96 +++
arch/arc/lib/strcpy-700.S | 70 ++
arch/arc/lib/strlen.S | 83 +++
arch/arc/mm/Makefile | 10 +
arch/arc/mm/cache_arc700.c | 838 +++++++++++++++++++++++++
arch/arc/mm/dma.c | 91 +++
arch/arc/mm/extable.c | 26 +
arch/arc/mm/fault.c | 228 +++++++
arch/arc/mm/init.c | 185 ++++++
arch/arc/mm/ioremap.c | 62 ++
arch/arc/mm/mmap.c | 21 +
arch/arc/mm/tlb.c | 586 +++++++++++++++++
arch/arc/mm/tlbex.S | 351 +++++++++++
arch/arc/plat-arcfpga/Kconfig | 33 +
arch/arc/plat-arcfpga/Makefile | 9 +
arch/arc/plat-arcfpga/include/plat/dma_addr.h | 45 ++
arch/arc/plat-arcfpga/include/plat/irq.h | 27 +
arch/arc/plat-arcfpga/include/plat/memmap.h | 31 +
arch/arc/plat-arcfpga/irq.c | 41 ++
arch/arc/plat-arcfpga/platform.c | 134 ++++
include/asm-generic/checksum.h | 4 +
include/asm-generic/uaccess.h | 3 +-
lib/checksum.c | 2 +
108 files changed, 12585 insertions(+), 1 deletions(-)
create mode 100644 arch/arc/Kbuild
create mode 100644 arch/arc/Kconfig
create mode 100644 arch/arc/Kconfig.debug
create mode 100644 arch/arc/Makefile
create mode 100644 arch/arc/boot/Makefile
create mode 100644 arch/arc/configs/fpga_defconfig
create mode 100644 arch/arc/include/asm/Kbuild
create mode 100644 arch/arc/include/asm/arcregs.h
create mode 100644 arch/arc/include/asm/asm-offsets.h
create mode 100644 arch/arc/include/asm/atomic.h
create mode 100644 arch/arc/include/asm/barrier.h
create mode 100644 arch/arc/include/asm/bitops.h
create mode 100644 arch/arc/include/asm/bug.h
create mode 100644 arch/arc/include/asm/byteorder.h
create mode 100644 arch/arc/include/asm/cache.h
create mode 100644 arch/arc/include/asm/cachectl.h
create mode 100644 arch/arc/include/asm/cacheflush.h
create mode 100644 arch/arc/include/asm/checksum.h
create mode 100644 arch/arc/include/asm/cmpxchg.h
create mode 100644 arch/arc/include/asm/delay.h
create mode 100644 arch/arc/include/asm/dma-mapping.h
create mode 100644 arch/arc/include/asm/dma.h
create mode 100644 arch/arc/include/asm/elf.h
create mode 100644 arch/arc/include/asm/entry.h
create mode 100644 arch/arc/include/asm/exec.h
create mode 100644 arch/arc/include/asm/io.h
create mode 100644 arch/arc/include/asm/irq.h
create mode 100644 arch/arc/include/asm/irqflags.h
create mode 100644 arch/arc/include/asm/kdebug.h
create mode 100644 arch/arc/include/asm/linkage.h
create mode 100644 arch/arc/include/asm/mmu.h
create mode 100644 arch/arc/include/asm/mmu_context.h
create mode 100644 arch/arc/include/asm/module.h
create mode 100644 arch/arc/include/asm/mutex.h
create mode 100644 arch/arc/include/asm/page.h
create mode 100644 arch/arc/include/asm/pgalloc.h
create mode 100644 arch/arc/include/asm/pgtable.h
create mode 100644 arch/arc/include/asm/processor.h
create mode 100644 arch/arc/include/asm/ptrace.h
create mode 100644 arch/arc/include/asm/sections.h
create mode 100644 arch/arc/include/asm/segment.h
create mode 100644 arch/arc/include/asm/setup.h
create mode 100644 arch/arc/include/asm/sigcontext.h
create mode 100644 arch/arc/include/asm/signal.h
create mode 100644 arch/arc/include/asm/smp.h
create mode 100644 arch/arc/include/asm/spinlock.h
create mode 100644 arch/arc/include/asm/spinlock_types.h
create mode 100644 arch/arc/include/asm/string.h
create mode 100644 arch/arc/include/asm/swab.h
create mode 100644 arch/arc/include/asm/switch_to.h
create mode 100644 arch/arc/include/asm/syscall.h
create mode 100644 arch/arc/include/asm/syscalls.h
create mode 100644 arch/arc/include/asm/thread_info.h
create mode 100644 arch/arc/include/asm/timex.h
create mode 100644 arch/arc/include/asm/tlb-mmu1.h
create mode 100644 arch/arc/include/asm/tlb.h
create mode 100644 arch/arc/include/asm/tlbflush.h
create mode 100644 arch/arc/include/asm/uaccess.h
create mode 100644 arch/arc/include/asm/unistd.h
create mode 100644 arch/arc/include/asm/vmlinux.lds.h
create mode 100644 arch/arc/kernel/Makefile
create mode 100644 arch/arc/kernel/arcksyms.c
create mode 100644 arch/arc/kernel/asm-offsets.c
create mode 100644 arch/arc/kernel/ctx_sw.c
create mode 100644 arch/arc/kernel/ctx_sw_asm.S
create mode 100644 arch/arc/kernel/entry.S
create mode 100644 arch/arc/kernel/fpu.c
create mode 100644 arch/arc/kernel/head.S
create mode 100644 arch/arc/kernel/irq.c
create mode 100644 arch/arc/kernel/process.c
create mode 100644 arch/arc/kernel/ptrace.c
create mode 100644 arch/arc/kernel/reset.c
create mode 100644 arch/arc/kernel/setup.c
create mode 100644 arch/arc/kernel/signal.c
create mode 100644 arch/arc/kernel/stacktrace.c
create mode 100644 arch/arc/kernel/sys.c
create mode 100644 arch/arc/kernel/time.c
create mode 100644 arch/arc/kernel/traps.c
create mode 100644 arch/arc/kernel/troubleshoot.c
create mode 100644 arch/arc/kernel/vmlinux.lds.S
create mode 100644 arch/arc/lib/Makefile
create mode 100644 arch/arc/lib/memcmp.S
create mode 100644 arch/arc/lib/memcpy-700.S
create mode 100644 arch/arc/lib/memset.S
create mode 100644 arch/arc/lib/strchr-700.S
create mode 100644 arch/arc/lib/strcmp.S
create mode 100644 arch/arc/lib/strcpy-700.S
create mode 100644 arch/arc/lib/strlen.S
create mode 100644 arch/arc/mm/Makefile
create mode 100644 arch/arc/mm/cache_arc700.c
create mode 100644 arch/arc/mm/dma.c
create mode 100644 arch/arc/mm/extable.c
create mode 100644 arch/arc/mm/fault.c
create mode 100644 arch/arc/mm/init.c
create mode 100644 arch/arc/mm/ioremap.c
create mode 100644 arch/arc/mm/mmap.c
create mode 100644 arch/arc/mm/tlb.c
create mode 100644 arch/arc/mm/tlbex.S
create mode 100644 arch/arc/plat-arcfpga/Kconfig
create mode 100644 arch/arc/plat-arcfpga/Makefile
create mode 100644 arch/arc/plat-arcfpga/include/plat/dma_addr.h
create mode 100644 arch/arc/plat-arcfpga/include/plat/irq.h
create mode 100644 arch/arc/plat-arcfpga/include/plat/memmap.h
create mode 100644 arch/arc/plat-arcfpga/irq.c
create mode 100644 arch/arc/plat-arcfpga/platform.c
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/arcregs.h | 114 +++++++++++++++++++++++++++++++
arch/arc/include/asm/irqflags.h | 143 +++++++++++++++++++++++++++++++++++++++
arch/arc/kernel/irq.c | 32 +++++++++
3 files changed, 289 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/arcregs.h
create mode 100644 arch/arc/include/asm/irqflags.h
create mode 100644 arch/arc/kernel/irq.c
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
new file mode 100644
index 0000000..8ca8faf
--- /dev/null
+++ b/arch/arc/include/asm/arcregs.h
@@ -0,0 +1,114 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_ARCREGS_H
+#define _ASM_ARC_ARCREGS_H
+
+#ifdef __KERNEL__
+
+/* status32 Bits Positions */
+#define STATUS_H_BIT 0 /* CPU Halted */
+#define STATUS_E1_BIT 1 /* Int 1 enable */
+#define STATUS_E2_BIT 2 /* Int 2 enable */
+#define STATUS_A1_BIT 3 /* Int 1 active */
+#define STATUS_A2_BIT 4 /* Int 2 active */
+#define STATUS_AE_BIT 5 /* Exception active */
+#define STATUS_DE_BIT 6 /* PC is in delay slot */
+#define STATUS_U_BIT 7 /* User/Kernel mode */
+#define STATUS_L_BIT 12 /* Loop inhibit */
+
+/* These masks correspond to the status word(STATUS_32) bits */
+#define STATUS_H_MASK (1<<STATUS_H_BIT)
+#define STATUS_E1_MASK (1<<STATUS_E1_BIT)
+#define STATUS_E2_MASK (1<<STATUS_E2_BIT)
+#define STATUS_A1_MASK (1<<STATUS_A1_BIT)
+#define STATUS_A2_MASK (1<<STATUS_A2_BIT)
+#define STATUS_AE_MASK (1<<STATUS_AE_BIT)
+#define STATUS_DE_MASK (1<<STATUS_DE_BIT)
+#define STATUS_U_MASK (1<<STATUS_U_BIT)
+#define STATUS_L_MASK (1<<STATUS_L_BIT)
+
+/* Auxiliary registers */
+#define AUX_IDENTITY 4
+#define AUX_INTR_VEC_BASE 0x25
+#define AUX_IRQ_LEV 0x200 /* IRQ Priority: L1 or L2 */
+#define AUX_IRQ_HINT 0x201 /* For generating Soft Interrupts */
+#define AUX_IRQ_LV12 0x43 /* interrupt level register */
+
+#define AUX_IENABLE 0x40c
+#define AUX_ITRIGGER 0x40d
+#define AUX_IPULSE 0x415
+
+#ifndef __ASSEMBLY__
+
+/*
+ ******************************************************************
+ * Inline ASM macros to read/write AUX Regs
+ * Essentially invocation of lr/sr insns from "C"
+ */
+
+#if 1
+
+#define read_aux_reg(reg) __builtin_arc_lr(reg)
+
+/* gcc builtin sr needs reg param to be long immediate */
+#define write_aux_reg(reg_immed, val) \
+ __builtin_arc_sr((unsigned int)val, reg_immed)
+
+#else
+
+#define read_aux_reg(reg) \
+({ \
+ unsigned int __ret; \
+ __asm__ __volatile__( \
+ " lr %0, [%1]" \
+ : "=r"(__ret) \
+ : "i"(reg)); \
+ __ret; \
+})
+
+/*
+ * Aux Reg address is specified as long immediate by caller
+ * e.g.
+ * write_aux_reg(0x69, some_val);
+ * This generates tightest code.
+ */
+#define write_aux_reg(reg_imm, val) \
+({ \
+ __asm__ __volatile__( \
+ " sr %0, [%1] \n" \
+ : \
+ : "ir"(val), "i"(reg_imm)); \
+})
+
+/*
+ * Aux Reg address is specified in a variable
+ * * e.g.
+ * reg_num = 0x69
+ * write_aux_reg2(reg_num, some_val);
+ * This has to generate glue code to load the reg num from
+ * memory to a reg hence not recommended.
+ */
+#define write_aux_reg2(reg_in_var, val) \
+({ \
+ unsigned int tmp; \
+ \
+ __asm__ __volatile__( \
+ " ld %0, [%2] \n\t" \
+ " sr %1, [%0] \n\t" \
+ : "=&r"(tmp) \
+ : "r"(val), "memory"(®_in_var)); \
+})
+
+#endif
+
+#endif /* __ASEMBLY__ */
+
+#endif /* __KERNEL__ */
+
+#endif /* _ASM_ARC_ARCREGS_H */
diff --git a/arch/arc/include/asm/irqflags.h b/arch/arc/include/asm/irqflags.h
new file mode 100644
index 0000000..752e55d
--- /dev/null
+++ b/arch/arc/include/asm/irqflags.h
@@ -0,0 +1,143 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_IRQFLAGS_H
+#define __ASM_ARC_IRQFLAGS_H
+
+/* vineetg: March 2010 : local_irq_save( ) optimisation
+ * -Remove explicit mov of current status32 into reg, that is not needed
+ * -Use BIC insn instead of INVERTED + AND
+ * -Conditionally disable interrupts (if they are not enabled, don't disable)
+*/
+
+#ifdef __KERNEL__
+
+#include <asm/arcregs.h>
+
+#ifndef __ASSEMBLY__
+
+/******************************************************************
+ * IRQ Control Macros
+ ******************************************************************/
+
+/*
+ * Save IRQ state and disable IRQs
+ */
+static inline long arch_local_irq_save(void)
+{
+ unsigned long temp, flags;
+
+ __asm__ __volatile__(
+ " lr %1, [status32] \n"
+ " bic %0, %1, %2 \n"
+ " and.f 0, %1, %2 \n"
+ " flag.nz %0 \n"
+ : "=r"(temp), "=r"(flags)
+ : "n"((STATUS_E1_MASK | STATUS_E2_MASK))
+ : "cc");
+
+ return flags;
+}
+
+/*
+ * restore saved IRQ state
+ */
+static inline void arch_local_irq_restore(unsigned long flags)
+{
+
+ __asm__ __volatile__(
+ " flag %0 \n"
+ :
+ : "r"(flags));
+}
+
+/*
+ * Conditionally Enable IRQs
+ */
+extern void arch_local_irq_enable(void);
+
+/*
+ * Unconditionally Disable IRQs
+ */
+static inline void arch_local_irq_disable(void)
+{
+ unsigned long temp;
+
+ __asm__ __volatile__(
+ " lr %0, [status32] \n"
+ " and %0, %0, %1 \n"
+ " flag %0 \n"
+ : "=&r"(temp)
+ : "n"(~(STATUS_E1_MASK | STATUS_E2_MASK)));
+}
+
+/*
+ * save IRQ state
+ */
+static inline long arch_local_save_flags(void)
+{
+ unsigned long temp;
+
+ __asm__ __volatile__(
+ " lr %0, [status32] \n"
+ : "=&r"(temp));
+
+ return temp;
+}
+
+/*
+ * Query IRQ state
+ */
+static inline int arch_irqs_disabled_flags(unsigned long flags)
+{
+ return !(flags & (STATUS_E1_MASK));
+}
+
+static inline int arch_irqs_disabled(void)
+{
+ return arch_irqs_disabled_flags(arch_local_save_flags());
+}
+
+static inline void arch_mask_irq(unsigned int irq)
+{
+ unsigned int ienb;
+
+ ienb = read_aux_reg(AUX_IENABLE);
+ ienb &= ~(1 << irq);
+ write_aux_reg(AUX_IENABLE, ienb);
+}
+
+static inline void arch_unmask_irq(unsigned int irq)
+{
+ unsigned int ienb;
+
+ ienb = read_aux_reg(AUX_IENABLE);
+ ienb |= (1 << irq);
+ write_aux_reg(AUX_IENABLE, ienb);
+}
+
+#else
+
+.macro IRQ_DISABLE scratch
+ lr \scratch, [status32]
+ bic \scratch, \scratch, (STATUS_E1_MASK | STATUS_E2_MASK)
+ flag \scratch
+.endm
+
+.macro IRQ_DISABLE_SAVE scratch, save
+ lr \scratch, [status32]
+ mov \save, \scratch /* Make a copy */
+ bic \scratch, \scratch, (STATUS_E1_MASK | STATUS_E2_MASK)
+ flag \scratch
+.endm
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* KERNEL */
+
+#endif
diff --git a/arch/arc/kernel/irq.c b/arch/arc/kernel/irq.c
new file mode 100644
index 0000000..16fcbe8
--- /dev/null
+++ b/arch/arc/kernel/irq.c
@@ -0,0 +1,32 @@
+/*
+ * Copyright (C) 2011-12 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <asm/irqflags.h>
+#include <asm/arcregs.h>
+
+void arch_local_irq_enable(void)
+{
+
+ unsigned long flags;
+ flags = arch_local_save_flags();
+ flags |= (STATUS_E1_MASK | STATUS_E2_MASK);
+
+ /*
+ * If called from hard ISR (between irq_enter and irq_exit)
+ * don't allow Level 1. In Soft ISR we allow further Level 1s
+ */
+
+ if (in_irq())
+ flags &= ~(STATUS_E1_MASK | STATUS_E2_MASK);
+
+ arch_local_irq_restore(flags);
+}
+EXPORT_SYMBOL(arch_local_irq_enable);
--
1.7.4.1
This is because mm_segment_t is exported by arch code, while seqment_eq
assumes it will have .seg element.
Acked-by: Arnd Bergmann <[email protected]>
Signed-off-by: Vineet Gupta <[email protected]>
---
include/asm-generic/uaccess.h | 3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/include/asm-generic/uaccess.h b/include/asm-generic/uaccess.h
index 9788568..5f6ee61 100644
--- a/include/asm-generic/uaccess.h
+++ b/include/asm-generic/uaccess.h
@@ -7,7 +7,6 @@
* address space, e.g. all NOMMU machines.
*/
#include <linux/sched.h>
-#include <linux/mm.h>
#include <linux/string.h>
#include <asm/segment.h>
@@ -32,7 +31,9 @@ static inline void set_fs(mm_segment_t fs)
}
#endif
+#ifndef segment_eq
#define segment_eq(a, b) ((a).seg == (b).seg)
+#endif
#define VERIFY_READ 0
#define VERIFY_WRITE 1
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/mutex.h | 9 ++
arch/arc/include/asm/spinlock.h | 144 +++++++++++++++++++++++++++++++++
arch/arc/include/asm/spinlock_types.h | 35 ++++++++
3 files changed, 188 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/mutex.h
create mode 100644 arch/arc/include/asm/spinlock.h
create mode 100644 arch/arc/include/asm/spinlock_types.h
diff --git a/arch/arc/include/asm/mutex.h b/arch/arc/include/asm/mutex.h
new file mode 100644
index 0000000..3be5e64
--- /dev/null
+++ b/arch/arc/include/asm/mutex.h
@@ -0,0 +1,9 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <asm-generic/mutex-xchg.h>
diff --git a/arch/arc/include/asm/spinlock.h b/arch/arc/include/asm/spinlock.h
new file mode 100644
index 0000000..f158197
--- /dev/null
+++ b/arch/arc/include/asm/spinlock.h
@@ -0,0 +1,144 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_SPINLOCK_H
+#define __ASM_SPINLOCK_H
+
+#include <asm/spinlock_types.h>
+#include <asm/processor.h>
+#include <asm/barrier.h>
+
+#define arch_spin_is_locked(x) ((x)->slock != __ARCH_SPIN_LOCK_UNLOCKED__)
+#define arch_spin_lock_flags(lock, flags) arch_spin_lock(lock)
+#define arch_spin_unlock_wait(x) \
+ do { while (arch_spin_is_locked(x)) cpu_relax(); } while (0)
+
+static inline void arch_spin_lock(arch_spinlock_t *lock)
+{
+ unsigned int tmp = __ARCH_SPIN_LOCK_LOCKED__;
+
+ __asm__ __volatile__(
+ "1: ex %0, [%1] \n"
+ " breq %0, %2, 1b \n"
+ : "+&r" (tmp)
+ : "r"(&(lock->slock)), "ir"(__ARCH_SPIN_LOCK_LOCKED__)
+ : "memory");
+}
+
+static inline int arch_spin_trylock(arch_spinlock_t *lock)
+{
+ unsigned int tmp = __ARCH_SPIN_LOCK_LOCKED__;
+
+ __asm__ __volatile__(
+ "1: ex %0, [%1] \n"
+ : "+r" (tmp)
+ : "r"(&(lock->slock))
+ : "memory");
+
+ return (tmp == __ARCH_SPIN_LOCK_UNLOCKED__);
+}
+
+static inline void arch_spin_unlock(arch_spinlock_t *lock)
+{
+ lock->slock = __ARCH_SPIN_LOCK_UNLOCKED__;
+ smp_mb();
+}
+
+/*
+ * Read-write spinlocks, allowing multiple readers but only one writer.
+ *
+ * The spinlock itself is contained in @counter and access to it is
+ * serialized with @lock_mutex.
+ *
+ * Unfair locking as Writers could be starved indefinitely by Reader(s)
+ */
+
+/* Would read_trylock() succeed? */
+#define arch_read_can_lock(x) ((x)->counter > 0)
+
+/* Would write_trylock() succeed? */
+#define arch_write_can_lock(x) ((x)->counter == __ARCH_RW_LOCK_UNLOCKED__)
+
+/* 1 - lock taken successfully */
+static inline int arch_read_trylock(arch_rwlock_t *rw)
+{
+ int ret = 0;
+
+ arch_spin_lock(&(rw->lock_mutex));
+
+ /*
+ * zero means writer holds the lock exclusively, deny Reader.
+ * Otherwise grant lock to first/subseq reader
+ */
+ if (rw->counter > 0) {
+ rw->counter--;
+ ret = 1;
+ }
+
+ arch_spin_unlock(&(rw->lock_mutex));
+
+ smp_mb();
+ return ret;
+}
+
+/* 1 - lock taken successfully */
+static inline int arch_write_trylock(arch_rwlock_t *rw)
+{
+ int ret = 0;
+
+ arch_spin_lock(&(rw->lock_mutex));
+
+ /*
+ * If reader(s) hold lock (lock < __ARCH_RW_LOCK_UNLOCKED__),
+ * deny writer. Otherwise if unlocked grant to writer
+ * Hence the claim that Linux rwlocks are unfair to writers.
+ * (can be starved for an indefinite time by readers).
+ */
+ if (rw->counter == __ARCH_RW_LOCK_UNLOCKED__) {
+ rw->counter = 0;
+ ret = 1;
+ }
+ arch_spin_unlock(&(rw->lock_mutex));
+
+ return ret;
+}
+
+static inline void arch_read_lock(arch_rwlock_t *rw)
+{
+ while (!arch_read_trylock(rw))
+ cpu_relax();
+}
+
+static inline void arch_write_lock(arch_rwlock_t *rw)
+{
+ while (!arch_write_trylock(rw))
+ cpu_relax();
+}
+
+static inline void arch_read_unlock(arch_rwlock_t *rw)
+{
+ arch_spin_lock(&(rw->lock_mutex));
+ rw->counter++;
+ arch_spin_unlock(&(rw->lock_mutex));
+}
+
+static inline void arch_write_unlock(arch_rwlock_t *rw)
+{
+ arch_spin_lock(&(rw->lock_mutex));
+ rw->counter = __ARCH_RW_LOCK_UNLOCKED__;
+ arch_spin_unlock(&(rw->lock_mutex));
+}
+
+#define arch_read_lock_flags(lock, flags) arch_read_lock(lock)
+#define arch_write_lock_flags(lock, flags) arch_write_lock(lock)
+
+#define arch_spin_relax(lock) cpu_relax()
+#define arch_read_relax(lock) cpu_relax()
+#define arch_write_relax(lock) cpu_relax()
+
+#endif /* __ASM_SPINLOCK_H */
diff --git a/arch/arc/include/asm/spinlock_types.h b/arch/arc/include/asm/spinlock_types.h
new file mode 100644
index 0000000..8276bfd
--- /dev/null
+++ b/arch/arc/include/asm/spinlock_types.h
@@ -0,0 +1,35 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_SPINLOCK_TYPES_H
+#define __ASM_SPINLOCK_TYPES_H
+
+typedef struct {
+ volatile unsigned int slock;
+} arch_spinlock_t;
+
+#define __ARCH_SPIN_LOCK_UNLOCKED__ 0
+#define __ARCH_SPIN_LOCK_LOCKED__ 1
+
+#define __ARCH_SPIN_LOCK_UNLOCKED { __ARCH_SPIN_LOCK_UNLOCKED__ }
+#define __ARCH_SPIN_LOCK_LOCKED { __ARCH_SPIN_LOCK_LOCKED__ }
+
+/*
+ * Unlocked: 0x01_00_00_00
+ * Read lock(s): 0x00_FF_00_00 to say 0x01
+ * Write lock: 0x0, but only possible if prior value "unlocked" 0x0100_0000
+ */
+typedef struct {
+ volatile unsigned int counter;
+ arch_spinlock_t lock_mutex;
+} arch_rwlock_t;
+
+#define __ARCH_RW_LOCK_UNLOCKED__ 0x01000000
+#define __ARCH_RW_LOCK_UNLOCKED { .counter = __ARCH_RW_LOCK_UNLOCKED__ }
+
+#endif
--
1.7.4.1
TBD: do_csum still needs to be written in asm
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/byteorder.h | 18 +++++++
arch/arc/include/asm/checksum.h | 101 ++++++++++++++++++++++++++++++++++++++
arch/arc/include/asm/swab.h | 99 +++++++++++++++++++++++++++++++++++++
3 files changed, 218 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/byteorder.h
create mode 100644 arch/arc/include/asm/checksum.h
create mode 100644 arch/arc/include/asm/swab.h
diff --git a/arch/arc/include/asm/byteorder.h b/arch/arc/include/asm/byteorder.h
new file mode 100644
index 0000000..9da71d4
--- /dev/null
+++ b/arch/arc/include/asm/byteorder.h
@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_BYTEORDER_H
+#define __ASM_ARC_BYTEORDER_H
+
+#ifdef CONFIG_CPU_BIG_ENDIAN
+#include <linux/byteorder/big_endian.h>
+#else
+#include <linux/byteorder/little_endian.h>
+#endif
+
+#endif /* ASM_ARC_BYTEORDER_H */
diff --git a/arch/arc/include/asm/checksum.h b/arch/arc/include/asm/checksum.h
new file mode 100644
index 0000000..1095729
--- /dev/null
+++ b/arch/arc/include/asm/checksum.h
@@ -0,0 +1,101 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Joern Rennecke <[email protected]>: Jan 2012
+ * -Insn Scheduling improvements to csum core routines.
+ * = csum_fold( ) largely derived from ARM version.
+ * = ip_fast_cum( ) to have module scheduling
+ * -gcc 4.4.x broke networking. Alias analysis needed to be primed.
+ * worked around by adding memory clobber to ip_fast_csum( )
+ *
+ * vineetg: May 2010
+ * -Rewrote ip_fast_cscum( ) and csum_fold( ) with fast inline asm
+ */
+
+#ifndef _ASM_ARC_CHECKSUM_H
+#define _ASM_ARC_CHECKSUM_H
+
+/*
+ * Fold a partial checksum
+ *
+ * The 2 swords comprising the 32bit sum are added, any carry to 16th bit
+ * added back and final sword result inverted.
+ */
+static inline __sum16 csum_fold(__wsum s)
+{
+ unsigned r = s << 16 | s >> 16; /* ror */
+ s = ~s;
+ s -= r;
+ return s >> 16;
+}
+
+/*
+ * This is a version of ip_compute_csum() optimized for IP headers,
+ * which always checksum on 4 octet boundaries.
+ */
+static inline __sum16
+ip_fast_csum(const void *iph, unsigned int ihl)
+{
+ const void *ptr = iph;
+ unsigned int tmp, tmp2, sum;
+
+ __asm__(
+ " ld.ab %0, [%3, 4] \n"
+ " ld.ab %2, [%3, 4] \n"
+ " sub %1, %4, 2 \n"
+ " lsr.f lp_count, %1, 1 \n"
+ " bcc 0f \n"
+ " add.f %0, %0, %2 \n"
+ " ld.ab %2, [%3, 4] \n"
+ "0: lp 1f \n"
+ " ld.ab %1, [%3, 4] \n"
+ " adc.f %0, %0, %2 \n"
+ " ld.ab %2, [%3, 4] \n"
+ " adc.f %0, %0, %1 \n"
+ "1: adc.f %0, %0, %2 \n"
+ " add.cs %0,%0,1 \n"
+ : "=&r"(sum), "=r"(tmp), "=&r"(tmp2), "+&r" (ptr)
+ : "r"(ihl)
+ : "cc", "lp_count", "memory");
+
+ return csum_fold(sum);
+}
+
+/*
+ * TCP pseudo Header is 12 bytes:
+ * SA [4], DA [4], zeroes [1], Proto[1], TCP Seg(hdr+data) Len [2]
+ */
+static inline __wsum
+csum_tcpudp_nofold(__be32 saddr, __be32 daddr, unsigned short len,
+ unsigned short proto, __wsum sum)
+{
+ __asm__ __volatile__(
+ " add.f %0, %0, %1 \n"
+ " adc.f %0, %0, %2 \n"
+ " adc.f %0, %0, %3 \n"
+ " adc.f %0, %0, %4 \n"
+ " adc %0, %0, 0 \n"
+ : "+&r"(sum)
+ : "r"(saddr), "r"(daddr),
+#ifdef CONFIG_CPU_BIG_ENDIAN
+ "r"(len),
+#else
+ "r"(len << 8),
+#endif
+ "r"(htons(proto))
+ : "cc");
+
+ return sum;
+}
+
+#define csum_fold csum_fold
+#define ip_fast_csum ip_fast_csum
+#define csum_tcpudp_nofold csum_tcpudp_nofold
+
+#include <asm-generic/checksum.h>
+
+#endif /* _ASM_ARC_CHECKSUM_H */
diff --git a/arch/arc/include/asm/swab.h b/arch/arc/include/asm/swab.h
new file mode 100644
index 0000000..9f6dcbb
--- /dev/null
+++ b/arch/arc/include/asm/swab.h
@@ -0,0 +1,99 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: May 2011
+ * -Support single cycle endian-swap insn in ARC700 4.10
+ *
+ * vineetg: June 2009
+ * -Better htonl implementation (5 instead of 9 ALU instructions)
+ * -Hardware assisted single cycle bswap (Use Case of ARC custom instrn)
+ */
+
+#ifndef __ASM_ARC_SWAB_H
+#define __ASM_ARC_SWAB_H
+
+#include <linux/types.h>
+
+/* Native single cycle endian swap insn */
+#ifdef CONFIG_ARC_HAS_SWAPE
+
+#define __arch_swab32(x) \
+({ \
+ unsigned int tmp = x; \
+ __asm__( \
+ " swape %0, %1 \n" \
+ : "=r" (tmp) \
+ : "r" (tmp)); \
+ tmp; \
+})
+
+#else
+
+/* Several ways of Endian-Swap Emulation for ARC
+ * 0: kernel generic
+ * 1: ARC optimised "C"
+ * 2: ARC Custom instruction
+ */
+#define ARC_BSWAP_TYPE 1
+
+#if (ARC_BSWAP_TYPE == 1) /******* Software only ********/
+
+/* The kernel default implementation of htonl is
+ * return x<<24 | x>>24 |
+ * (x & (__u32)0x0000ff00UL)<<8 | (x & (__u32)0x00ff0000UL)>>8;
+ *
+ * This generates 9 instructions on ARC (excluding the ld/st)
+ *
+ * 8051fd8c: ld r3,[r7,20] ; Mem op : Get the value to be swapped
+ * 8051fd98: asl r5,r3,24 ; get 3rd Byte
+ * 8051fd9c: lsr r2,r3,24 ; get 0th Byte
+ * 8051fda0: and r4,r3,0xff00
+ * 8051fda8: asl r4,r4,8 ; get 1st Byte
+ * 8051fdac: and r3,r3,0x00ff0000
+ * 8051fdb4: or r2,r2,r5 ; combine 0th and 3rd Bytes
+ * 8051fdb8: lsr r3,r3,8 ; 2nd Byte at correct place in Dst Reg
+ * 8051fdbc: or r2,r2,r4 ; combine 0,3 Bytes with 1st Byte
+ * 8051fdc0: or r2,r2,r3 ; combine 0,3,1 Bytes with 2nd Byte
+ * 8051fdc4: st r2,[r1,20] ; Mem op : save result back to mem
+ *
+ * Joern suggested a better "C" algorithm which is great since
+ * (1) It is portable to any architecure
+ * (2) At the same time it takes advantage of ARC ISA (rotate intrns)
+ */
+
+#define __arch_swab32(x) \
+({ unsigned long __in = (x), __tmp; \
+ __tmp = __in << 8 | __in >> 24; /* ror tmp,in,24 */ \
+ __in = __in << 24 | __in >> 8; /* ror in,in,8 */ \
+ __tmp ^= __in; \
+ __tmp &= 0xff00ff; \
+ __tmp ^ __in; \
+})
+
+#elif (ARC_BSWAP_TYPE == 2) /* Custom single cycle bwap instruction */
+
+#define __arch_swab32(x) \
+({ \
+ unsigned int tmp = x; \
+ __asm__( \
+ " .extInstruction bswap, 7, 0x00, SUFFIX_NONE, SYNTAX_2OP \n"\
+ " bswap %0, %1 \n"\
+ : "=r" (tmp) \
+ : "r" (tmp)); \
+ tmp; \
+})
+
+#endif /* ARC_BSWAP_TYPE=zzz */
+
+#endif /* CONFIG_ARC_HAS_SWAPE */
+
+#if !defined(__STRICT_ANSI__) || defined(__KERNEL__)
+#define __BYTEORDER_HAS_U64__
+#define __SWAB_64_THRU_32__
+#endif
+
+#endif
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/arcregs.h | 3 ++
arch/arc/include/asm/hw_irq.h | 7 ++++
arch/arc/include/asm/irq.h | 25 +++++++++++++
arch/arc/kernel/irq.c | 79 +++++++++++++++++++++++++++++++++++++++-
arch/arc/plat-arcfpga/irq.c | 41 +++++++++++++++++++++
5 files changed, 153 insertions(+), 2 deletions(-)
create mode 100644 arch/arc/include/asm/hw_irq.h
create mode 100644 arch/arc/include/asm/irq.h
create mode 100644 arch/arc/plat-arcfpga/irq.c
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
index 8ca8faf..3fccb04 100644
--- a/arch/arc/include/asm/arcregs.h
+++ b/arch/arc/include/asm/arcregs.h
@@ -11,6 +11,9 @@
#ifdef __KERNEL__
+/* Build Configuration Registers */
+#define ARC_REG_VECBASE_BCR 0x68
+
/* status32 Bits Positions */
#define STATUS_H_BIT 0 /* CPU Halted */
#define STATUS_E1_BIT 1 /* Int 1 enable */
diff --git a/arch/arc/include/asm/hw_irq.h b/arch/arc/include/asm/hw_irq.h
new file mode 100644
index 0000000..fd565ab
--- /dev/null
+++ b/arch/arc/include/asm/hw_irq.h
@@ -0,0 +1,7 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
diff --git a/arch/arc/include/asm/irq.h b/arch/arc/include/asm/irq.h
new file mode 100644
index 0000000..633416c
--- /dev/null
+++ b/arch/arc/include/asm/irq.h
@@ -0,0 +1,25 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_IRQ_H
+#define __ASM_ARC_IRQ_H
+
+/* Platform Inependent IRQs */
+#define TIMER0_IRQ 3
+#define TIMER1_IRQ 4
+
+#include <asm-generic/irq.h>
+
+extern void __init arc_init_IRQ(void);
+extern void __init plat_init_IRQ(void);
+extern int __init get_hw_config_num_irq(void);
+
+void __cpuinit arc_clockevent_init(void);
+void __cpuinit arc_clock_counter_setup(void);
+
+#endif
diff --git a/arch/arc/kernel/irq.c b/arch/arc/kernel/irq.c
index 16fcbe8..03a125c 100644
--- a/arch/arc/kernel/irq.c
+++ b/arch/arc/kernel/irq.c
@@ -9,8 +9,83 @@
#include <linux/interrupt.h>
#include <linux/module.h>
-#include <asm/irqflags.h>
-#include <asm/arcregs.h>
+#include <asm/sections.h>
+#include <asm/irq.h>
+
+/*
+ * Early Interrupt sub-system setup
+ * -Called very early (start_kernel -> setup_arch -> setup_processor)
+ * -Platform Independent (must for any ARC700)
+ * -Needed for each CPU (hence not foldable into init_IRQ)
+ *
+ * what it does ?
+ * -setup Vector Table Base Reg - in case Linux not linked at 0x8000_0000
+ * -Disable all IRQs (on CPU side)
+ */
+void __init arc_init_IRQ(void)
+{
+ int level_mask = level_mask;
+
+ write_aux_reg(AUX_INTR_VEC_BASE, _int_vec_base_lds);
+
+ /* Disable all IRQs: enable them as devices request */
+ write_aux_reg(AUX_IENABLE, 0);
+}
+
+/*
+ * Late Interrupt system init called from start_kernel for Boot CPU only
+ *
+ * Since slab must already be initialized, platforms can start doing any
+ * needed request_irq( )s
+ */
+void __init init_IRQ(void)
+{
+ const int irq = TIMER0_IRQ;
+
+ /*
+ * Each CPU needs to register irq of it's private TIMER0.
+ * The APIs request_percpu_irq()/enable_percpu_irq() will not be
+ * functional, if we don't "prep" the generic IRQ sub-system with
+ * the following:
+ * -Ensure that devid passed to request_percpu_irq() is indeed per cpu
+ * -disable NOAUTOEN, w/o which the device handler never gets called
+ */
+ irq_set_percpu_devid(irq);
+ irq_modify_status(irq, IRQ_NOAUTOEN, 0);
+
+ plat_init_IRQ();
+}
+
+/*
+ * "C" Entry point for any ARC ISR, called from low level vector handler
+ */
+void arch_do_IRQ(unsigned int irq, struct pt_regs *regs)
+{
+ struct pt_regs *old_regs = set_irq_regs(regs);
+
+ irq_enter();
+ generic_handle_irq(irq);
+ irq_exit();
+ set_irq_regs(old_regs);
+}
+
+int __init get_hw_config_num_irq(void)
+{
+ uint32_t val = read_aux_reg(ARC_REG_VECBASE_BCR);
+
+ switch (val & 0x03) {
+ case 0:
+ return 16;
+ case 1:
+ return 32;
+ case 2:
+ return 8;
+ default:
+ return 0;
+ }
+
+ return 0;
+}
void arch_local_irq_enable(void)
{
diff --git a/arch/arc/plat-arcfpga/irq.c b/arch/arc/plat-arcfpga/irq.c
new file mode 100644
index 0000000..a308057
--- /dev/null
+++ b/arch/arc/plat-arcfpga/irq.c
@@ -0,0 +1,41 @@
+/*
+ * ARC FPGA Platform IRQ hookups
+ *
+ * Copyright (C) 2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/interrupt.h>
+#include <asm/irq.h>
+#include <plat/memmap.h>
+
+static void arc_mask_irq(struct irq_data *data)
+{
+ arch_mask_irq(data->irq);
+}
+
+static void arc_unmask_irq(struct irq_data *data)
+{
+ arch_unmask_irq(data->irq);
+}
+
+/*
+ * There's no off-chip Interrupt Controller in the FPGA builds
+ * Below sufficies a simple model for the on-chip controller, with
+ * all interrupts being level triggered.
+ */
+static struct irq_chip fpga_chip = {
+ .irq_mask = arc_mask_irq,
+ .irq_unmask = arc_unmask_irq,
+};
+
+void __init plat_init_IRQ(void)
+{
+ int i;
+
+ for (i = 0; i < NR_IRQS; i++)
+ irq_set_chip_and_handler(i, &fpga_chip, handle_level_irq);
+}
--
1.7.4.1
ARC700 MMU provides for tagging TLB entries with a 8-bit ASID to avoid
having to flush the TLB every task switch.
It also allows for a quick way to invalidate all the TLB entries for
task useful for:
* COW sementics during fork()
* task exit()ing
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/arcregs.h | 7 ++
arch/arc/include/asm/mmu.h | 23 ++++
arch/arc/include/asm/mmu_context.h | 209 ++++++++++++++++++++++++++++++++++++
arch/arc/mm/tlb.c | 23 ++++
4 files changed, 262 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/mmu.h
create mode 100644 arch/arc/include/asm/mmu_context.h
create mode 100644 arch/arc/mm/tlb.c
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
index c6e2805..c12eb9b 100644
--- a/arch/arc/include/asm/arcregs.h
+++ b/arch/arc/include/asm/arcregs.h
@@ -85,6 +85,13 @@
#define DC_CTRL_INV_MODE_FLUSH 0x40
#define DC_CTRL_FLUSH_STATUS 0x100
+/* MMU Management regs */
+#define ARC_REG_PID 0x409
+#define ARC_REG_SCRATCH_DATA0 0x418
+
+/* Bits in MMU PID register */
+#define MMU_ENABLE (1 << 31) /* Enable MMU for process */
+
/*
* Floating Pt Registers
* Status regs are read-only (build-time) so need not be saved/restored
diff --git a/arch/arc/include/asm/mmu.h b/arch/arc/include/asm/mmu.h
new file mode 100644
index 0000000..56b0232
--- /dev/null
+++ b/arch/arc/include/asm/mmu.h
@@ -0,0 +1,23 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_MMU_H
+#define _ASM_ARC_MMU_H
+
+#ifndef __ASSEMBLY__
+
+typedef struct {
+ unsigned long asid; /* Pvt Addr-Space ID for mm */
+#ifdef CONFIG_ARC_TLB_DBG
+ struct task_struct *tsk;
+#endif
+} mm_context_t;
+
+#endif
+
+#endif
diff --git a/arch/arc/include/asm/mmu_context.h b/arch/arc/include/asm/mmu_context.h
new file mode 100644
index 0000000..d12f3de
--- /dev/null
+++ b/arch/arc/include/asm/mmu_context.h
@@ -0,0 +1,209 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: May 2011
+ * -Refactored get_new_mmu_context( ) to only handle live-mm.
+ * retiring-mm handled in other hooks
+ *
+ * Vineetg: March 25th, 2008: Bug #92690
+ * -Major rewrite of Core ASID allocation routine get_new_mmu_context
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef _ASM_ARC_MMU_CONTEXT_H
+#define _ASM_ARC_MMU_CONTEXT_H
+
+#include <asm/arcregs.h>
+#include <asm/tlb.h>
+
+#include <asm-generic/mm_hooks.h>
+
+/* ARC700 ASID Management
+ *
+ * ARC MMU provides 8-bit ASID (0..255) to TAG TLB entries, allowing entries
+ * with same vaddr (different tasks) to co-exit. This provides for
+ * "Fast Context Switch" i.e. no TLB flush on ctxt-switch
+ *
+ * Linux assigns each task a unique ASID. A simple round-robin allocation
+ * of H/w ASID is done using software tracker @asid_cache.
+ * When it reaches max 255, the allocation cycle starts afresh by flushing
+ * the entire TLB and wrapping ASID back to zero.
+ *
+ * For book-keeping, Linux uses a couple of data-structures:
+ * -mm_struct has an @asid field to keep a note of task's ASID (needed at the
+ * time of say switch_mm( )
+ * -An array of mm structs @asid_mm_map[] for asid->mm the reverse mapping,
+ * given an ASID, finding the mm struct associated.
+ *
+ * The round-robin allocation algorithm allows for ASID stealing.
+ * If asid tracker is at "x-1", a new req will allocate "x", even if "x" was
+ * already assigned to another (switched-out) task. Obviously the prev owner
+ * is marked with an invalid ASID to make it request for a new ASID when it
+ * gets scheduled next time. However its TLB entries (with ASID "x") could
+ * exist, which must be cleared before the same ASID is used by the new owner.
+ * Flushing them would be plausible but costly solution. Instead we force a
+ * allocation policy quirk, which ensures that a stolen ASID won't have any
+ * TLB entries associates, alleviating the need to flush.
+ * The quirk essentially is not allowing ASID allocated in prev cycle
+ * to be used past a roll-over in the next cycle.
+ * When this happens (i.e. task ASID > asid tracker), task needs to refresh
+ * its ASID, aligning it to current value of tracker. If the task doesn't get
+ * scheduled past a roll-over, hence its ASID is not yet realigned with
+ * tracker, such ASID is anyways safely reusable because it is
+ * gauranteed that TLB entries with that ASID wont exist.
+ */
+
+#define FIRST_ASID 0
+#define MAX_ASID 255 /* 8 bit PID field in PID Aux reg */
+#define NO_ASID (MAX_ASID + 1) /* ASID Not alloc to mmu ctxt */
+#define NUM_ASID ((MAX_ASID - FIRST_ASID) + 1)
+
+/* ASID to mm struct mapping */
+extern struct mm_struct *asid_mm_map[NUM_ASID + 1];
+
+extern int asid_cache;
+
+/*
+ * Assign a new ASID to task. If the task already has an ASID, it is
+ * relinquished.
+ */
+static inline void get_new_mmu_context(struct mm_struct *mm)
+{
+ struct mm_struct *prev_owner;
+ unsigned long flags;
+
+ local_irq_save(flags);
+
+ /*
+ * Relinquish the currently owned ASID (if any).
+ * Doing unconditionally saves a cmp-n-branch; for already unused
+ * ASID slot, the value was/remains NULL
+ */
+ asid_mm_map[mm->context.asid] = (struct mm_struct *)NULL;
+
+ /* move to new ASID */
+ if (++asid_cache > MAX_ASID) { /* ASID roll-over */
+ asid_cache = FIRST_ASID;
+ flush_tlb_all();
+ }
+
+ /*
+ * Is next ASID already owned by some-one else (we are stealing it).
+ * If so, let the orig owner be aware of this, so when it runs, it
+ * asks for a brand new ASID. This would only happen for a long-lived
+ * task with ASID from prev allocation cycle (before ASID roll-over).
+ *
+ * This might look wrong - if we are re-using some other task's ASID,
+ * won't we use it's stale TLB entries too. Actually switch_mm( ) takes
+ * care of such a case: it ensures that task with ASID from prev alloc
+ * cycle, when scheduled will refresh it's ASID: see switch_mm( ) below
+ * The stealing scenario described here will only happen if that task
+ * didn't get a chance to refresh it's ASID - implying stale entries
+ * won't exist.
+ */
+ prev_owner = asid_mm_map[asid_cache];
+ if (prev_owner)
+ prev_owner->context.asid = NO_ASID;
+
+ /* Assign new ASID to tsk */
+ asid_mm_map[asid_cache] = mm;
+ mm->context.asid = asid_cache;
+
+#ifdef CONFIG_ARC_TLB_DBG
+ pr_info("ARC_TLB_DBG: NewMM=0x%x OldMM=0x%x task_struct=0x%x Task: %s,"
+ " pid:%u, assigned asid:%lu\n",
+ (unsigned int)mm, (unsigned int)prev_owner,
+ (unsigned int)(mm->context.tsk), (mm->context.tsk)->comm,
+ (mm->context.tsk)->pid, mm->context.asid);
+#endif
+
+ write_aux_reg(ARC_REG_PID, asid_cache | MMU_ENABLE);
+
+ local_irq_restore(flags);
+}
+
+/*
+ * Initialize the context related info for a new mm_struct
+ * instance.
+ */
+static inline int
+init_new_context(struct task_struct *tsk, struct mm_struct *mm)
+{
+ mm->context.asid = NO_ASID;
+#ifdef CONFIG_ARC_TLB_DBG
+ mm->context.tsk = tsk;
+#endif
+ return 0;
+}
+
+/* Prepare the MMU for task: setup PID reg with allocated ASID
+ If task doesn't have an ASID (never alloc or stolen, get a new ASID)
+*/
+static inline void switch_mm(struct mm_struct *prev, struct mm_struct *next,
+ struct task_struct *tsk)
+{
+ /* PGD cached in MMU reg to avoid 3 mem lookups: task->mm->pgd */
+ write_aux_reg(ARC_REG_SCRATCH_DATA0, next->pgd);
+
+ /*
+ * Get a new ASID if task doesn't have a valid one. Possible when
+ * -task never had an ASID (fresh after fork)
+ * -it's ASID was stolen - past an ASID roll-over.
+ * -There's a third obscure scenario (if this task is running for the
+ * first time afer an ASID rollover), where despite having a valid
+ * ASID, we force a get for new ASID - see comments at top.
+ *
+ * Both the non-alloc scenario and first-use-after-rollover can be
+ * detected using the single condition below: NO_ASID = 256
+ * while asid_cache is always a valid ASID value (0-255).
+ */
+ if (next->context.asid > asid_cache) {
+ get_new_mmu_context(next);
+ } else {
+ /*
+ * XXX: This will never happen given the chks above
+ * BUG_ON(next->context.asid > MAX_ASID);
+ */
+ write_aux_reg(ARC_REG_PID, next->context.asid | MMU_ENABLE);
+ }
+
+}
+
+static inline void destroy_context(struct mm_struct *mm)
+{
+ unsigned long flags;
+
+ local_irq_save(flags);
+
+ asid_mm_map[mm->context.asid] = NULL;
+ mm->context.asid = NO_ASID;
+
+ local_irq_restore(flags);
+}
+
+/* it seemed that deactivate_mm( ) is a reasonable place to do book-keeping
+ * for retiring-mm. However destroy_context( ) still needs to do that because
+ * between mm_release( ) = >deactive_mm( ) and
+ * mmput => .. => __mmdrop( ) => destroy_context( )
+ * there is a good chance that task gets sched-out/in, making it's ASID valid
+ * again (this teased me for a whole day).
+ */
+#define deactivate_mm(tsk, mm) do { } while (0)
+
+static inline void activate_mm(struct mm_struct *prev, struct mm_struct *next)
+{
+ write_aux_reg(ARC_REG_SCRATCH_DATA0, next->pgd);
+
+ /* Unconditionally get a new ASID */
+ get_new_mmu_context(next);
+
+}
+
+#define enter_lazy_tlb(mm, tsk)
+
+#endif /* __ASM_ARC_MMU_CONTEXT_H */
diff --git a/arch/arc/mm/tlb.c b/arch/arc/mm/tlb.c
new file mode 100644
index 0000000..f1edae2
--- /dev/null
+++ b/arch/arc/mm/tlb.c
@@ -0,0 +1,23 @@
+/*
+ * TLB Management (flush/create/diagnostics) for ARC700
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/module.h>
+#include <asm/arcregs.h>
+#include <asm/mmu_context.h>
+#include <asm/tlb.h>
+
+/* A copy of the ASID from the PID reg is kept in asid_cache */
+int asid_cache = FIRST_ASID;
+
+/* ASID to mm struct mapping. We have one extra entry corresponding to
+ * NO_ASID to save us a compare when clearing the mm entry for old asid
+ * see get_new_mmu_context (asm-arc/mmu_context.h)
+ */
+struct mm_struct *asid_mm_map[NUM_ASID + 1];
--
1.7.4.1
* ARC700 has VIPT L1 Caches (L1 only)
* Caches don't snoop and are not coherent
* Given the PAGE_SIZE and Cache associativity, we don't support aliasing
D$ configurations, but so allow aliasing I$ configs
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/arcregs.h | 80 ++++
arch/arc/include/asm/cache.h | 54 +++
arch/arc/include/asm/cachectl.h | 30 ++
arch/arc/include/asm/cacheflush.h | 67 +++
arch/arc/mm/cache_arc700.c | 838 +++++++++++++++++++++++++++++++++++++
5 files changed, 1069 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/cachectl.h
create mode 100644 arch/arc/include/asm/cacheflush.h
create mode 100644 arch/arc/mm/cache_arc700.c
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
index 5131bb3..c6e2805 100644
--- a/arch/arc/include/asm/arcregs.h
+++ b/arch/arc/include/asm/arcregs.h
@@ -58,6 +58,33 @@
#define TIMER_CTRL_IE (1 << 0) /* Interupt when Count reachs limit */
#define TIMER_CTRL_NH (1 << 1) /* Count only when CPU NOT halted */
+/* Instruction cache related Auxiliary registers */
+#define ARC_REG_IC_BCR 0x77 /* Build Config reg */
+#define ARC_REG_IC_IVIC 0x10
+#define ARC_REG_IC_CTRL 0x11
+#define ARC_REG_IC_IVIL 0x19
+#if (CONFIG_ARC_MMU_VER > 2)
+#define ARC_REG_IC_PTAG 0x1E
+#endif
+
+/* Bit val in IC_CTRL */
+#define IC_CTRL_CACHE_DISABLE 0x1
+
+/* Data cache related Auxiliary registers */
+#define ARC_REG_DC_BCR 0x72
+#define ARC_REG_DC_IVDC 0x47
+#define ARC_REG_DC_CTRL 0x48
+#define ARC_REG_DC_IVDL 0x4A
+#define ARC_REG_DC_FLSH 0x4B
+#define ARC_REG_DC_FLDL 0x4C
+#if (CONFIG_ARC_MMU_VER > 2)
+#define ARC_REG_DC_PTAG 0x5C
+#endif
+
+/* Bit val in DC_CTRL */
+#define DC_CTRL_INV_MODE_FLUSH 0x40
+#define DC_CTRL_FLUSH_STATUS 0x100
+
/*
* Floating Pt Registers
* Status regs are read-only (build-time) so need not be saved/restored
@@ -132,6 +159,31 @@
#endif
+#define READ_BCR(reg, into) \
+{ \
+ unsigned int tmp; \
+ tmp = read_aux_reg(reg); \
+ if (sizeof(tmp) == sizeof(into)) { \
+ into = *((typeof(into) *)&tmp); \
+ } else { \
+ extern void bogus_undefined(void); \
+ bogus_undefined(); \
+ } \
+}
+
+#define WRITE_BCR(reg, into) \
+{ \
+ unsigned int tmp; \
+ if (sizeof(tmp) == sizeof(into)) { \
+ tmp = (*(unsigned int *)(into)); \
+ write_aux_reg(reg, tmp); \
+ } else { \
+ extern void bogus_undefined(void); \
+ bogus_undefined(); \
+ } \
+}
+
+
#ifdef CONFIG_ARC_FPU_SAVE_RESTORE
/* These DPFP regs need to be saved/restored across ctx-sw */
struct arc_fpu {
@@ -141,6 +193,34 @@ struct arc_fpu {
};
#endif
+/*
+ ***************************************************************
+ * Build Configuration Registers, with encoded hardware config
+ */
+
+struct bcr_cache {
+#ifdef CONFIG_CPU_BIG_ENDIAN
+ unsigned int pad:12, line_len:4, sz:4, config:4, ver:8;
+#else
+ unsigned int ver:8, config:4, sz:4, line_len:4, pad:12;
+#endif
+};
+
+/*
+ *******************************************************************
+ * Generic structures to hold build configuration used at runtime
+ */
+
+struct cpuinfo_arc_cache {
+ unsigned int has_aliasing, sz, line_len, assoc, ver;
+};
+
+struct cpuinfo_arc {
+ struct cpuinfo_arc_cache icache, dcache;
+};
+
+extern struct cpuinfo_arc cpuinfo_arc700[];
+
#endif /* __ASEMBLY__ */
#endif /* __KERNEL__ */
diff --git a/arch/arc/include/asm/cache.h b/arch/arc/include/asm/cache.h
index 30c72a4..6632273 100644
--- a/arch/arc/include/asm/cache.h
+++ b/arch/arc/include/asm/cache.h
@@ -18,4 +18,58 @@
#define L1_CACHE_BYTES (1 << L1_CACHE_SHIFT)
+#define ARC_ICACHE_WAYS 2
+#define ARC_DCACHE_WAYS 4
+
+/* Helpers */
+#define ARC_ICACHE_LINE_LEN L1_CACHE_BYTES
+#define ARC_DCACHE_LINE_LEN L1_CACHE_BYTES
+
+#define ICACHE_LINE_MASK (~(ARC_ICACHE_LINE_LEN - 1))
+#define DCACHE_LINE_MASK (~(ARC_DCACHE_LINE_LEN - 1))
+
+#if ARC_ICACHE_LINE_LEN != ARC_DCACHE_LINE_LEN
+#error "Need to fix some code as I/D cache lines not same"
+#else
+#define is_not_cache_aligned(p) ((unsigned long)p & (~DCACHE_LINE_MASK))
+#endif
+
+#ifndef __ASSEMBLY__
+
+/* Uncached access macros */
+#define arc_read_uncached_32(ptr) \
+({ \
+ unsigned int __ret; \
+ __asm__ __volatile__( \
+ " ld.di %0, [%1] \n" \
+ : "=r"(__ret) \
+ : "r"(ptr)); \
+ __ret; \
+})
+
+#define arc_write_uncached_32(ptr, data)\
+({ \
+ __asm__ __volatile__( \
+ " st.di %0, [%1] \n" \
+ : \
+ : "r"(data), "r"(ptr)); \
+})
+
+/* used to give SHMLBA a value to avoid Cache Aliasing */
+extern unsigned int ARC_shmlba;
+
+#define ARCH_DMA_MINALIGN L1_CACHE_BYTES
+
+/*
+ * ARC700 doesn't cache any access in top 256M.
+ * Ideal for wiring memory mapped peripherals as we don't need to do
+ * explicit uncached accesses (LD.di/ST.di) hence more portable drivers
+ */
+#define ARC_UNCACHED_ADDR_SPACE 0xc0000000
+
+extern void arc_cache_init(void);
+extern char *arc_cache_mumbojumbo(int cpu_id, char *buf, int len);
+extern void __init read_decode_cache_bcr(void);
+#endif
+
#endif /* _ASM_CACHE_H */
diff --git a/arch/arc/include/asm/cachectl.h b/arch/arc/include/asm/cachectl.h
new file mode 100644
index 0000000..d0ba876
--- /dev/null
+++ b/arch/arc/include/asm/cachectl.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ARC_ASM_CACHECTL_H
+#define __ARC_ASM_CACHECTL_H
+
+/*
+ * ARC ABI flags defined for Android's finegrained cacheflush requirements
+ */
+#define CF_I_INV 0x0002
+#define CF_D_FLUSH 0x0010
+#define CF_D_INV 0x0020
+#define CF_D_FLUSH_INV 0x0030
+#define CF_PHY_ADDR 0x1000
+
+#define CF_DEFAULT (CF_I_INV | CF_D_FLUSH)
+
+/*
+ * Standard flags expected by cacheflush system call users
+ */
+#define ICACHE CF_I_INV
+#define DCACHE CF_D_FLUSH
+#define BCACHE (CF_I_INV | CF_D_FLUSH)
+
+#endif
diff --git a/arch/arc/include/asm/cacheflush.h b/arch/arc/include/asm/cacheflush.h
new file mode 100644
index 0000000..97ee96f
--- /dev/null
+++ b/arch/arc/include/asm/cacheflush.h
@@ -0,0 +1,67 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: May 2011: for Non-aliasing VIPT D-cache following can be NOPs
+ * -flush_cache_dup_mm (fork)
+ * -likewise for flush_cache_mm (exit/execve)
+ * -likewise for flush_cache_{range,page} (munmap, exit, COW-break)
+ *
+ * vineetg: April 2008
+ * -Added a critical CacheLine flush to copy_to_user_page( ) which
+ * was causing gdbserver to not setup breakpoints consistently
+ */
+
+#ifndef _ASM_CACHEFLUSH_H
+#define _ASM_CACHEFLUSH_H
+
+#include <linux/mm.h>
+
+void flush_cache_all(void);
+
+void flush_icache_range(unsigned long start, unsigned long end);
+void flush_icache_page(struct vm_area_struct *vma, struct page *page);
+void flush_icache_range_vaddr(unsigned long paddr, unsigned long u_vaddr,
+ int len);
+
+#define ARCH_IMPLEMENTS_FLUSH_DCACHE_PAGE 1
+
+void flush_dcache_page(struct page *page);
+
+void dma_cache_wback_inv(unsigned long start, unsigned long sz);
+void dma_cache_inv(unsigned long start, unsigned long sz);
+void dma_cache_wback(unsigned long start, unsigned long sz);
+
+#define flush_dcache_mmap_lock(mapping) do { } while (0)
+#define flush_dcache_mmap_unlock(mapping) do { } while (0)
+
+/* TBD: optimize this */
+#define flush_cache_vmap(start, end) flush_cache_all()
+#define flush_cache_vunmap(start, end) flush_cache_all()
+
+/*
+ * VM callbacks when entire/range of user-space V-P mappings are
+ * torn-down/get-invalidated
+ *
+ * Currently we don't support D$ aliasing configs for our VIPT caches
+ * NOPS for VIPT Cache with non-aliasing D$ configurations only
+ */
+#define flush_cache_dup_mm(mm) /* called on fork */
+#define flush_cache_mm(mm) /* called on munmap/exit */
+#define flush_cache_range(mm, u_vstart, u_vend)
+#define flush_cache_page(vma, u_vaddr, pfn) /* PF handling/COW-break */
+
+#define copy_to_user_page(vma, page, vaddr, dst, src, len) \
+do { \
+ memcpy(dst, src, len); \
+ if (vma->vm_flags & VM_EXEC) \
+ flush_icache_range_vaddr((unsigned long)(dst), vaddr, len);\
+} while (0)
+
+#define copy_from_user_page(vma, page, vaddr, dst, src, len) \
+ memcpy(dst, src, len); \
+
+#endif
diff --git a/arch/arc/mm/cache_arc700.c b/arch/arc/mm/cache_arc700.c
new file mode 100644
index 0000000..9f8f567
--- /dev/null
+++ b/arch/arc/mm/cache_arc700.c
@@ -0,0 +1,838 @@
+/*
+ * ARC700 VIPT Cache Management
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: May 2011: for Non-aliasing VIPT D-cache following can be NOPs
+ * -flush_cache_dup_mm (fork)
+ * -likewise for flush_cache_mm (exit/execve)
+ * -likewise for flush_cache_range,flush_cache_page (munmap, exit, COW-break)
+ *
+ * vineetg: Apr 2011
+ * -Now that MMU can support larger pg sz (16K), the determiniation of
+ * aliasing shd not be based on assumption of 8k pg
+ *
+ * vineetg: Mar 2011
+ * -optimised version of flush_icache_range( ) for making I/D coherent
+ * when vaddr is available (agnostic of num of aliases)
+ *
+ * vineetg: Mar 2011
+ * -Added documentation about I-cache aliasing on ARC700 and the way it
+ * was handled up until MMU V2.
+ * -Spotted a three year old bug when killing the 4 aliases, which needs
+ * bottom 2 bits, so we need to do paddr | {0x00, 0x01, 0x02, 0x03}
+ * instead of paddr | {0x00, 0x01, 0x10, 0x11}
+ * (Rajesh you owe me one now)
+ *
+ * vineetg: Dec 2010
+ * -Off-by-one error when computing num_of_lines to flush
+ * This broke signal handling with bionic which uses synthetic sigret stub
+ *
+ * vineetg: Mar 2010
+ * -GCC can't generate ZOL for core cache flush loops.
+ * Conv them into iterations based as opposed to while (start < end) types
+ *
+ * Vineetg: July 2009
+ * -In I-cache flush routine we used to chk for aliasing for every line INV.
+ * Instead now we setup routines per cache geometry and invoke them
+ * via function pointers.
+ *
+ * Vineetg: Jan 2009
+ * -Cache Line flush routines used to flush an extra line beyond end addr
+ * because check was while (end >= start) instead of (end > start)
+ * =Some call sites had to work around by doing -1, -4 etc to end param
+ * =Some callers didnt care. This was spec bad in case of INV routines
+ * which would discard valid data (cause of the horrible ext2 bug
+ * in ARC IDE driver)
+ *
+ * vineetg: June 11th 2008: Fixed flush_icache_range( )
+ * -Since ARC700 caches are not coherent (I$ doesnt snoop D$) both need
+ * to be flushed, which it was not doing.
+ * -load_module( ) passes vmalloc addr (Kernel Virtual Addr) to the API,
+ * however ARC cache maintenance OPs require PHY addr. Thus need to do
+ * vmalloc_to_phy.
+ * -Also added optimisation there, that for range > PAGE SIZE we flush the
+ * entire cache in one shot rather than line by line. For e.g. a module
+ * with Code sz 600k, old code flushed 600k worth of cache (line-by-line),
+ * while cache is only 16 or 32k.
+ */
+
+#include <linux/module.h>
+#include <linux/mm.h>
+#include <linux/sched.h>
+#include <linux/cache.h>
+#include <linux/mmu_context.h>
+#include <linux/syscalls.h>
+#include <linux/uaccess.h>
+#include <asm/cacheflush.h>
+#include <asm/cachectl.h>
+#include <asm/setup.h>
+
+
+#ifdef CONFIG_ARC_HAS_ICACHE
+static void __ic_line_inv_no_alias(unsigned long, int);
+static void __ic_line_inv_2_alias(unsigned long, int);
+static void __ic_line_inv_4_alias(unsigned long, int);
+
+/* Holds the ptr to flush routine, dependign on size due to aliasing issues */
+static void (*___flush_icache_rtn) (unsigned long, int);
+#endif
+
+/*
+ * Read the Cache Build Confuration Registers, Decode them and save into
+ * the cpuinfo structure for later use.
+ * No Validation done here, simply read/convert the BCRs
+ */
+void __init read_decode_cache_bcr(void)
+{
+ struct bcr_cache ibcr, dbcr;
+ struct cpuinfo_arc_cache *p_ic, *p_dc;
+ unsigned int cpu = smp_processor_id();
+
+ p_ic = &cpuinfo_arc700[cpu].icache;
+ READ_BCR(ARC_REG_IC_BCR, ibcr);
+
+ if (ibcr.config == 0x3)
+ p_ic->assoc = 2;
+ p_ic->line_len = 8 << ibcr.line_len;
+ p_ic->sz = 0x200 << ibcr.sz;
+ p_ic->ver = ibcr.ver;
+
+ p_dc = &cpuinfo_arc700[cpu].dcache;
+ READ_BCR(ARC_REG_DC_BCR, dbcr);
+
+ if (dbcr.config == 0x2)
+ p_dc->assoc = 4;
+ p_dc->line_len = 16 << dbcr.line_len;
+ p_dc->sz = 0x200 << dbcr.sz;
+ p_dc->ver = dbcr.ver;
+}
+
+/*
+ * 1. Validate the Cache Geomtery (compile time config matches hardware)
+ * 2. If I-cache suffers from aliasing, setup work arounds (difft flush rtn)
+ * (aliasing D-cache configurations are not supported YET)
+ * 3. Enable the Caches, setup default flush mode for D-Cache
+ * 3. Calculate the SHMLBA used by user space
+ */
+void __init arc_cache_init(void)
+{
+ unsigned int temp;
+#ifdef CONFIG_ARC_CACHE
+ unsigned int cpu = smp_processor_id();
+#endif
+#ifdef CONFIG_ARC_HAS_ICACHE
+ struct cpuinfo_arc_cache *ic;
+#endif
+#ifdef CONFIG_ARC_HAS_DCACHE
+ struct cpuinfo_arc_cache *dc;
+#endif
+ int way_pg_ratio = way_pg_ratio;
+
+#ifdef CONFIG_ARC_HAS_ICACHE
+ ic = &cpuinfo_arc700[cpu].icache;
+
+ /*
+ * if Cache way size is <= page size then no aliasing exhibited
+ * otherwise ratio determines num of aliases.
+ * e.g. 32K I$, 2 way set assoc, 8k pg size
+ * way-sz = 32k/2 = 16k
+ * way-pg-ratio = 16k/8k = 2, so 2 aliases possible
+ * (meaning 1 line could be in 2 possible locations).
+ */
+ way_pg_ratio = ic->sz / ARC_ICACHE_WAYS / PAGE_SIZE;
+ switch (way_pg_ratio) {
+ case 0:
+ case 1:
+ ___flush_icache_rtn = __ic_line_inv_no_alias;
+ break;
+ case 2:
+ ___flush_icache_rtn = __ic_line_inv_2_alias;
+ break;
+ case 4:
+ ___flush_icache_rtn = __ic_line_inv_4_alias;
+ break;
+ default:
+ panic("Unsupported I-Cache Sz\n");
+ }
+#endif
+
+ /* Enable/disable I-Cache */
+ temp = read_aux_reg(ARC_REG_IC_CTRL);
+
+#ifdef CONFIG_ARC_HAS_ICACHE
+ temp &= ~IC_CTRL_CACHE_DISABLE;
+#else
+ temp |= IC_CTRL_CACHE_DISABLE;
+#endif
+
+ write_aux_reg(ARC_REG_IC_CTRL, temp);
+
+#ifdef CONFIG_ARC_HAS_DCACHE
+ dc = &cpuinfo_arc700[cpu].dcache;
+
+ /* check for D-Cache aliasing */
+ if ((dc->sz / ARC_DCACHE_WAYS) > PAGE_SIZE)
+ panic("D$ aliasing not handled right now\n");
+#endif
+
+ /* Set the default Invalidate Mode to "simpy discard dirty lines"
+ * as this is more frequent then flush before invalidate
+ * Ofcourse we toggle this default behviour when desired
+ */
+ temp = read_aux_reg(ARC_REG_DC_CTRL);
+ temp &= ~DC_CTRL_INV_MODE_FLUSH;
+
+#ifdef CONFIG_ARC_HAS_DCACHE
+ /* Enable D-Cache: Clear Bit 0 */
+ write_aux_reg(ARC_REG_DC_CTRL, temp & ~IC_CTRL_CACHE_DISABLE);
+#else
+ /* Flush D cache */
+ write_aux_reg(ARC_REG_DC_FLSH, 0x1);
+ /* Disable D cache */
+ write_aux_reg(ARC_REG_DC_CTRL, temp | IC_CTRL_CACHE_DISABLE);
+#endif
+
+ return;
+}
+
+#define OP_INV 0x1
+#define OP_FLUSH 0x2
+#define OP_FLUSH_N_INV 0x3
+
+#ifdef CONFIG_ARC_HAS_DCACHE
+
+/***************************************************************
+ * Machine specific helpers for Entire D-Cache or Per Line ops
+ */
+
+static inline void wait_for_flush(void)
+{
+ while (read_aux_reg(ARC_REG_DC_CTRL) & DC_CTRL_FLUSH_STATUS)
+ ;
+}
+
+/*
+ * Operation on Entire D-Cache
+ * @cacheop = {OP_INV, OP_FLUSH, OP_FLUSH_N_INV}
+ * Note that constant propagation ensures all the checks are gone
+ * in generated code
+ */
+static inline void __dc_entire_op(const int cacheop)
+{
+ unsigned long flags, tmp = tmp;
+ int aux;
+
+ local_irq_save(flags);
+
+ if (cacheop == OP_FLUSH_N_INV) {
+ /* Dcache provides 2 cmd: FLUSH or INV
+ * INV inturn has sub-modes: DISCARD or FLUSH-BEFORE
+ * flush-n-inv is achieved by INV cmd but with IM=1
+ * Default INV sub-mode is DISCARD, which needs to be toggled
+ */
+ tmp = read_aux_reg(ARC_REG_DC_CTRL);
+ write_aux_reg(ARC_REG_DC_CTRL, tmp | DC_CTRL_INV_MODE_FLUSH);
+ }
+
+ if (cacheop & OP_INV) /* Inv or flush-n-inv use same cmd reg */
+ aux = ARC_REG_DC_IVDC;
+ else
+ aux = ARC_REG_DC_FLSH;
+
+ write_aux_reg(aux, 0x1);
+
+ if (cacheop & OP_FLUSH) /* flush / flush-n-inv both wait */
+ wait_for_flush();
+
+ /* Switch back the DISCARD ONLY Invalidate mode */
+ if (cacheop == OP_FLUSH_N_INV)
+ write_aux_reg(ARC_REG_DC_CTRL, tmp & ~DC_CTRL_INV_MODE_FLUSH);
+
+ local_irq_restore(flags);
+}
+
+/*
+ * Per Line Operation on D-Cache
+ * Doesn't deal with type-of-op/IRQ-disabling/waiting-for-flush-to-complete
+ * It's sole purpose is to help gcc generate ZOL
+ */
+static inline void __dc_line_loop(unsigned long start, unsigned long sz,
+ int aux_reg)
+{
+ int num_lines, slack;
+
+ /* Ensure we properly floor/ceil the non-line aligned/sized requests
+ * and have @start - aligned to cache line and integral @num_lines.
+ * This however can be avoided for page sized since:
+ * -@start will be cache-line aligned already (being page aligned)
+ * -@sz will be integral multiple of line size (being page sized).
+ */
+ if (!(__builtin_constant_p(sz) && sz == PAGE_SIZE)) {
+ slack = start & ~DCACHE_LINE_MASK;
+ sz += slack;
+ start -= slack;
+ }
+
+ num_lines = DIV_ROUND_UP(sz, ARC_DCACHE_LINE_LEN);
+
+ while (num_lines-- > 0) {
+#if (CONFIG_ARC_MMU_VER > 2)
+ /*
+ * Just as for I$, in MMU v3, D$ ops also require
+ * "tag" bits in DC_PTAG, "index" bits in FLDL,IVDL ops
+ * But we pass phy addr for both. This works since Linux
+ * doesn't support aliasing configs for D$, yet.
+ * Thus paddr is enough to provide both tag and index.
+ */
+ write_aux_reg(ARC_REG_DC_PTAG, start);
+#endif
+ write_aux_reg(aux_reg, start);
+ start += ARC_DCACHE_LINE_LEN;
+ }
+}
+
+/*
+ * D-Cache : Per Line INV (discard or wback+discard) or FLUSH (wback)
+ */
+static inline void __dc_line_op(unsigned long start, unsigned long sz,
+ const int cacheop)
+{
+ unsigned long flags, tmp = tmp;
+ int aux;
+
+ local_irq_save(flags);
+
+ if (cacheop == OP_FLUSH_N_INV) {
+ /*
+ * Dcache provides 2 cmd: FLUSH or INV
+ * INV inturn has sub-modes: DISCARD or FLUSH-BEFORE
+ * flush-n-inv is achieved by INV cmd but with IM=1
+ * Default INV sub-mode is DISCARD, which needs to be toggled
+ */
+ tmp = read_aux_reg(ARC_REG_DC_CTRL);
+ write_aux_reg(ARC_REG_DC_CTRL, tmp | DC_CTRL_INV_MODE_FLUSH);
+ }
+
+ if (cacheop & OP_INV) /* Inv / flush-n-inv use same cmd reg */
+ aux = ARC_REG_DC_IVDL;
+ else
+ aux = ARC_REG_DC_FLDL;
+
+ __dc_line_loop(start, sz, aux);
+
+ if (cacheop & OP_FLUSH) /* flush / flush-n-inv both wait */
+ wait_for_flush();
+
+ /* Switch back the DISCARD ONLY Invalidate mode */
+ if (cacheop == OP_FLUSH_N_INV)
+ write_aux_reg(ARC_REG_DC_CTRL, tmp & ~DC_CTRL_INV_MODE_FLUSH);
+
+ local_irq_restore(flags);
+}
+
+#else
+
+#define __dc_entire_op(cacheop)
+#define __dc_line_op(start, sz, cacheop)
+
+#endif /* CONFIG_ARC_HAS_DCACHE */
+
+
+#ifdef CONFIG_ARC_HAS_ICACHE
+
+/*
+ * I-Cache Aliasing in ARC700 VIPT caches
+ *
+ * For fetching code from I$, ARC700 uses vaddr (embedded in program code)
+ * to "index" into SET of cache-line and paddr from MMU to match the TAG
+ * in the WAYS of SET.
+ *
+ * However the CDU iterface (to flush/inv) lines from software, only takes
+ * paddr (to have simpler hardware interface). For simpler cases, using paddr
+ * alone suffices.
+ * e.g. 2-way-set-assoc, 16K I$ (8k MMU pg sz, 32b cache line size):
+ * way_sz = cache_sz / num_ways = 16k/2 = 8k
+ * num_sets = way_sz / line_sz = 8k/32 = 256 => 8 bits
+ * Ignoring the bottom 5 bits corresp to the off within a 32b cacheline,
+ * bits req for calc set-index = bits 12:5 (0 based). Since this range fits
+ * inside the bottom 13 bits of paddr, which are same for vaddr and paddr
+ * (with 8k pg sz), paddr alone can be safely used by CDU to unambigously
+ * locate a cache-line.
+ *
+ * However for a difft sized cache, say 32k I$, above math yields need
+ * for 14 bits of vaddr to locate a cache line, which can't be provided by
+ * paddr, since the bit 13 (0 based) might differ between the two.
+ *
+ * This lack of extra bits needed for correct line addressing, defines the
+ * classical problem of Cache aliasing with VIPT architectures
+ * num_aliases = 1 << extra_bits
+ * e.g. 2-way-set-assoc, 32K I$ with 8k MMU pg sz => 2 aliases
+ * 2-way-set-assoc, 64K I$ with 8k MMU pg sz => 4 aliases
+ * 2-way-set-assoc, 16K I$ with 8k MMU pg sz => NO aliases
+ *
+ * ------------------
+ * MMU v1/v2 (Fixed Page Size 8k)
+ * ------------------
+ * The solution was to provide CDU with these additonal vaddr bits. These
+ * would be bits [x:13], x would depend on cache-geom.
+ * H/w folks chose [17:13] to be a future safe range, and moreso these 5 bits
+ * of vaddr could easily be "stuffed" in the paddr as bits [4:0] since the
+ * orig 5 bits of paddr were anyways ignored by CDU line ops, as they
+ * represent the offset within cache-line. The adv of using this "clumsy"
+ * interface for additional info was no new reg was needed in CDU.
+ *
+ * 17:13 represented the max num of bits passable, actual bits needed were
+ * fewer, based on the num-of-aliases possible.
+ * -for 2 alias possibility, only bit 13 needed (32K cache)
+ * -for 4 alias possibility, bits 14:13 needed (64K cache)
+ *
+ * Since vaddr was not available for all instances of I$ flush req by core
+ * kernel, the only safe way (non-optimal though) was to kill all possible
+ * lines which could represent an alias (even if they didnt represent one
+ * in execution).
+ * e.g. for 64K I$, 4 aliases possible, so we did
+ * flush start
+ * flush start | 0x01
+ * flush start | 0x2
+ * flush start | 0x3
+ *
+ * The penalty was invoking the operation itself, since tag match is anyways
+ * paddr based, a line which didn't represent an alias would not match the
+ * paddr, hence wont be killed
+ *
+ * Note that aliasing concerns are independent of line-sz for a given cache
+ * geometry (size + set_assoc) because the extra bits required by line-sz are
+ * reduced from the set calc.
+ * e.g. 2-way-set-assoc, 32K I$ with 8k MMU pg sz and using math above
+ * 32b line-sz: 9 bits set-index-calc, 5 bits offset-in-line => 1 extra bit
+ * 64b line-sz: 8 bits set-index-calc, 6 bits offset-in-line => 1 extra bit
+ *
+ * ------------------
+ * MMU v3
+ * ------------------
+ * This ver of MMU supports var page sizes (1k-16k) - Linux will support
+ * 8k (default), 16k and 4k.
+ * However from hardware perspective, smaller page sizes aggrevate aliasing
+ * meaning more vaddr bits needed to disambiguate the cache-line-op ;
+ * the existing scheme of piggybacking won't work for certain configurations.
+ * Two new registers IC_PTAG and DC_PTAG inttoduced.
+ * "tag" bits are provided in PTAG, index bits in existing IVIL/IVDL/FLDL regs
+ */
+
+/***********************************************************
+ * Machine specific helpers for per line I-Cache invalidate.
+ * 3 routines to accpunt for 1, 2, 4 aliases possible
+ */
+
+static void __ic_line_inv_no_alias(unsigned long start, int num_lines)
+{
+ while (num_lines-- > 0) {
+#if (CONFIG_ARC_MMU_VER > 2)
+ write_aux_reg(ARC_REG_IC_PTAG, start);
+#endif
+ write_aux_reg(ARC_REG_IC_IVIL, start);
+ start += ARC_ICACHE_LINE_LEN;
+ }
+}
+
+static void __ic_line_inv_2_alias(unsigned long start, int num_lines)
+{
+ while (num_lines-- > 0) {
+
+#if (CONFIG_ARC_MMU_VER > 2)
+ /*
+ * MMU v3, CDU prog model (for line ops) now uses a new IC_PTAG
+ * reg to pass the "tag" bits and existing IVIL reg only looks
+ * at bits relevant for "index" (details above)
+ * Programming Notes:
+ * -when writing tag to PTAG reg, bit chopping can be avoided,
+ * CDU ignores non-tag bits.
+ * -Ideally "index" must be computed from vaddr, but it is not
+ * avail in these rtns. So to be safe, we kill the lines in all
+ * possible indexes corresp to num of aliases possible for
+ * given cache config.
+ */
+ write_aux_reg(ARC_REG_IC_PTAG, start);
+ write_aux_reg(ARC_REG_IC_IVIL,
+ start & ~(0x1 << PAGE_SHIFT));
+ write_aux_reg(ARC_REG_IC_IVIL, start | (0x1 << PAGE_SHIFT));
+#else
+ write_aux_reg(ARC_REG_IC_IVIL, start);
+ write_aux_reg(ARC_REG_IC_IVIL, start | 0x01);
+#endif
+ start += ARC_ICACHE_LINE_LEN;
+ }
+}
+
+static void __ic_line_inv_4_alias(unsigned long start, int num_lines)
+{
+ while (num_lines-- > 0) {
+
+#if (CONFIG_ARC_MMU_VER > 2)
+ write_aux_reg(ARC_REG_IC_PTAG, start);
+
+ write_aux_reg(ARC_REG_IC_IVIL,
+ start & ~(0x3 << PAGE_SHIFT));
+ write_aux_reg(ARC_REG_IC_IVIL,
+ start & ~(0x2 << PAGE_SHIFT));
+ write_aux_reg(ARC_REG_IC_IVIL,
+ start & ~(0x1 << PAGE_SHIFT));
+ write_aux_reg(ARC_REG_IC_IVIL, start | (0x3 << PAGE_SHIFT));
+#else
+ write_aux_reg(ARC_REG_IC_IVIL, start);
+ write_aux_reg(ARC_REG_IC_IVIL, start | 0x01);
+ write_aux_reg(ARC_REG_IC_IVIL, start | 0x02);
+ write_aux_reg(ARC_REG_IC_IVIL, start | 0x03);
+#endif
+ start += ARC_ICACHE_LINE_LEN;
+ }
+}
+
+static void __ic_line_inv(unsigned long start, unsigned long sz)
+{
+ unsigned long flags;
+ int num_lines, slack;
+
+ /*
+ * Ensure we properly floor/ceil the non-line aligned/sized requests
+ * and have @start - aligned to cache line, and integral @num_lines
+ * However page sized flushes can be compile time optimised.
+ * -@start will be cache-line aligned already (being page aligned)
+ * -@sz will be integral multiple of line size (being page sized).
+ */
+ if (!(__builtin_constant_p(sz) && sz == PAGE_SIZE)) {
+ slack = start & ~ICACHE_LINE_MASK;
+ sz += slack;
+ start -= slack;
+ }
+
+ num_lines = DIV_ROUND_UP(sz, ARC_ICACHE_LINE_LEN);
+
+ local_irq_save(flags);
+ (*___flush_icache_rtn) (start, num_lines);
+ local_irq_restore(flags);
+}
+
+/* Unlike routines above, having vaddr for flush op (along with paddr),
+ * prevents the need to speculatively kill the lines in multiple sets
+ * based on ratio of way_sz : pg_sz
+ */
+static void __ic_line_inv_vaddr(unsigned long phy_start,
+ unsigned long vaddr, unsigned long sz)
+{
+ unsigned long flags;
+ int num_lines, slack;
+ unsigned int addr;
+
+ slack = phy_start & ~ICACHE_LINE_MASK;
+ sz += slack;
+ phy_start -= slack;
+ num_lines = DIV_ROUND_UP(sz, ARC_ICACHE_LINE_LEN);
+
+#if (CONFIG_ARC_MMU_VER > 2)
+ vaddr &= ~ICACHE_LINE_MASK;
+ addr = phy_start;
+#else
+ /* bits 17:13 of vaddr go as bits 4:0 of paddr */
+ addr = phy_start | ((vaddr >> 13) & 0x1F);
+#endif
+
+ local_irq_save(flags);
+ while (num_lines-- > 0) {
+#if (CONFIG_ARC_MMU_VER > 2)
+ /* tag comes from phy addr */
+ write_aux_reg(ARC_REG_IC_PTAG, addr);
+
+ /* index bits come from vaddr */
+ write_aux_reg(ARC_REG_IC_IVIL, vaddr);
+ vaddr += ARC_ICACHE_LINE_LEN;
+#else
+ /* this paddr contains vaddrs bits as needed */
+ write_aux_reg(ARC_REG_IC_IVIL, addr);
+#endif
+ addr += ARC_ICACHE_LINE_LEN;
+ }
+ local_irq_restore(flags);
+}
+
+#else
+
+#define __ic_line_inv(start, sz)
+#define __ic_line_inv_vaddr(pstart, vstart, sz)
+
+#endif /* CONFIG_ARC_HAS_ICACHE */
+
+
+/***********************************************************
+ * Exported APIs
+ */
+
+/* TBD: use pg_arch_1 to optimize this */
+void flush_dcache_page(struct page *page)
+{
+ __dc_line_op((unsigned long)page_address(page), PAGE_SIZE, OP_FLUSH);
+}
+EXPORT_SYMBOL(flush_dcache_page);
+
+
+void dma_cache_wback_inv(unsigned long start, unsigned long sz)
+{
+ __dc_line_op(start, sz, OP_FLUSH_N_INV);
+}
+EXPORT_SYMBOL(dma_cache_wback_inv);
+
+void dma_cache_inv(unsigned long start, unsigned long sz)
+{
+ __dc_line_op(start, sz, OP_INV);
+}
+EXPORT_SYMBOL(dma_cache_inv);
+
+void dma_cache_wback(unsigned long start, unsigned long sz)
+{
+ __dc_line_op(start, sz, OP_FLUSH);
+}
+EXPORT_SYMBOL(dma_cache_wback);
+
+/*
+ * This is API for making I/D Caches consistent when modifying code
+ * (loadable modules, kprobes, etc)
+ * This is called on insmod, with kernel virtual address for CODE of
+ * the module. ARC cache maintenance ops require PHY address thus we
+ * need to convert vmalloc addr to PHY addr
+ */
+void flush_icache_range(unsigned long kstart, unsigned long kend)
+{
+ unsigned int tot_sz, off, sz;
+ unsigned long phy, pfn;
+ unsigned long flags;
+
+ /* printk("Kernel Cache Cohenercy: %lx to %lx\n",kstart, kend); */
+
+ /* This is not the right API for user virtual address */
+ if (kstart < TASK_SIZE) {
+ BUG_ON("Flush icache range for user virtual addr space");
+ return;
+ }
+
+ /* Shortcut for bigger flush ranges.
+ * Here we don't care if this was kernel virtual or phy addr
+ */
+ tot_sz = kend - kstart;
+ if (tot_sz > PAGE_SIZE) {
+ flush_cache_all();
+ return;
+ }
+
+ /* Case: Kernel Phy addr (0x8000_0000 onwards) */
+ if (likely(kstart > PAGE_OFFSET)) {
+ __ic_line_inv(kstart, kend - kstart);
+ __dc_line_op(kstart, kend - kstart, OP_FLUSH);
+ return;
+ }
+
+ /*
+ * Case: Kernel Vaddr (0x7000_0000 to 0x7fff_ffff)
+ * (1) ARC Cache Maintenance ops only take Phy addr, hence special
+ * handling of kernel vaddr.
+ *
+ * (2) Despite @tot_sz being < PAGE_SIZE (bigger cases handled already),
+ * it still needs to handle a 2 page scenario, where the range
+ * straddles across 2 virtual pages and hence need for loop
+ */
+ while (tot_sz > 0) {
+ off = kstart % PAGE_SIZE;
+ pfn = vmalloc_to_pfn((void *)kstart);
+ phy = (pfn << PAGE_SHIFT) + off;
+ sz = min_t(unsigned int, tot_sz, PAGE_SIZE - off);
+ local_irq_save(flags);
+ __dc_line_op(phy, sz, OP_FLUSH);
+ __ic_line_inv(phy, sz);
+ local_irq_restore(flags);
+ kstart += sz;
+ tot_sz -= sz;
+ }
+}
+
+/*
+ * Optimised ver of flush_icache_range() with spec callers: ptrace/signals
+ * where vaddr is also available. This allows passing both vaddr and paddr
+ * bits to CDU for cache flush, short-circuting the current pessimistic algo
+ * which kills all possible aliases.
+ * An added adv of knowing that vaddr is user-vaddr avoids various checks
+ * and handling for k-vaddr, k-paddr as done in orig ver above
+ */
+void flush_icache_range_vaddr(unsigned long paddr, unsigned long u_vaddr,
+ int len)
+{
+ __ic_line_inv_vaddr(paddr, u_vaddr, len);
+ __dc_line_op(paddr, len, OP_FLUSH);
+}
+
+/*
+ * XXX: This also needs to be optim using pg_arch_1
+ * This is called when a page-cache page is about to be mapped into a
+ * user process' address space. It offers an opportunity for a
+ * port to ensure d-cache/i-cache coherency if necessary.
+ */
+void flush_icache_page(struct vm_area_struct *vma, struct page *page)
+{
+ if (!(vma->vm_flags & VM_EXEC))
+ return;
+
+ __ic_line_inv((unsigned long)page_address(page), PAGE_SIZE);
+}
+
+void flush_icache_all(void)
+{
+ unsigned long flags;
+
+ local_irq_save(flags);
+
+ write_aux_reg(ARC_REG_IC_IVIC, 1);
+
+ /* lr will not complete till the icache inv operation is not over */
+ read_aux_reg(ARC_REG_IC_CTRL);
+ local_irq_restore(flags);
+}
+
+noinline void flush_cache_all(void)
+{
+ unsigned long flags;
+
+ local_irq_save(flags);
+
+ flush_icache_all();
+ __dc_entire_op(OP_FLUSH_N_INV);
+
+ local_irq_restore(flags);
+
+}
+
+/* Helper functions for sys_cacheflush */
+
+
+static void __arc_cf_all(uint32_t flags)
+{
+ unsigned long irq_flags;
+
+ local_irq_save(irq_flags);
+
+ switch (flags & CF_D_FLUSH_INV) {
+ case CF_D_FLUSH:
+ __dc_entire_op(OP_FLUSH);
+ break;
+ case CF_D_INV:
+ __dc_entire_op(OP_INV);
+ break;
+ case CF_D_FLUSH_INV:
+ __dc_entire_op(OP_FLUSH_N_INV);
+ break;
+ }
+
+ if (flags & CF_I_INV)
+ flush_icache_all();
+
+ local_irq_restore(irq_flags);
+}
+
+static void __arc_cf_lines(uint32_t phy, uint32_t sz, uint32_t flags)
+{
+ unsigned long irq_flags;
+
+ local_irq_save(irq_flags);
+
+ switch (flags & CF_D_FLUSH_INV) {
+ case CF_D_FLUSH:
+ __dc_line_op(phy, sz, OP_FLUSH);
+ break;
+ case CF_D_INV:
+ __dc_line_op(phy, sz, OP_INV);
+ break;
+ case CF_D_FLUSH_INV:
+ __dc_line_op(phy, sz, OP_FLUSH_N_INV);
+ break;
+ default:
+ break;
+ }
+
+ if (flags & CF_I_INV)
+ __ic_line_inv(phy, sz);
+
+ local_irq_restore(irq_flags);
+}
+
+/**********************************************************************
+ * Explicit Cache flush request from user space via syscall
+ * Needed for JITs which generate code on the fly
+ */
+SYSCALL_DEFINE3(cacheflush, uint32_t, start, uint32_t, sz, uint32_t, flags)
+{
+ struct vm_area_struct *vma;
+ uint32_t end = start + sz;
+
+ if (!flags)
+ flags = CF_DEFAULT;
+
+ /* make sure that the address is valid, in case of virtual address */
+ if ((!(flags & CF_PHY_ADDR)) && access_ok(VERIFY_READ, start, sz))
+ return -EFAULT;
+
+ /* optimization for large areas: flush/inv entire D$ instead */
+ if (sz > PAGE_SIZE) {
+ __arc_cf_all(flags);
+ return 0;
+ }
+
+ if (flags & CF_PHY_ADDR) {
+ __arc_cf_lines(start, sz, flags);
+ return 0;
+ }
+
+ vma = find_vma(current->mm, start);
+ while ((vma) && (vma->vm_start < end)) {
+ uint32_t lstart, lend, laddr;
+ lstart = (vma->vm_start < start) ? start : vma->vm_start;
+ lend = (vma->vm_end > end) ? end : vma->vm_end;
+ lstart &= PAGE_MASK;
+
+ for (laddr = lstart; laddr <= lend; laddr += PAGE_SIZE) {
+ uint32_t pfn, lsz, phy;
+ pte_t *page_table, pte;
+
+ pgd_t *pgd = pgd_offset(current->mm, laddr);
+ pud_t *pud = pud_offset(pgd, laddr);
+ pmd_t *pmd = pmd_offset(pud, laddr);
+
+ /*
+ * FIXME: some times these values are 0, not sure why,
+ * but at least this prevents a crash
+ */
+ if (!pmd)
+ return -EINVAL;
+
+ page_table = pte_offset_kernel(pmd, laddr);
+ if (!page_table)
+ return -EINVAL;
+
+ pte = *page_table;
+ pfn = pte_pfn(pte);
+ phy = pfn << PAGE_SHIFT;
+ lsz = (lend < (laddr + PAGE_SIZE)) ?
+ (lend - laddr) : PAGE_SIZE;
+ if (start > laddr) {
+ phy += start - laddr;
+ lsz -= start - laddr;
+ }
+ __arc_cf_lines(phy, lsz, flags);
+ }
+ start = vma->vm_end;
+ vma = find_vma(current->mm, start);
+ }
+
+ return 0;
+}
--
1.7.4.1
This includes recent changes to make handler "retry" and/or "killable"
The killable (early exit) logic is loosely based on how SH implements it
return if SIGKILL + either of VM_FAULT_OOM or VM_FAULT_RETRY
which is different from Hexagon implementation which would NOT early
exit for
SIGKILL + VM_FAULT_OOM + !VM_FAULT_RETRY
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/mm/extable.c | 26 ++++++
arch/arc/mm/fault.c | 228 +++++++++++++++++++++++++++++++++++++++++++++++++
2 files changed, 254 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/mm/extable.c
create mode 100644 arch/arc/mm/fault.c
diff --git a/arch/arc/mm/extable.c b/arch/arc/mm/extable.c
new file mode 100644
index 0000000..f05b677
--- /dev/null
+++ b/arch/arc/mm/extable.c
@@ -0,0 +1,26 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Borrowed heavily from MIPS
+ */
+
+#include <linux/module.h>
+#include <linux/uaccess.h>
+
+int fixup_exception(struct pt_regs *regs)
+{
+ const struct exception_table_entry *fixup;
+
+ fixup = search_exception_tables(instruction_pointer(regs));
+ if (fixup) {
+ regs->ret = fixup->fixup;
+
+ return 1;
+ }
+
+ return 0;
+}
diff --git a/arch/arc/mm/fault.c b/arch/arc/mm/fault.c
new file mode 100644
index 0000000..af55aab
--- /dev/null
+++ b/arch/arc/mm/fault.c
@@ -0,0 +1,228 @@
+/* Page Fault Handling for ARC (TLB Miss / ProtV)
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/signal.h>
+#include <linux/interrupt.h>
+#include <linux/sched.h>
+#include <linux/errno.h>
+#include <linux/ptrace.h>
+#include <linux/version.h>
+#include <linux/uaccess.h>
+#include <linux/kdebug.h>
+#include <asm/pgalloc.h>
+
+static int handle_vmalloc_fault(struct mm_struct *mm, unsigned long address)
+{
+ /*
+ * Synchronize this task's top level page-table
+ * with the 'reference' page table.
+ */
+ pgd_t *pgd, *pgd_k;
+ pud_t *pud, *pud_k;
+ pmd_t *pmd, *pmd_k;
+
+ pgd = pgd_offset_fast(mm, address);
+ pgd_k = pgd_offset_k(address);
+
+ if (!pgd_present(*pgd_k))
+ goto bad_area;
+
+ pud = pud_offset(pgd, address);
+ pud_k = pud_offset(pgd_k, address);
+ if (!pud_present(*pud_k))
+ goto bad_area;
+
+ pmd = pmd_offset(pud, address);
+ pmd_k = pmd_offset(pud_k, address);
+ if (!pmd_present(*pmd_k))
+ goto bad_area;
+
+ set_pmd(pmd, *pmd_k);
+
+ /* XXX: create the TLB entry here */
+ return 0;
+
+bad_area:
+ return 1;
+}
+
+void do_page_fault(struct pt_regs *regs, int write, unsigned long address,
+ unsigned long cause_code)
+{
+ struct vm_area_struct *vma = NULL;
+ struct task_struct *tsk = current;
+ struct mm_struct *mm = tsk->mm;
+ siginfo_t info;
+ int fault, ret;
+ unsigned int flags = FAULT_FLAG_ALLOW_RETRY | FAULT_FLAG_KILLABLE |
+ (write ? FAULT_FLAG_WRITE : 0);
+
+ /*
+ * We fault-in kernel-space virtual memory on-demand. The
+ * 'reference' page table is init_mm.pgd.
+ *
+ * NOTE! We MUST NOT take any locks for this case. We may
+ * be in an interrupt or a critical region, and should
+ * only copy the information from the master page table,
+ * nothing more.
+ */
+ if (address >= VMALLOC_START && address <= VMALLOC_END) {
+ ret = handle_vmalloc_fault(mm, address);
+ if (unlikely(ret))
+ goto bad_area_nosemaphore;
+ else
+ return;
+ }
+
+ info.si_code = SEGV_MAPERR;
+
+ /*
+ * If we're in an interrupt or have no user
+ * context, we must not take the fault..
+ */
+ if (in_atomic() || !mm)
+ goto no_context;
+
+retry:
+ down_read(&mm->mmap_sem);
+ vma = find_vma(mm, address);
+ if (!vma)
+ goto bad_area;
+ if (vma->vm_start <= address)
+ goto good_area;
+ if (!(vma->vm_flags & VM_GROWSDOWN))
+ goto bad_area;
+ if (expand_stack(vma, address))
+ goto bad_area;
+
+ /*
+ * Ok, we have a good vm_area for this memory access, so
+ * we can handle it..
+ */
+good_area:
+ info.si_code = SEGV_ACCERR;
+
+ /* Handle protection violation, execute on heap or stack */
+
+ if (cause_code == ((ECR_V_PROTV << 16) | ECR_C_PROTV_INST_FETCH))
+ goto bad_area;
+
+ if (write) {
+ if (!(vma->vm_flags & VM_WRITE))
+ goto bad_area;
+ } else {
+ if (!(vma->vm_flags & (VM_READ | VM_EXEC)))
+ goto bad_area;
+ }
+
+survive:
+ /*
+ * If for any reason at all we couldn't handle the fault,
+ * make sure we exit gracefully rather than endlessly redo
+ * the fault.
+ */
+ fault = handle_mm_fault(mm, vma, address, flags);
+
+ /* If Pagefault was interrupted by SIGKILL, exit page fault "early" */
+ if (unlikely(fatal_signal_pending(current))) {
+ if ((fault & VM_FAULT_ERROR) && !(fault & VM_FAULT_RETRY))
+ up_read(&mm->mmap_sem);
+ if (user_mode(regs))
+ return;
+ }
+
+ if (likely(!(fault & VM_FAULT_ERROR))) {
+ if (flags & FAULT_FLAG_ALLOW_RETRY) {
+ /* To avoid updating stats twice for retry case */
+ if (fault & VM_FAULT_MAJOR)
+ tsk->maj_flt++;
+ else
+ tsk->min_flt++;
+
+ if (fault & VM_FAULT_RETRY) {
+ flags &= ~FAULT_FLAG_ALLOW_RETRY;
+ flags |= FAULT_FLAG_TRIED;
+ goto retry;
+ }
+ }
+
+ /* Fault Handled Gracefully */
+ up_read(&mm->mmap_sem);
+ return;
+ }
+
+ /* TBD: switch to pagefault_out_of_memory() */
+ if (fault & VM_FAULT_OOM)
+ goto out_of_memory;
+ else if (fault & VM_FAULT_SIGBUS)
+ goto do_sigbus;
+
+ /* no man's land */
+ BUG();
+
+ /*
+ * Something tried to access memory that isn't in our memory map..
+ * Fix it, but check if it's kernel or user first..
+ */
+bad_area:
+ up_read(&mm->mmap_sem);
+
+bad_area_nosemaphore:
+ /* User mode accesses just cause a SIGSEGV */
+ if (user_mode(regs)) {
+ tsk->thread.fault_address = address;
+ tsk->thread.cause_code = cause_code;
+ info.si_signo = SIGSEGV;
+ info.si_errno = 0;
+ /* info.si_code has been set above */
+ info.si_addr = (void __user *)address;
+ force_sig_info(SIGSEGV, &info, tsk);
+ return;
+ }
+
+no_context:
+ /* Are we prepared to handle this kernel fault?
+ *
+ * (The kernel has valid exception-points in the source
+ * when it acesses user-memory. When it fails in one
+ * of those points, we find it in a table and do a jump
+ * to some fixup code that loads an appropriate error
+ * code)
+ */
+ if (fixup_exception(regs))
+ return;
+
+ die("Oops", regs, address, cause_code);
+
+out_of_memory:
+ if (is_global_init(tsk)) {
+ yield();
+ goto survive;
+ }
+ up_read(&mm->mmap_sem);
+
+ if (user_mode(regs))
+ do_group_exit(SIGKILL); /* This will never return */
+
+ goto no_context;
+
+do_sigbus:
+ up_read(&mm->mmap_sem);
+
+ if (!user_mode(regs))
+ goto no_context;
+
+ tsk->thread.fault_address = address;
+ tsk->thread.cause_code = cause_code;
+ info.si_signo = SIGBUS;
+ info.si_errno = 0;
+ info.si_code = BUS_ADRERR;
+ info.si_addr = (void __user *)address;
+ force_sig_info(SIGBUS, &info, tsk);
+}
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/configs/fpga_defconfig | 607 +++++++++++++++++++++++++++++++++++++++
1 files changed, 607 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/configs/fpga_defconfig
diff --git a/arch/arc/configs/fpga_defconfig b/arch/arc/configs/fpga_defconfig
new file mode 100644
index 0000000..cfdca3c
--- /dev/null
+++ b/arch/arc/configs/fpga_defconfig
@@ -0,0 +1,607 @@
+#
+# Automatically generated file; DO NOT EDIT.
+# Linux/arc 3.7.0-rc3 Kernel Configuration
+#
+CONFIG_ARC=y
+CONFIG_SCHED_OMIT_FRAME_POINTER=y
+CONFIG_GENERIC_CSUM=y
+CONFIG_RWSEM_GENERIC_SPINLOCK=y
+CONFIG_UID16=y
+CONFIG_ARCH_FLATMEM_ENABLE=y
+CONFIG_MMU=y
+CONFIG_NO_IOPORT=y
+CONFIG_GENERIC_CALIBRATE_DELAY=y
+CONFIG_GENERIC_HWEIGHT=y
+CONFIG_BINFMT_ELF=y
+CONFIG_STACKTRACE_SUPPORT=y
+CONFIG_HAVE_LATENCYTOP_SUPPORT=y
+# CONFIG_NO_DMA is not set
+CONFIG_DEFCONFIG_LIST="/lib/modules/$UNAME_RELEASE/.config"
+
+#
+# General setup
+#
+# CONFIG_EXPERIMENTAL is not set
+CONFIG_BROKEN_ON_SMP=y
+CONFIG_INIT_ENV_ARG_LIMIT=32
+CONFIG_CROSS_COMPILE="arc-elf32-"
+CONFIG_LOCALVERSION=""
+# CONFIG_LOCALVERSION_AUTO is not set
+CONFIG_DEFAULT_HOSTNAME="(none)"
+# CONFIG_SWAP is not set
+# CONFIG_SYSVIPC is not set
+# CONFIG_FHANDLE is not set
+CONFIG_HAVE_GENERIC_HARDIRQS=y
+
+#
+# IRQ subsystem
+#
+CONFIG_GENERIC_HARDIRQS=y
+CONFIG_GENERIC_IRQ_SHOW=y
+CONFIG_GENERIC_CLOCKEVENTS=y
+CONFIG_GENERIC_CLOCKEVENTS_BUILD=y
+
+#
+# Timers subsystem
+#
+CONFIG_TICK_ONESHOT=y
+# CONFIG_NO_HZ is not set
+CONFIG_HIGH_RES_TIMERS=y
+
+#
+# CPU/Task time and stats accounting
+#
+CONFIG_TICK_CPU_ACCOUNTING=y
+# CONFIG_BSD_PROCESS_ACCT is not set
+
+#
+# RCU Subsystem
+#
+CONFIG_TINY_RCU=y
+# CONFIG_PREEMPT_RCU is not set
+# CONFIG_TREE_RCU_TRACE is not set
+# CONFIG_IKCONFIG is not set
+CONFIG_LOG_BUF_SHIFT=17
+# CONFIG_CHECKPOINT_RESTORE is not set
+CONFIG_NAMESPACES=y
+# CONFIG_UTS_NS is not set
+# CONFIG_PID_NS is not set
+CONFIG_UIDGID_CONVERTED=y
+# CONFIG_UIDGID_STRICT_TYPE_CHECKS is not set
+# CONFIG_SCHED_AUTOGROUP is not set
+# CONFIG_SYSFS_DEPRECATED is not set
+# CONFIG_RELAY is not set
+CONFIG_BLK_DEV_INITRD=y
+CONFIG_INITRAMFS_SOURCE="../arc_initramfs"
+CONFIG_INITRAMFS_ROOT_UID=0
+CONFIG_INITRAMFS_ROOT_GID=0
+CONFIG_RD_GZIP=y
+# CONFIG_RD_BZIP2 is not set
+# CONFIG_RD_LZMA is not set
+# CONFIG_RD_XZ is not set
+# CONFIG_RD_LZO is not set
+CONFIG_INITRAMFS_COMPRESSION_NONE=y
+# CONFIG_INITRAMFS_COMPRESSION_GZIP is not set
+# CONFIG_CC_OPTIMIZE_FOR_SIZE is not set
+CONFIG_SYSCTL=y
+CONFIG_EXPERT=y
+# CONFIG_SYSCTL_SYSCALL is not set
+CONFIG_KALLSYMS=y
+# CONFIG_KALLSYMS_ALL is not set
+CONFIG_HOTPLUG=y
+CONFIG_PRINTK=y
+CONFIG_BUG=y
+# CONFIG_BASE_FULL is not set
+# CONFIG_FUTEX is not set
+# CONFIG_EPOLL is not set
+# CONFIG_SIGNALFD is not set
+# CONFIG_TIMERFD is not set
+# CONFIG_EVENTFD is not set
+# CONFIG_SHMEM is not set
+# CONFIG_AIO is not set
+CONFIG_EMBEDDED=y
+
+#
+# Kernel Performance Events And Counters
+#
+# CONFIG_VM_EVENT_COUNTERS is not set
+# CONFIG_SLUB_DEBUG is not set
+# CONFIG_COMPAT_BRK is not set
+# CONFIG_SLAB is not set
+CONFIG_SLUB=y
+# CONFIG_SLOB is not set
+# CONFIG_PROFILING is not set
+CONFIG_GENERIC_SMP_IDLE_THREAD=y
+CONFIG_ARCH_WANT_IPC_PARSE_VERSION=y
+CONFIG_GENERIC_KERNEL_THREAD=y
+CONFIG_MODULES_USE_ELF_RELA=y
+
+#
+# GCOV-based kernel profiling
+#
+# CONFIG_HAVE_GENERIC_DMA_COHERENT is not set
+CONFIG_BASE_SMALL=1
+# CONFIG_MODULES is not set
+CONFIG_BLOCK=y
+# CONFIG_LBDAF is not set
+# CONFIG_BLK_DEV_BSG is not set
+# CONFIG_BLK_DEV_BSGLIB is not set
+# CONFIG_BLK_DEV_INTEGRITY is not set
+
+#
+# Partition Types
+#
+# CONFIG_PARTITION_ADVANCED is not set
+CONFIG_MSDOS_PARTITION=y
+
+#
+# IO Schedulers
+#
+CONFIG_IOSCHED_NOOP=y
+# CONFIG_IOSCHED_DEADLINE is not set
+# CONFIG_IOSCHED_CFQ is not set
+CONFIG_DEFAULT_NOOP=y
+CONFIG_DEFAULT_IOSCHED="noop"
+CONFIG_INLINE_SPIN_UNLOCK_IRQ=y
+CONFIG_INLINE_READ_UNLOCK=y
+CONFIG_INLINE_READ_UNLOCK_IRQ=y
+CONFIG_INLINE_WRITE_UNLOCK=y
+CONFIG_INLINE_WRITE_UNLOCK_IRQ=y
+# CONFIG_FREEZER is not set
+
+#
+# ARC Architecture Configuration
+#
+CONFIG_ARC_PLAT_FPGA_LEGACY=y
+
+#
+# ARC CPU Configuration
+#
+# CONFIG_ARC_CPU_750D is not set
+CONFIG_ARC_CPU_770=y
+# CONFIG_CPU_BIG_ENDIAN is not set
+CONFIG_ARC_CACHE=y
+CONFIG_ARC_CACHE_LINE_SHIFT=6
+CONFIG_ARC_HAS_ICACHE=y
+CONFIG_ARC_HAS_DCACHE=y
+CONFIG_ARC_CACHE_PAGES=y
+CONFIG_ARC_HAS_HW_MPY=y
+# CONFIG_ARC_MMU_V1 is not set
+# CONFIG_ARC_MMU_V2 is not set
+CONFIG_ARC_MMU_V3=y
+CONFIG_ARC_PAGE_SIZE_8K=y
+# CONFIG_ARC_PAGE_SIZE_16K is not set
+# CONFIG_ARC_PAGE_SIZE_4K is not set
+# CONFIG_ARC_FPU_SAVE_RESTORE is not set
+CONFIG_ARC_CPU_REL_4_10=y
+CONFIG_ARC_HAS_LLSC=y
+CONFIG_ARC_HAS_SWAPE=y
+CONFIG_ARC_HAS_RTSC=y
+
+#
+# Platform Board Configuration
+#
+CONFIG_ARC_BOARD_ANGEL4=y
+# CONFIG_ARC_BOARD_ML509 is not set
+CONFIG_ARC_SERIAL_BAUD=115200
+CONFIG_ARC_PLAT_CLK=80000000
+CONFIG_LINUX_LINK_BASE=0x80000000
+CONFIG_ARC_PLAT_SDRAM_SIZE=0x10000000
+# CONFIG_ARC_STACK_NONEXEC is not set
+CONFIG_HZ=100
+# CONFIG_ARC_DBG is not set
+CONFIG_CMDLINE="print-fatal-signals=1"
+# CONFIG_CMDLINE_UBOOT is not set
+CONFIG_PREEMPT_NONE=y
+# CONFIG_PREEMPT_VOLUNTARY is not set
+# CONFIG_PREEMPT is not set
+CONFIG_FLATMEM=y
+CONFIG_FLAT_NODE_MEM_MAP=y
+CONFIG_PAGEFLAGS_EXTENDED=y
+CONFIG_SPLIT_PTLOCK_CPUS=4
+# CONFIG_COMPACTION is not set
+# CONFIG_PHYS_ADDR_T_64BIT is not set
+CONFIG_ZONE_DMA_FLAG=0
+CONFIG_VIRT_TO_BUS=y
+# CONFIG_KSM is not set
+CONFIG_DEFAULT_MMAP_MIN_ADDR=4096
+# CONFIG_CROSS_MEMORY_ATTACH is not set
+CONFIG_NEED_PER_CPU_KM=y
+# CONFIG_CLEANCACHE is not set
+# CONFIG_NET is not set
+
+#
+# Device Drivers
+#
+
+#
+# Generic Driver Options
+#
+CONFIG_UEVENT_HELPER_PATH=""
+CONFIG_DEVTMPFS=y
+CONFIG_DEVTMPFS_MOUNT=y
+CONFIG_STANDALONE=y
+# CONFIG_PREVENT_FIRMWARE_BUILD is not set
+CONFIG_FW_LOADER=y
+# CONFIG_FIRMWARE_IN_KERNEL is not set
+CONFIG_EXTRA_FIRMWARE=""
+# CONFIG_DEBUG_DRIVER is not set
+# CONFIG_DEBUG_DEVRES is not set
+# CONFIG_SYS_HYPERVISOR is not set
+# CONFIG_GENERIC_CPU_DEVICES is not set
+
+#
+# Bus devices
+#
+# CONFIG_OMAP_OCP2SCP is not set
+# CONFIG_MTD is not set
+# CONFIG_PARPORT is not set
+# CONFIG_BLK_DEV is not set
+
+#
+# Misc devices
+#
+# CONFIG_SENSORS_LIS3LV02D is not set
+# CONFIG_ENCLOSURE_SERVICES is not set
+# CONFIG_C2PORT is not set
+
+#
+# EEPROM support
+#
+# CONFIG_EEPROM_93CX6 is not set
+
+#
+# Texas Instruments shared transport line discipline
+#
+
+#
+# Altera FPGA firmware download module
+#
+
+#
+# SCSI device support
+#
+CONFIG_SCSI_MOD=y
+# CONFIG_RAID_ATTRS is not set
+# CONFIG_SCSI is not set
+# CONFIG_SCSI_DMA is not set
+# CONFIG_SCSI_NETLINK is not set
+# CONFIG_ATA is not set
+# CONFIG_MD is not set
+
+#
+# Input device support
+#
+CONFIG_INPUT=y
+# CONFIG_INPUT_FF_MEMLESS is not set
+# CONFIG_INPUT_POLLDEV is not set
+# CONFIG_INPUT_SPARSEKMAP is not set
+# CONFIG_INPUT_MATRIXKMAP is not set
+
+#
+# Userland interfaces
+#
+# CONFIG_INPUT_MOUSEDEV is not set
+# CONFIG_INPUT_JOYDEV is not set
+# CONFIG_INPUT_EVDEV is not set
+# CONFIG_INPUT_EVBUG is not set
+
+#
+# Input Device Drivers
+#
+# CONFIG_INPUT_KEYBOARD is not set
+# CONFIG_INPUT_MOUSE is not set
+# CONFIG_INPUT_JOYSTICK is not set
+# CONFIG_INPUT_TABLET is not set
+# CONFIG_INPUT_TOUCHSCREEN is not set
+# CONFIG_INPUT_MISC is not set
+
+#
+# Hardware I/O ports
+#
+# CONFIG_SERIO is not set
+# CONFIG_GAMEPORT is not set
+
+#
+# Character devices
+#
+CONFIG_VT=y
+CONFIG_CONSOLE_TRANSLATIONS=y
+CONFIG_VT_CONSOLE=y
+CONFIG_HW_CONSOLE=y
+# CONFIG_VT_HW_CONSOLE_BINDING is not set
+CONFIG_UNIX98_PTYS=y
+# CONFIG_DEVPTS_MULTIPLE_INSTANCES is not set
+# CONFIG_LEGACY_PTYS is not set
+# CONFIG_SERIAL_NONSTANDARD is not set
+# CONFIG_TRACE_SINK is not set
+# CONFIG_DEVKMEM is not set
+
+#
+# Serial drivers
+#
+# CONFIG_SERIAL_8250 is not set
+
+#
+# Non-8250 serial port support
+#
+CONFIG_SERIAL_CORE=y
+CONFIG_SERIAL_CORE_CONSOLE=y
+# CONFIG_SERIAL_SCCNXP is not set
+# CONFIG_SERIAL_TIMBERDALE is not set
+# CONFIG_SERIAL_ALTERA_JTAGUART is not set
+# CONFIG_SERIAL_ALTERA_UART is not set
+# CONFIG_SERIAL_XILINX_PS_UART is not set
+CONFIG_SERIAL_ARC=y
+CONFIG_SERIAL_ARC_CONSOLE=y
+CONFIG_SERIAL_ARC_NR_PORTS=1
+# CONFIG_TTY_PRINTK is not set
+# CONFIG_IPMI_HANDLER is not set
+# CONFIG_HW_RANDOM is not set
+# CONFIG_RTC is not set
+# CONFIG_GEN_RTC is not set
+# CONFIG_R3964 is not set
+
+#
+# PCMCIA character devices
+#
+# CONFIG_RAW_DRIVER is not set
+# CONFIG_TCG_TPM is not set
+# CONFIG_I2C is not set
+# CONFIG_SPI is not set
+# CONFIG_HSI is not set
+
+#
+# PPS support
+#
+
+#
+# PPS generators support
+#
+
+#
+# PTP clock support
+#
+
+#
+# Enable Device Drivers -> PPS to see the PTP clock options.
+#
+# CONFIG_W1 is not set
+# CONFIG_POWER_SUPPLY is not set
+# CONFIG_POWER_AVS is not set
+# CONFIG_HWMON is not set
+# CONFIG_THERMAL is not set
+# CONFIG_WATCHDOG is not set
+CONFIG_SSB_POSSIBLE=y
+
+#
+# Sonics Silicon Backplane
+#
+# CONFIG_SSB is not set
+CONFIG_BCMA_POSSIBLE=y
+
+#
+# Broadcom specific AMBA
+#
+# CONFIG_BCMA is not set
+
+#
+# Multifunction device drivers
+#
+# CONFIG_MFD_CORE is not set
+# CONFIG_MFD_SM501 is not set
+# CONFIG_HTC_PASIC3 is not set
+# CONFIG_MFD_TMIO is not set
+# CONFIG_ABX500_CORE is not set
+# CONFIG_REGULATOR is not set
+# CONFIG_MEDIA_SUPPORT is not set
+
+#
+# Graphics support
+#
+# CONFIG_DRM is not set
+# CONFIG_VGASTATE is not set
+# CONFIG_VIDEO_OUTPUT_CONTROL is not set
+# CONFIG_FB is not set
+# CONFIG_EXYNOS_VIDEO is not set
+# CONFIG_BACKLIGHT_LCD_SUPPORT is not set
+
+#
+# Console display driver support
+#
+# CONFIG_VGA_CONSOLE is not set
+CONFIG_DUMMY_CONSOLE=y
+# CONFIG_SOUND is not set
+
+#
+# HID support
+#
+# CONFIG_HID is not set
+# CONFIG_USB_ARCH_HAS_OHCI is not set
+# CONFIG_USB_ARCH_HAS_EHCI is not set
+# CONFIG_USB_ARCH_HAS_XHCI is not set
+# CONFIG_USB_SUPPORT is not set
+# CONFIG_MMC is not set
+# CONFIG_MEMSTICK is not set
+# CONFIG_NEW_LEDS is not set
+# CONFIG_ACCESSIBILITY is not set
+# CONFIG_RTC_CLASS is not set
+# CONFIG_DMADEVICES is not set
+# CONFIG_AUXDISPLAY is not set
+# CONFIG_UIO is not set
+
+#
+# Virtio drivers
+#
+
+#
+# Microsoft Hyper-V guest support
+#
+# CONFIG_STAGING is not set
+
+#
+# Hardware Spinlock drivers
+#
+# CONFIG_IOMMU_SUPPORT is not set
+
+#
+# Remoteproc drivers (EXPERIMENTAL)
+#
+
+#
+# Rpmsg drivers (EXPERIMENTAL)
+#
+# CONFIG_VIRT_DRIVERS is not set
+# CONFIG_PM_DEVFREQ is not set
+# CONFIG_EXTCON is not set
+# CONFIG_MEMORY is not set
+# CONFIG_IIO is not set
+# CONFIG_PWM is not set
+
+#
+# File systems
+#
+# CONFIG_EXT2_FS is not set
+# CONFIG_EXT3_FS is not set
+# CONFIG_EXT4_FS is not set
+# CONFIG_REISERFS_FS is not set
+# CONFIG_JFS_FS is not set
+# CONFIG_XFS_FS is not set
+# CONFIG_FS_POSIX_ACL is not set
+CONFIG_FILE_LOCKING=y
+# CONFIG_FSNOTIFY is not set
+# CONFIG_DNOTIFY is not set
+# CONFIG_INOTIFY_USER is not set
+# CONFIG_FANOTIFY is not set
+# CONFIG_QUOTA is not set
+# CONFIG_QUOTACTL is not set
+# CONFIG_AUTOFS4_FS is not set
+# CONFIG_FUSE_FS is not set
+
+#
+# Caches
+#
+# CONFIG_FSCACHE is not set
+
+#
+# CD-ROM/DVD Filesystems
+#
+# CONFIG_ISO9660_FS is not set
+# CONFIG_UDF_FS is not set
+
+#
+# DOS/FAT/NT Filesystems
+#
+# CONFIG_MSDOS_FS is not set
+# CONFIG_VFAT_FS is not set
+# CONFIG_NTFS_FS is not set
+
+#
+# Pseudo filesystems
+#
+CONFIG_PROC_FS=y
+# CONFIG_PROC_KCORE is not set
+CONFIG_PROC_SYSCTL=y
+CONFIG_PROC_PAGE_MONITOR=y
+CONFIG_SYSFS=y
+# CONFIG_HUGETLB_PAGE is not set
+# CONFIG_CONFIGFS_FS is not set
+# CONFIG_MISC_FILESYSTEMS is not set
+# CONFIG_NLS is not set
+
+#
+# Kernel hacking
+#
+# CONFIG_PRINTK_TIME is not set
+CONFIG_DEFAULT_MESSAGE_LOGLEVEL=4
+CONFIG_ENABLE_WARN_DEPRECATED=y
+CONFIG_ENABLE_MUST_CHECK=y
+CONFIG_FRAME_WARN=1024
+# CONFIG_MAGIC_SYSRQ is not set
+# CONFIG_STRIP_ASM_SYMS is not set
+# CONFIG_READABLE_ASM is not set
+# CONFIG_UNUSED_SYMBOLS is not set
+# CONFIG_DEBUG_FS is not set
+# CONFIG_HEADERS_CHECK is not set
+# CONFIG_DEBUG_SECTION_MISMATCH is not set
+CONFIG_DEBUG_KERNEL=y
+# CONFIG_DEBUG_SHIRQ is not set
+# CONFIG_LOCKUP_DETECTOR is not set
+# CONFIG_PANIC_ON_OOPS is not set
+CONFIG_PANIC_ON_OOPS_VALUE=0
+# CONFIG_DETECT_HUNG_TASK is not set
+CONFIG_SCHED_DEBUG=y
+# CONFIG_SCHEDSTATS is not set
+# CONFIG_TIMER_STATS is not set
+# CONFIG_DEBUG_OBJECTS is not set
+# CONFIG_SLUB_STATS is not set
+# CONFIG_DEBUG_SPINLOCK is not set
+# CONFIG_DEBUG_MUTEXES is not set
+# CONFIG_SPARSE_RCU_POINTER is not set
+# CONFIG_DEBUG_ATOMIC_SLEEP is not set
+# CONFIG_DEBUG_LOCKING_API_SELFTESTS is not set
+CONFIG_STACKTRACE=y
+# CONFIG_DEBUG_STACK_USAGE is not set
+# CONFIG_DEBUG_KOBJECT is not set
+# CONFIG_DEBUG_INFO is not set
+# CONFIG_DEBUG_VM is not set
+# CONFIG_DEBUG_WRITECOUNT is not set
+# CONFIG_DEBUG_MEMORY_INIT is not set
+# CONFIG_DEBUG_LIST is not set
+# CONFIG_TEST_LIST_SORT is not set
+# CONFIG_DEBUG_SG is not set
+# CONFIG_DEBUG_NOTIFIERS is not set
+# CONFIG_DEBUG_CREDENTIALS is not set
+# CONFIG_BOOT_PRINTK_DELAY is not set
+# CONFIG_RCU_TORTURE_TEST is not set
+# CONFIG_RCU_TRACE is not set
+# CONFIG_BACKTRACE_SELF_TEST is not set
+# CONFIG_DEBUG_BLOCK_EXT_DEVT is not set
+# CONFIG_DEBUG_FORCE_WEAK_PER_CPU is not set
+# CONFIG_NOTIFIER_ERROR_INJECTION is not set
+# CONFIG_FAULT_INJECTION is not set
+# CONFIG_LATENCYTOP is not set
+# CONFIG_DEBUG_PAGEALLOC is not set
+# CONFIG_ATOMIC64_SELFTEST is not set
+# CONFIG_SAMPLES is not set
+# CONFIG_TEST_KSTRTOX is not set
+CONFIG_EARLY_PRINTK=y
+# CONFIG_DEBUG_STACKOVERFLOW is not set
+# CONFIG_16KSTACKS is not set
+
+#
+# Security options
+#
+# CONFIG_KEYS is not set
+# CONFIG_SECURITY_DMESG_RESTRICT is not set
+# CONFIG_SECURITY is not set
+# CONFIG_SECURITYFS is not set
+CONFIG_DEFAULT_SECURITY_DAC=y
+CONFIG_DEFAULT_SECURITY=""
+# CONFIG_CRYPTO is not set
+# CONFIG_BINARY_PRINTF is not set
+
+#
+# Library routines
+#
+CONFIG_GENERIC_FIND_FIRST_BIT=y
+CONFIG_GENERIC_IO=y
+# CONFIG_CRC_CCITT is not set
+# CONFIG_CRC16 is not set
+# CONFIG_CRC_T10DIF is not set
+# CONFIG_CRC_ITU_T is not set
+# CONFIG_CRC32 is not set
+# CONFIG_CRC7 is not set
+# CONFIG_LIBCRC32C is not set
+# CONFIG_CRC8 is not set
+CONFIG_ZLIB_INFLATE=y
+# CONFIG_XZ_DEC is not set
+# CONFIG_XZ_DEC_BCJ is not set
+CONFIG_DECOMPRESS_GZIP=y
+CONFIG_HAS_IOMEM=y
+CONFIG_HAS_DMA=y
+CONFIG_GENERIC_ATOMIC64=y
+CONFIG_ARCH_HAS_ATOMIC64_DEC_IF_POSITIVE=y
+# CONFIG_AVERAGE is not set
+# CONFIG_CORDIC is not set
+# CONFIG_DDR is not set
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/Kconfig | 1 +
arch/arc/kernel/process.c | 29 +++++++++--------------------
2 files changed, 10 insertions(+), 20 deletions(-)
diff --git a/arch/arc/Kconfig b/arch/arc/Kconfig
index d6983afa..f3aacfc 100644
--- a/arch/arc/Kconfig
+++ b/arch/arc/Kconfig
@@ -17,6 +17,7 @@ config ARC
select GENERIC_FIND_FIRST_BIT
# for now, we don't need GENERIC_IRQ_PROBE, CONFIG_GENERIC_IRQ_CHIP
select GENERIC_IRQ_SHOW
+ select GENERIC_KERNEL_THREAD
select GENERIC_PENDING_IRQ if SMP
select GENERIC_SMP_IDLE_THREAD
select HAVE_GENERIC_HARDIRQS
diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c
index fee3e11..d2bb27f 100644
--- a/arch/arc/kernel/process.c
+++ b/arch/arc/kernel/process.c
@@ -142,22 +142,6 @@ void cpu_idle(void)
}
}
-int kernel_thread(int (*fn) (void *), void *arg, unsigned long flags)
-{
- struct pt_regs regs = {
- .r0 = (unsigned long)arg,
- .r1 = (unsigned long)fn
- };
-
- regs.status32 = read_aux_reg(0xa);
-
- /* Ok, create the new process.. */
- return do_fork(flags | CLONE_VM | CLONE_UNTRACED, 0, ®s, 0, NULL,
- NULL);
-
-}
-EXPORT_SYMBOL(kernel_thread);
-
asmlinkage void ret_from_fork(void);
asmlinkage void ret_from_kernel_thread(void) __attribute__((noreturn));
@@ -192,7 +176,7 @@ asmlinkage void ret_from_kernel_thread(void) __attribute__((noreturn));
* ------------------ <===== END of PAGE
*/
int copy_thread(unsigned long clone_flags,
- unsigned long usp, unsigned long topstk,
+ unsigned long usp, unsigned long arg,
struct task_struct *p, struct pt_regs *regs)
{
struct pt_regs *c_regs; /* child's pt_regs */
@@ -213,10 +197,12 @@ int copy_thread(unsigned long clone_flags,
*/
p->thread.ksp = (unsigned long)c_callee; /* THREAD_KSP */
- /* Copy parents pt regs on child's kernel mode stack */
- *c_regs = *regs;
+ if (unlikely(p->flags & PF_KTHREAD)) {
+ memset(c_regs, 0, sizeof(struct pt_regs));
+ c_regs->r0 = arg; /* argument to kernel thread */
+ c_regs->r1 = usp; /* function */
+ c_regs->status32 = read_aux_reg(0xa);
- if (!(user_mode(regs))) {
c_regs->sp =
(unsigned long)task_thread_info(p) + (THREAD_SIZE - 4);
@@ -232,6 +218,9 @@ int copy_thread(unsigned long clone_flags,
childksp[0] = 0; /* for POP fp */
childksp[1] = (unsigned long)ret_from_fork; /* for POP blink */
+ /* Copy parents pt regs on child's kernel mode stack */
+ *c_regs = *regs;
+
c_regs->sp = usp;
c_regs->r0 = 0; /* fork returns 0 in child */
--
1.7.4.1
=========================>8=============================
Linux version 3.7.0-rc3+ (vineetg@vineetg-Latitude) (gcc version 4.4.7
(ARCompact elf32 toolchain (built 20120928)) ) #5 Tue Nov 6 17:05:37 CET
2012
[plat-arcfpga]: registering early dev resources
bootconsole [early_ARCuart0] enabled
pcpu-alloc: s0 r0 d32768 u32768 alloc=1*32768
pcpu-alloc: [0] 0
Built 1 zonelists in Zone order, mobility grouping on. Total pages:
32624
Kernel command line: print-fatal-signals=1
PID hash table entries: 1024 (order: -1, 4096 bytes)
Dentry cache hash table entries: 32768 (order: 4, 131072 bytes)
Inode-cache hash table entries: 16384 (order: 3, 65536 bytes)
Memory Available: 248M / 256M (1219K code, 412K data, 4192K init, 1400K
reserv)
SLUB: Genslabs=12, HWalign=64, Order=0-3, MinObjects=0, CPUs=1, Nodes=1
NR_IRQS:16
Device [ARC Timer0] clockevent mode now [1]
Device [ARC Timer0] clockevent mode now [2]
Console: colour dummy device 80x25
Calibrating delay loop... 39.73 BogoMIPS (lpj=198656)
pid_max: default: 4096 minimum: 301
Mount-cache hash table entries: 1024
devtmpfs: initialized
[plat-arcfpga]: registering device resources
bio: create slab <bio-0> at 0
Switching to clocksource ARC Timer1
Device [ARC Timer0] clockevent mode now [3]
io scheduler noop registered (default)
arc-uart.0: ttyARC0 at MMIO 0xc0fc1000 (irq = 5) is a arc-uart
console [ttyARC0] enabled, bootconsole disabled
console [ttyARC0] enabled, bootconsole disabled
Warning: unable to open an initial console.
Freeing unused kernel memory: 4192k [80002000] to [8041a000]
Mounting proc
Mounting sysfs
Mounting devpts
Setting hostname to ARCLinux
Starting System logger (syslogd)
Bringing up loopback device
ifconfig: socket: Function not implemented
route: socket: Function not implemented
Disk not detected !
Mounting tmpfs
/etc/init.d/rcS: line 76: can't create /proc/sys/kernel/msgmni:
nonexistent directory
Please press Enter to activate this console.
***********************************************************************
Welcome to ARCLinux
***********************************************************************
[ARCLinux]$
=========================>8=============================
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/kernel/ptrace.c | 26 ++++++++++++++++++++++++
arch/arc/kernel/stacktrace.c | 43 ++++++++++++++++++++++++++++++++++++++++
arch/arc/kernel/troubleshoot.c | 17 +++++++++++++++
3 files changed, 86 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/kernel/ptrace.c
create mode 100644 arch/arc/kernel/stacktrace.c
create mode 100644 arch/arc/kernel/troubleshoot.c
diff --git a/arch/arc/kernel/ptrace.c b/arch/arc/kernel/ptrace.c
new file mode 100644
index 0000000..1cf944a
--- /dev/null
+++ b/arch/arc/kernel/ptrace.c
@@ -0,0 +1,26 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/ptrace.h>
+
+void ptrace_disable(struct task_struct *child)
+{
+}
+
+long arch_ptrace(struct task_struct *child, long request,
+ unsigned long addr, unsigned long data)
+{
+ int ret = -EIO;
+ return ret;
+}
+
+
+const struct user_regset_view *task_user_regset_view(struct task_struct *task)
+{
+ return (const struct user_regset_view *)NULL;
+}
diff --git a/arch/arc/kernel/stacktrace.c b/arch/arc/kernel/stacktrace.c
new file mode 100644
index 0000000..b9d1646
--- /dev/null
+++ b/arch/arc/kernel/stacktrace.c
@@ -0,0 +1,43 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/ptrace.h>
+#include <linux/export.h>
+
+/*-------------------------------------------------------------------------
+ * APIs expected by various kernel sub-systems
+ *-------------------------------------------------------------------------
+ */
+
+noinline void show_stacktrace(struct task_struct *tsk, struct pt_regs *regs)
+{
+ pr_info("\nStack Trace: NOT Available\n");
+}
+EXPORT_SYMBOL(show_stacktrace);
+
+/* Expected by sched Code */
+void show_stack(struct task_struct *tsk, unsigned long *sp)
+{
+ show_stacktrace(tsk, NULL);
+}
+
+/* Expected by Rest of kernel code */
+void dump_stack(void)
+{
+ show_stacktrace(NULL, NULL);
+}
+EXPORT_SYMBOL(dump_stack);
+
+/* Another API expected by schedular, shows up in "ps" as Wait Channel
+ * Ofcourse just returning schedule( ) would be pointless so unwind until
+ * the function is not in schedular code
+ */
+unsigned int get_wchan(struct task_struct *tsk)
+{
+ return 0;
+}
diff --git a/arch/arc/kernel/troubleshoot.c b/arch/arc/kernel/troubleshoot.c
new file mode 100644
index 0000000..80bfe2a
--- /dev/null
+++ b/arch/arc/kernel/troubleshoot.c
@@ -0,0 +1,17 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ */
+
+#include <linux/ptrace.h>
+
+void show_regs(struct pt_regs *regs)
+{
+}
+
+void show_kernel_fault_diag(const char *str, struct pt_regs *regs,
+ unsigned long address, unsigned long cause_reg)
+{
+}
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/irq.h | 1 +
arch/arc/plat-arcfpga/include/plat/irq.h | 27 +++++++
arch/arc/plat-arcfpga/include/plat/memmap.h | 31 ++++++++
arch/arc/plat-arcfpga/platform.c | 105 +++++++++++++++++++++++++++
4 files changed, 164 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/plat-arcfpga/include/plat/irq.h
create mode 100644 arch/arc/plat-arcfpga/include/plat/memmap.h
diff --git a/arch/arc/include/asm/irq.h b/arch/arc/include/asm/irq.h
index 633416c..8e771f6 100644
--- a/arch/arc/include/asm/irq.h
+++ b/arch/arc/include/asm/irq.h
@@ -13,6 +13,7 @@
#define TIMER0_IRQ 3
#define TIMER1_IRQ 4
+#include <plat/irq.h> /* Board Specific IRQ assignments */
#include <asm-generic/irq.h>
extern void __init arc_init_IRQ(void);
diff --git a/arch/arc/plat-arcfpga/include/plat/irq.h b/arch/arc/plat-arcfpga/include/plat/irq.h
new file mode 100644
index 0000000..b34e087
--- /dev/null
+++ b/arch/arc/plat-arcfpga/include/plat/irq.h
@@ -0,0 +1,27 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: Feb 2009
+ * -For AA4 board, IRQ assignments to peripherals
+ */
+
+#ifndef __PLAT_IRQ_H
+#define __PLAT_IRQ_H
+
+#define NR_IRQS 16
+
+#define UART0_IRQ 5
+#define UART1_IRQ 10
+#define UART2_IRQ 11
+
+#define VMAC_IRQ 6
+
+#define IDE_IRQ 13
+#define PCI_IRQ 14
+#define PS2_IRQ 15
+
+#endif
diff --git a/arch/arc/plat-arcfpga/include/plat/memmap.h b/arch/arc/plat-arcfpga/include/plat/memmap.h
new file mode 100644
index 0000000..1663f33
--- /dev/null
+++ b/arch/arc/plat-arcfpga/include/plat/memmap.h
@@ -0,0 +1,31 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: Feb 2009
+ * -For AA4 board, System Memory Map for Peripherals etc
+ */
+
+#ifndef __PLAT_MEMMAP_H
+#define __PLAT_MEMMAP_H
+
+#define UART0_BASE 0xC0FC1000
+#define UART1_BASE 0xC0FC1100
+
+#define VMAC_REG_BASEADDR 0xC0FC2000
+
+#define IDE_CONTROLLER_BASE 0xC0FC9000
+
+#define AHB_PCI_HOST_BRG_BASE 0xC0FD0000
+
+#define PGU_BASEADDR 0xC0FC8000
+#define VLCK_ADDR 0xC0FCF028
+
+#define BVCI_LAT_UNIT_BASE 0xC0FED000
+
+#define PS2_BASE_ADDR 0xC0FCC000
+
+#endif
diff --git a/arch/arc/plat-arcfpga/platform.c b/arch/arc/plat-arcfpga/platform.c
index 7b5dcab..3243091 100644
--- a/arch/arc/plat-arcfpga/platform.c
+++ b/arch/arc/plat-arcfpga/platform.c
@@ -10,7 +10,99 @@
#include <linux/types.h>
#include <linux/init.h>
+#include <linux/device.h>
#include <linux/platform_device.h>
+#include <linux/console.h>
+#include <asm/setup.h>
+#include <asm/irq.h>
+#include <plat/memmap.h>
+
+/*----------------------- Platform Devices -----------------------------*/
+
+#if defined(CONFIG_SERIAL_ARC) || defined(CONFIG_SERIAL_ARC_MODULE)
+
+static unsigned long arc_uart_info[] = {
+ CONFIG_ARC_SERIAL_BAUD, /* uart->baud */
+ CONFIG_ARC_PLAT_CLK, /* uart->port.uartclk */
+ -1, /* uart->is_emulated (runtime @running_on_hw) */
+ 0
+};
+
+#define ARC_UART_DEV(n) \
+ \
+static struct resource arc_uart##n##_res[] = { \
+ { \
+ .start = UART##n##_BASE, \
+ .end = UART##n##_BASE + 0xFF, \
+ .flags = IORESOURCE_MEM, \
+ }, \
+ { \
+ .start = UART##n##_IRQ, \
+ .end = UART##n##_IRQ, \
+ .flags = IORESOURCE_IRQ, \
+ }, \
+}; \
+ \
+static struct platform_device arc_uart##n##_dev = { \
+ .name = "arc-uart", \
+ .id = n, \
+ .num_resources = ARRAY_SIZE(arc_uart##n##_res), \
+ .resource = arc_uart##n##_res, \
+ .dev = { \
+ .platform_data = &arc_uart_info, \
+ }, \
+}
+
+ARC_UART_DEV(0);
+#if CONFIG_SERIAL_ARC_NR_PORTS > 1
+ARC_UART_DEV(1);
+#endif
+
+static struct platform_device *fpga_early_devs[] __initdata = {
+#if defined(CONFIG_SERIAL_ARC_CONSOLE)
+ &arc_uart0_dev,
+#endif
+};
+
+static void arc_fpga_serial_init(void)
+{
+ /* To let driver workaround ISS bug: baudh Reg can't be set to 0 */
+ arc_uart_info[2] = !running_on_hw;
+
+ early_platform_add_devices(fpga_early_devs,
+ ARRAY_SIZE(fpga_early_devs));
+
+ /*
+ * ARC console driver registers itself as an early platform driver
+ * of class "earlyprintk".
+ * Install it here, followed by probe of devices.
+ * The installation here doesn't require earlyprintk in command line
+ * To do so however, replace the lines below with
+ * parse_early_param();
+ * early_platform_driver_probe("earlyprintk", 1, 1);
+ * ^^
+ */
+ early_platform_driver_register_all("earlyprintk");
+ early_platform_driver_probe("earlyprintk", 1, 0);
+
+ /*
+ * This is to make sure that arc uart would be preferred console
+ * despite one/more of following:
+ * -command line lacked "console=ttyARC0" or
+ * -CONFIG_VT_CONSOLE was enabled (for no reason whatsoever)
+ * Note that this needs to be done after above early console is reg,
+ * otherwise the early console never gets a chance to run.
+ */
+ add_preferred_console("ttyARC", 0, "115200");
+}
+
+#else
+
+static void arc_fpga_serial_init(void)
+{
+}
+
+#endif /* CONFIG_SERIAL_ARC */
/*
* Early Platform Initialization called from setup_arch()
@@ -18,12 +110,25 @@
void __init arc_platform_early_init(void)
{
pr_info("[plat-arcfpga]: registering early dev resources\n");
+
+ arc_fpga_serial_init();
}
+static struct platform_device *fpga_devs[] __initdata = {
+#if defined(CONFIG_SERIAL_ARC) || defined(CONFIG_SERIAL_ARC_MODULE)
+ &arc_uart0_dev,
+#if CONFIG_SERIAL_ARC_NR_PORTS > 1
+ &arc_uart1_dev,
+#endif
+#endif
+};
+
int __init fpga_plat_init(void)
{
pr_info("[plat-arcfpga]: registering device resources\n");
+ platform_add_devices(fpga_devs, ARRAY_SIZE(fpga_devs));
+
return 0;
}
arch_initcall(fpga_plat_init);
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/page.h | 92 +++++++++
arch/arc/include/asm/pgalloc.h | 134 +++++++++++++
arch/arc/include/asm/pgtable.h | 401 ++++++++++++++++++++++++++++++++++++++++
3 files changed, 627 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/pgalloc.h
create mode 100644 arch/arc/include/asm/pgtable.h
diff --git a/arch/arc/include/asm/page.h b/arch/arc/include/asm/page.h
index 7cd1224..d111d0c 100644
--- a/arch/arc/include/asm/page.h
+++ b/arch/arc/include/asm/page.h
@@ -35,6 +35,98 @@
#define PAGE_MASK (~(PAGE_SIZE-1))
+#ifdef __KERNEL__
+
+#ifndef __ASSEMBLY__
+
+#define get_user_page(vaddr) __get_free_page(GFP_KERNEL)
+#define free_user_page(page, addr) free_page(addr)
+
+/* TBD: for now don't worry about VIPT D$ aliasing */
+#define clear_page(paddr) memset((paddr), 0, PAGE_SIZE)
+#define copy_page(to, from) memcpy((to), (from), PAGE_SIZE)
+
+#define clear_user_page(addr, vaddr, pg) clear_page(addr)
+#define copy_user_page(vto, vfrom, vaddr, pg) copy_page(vto, vfrom)
+
+#undef STRICT_MM_TYPECHECKS
+
+#ifdef STRICT_MM_TYPECHECKS
+/*
+ * These are used to make use of C type-checking..
+ */
+typedef struct {
+ unsigned long pte;
+} pte_t;
+typedef struct {
+ unsigned long pgd;
+} pgd_t;
+typedef struct {
+ unsigned long pgprot;
+} pgprot_t;
+typedef unsigned long pgtable_t;
+
+#define pte_val(x) ((x).pte)
+#define pgd_val(x) ((x).pgd)
+#define pgprot_val(x) ((x).pgprot)
+
+#define __pte(x) ((pte_t) { (x) })
+#define __pgd(x) ((pgd_t) { (x) })
+#define __pgprot(x) ((pgprot_t) { (x) })
+
+#else /* !STRICT_MM_TYPECHECKS */
+
+typedef unsigned long pte_t;
+typedef unsigned long pgd_t;
+typedef unsigned long pgprot_t;
+typedef unsigned long pgtable_t;
+
+#define pte_val(x) (x)
+#define pgd_val(x) (x)
+#define pgprot_val(x) (x)
+#define __pte(x) (x)
+#define __pgprot(x) (x)
+
+#endif
+
+#define ARCH_PFN_OFFSET (CONFIG_LINUX_LINK_BASE >> PAGE_SHIFT)
+
+#define pfn_valid(pfn) (((pfn) - ARCH_PFN_OFFSET) < max_mapnr)
+
+/*
+ * __pa, __va, virt_to_page (ALERT: deprecated, don't use them)
+ *
+ * These macros have historically been misnamed
+ * virt here means link-address/program-address as embedded in object code.
+ * So if kernel img is linked at 0x8000_0000 onwards, 0x8010_0000 will be
+ * 128th page, and virt_to_page( ) will return the struct page corresp to it.
+ * mem_map[ ] is an array of struct page for each page frame in the system
+ *
+ * Independent of where linux is linked at, link-addr = physical address
+ * So the old macro __pa = vaddr + PAGE_OFFSET - CONFIG_LINUX_LINK_BASE
+ * would have been wrong in case kernel is not at 0x8zs
+ */
+#define __pa(vaddr) ((unsigned long)vaddr)
+#define __va(paddr) ((void *)((unsigned long)(paddr)))
+
+#define virt_to_page(kaddr) \
+ (mem_map + ((__pa(kaddr) - CONFIG_LINUX_LINK_BASE) >> PAGE_SHIFT))
+
+#define virt_addr_valid(kaddr) pfn_valid(__pa(kaddr) >> PAGE_SHIFT)
+
+/* Default Permissions for page, used in mmap.c */
+#ifdef CONFIG_ARC_STACK_NONEXEC
+#define VM_DATA_DEFAULT_FLAGS (VM_READ | VM_WRITE | VM_MAYREAD | VM_MAYWRITE)
+#else
+#define VM_DATA_DEFAULT_FLAGS (VM_READ | VM_WRITE | VM_EXEC | \
+ VM_MAYREAD | VM_MAYWRITE | VM_MAYEXEC)
+#endif
+
+#define WANT_PAGE_VIRTUAL 1
+
+#include <asm-generic/memory_model.h> /* page_to_pfn, pfn_to_page */
+#include <asm-generic/getorder.h>
+
#endif /* !__ASSEMBLY__ */
#endif /* __KERNEL__ */
diff --git a/arch/arc/include/asm/pgalloc.h b/arch/arc/include/asm/pgalloc.h
new file mode 100644
index 0000000..36a9f20
--- /dev/null
+++ b/arch/arc/include/asm/pgalloc.h
@@ -0,0 +1,134 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: June 2011
+ * -"/proc/meminfo | grep PageTables" kept on increasing
+ * Recently added pgtable dtor was not getting called.
+ *
+ * vineetg: May 2011
+ * -Variable pg-sz means that Page Tables could be variable sized themselves
+ * So calculate it based on addr traversal split [pgd-bits:pte-bits:xxx]
+ * -Page Table size capped to max 1 to save memory - hence verified.
+ * -Since these deal with constants, gcc compile-time optimizes them.
+ *
+ * vineetg: Nov 2010
+ * -Added pgtable ctor/dtor used for pgtable mem accounting
+ *
+ * vineetg: April 2010
+ * -Switched pgtable_t from being struct page * to unsigned long
+ * =Needed so that Page Table allocator (pte_alloc_one) is not forced to
+ * to deal with struct page. Thay way in future we can make it allocate
+ * multiple PG Tbls in one Page Frame
+ * =sweet side effect is avoiding calls to ugly page_address( ) from the
+ * pg-tlb allocator sub-sys (pte_alloc_one, ptr_free, pmd_populate
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef _ASM_ARC_PGALLOC_H
+#define _ASM_ARC_PGALLOC_H
+
+#include <linux/mm.h>
+#include <linux/log2.h>
+
+static inline void
+pmd_populate_kernel(struct mm_struct *mm, pmd_t *pmd, pte_t *pte)
+{
+ pmd_set(pmd, pte);
+}
+
+static inline void
+pmd_populate(struct mm_struct *mm, pmd_t *pmd, pgtable_t ptep)
+{
+ pmd_set(pmd, (pte_t *) ptep);
+}
+
+static inline int __get_order_pgd(void)
+{
+ return get_order(PTRS_PER_PGD * 4);
+}
+
+static inline pgd_t *pgd_alloc(struct mm_struct *mm)
+{
+ int num, num2;
+ pgd_t *ret = (pgd_t *) __get_free_pages(GFP_KERNEL, __get_order_pgd());
+
+ if (ret) {
+ num = USER_PTRS_PER_PGD + USER_KERNEL_GUTTER / PGDIR_SIZE;
+ memzero(ret, num * sizeof(pgd_t));
+
+ num2 = VMALLOC_SIZE / PGDIR_SIZE;
+ memcpy(ret + num, swapper_pg_dir + num, num2 * sizeof(pgd_t));
+
+ memzero(ret + num + num2,
+ (PTRS_PER_PGD - num - num2) * sizeof(pgd_t));
+
+ }
+ return ret;
+}
+
+static inline void pgd_free(struct mm_struct *mm, pgd_t *pgd)
+{
+ free_pages((unsigned long)pgd, __get_order_pgd());
+}
+
+
+/*
+ * With software-only page-tables, addr-split for traversal is tweakable and
+ * that directly governs how big tables would be at each level.
+ * Further, the MMU page size is configurable.
+ * Thus we need to programatically assert the size constraint
+ * All of this is const math, allowing gcc to do constant folding/propagation.
+ */
+
+static inline int __get_order_pte(void)
+{
+ return get_order(PTRS_PER_PTE * 4);
+}
+
+static inline pte_t *pte_alloc_one_kernel(struct mm_struct *mm,
+ unsigned long address)
+{
+ pte_t *pte;
+
+ pte = (pte_t *) __get_free_pages(GFP_KERNEL | __GFP_REPEAT | __GFP_ZERO,
+ __get_order_pte());
+
+ return pte;
+}
+
+static inline pgtable_t
+pte_alloc_one(struct mm_struct *mm, unsigned long address)
+{
+ pgtable_t pte_pg;
+
+ pte_pg = __get_free_pages(GFP_KERNEL | __GFP_REPEAT, __get_order_pte());
+ if (pte_pg) {
+ memzero((void *)pte_pg, PTRS_PER_PTE * 4);
+ pgtable_page_ctor(virt_to_page(pte_pg));
+ }
+
+ return pte_pg;
+}
+
+static inline void pte_free_kernel(struct mm_struct *mm, pte_t *pte)
+{
+ free_pages((unsigned long)pte, __get_order_pte()); /* takes phy addr */
+}
+
+static inline void pte_free(struct mm_struct *mm, pgtable_t ptep)
+{
+ pgtable_page_dtor(virt_to_page(ptep));
+ free_pages(ptep, __get_order_pte());
+}
+
+#define __pte_free_tlb(tlb, pte, addr) pte_free((tlb)->mm, pte)
+
+#define check_pgt_cache() do { } while (0)
+#define pmd_pgtable(pmd) pmd_page_vaddr(pmd)
+
+#endif /* _ASM_ARC_PGALLOC_H */
diff --git a/arch/arc/include/asm/pgtable.h b/arch/arc/include/asm/pgtable.h
new file mode 100644
index 0000000..dcb07015
--- /dev/null
+++ b/arch/arc/include/asm/pgtable.h
@@ -0,0 +1,401 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: May 2011
+ * -Folded PAGE_PRESENT (used by VM) and PAGE_VALID (used by MMU) into 1.
+ * They are semantically the same although in different contexts
+ * VALID marks a TLB entry exists and it will only happen if PRESENT
+ * - Utilise some unused free bits to confine PTE flags to 12 bits
+ * This is a must for 4k pg-sz
+ *
+ * vineetg: Mar 2011 - changes to accomodate MMU TLB Page Descriptor mods
+ * -TLB Locking never really existed, except for initial specs
+ * -SILENT_xxx not needed for our port
+ * -Per my request, MMU V3 changes the layout of some of the bits
+ * to avoid a few shifts in TLB Miss handlers.
+ *
+ * vineetg: April 2010
+ * -PGD entry no longer contains any flags. If empty it is 0, otherwise has
+ * Pg-Tbl ptr. Thus pmd_present(), pmd_valid(), pmd_set( ) become simpler
+ *
+ * vineetg: April 2010
+ * -Switched form 8:11:13 split for page table lookup to 11:8:13
+ * -this speeds up page table allocation itself as we now have to memset 1K
+ * instead of 8k per page table.
+ * -TODO: Right now page table alloc is 8K and rest 7K is unused
+ * need to optimise it
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef _ASM_ARC_PGTABLE_H
+#define _ASM_ARC_PGTABLE_H
+
+#include <asm/page.h>
+#include <asm/mmu.h>
+#include <asm-generic/pgtable-nopmd.h>
+
+/**************************************************************************
+ * Page Table Flags
+ *
+ * ARC700 MMU only deals with softare managed TLB entries.
+ * Page Tables are purely for Linux VM's consumption and the bits below are
+ * suited to that (uniqueness). Hence some are not implemented in the TLB and
+ * some have different value in TLB.
+ * e.g. MMU v2: K_READ bit is 8 and so is GLOBAL (possible becoz they live in
+ * seperate PD0 and PD1, which combined forms a translation entry)
+ * while for PTE perspective, they are 8 and 9 respectively
+ * with MMU v3: Most bits (except SHARED) represent the exact hardware pos
+ * (saves some bit shift ops in TLB Miss hdlrs)
+ */
+
+#if (CONFIG_ARC_MMU_VER <= 2)
+
+#define _PAGE_ACCESSED (1<<1) /* Page is accessed (S) */
+#define _PAGE_CACHEABLE (1<<2) /* Page is cached (H) */
+#define _PAGE_EXECUTE (1<<3) /* Page has user execute perm (H) */
+#define _PAGE_WRITE (1<<4) /* Page has user write perm (H) */
+#define _PAGE_READ (1<<5) /* Page has user read perm (H) */
+#define _PAGE_K_EXECUTE (1<<6) /* Page has kernel execute perm (H) */
+#define _PAGE_K_WRITE (1<<7) /* Page has kernel write perm (H) */
+#define _PAGE_K_READ (1<<8) /* Page has kernel perm (H) */
+#define _PAGE_GLOBAL (1<<9) /* Page is global (H) */
+#define _PAGE_MODIFIED (1<<10) /* Page modified (dirty) (S) */
+#define _PAGE_FILE (1<<10) /* page cache/ swap (S) */
+#define _PAGE_PRESENT (1<<11) /* TLB entry is valid (H) */
+
+#else
+
+/* PD1 */
+#define _PAGE_CACHEABLE (1<<0) /* Page is cached (H) */
+#define _PAGE_EXECUTE (1<<1) /* Page has user execute perm (H) */
+#define _PAGE_WRITE (1<<2) /* Page has user write perm (H) */
+#define _PAGE_READ (1<<3) /* Page has user read perm (H) */
+#define _PAGE_K_EXECUTE (1<<4) /* Page has kernel execute perm (H) */
+#define _PAGE_K_WRITE (1<<5) /* Page has kernel write perm (H) */
+#define _PAGE_K_READ (1<<6) /* Page has kernel perm (H) */
+#define _PAGE_ACCESSED (1<<7) /* Page is accessed (S) */
+
+/* PD0 */
+#define _PAGE_GLOBAL (1<<8) /* Page is global (H) */
+#define _PAGE_PRESENT (1<<9) /* TLB entry is valid (H) */
+#define _PAGE_SHARED_CODE (1<<10) /* Shared Code page with cmn vaddr
+ usable for shared TLB entries (H) */
+
+#define _PAGE_MODIFIED (1<<11) /* Page modified (dirty) (S) */
+#define _PAGE_FILE (1<<12) /* page cache/ swap (S) */
+
+#define _PAGE_SHARED_CODE_H (1<<31) /* Hardware counterpart of above */
+#endif
+
+/* Kernel allowed all permissions for all pages */
+#define _K_PAGE_PERMS (_PAGE_K_EXECUTE | _PAGE_K_WRITE | _PAGE_K_READ)
+
+#ifdef CONFIG_ARC_CACHE_PAGES
+#define _PAGE_DEF_CACHEABLE _PAGE_CACHEABLE
+#else
+#define _PAGE_DEF_CACHEABLE (0)
+#endif
+
+/* Helper for every "user" page
+ * -kernel can R/W/X
+ * -by default cached, unless config otherwise
+ * -present in memory
+ */
+#define ___DEF (_PAGE_PRESENT | _K_PAGE_PERMS | _PAGE_DEF_CACHEABLE)
+
+/* Set of bits not changed in pte_modify */
+#define _PAGE_CHG_MASK (PAGE_MASK | _PAGE_ACCESSED | _PAGE_MODIFIED)
+
+/* More Abbrevaited helpers */
+#define PAGE_U_NONE __pgprot(___DEF)
+#define PAGE_U_R __pgprot(___DEF | _PAGE_READ)
+#define PAGE_U_W_R __pgprot(___DEF | _PAGE_READ | _PAGE_WRITE)
+#define PAGE_U_X_R __pgprot(___DEF | _PAGE_READ | _PAGE_EXECUTE)
+#define PAGE_U_X_W_R __pgprot(___DEF | _PAGE_READ | _PAGE_WRITE | \
+ _PAGE_EXECUTE)
+
+#define PAGE_SHARED PAGE_U_W_R
+
+/* While kernel runs out of unstrslated space, vmalloc/modules use a chunk of
+ * kernel vaddr space - visible in all addr spaces, but kernel mode only
+ * Thus Global, all-kernel-access, no-user-access, cached
+ */
+#define PAGE_KERNEL __pgprot(___DEF | _PAGE_GLOBAL)
+
+/* ioremap */
+#define PAGE_KERNEL_NO_CACHE __pgprot(_PAGE_PRESENT | _K_PAGE_PERMS | \
+ _PAGE_GLOBAL)
+
+/**************************************************************************
+ * Mapping of vm_flags (Generic VM) to PTE flags (arch specific)
+ *
+ * Certain cases have 1:1 mapping
+ * e.g. __P101 means VM_READ, VM_EXEC and !VM_SHARED
+ * which directly corresponds to PAGE_U_X_R
+ *
+ * Other rules which cause the divergence from 1:1 mapping
+ *
+ * 1. Although ARC700 can do exclusive execute/write protection (meaning R
+ * can be tracked independet of X/W unlike some other CPUs), still to
+ * keep things consistent with other archs:
+ * -Write implies Read: W => R
+ * -Execute implies Read: X => R
+ *
+ * 2. Pvt Writable doesn't have Write Enabled initially: Pvt-W => !W
+ * This is to enable COW mechanism
+ */
+ /* xwr */
+#define __P000 PAGE_U_NONE
+#define __P001 PAGE_U_R
+#define __P010 PAGE_U_R /* Pvt-W => !W */
+#define __P011 PAGE_U_R /* Pvt-W => !W */
+#define __P100 PAGE_U_X_R /* X => R */
+#define __P101 PAGE_U_X_R
+#define __P110 PAGE_U_X_R /* Pvt-W => !W and X => R */
+#define __P111 PAGE_U_X_R /* Pvt-W => !W */
+
+#define __S000 PAGE_U_NONE
+#define __S001 PAGE_U_R
+#define __S010 PAGE_U_W_R /* W => R */
+#define __S011 PAGE_U_W_R
+#define __S100 PAGE_U_X_R /* X => R */
+#define __S101 PAGE_U_X_R
+#define __S110 PAGE_U_X_W_R /* X => R */
+#define __S111 PAGE_U_X_W_R
+
+/****************************************************************
+ * Page Table Lookup split
+ *
+ * We implement 2 tier paging and since this is all software, we are free
+ * to customize the span of a PGD / PTE entry to suit us
+ *
+ * 32 bit virtual address
+ * -------------------------------------------------------
+ * | BITS_FOR_PGD | BITS_FOR_PTE | BITS_IN_PAGE |
+ * -------------------------------------------------------
+ * | | |
+ * | | --> off in page frame
+ * | |
+ * | ---> index into Page Table
+ * |
+ * ----> index into Page Directory
+ */
+
+#define BITS_IN_PAGE PAGE_SHIFT
+
+/* Optimal Sizing of Pg Tbl - based on MMU page size */
+#if defined(CONFIG_ARC_PAGE_SIZE_8K)
+#define BITS_FOR_PTE 8
+#elif defined(CONFIG_ARC_PAGE_SIZE_16K)
+#define BITS_FOR_PTE 8
+#elif defined(CONFIG_ARC_PAGE_SIZE_4K)
+#define BITS_FOR_PTE 9
+#endif
+
+#define BITS_FOR_PGD (32 - BITS_FOR_PTE - BITS_IN_PAGE)
+
+#define PGDIR_SHIFT (BITS_FOR_PTE + BITS_IN_PAGE)
+#define PGDIR_SIZE (1UL << PGDIR_SHIFT) /* vaddr span, not PDG sz */
+#define PGDIR_MASK (~(PGDIR_SIZE-1))
+
+#ifdef __ASSEMBLY__
+#define PTRS_PER_PTE (1 << BITS_FOR_PTE)
+#define PTRS_PER_PGD (1 << BITS_FOR_PGD)
+#else
+#define PTRS_PER_PTE (1UL << BITS_FOR_PTE)
+#define PTRS_PER_PGD (1UL << BITS_FOR_PGD)
+#endif
+/*
+ * Number of entries a user land program use.
+ * TASK_SIZE is the maximum vaddr that can be used by a userland program.
+ */
+#define USER_PTRS_PER_PGD (TASK_SIZE / PGDIR_SIZE)
+
+/*
+ * No special requirements for lowest virtual address we permit any user space
+ * mapping to be mapped at.
+ */
+#define FIRST_USER_ADDRESS 0
+
+
+/****************************************************************
+ * Bucket load of VM Helpers
+ */
+
+#ifndef __ASSEMBLY__
+
+#define pte_ERROR(e) \
+ pr_crit("%s:%d: bad pte %08lx.\n", __FILE__, __LINE__, pte_val(e))
+#define pgd_ERROR(e) \
+ pr_crit("%s:%d: bad pgd %08lx.\n", __FILE__, __LINE__, pgd_val(e))
+
+/* the zero page used for uninitialized and anonymous pages */
+extern char empty_zero_page[PAGE_SIZE];
+#define ZERO_PAGE(vaddr) (virt_to_page(empty_zero_page))
+
+#define pte_unmap(pte) do { } while (0)
+#define pte_unmap_nested(pte) do { } while (0)
+
+#define set_pte(pteptr, pteval) ((*(pteptr)) = (pteval))
+#define set_pmd(pmdptr, pmdval) (*(pmdptr) = pmdval)
+
+/* find the page descriptor of the Page Tbl ref by PMD entry */
+#define pmd_page(pmd) virt_to_page(pmd_val(pmd) & PAGE_MASK)
+
+/* find the logical addr (phy for ARC) of the Page Tbl ref by PMD entry */
+#define pmd_page_vaddr(pmd) (pmd_val(pmd) & PAGE_MASK)
+
+/* In a 2 level sys, setup the PGD entry with PTE value */
+static inline void pmd_set(pmd_t *pmdp, pte_t *ptep)
+{
+ pmd_val(*pmdp) = (unsigned long)ptep;
+}
+
+#define pte_none(x) (!pte_val(x))
+#define pte_present(x) (pte_val(x) & _PAGE_PRESENT)
+#define pte_clear(mm, addr, ptep) set_pte_at(mm, addr, ptep, __pte(0))
+
+#define pmd_none(x) (!pmd_val(x))
+#define pmd_bad(x) ((pmd_val(x) & ~PAGE_MASK))
+#define pmd_present(x) (pmd_val(x))
+#define pmd_clear(xp) do { pmd_val(*(xp)) = 0; } while (0)
+
+#define pte_page(x) (mem_map + \
+ (unsigned long)(((pte_val(x) - PAGE_OFFSET) >> PAGE_SHIFT)))
+
+#define mk_pte(page, pgprot) \
+({ \
+ pte_t pte; \
+ pte_val(pte) = __pa(page_address(page)) + pgprot_val(pgprot); \
+ pte; \
+})
+
+/* TBD: Non linear mapping stuff */
+static inline int pte_file(pte_t pte)
+{
+ return pte_val(pte) & _PAGE_FILE;
+}
+
+#define PTE_FILE_MAX_BITS 30
+#define pgoff_to_pte(x) __pte(x)
+#define pte_to_pgoff(x) (pte_val(x) >> 2)
+#define pte_pfn(pte) (pte_val(pte) >> PAGE_SHIFT)
+#define pfn_pte(pfn, prot) (__pte(((pfn) << PAGE_SHIFT) | pgprot_val(prot)))
+#define __pte_index(addr) (((addr) >> PAGE_SHIFT) & (PTRS_PER_PTE - 1))
+
+/*
+ * pte_offset gets a @ptr to PMD entry (PGD in our 2-tier paging system)
+ * and returns ptr to PTE entry corresponding to @addr
+ */
+#define pte_offset(dir, addr) ((pte_t *)(pmd_page_vaddr(*dir)) +\
+ __pte_index(addr))
+
+/* No mapping of Page Tables in high mem etc, so following same as above */
+#define pte_offset_kernel(dir, addr) pte_offset(dir, addr)
+#define pte_offset_map(dir, addr) pte_offset(dir, addr)
+
+/* Zoo of pte_xxx function */
+#define pte_read(pte) (pte_val(pte) & _PAGE_READ)
+#define pte_write(pte) (pte_val(pte) & _PAGE_WRITE)
+#define pte_dirty(pte) (pte_val(pte) & _PAGE_MODIFIED)
+#define pte_young(pte) (pte_val(pte) & _PAGE_ACCESSED)
+#define pte_special(pte) (0)
+
+#define PTE_BIT_FUNC(fn, op) \
+ static inline pte_t pte_##fn(pte_t pte) { pte_val(pte) op; return pte; }
+
+PTE_BIT_FUNC(wrprotect, &= ~(_PAGE_WRITE));
+PTE_BIT_FUNC(mkwrite, |= (_PAGE_WRITE));
+PTE_BIT_FUNC(mkclean, &= ~(_PAGE_MODIFIED));
+PTE_BIT_FUNC(mkdirty, |= (_PAGE_MODIFIED));
+PTE_BIT_FUNC(mkold, &= ~(_PAGE_ACCESSED));
+PTE_BIT_FUNC(mkyoung, |= (_PAGE_ACCESSED));
+PTE_BIT_FUNC(exprotect, &= ~(_PAGE_EXECUTE));
+PTE_BIT_FUNC(mkexec, |= (_PAGE_EXECUTE));
+
+static inline pte_t pte_mkspecial(pte_t pte) { return pte; }
+
+static inline pte_t pte_modify(pte_t pte, pgprot_t newprot)
+{
+ return __pte((pte_val(pte) & _PAGE_CHG_MASK) | pgprot_val(newprot));
+}
+
+/* Macro to mark a page protection as uncacheable */
+#define pgprot_noncached(prot) (__pgprot(pgprot_val(prot) & ~_PAGE_CACHEABLE))
+
+static inline void set_pte_at(struct mm_struct *mm, unsigned long addr,
+ pte_t *ptep, pte_t pteval)
+{
+ set_pte(ptep, pteval);
+}
+
+/*
+ * All kernel related VM pages are in init's mm.
+ */
+#define pgd_offset_k(address) pgd_offset(&init_mm, address)
+#define pgd_index(addr) ((addr) >> PGDIR_SHIFT)
+#define pgd_offset(mm, addr) (((mm)->pgd)+pgd_index(addr))
+
+/*
+ * Macro to quickly access the PGD entry, utlising the fact that some
+ * arch may cache the pointer to Page Directory of "current" task
+ * in a MMU register
+ *
+ * Thus task->mm->pgd (3 pointer dereferences, cache misses etc simply
+ * becomes read a register
+ *
+ * ********CAUTION*******:
+ * Kernel code might be dealing with some mm_struct of NON "current"
+ * Thus use this macro only when you are certain that "current" is current
+ * e.g. when dealing with signal frame setup code etc
+ */
+#define pgd_offset_fast(mm, addr) \
+({ \
+ pgd_t *pgd_base = (pgd_t *) read_aux_reg(ARC_REG_SCRATCH_DATA0); \
+ pgd_base + pgd_index(addr); \
+})
+
+extern void paging_init(void);
+extern pgd_t swapper_pg_dir[] __aligned(PAGE_SIZE);
+void update_mmu_cache(struct vm_area_struct *vma, unsigned long address,
+ pte_t *ptep);
+
+/* Encode swap {type,off} tuple into PTE
+ * We reserve 13 bits for 5-bit @type, keeping bits 12-5 zero, ensuring that
+ * both PAGE_FILE and PAGE_PRESENT are zero in a PTE holding swap "identifier"
+ */
+#define __swp_entry(type, off) ((swp_entry_t) { \
+ ((type) & 0x1f) | ((off) << 13) })
+
+/* Decode a PTE containing swap "identifier "into constituents */
+#define __swp_type(pte_lookalike) (((pte_lookalike).val) & 0x1f)
+#define __swp_offset(pte_lookalike) ((pte_lookalike).val << 13)
+
+/* NOPs, to keep generic kernel happy */
+#define __pte_to_swp_entry(pte) ((swp_entry_t) { pte_val(pte) })
+#define __swp_entry_to_pte(x) ((pte_t) { (x).val })
+
+#define kern_addr_valid(addr) (1)
+
+/*
+ * remap a physical page `pfn' of size `size' with page protection `prot'
+ * into virtual address `from'
+ */
+#define io_remap_pfn_range(vma, from, pfn, size, prot) \
+ remap_pfn_range(vma, from, pfn, size, prot)
+
+#include <asm-generic/pgtable.h>
+
+/*
+ * No page table caches to initialise
+ */
+#define pgtable_cache_init() do { } while (0)
+
+#endif /* __ASSEMBLY__ */
+
+#endif
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/arcregs.h | 5 +
arch/arc/include/asm/setup.h | 22 +++++
arch/arc/kernel/head.S | 78 ++++++++++++++++
arch/arc/kernel/reset.c | 33 +++++++
arch/arc/kernel/setup.c | 166 ++++++++++++++++++++++++++++++++++
arch/arc/mm/init.c | 185 ++++++++++++++++++++++++++++++++++++++
arch/arc/plat-arcfpga/platform.c | 29 ++++++
7 files changed, 518 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/setup.h
create mode 100644 arch/arc/kernel/head.S
create mode 100644 arch/arc/kernel/reset.c
create mode 100644 arch/arc/kernel/setup.c
create mode 100644 arch/arc/mm/init.c
create mode 100644 arch/arc/plat-arcfpga/platform.c
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
index aedfbf1..725b94c 100644
--- a/arch/arc/include/asm/arcregs.h
+++ b/arch/arc/include/asm/arcregs.h
@@ -257,6 +257,11 @@
} \
}
+/* Helpers */
+#define TO_KB(bytes) ((bytes) >> 10)
+#define TO_MB(bytes) (TO_KB(bytes) >> 10)
+#define PAGES_TO_KB(n_pages) ((n_pages) << (PAGE_SHIFT - 10))
+#define PAGES_TO_MB(n_pages) (PAGES_TO_KB(n_pages) >> 10)
#ifdef CONFIG_ARC_FPU_SAVE_RESTORE
/* These DPFP regs need to be saved/restored across ctx-sw */
diff --git a/arch/arc/include/asm/setup.h b/arch/arc/include/asm/setup.h
new file mode 100644
index 0000000..ab427e6
--- /dev/null
+++ b/arch/arc/include/asm/setup.h
@@ -0,0 +1,22 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASMARC_SETUP_H
+#define __ASMARC_SETUP_H
+
+#include <linux/types.h>
+
+#define COMMAND_LINE_SIZE 256
+
+extern int root_mountflags, end_mem;
+extern int running_on_hw;
+
+void __init setup_processor(void);
+void __init setup_arch_memory(void);
+
+#endif /* __ASMARC_SETUP_H */
diff --git a/arch/arc/kernel/head.S b/arch/arc/kernel/head.S
new file mode 100644
index 0000000..e63f6a4
--- /dev/null
+++ b/arch/arc/kernel/head.S
@@ -0,0 +1,78 @@
+/*
+ * ARC CPU startup Code
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Vineetg: Dec 2007
+ * -Check if we are running on Simulator or on real hardware
+ * to skip certain things during boot on simulator
+ */
+
+#include <asm/asm-offsets.h>
+#include <asm/entry.h>
+#include <linux/linkage.h>
+#include <asm/arcregs.h>
+
+ .cpu A7
+
+ .section .init.text, "ax",@progbits
+ .type stext, @function
+ .globl stext
+stext:
+ ;-------------------------------------------------------------------
+ ; Don't clobber r0-r4 yet. It might have bootloader provided info
+ ;-------------------------------------------------------------------
+
+ ; Clear BSS before updating any globals
+ ; XXX: use ZOL here
+ mov r5, __bss_start
+ mov r6, __bss_stop
+1:
+ st.ab 0, [r5,4]
+ brlt r5, r6, 1b
+
+#ifdef CONFIG_CMDLINE_UBOOT
+ ; support for bootloader provided cmdline
+ ; If cmdline passed by u-boot, then
+ ; r0 = 1 (because ATAGS parsing, now retired, used to use 0)
+ ; r1 = magic number (board identity)
+ ; r2 = addr of cmdline string (somewhere in memory/flash)
+
+ brne r0, 1, .Lother_bootup_chores ; u-boot didn't pass cmdline
+ breq r2, 0, .Lother_bootup_chores ; or cmdline is NULL
+
+ mov r5, @command_line
+1:
+ ldb.ab r6, [r2, 1]
+ breq r6, 0, .Lother_bootup_chores
+ b.d 1b
+ stb.ab r6, [r5, 1]
+#endif
+
+.Lother_bootup_chores:
+
+ ; Identify if running on ISS vs Silicon
+ ; IDENTITY Reg [ 3 2 1 0 ]
+ ; (chip-id) ^^^^^ ==> 0xffff for ISS
+ lr r0, [identity]
+ lsr r3, r0, 16
+ cmp r3, 0xffff
+ mov.z r4, 0
+ mov.nz r4, 1
+ st r4, [@running_on_hw]
+
+ ; setup "current" tsk and optionally cache it in dedicated r25
+ mov r9, @init_task
+ SET_CURR_TASK_ON_CPU r9, r0 ; r9 = tsk, r0 = scratch
+
+ ; setup stack (fp, sp)
+ mov fp, 0
+
+ ; tsk->thread_info is really a PAGE, whose bottom hoists stack
+ GET_TSK_STACK_BASE r9, sp ; r9 = tsk, sp = stack base(output)
+
+ j start_kernel ; "C" entry point
diff --git a/arch/arc/kernel/reset.c b/arch/arc/kernel/reset.c
new file mode 100644
index 0000000..e227a2b
--- /dev/null
+++ b/arch/arc/kernel/reset.c
@@ -0,0 +1,33 @@
+/*
+ * Copyright (C) 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/kernel.h>
+#include <linux/printk.h>
+#include <linux/reboot.h>
+#include <linux/pm.h>
+
+void machine_halt(void)
+{
+ /* Halt the processor */
+ __asm__ __volatile__("flag 1\n");
+}
+
+void machine_restart(char *__unused)
+{
+ /* Soft reset : jump to reset vector */
+ pr_info("Put your restart handler here\n");
+ machine_halt();
+}
+
+void machine_power_off(void)
+{
+ /* FIXME :: power off ??? */
+ machine_halt();
+}
+
+void (*pm_power_off) (void) = NULL;
diff --git a/arch/arc/kernel/setup.c b/arch/arc/kernel/setup.c
new file mode 100644
index 0000000..82ac206
--- /dev/null
+++ b/arch/arc/kernel/setup.c
@@ -0,0 +1,166 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/seq_file.h>
+#include <linux/fs.h>
+#include <linux/delay.h>
+#include <linux/root_dev.h>
+#include <linux/console.h>
+#include <linux/module.h>
+#include <linux/cpu.h>
+#include <asm/arcregs.h>
+#include <asm/tlb.h>
+#include <asm/cache.h>
+#include <asm/setup.h>
+#include <asm/page.h>
+#include <asm/irq.h>
+#include <asm/arcregs.h>
+
+#define FIX_PTR(x) __asm__ __volatile__(";" : "+r"(x))
+
+int running_on_hw = 1; /* vs. on ISS */
+
+char __initdata command_line[COMMAND_LINE_SIZE];
+
+struct task_struct *_current_task[NR_CPUS]; /* For stack switching */
+
+struct cpuinfo_arc cpuinfo_arc700[NR_CPUS];
+
+void __init read_arc_build_cfg_regs(void)
+{
+ read_decode_mmu_bcr();
+ read_decode_cache_bcr();
+}
+
+/*
+ * Initialize and setup the processor core
+ * This is called by all the CPUs thus should not do special case stuff
+ * such as only for boot CPU etc
+ */
+
+void __init setup_processor(void)
+{
+ read_arc_build_cfg_regs();
+ arc_init_IRQ();
+ arc_mmu_init();
+ arc_cache_init();
+}
+
+void __init __attribute__((weak)) arc_platform_early_init(void)
+{
+}
+
+void __init setup_arch(char **cmdline_p)
+{
+#ifdef CONFIG_CMDLINE_UBOOT
+ /* Make sure that a whitespace is inserted before */
+ strlcat(command_line, " ", sizeof(command_line));
+#endif
+ /*
+ * Append .config cmdline to base command line, which might already
+ * contain u-boot "bootargs" (handled by head.S, if so configured)
+ */
+ strlcat(command_line, CONFIG_CMDLINE, sizeof(command_line));
+
+ /* Save unparsed command line copy for /proc/cmdline */
+ strlcpy(boot_command_line, command_line, COMMAND_LINE_SIZE);
+ *cmdline_p = command_line;
+
+ /* To force early parsing of things like mem=xxx */
+ parse_early_param();
+
+ /* Platform/board specific: e.g. early console registration */
+ arc_platform_early_init();
+
+ setup_processor();
+
+ setup_arch_memory();
+
+ /* Can be issue if someone passes cmd line arg "ro"
+ * But that is unlikely so keeping it as it is
+ */
+ root_mountflags &= ~MS_RDONLY;
+
+ console_verbose();
+
+#if defined(CONFIG_VT) && defined(CONFIG_DUMMY_CONSOLE)
+ conswitchp = &dummy_con;
+#endif
+
+}
+
+/*
+ * Get CPU information for use by the procfs.
+ */
+
+#define cpu_to_ptr(c) ((void *)(0xFFFF0000 | (unsigned int)(c)))
+#define ptr_to_cpu(p) (~0xFFFF0000UL & (unsigned int)(p))
+
+static int show_cpuinfo(struct seq_file *m, void *v)
+{
+ char *str;
+ int cpu_id = ptr_to_cpu(v);
+
+ str = (char *)__get_free_page(GFP_TEMPORARY);
+ if (!str)
+ goto done;
+
+ seq_printf(m, "ARC700 #%d\n", cpu_id);
+
+ seq_printf(m, "Bogo MIPS : \t%lu.%02lu\n",
+ loops_per_jiffy / (500000 / HZ),
+ (loops_per_jiffy / (5000 / HZ)) % 100);
+
+ free_page((unsigned long)str);
+done:
+ seq_printf(m, "\n\n");
+
+ return 0;
+}
+
+static void *c_start(struct seq_file *m, loff_t *pos)
+{
+ /*
+ * Callback returns cpu-id to iterator for show routine, NULL to stop.
+ * However since NULL is also a valid cpu-id (0), we use a round-about
+ * way to pass it w/o having to kmalloc/free a 2 byte string.
+ * Encode cpu-id as 0xFFcccc, which is decoded by show routine.
+ */
+ return *pos < num_possible_cpus() ? cpu_to_ptr(*pos) : NULL;
+}
+
+static void *c_next(struct seq_file *m, void *v, loff_t *pos)
+{
+ ++*pos;
+ return c_start(m, pos);
+}
+
+static void c_stop(struct seq_file *m, void *v)
+{
+}
+
+const struct seq_operations cpuinfo_op = {
+ .start = c_start,
+ .next = c_next,
+ .stop = c_stop,
+ .show = show_cpuinfo
+};
+
+static DEFINE_PER_CPU(struct cpu, cpu_topology);
+
+static int __init topology_init(void)
+{
+ int cpu;
+
+ for_each_present_cpu(cpu)
+ register_cpu(&per_cpu(cpu_topology, cpu), cpu);
+
+ return 0;
+}
+
+subsys_initcall(topology_init);
diff --git a/arch/arc/mm/init.c b/arch/arc/mm/init.c
new file mode 100644
index 0000000..2373ebf
--- /dev/null
+++ b/arch/arc/mm/init.c
@@ -0,0 +1,185 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/kernel.h>
+#include <linux/mm.h>
+#include <linux/bootmem.h>
+#ifdef CONFIG_BLOCK_DEV_RAM
+#include <linux/blk.h>
+#endif
+#include <linux/swap.h>
+#include <linux/module.h>
+#include <asm/page.h>
+#include <asm/pgalloc.h>
+#include <asm/sections.h>
+#include <asm/arcregs.h>
+
+pgd_t swapper_pg_dir[PTRS_PER_PGD] __aligned(PAGE_SIZE);
+char empty_zero_page[PAGE_SIZE] __aligned(PAGE_SIZE);
+EXPORT_SYMBOL(empty_zero_page);
+
+/* Default tot mem from .config */
+static unsigned long end_mem = CONFIG_ARC_PLAT_SDRAM_SIZE +
+ CONFIG_LINUX_LINK_BASE;
+
+/* User can over-ride above with "mem=nnn[KkMm]" in cmdline */
+static int __init setup_mem_sz(char *str)
+{
+ unsigned long sz_bytes;
+
+ sz_bytes = memparse(str, NULL);
+ end_mem = CONFIG_LINUX_LINK_BASE + (sz_bytes & PAGE_MASK);
+
+ /* early console might not be setup yet - it will show up later */
+ pr_info("\"mem=%s\": End mem set to 0x%lx\n", str, end_mem);
+
+ return 0;
+}
+early_param("mem", setup_mem_sz);
+
+/*
+ * First memory setup routine called from setup_arch()
+ * 1. setup swapper's mm @init_mm
+ * 2. Count the pages we have and setup bootmem allocator
+ * 3. zone setup
+ */
+void __init setup_arch_memory(void)
+{
+ int bootmap_sz;
+ unsigned int first_free_pfn;
+ unsigned long kernel_img_end, alloc_start;
+ unsigned long zones_size[MAX_NR_ZONES] = { 0, 0 };
+
+ init_mm.start_code = (unsigned long)_text;
+ init_mm.end_code = (unsigned long)_etext;
+ init_mm.end_data = (unsigned long)_edata;
+ init_mm.brk = (unsigned long)_end;
+
+ /* _end needs to be page aligned */
+ kernel_img_end = (unsigned long)_end;
+ BUG_ON(kernel_img_end & ~PAGE_MASK);
+
+ /* first page of system - kernel .vector starts here */
+ min_low_pfn = PFN_DOWN(CONFIG_LINUX_LINK_BASE);
+
+ /* First free page beyond kernel image */
+ first_free_pfn = PFN_DOWN(kernel_img_end);
+
+ /*
+ * Last usable page of low mem (no HIGHMEM yet for ARC port)
+ * -must be BASE + SIZE
+ */
+ max_low_pfn = max_pfn = PFN_DOWN(end_mem);
+
+ max_mapnr = num_physpages = max_low_pfn - min_low_pfn;
+
+ /* setup bootmem allocator */
+ bootmap_sz = init_bootmem_node(NODE_DATA(0),
+ first_free_pfn,/* bitmap start */
+ min_low_pfn, /* First pg to track */
+ max_low_pfn); /* Last pg to track */
+
+ /*
+ * init_bootmem above marks all tracked Page-frames as inuse "allocated"
+ * This includes pages occupied by kernel's elf segments.
+ * Beyond that, excluding bootmem bitmap itself, mark the rest of
+ * free-mem as "allocatable"
+ */
+ alloc_start = kernel_img_end + bootmap_sz;
+ free_bootmem(alloc_start, end_mem - alloc_start);
+
+ memset(zones_size, 0, sizeof(zones_size));
+ zones_size[ZONE_NORMAL] = num_physpages;
+
+ /*
+ * Must not use free_area_init() as that uses PAGE_OFFSET, which
+ * need not be the case when our kernel linked at non-default addr
+ * i.e. when CONFIG_LINUX_LINK_BASE != PAGE_OFFSET
+ */
+ free_area_init_node(0, /* node-id */
+ zones_size, /* num pages per zone */
+ min_low_pfn, /* first pfn of node */
+ NULL); /* NO holes */
+}
+
+/*
+ * mem_init - initializes memory
+ *
+ * Frees up bootmem
+ * Calculates and displays memory available/used
+ */
+void __init mem_init(void)
+{
+ int codesize, datasize, initsize, reserved_pages, free_pages;
+ int tmp;
+
+ high_memory = (void *)end_mem;
+
+ totalram_pages = free_all_bootmem();
+
+ /* count all reserved pages [kernel code/data/mem_map..] */
+ reserved_pages = 0;
+ for (tmp = 0; tmp < max_mapnr; tmp++)
+ if (PageReserved(mem_map + tmp))
+ reserved_pages++;
+
+ /* XXX: nr_free_pages() is equivalent */
+ free_pages = max_mapnr - reserved_pages;
+
+ /*
+ * For the purpose of display below, split the "reserve mem"
+ * kernel code/data is already shown explicitly,
+ * Show any other reservations (mem_map[ ] et al)
+ */
+ reserved_pages -= (((unsigned int)_end - CONFIG_LINUX_LINK_BASE) >>
+ PAGE_SHIFT);
+
+ codesize = _etext - _text;
+ datasize = _end - _etext;
+ initsize = __init_end - __init_begin;
+
+ pr_info("Memory Available: %dM / %uM (%dK code, %dK data, %dK init, %dK reserv)\n",
+ PAGES_TO_MB(free_pages),
+ TO_MB(CONFIG_ARC_PLAT_SDRAM_SIZE),
+ TO_KB(codesize), TO_KB(datasize), TO_KB(initsize),
+ PAGES_TO_KB(reserved_pages));
+}
+
+static void __init free_init_pages(const char *what, unsigned long begin,
+ unsigned long end)
+{
+ unsigned long addr;
+
+ pr_info("Freeing %s: %ldk [%lx] to [%lx]\n",
+ what, TO_KB(end - begin), begin, end);
+
+ /* need to check that the page we free is not a partial page */
+ for (addr = begin; addr + PAGE_SIZE <= end; addr += PAGE_SIZE) {
+ ClearPageReserved(virt_to_page(addr));
+ init_page_count(virt_to_page(addr));
+ free_page(addr);
+ totalram_pages++;
+ }
+}
+
+/*
+ * free_initmem: Free all the __init memory.
+ */
+void __init_refok free_initmem(void)
+{
+ free_init_pages("unused kernel memory",
+ (unsigned long)__init_begin,
+ (unsigned long)__init_end);
+}
+
+#ifdef CONFIG_BLK_DEV_INITRD
+void __init free_initrd_mem(unsigned long start, unsigned long end)
+{
+ free_init_pages("initrd memory", start, end);
+}
+#endif
diff --git a/arch/arc/plat-arcfpga/platform.c b/arch/arc/plat-arcfpga/platform.c
new file mode 100644
index 0000000..7b5dcab
--- /dev/null
+++ b/arch/arc/plat-arcfpga/platform.c
@@ -0,0 +1,29 @@
+/*
+ * ARC FPGA Platform support code
+ *
+ * Copyright (C) 2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/types.h>
+#include <linux/init.h>
+#include <linux/platform_device.h>
+
+/*
+ * Early Platform Initialization called from setup_arch()
+ */
+void __init arc_platform_early_init(void)
+{
+ pr_info("[plat-arcfpga]: registering early dev resources\n");
+}
+
+int __init fpga_plat_init(void)
+{
+ pr_info("[plat-arcfpga]: registering device resources\n");
+
+ return 0;
+}
+arch_initcall(fpga_plat_init);
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/hw_irq.h | 7 --
arch/arc/kernel/traps.c | 125 +++++++++++++++++++++++++++++++++++++++++
2 files changed, 125 insertions(+), 7 deletions(-)
delete mode 100644 arch/arc/include/asm/hw_irq.h
create mode 100644 arch/arc/kernel/traps.c
diff --git a/arch/arc/include/asm/hw_irq.h b/arch/arc/include/asm/hw_irq.h
deleted file mode 100644
index fd565ab..0000000
--- a/arch/arc/include/asm/hw_irq.h
+++ /dev/null
@@ -1,7 +0,0 @@
-/*
- * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License version 2 as
- * published by the Free Software Foundation.
- */
diff --git a/arch/arc/kernel/traps.c b/arch/arc/kernel/traps.c
new file mode 100644
index 0000000..fd2457c
--- /dev/null
+++ b/arch/arc/kernel/traps.c
@@ -0,0 +1,125 @@
+/*
+ * Traps/Non-MMU Exception handling for ARC
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Rahul Trivedi: Codito Technologies 2004
+ */
+
+#include <linux/sched.h>
+#include <linux/kdebug.h>
+#include <linux/uaccess.h>
+#include <asm/ptrace.h>
+#include <asm/setup.h>
+
+void __init trap_init(void)
+{
+ return;
+}
+
+void die(const char *str, struct pt_regs *regs, unsigned long address,
+ unsigned long cause_reg)
+{
+ show_kernel_fault_diag(str, regs, address, cause_reg);
+
+ /* DEAD END */
+ __asm__("flag 1");
+}
+
+/*
+ * Helper called for bulk of exceptions NOT needing specific handling
+ * -for user faults enqueues requested signal
+ * -for kernel, chk if due to copy_(to|from)_user, otherwise die()
+ */
+static noinline int handle_exception(unsigned long cause, char *str,
+ struct pt_regs *regs, siginfo_t *info)
+{
+ if (user_mode(regs)) {
+ struct task_struct *tsk = current;
+
+ tsk->thread.fault_address = (__force unsigned int)info->si_addr;
+ tsk->thread.cause_code = cause;
+
+ force_sig_info(info->si_signo, info, tsk);
+
+ } else {
+ /* If not due to copy_(to|from)_user, we are doomed */
+ if (fixup_exception(regs))
+ return 0;
+
+ die(str, regs, (unsigned long)info->si_addr, cause);
+ }
+
+ return 1;
+}
+
+#define DO_ERROR_INFO(signr, str, name, sicode) \
+int name(unsigned long cause, unsigned long address, struct pt_regs *regs) \
+{ \
+ siginfo_t info = { \
+ .si_signo = signr, \
+ .si_errno = 0, \
+ .si_code = sicode, \
+ .si_addr = (void __user *)address, \
+ }; \
+ return handle_exception(cause, str, regs, &info);\
+}
+
+/*
+ * Entry points for exceptions NOT needing specific handling
+ */
+DO_ERROR_INFO(SIGILL, "Priv Op/Disabled Extn", do_privilege_fault, ILL_PRVOPC)
+DO_ERROR_INFO(SIGILL, "Invalid Extn Insn", do_extension_fault, ILL_ILLOPC)
+DO_ERROR_INFO(SIGILL, "Illegal Insn (or Seq)", insterror_is_error, ILL_ILLOPC)
+DO_ERROR_INFO(SIGBUS, "Invalid Mem Access", do_memory_error, BUS_ADRERR)
+DO_ERROR_INFO(SIGTRAP, "Breakpoint Set", trap_is_brkpt, TRAP_BRKPT)
+
+DO_ERROR_INFO(SIGSEGV, "Misaligned Access", do_misaligned_access, SEGV_ACCERR)
+
+/*
+ * Entry point for miscll errors such as Nested Exceptions
+ * -Duplicate TLB entry is handled seperately though
+ */
+void do_machine_check_fault(unsigned long cause, unsigned long address,
+ struct pt_regs *regs)
+{
+ die("Machine Check Exception", regs, address, cause);
+}
+
+/*
+ * Entry point for traps induced by ARCompact TRAP_S <n> insn
+ * This is same family as TRAP0/SWI insn (use the same vector).
+ * The only difference being SWI insn take no operand, while TRAP_S does
+ * which reflects in ECR Reg as 8 bit param.
+ * Thus TRAP_S <n> can be used for specific purpose
+ * -1 used for software breakpointing (gdb)
+ * -2 used by kprobes
+ */
+void do_non_swi_trap(unsigned long cause, unsigned long address,
+ struct pt_regs *regs)
+{
+ unsigned int param = cause & 0xff;
+
+ switch (param) {
+ case 1:
+ trap_is_brkpt(cause, address, regs);
+ break;
+
+ default:
+ break;
+ }
+}
+
+/*
+ * Entry point for Instruction Error Exception
+ */
+void do_insterror_or_kprobe(unsigned long cause,
+ unsigned long address,
+ struct pt_regs *regs)
+{
+ insterror_is_error(cause, address, regs);
+}
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/sigcontext.h | 23 +++
arch/arc/include/asm/signal.h | 27 +++
arch/arc/kernel/signal.c | 360 +++++++++++++++++++++++++++++++++++++
3 files changed, 410 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/sigcontext.h
create mode 100644 arch/arc/include/asm/signal.h
create mode 100644 arch/arc/kernel/signal.c
diff --git a/arch/arc/include/asm/sigcontext.h b/arch/arc/include/asm/sigcontext.h
new file mode 100644
index 0000000..f21b541
--- /dev/null
+++ b/arch/arc/include/asm/sigcontext.h
@@ -0,0 +1,23 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_SIGCONTEXT_H
+#define _ASM_ARC_SIGCONTEXT_H
+
+#include <asm/ptrace.h>
+
+/*
+ * Signal context structure - contains all info to do with the state
+ * before the signal handler was invoked. Note: only add new entries
+ * to the end of the structure.
+ */
+struct sigcontext {
+ struct pt_regs regs;
+};
+
+#endif /* _ASM_ARC_SIGCONTEXT_H */
diff --git a/arch/arc/include/asm/signal.h b/arch/arc/include/asm/signal.h
new file mode 100644
index 0000000..fad62f7
--- /dev/null
+++ b/arch/arc/include/asm/signal.h
@@ -0,0 +1,27 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef _ASM_ARC_SIGNAL_H
+#define _ASM_ARC_SIGNAL_H
+
+/*
+ * This is much needed for ARC sigreturn optimization.
+ * This allows uClibc to piggback the addr of a sigreturn stub in sigaction,
+ * which allows sigreturn based re-entry into kernel after handling signal.
+ * W/o this kernel needs to "synthesize" the sigreturn trampoline on user
+ * mode stack which in turn forces the following:
+ * -TLB Flush (after making the stack page executable)
+ * -Cache line Flush (to make I/D Cache lines coherent)
+ */
+#define SA_RESTORER 0x04000000
+
+#include <asm-generic/signal.h>
+
+#endif /* _ASM_ARC_SIGNAL_H */
diff --git a/arch/arc/kernel/signal.c b/arch/arc/kernel/signal.c
new file mode 100644
index 0000000..bc834da
--- /dev/null
+++ b/arch/arc/kernel/signal.c
@@ -0,0 +1,360 @@
+/*
+ * Signal Handling for ARC
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: Jan 2010 (Restarting of timer related syscalls)
+ *
+ * vineetg: Nov 2009 (Everything needed for TIF_RESTORE_SIGMASK)
+ * -do_signal() supports TIF_RESTORE_SIGMASK
+ * -do_signal() no loner needs oldset, required by OLD sys_sigsuspend
+ * -sys_rt_sigsuspend() now comes from generic code, so discard arch implemen
+ * -sys_sigsuspend() no longer needs to fudge ptregs, hence that arg removed
+ * -sys_sigsuspend() no longer loops for do_signal(), sets TIF_xxx and leaves
+ * the job to do_signal()
+ *
+ * vineetg: July 2009
+ * -Modified Code to support the uClibc provided userland sigreturn stub
+ * to avoid kernel synthesing it on user stack at runtime, costing TLB
+ * probes and Cache line flushes.
+ *
+ * vineetg: July 2009
+ * -In stash_usr_regs( ) and restore_usr_regs( ), save/restore of user regs
+ * in done in block copy rather than one word at a time.
+ * This saves around 2K of code and improves LMBench lat_sig <catch>
+ *
+ * rajeshwarr: Feb 2009
+ * - Support for Realtime Signals
+ *
+ * vineetg: Aug 11th 2008: Bug #94183
+ * -ViXS were still seeing crashes when using insmod to load drivers.
+ * It turned out that the code to change Execute permssions for TLB entries
+ * of user was not guarded for interrupts (mod_tlb_permission)
+ * This was cauing TLB entries to be overwritten on unrelated indexes
+ *
+ * Vineetg: July 15th 2008: Bug #94183
+ * -Exception happens in Delay slot of a JMP, and before user space resumes,
+ * Signal is delivered (Ctrl + C) = >SIGINT.
+ * setup_frame( ) sets up PC,SP,BLINK to enable user space signal handler
+ * to run, but doesn't clear the Delay slot bit from status32. As a result,
+ * on resuming user mode, signal handler branches off to BTA of orig JMP
+ * -FIX: clear the DE bit from status32 in setup_frame( )
+ *
+ * Rahul Trivedi, Kanika Nema: Codito Technologies 2004
+ */
+
+#include <linux/signal.h>
+#include <linux/ptrace.h>
+#include <linux/personality.h>
+#include <linux/uaccess.h>
+#include <linux/syscalls.h>
+#include <linux/tracehook.h>
+#include <asm/ucontext.h>
+
+#define _BLOCKABLE (~(sigmask(SIGKILL) | sigmask(SIGSTOP)))
+
+asmlinkage int sys_sigaltstack(const stack_t __user *uss, stack_t __user *uoss)
+{
+ struct pt_regs *regs = task_pt_regs(current);
+ return do_sigaltstack(uss, uoss, regs->sp);
+}
+
+struct rt_sigframe {
+ struct siginfo info;
+ struct ucontext uc;
+#define MAGIC_SIGALTSTK 0x07302004
+ unsigned int sigret_magic;
+};
+
+static int
+stash_usr_regs(struct rt_sigframe __user *sf, struct pt_regs *regs,
+ sigset_t *set)
+{
+ int err;
+ err = __copy_to_user(&(sf->uc.uc_mcontext.regs), regs, sizeof(*regs));
+ err |= __copy_to_user(&sf->uc.uc_sigmask, set, sizeof(sigset_t));
+
+ return err;
+}
+
+static int restore_usr_regs(struct pt_regs *regs, struct rt_sigframe __user *sf)
+{
+ sigset_t set;
+ int err;
+
+ err = __copy_from_user(&set, &sf->uc.uc_sigmask, sizeof(set));
+ if (err == 0) {
+ sigdelsetmask(&set, ~_BLOCKABLE);
+ set_current_blocked(&set);
+ }
+
+ err |= __copy_from_user(regs, &(sf->uc.uc_mcontext.regs),
+ sizeof(*regs));
+
+ return err;
+}
+
+static inline int is_do_ss_needed(unsigned int magic)
+{
+ if (MAGIC_SIGALTSTK == magic)
+ return 1;
+ else
+ return 0;
+}
+
+SYSCALL_DEFINE0(rt_sigreturn)
+{
+ struct rt_sigframe __user *sf;
+ unsigned int magic;
+ int err;
+ struct pt_regs *regs = task_pt_regs(current);
+
+ /* Always make any pending restarted system calls return -EINTR */
+ current_thread_info()->restart_block.fn = do_no_restart_syscall;
+
+ /* Since we stacked the signal on a word boundary,
+ * then 'sp' should be word aligned here. If it's
+ * not, then the user is trying to mess with us.
+ */
+ if (regs->sp & 3)
+ goto badframe;
+
+ sf = (struct rt_sigframe __force __user *)(regs->sp);
+
+ if (!access_ok(VERIFY_READ, sf, sizeof(*sf)))
+ goto badframe;
+
+ err = restore_usr_regs(regs, sf);
+ err |= __get_user(magic, &sf->sigret_magic);
+ if (err)
+ goto badframe;
+
+ if (unlikely(is_do_ss_needed(magic)))
+ if (do_sigaltstack(&sf->uc.uc_stack, NULL, regs->sp) == -EFAULT)
+ goto badframe;
+
+ return regs->r0;
+
+badframe:
+ force_sig(SIGSEGV, current);
+ return 0;
+}
+
+/*
+ * Determine which stack to use..
+ */
+static inline void __user *get_sigframe(struct k_sigaction *ka,
+ struct pt_regs *regs,
+ unsigned long framesize)
+{
+ unsigned long sp = regs->sp;
+ void __user *frame;
+
+ /* This is the X/Open sanctioned signal stack switching */
+ if ((ka->sa.sa_flags & SA_ONSTACK) && !sas_ss_flags(sp))
+ sp = current->sas_ss_sp + current->sas_ss_size;
+
+ /* No matter what happens, 'sp' must be word
+ * aligned otherwise nasty things could happen
+ */
+
+ /* ATPCS B01 mandates 8-byte alignment */
+ frame = (void __user *)((sp - framesize) & ~7);
+
+ /* Check that we can actually write to the signal frame */
+ if (!access_ok(VERIFY_WRITE, frame, framesize))
+ frame = NULL;
+
+ return frame;
+}
+
+static int
+setup_rt_frame(int sig, struct k_sigaction *ka, siginfo_t *info,
+ sigset_t *set, struct pt_regs *regs)
+{
+ struct rt_sigframe __user *sf;
+ unsigned int magic = 0;
+ stack_t stk;
+ int err = 0;
+
+ sf = get_sigframe(ka, regs, sizeof(struct rt_sigframe));
+ if (!sf)
+ return 1;
+
+ /*
+ * SA_SIGINFO requires 3 args to signal handler:
+ * #1: sig-no (common to any handler)
+ * #2: struct siginfo
+ * #3: struct ucontext (completely populated)
+ */
+ if (unlikely(ka->sa.sa_flags & SA_SIGINFO)) {
+ err |= copy_siginfo_to_user(&sf->info, info);
+ err |= __put_user(0, &sf->uc.uc_flags);
+ err |= __put_user(NULL, &sf->uc.uc_link);
+ stk.ss_sp = (void __user *)current->sas_ss_sp;
+ stk.ss_flags = sas_ss_flags(regs->sp);
+ stk.ss_size = current->sas_ss_size;
+ err |= __copy_to_user(&sf->uc.uc_stack, &stk, sizeof(stk));
+
+ /* setup args 2 and 3 fo ruse rmode handler */
+ regs->r1 = (unsigned long)&sf->info;
+ regs->r2 = (unsigned long)&sf->uc;
+
+ /*
+ * small optim to avoid unconditonally calling do_sigaltstack
+ * in sigreturn path, now that we only have rt_sigreturn
+ */
+ magic = MAGIC_SIGALTSTK;
+ }
+
+ /*
+ * w/o SA_SIGINFO, struct ucontext is partially populated (only
+ * uc_mcontext/uc_sigmask) for kernel's normal user state preservation
+ * during signal handler execution. This works for SA_SIGINFO as well
+ * although the semantics are now overloaded (the same reg state can be
+ * inspected by userland: but are they allowed to fiddle with it ?
+ */
+ err |= stash_usr_regs(sf, regs, set);
+ err |= __put_user(magic, &sf->sigret_magic);
+ if (err)
+ return err;
+
+ /* #1 arg to the user Signal handler */
+ regs->r0 = sig;
+
+ /* setup PC of user space signal handler */
+ regs->ret = (unsigned long)ka->sa.sa_handler;
+
+ /*
+ * handler returns using sigreturn stub provided already by userpsace
+ */
+ BUG_ON(!(ka->sa.sa_flags & SA_RESTORER));
+ regs->blink = (unsigned long)ka->sa.sa_restorer;
+
+ /* User Stack for signal handler will be above the frame just carved */
+ regs->sp = (unsigned long)sf;
+
+ /*
+ * Bug 94183, Clear the DE bit, so that when signal handler
+ * starts to run, it doesn't use BTA
+ */
+ regs->status32 &= ~STATUS_DE_MASK;
+ regs->status32 |= STATUS_L_MASK;
+
+ return err;
+}
+
+/*
+ * OK, we're invoking a handler
+ */
+static void
+handle_signal(unsigned long sig, struct k_sigaction *ka, siginfo_t *info,
+ struct pt_regs *regs, int in_syscall)
+{
+ struct thread_info *thread = current_thread_info();
+ sigset_t *oldset = sigmask_to_save();
+ unsigned long usig = sig;
+ int ret;
+
+ /* Syscall restarting based on the ret code and other criteria */
+ if (in_syscall) {
+ switch (regs->r0) {
+ case -ERESTART_RESTARTBLOCK:
+ case -ERESTARTNOHAND:
+ /*
+ * ERESTARTNOHAND means that the syscall should
+ * only be restarted if there was no handler for
+ * the signal, and since we only get here if there
+ * is a handler, we don't restart
+ */
+ regs->r0 = -EINTR; /* ERESTART_xxx is internal */
+ break;
+
+ case -ERESTARTSYS:
+ /*
+ * ERESTARTSYS means to restart the syscall if
+ * there is no handler or the handler was
+ * registered with SA_RESTART
+ */
+ if (!(ka->sa.sa_flags & SA_RESTART)) {
+ regs->r0 = -EINTR;
+ break;
+ }
+ /* fallthrough */
+
+ case -ERESTARTNOINTR:
+ /*
+ * ERESTARTNOINTR means that the syscall should
+ * be called again after the signal handler returns.
+ * Setup reg state just as it was before doing the trap
+ * r0 has been clobbered with sys call ret code thus it
+ * needs to be reloaded with orig first arg to syscall
+ * in orig_r0. Rest of relevant reg-file:
+ * r8 (syscall num) and (r1 - r7) will be reset to
+ * their orig user space value when we ret from kernel
+ */
+ regs->r0 = regs->orig_r0;
+ regs->ret -= 4;
+ break;
+ }
+ }
+
+ if (thread->exec_domain && thread->exec_domain->signal_invmap
+ && usig < 32)
+ usig = thread->exec_domain->signal_invmap[usig];
+
+ /* Set up the stack frame */
+ ret = setup_rt_frame(usig, ka, info, oldset, regs);
+
+ if (ret)
+ force_sigsegv(sig, current);
+ else
+ signal_delivered(sig, info, ka, regs, 0);
+}
+
+void do_signal(struct pt_regs *regs)
+{
+ struct k_sigaction ka;
+ siginfo_t info;
+ int signr;
+ int insyscall;
+
+ signr = get_signal_to_deliver(&info, &ka, regs, NULL);
+
+ /* Are we from a system call? */
+ insyscall = in_syscall(regs);
+
+ if (signr > 0) {
+ handle_signal(signr, &ka, &info, regs, insyscall);
+ return;
+ }
+
+ if (insyscall) {
+ /* No handler for syscall: restart it */
+ if (regs->r0 == -ERESTARTNOHAND ||
+ regs->r0 == -ERESTARTSYS || regs->r0 == -ERESTARTNOINTR) {
+ regs->r0 = regs->orig_r0;
+ regs->ret -= 4;
+ } else if (regs->r0 == -ERESTART_RESTARTBLOCK) {
+ regs->r8 = __NR_restart_syscall;
+ regs->ret -= 4;
+ }
+ }
+
+ /* If there's no signal to deliver, restore the saved sigmask back */
+ restore_saved_sigmask();
+}
+
+void do_notify_resume(struct pt_regs *regs)
+{
+ /*
+ * ASM glue gaurantees that this is only called when returning to
+ * user mode
+ */
+ if (test_and_clear_thread_flag(TIF_NOTIFY_RESUME))
+ tracehook_notify_resume(regs);
+}
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/ptrace.h | 5 ++
arch/arc/include/asm/syscall.h | 72 +++++++++++++++++++++++
arch/arc/include/asm/syscalls.h | 30 ++++++++++
arch/arc/include/asm/unistd.h | 44 ++++++++++++++
arch/arc/kernel/entry.S | 49 ++++++++++++++++
arch/arc/kernel/process.c | 119 +++++++++++++++++++++++++++++++++++++++
arch/arc/kernel/sys.c | 19 ++++++
arch/arc/mm/mmap.c | 21 +++++++
8 files changed, 359 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/syscall.h
create mode 100644 arch/arc/include/asm/syscalls.h
create mode 100644 arch/arc/include/asm/unistd.h
create mode 100644 arch/arc/kernel/process.c
create mode 100644 arch/arc/kernel/sys.c
create mode 100644 arch/arc/mm/mmap.c
diff --git a/arch/arc/include/asm/ptrace.h b/arch/arc/include/asm/ptrace.h
index c178357..ee31d8a 100644
--- a/arch/arc/include/asm/ptrace.h
+++ b/arch/arc/include/asm/ptrace.h
@@ -115,6 +115,11 @@ struct user_regs_struct {
/* return 1 if user mode or 0 if kernel mode */
#define user_mode(regs) (regs->status32 & STATUS_U_MASK)
+
+/* return 1 if in syscall, 0 if Intr or Exception */
+#define in_syscall(regs) (((regs->orig_r8) >= 0 && \
+ (regs->orig_r8 <= NR_syscalls)) ? 1 : 0)
+
#endif /* __ASSEMBLY__ */
#endif /* __ASM_PTRACE_H */
diff --git a/arch/arc/include/asm/syscall.h b/arch/arc/include/asm/syscall.h
new file mode 100644
index 0000000..33ab304
--- /dev/null
+++ b/arch/arc/include/asm/syscall.h
@@ -0,0 +1,72 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_SYSCALL_H
+#define _ASM_ARC_SYSCALL_H 1
+
+#include <linux/err.h>
+#include <linux/sched.h>
+#include <asm/unistd.h>
+#include <asm/ptrace.h> /* in_syscall() */
+
+static inline long
+syscall_get_nr(struct task_struct *task, struct pt_regs *regs)
+{
+ if (user_mode(regs) && in_syscall(regs))
+ return regs->orig_r8;
+ else
+ return -1;
+}
+
+static inline void
+syscall_rollback(struct task_struct *task, struct pt_regs *regs)
+{
+ /* XXX: I can't fathom how pt_regs->r8 will be clobbered ? */
+ regs->r8 = regs->orig_r8;
+}
+
+static inline long
+syscall_get_error(struct task_struct *task, struct pt_regs *regs)
+{
+ /* 0 if syscall succeeded, otherwise -Errorcode */
+ return IS_ERR_VALUE(regs->r0) ? regs->r0 : 0;
+}
+
+static inline long
+syscall_get_return_value(struct task_struct *task, struct pt_regs *regs)
+{
+ return regs->r0;
+}
+
+static inline void
+syscall_set_return_value(struct task_struct *task, struct pt_regs *regs,
+ int error, long val)
+{
+ regs->r0 = (long) error ?: val;
+}
+
+/*
+ * @i: argument index [0,5]
+ * @n: number of arguments; n+i must be [1,6].
+ */
+static inline void
+syscall_get_arguments(struct task_struct *task, struct pt_regs *regs,
+ unsigned int i, unsigned int n, unsigned long *args)
+{
+ unsigned long *inside_ptregs = &(regs->r0);
+ inside_ptregs -= i;
+
+ BUG_ON((i + n) > 6);
+
+ while (n--) {
+ args[i++] = (*inside_ptregs);
+ inside_ptregs--;
+ }
+}
+
+#endif
diff --git a/arch/arc/include/asm/syscalls.h b/arch/arc/include/asm/syscalls.h
new file mode 100644
index 0000000..cf5d2f5
--- /dev/null
+++ b/arch/arc/include/asm/syscalls.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_SYSCALLS_H
+#define _ASM_ARC_SYSCALLS_H 1
+
+#ifdef __KERNEL__
+
+#include <linux/compiler.h>
+#include <linux/linkage.h>
+#include <linux/types.h>
+
+int sys_execve_wrapper(int, int, int);
+int sys_clone_wrapper(int, int, int, int, int);
+int sys_fork_wrapper(void);
+int sys_vfork_wrapper(void);
+int sys_cacheflush(uint32_t, uint32_t uint32_t);
+int sys_arc_settls(void *);
+int sys_arc_gettls(void);
+
+#include <asm-generic/syscalls.h>
+
+#endif /* __KERNEL__ */
+
+#endif
diff --git a/arch/arc/include/asm/unistd.h b/arch/arc/include/asm/unistd.h
new file mode 100644
index 0000000..754d588
--- /dev/null
+++ b/arch/arc/include/asm/unistd.h
@@ -0,0 +1,44 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+/*
+ * Being uClibc based we need some of the deprecated syscalls:
+ * -Not emulated by uClibc at all
+ * unlink, mkdir,... (needed by Busybox, LTP etc)
+ * times (needed by LTP pan test harness)
+ * -Not emulated efficiently
+ * select: emulated using pselect (but extra code to chk usec > 1sec)
+ *
+ * some (send/recv) correctly emulated using (recfrom/sendto) and
+ * some arch specific ones (fork/vfork)can easily be emulated using clone but
+ * thats the price of using common-denominator....
+ */
+#define __ARCH_WANT_SYSCALL_NO_AT
+#define __ARCH_WANT_SYSCALL_NO_FLAGS
+#define __ARCH_WANT_SYSCALL_OFF_T
+#define __ARCH_WANT_SYSCALL_DEPRECATED
+
+#define sys_mmap2 sys_mmap_pgoff
+
+#include <asm-generic/unistd.h>
+
+#define NR_syscalls __NR_syscalls
+
+/* ARC specific syscall */
+#define __NR_cacheflush (__NR_arch_specific_syscall + 0)
+#define __NR_arc_settls (__NR_arch_specific_syscall + 1)
+#define __NR_arc_gettls (__NR_arch_specific_syscall + 2)
+
+__SYSCALL(__NR_cacheflush, sys_cacheflush)
+__SYSCALL(__NR_arc_settls, sys_arc_settls)
+__SYSCALL(__NR_arc_gettls, sys_arc_gettls)
+
+
+/* Generic syscall (fs/filesystems.c - lost in asm-generic/unistd.h */
+#define __NR_sysfs (__NR_arch_specific_syscall + 3)
+__SYSCALL(__NR_sysfs, sys_sysfs)
diff --git a/arch/arc/kernel/entry.S b/arch/arc/kernel/entry.S
index b829052..fe9a17c 100644
--- a/arch/arc/kernel/entry.S
+++ b/arch/arc/kernel/entry.S
@@ -7,6 +7,10 @@
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
*
+ * TBD:
+ * -remove the low level wrappers for execve/fork/vfork
+ * needed before task_pt_regs was invented. This saves a branch per call
+ *
* vineetg: Nov 2010:
* -Vector table jumps (@8 bytes) converted into branches (@4 bytes)
* -To maintain the slot size of 8 bytes/vector, added nop, which is
@@ -575,3 +579,48 @@ ARC_ENTRY ret_from_fork
bl @schedule_tail
b @ret_from_exception
ARC_EXIT ret_from_fork
+
+;################### Special Sys Call Wrappers ##########################
+
+ARC_ENTRY sys_execve_wrapper
+ ; copy pointer to pt_regs as a parameter
+ mov r3, sp
+ bl @sys_execve
+
+ b ret_from_system_call
+ARC_EXIT sys_execve_wrapper
+
+; TBD: call do_fork directly from here
+ARC_ENTRY sys_fork_wrapper
+ ; copy pointer to pt_regs as a parameter
+ mov r0, sp
+ SAVE_CALLEE_SAVED_USER
+ bl @sys_fork
+ DISCARD_CALLEE_SAVED_USER
+
+ b ret_from_system_call
+ARC_EXIT sys_fork_wrapper
+
+ARC_ENTRY sys_vfork_wrapper
+ ; copy pointer to pt_regs as a parameter
+ mov r0, sp
+ SAVE_CALLEE_SAVED_USER
+ bl @sys_vfork
+ DISCARD_CALLEE_SAVED_USER
+
+ b ret_from_system_call
+ARC_EXIT sys_vfork_wrapper
+
+ARC_ENTRY sys_clone_wrapper
+ ; clone sys-call takes 2 mandatary args: @flags and @child-stack
+ ; and it can take 3 var-args, depending on flags
+ ; To keep sys_clone( ) signature constant, we assume all 5 args
+ ; and set the helper @prtegs in next free reg
+ ; this is cheap since our args are in regs, not on stack
+ mov r5, sp ; pt_regs
+ SAVE_CALLEE_SAVED_USER
+ bl @sys_clone
+ DISCARD_CALLEE_SAVED_USER
+
+ b ret_from_system_call
+ARC_EXIT sys_clone_wrapper
diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c
new file mode 100644
index 0000000..21d1889
--- /dev/null
+++ b/arch/arc/kernel/process.c
@@ -0,0 +1,119 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Amit Bhor, Kanika Nema: Codito Technologies 2004
+ */
+
+#include <linux/errno.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/fs.h>
+#include <linux/unistd.h>
+#include <linux/ptrace.h>
+#include <linux/slab.h>
+#include <linux/syscalls.h>
+#include <linux/elf.h>
+#include <linux/tick.h>
+
+asmlinkage int sys_fork(struct pt_regs *regs)
+{
+ return do_fork(SIGCHLD, regs->sp, regs, 0, NULL, NULL);
+}
+
+asmlinkage int sys_vfork(struct pt_regs *regs)
+{
+ return do_fork(CLONE_VFORK | CLONE_VM | SIGCHLD, regs->sp, regs, 0,
+ NULL, NULL);
+}
+
+/* Per man, C-lib clone( ) is as follows
+ *
+ * int clone(int (*fn)(void *), void *child_stack,
+ * int flags, void *arg, ...
+ * pid_t *ptid, struct user_desc *tls, pid_t *ctid);
+ *
+ * @fn and @arg are of userland thread-hnalder and thus of no use
+ * in sys-call, hence excluded in sys_clone arg list.
+ * The only addition is ptregs, needed by fork core, although now-a-days
+ * task_pt_regs() can be called anywhere to get that.
+ */
+asmlinkage int sys_clone(unsigned long clone_flags, unsigned long newsp,
+ int __user *parent_tidptr, void *tls,
+ int __user *child_tidptr, struct pt_regs *regs)
+{
+ if (!newsp)
+ newsp = regs->sp;
+
+ return do_fork(clone_flags, newsp, regs, 0, parent_tidptr,
+ child_tidptr);
+}
+
+int sys_execve(const char __user *filenamei, const char __user *__user *argv,
+ const char __user *__user *envp, struct pt_regs *regs)
+{
+ long error;
+ struct filename *filename;
+
+ filename = getname(filenamei);
+ error = PTR_ERR(filename);
+ if (IS_ERR(filename))
+ goto out;
+
+ error = do_execve(filename->name, argv, envp, regs);
+ putname(filename);
+out:
+ return error;
+}
+
+int kernel_execve(const char *filename, const char *const argv[],
+ const char *const envp[])
+{
+ /*
+ * Although the arguments (order, number) to this function are
+ * same as sys call, we don't need to setup args in regs again.
+ * However in case mainline kernel changes the order of args to
+ * kernel_execve, that assumtion will break.
+ * So to be safe, let gcc know the args for sys call.
+ * If they match no extra code will be generated
+ */
+ register int arg2 asm("r1") = (int)argv;
+ register int arg3 asm("r2") = (int)envp;
+
+ register int filenm_n_ret asm("r0") = (int)filename;
+
+ __asm__ __volatile__(
+ "mov r8, %1 \n\t"
+ "trap0 \n\t"
+ : "+r"(filenm_n_ret)
+ : "i"(__NR_execve), "r"(arg2), "r"(arg3)
+ : "r8", "memory");
+
+ return filenm_n_ret;
+}
+EXPORT_SYMBOL(kernel_execve);
+
+SYSCALL_DEFINE1(arc_settls, void *, user_tls_data_ptr)
+{
+ task_thread_info(current)->thr_ptr = (unsigned int)user_tls_data_ptr;
+ return 0;
+}
+
+/*
+ * We return the user space TLS data ptr as sys-call return code
+ * Ideally it should be copy to user.
+ * However we can cheat by the fact that some sys-calls do return
+ * absurdly high values
+ * Since the tls dat aptr is not going to be in range of 0xFFFF_xxxx
+ * it won't be considered a sys-call error
+ * and it will be loads better than copy-to-user, which is a definite
+ * D-TLB Miss
+ */
+SYSCALL_DEFINE0(arc_gettls)
+{
+ return task_thread_info(current)->thr_ptr;
+}
diff --git a/arch/arc/kernel/sys.c b/arch/arc/kernel/sys.c
new file mode 100644
index 0000000..4c30345
--- /dev/null
+++ b/arch/arc/kernel/sys.c
@@ -0,0 +1,19 @@
+
+#include <linux/syscalls.h>
+#include <linux/signal.h>
+#include <linux/unistd.h>
+
+#include <asm/syscalls.h>
+
+#define sys_execve sys_execve_wrapper
+#define sys_clone sys_clone_wrapper
+#define sys_fork sys_fork_wrapper
+#define sys_vfork sys_vfork_wrapper
+
+#undef __SYSCALL
+#define __SYSCALL(nr, call) [nr] = (call),
+
+void *sys_call_table[NR_syscalls] = {
+ [0 ... NR_syscalls-1] = sys_ni_syscall,
+#include <asm/unistd.h>
+};
diff --git a/arch/arc/mm/mmap.c b/arch/arc/mm/mmap.c
new file mode 100644
index 0000000..e1a9d61
--- /dev/null
+++ b/arch/arc/mm/mmap.c
@@ -0,0 +1,21 @@
+/* mmap for ARC
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/mman.h>
+#include <linux/mm.h>
+#include <linux/syscalls.h>
+
+/* Gets dragged in due to __ARCH_WANT_SYSCALL_OFF_T */
+SYSCALL_DEFINE6(mmap, unsigned long, addr_hint, unsigned long, len,
+ unsigned long, prot, unsigned long, flags, unsigned long, fd,
+ unsigned long, off)
+{
+ pr_err("old mmap not supported\n");
+ return -EINVAL;
+}
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/arcregs.h | 31 +++++
arch/arc/include/asm/delay.h | 68 +++++++++++
arch/arc/include/asm/processor.h | 3 +
arch/arc/include/asm/switch_to.h | 41 +++++++
arch/arc/include/asm/timex.h | 18 +++
arch/arc/kernel/ctx_sw.c | 91 +++++++++++++++
arch/arc/kernel/ctx_sw_asm.S | 58 +++++++++
arch/arc/kernel/fpu.c | 55 +++++++++
arch/arc/kernel/process.c | 201 ++++++++++++++++++++++++++++++++
arch/arc/kernel/time.c | 237 ++++++++++++++++++++++++++++++++++++++
10 files changed, 803 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/delay.h
create mode 100644 arch/arc/include/asm/switch_to.h
create mode 100644 arch/arc/include/asm/timex.h
create mode 100644 arch/arc/kernel/ctx_sw.c
create mode 100644 arch/arc/kernel/ctx_sw_asm.S
create mode 100644 arch/arc/kernel/fpu.c
create mode 100644 arch/arc/kernel/time.c
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
index 3fccb04..5131bb3 100644
--- a/arch/arc/include/asm/arcregs.h
+++ b/arch/arc/include/asm/arcregs.h
@@ -47,6 +47,28 @@
#define AUX_ITRIGGER 0x40d
#define AUX_IPULSE 0x415
+/* Timer related Aux registers */
+#define ARC_REG_TIMER0_LIMIT 0x23 /* timer 0 limit */
+#define ARC_REG_TIMER0_CTRL 0x22 /* timer 0 control */
+#define ARC_REG_TIMER0_CNT 0x21 /* timer 0 count */
+#define ARC_REG_TIMER1_LIMIT 0x102 /* timer 1 limit */
+#define ARC_REG_TIMER1_CTRL 0x101 /* timer 1 control */
+#define ARC_REG_TIMER1_CNT 0x100 /* timer 1 count */
+
+#define TIMER_CTRL_IE (1 << 0) /* Interupt when Count reachs limit */
+#define TIMER_CTRL_NH (1 << 1) /* Count only when CPU NOT halted */
+
+/*
+ * Floating Pt Registers
+ * Status regs are read-only (build-time) so need not be saved/restored
+ */
+#define ARC_AUX_FP_STAT 0x300
+#define ARC_AUX_DPFP_1L 0x301
+#define ARC_AUX_DPFP_1H 0x302
+#define ARC_AUX_DPFP_2L 0x303
+#define ARC_AUX_DPFP_2H 0x304
+#define ARC_AUX_DPFP_STAT 0x305
+
#ifndef __ASSEMBLY__
/*
@@ -110,6 +132,15 @@
#endif
+#ifdef CONFIG_ARC_FPU_SAVE_RESTORE
+/* These DPFP regs need to be saved/restored across ctx-sw */
+struct arc_fpu {
+ struct {
+ unsigned int l, h;
+ } aux_dpfp[2];
+};
+#endif
+
#endif /* __ASEMBLY__ */
#endif /* __KERNEL__ */
diff --git a/arch/arc/include/asm/delay.h b/arch/arc/include/asm/delay.h
new file mode 100644
index 0000000..442ce5d
--- /dev/null
+++ b/arch/arc/include/asm/delay.h
@@ -0,0 +1,68 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Delay routines using pre computed loops_per_jiffy value.
+ *
+ * vineetg: Feb 2012
+ * -Rewrote in "C" to avoid dealing with availability of H/w MPY
+ * -Also reduced the num of MPY operations from 3 to 2
+ *
+ * Amit Bhor: Codito Technologies 2004
+ */
+
+#ifndef __ASM_ARC_UDELAY_H
+#define __ASM_ARC_UDELAY_H
+
+#include <asm/param.h> /* HZ */
+
+static inline void __delay(unsigned long loops)
+{
+ __asm__ __volatile__(
+ "1: sub.f %0, %0, 1 \n"
+ " jpnz 1b \n"
+ : "+r"(loops)
+ :
+ : "cc");
+}
+
+extern void __bad_udelay(void);
+
+/*
+ * Normal Math for computing loops in "N" usecs
+ * -we have precomputed @loops_per_jiffy
+ * -1 sec has HZ jiffies
+ * loops per "N" usecs = ((loops_per_jiffy * HZ / 1000000) * N)
+ *
+ * Approximate Division by multiplication:
+ * -Mathematically if we multiply and divide a number by same value the
+ * result remains unchanged: In this case, we use 2^32
+ * -> (loops_per_N_usec * 2^32 ) / 2^32
+ * -> (((loops_per_jiffy * HZ / 1000000) * N) * 2^32) / 2^32
+ * -> (loops_per_jiffy * HZ * N * 4295) / 2^32
+ *
+ * -Divide by 2^32 is very simply right shift by 32
+ * -We simply need to ensure that the multiply per above eqn happens in
+ * 64-bit precision (if CPU doesn't support it - gcc can emaulate it)
+ */
+
+static inline void __udelay(unsigned long usecs)
+{
+ unsigned long loops;
+
+ /* (long long) cast ensures 64 bit MPY - real or emulated
+ * HZ * 4295 is pre-evaluated by gcc - hence only 2 mpy ops
+ */
+ loops = ((long long)(usecs * 4295 * HZ) *
+ (long long)(loops_per_jiffy)) >> 32;
+
+ __delay(loops);
+}
+
+#define udelay(n) (__builtin_constant_p(n) ? ((n) > 20000 ? __bad_udelay() \
+ : __udelay(n)) : __udelay(n))
+
+#endif /* __ASM_ARC_UDELAY_H */
diff --git a/arch/arc/include/asm/processor.h b/arch/arc/include/asm/processor.h
index e2445bc..38ea5fb 100644
--- a/arch/arc/include/asm/processor.h
+++ b/arch/arc/include/asm/processor.h
@@ -29,6 +29,9 @@ struct thread_struct {
unsigned long callee_reg; /* pointer to callee regs */
unsigned long fault_address; /* dbls as brkpt holder as well */
unsigned long cause_code; /* Exception Cause Code (ECR) */
+#ifdef CONFIG_ARC_FPU_SAVE_RESTORE
+ struct arc_fpu fpu;
+#endif
};
#define INIT_THREAD { \
diff --git a/arch/arc/include/asm/switch_to.h b/arch/arc/include/asm/switch_to.h
new file mode 100644
index 0000000..1b171ab
--- /dev/null
+++ b/arch/arc/include/asm/switch_to.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_SWITCH_TO_H
+#define _ASM_ARC_SWITCH_TO_H
+
+#ifndef __ASSEMBLY__
+
+#include <linux/sched.h>
+
+#ifdef CONFIG_ARC_FPU_SAVE_RESTORE
+
+extern void fpu_save_restore(struct task_struct *p, struct task_struct *n);
+#define ARC_FPU_PREV(p, n) fpu_save_restore(p, n)
+#define ARC_FPU_NEXT(t)
+
+#else
+
+#define ARC_FPU_PREV(p, n)
+#define ARC_FPU_NEXT(n)
+
+#endif /* !CONFIG_ARC_FPU_SAVE_RESTORE */
+
+struct task_struct *__switch_to(struct task_struct *p, struct task_struct *n);
+
+#define switch_to(prev, next, last) \
+do { \
+ ARC_FPU_PREV(prev, next); \
+ last = __switch_to(prev, next);\
+ ARC_FPU_NEXT(next); \
+ mb(); \
+} while (0)
+
+#endif
+
+#endif
diff --git a/arch/arc/include/asm/timex.h b/arch/arc/include/asm/timex.h
new file mode 100644
index 0000000..2c9bfb2
--- /dev/null
+++ b/arch/arc/include/asm/timex.h
@@ -0,0 +1,18 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_TIMEX_H
+#define _ASM_ARC_TIMEX_H
+
+#define CLOCK_TICK_RATE CONFIG_ARC_PLAT_CLK /* Underlying HZ */
+
+#include <asm-generic/timex.h>
+
+/* XXX: get_cycles() to be implemented with RTSC insn */
+
+#endif /* _ASM_ARC_TIMEX_H */
diff --git a/arch/arc/kernel/ctx_sw.c b/arch/arc/kernel/ctx_sw.c
new file mode 100644
index 0000000..647e37a
--- /dev/null
+++ b/arch/arc/kernel/ctx_sw.c
@@ -0,0 +1,91 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Vineetg: Aug 2009
+ * -"C" version of lowest level context switch asm macro called by schedular
+ * gcc doesn't generate the dward CFI info for hand written asm, hence can't
+ * backtrace out of it (e.g. tasks sleeping in kernel).
+ * So we cheat a bit by writing almost similar code in inline-asm.
+ * -This is a hacky way of doing things, but there is no other simple way.
+ * I don't want/intend to extend unwinding code to understand raw asm
+ */
+
+#include <asm/asm-offsets.h>
+#include <linux/sched.h>
+
+struct task_struct *__sched
+__switch_to(struct task_struct *prev_task, struct task_struct *next_task)
+{
+ unsigned int tmp;
+ unsigned int prev = (unsigned int)prev_task;
+ unsigned int next = (unsigned int)next_task;
+ int num_words_to_skip = 1;
+
+ __asm__ __volatile__(
+ /* FP/BLINK save generated by gcc (standard function prologue */
+ "st.a r13, [sp, -4] \n\t"
+ "st.a r14, [sp, -4] \n\t"
+ "st.a r15, [sp, -4] \n\t"
+ "st.a r16, [sp, -4] \n\t"
+ "st.a r17, [sp, -4] \n\t"
+ "st.a r18, [sp, -4] \n\t"
+ "st.a r19, [sp, -4] \n\t"
+ "st.a r20, [sp, -4] \n\t"
+ "st.a r21, [sp, -4] \n\t"
+ "st.a r22, [sp, -4] \n\t"
+ "st.a r23, [sp, -4] \n\t"
+ "st.a r24, [sp, -4] \n\t"
+ "st.a r25, [sp, -4] \n\t"
+ "sub sp, sp, %4 \n\t" /* create gutter at top */
+
+ /* set ksp of outgoing task in tsk->thread.ksp */
+ "st.as sp, [%3, %1] \n\t"
+
+ "sync \n\t"
+
+ /*
+ * setup _current_task with incoming tsk.
+ * optionally, set r25 to that as well
+ * For SMP extra work to get to &_current_task[cpu]
+ * (open coded SET_CURR_TASK_ON_CPU)
+ */
+ "st %2, [@_current_task] \n\t"
+
+ /* get ksp of incoming task from tsk->thread.ksp */
+ "ld.as sp, [%2, %1] \n\t"
+
+ /* start loading it's CALLEE reg file */
+
+ "add sp, sp, %4 \n\t" /* skip gutter at top */
+
+ "ld.ab r25, [sp, 4] \n\t"
+ "ld.ab r24, [sp, 4] \n\t"
+ "ld.ab r23, [sp, 4] \n\t"
+ "ld.ab r22, [sp, 4] \n\t"
+ "ld.ab r21, [sp, 4] \n\t"
+ "ld.ab r20, [sp, 4] \n\t"
+ "ld.ab r19, [sp, 4] \n\t"
+ "ld.ab r18, [sp, 4] \n\t"
+ "ld.ab r17, [sp, 4] \n\t"
+ "ld.ab r16, [sp, 4] \n\t"
+ "ld.ab r15, [sp, 4] \n\t"
+ "ld.ab r14, [sp, 4] \n\t"
+ "ld.ab r13, [sp, 4] \n\t"
+
+ /* last (ret value) = prev : although for ARC it mov r0, r0 */
+ "mov %0, %3 \n\t"
+
+ /* FP/BLINK restore generated by gcc (standard func epilogue */
+
+ : "=r"(tmp)
+ : "n"((TASK_THREAD + THREAD_KSP) / 4), "r"(next), "r"(prev),
+ "n"(num_words_to_skip * 4)
+ : "blink"
+ );
+
+ return (struct task_struct *)tmp;
+}
diff --git a/arch/arc/kernel/ctx_sw_asm.S b/arch/arc/kernel/ctx_sw_asm.S
new file mode 100644
index 0000000..d897234
--- /dev/null
+++ b/arch/arc/kernel/ctx_sw_asm.S
@@ -0,0 +1,58 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Vineetg: Aug 2009
+ * -Moved core context switch macro out of entry.S into this file.
+ * -This is the more "natural" hand written assembler
+ */
+
+#include <asm/entry.h> /* For the SAVE_* macros */
+#include <asm/asm-offsets.h>
+#include <asm/linkage.h>
+
+;################### Low Level Context Switch ##########################
+
+ .section .sched.text,"ax",@progbits
+ .align 4
+ .global __switch_to
+ .type __switch_to, @function
+__switch_to:
+
+ /* Save regs on kernel mode stack of task */
+ st.a blink, [sp, -4]
+ st.a fp, [sp, -4]
+ SAVE_CALLEE_SAVED_KERNEL
+
+ /* Save the now KSP in task->thread.ksp */
+ st.as sp, [r0, (TASK_THREAD + THREAD_KSP)/4]
+
+ /*
+ * Return last task in r0 (return reg)
+ * On ARC, Return reg = First Arg reg = r0.
+ * Since we already have last task in r0,
+ * don't need to do anything special to return it
+ */
+
+ /* hardware memory barrier */
+ sync
+
+ /*
+ * switch to new task, contained in r1
+ * Temp reg r3 is required to get the ptr to store val
+ */
+ SET_CURR_TASK_ON_CPU r1, r3
+
+ /* reload SP with kernel mode stack pointer in task->thread.ksp */
+ ld.as sp, [r1, (TASK_THREAD + THREAD_KSP)/4]
+
+ /* restore the registers */
+ RESTORE_CALLEE_SAVED_KERNEL
+ ld.ab fp, [sp, 4]
+ ld.ab blink, [sp, 4]
+ j [blink]
+
+ARC_EXIT __switch_to
diff --git a/arch/arc/kernel/fpu.c b/arch/arc/kernel/fpu.c
new file mode 100644
index 0000000..f352e51
--- /dev/null
+++ b/arch/arc/kernel/fpu.c
@@ -0,0 +1,55 @@
+/*
+ * fpu.c - save/restore of Floating Point Unit Registers on task switch
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/sched.h>
+#include <asm/switch_to.h>
+
+/*
+ * To save/restore FPU regs, simplest scheme would use LR/SR insns.
+ * However since SR serializes the pipeline, an alternate "hack" can be used
+ * which uses the FPU Exchange insn (DEXCL) to r/w FPU regs.
+ *
+ * Store to 64bit dpfp1 reg from a pair of core regs:
+ * dexcl1 0, r1, r0 ; where r1:r0 is the 64 bit val
+ *
+ * Read from dpfp1 into pair of core regs (w/o clobbering dpfp1)
+ * mov_s r3, 0
+ * daddh11 r1, r3, r3 ; get "hi" into r1 (dpfp1 unchanged)
+ * dexcl1 r0, r1, r3 ; get "low" into r0 (dpfp1 low clobbered)
+ * dexcl1 0, r1, r0 ; restore dpfp1 to orig value
+ *
+ * However we can tweak the read, so that read-out of outgoing task's FPU regs
+ * and write of incoming task's regs happen in one shot. So all the work is
+ * done before context switch
+ */
+
+void fpu_save_restore(struct task_struct *prev, struct task_struct *next)
+{
+ unsigned int *saveto = &prev->thread.fpu.aux_dpfp[0].l;
+ unsigned int *readfrom = &next->thread.fpu.aux_dpfp[0].l;
+
+ const unsigned int zero = 0;
+
+ __asm__ __volatile__(
+ "daddh11 %0, %2, %2\n"
+ "dexcl1 %1, %3, %4\n"
+ : "=&r" (*(saveto + 1)), /* early clobber must here */
+ "=&r" (*(saveto))
+ : "r" (zero), "r" (*(readfrom + 1)), "r" (*(readfrom))
+ );
+
+ __asm__ __volatile__(
+ "daddh22 %0, %2, %2\n"
+ "dexcl2 %1, %3, %4\n"
+ : "=&r"(*(saveto + 3)), /* early clobber must here */
+ "=&r"(*(saveto + 2))
+ : "r" (zero), "r" (*(readfrom + 3)), "r" (*(readfrom + 2))
+ );
+}
diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c
index 21d1889..c116fa5 100644
--- a/arch/arc/kernel/process.c
+++ b/arch/arc/kernel/process.c
@@ -117,3 +117,204 @@ SYSCALL_DEFINE0(arc_gettls)
{
return task_thread_info(current)->thr_ptr;
}
+
+static inline void arch_idle(void)
+{
+ __asm__("sleep");
+}
+
+void cpu_idle(void)
+{
+ /* Since we SLEEP in idle loop, TIF_POLLING_NRFLAG can't be set */
+
+ /* endless idle loop with no priority at all */
+ while (1) {
+ tick_nohz_idle_enter();
+
+ while (!need_resched())
+ arch_idle();
+
+ tick_nohz_idle_exit();
+
+ preempt_enable_no_resched();
+ schedule();
+ preempt_disable();
+ }
+}
+
+void kernel_thread_helper(void)
+{
+ __asm__ __volatile__(
+ "mov r0, r2 \n\t"
+ "mov r1, r3 \n\t"
+ "j [r1] \n\t");
+}
+
+int kernel_thread(int (*fn) (void *), void *arg, unsigned long flags)
+{
+ struct pt_regs regs;
+
+ memset(®s, 0, sizeof(regs));
+
+ regs.r2 = (unsigned long)arg;
+ regs.r3 = (unsigned long)fn;
+ regs.blink = (unsigned long)do_exit;
+ regs.ret = (unsigned long)kernel_thread_helper;
+ regs.status32 = read_aux_reg(0xa);
+
+ /* Ok, create the new process.. */
+ return do_fork(flags | CLONE_VM | CLONE_UNTRACED, 0, ®s, 0, NULL,
+ NULL);
+
+}
+EXPORT_SYMBOL(kernel_thread);
+
+asmlinkage void ret_from_fork(void);
+
+/* Layout of Child kernel mode stack as setup at the end of this function is
+ *
+ * | ... |
+ * | ... |
+ * | unused |
+ * | |
+ * ------------------ <==== top of Stack (thread.ksp)
+ * | UNUSED 1 word|
+ * ------------------
+ * | r25 |
+ * ~ ~
+ * | --to-- | (CALLEE Regs of user mode)
+ * | r13 |
+ * ------------------
+ * | fp |
+ * | blink | @ret_from_fork
+ * ------------------
+ * | |
+ * ~ ~
+ * ~ ~
+ * | |
+ * ------------------
+ * | r12 |
+ * ~ ~
+ * | --to-- | (scratch Regs of user mode)
+ * | r0 |
+ * ------------------
+ * | UNUSED 1 word|
+ * ------------------ <===== END of PAGE
+ */
+int copy_thread(unsigned long clone_flags,
+ unsigned long usp, unsigned long topstk,
+ struct task_struct *p, struct pt_regs *regs)
+{
+ struct pt_regs *c_regs; /* child's pt_regs */
+ unsigned long *childksp; /* to unwind out of __switch_to() */
+ struct callee_regs *c_callee; /* child's callee regs */
+ struct callee_regs *parent_callee; /* paren't callee */
+
+ /* Mark the specific anchors to begin with (see pic above) */
+ c_regs = task_pt_regs(p);
+ childksp = (unsigned long *)c_regs - 2; /* 2 words for FP/BLINK */
+ c_callee = ((struct callee_regs *)childksp) - 1;
+
+ /*
+ * At the end of this function, kernel SP is all set for
+ * switch_to to start unwinding.
+ * For kernel threads we don't have callee regs, but the stack
+ * layout nevertheless needs to remain the same
+ */
+ p->thread.ksp = (unsigned long)c_callee; /* THREAD_KSP */
+
+ /* Copy parents pt regs on child's kernel mode stack */
+ *c_regs = *regs;
+
+ /* __switch_to expects FP(0), BLINK(return addr) at top of stack */
+ childksp[0] = 0; /* for POP fp */
+ childksp[1] = (unsigned long)ret_from_fork; /* for POP blink */
+
+ if (!(user_mode(regs))) {
+ c_regs->sp =
+ (unsigned long)task_thread_info(p) + (THREAD_SIZE - 4);
+ return 0;
+ }
+
+ /*--------- User Task Only --------------*/
+
+ c_regs->sp = usp;
+ c_regs->r0 = 0; /* fork returns 0 in child */
+
+ parent_callee = ((struct callee_regs *)regs) - 1;
+ *c_callee = *parent_callee;
+
+ if (unlikely(clone_flags & CLONE_SETTLS)) {
+ /*
+ * set task's userland tls data ptr from 4th arg
+ * clone C-lib call is difft from clone sys-call
+ */
+ task_thread_info(p)->thr_ptr = regs->r3;
+ } else {
+ /* Normal fork case: set parent's TLS ptr in child */
+ task_thread_info(p)->thr_ptr =
+ task_thread_info(current)->thr_ptr;
+ }
+
+ return 0;
+}
+
+/*
+ * Some archs flush debug and FPU info here
+ */
+void flush_thread(void)
+{
+}
+
+/*
+ * Free any architecture-specific thread data structures, etc.
+ */
+void exit_thread(void)
+{
+}
+
+int dump_fpu(struct pt_regs *regs, elf_fpregset_t *fpu)
+{
+ return 0;
+}
+
+/*
+ * API: expected by schedular Code: If thread is sleeping where is that.
+ * What is this good for? it will be always the scheduler or ret_from_fork.
+ * So we hard code that anyways.
+ */
+unsigned long thread_saved_pc(struct task_struct *t)
+{
+ struct pt_regs *regs = task_pt_regs(t);
+ unsigned long blink = 0;
+
+ /*
+ * If the thread being queried for in not itself calling this, then it
+ * implies it is not executing, which in turn implies it is sleeping,
+ * which in turn implies it got switched OUT by the schedular.
+ * In that case, it's kernel mode blink can reliably retrieved as per
+ * the picture above (right above pt_regs).
+ */
+ if (t != current && t->state != TASK_RUNNING)
+ blink = *((unsigned int *)regs - 1);
+
+ return blink;
+}
+
+int elf_check_arch(const struct elf32_hdr *x)
+{
+ unsigned int eflags;
+
+ if (x->e_machine != EM_ARCOMPACT)
+ return 0;
+
+ eflags = x->e_flags;
+ if ((eflags & EF_ARC_OSABI_MSK) < EF_ARC_OSABI_V2) {
+ pr_err("ABI mismatch - you need newer toolchain\n");
+ force_sigsegv(SIGSEGV, current);
+ return 0;
+ }
+
+ return 1;
+}
+EXPORT_SYMBOL(elf_check_arch);
diff --git a/arch/arc/kernel/time.c b/arch/arc/kernel/time.c
new file mode 100644
index 0000000..a0aa884
--- /dev/null
+++ b/arch/arc/kernel/time.c
@@ -0,0 +1,237 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: Jan 1011
+ * -sched_clock( ) no longer jiffies based. Uses the same clocksource
+ * as gtod
+ *
+ * Rajeshwarr/Vineetg: Mar 2008
+ * -Implemented CONFIG_GENERIC_TIME (rather deleted arch specific code)
+ * for arch independent gettimeofday()
+ * -Implemented CONFIG_GENERIC_CLOCKEVENTS as base for hrtimers
+ *
+ * Vineetg: Mar 2008: Forked off from time.c which now is time-jiff.c
+ */
+
+#include <linux/spinlock.h>
+#include <linux/interrupt.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/kernel.h>
+#include <linux/interrupt.h>
+#include <linux/time.h>
+#include <linux/init.h>
+#include <linux/timex.h>
+#include <linux/profile.h>
+#include <linux/clocksource.h>
+#include <linux/clockchips.h>
+#include <asm/irq.h>
+#include <asm/arcregs.h>
+
+/* ARC700 has two 32bit independent prog Timers: TIMER0 and TIMER1
+ * Each can programmed to go from @count to @limit and optionally
+ * interrupt when that happens
+ *
+ * We've designated TIMER0 for events (clockevents)
+ * while TIMER1 for free running (clocksource)
+ */
+#define ARC_TIMER_MAX 0xFFFFFFFF
+
+/******************************************************************
+ * Hardware Interface routines to program the ARC TIMERs
+ ******************************************************************/
+
+/*
+ * Arm the timer to interrupt after @limit cycles
+ */
+static void arc_periodic_timer_setup(unsigned int limit)
+{
+ /* setup start and end markers */
+ write_aux_reg(ARC_REG_TIMER0_LIMIT, limit);
+ write_aux_reg(ARC_REG_TIMER0_CNT, 0); /* start from 0 */
+
+ /* IE: Interrupt on count = limit,
+ * NH: Count cycles only when CPU running (NOT Halted)
+ */
+ write_aux_reg(ARC_REG_TIMER0_CTRL, TIMER_CTRL_IE | TIMER_CTRL_NH);
+}
+
+/*
+ * Acknowledge the interrupt & enable/disable the interrupt
+ */
+static void arc_periodic_timer_ack(unsigned int irq_reenable)
+{
+ /* 1. Ack the interrupt by writing to CTRL reg.
+ * Any write will cause intr to be ack, however it has to be one of
+ * writable bits (NH: Count when not halted)
+ * 2. If required by caller, re-arm timer to Interrupt at the end of
+ * next cycle.
+ *
+ * Small optimisation:
+ * Normal code would have been
+ * if (irq_reenable) CTRL_REG = (IE | NH); else CTRL_REG = NH;
+ * However since IE is BIT0 we can fold the branch
+ */
+ write_aux_reg(ARC_REG_TIMER0_CTRL, irq_reenable | TIMER_CTRL_NH);
+}
+
+/*
+ * Arm the timer to keep counting monotonically
+ */
+void __cpuinit arc_clock_counter_setup(void)
+{
+ unsigned int limit = ARC_TIMER_MAX;
+
+ /* although for free flowing case, limit would alway be max 32 bits
+ * still we've kept the interface open, just in case ...
+ */
+ write_aux_reg(ARC_REG_TIMER1_LIMIT, limit);
+ write_aux_reg(ARC_REG_TIMER1_CNT, 0);
+ write_aux_reg(ARC_REG_TIMER1_CTRL, TIMER_CTRL_NH);
+}
+
+/********** Clock Source Device *********/
+
+static cycle_t cycle_read_t1(struct clocksource *cs)
+{
+ return (cycle_t) read_aux_reg(ARC_REG_TIMER1_CNT);
+}
+
+static struct clocksource clocksource_t1 = {
+ .name = "ARC Timer1",
+ .rating = 300,
+ .read = cycle_read_t1,
+ .mask = CLOCKSOURCE_MASK(32),
+ .flags = CLOCK_SOURCE_IS_CONTINUOUS,
+};
+
+void __cpuinit arc_clocksource_init(void)
+{
+ arc_clock_counter_setup();
+
+ /*
+ * CLK upto 4.29 GHz can be safely represented in 32 bits because
+ * Max 32 bit number is 4,294,967,295
+ */
+ clocksource_register_hz(&clocksource_t1, CONFIG_ARC_PLAT_CLK);
+}
+
+/********** Clock Event Device *********/
+
+static int arc_clkevent_set_next_event(unsigned long delta,
+ struct clock_event_device *dev)
+{
+ arc_periodic_timer_setup(delta);
+ return 0;
+}
+
+static void arc_clkevent_set_mode(enum clock_event_mode mode,
+ struct clock_event_device *dev)
+{
+ pr_info("Device [%s] clockevent mode now [%d]\n", dev->name, mode);
+ switch (mode) {
+ case CLOCK_EVT_MODE_PERIODIC:
+ arc_periodic_timer_setup(CONFIG_ARC_PLAT_CLK / HZ);
+ break;
+ case CLOCK_EVT_MODE_ONESHOT:
+ break;
+ default:
+ break;
+ }
+
+ return;
+}
+
+static DEFINE_PER_CPU(struct clock_event_device, arc_clockevent_device) = {
+ .name = "ARC Timer0",
+ .features = CLOCK_EVT_FEAT_ONESHOT | CLOCK_EVT_FEAT_PERIODIC,
+ .mode = CLOCK_EVT_MODE_UNUSED,
+ .rating = 300,
+ .irq = TIMER0_IRQ, /* hardwired, no need for resources */
+ .set_next_event = arc_clkevent_set_next_event,
+ .set_mode = arc_clkevent_set_mode,
+};
+
+irqreturn_t timer_irq_handler(int irq, void *dev_id)
+{
+ unsigned int cpu = smp_processor_id();
+ struct clock_event_device *evt = &per_cpu(arc_clockevent_device, cpu);
+
+ arc_periodic_timer_ack(evt->mode == CLOCK_EVT_MODE_PERIODIC);
+ evt->event_handler(evt);
+ return IRQ_HANDLED;
+}
+
+void __cpuinit arc_clockevent_init(void)
+{
+ int rc;
+ unsigned int cpu = smp_processor_id();
+ struct clock_event_device *evt = &per_cpu(arc_clockevent_device, cpu);
+
+ clockevents_calc_mult_shift(evt, CONFIG_ARC_PLAT_CLK, 5);
+
+ evt->max_delta_ns = clockevent_delta2ns(ARC_TIMER_MAX, evt);
+ evt->cpumask = cpumask_of(cpu);
+
+ clockevents_register_device(evt);
+
+ /*
+ * Done only on Boot CPU as it would fail on others.
+ * (Treated a re-registration for a !IRQF_SHARED irq)
+ */
+ if (cpu == 0) {
+ rc = request_percpu_irq(TIMER0_IRQ, timer_irq_handler,
+ "Timer0 (clock-evt-dev)", evt);
+
+ if (rc)
+ panic("TIMER0 IRQ reg failed on BOOT cpu\n");
+ }
+
+ /* This is done on each CPU */
+ enable_percpu_irq(TIMER0_IRQ, 0);
+
+}
+
+void __init time_init(void)
+{
+ arc_clocksource_init();
+ arc_clockevent_init();
+}
+
+static int arc_finished_booting;
+
+/*
+ * Scheduler clock - returns current time in nanosec units.
+ * It's return value must NOT wrap around.
+ *
+ * Although the return value is nanosec units based, what's more important
+ * is whats the "source" of this value. The orig jiffies based computation
+ * was only as granular as jiffies itself (10ms on ARC).
+ * We need something that is more granular, so use the same mechanism as
+ * gettimeofday(), which uses ARC Timer T1 wrapped as a clocksource.
+ * Unfortunately the first call to sched_clock( ) is way before that subsys
+ * is initialiased, thus use the jiffies based value in the interim.
+ */
+unsigned long long sched_clock(void)
+{
+ if (!arc_finished_booting) {
+ return (unsigned long long)(jiffies - INITIAL_JIFFIES)
+ * (NSEC_PER_SEC / HZ);
+ } else {
+ struct timespec ts;
+ getrawmonotonic(&ts);
+ return (unsigned long long)timespec_to_ns(&ts);
+ }
+}
+
+static int __init arc_clocksource_done_booting(void)
+{
+ arc_finished_booting = 1;
+ return 0;
+}
+
+fs_initcall(arc_clocksource_done_booting);
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/string.h | 40 +++++++++++++
arch/arc/lib/memcmp.S | 124 +++++++++++++++++++++++++++++++++++++++++
arch/arc/lib/memcpy-700.S | 66 ++++++++++++++++++++++
arch/arc/lib/memset.S | 59 +++++++++++++++++++
arch/arc/lib/strchr-700.S | 123 ++++++++++++++++++++++++++++++++++++++++
arch/arc/lib/strcmp.S | 96 +++++++++++++++++++++++++++++++
arch/arc/lib/strcpy-700.S | 70 +++++++++++++++++++++++
arch/arc/lib/strlen.S | 83 +++++++++++++++++++++++++++
8 files changed, 661 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/string.h
create mode 100644 arch/arc/lib/memcmp.S
create mode 100644 arch/arc/lib/memcpy-700.S
create mode 100644 arch/arc/lib/memset.S
create mode 100644 arch/arc/lib/strchr-700.S
create mode 100644 arch/arc/lib/strcmp.S
create mode 100644 arch/arc/lib/strcpy-700.S
create mode 100644 arch/arc/lib/strlen.S
diff --git a/arch/arc/include/asm/string.h b/arch/arc/include/asm/string.h
new file mode 100644
index 0000000..87676c8
--- /dev/null
+++ b/arch/arc/include/asm/string.h
@@ -0,0 +1,40 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: May 2011
+ * -We had half-optimised memset/memcpy, got better versions of those
+ * -Added memcmp, strchr, strcpy, strcmp, strlen
+ *
+ * Amit Bhor: Codito Technologies 2004
+ */
+
+#ifndef _ASM_ARC_STRING_H
+#define _ASM_ARC_STRING_H
+
+#include <linux/types.h>
+
+#ifdef __KERNEL__
+
+#define __HAVE_ARCH_MEMSET
+#define __HAVE_ARCH_MEMCPY
+#define __HAVE_ARCH_MEMCMP
+#define __HAVE_ARCH_STRCHR
+#define __HAVE_ARCH_STRCPY
+#define __HAVE_ARCH_STRCMP
+#define __HAVE_ARCH_STRLEN
+
+extern void *memset(void *ptr, int, __kernel_size_t);
+extern void *memcpy(void *, const void *, __kernel_size_t);
+extern void memzero(void *ptr, __kernel_size_t n);
+extern int memcmp(const void *, const void *, __kernel_size_t);
+extern char *strchr(const char *s, int c);
+extern char *strcpy(char *dest, const char *src);
+extern int strcmp(const char *cs, const char *ct);
+extern __kernel_size_t strlen(const char *);
+
+#endif /* __KERNEL__ */
+#endif /* _ASM_ARC_STRING_H */
diff --git a/arch/arc/lib/memcmp.S b/arch/arc/lib/memcmp.S
new file mode 100644
index 0000000..bc813d5
--- /dev/null
+++ b/arch/arc/lib/memcmp.S
@@ -0,0 +1,124 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <asm/linkage.h>
+
+#ifdef __LITTLE_ENDIAN__
+#define WORD2 r2
+#define SHIFT r3
+#else /* BIG ENDIAN */
+#define WORD2 r3
+#define SHIFT r2
+#endif
+
+ARC_ENTRY memcmp
+ or r12,r0,r1
+ asl_s r12,r12,30
+ sub r3,r2,1
+ brls r2,r12,.Lbytewise
+ ld r4,[r0,0]
+ ld r5,[r1,0]
+ lsr.f lp_count,r3,3
+ lpne .Loop_end
+ ld_s WORD2,[r0,4]
+ ld_s r12,[r1,4]
+ brne r4,r5,.Leven
+ ld.a r4,[r0,8]
+ ld.a r5,[r1,8]
+ brne WORD2,r12,.Lodd
+.Loop_end:
+ asl_s SHIFT,SHIFT,3
+ bhs_s .Last_cmp
+ brne r4,r5,.Leven
+ ld r4,[r0,4]
+ ld r5,[r1,4]
+#ifdef __LITTLE_ENDIAN__
+ nop_s
+ ; one more load latency cycle
+.Last_cmp:
+ xor r0,r4,r5
+ bset r0,r0,SHIFT
+ sub_s r1,r0,1
+ bic_s r1,r1,r0
+ norm r1,r1
+ b.d .Leven_cmp
+ and r1,r1,24
+.Leven:
+ xor r0,r4,r5
+ sub_s r1,r0,1
+ bic_s r1,r1,r0
+ norm r1,r1
+ ; slow track insn
+ and r1,r1,24
+.Leven_cmp:
+ asl r2,r4,r1
+ asl r12,r5,r1
+ lsr_s r2,r2,1
+ lsr_s r12,r12,1
+ j_s.d [blink]
+ sub r0,r2,r12
+ .balign 4
+.Lodd:
+ xor r0,WORD2,r12
+ sub_s r1,r0,1
+ bic_s r1,r1,r0
+ norm r1,r1
+ ; slow track insn
+ and r1,r1,24
+ asl_s r2,r2,r1
+ asl_s r12,r12,r1
+ lsr_s r2,r2,1
+ lsr_s r12,r12,1
+ j_s.d [blink]
+ sub r0,r2,r12
+#else /* BIG ENDIAN */
+.Last_cmp:
+ neg_s SHIFT,SHIFT
+ lsr r4,r4,SHIFT
+ lsr r5,r5,SHIFT
+ ; slow track insn
+.Leven:
+ sub.f r0,r4,r5
+ mov.ne r0,1
+ j_s.d [blink]
+ bset.cs r0,r0,31
+.Lodd:
+ cmp_s WORD2,r12
+
+ mov_s r0,1
+ j_s.d [blink]
+ bset.cs r0,r0,31
+#endif /* ENDIAN */
+ .balign 4
+.Lbytewise:
+ breq r2,0,.Lnil
+ ldb r4,[r0,0]
+ ldb r5,[r1,0]
+ lsr.f lp_count,r3
+ lpne .Lbyte_end
+ ldb_s r3,[r0,1]
+ ldb r12,[r1,1]
+ brne r4,r5,.Lbyte_even
+ ldb.a r4,[r0,2]
+ ldb.a r5,[r1,2]
+ brne r3,r12,.Lbyte_odd
+.Lbyte_end:
+ bcc .Lbyte_even
+ brne r4,r5,.Lbyte_even
+ ldb_s r3,[r0,1]
+ ldb_s r12,[r1,1]
+.Lbyte_odd:
+ j_s.d [blink]
+ sub r0,r3,r12
+.Lbyte_even:
+ j_s.d [blink]
+ sub r0,r4,r5
+.Lnil:
+ j_s.d [blink]
+ mov r0,0
+ARC_EXIT memcmp
diff --git a/arch/arc/lib/memcpy-700.S b/arch/arc/lib/memcpy-700.S
new file mode 100644
index 0000000..b64cc10
--- /dev/null
+++ b/arch/arc/lib/memcpy-700.S
@@ -0,0 +1,66 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <asm/linkage.h>
+
+ARC_ENTRY memcpy
+ or r3,r0,r1
+ asl_s r3,r3,30
+ mov_s r5,r0
+ brls.d r2,r3,.Lcopy_bytewise
+ sub.f r3,r2,1
+ ld_s r12,[r1,0]
+ asr.f lp_count,r3,3
+ bbit0.d r3,2,.Lnox4
+ bmsk_s r2,r2,1
+ st.ab r12,[r5,4]
+ ld.a r12,[r1,4]
+.Lnox4:
+ lppnz .Lendloop
+ ld_s r3,[r1,4]
+ st.ab r12,[r5,4]
+ ld.a r12,[r1,8]
+ st.ab r3,[r5,4]
+.Lendloop:
+ breq r2,0,.Last_store
+ ld r3,[r5,0]
+#ifdef __LITTLE_ENDIAN__
+ add3 r2,-1,r2
+ ; uses long immediate
+ xor_s r12,r12,r3
+ bmsk r12,r12,r2
+ xor_s r12,r12,r3
+#else /* BIG ENDIAN */
+ sub3 r2,31,r2
+ ; uses long immediate
+ xor_s r3,r3,r12
+ bmsk r3,r3,r2
+ xor_s r12,r12,r3
+#endif /* ENDIAN */
+.Last_store:
+ j_s.d [blink]
+ st r12,[r5,0]
+
+ .balign 4
+.Lcopy_bytewise:
+ jcs [blink]
+ ldb_s r12,[r1,0]
+ lsr.f lp_count,r3
+ bhs_s .Lnox1
+ stb.ab r12,[r5,1]
+ ldb.a r12,[r1,1]
+.Lnox1:
+ lppnz .Lendbloop
+ ldb_s r3,[r1,1]
+ stb.ab r12,[r5,1]
+ ldb.a r12,[r1,2]
+ stb.ab r3,[r5,1]
+.Lendbloop:
+ j_s.d [blink]
+ stb r12,[r5,0]
+ARC_EXIT memcpy
diff --git a/arch/arc/lib/memset.S b/arch/arc/lib/memset.S
new file mode 100644
index 0000000..9b2d88d
--- /dev/null
+++ b/arch/arc/lib/memset.S
@@ -0,0 +1,59 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <asm/linkage.h>
+
+#define SMALL 7 /* Must be at least 6 to deal with alignment/loop issues. */
+
+ARC_ENTRY memset
+ mov_s r4,r0
+ or r12,r0,r2
+ bmsk.f r12,r12,1
+ extb_s r1,r1
+ asl r3,r1,8
+ beq.d .Laligned
+ or_s r1,r1,r3
+ brls r2,SMALL,.Ltiny
+ add r3,r2,r0
+ stb r1,[r3,-1]
+ bclr_s r3,r3,0
+ stw r1,[r3,-2]
+ bmsk.f r12,r0,1
+ add_s r2,r2,r12
+ sub.ne r2,r2,4
+ stb.ab r1,[r4,1]
+ and r4,r4,-2
+ stw.ab r1,[r4,2]
+ and r4,r4,-4
+.Laligned: ; This code address should be aligned for speed.
+ asl r3,r1,16
+ lsr.f lp_count,r2,2
+ or_s r1,r1,r3
+ lpne .Loop_end
+ st.ab r1,[r4,4]
+.Loop_end:
+ j_s [blink]
+
+ .balign 4
+.Ltiny:
+ mov.f lp_count,r2
+ lpne .Ltiny_end
+ stb.ab r1,[r4,1]
+.Ltiny_end:
+ j_s [blink]
+ARC_EXIT memset
+
+; memzero: @r0 = mem, @r1 = size_t
+; memset: @r0 = mem, @r1 = char, @r2 = size_t
+
+ARC_ENTRY memzero
+ ; adjust bzero args to memset args
+ mov r2, r1
+ mov r1, 0
+ b memset ;tail call so need to tinker with blink
+ARC_EXIT memzero
diff --git a/arch/arc/lib/strchr-700.S b/arch/arc/lib/strchr-700.S
new file mode 100644
index 0000000..99c1047
--- /dev/null
+++ b/arch/arc/lib/strchr-700.S
@@ -0,0 +1,123 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+/* ARC700 has a relatively long pipeline and branch prediction, so we want
+ to avoid branches that are hard to predict. On the other hand, the
+ presence of the norm instruction makes it easier to operate on whole
+ words branch-free. */
+
+#include <asm/linkage.h>
+
+ARC_ENTRY strchr
+ extb_s r1,r1
+ asl r5,r1,8
+ bmsk r2,r0,1
+ or r5,r5,r1
+ mov_s r3,0x01010101
+ breq.d r2,r0,.Laligned
+ asl r4,r5,16
+ sub_s r0,r0,r2
+ asl r7,r2,3
+ ld_s r2,[r0]
+#ifdef __LITTLE_ENDIAN__
+ asl r7,r3,r7
+#else
+ lsr r7,r3,r7
+#endif
+ or r5,r5,r4
+ ror r4,r3
+ sub r12,r2,r7
+ bic_s r12,r12,r2
+ and r12,r12,r4
+ brne.d r12,0,.Lfound0_ua
+ xor r6,r2,r5
+ ld.a r2,[r0,4]
+ sub r12,r6,r7
+ bic r12,r12,r6
+ and r7,r12,r4
+ breq r7,0,.Loop ; For speed, we want this branch to be unaligned.
+ b .Lfound_char ; Likewise this one.
+; /* We require this code address to be unaligned for speed... */
+.Laligned:
+ ld_s r2,[r0]
+ or r5,r5,r4
+ ror r4,r3
+; /* ... so that this code address is aligned, for itself and ... */
+.Loop:
+ sub r12,r2,r3
+ bic_s r12,r12,r2
+ and r12,r12,r4
+ brne.d r12,0,.Lfound0
+ xor r6,r2,r5
+ ld.a r2,[r0,4]
+ sub r12,r6,r3
+ bic r12,r12,r6
+ and r7,r12,r4
+ breq r7,0,.Loop /* ... so that this branch is unaligned. */
+ ; Found searched-for character. r0 has already advanced to next word.
+#ifdef __LITTLE_ENDIAN__
+/* We only need the information about the first matching byte
+ (i.e. the least significant matching byte) to be exact,
+ hence there is no problem with carry effects. */
+.Lfound_char:
+ sub r3,r7,1
+ bic r3,r3,r7
+ norm r2,r3
+ sub_s r0,r0,1
+ asr_s r2,r2,3
+ j.d [blink]
+ sub_s r0,r0,r2
+
+ .balign 4
+.Lfound0_ua:
+ mov r3,r7
+.Lfound0:
+ sub r3,r6,r3
+ bic r3,r3,r6
+ and r2,r3,r4
+ or_s r12,r12,r2
+ sub_s r3,r12,1
+ bic_s r3,r3,r12
+ norm r3,r3
+ add_s r0,r0,3
+ asr_s r12,r3,3
+ asl.f 0,r2,r3
+ sub_s r0,r0,r12
+ j_s.d [blink]
+ mov.pl r0,0
+#else /* BIG ENDIAN */
+.Lfound_char:
+ lsr r7,r7,7
+
+ bic r2,r7,r6
+ norm r2,r2
+ sub_s r0,r0,4
+ asr_s r2,r2,3
+ j.d [blink]
+ add_s r0,r0,r2
+
+.Lfound0_ua:
+ mov_s r3,r7
+.Lfound0:
+ asl_s r2,r2,7
+ or r7,r6,r4
+ bic_s r12,r12,r2
+ sub r2,r7,r3
+ or r2,r2,r6
+ bic r12,r2,r12
+ bic.f r3,r4,r12
+ norm r3,r3
+
+ add.pl r3,r3,1
+ asr_s r12,r3,3
+ asl.f 0,r2,r3
+ add_s r0,r0,r12
+ j_s.d [blink]
+ mov.mi r0,0
+#endif /* ENDIAN */
+ARC_EXIT strchr
diff --git a/arch/arc/lib/strcmp.S b/arch/arc/lib/strcmp.S
new file mode 100644
index 0000000..5dc802b
--- /dev/null
+++ b/arch/arc/lib/strcmp.S
@@ -0,0 +1,96 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+/* This is optimized primarily for the ARC700.
+ It would be possible to speed up the loops by one cycle / word
+ respective one cycle / byte by forcing double source 1 alignment, unrolling
+ by a factor of two, and speculatively loading the second word / byte of
+ source 1; however, that would increase the overhead for loop setup / finish,
+ and strcmp might often terminate early. */
+
+#include <asm/linkage.h>
+
+ARC_ENTRY strcmp
+ or r2,r0,r1
+ bmsk_s r2,r2,1
+ brne r2,0,.Lcharloop
+ mov_s r12,0x01010101
+ ror r5,r12
+.Lwordloop:
+ ld.ab r2,[r0,4]
+ ld.ab r3,[r1,4]
+ nop_s
+ sub r4,r2,r12
+ bic r4,r4,r2
+ and r4,r4,r5
+ brne r4,0,.Lfound0
+ breq r2,r3,.Lwordloop
+#ifdef __LITTLE_ENDIAN__
+ xor r0,r2,r3 ; mask for difference
+ sub_s r1,r0,1
+ bic_s r0,r0,r1 ; mask for least significant difference bit
+ sub r1,r5,r0
+ xor r0,r5,r1 ; mask for least significant difference byte
+ and_s r2,r2,r0
+ and_s r3,r3,r0
+#endif /* LITTLE ENDIAN */
+ cmp_s r2,r3
+ mov_s r0,1
+ j_s.d [blink]
+ bset.lo r0,r0,31
+
+ .balign 4
+#ifdef __LITTLE_ENDIAN__
+.Lfound0:
+ xor r0,r2,r3 ; mask for difference
+ or r0,r0,r4 ; or in zero indicator
+ sub_s r1,r0,1
+ bic_s r0,r0,r1 ; mask for least significant difference bit
+ sub r1,r5,r0
+ xor r0,r5,r1 ; mask for least significant difference byte
+ and_s r2,r2,r0
+ and_s r3,r3,r0
+ sub.f r0,r2,r3
+ mov.hi r0,1
+ j_s.d [blink]
+ bset.lo r0,r0,31
+#else /* BIG ENDIAN */
+ /* The zero-detection above can mis-detect 0x01 bytes as zeroes
+ because of carry-propagateion from a lower significant zero byte.
+ We can compensate for this by checking that bit0 is zero.
+ This compensation is not necessary in the step where we
+ get a low estimate for r2, because in any affected bytes
+ we already have 0x00 or 0x01, which will remain unchanged
+ when bit 7 is cleared. */
+ .balign 4
+.Lfound0:
+ lsr r0,r4,8
+ lsr_s r1,r2
+ bic_s r2,r2,r0 ; get low estimate for r2 and get ...
+ bic_s r0,r0,r1 ; <this is the adjusted mask for zeros>
+ or_s r3,r3,r0 ; ... high estimate r3 so that r2 > r3 will ...
+ cmp_s r3,r2 ; ... be independent of trailing garbage
+ or_s r2,r2,r0 ; likewise for r3 > r2
+ bic_s r3,r3,r0
+ rlc r0,0 ; r0 := r2 > r3 ? 1 : 0
+ cmp_s r2,r3
+ j_s.d [blink]
+ bset.lo r0,r0,31
+#endif /* ENDIAN */
+
+ .balign 4
+.Lcharloop:
+ ldb.ab r2,[r0,1]
+ ldb.ab r3,[r1,1]
+ nop_s
+ breq r2,0,.Lcmpend
+ breq r2,r3,.Lcharloop
+.Lcmpend:
+ j_s.d [blink]
+ sub r0,r2,r3
+ARC_EXIT strcmp
diff --git a/arch/arc/lib/strcpy-700.S b/arch/arc/lib/strcpy-700.S
new file mode 100644
index 0000000..b7ca4ae
--- /dev/null
+++ b/arch/arc/lib/strcpy-700.S
@@ -0,0 +1,70 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+/* If dst and src are 4 byte aligned, copy 8 bytes at a time.
+ If the src is 4, but not 8 byte aligned, we first read 4 bytes to get
+ it 8 byte aligned. Thus, we can do a little read-ahead, without
+ dereferencing a cache line that we should not touch.
+ Note that short and long instructions have been scheduled to avoid
+ branch stalls.
+ The beq_s to r3z could be made unaligned & long to avoid a stall
+ there, but the it is not likely to be taken often, and it
+ would also be likey to cost an unaligned mispredict at the next call. */
+
+#include <asm/linkage.h>
+
+ARC_ENTRY strcpy
+ or r2,r0,r1
+ bmsk_s r2,r2,1
+ brne.d r2,0,charloop
+ mov_s r10,r0
+ ld_s r3,[r1,0]
+ mov r8,0x01010101
+ bbit0.d r1,2,loop_start
+ ror r12,r8
+ sub r2,r3,r8
+ bic_s r2,r2,r3
+ tst_s r2,r12
+ bne r3z
+ mov_s r4,r3
+ .balign 4
+loop:
+ ld.a r3,[r1,4]
+ st.ab r4,[r10,4]
+loop_start:
+ ld.a r4,[r1,4]
+ sub r2,r3,r8
+ bic_s r2,r2,r3
+ tst_s r2,r12
+ bne_s r3z
+ st.ab r3,[r10,4]
+ sub r2,r4,r8
+ bic r2,r2,r4
+ tst r2,r12
+ beq loop
+ mov_s r3,r4
+#ifdef __LITTLE_ENDIAN__
+r3z: bmsk.f r1,r3,7
+ lsr_s r3,r3,8
+#else
+r3z: lsr.f r1,r3,24
+ asl_s r3,r3,8
+#endif
+ bne.d r3z
+ stb.ab r1,[r10,1]
+ j_s [blink]
+
+ .balign 4
+charloop:
+ ldb.ab r3,[r1,1]
+
+
+ brne.d r3,0,charloop
+ stb.ab r3,[r10,1]
+ j [blink]
+ARC_EXIT strcpy
diff --git a/arch/arc/lib/strlen.S b/arch/arc/lib/strlen.S
new file mode 100644
index 0000000..39759e0
--- /dev/null
+++ b/arch/arc/lib/strlen.S
@@ -0,0 +1,83 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <asm/linkage.h>
+
+ARC_ENTRY strlen
+ or r3,r0,7
+ ld r2,[r3,-7]
+ ld.a r6,[r3,-3]
+ mov r4,0x01010101
+ ; uses long immediate
+#ifdef __LITTLE_ENDIAN__
+ asl_s r1,r0,3
+ btst_s r0,2
+ asl r7,r4,r1
+ ror r5,r4
+ sub r1,r2,r7
+ bic_s r1,r1,r2
+ mov.eq r7,r4
+ sub r12,r6,r7
+ bic r12,r12,r6
+ or.eq r12,r12,r1
+ and r12,r12,r5
+ brne r12,0,.Learly_end
+#else /* BIG ENDIAN */
+ ror r5,r4
+ btst_s r0,2
+ mov_s r1,31
+ sub3 r7,r1,r0
+ sub r1,r2,r4
+ bic_s r1,r1,r2
+ bmsk r1,r1,r7
+ sub r12,r6,r4
+ bic r12,r12,r6
+ bmsk.ne r12,r12,r7
+ or.eq r12,r12,r1
+ and r12,r12,r5
+ brne r12,0,.Learly_end
+#endif /* ENDIAN */
+
+.Loop:
+ ld_s r2,[r3,4]
+ ld.a r6,[r3,8]
+ ; stall for load result
+ sub r1,r2,r4
+ bic_s r1,r1,r2
+ sub r12,r6,r4
+ bic r12,r12,r6
+ or r12,r12,r1
+ and r12,r12,r5
+ breq r12,0,.Loop
+.Lend:
+ and.f r1,r1,r5
+ sub.ne r3,r3,4
+ mov.eq r1,r12
+#ifdef __LITTLE_ENDIAN__
+ sub_s r2,r1,1
+ bic_s r2,r2,r1
+ norm r1,r2
+ sub_s r0,r0,3
+ lsr_s r1,r1,3
+ sub r0,r3,r0
+ j_s.d [blink]
+ sub r0,r0,r1
+#else /* BIG ENDIAN */
+ lsr_s r1,r1,7
+ mov.eq r2,r6
+ bic_s r1,r1,r2
+ norm r1,r1
+ sub r0,r3,r0
+ lsr_s r1,r1,3
+ j_s.d [blink]
+ add r0,r0,r1
+#endif /* ENDIAN */
+.Learly_end:
+ b.d .Lend
+ sub_s.ne r1,r1,r1
+ARC_EXIT strlen
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/entry.h | 495 ++++++++++++++++++++++++++++++++++++
arch/arc/kernel/entry.S | 577 ++++++++++++++++++++++++++++++++++++++++++
2 files changed, 1072 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/entry.h
create mode 100644 arch/arc/kernel/entry.S
diff --git a/arch/arc/include/asm/entry.h b/arch/arc/include/asm/entry.h
new file mode 100644
index 0000000..3e0a9f4
--- /dev/null
+++ b/arch/arc/include/asm/entry.h
@@ -0,0 +1,495 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Vineetg: Aug 28th 2008: Bug #94984
+ * -Zero Overhead Loop Context shd be cleared when entering IRQ/EXcp/Trap
+ * Normally CPU does this automatically, however when doing FAKE rtie,
+ * we also need to explicitly do this. The problem in macros
+ * FAKE_RET_FROM_EXCPN and FAKE_RET_FROM_EXCPN_LOCK_IRQ was that this bit
+ * was being "CLEARED" rather then "SET". Actually "SET" clears ZOL context
+ *
+ * Vineetg: May 5th 2008
+ * - Defined Stack Switching Macro to be reused in all intr/excp hdlrs
+ * - Shaved off 11 instructions from RESTORE_ALL_INT1 by using the
+ * address Write back load ld.ab instead of seperate ld/add instn
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef __ASM_ARC_ENTRY_H
+#define __ASM_ARC_ENTRY_H
+
+#ifdef __ASSEMBLY__
+#include <asm/unistd.h> /* For NR_syscalls defination */
+#include <asm/asm-offsets.h>
+#include <asm/arcregs.h>
+#include <asm/ptrace.h>
+#include <asm/thread_info.h> /* For THREAD_SIZE */
+
+/* Note on the LD/ST addr modes with addr reg wback
+ *
+ * LD.a same as LD.aw
+ *
+ * LD.a reg1, [reg2, x] => Pre Incr
+ * Eff Addr for load = [reg2 + x]
+ *
+ * LD.ab reg1, [reg2, x] => Post Incr
+ * Eff Addr for load = [reg2]
+ */
+
+/*--------------------------------------------------------------
+ * Save caller saved registers (scratch registers) ( r0 - r12 )
+ * Registers are pushed / popped in the order defined in struct ptregs
+ * in asm/ptrace.h
+ *-------------------------------------------------------------*/
+.macro SAVE_CALLER_SAVED
+ st.a r0, [sp, -4]
+ st.a r1, [sp, -4]
+ st.a r2, [sp, -4]
+ st.a r3, [sp, -4]
+ st.a r4, [sp, -4]
+ st.a r5, [sp, -4]
+ st.a r6, [sp, -4]
+ st.a r7, [sp, -4]
+ st.a r8, [sp, -4]
+ st.a r9, [sp, -4]
+ st.a r10, [sp, -4]
+ st.a r11, [sp, -4]
+ st.a r12, [sp, -4]
+.endm
+
+/*--------------------------------------------------------------
+ * Restore caller saved registers (scratch registers)
+ *-------------------------------------------------------------*/
+.macro RESTORE_CALLER_SAVED
+ ld.ab r12, [sp, 4]
+ ld.ab r11, [sp, 4]
+ ld.ab r10, [sp, 4]
+ ld.ab r9, [sp, 4]
+ ld.ab r8, [sp, 4]
+ ld.ab r7, [sp, 4]
+ ld.ab r6, [sp, 4]
+ ld.ab r5, [sp, 4]
+ ld.ab r4, [sp, 4]
+ ld.ab r3, [sp, 4]
+ ld.ab r2, [sp, 4]
+ ld.ab r1, [sp, 4]
+ ld.ab r0, [sp, 4]
+.endm
+
+
+/*--------------------------------------------------------------
+ * Save callee saved registers (non scratch registers) ( r13 - r25 )
+ * on kernel stack.
+ * User mode callee regs need to be saved in case of
+ * -fork and friends for replicating from parent to child
+ * -before going into do_signal( ) for ptrace/core-dump
+ * Special case handling is required for r25 in case it is used by kernel
+ * for caching task ptr. Low level exception/ISR save user mode r25
+ * into task->thread.user_r25. So it needs to be retrieved from there and
+ * saved into kernel stack wit rest of callee reg-file
+ *-------------------------------------------------------------*/
+.macro SAVE_CALLEE_SAVED_USER
+ st.a r13, [sp, -4]
+ st.a r14, [sp, -4]
+ st.a r15, [sp, -4]
+ st.a r16, [sp, -4]
+ st.a r17, [sp, -4]
+ st.a r18, [sp, -4]
+ st.a r19, [sp, -4]
+ st.a r20, [sp, -4]
+ st.a r21, [sp, -4]
+ st.a r22, [sp, -4]
+ st.a r23, [sp, -4]
+ st.a r24, [sp, -4]
+ st.a r25, [sp, -4]
+
+ /* move up by 1 word to "create" callee_regs->"stack_place_holder" */
+ sub sp, sp, 4
+.endm
+
+/*--------------------------------------------------------------
+ * Save callee saved registers (non scratch registers) ( r13 - r25 )
+ * kernel mode callee regs needed to be saved in case of context switch
+ * If r25 is used for caching task pointer then that need not be saved
+ * as it can be re-created from current task global
+ *-------------------------------------------------------------*/
+.macro SAVE_CALLEE_SAVED_KERNEL
+ st.a r13, [sp, -4]
+ st.a r14, [sp, -4]
+ st.a r15, [sp, -4]
+ st.a r16, [sp, -4]
+ st.a r17, [sp, -4]
+ st.a r18, [sp, -4]
+ st.a r19, [sp, -4]
+ st.a r20, [sp, -4]
+ st.a r21, [sp, -4]
+ st.a r22, [sp, -4]
+ st.a r23, [sp, -4]
+ st.a r24, [sp, -4]
+ st.a r25, [sp, -4]
+ sub sp, sp, 4
+.endm
+
+/*--------------------------------------------------------------
+ * RESTORE_CALLEE_SAVED:
+ * Loads callee (non scratch) Reg File by popping from Kernel mode stack.
+ * This is reverse of SAVE_CALLEE_SAVED,
+ *
+ * NOTE:
+ * Ideally this shd only be called in switch_to for loading
+ * switched-IN task's CALLEE Reg File.
+ * For all other cases RESTORE_CALLEE_SAVED_FAST must be used
+ * which simply pops the stack w/o touching regs.
+ *-------------------------------------------------------------*/
+.macro RESTORE_CALLEE_SAVED_KERNEL
+
+ add sp, sp, 4 /* skip "callee_regs->stack_place_holder" */
+ ld.ab r25, [sp, 4]
+ ld.ab r24, [sp, 4]
+ ld.ab r23, [sp, 4]
+ ld.ab r22, [sp, 4]
+ ld.ab r21, [sp, 4]
+ ld.ab r20, [sp, 4]
+ ld.ab r19, [sp, 4]
+ ld.ab r18, [sp, 4]
+ ld.ab r17, [sp, 4]
+ ld.ab r16, [sp, 4]
+ ld.ab r15, [sp, 4]
+ ld.ab r14, [sp, 4]
+ ld.ab r13, [sp, 4]
+
+.endm
+
+/*--------------------------------------------------------------
+ * Super FAST Restore callee saved regs by simply re-adjusting SP
+ *-------------------------------------------------------------*/
+.macro DISCARD_CALLEE_SAVED_USER
+ add sp, sp, 14 * 4
+.endm
+
+/*--------------------------------------------------------------
+ * Restore User mode r25 saved in task_struct->thread.user_r25
+ *-------------------------------------------------------------*/
+.macro RESTORE_USER_R25
+ ld r25, [r25, TASK_THREAD + THREAD_USER_R25]
+.endm
+
+/*-------------------------------------------------------------
+ * given a tsk struct, get to the base of it's kernel mode stack
+ * tsk->thread_info is really a PAGE, whose bottom hoists stack
+ * which grows upwards towards thread_info
+ *------------------------------------------------------------*/
+
+.macro GET_TSK_STACK_BASE tsk, out
+
+ /* Get task->thread_info (this is essentially start of a PAGE) */
+ ld \out, [\tsk, TASK_THREAD_INFO]
+
+ /* Go to end of page where stack begins (grows upwards) */
+ add2 \out, \out, (THREAD_SIZE - 4)/4 /* one word GUTTER */
+
+.endm
+
+/*--------------------------------------------------------------
+ * Switch to Kernel Mode stack if SP points to User Mode stack
+ *
+ * Entry : r9 contains pre-IRQ/exception/trap status32
+ * Exit : SP is set to kernel mode stack pointer
+ * Clobbers: r9
+ *-------------------------------------------------------------*/
+
+.macro SWITCH_TO_KERNEL_STK
+
+ /* User Mode when this happened ? Yes: Proceed to switch stack */
+ bbit1 r9, STATUS_U_BIT, 88f
+
+ /* OK we were already in kernel mode when this event happened, thus can
+ * assume SP is kernel mode SP. _NO_ need to do any stack switching
+ */
+
+ /* Save Pre Intr/Exception KERNEL MODE SP on kernel stack
+ * safe-keeping not really needed, but it keeps the epilogue code
+ * (SP restore) simpler/uniform.
+ */
+ b.d 77f
+
+ st.a sp, [sp, -12] ; Make room for orig_r0 and orig_r8
+
+88: /*------Intr/Ecxp happened in user mode, "switch" stack ------ */
+
+ GET_CURR_TASK_ON_CPU r9
+
+ /* With current tsk in r9, get it's kernel mode stack base */
+ GET_TSK_STACK_BASE r9, r9
+
+#ifdef PT_REGS_CANARY
+ st 0xabcdabcd, [r9, 0]
+#endif
+
+ /* Save Pre Intr/Exception User SP on kernel stack */
+ st.a sp, [r9, -12] ; Make room for orig_r0 and orig_r8
+
+ /* CAUTION:
+ * SP should be set at the very end when we are done with everything
+ * In case of 2 levels of interrupt we depend on value of SP to assume
+ * that everything else is done (loading r25 etc)
+ */
+
+ /* set SP to point to kernel mode stack */
+ mov sp, r9
+
+77: /* ----- Stack Switched to kernel Mode, Now save REG FILE ----- */
+
+.endm
+
+/*------------------------------------------------------------
+ * "FAKE" a rtie to return from CPU Exception context
+ * This is to re-enable Exceptions within exception
+ * Look at EV_ProtV to see how this is actually used
+ *-------------------------------------------------------------*/
+
+.macro FAKE_RET_FROM_EXCPN reg
+
+ ld \reg, [sp, PT_status32]
+ bic \reg, \reg, (STATUS_U_MASK|STATUS_DE_MASK)
+ bset \reg, \reg, STATUS_L_BIT
+ sr \reg, [erstatus]
+ mov \reg, 55f
+ sr \reg, [eret]
+
+ rtie
+55:
+.endm
+
+/*
+ * @reg [OUT] &thread_info of "current"
+ */
+.macro GET_CURR_THR_INFO_FROM_SP reg
+ and \reg, sp, ~(THREAD_SIZE - 1)
+.endm
+
+/*
+ * @reg [OUT] thread_info->flags of "current"
+ */
+.macro GET_CURR_THR_INFO_FLAGS reg
+ GET_CURR_THR_INFO_FROM_SP \reg
+ ld \reg, [\reg, THREAD_INFO_FLAGS]
+.endm
+
+/*--------------------------------------------------------------
+ * For early Exception Prologue, a core reg is temporarily needed to
+ * code the rest of prolog (stack switching). This is done by stashing
+ * it to memory (non-SMP case) or SCRATCH0 Aux Reg (SMP).
+ *
+ * Before saving the full regfile - this reg is restored back, only
+ * to be saved again on kernel mode stack, as part of ptregs.
+ *-------------------------------------------------------------*/
+.macro EXCPN_PROLOG_FREEUP_REG reg
+ st \reg, [@ex_saved_reg1]
+.endm
+
+.macro EXCPN_PROLOG_RESTORE_REG reg
+ ld \reg, [@ex_saved_reg1]
+.endm
+
+/*--------------------------------------------------------------
+ * Save all registers used by Exceptions (TLB Miss, Prot-V, Mem err etc)
+ * Requires SP to be already switched to kernel mode Stack
+ * sp points to the next free element on the stack at exit of this macro.
+ * Registers are pushed / popped in the order defined in struct ptregs
+ * in asm/ptrace.h
+ * Note that syscalls are implemented via TRAP which is also a exception
+ * from CPU's point of view
+ *-------------------------------------------------------------*/
+.macro SAVE_ALL_EXCEPTION marker
+
+ /* Restore r9 used to code the early prologue */
+ EXCPN_PROLOG_RESTORE_REG r9
+
+ /* Save the complete regfile now */
+
+ /* orig_r8 marker:
+ * syscalls -> 1 to NR_SYSCALLS
+ * Exceptions -> NR_SYSCALLS + 1
+ * Break-point-> NR_SYSCALLS + 2
+ */
+ st \marker, [sp, 8]
+ st r0, [sp, 4] /* orig_r0, needed only for sys calls */
+ SAVE_CALLER_SAVED
+ st.a r26, [sp, -4] /* gp */
+ st.a fp, [sp, -4]
+ st.a blink, [sp, -4]
+ lr r9, [eret]
+ st.a r9, [sp, -4]
+ lr r9, [erstatus]
+ st.a r9, [sp, -4]
+ st.a lp_count, [sp, -4]
+ lr r9, [lp_end]
+ st.a r9, [sp, -4]
+ lr r9, [lp_start]
+ st.a r9, [sp, -4]
+ lr r9, [erbta]
+ st.a r9, [sp, -4]
+
+#ifdef PT_REGS_CANARY
+ mov r9, 0xdeadbeef
+ st r9, [sp, -4]
+#endif
+
+ /* move up by 1 word to "create" pt_regs->"stack_place_holder" */
+ sub sp, sp, 4
+.endm
+
+/*--------------------------------------------------------------
+ * Save scratch regs for exceptions
+ *-------------------------------------------------------------*/
+.macro SAVE_ALL_SYS
+ SAVE_ALL_EXCEPTION (NR_syscalls + 1)
+.endm
+
+/*--------------------------------------------------------------
+ * Save scratch regs for sys calls
+ *-------------------------------------------------------------*/
+.macro SAVE_ALL_TRAP
+ SAVE_ALL_EXCEPTION r8
+.endm
+
+/*--------------------------------------------------------------
+ * Restore all registers used by system call or Exceptions
+ * SP should always be pointing to the next free stack element
+ * when entering this macro.
+ *
+ * NOTE:
+ *
+ * It is recommended that lp_count/ilink1/ilink2 not be used as a dest reg
+ * for memory load operations. If used in that way interrupts are deffered
+ * by hardware and that is not good.
+ *-------------------------------------------------------------*/
+.macro RESTORE_ALL_SYS
+
+ add sp, sp, 4 /* hop over unused "pt_regs->stack_place_holder" */
+
+ ld.ab r9, [sp, 4]
+ sr r9, [erbta]
+ ld.ab r9, [sp, 4]
+ sr r9, [lp_start]
+ ld.ab r9, [sp, 4]
+ sr r9, [lp_end]
+ ld.ab r9, [sp, 4]
+ mov lp_count, r9
+ ld.ab r9, [sp, 4]
+ sr r9, [erstatus]
+ ld.ab r9, [sp, 4]
+ sr r9, [eret]
+ ld.ab blink, [sp, 4]
+ ld.ab fp, [sp, 4]
+ ld.ab r26, [sp, 4] /* gp */
+ RESTORE_CALLER_SAVED
+
+ ld sp, [sp] /* restore original sp */
+ /* orig_r0 and orig_r8 skipped automatically */
+.endm
+
+
+/*--------------------------------------------------------------
+ * Save all registers used by interrupt handlers.
+ *-------------------------------------------------------------*/
+.macro SAVE_ALL_INT1
+
+ /* restore original r9 , saved in int1_saved_reg
+ * It will be saved on stack in macro: SAVE_CALLER_SAVED
+ */
+ ld r9, [@int1_saved_reg]
+
+ /* now we are ready to save the remaining context :) */
+ st -1, [sp, 8] /* orig_r8, -1 for interuppt level one */
+ st 0, [sp, 4] /* orig_r0 , N/A for IRQ */
+ SAVE_CALLER_SAVED
+ st.a r26, [sp, -4] /* gp */
+ st.a fp, [sp, -4]
+ st.a blink, [sp, -4]
+ st.a ilink1, [sp, -4]
+ lr r9, [status32_l1]
+ st.a r9, [sp, -4]
+ st.a lp_count, [sp, -4]
+ lr r9, [lp_end]
+ st.a r9, [sp, -4]
+ lr r9, [lp_start]
+ st.a r9, [sp, -4]
+ lr r9, [bta_l1]
+ st.a r9, [sp, -4]
+
+#ifdef PT_REGS_CANARY
+ mov r9, 0xdeadbee1
+ st r9, [sp, -4]
+#endif
+ /* move up by 1 word to "create" pt_regs->"stack_place_holder" */
+ sub sp, sp, 4
+.endm
+
+/*--------------------------------------------------------------
+ * Restore all registers used by interrupt handlers.
+ *
+ * NOTE:
+ *
+ * It is recommended that lp_count/ilink1/ilink2 not be used as a dest reg
+ * for memory load operations. If used in that way interrupts are deffered
+ * by hardware and that is not good.
+ *-------------------------------------------------------------*/
+
+.macro RESTORE_ALL_INT1
+ add sp, sp, 4 /* hop over unused "pt_regs->stack_place_holder" */
+
+ ld.ab r9, [sp, 4] /* Actual reg file */
+ sr r9, [bta_l1]
+ ld.ab r9, [sp, 4]
+ sr r9, [lp_start]
+ ld.ab r9, [sp, 4]
+ sr r9, [lp_end]
+ ld.ab r9, [sp, 4]
+ mov lp_count, r9
+ ld.ab r9, [sp, 4]
+ sr r9, [status32_l1]
+ ld.ab r9, [sp, 4]
+ mov ilink1, r9
+ ld.ab blink, [sp, 4]
+ ld.ab fp, [sp, 4]
+ ld.ab r26, [sp, 4] /* gp */
+ RESTORE_CALLER_SAVED
+
+ ld sp, [sp] /* restore original sp */
+ /* orig_r0 and orig_r8 skipped automatically */
+.endm
+
+/* Get CPU-ID of this core */
+.macro GET_CPU_ID reg
+ lr \reg, [identity]
+ lsr \reg, \reg, 8
+ bmsk \reg, \reg, 7
+.endm
+
+.macro GET_CURR_TASK_ON_CPU reg
+ ld \reg, [@_current_task]
+.endm
+
+.macro SET_CURR_TASK_ON_CPU tsk, tmp
+ st \tsk, [@_current_task]
+.endm
+
+/* ------------------------------------------------------------------
+ * Get the ptr to some field of Current Task at @off in task struct
+ */
+
+.macro GET_CURR_TASK_FIELD_PTR off, reg
+ GET_CURR_TASK_ON_CPU \reg
+ add \reg, \reg, \off
+.endm
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* __ASM_ARC_ENTRY_H */
diff --git a/arch/arc/kernel/entry.S b/arch/arc/kernel/entry.S
new file mode 100644
index 0000000..b829052
--- /dev/null
+++ b/arch/arc/kernel/entry.S
@@ -0,0 +1,577 @@
+/*
+ * Low Level Interrupts/Traps/Exceptions(non-TLB) Handling for ARC
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: Nov 2010:
+ * -Vector table jumps (@8 bytes) converted into branches (@4 bytes)
+ * -To maintain the slot size of 8 bytes/vector, added nop, which is
+ * not executed at runtime.
+ *
+ * vineetg: Nov 2009 (Everything needed for TIF_RESTORE_SIGMASK)
+ * -do_signal()invoked upon TIF_RESTORE_SIGMASK as well
+ * -Wrappers for sys_{,rt_}sigsuspend() nolonger needed as they don't
+ * need ptregs anymore
+ *
+ * Vineetg: Oct 2009
+ * -In a rare scenario, Process gets a Priv-V exception and gets scheduled
+ * out. Since we don't do FAKE RTIE for Priv-V, CPU excpetion state remains
+ * active (AE bit enabled). This causes a double fault for a subseq valid
+ * exception. Thus FAKE RTIE needed in low level Priv-Violation handler.
+ * Instr Error could also cause similar scenario, so same there as well.
+ *
+ * Vineetg: Aug 28th 2008: Bug #94984
+ * -Zero Overhead Loop Context shd be cleared when entering IRQ/EXcp/Trap
+ * Normally CPU does this automatically, however when doing FAKE rtie,
+ * we need to explicitly do this. The problem in macros
+ * FAKE_RET_FROM_EXCPN and FAKE_RET_FROM_EXCPN_LOCK_IRQ was that this bit
+ * was being "CLEARED" rather then "SET". Since it is Loop INHIBIT Bit,
+ * setting it and not clearing it clears ZOL context
+ *
+ * Vineetg: Dec 22, 2007
+ * Minor Surgery of Low Level ISR to make it SMP safe
+ * - MMU_SCRATCH0 Reg used for freeing up r9 in Level 1 ISR
+ * - _current_task is made an array of NR_CPUS
+ * - Access of _current_task wrapped inside a macro so that if hardware
+ * team agrees for a dedicated reg, no other code is touched
+ *
+ * Amit Bhor, Rahul Trivedi, Kanika Nema, Sameer Dhavale : Codito Tech 2004
+ */
+
+/*------------------------------------------------------------------
+ * Function ABI
+ *------------------------------------------------------------------
+ *
+ * Arguments r0 - r7
+ * Caller Saved Registers r0 - r12
+ * Callee Saved Registers r13- r25
+ * Global Pointer (gp) r26
+ * Frame Pointer (fp) r27
+ * Stack Pointer (sp) r28
+ * Interrupt link register (ilink1) r29
+ * Interrupt link register (ilink2) r30
+ * Branch link register (blink) r31
+ *------------------------------------------------------------------
+ */
+
+ .cpu A7
+
+;############################ Vector Table #################################
+
+.macro VECTOR lbl
+#if 1 /* Just in case, build breaks */
+ j \lbl
+#else
+ b \lbl
+ nop
+#endif
+.endm
+
+ .section .vector, "ax",@progbits
+ .align 4
+
+/* Each entry in the vector table must occupy 2 words. Since it is a jump
+ * across sections (.vector to .text) we are gauranteed that 'j somewhere'
+ * will use the 'j limm' form of the intrsuction as long as somewhere is in
+ * a section other than .vector.
+ */
+
+; ********* Critical System Events **********************
+VECTOR res_service ; 0x0, Restart Vector (0x0)
+VECTOR mem_service ; 0x8, Mem exception (0x1)
+VECTOR instr_service ; 0x10, Instrn Error (0x2)
+
+; ******************** Device ISRs **********************
+VECTOR handle_interrupt_level1
+
+VECTOR handle_interrupt_level1
+
+VECTOR handle_interrupt_level1
+
+VECTOR handle_interrupt_level1
+
+.rept 25
+VECTOR handle_interrupt_level1 ; Other devices
+.endr
+
+/* FOR ARC600: timer = 0x3, uart = 0x8, emac = 0x10 */
+
+; ******************** Exceptions **********************
+VECTOR EV_MachineCheck ; 0x100, Fatal Machine check (0x20)
+VECTOR EV_TLBMissI ; 0x108, Intruction TLB miss (0x21)
+VECTOR EV_TLBMissD ; 0x110, Data TLB miss (0x22)
+VECTOR EV_TLBProtV ; 0x118, Protection Violation (0x23)
+ ; or Misaligned Access
+VECTOR EV_PrivilegeV ; 0x120, Privilege Violation (0x24)
+VECTOR EV_Trap ; 0x128, Trap exception (0x25)
+VECTOR EV_Extension ; 0x130, Extn Intruction Excp (0x26)
+
+.rept 24
+VECTOR reserved ; Reserved Exceptions
+.endr
+
+#include <linux/linkage.h> /* ARC_{EXTRY,EXIT} */
+#include <asm/entry.h> /* SAVE_ALL_{INT1,INT2,TRAP...} */
+#include <asm/errno.h>
+#include <asm/arcregs.h>
+#include <asm/irqflags.h>
+
+;##################### Scratch Mem for IRQ stack switching #############
+
+ .section .data ; NOT .global
+ .align 32
+ .type int1_saved_reg, @object
+ .size int1_saved_reg, 4
+int1_saved_reg:
+ .zero 4
+
+; ---------------------------------------------
+ .section .text, "ax",@progbits
+
+res_service: ; processor restart
+ flag 0x1 ; not implemented
+ nop
+ nop
+
+reserved: ; processor restart
+ rtie ; jump to processor initializations
+
+;##################### Interrupt Handling ##############################
+
+; ---------------------------------------------
+; Level 1 ISR
+; ---------------------------------------------
+ARC_ENTRY handle_interrupt_level1
+
+ /* free up r9 as scratchpad */
+ st r9, [@int1_saved_reg]
+
+ ;Which mode (user/kernel) was the system in when intr occured
+ lr r9, [status32_l1]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_INT1
+
+ lr r0, [icause1]
+ and r0, r0, 0x1f
+
+ bl.d @arch_do_IRQ
+ mov r1, sp
+
+ mov r8,0x1
+ sr r8, [AUX_IRQ_LV12] ; clear bit in Sticky Status Reg
+
+ b ret_from_exception
+ARC_EXIT handle_interrupt_level1
+
+;################### Non TLB Exception Handling #############################
+
+; ---------------------------------------------
+; Instruction Error Exception Handler
+; ---------------------------------------------
+
+ARC_ENTRY instr_service
+
+ EXCPN_PROLOG_FREEUP_REG r9
+
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_SYS
+
+ lr r0, [ecr]
+ lr r1, [efa]
+
+ mov r2, sp
+
+ FAKE_RET_FROM_EXCPN r9
+
+ bl do_insterror_or_kprobe
+ b ret_from_exception
+ARC_EXIT instr_service
+
+; ---------------------------------------------
+; Memory Error Exception Handler
+; ---------------------------------------------
+
+ARC_ENTRY mem_service
+
+ EXCPN_PROLOG_FREEUP_REG r9
+
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_SYS
+
+ lr r0, [ecr]
+ lr r1, [efa]
+ mov r2, sp
+ bl do_memory_error
+ b ret_from_exception
+ARC_EXIT mem_service
+
+; ---------------------------------------------
+; Machine Check Exception Handler
+; ---------------------------------------------
+
+ARC_ENTRY EV_MachineCheck
+
+ EXCPN_PROLOG_FREEUP_REG r9
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_SYS
+
+ lr r0, [ecr]
+ lr r1, [efa]
+ mov r2, sp
+
+ brne r0, 0x200100, 1f
+ bl do_tlb_overlap_fault
+ b ret_from_exception
+
+1:
+ ; DEAD END: can't do much, display Regs and HALT
+ SAVE_CALLEE_SAVED_USER
+
+ GET_CURR_TASK_FIELD_PTR TASK_THREAD, r10
+ st sp, [r10, THREAD_CALLEE_REG]
+
+ j do_machine_check_fault
+
+ARC_EXIT EV_MachineCheck
+
+; ---------------------------------------------
+; Protection Violation Exception Handler
+; ---------------------------------------------
+
+ARC_ENTRY EV_TLBProtV
+
+ EXCPN_PROLOG_FREEUP_REG r9
+
+ ;Which mode (user/kernel) was the system in when Exception occured
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_SYS
+
+ ;---------(3) Save some more regs-----------------
+ ; vineetg: Mar 6th: Random Seg Fault issue #1
+ ; ecr and efa were not saved in case an Intr sneaks in
+ ; after fake rtie
+ ;
+ lr r3, [ecr]
+ lr r4, [efa]
+
+ ; --------(4) Return from CPU Exception Mode ---------
+ ; Fake a rtie, but rtie to next label
+ ; That way, subsequently, do_page_fault ( ) executes in pure kernel
+ ; mode with further Exceptions enabled
+
+ FAKE_RET_FROM_EXCPN r9
+
+ ;------ (5) Type of Protection Violation? ----------
+ ;
+ ; ProtV Hardware Exception is triggered for Access Faults of 2 types
+ ; -Access Violaton (WRITE to READ ONLY Page) - for linux COW
+ ; -Unaligned Access (READ/WRITE on odd boundary)
+ ;
+ cmp r3, 0x230400 ; Misaligned data access ?
+ beq 4f
+
+ ;========= (6a) Access Violation Processing ========
+ cmp r3, 0x230100
+ mov r1, 0x0 ; if LD exception ? write = 0
+ mov.ne r1, 0x1 ; else write = 1
+
+ mov r2, r4 ; faulting address
+ mov r0, sp ; pt_regs
+ bl do_page_fault
+ b ret_from_exception
+
+ ;========== (6b) Non aligned access ============
+4:
+ mov r0, r3 ; cause code
+ mov r1, r4 ; faulting address
+ mov r2, sp ; pt_regs
+
+ bl do_misaligned_access
+ b ret_from_exception
+
+ARC_EXIT EV_TLBProtV
+
+; ---------------------------------------------
+; Privilege Violation Exception Handler
+; ---------------------------------------------
+ARC_ENTRY EV_PrivilegeV
+
+ EXCPN_PROLOG_FREEUP_REG r9
+
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_SYS
+
+ lr r0, [ecr]
+ lr r1, [efa]
+ mov r2, sp
+
+ FAKE_RET_FROM_EXCPN r9
+
+ bl do_privilege_fault
+ b ret_from_exception
+ARC_EXIT EV_PrivilegeV
+
+; ---------------------------------------------
+; Extension Instruction Exception Handler
+; ---------------------------------------------
+ARC_ENTRY EV_Extension
+
+ EXCPN_PROLOG_FREEUP_REG r9
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_SYS
+
+ lr r0, [ecr]
+ lr r1, [efa]
+ mov r2, sp
+ bl do_extension_fault
+ b ret_from_exception
+ARC_EXIT EV_Extension
+
+;################### Break Point TRAP ##########################
+
+ ; ======= (5b) Trap is due to Break-Point =========
+
+trap_with_param:
+
+ ;make sure orig_r8 is a positive value
+ st NR_syscalls + 2, [sp, PT_orig_r8]
+
+ mov r0, r12
+ lr r1, [efa]
+ mov r2, sp
+
+ ; Now that we have read EFA, its safe to do "fake" rtie
+ ; and get out of CPU exception mode
+ FAKE_RET_FROM_EXCPN r11
+
+ ; Save callee regs in case gdb wants to have a look
+ ; SP will grow up by size of CALLEE Reg-File
+ ; NOTE: clobbers r12
+ SAVE_CALLEE_SAVED_USER
+
+ ; save location of saved Callee Regs @ thread_struct->pc
+ GET_CURR_TASK_FIELD_PTR TASK_THREAD, r10
+ st sp, [r10, THREAD_CALLEE_REG]
+
+ ; Call the trap handler
+ bl do_non_swi_trap
+
+ ; unwind stack to discard Callee saved Regs
+ DISCARD_CALLEE_SAVED_USER
+
+ b ret_from_exception
+
+;##################### Trap Handling ##############################
+;
+; EV_Trap caused by TRAP_S and TRAP0 instructions.
+;------------------------------------------------------------------
+; (1) System Calls
+; :parameters in r0-r7.
+; :r8 has the system call number
+; (2) Break Points
+;------------------------------------------------------------------
+
+ARC_ENTRY EV_Trap
+
+ ; Need at least 1 reg to code the early exception prolog
+ EXCPN_PROLOG_FREEUP_REG r9
+
+ ;Which mode (user/kernel) was the system in when intr occured
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_TRAP
+
+ ;------- (4) What caused the Trap --------------
+ lr r12, [ecr]
+ and.f 0, r12, ECR_PARAM_MASK
+ bnz trap_with_param
+
+ ; ======= (5a) Trap is due to System Call ========
+
+ ; Before doing anything, return from CPU Exception Mode
+ FAKE_RET_FROM_EXCPN r11
+
+ ;============ This is normal System Call case ==========
+ ; Sys-call num shd not exceed the total system calls avail
+ cmp r8, NR_syscalls
+ mov.hi r0, -ENOSYS
+ bhi ret_from_system_call
+
+ ; Offset into the syscall_table and call handler
+ ld.as r9,[sys_call_table, r8]
+ jl [r9] ; Entry into Sys Call Handler
+
+ ; fall through to ret_from_system_call
+ARC_EXIT EV_Trap
+
+ARC_ENTRY ret_from_system_call
+
+ st r0, [sp, PT_r0] ; sys call return value in pt_regs
+
+ ; fall through yet again to ret_from_exception
+
+;############# Return from Intr/Excp/Trap (Linux Specifics) ##############
+;
+; If ret to user mode do we need to handle signals, schedule() et al.
+
+ARC_ENTRY ret_from_exception
+
+ ; Pre-{IRQ,Trap,Exception} K/U mode from pt_regs->status32
+ ld r8, [sp, PT_status32] ; returning to User/Kernel Mode
+
+#ifdef CONFIG_PREEMPT
+ bbit0 r8, STATUS_U_BIT, resume_kernel_mode
+#else
+ bbit0 r8, STATUS_U_BIT, restore_regs
+#endif
+
+resume_user_mode_begin:
+
+ ; Before returning to User mode check-for-and-complete any pending work
+ ; such as rescheduling/signal-delivery etc.
+ ; Disable IRQs to ensures that chk for pending work itself is atomic
+ ; (and we don't end up missing a NEED_RESCHED/SIGPENDING due to an
+ ; interim IRQ).
+ ;
+ ; IRQs would be re-enabled by one of the ensuing functions such as
+ ; schedule()or worst case RTIE will automatically re-enable them,
+ ; based on cpu state prior to this event.
+
+ IRQ_DISABLE r10
+
+ ; Fast Path return to user mode if no pending work
+ GET_CURR_THR_INFO_FLAGS r9
+ and.f 0, r9, _TIF_WORK_MASK
+ bz restore_regs
+
+ ; --- task preemption ---
+ bbit0 r9, TIF_NEED_RESCHED, 1f
+ bl @schedule ; blnz yields relocation errors
+ b resume_user_mode_begin ; loop back to start of U mode ret
+
+1:
+ ; --- signal handling ---
+ mov r0, sp ; pt_regs for arg to do_signal()/do_notify_resume()
+
+ ; --------- check for signals/restore-sigmask ------------
+ bbit0 r9, TIF_SIGPENDING, chk_next_work
+
+ ; save CALLEE Regs.
+ ; (i) If this signal causes coredump - full regfile needed
+ ; (ii) If signal is SIGTRAP/SIGSTOP, task is being traced thus
+ ; tracer might call PEEKUSR for a CALLEE reg
+ ;
+ ; NOTE: SP will grow up by size of CALLEE Reg-File
+ SAVE_CALLEE_SAVED_USER ; clobbers r12
+
+ ; save location of saved Callee Regs @ thread_struct->callee
+ GET_CURR_TASK_FIELD_PTR TASK_THREAD, r10
+ st sp, [r10, THREAD_CALLEE_REG]
+
+ bl @do_signal
+
+ ; unwind SP for cheap discard of Callee saved Regs
+ DISCARD_CALLEE_SAVED_USER
+
+ b resume_user_mode_begin ; loop back to start of U mode ret
+
+ ; --- notify_resume ---
+chk_next_work:
+ btst r9, TIF_NOTIFY_RESUME
+ blnz @do_notify_resume
+
+ ;--------- All things done, go back to Userland ------
+
+ b restore_regs
+
+#ifdef CONFIG_PREEMPT
+
+resume_kernel_mode:
+
+ ; Can't preempt if preemption disabled
+ GET_CURR_THR_INFO_FROM_SP r10
+ ld r8, [r10, THREAD_INFO_PREEMPT_COUNT]
+ brne r8, 0, restore_regs
+
+ ; check if this task's NEED_RESCHED flag set
+ ld r9, [r10, THREAD_INFO_FLAGS]
+ bbit0 r9, TIF_NEED_RESCHED, restore_regs
+
+ IRQ_DISABLE r9
+
+ ; Invoke PREEMPTION
+ bl preempt_schedule_irq
+
+ ; preempt_schedule_irq() always returns with IRQ disabled
+#endif
+
+ ; fall through
+
+;############# Return from Intr/Excp/Trap (ARC Specifics) ##############
+;
+; Restore the saved sys context (common exit-path for EXCPN/IRQ/Trap)
+; IRQ shd definitely not happen between now and rtie
+
+restore_regs :
+
+ ; Disable Interrupts while restoring reg-file back
+ ; XXX can this be optimised out
+ IRQ_DISABLE_SAVE r9, r10 ;@r10 has prisitine (pre-disable) copy
+
+ ; Restore REG File. In case multiple Events outstanding,
+ ; use the same priorty as rtie: EXCPN, L2 IRQ, L1 IRQ, None
+ ; Note that we use realtime STATUS32 (not pt_regs->status32) to
+ ; decide that.
+
+ ; if Returning from Exception
+ bbit0 r10, STATUS_AE_BIT, not_exception
+ RESTORE_ALL_SYS
+ rtie
+
+ ; Not Exception so maybe Interrupts (Level 1 or 2)
+
+not_exception:
+
+ bbit0 r10, STATUS_A1_BIT, not_level1_interrupt
+
+ ;return from level 1
+
+ RESTORE_ALL_INT1
+debug_marker_l1:
+ rtie
+
+not_level1_interrupt:
+
+ ;this case is for syscalls or Exceptions (with fake rtie)
+
+ RESTORE_ALL_SYS
+debug_marker_syscall:
+ rtie
+
+ARC_EXIT ret_from_exception
+
+ARC_ENTRY ret_from_fork
+ ; when the forked child comes here from the __switch_to function
+ ; r0 has the last task pointer.
+ ; put last task in scheduler queue
+ bl @schedule_tail
+ b @ret_from_exception
+ARC_EXIT ret_from_fork
--
1.7.4.1
* L1_CACHE_SHIFT
* PAGE_SIZE, PAGE_OFFSET
* struct pt_regs, struct user_regs_struct
* struct thread_struct, cpu_relax(), task_pt_regs(), start_thread(), ...
* struct thread_info, THREAD_SIZE, INIT_THREAD_INFO(), TIF_*, ...
* BUG()
* ELF_*
* Elf_*
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/bug.h | 37 +++++++++
arch/arc/include/asm/cache.h | 21 +++++
arch/arc/include/asm/elf.h | 97 ++++++++++++++++++++++++
arch/arc/include/asm/exec.h | 15 ++++
arch/arc/include/asm/kdebug.h | 19 +++++
arch/arc/include/asm/linkage.h | 30 ++++++++
arch/arc/include/asm/module.h | 17 ++++
arch/arc/include/asm/page.h | 42 +++++++++++
arch/arc/include/asm/processor.h | 143 ++++++++++++++++++++++++++++++++++++
arch/arc/include/asm/ptrace.h | 120 ++++++++++++++++++++++++++++++
arch/arc/include/asm/sections.h | 17 ++++
arch/arc/include/asm/thread_info.h | 120 ++++++++++++++++++++++++++++++
12 files changed, 678 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/bug.h
create mode 100644 arch/arc/include/asm/cache.h
create mode 100644 arch/arc/include/asm/elf.h
create mode 100644 arch/arc/include/asm/exec.h
create mode 100644 arch/arc/include/asm/kdebug.h
create mode 100644 arch/arc/include/asm/linkage.h
create mode 100644 arch/arc/include/asm/module.h
create mode 100644 arch/arc/include/asm/page.h
create mode 100644 arch/arc/include/asm/processor.h
create mode 100644 arch/arc/include/asm/ptrace.h
create mode 100644 arch/arc/include/asm/sections.h
create mode 100644 arch/arc/include/asm/thread_info.h
diff --git a/arch/arc/include/asm/bug.h b/arch/arc/include/asm/bug.h
new file mode 100644
index 0000000..2ad8f9b
--- /dev/null
+++ b/arch/arc/include/asm/bug.h
@@ -0,0 +1,37 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_BUG_H
+#define _ASM_ARC_BUG_H
+
+#ifndef __ASSEMBLY__
+
+#include <asm/ptrace.h>
+
+struct task_struct;
+
+void show_regs(struct pt_regs *regs);
+void show_stacktrace(struct task_struct *tsk, struct pt_regs *regs);
+void show_kernel_fault_diag(const char *str, struct pt_regs *regs,
+ unsigned long address, unsigned long cause_reg);
+void die(const char *str, struct pt_regs *regs, unsigned long address,
+ unsigned long cause_reg);
+
+#define BUG() do { \
+ dump_stack(); \
+ pr_warn("Kernel BUG in %s: %s: %d!\n", \
+ __FILE__, __func__, __LINE__); \
+} while (0)
+
+#define HAVE_ARCH_BUG
+
+#include <asm-generic/bug.h>
+
+#endif /* !__ASSEMBLY__ */
+
+#endif
diff --git a/arch/arc/include/asm/cache.h b/arch/arc/include/asm/cache.h
new file mode 100644
index 0000000..30c72a4
--- /dev/null
+++ b/arch/arc/include/asm/cache.h
@@ -0,0 +1,21 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ARC_ASM_CACHE_H
+#define __ARC_ASM_CACHE_H
+
+/* In case $$ not config, setup a dummy number for rest of kernel */
+#ifndef CONFIG_ARC_CACHE_LINE_SHIFT
+#define L1_CACHE_SHIFT 6
+#else
+#define L1_CACHE_SHIFT CONFIG_ARC_CACHE_LINE_SHIFT
+#endif
+
+#define L1_CACHE_BYTES (1 << L1_CACHE_SHIFT)
+
+#endif /* _ASM_CACHE_H */
diff --git a/arch/arc/include/asm/elf.h b/arch/arc/include/asm/elf.h
new file mode 100644
index 0000000..147284f
--- /dev/null
+++ b/arch/arc/include/asm/elf.h
@@ -0,0 +1,97 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_ELF_H
+#define __ASM_ARC_ELF_H
+
+#include <asm/ptrace.h>
+
+#define EM_ARCOMPACT 93
+
+/* Machine specific ELF Hdr flags */
+#define EF_ARC_OSABI_MSK 0x00000f00
+#define EF_ARC_OSABI_ORIG 0x00000000 /* MUST be zero for back-compat */
+#define EF_ARC_OSABI_V2 0x00000200
+
+/* ARC Relocations (kernel Modules only) */
+#define R_ARC_32 0x4
+#define R_ARC_32_ME 0x1B
+#define R_ARC_S25H_PCREL 0x10
+#define R_ARC_S25W_PCREL 0x11
+
+typedef unsigned long elf_greg_t;
+typedef unsigned long elf_fpregset_t;
+
+
+/* core dump regs is in the order pt_regs, callee_regs, stop_pc (for gdb) */
+#define ELF_NGREG ((sizeof(struct pt_regs) + sizeof(struct callee_regs) \
+ + sizeof(unsigned long)) / sizeof(elf_greg_t))
+
+typedef elf_greg_t elf_gregset_t[ELF_NGREG];
+
+/*
+ * To ensure that
+ * -we don't load something for the wrong architecture.
+ * -The userspace is using the correct syscall ABI
+ */
+struct elf32_hdr;
+extern int elf_check_arch(const struct elf32_hdr *);
+#define elf_check_arch elf_check_arch
+
+/*
+ * These are used to set parameters in the core dumps.
+ */
+#define ELF_ARCH EM_ARCOMPACT
+#define ELF_CLASS ELFCLASS32
+
+#ifdef CONFIG_CPU_BIG_ENDIAN
+#define ELF_DATA ELFDATA2MSB
+#else
+#define ELF_DATA ELFDATA2LSB
+#endif
+
+#ifdef __KERNEL__
+
+#define CORE_DUMP_USE_REGSET
+
+#define ELF_EXEC_PAGESIZE PAGE_SIZE
+
+/*
+ * This is the location that an ET_DYN program is loaded if exec'ed. Typical
+ * use of this is to invoke "./ld.so someprog" to test out a new version of
+ * the loader. We need to make sure that it is out of the way of the program
+ * that it will "exec", and that there is sufficient room for the brk.
+ */
+#define ELF_ET_DYN_BASE (2 * TASK_SIZE / 3)
+
+/*
+ * When the program starts, a1 contains a pointer to a function to be
+ * registered with atexit, as per the SVR4 ABI. A value of 0 means we
+ * have no such handler.
+ */
+#define ELF_PLAT_INIT(_r, load_addr) ((_r)->r0 = 0)
+
+/*
+ * This yields a mask that user programs can use to figure out what
+ * instruction set this cpu supports.
+ */
+#define ELF_HWCAP (0)
+
+/*
+ * This yields a string that ld.so will use to load implementation
+ * specific libraries for optimization. This is more specific in
+ * intent than poking at uname or /proc/cpuinfo.
+ */
+#define ELF_PLATFORM (NULL)
+
+#define SET_PERSONALITY(ex) \
+ set_personality(PER_LINUX | (current->personality & (~PER_MASK)))
+
+#endif /* __KERNEL__ */
+
+#endif
diff --git a/arch/arc/include/asm/exec.h b/arch/arc/include/asm/exec.h
new file mode 100644
index 0000000..28abc69
--- /dev/null
+++ b/arch/arc/include/asm/exec.h
@@ -0,0 +1,15 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_EXEC_H
+#define __ASM_ARC_EXEC_H
+
+/* Align to 16b */
+#define arch_align_stack(p) ((unsigned long)(p) & ~0xf)
+
+#endif
diff --git a/arch/arc/include/asm/kdebug.h b/arch/arc/include/asm/kdebug.h
new file mode 100644
index 0000000..3fbe6c4
--- /dev/null
+++ b/arch/arc/include/asm/kdebug.h
@@ -0,0 +1,19 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_KDEBUG_H
+#define _ASM_ARC_KDEBUG_H
+
+enum die_val {
+ DIE_UNUSED,
+ DIE_TRAP,
+ DIE_IERR,
+ DIE_OOPS
+};
+
+#endif
diff --git a/arch/arc/include/asm/linkage.h b/arch/arc/include/asm/linkage.h
new file mode 100644
index 0000000..a45d1bb
--- /dev/null
+++ b/arch/arc/include/asm/linkage.h
@@ -0,0 +1,30 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_LINKAGE_H
+#define __ASM_LINKAGE_H
+
+#ifdef __ASSEMBLY__
+
+/* Can't use the ENTRY macro in linux/linkage.h
+ * gas considers ';' as comment vs. newline
+ */
+.macro ARC_ENTRY name
+ .global \name
+ .align 4
+ \name:
+.endm
+
+.macro ARC_EXIT name
+#define ASM_PREV_SYM_ADDR(name) .-##name
+ .size \ name, ASM_PREV_SYM_ADDR(\name)
+.endm
+
+#endif /* __ASSEMBLY__ */
+
+#endif
diff --git a/arch/arc/include/asm/module.h b/arch/arc/include/asm/module.h
new file mode 100644
index 0000000..165d768
--- /dev/null
+++ b/arch/arc/include/asm/module.h
@@ -0,0 +1,17 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+
+ */
+
+#ifndef _ASM_ARC_MODULE_H
+#define _ASM_ARC_MODULE_H
+
+#include <asm-generic/module.h>
+
+#endif /* _ASM_ARC_MODULE_H */
diff --git a/arch/arc/include/asm/page.h b/arch/arc/include/asm/page.h
new file mode 100644
index 0000000..7cd1224
--- /dev/null
+++ b/arch/arc/include/asm/page.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_PAGE_H
+#define __ASM_ARC_PAGE_H
+
+/* PAGE_SHIFT determines the page size */
+#if defined(CONFIG_ARC_PAGE_SIZE_16K)
+#define PAGE_SHIFT 14
+#elif defined(CONFIG_ARC_PAGE_SIZE_4K)
+#define PAGE_SHIFT 12
+#else
+/*
+ * Default 8k
+ * done this way (instead of under CONFIG_ARC_PAGE_SIZE_8K) because adhoc
+ * user code (busybox appletlib.h) expects PAGE_SHIFT to be defined w/o
+ * using the correct uClibc header and in their build our autoconf.h is
+ * not available
+ */
+#define PAGE_SHIFT 13
+#endif
+
+#ifdef __ASSEMBLY__
+#define PAGE_SIZE (1 << PAGE_SHIFT)
+#define PAGE_OFFSET (0x80000000)
+#else
+#define PAGE_SIZE (1UL << PAGE_SHIFT) /* Default 8K */
+#define PAGE_OFFSET (0x80000000UL) /* Kernel starts at 2G onwards */
+#endif
+
+#define PAGE_MASK (~(PAGE_SIZE-1))
+
+#endif /* !__ASSEMBLY__ */
+
+#endif /* __KERNEL__ */
+
+#endif
diff --git a/arch/arc/include/asm/processor.h b/arch/arc/include/asm/processor.h
new file mode 100644
index 0000000..e2445bc
--- /dev/null
+++ b/arch/arc/include/asm/processor.h
@@ -0,0 +1,143 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: March 2009
+ * -Implemented task_pt_regs( )
+ *
+ * Amit Bhor, Sameer Dhavale, Ashwin Chaugule: Codito Technologies 2004
+ */
+
+#ifndef __ASM_ARC_PROCESSOR_H
+#define __ASM_ARC_PROCESSOR_H
+
+#if defined(__KERNEL__)
+
+#if !defined(__ASSEMBLY__)
+
+#include <asm/arcregs.h> /* for STATUS_E1_MASK et all */
+
+/* Arch specific stuff which needs to be saved per task.
+ * However these items are not so important so as to earn a place in
+ * struct thread_info
+ */
+struct thread_struct {
+ unsigned long ksp; /* kernel mode stack pointer */
+ unsigned long callee_reg; /* pointer to callee regs */
+ unsigned long fault_address; /* dbls as brkpt holder as well */
+ unsigned long cause_code; /* Exception Cause Code (ECR) */
+};
+
+#define INIT_THREAD { \
+ .ksp = sizeof(init_stack) + (unsigned long) init_stack, \
+}
+
+/* Forward declaration, a strange C thing */
+struct task_struct;
+
+/*
+ * Return saved PC of a blocked thread.
+ */
+unsigned long thread_saved_pc(struct task_struct *t);
+
+#define task_pt_regs(p) \
+ ((struct pt_regs *)(THREAD_SIZE - 4 + (void *)task_stack_page(p)) - 1)
+
+/* Free all resources held by a thread. */
+#define release_thread(thread) do { } while (0)
+
+/* Prepare to copy thread state - unlazy all lazy status */
+#define prepare_to_copy(tsk) do { } while (0)
+
+#define cpu_relax() do { } while (0)
+
+/*
+ * Create a new kernel thread
+ */
+
+extern int kernel_thread(int (*fn) (void *), void *arg, unsigned long flags);
+
+#define copy_segments(tsk, mm) do { } while (0)
+#define release_segments(mm) do { } while (0)
+
+#define KSTK_EIP(tsk) (task_pt_regs(tsk)->ret)
+
+/*
+ * Where abouts of Task's sp, fp, blink when it was last seen in kernel mode.
+ * These can't be derived from pt_regs as that would give correp user-mode val
+ */
+#define KSTK_ESP(tsk) (tsk->thread.ksp)
+#define KSTK_BLINK(tsk) (*((unsigned int *)((KSTK_ESP(tsk)) + (13+1+1)*4)))
+#define KSTK_FP(tsk) (*((unsigned int *)((KSTK_ESP(tsk)) + (13+1)*4)))
+
+/*
+ * Do necessary setup to start up a newly executed thread.
+ *
+ * E1,E2 so that Interrupts are enabled in user mode
+ * L set, so Loop inhibited to begin with
+ * lp_start and lp_end seeded with bogus non-zero values so to easily catch
+ * the ARC700 sr to lp_start hardware bug
+ */
+#define start_thread(_regs, _pc, _usp) \
+do { \
+ set_fs(USER_DS); /* reads from user space */ \
+ (_regs)->ret = (_pc); \
+ /* Interrupts enabled in User Mode */ \
+ (_regs)->status32 = STATUS_U_MASK | STATUS_L_MASK \
+ | STATUS_E1_MASK | STATUS_E2_MASK; \
+ (_regs)->sp = (_usp); \
+ /* bogus seed values for debugging */ \
+ (_regs)->lp_start = 0x10; \
+ (_regs)->lp_end = 0x80; \
+} while (0)
+
+extern unsigned int get_wchan(struct task_struct *p);
+
+/*
+ * Default implementation of macro that returns current
+ * instruction pointer ("program counter").
+ * Should the PC register be read instead ? This macro does not seem to
+ * be used in many places so this wont be all that bad.
+ */
+#define current_text_addr() ({ __label__ _l; _l: &&_l; })
+
+#endif /* !__ASSEMBLY__ */
+
+/* Kernels Virtual memory area.
+ * Unlike other architectures(MIPS, sh, cris ) ARC 700 does not have a
+ * "kernel translated" region (like KSEG2 in MIPS). So we use a upper part
+ * of the translated bottom 2GB for kernel virtual memory and protect
+ * these pages from user accesses by disabling Ru, Eu and Wu.
+ */
+#define VMALLOC_SIZE (0x10000000) /* 256M */
+#define VMALLOC_START (PAGE_OFFSET - VMALLOC_SIZE)
+#define VMALLOC_END (PAGE_OFFSET)
+
+/* Most of the architectures seem to be keeping some kind of padding between
+ * userspace TASK_SIZE and PAGE_OFFSET. i.e TASK_SIZE != PAGE_OFFSET.
+ */
+#define USER_KERNEL_GUTTER 0x10000000
+
+/* User address space:
+ * On ARC700, CPU allows the entire lower half of 32 bit address space to be
+ * translated. Thus potentially 2G (0:0x7FFF_FFFF) could be User vaddr space.
+ * However we steal 256M for kernel addr (0x7000_0000:0x7FFF_FFFF) and another
+ * 256M (0x6000_0000:0x6FFF_FFFF) is gutter between user/kernel spaces
+ * Thus total User vaddr space is (0:0x5FFF_FFFF)
+ */
+#define TASK_SIZE (PAGE_OFFSET - VMALLOC_SIZE - USER_KERNEL_GUTTER)
+
+#define STACK_TOP TASK_SIZE
+#define STACK_TOP_MAX STACK_TOP
+
+/* This decides where the kernel will search for a free chunk of vm
+ * space during mmap's.
+ */
+#define TASK_UNMAPPED_BASE (TASK_SIZE / 3)
+
+#endif /* __KERNEL__ */
+
+#endif /* __ASM_ARC_PROCESSOR_H */
diff --git a/arch/arc/include/asm/ptrace.h b/arch/arc/include/asm/ptrace.h
new file mode 100644
index 0000000..c178357
--- /dev/null
+++ b/arch/arc/include/asm/ptrace.h
@@ -0,0 +1,120 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef __ASM_ARC_PTRACE_H
+#define __ASM_ARC_PTRACE_H
+
+/*
+ * XXX: ABI hack.
+ * The offset calc can be cleanly done in asm-offset.c, however gdb includes
+ * this header directly.
+ */
+#define PT_bta 4
+#define PT_lp_start 8
+#define PT_lp_end 12
+#define PT_lp_count 16
+#define PT_status32 20
+#define PT_ret 24
+#define PT_blink 28
+#define PT_fp 32
+#define PT_r26 36
+#define PT_r12 40
+#define PT_r11 44
+#define PT_r10 48
+#define PT_r9 52
+#define PT_r8 56
+#define PT_r7 60
+#define PT_r6 64
+#define PT_r5 68
+#define PT_r4 72
+#define PT_r3 76
+#define PT_r2 80
+#define PT_r1 84
+#define PT_r0 88
+#define PT_sp 92
+#define PT_orig_r0 96
+#define PT_orig_r8 100
+
+#ifndef __ASSEMBLY__
+
+/* THE pt_regs: Defines how regs are saved during entry into kernel */
+
+struct pt_regs {
+ /*
+ * 1 word gutter after reg-file has been saved
+ * Technically not needed, Since SP always points to a "full" location
+ * (vs. "empty"). But pt_regs is shared with tools....
+ */
+ long res;
+
+ /* Real registers */
+ long bta; /* bta_l1, bta_l2, erbta */
+ long lp_start;
+ long lp_end;
+ long lp_count;
+ long status32; /* status32_l1, status32_l2, erstatus */
+ long ret; /* ilink1, ilink2 or eret */
+ long blink;
+ long fp;
+ long r26; /* gp */
+ long r12;
+ long r11;
+ long r10;
+ long r9;
+ long r8;
+ long r7;
+ long r6;
+ long r5;
+ long r4;
+ long r3;
+ long r2;
+ long r1;
+ long r0;
+ long sp; /* user/kernel sp depending on where we came from */
+ long orig_r0;
+ long orig_r8; /*to distinguish bet excp, sys call, int1 or int2 */
+};
+
+/* Callee saved registers - need to be saved only when you are scheduled out */
+
+struct callee_regs {
+ long res; /* Again this is not needed */
+ long r25;
+ long r24;
+ long r23;
+ long r22;
+ long r21;
+ long r20;
+ long r19;
+ long r18;
+ long r17;
+ long r16;
+ long r15;
+ long r14;
+ long r13;
+};
+
+/* User mode registers, used for core dumps. */
+struct user_regs_struct {
+ struct pt_regs scratch;
+ struct callee_regs callee;
+ long efa; /* break pt addr, for break points in delay slots */
+ long stop_pc; /* give dbg stop_pc directly after checking orig_r8 */
+};
+
+#define instruction_pointer(regs) ((regs)->ret)
+#define profile_pc(regs) instruction_pointer(regs)
+
+/* return 1 if user mode or 0 if kernel mode */
+#define user_mode(regs) (regs->status32 & STATUS_U_MASK)
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* __ASM_PTRACE_H */
diff --git a/arch/arc/include/asm/sections.h b/arch/arc/include/asm/sections.h
new file mode 100644
index 0000000..fc15f2e
--- /dev/null
+++ b/arch/arc/include/asm/sections.h
@@ -0,0 +1,17 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_SECTIONS_H
+#define _ASM_ARC_SECTIONS_H
+
+#include <asm-generic/sections.h>
+
+extern char _int_vec_base_lds[];
+extern char __arc_dccm_base[];
+
+#endif
diff --git a/arch/arc/include/asm/thread_info.h b/arch/arc/include/asm/thread_info.h
new file mode 100644
index 0000000..937a818
--- /dev/null
+++ b/arch/arc/include/asm/thread_info.h
@@ -0,0 +1,120 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Vineetg: Oct 2009
+ * No need for ARC specific thread_info allocator (kmalloc/free). This is
+ * anyways one page allocation, thus slab alloc can be short-circuited and
+ * the generic version (get_free_page) would be loads better.
+ *
+ * Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef _ASM_THREAD_INFO_H
+#define _ASM_THREAD_INFO_H
+
+#ifdef __KERNEL__
+
+#include <asm/page.h>
+
+#ifdef CONFIG_16KSTACKS
+#define THREAD_SIZE_ORDER 1
+#else
+#define THREAD_SIZE_ORDER 0
+#endif
+
+#define THREAD_SIZE (PAGE_SIZE << THREAD_SIZE_ORDER)
+
+#ifndef __ASSEMBLY__
+
+#include <linux/thread_info.h>
+#include <asm/segment.h>
+
+/*
+ * low level task data that entry.S needs immediate access to
+ * - this struct should fit entirely inside of one cache line
+ * - this struct shares the supervisor stack pages
+ * - if the contents of this structure are changed, the assembly constants
+ * must also be changed
+ */
+struct thread_info {
+ unsigned long flags; /* low level flags */
+ int preempt_count; /* 0 => preemptable, <0 => BUG */
+ struct task_struct *task; /* main task structure */
+ mm_segment_t addr_limit; /* thread address space */
+ struct exec_domain *exec_domain;/* execution domain */
+ __u32 cpu; /* current CPU */
+ unsigned long thr_ptr; /* TLS ptr */
+ struct restart_block restart_block;
+};
+
+/*
+ * macros/functions for gaining access to the thread information structure
+ *
+ * preempt_count needs to be 1 initially, until the scheduler is functional.
+ */
+#define INIT_THREAD_INFO(tsk) \
+{ \
+ .task = &tsk, \
+ .exec_domain = &default_exec_domain, \
+ .flags = 0, \
+ .cpu = 0, \
+ .preempt_count = INIT_PREEMPT_COUNT, \
+ .addr_limit = KERNEL_DS, \
+ .restart_block = { \
+ .fn = do_no_restart_syscall, \
+ }, \
+}
+
+#define init_thread_info (init_thread_union.thread_info)
+#define init_stack (init_thread_union.stack)
+
+static inline __attribute_const__ struct thread_info *current_thread_info(void)
+{
+ register unsigned long sp asm("sp");
+ return (struct thread_info *)(sp & ~(THREAD_SIZE - 1));
+}
+
+#endif /* !__ASSEMBLY__ */
+
+#define PREEMPT_ACTIVE 0x10000000
+
+/*
+ * thread information flags
+ * - these are process state flags that various assembly files may need to
+ * access
+ * - pending work-to-be-done flags are in LSW
+ * - other flags in MSW
+ */
+#define TIF_RESTORE_SIGMASK 0 /* restore sig mask in do_signal() */
+#define TIF_NOTIFY_RESUME 1 /* resumption notification requested */
+#define TIF_SIGPENDING 2 /* signal pending */
+#define TIF_NEED_RESCHED 3 /* rescheduling necessary */
+#define TIF_SYSCALL_AUDIT 4 /* syscall auditing active */
+#define TIF_SYSCALL_TRACE 15 /* syscall trace active */
+
+/* true if poll_idle() is polling TIF_NEED_RESCHED */
+#define TIF_MEMDIE 16
+
+#define _TIF_SYSCALL_TRACE (1<<TIF_SYSCALL_TRACE)
+#define _TIF_NOTIFY_RESUME (1<<TIF_NOTIFY_RESUME)
+#define _TIF_SIGPENDING (1<<TIF_SIGPENDING)
+#define _TIF_NEED_RESCHED (1<<TIF_NEED_RESCHED)
+#define _TIF_SYSCALL_AUDIT (1<<TIF_SYSCALL_AUDIT)
+#define _TIF_MEMDIE (1<<TIF_MEMDIE)
+
+/* work to do on interrupt/exception return */
+#define _TIF_WORK_MASK (0x0000FFFF & ~_TIF_SYSCALL_TRACE)
+
+/*
+ * _TIF_ALLWORK_MASK includes SYSCALL_TRACE, but we don't need it.
+ * SYSCALL_TRACE is anways seperately/unconditionally tested right after a
+ * syscall, so all that reamins to be tested is _TIF_WORK_MASK
+ */
+
+#endif /* __KERNEL__ */
+
+#endif /* _ASM_THREAD_INFO_H */
--
1.7.4.1
arches can have more efficient implementation of these routines
Acked-by: Arnd Bergmann <[email protected]>
Signed-off-by: Vineet Gupta <[email protected]>
---
include/asm-generic/checksum.h | 4 ++++
lib/checksum.c | 2 ++
2 files changed, 6 insertions(+), 0 deletions(-)
diff --git a/include/asm-generic/checksum.h b/include/asm-generic/checksum.h
index c084767..59811df 100644
--- a/include/asm-generic/checksum.h
+++ b/include/asm-generic/checksum.h
@@ -38,12 +38,15 @@ extern __wsum csum_partial_copy_from_user(const void __user *src, void *dst,
csum_partial_copy((src), (dst), (len), (sum))
#endif
+#ifndef ip_fast_csum
/*
* This is a version of ip_compute_csum() optimized for IP headers,
* which always checksum on 4 octet boundaries.
*/
extern __sum16 ip_fast_csum(const void *iph, unsigned int ihl);
+#endif
+#ifndef csum_fold
/*
* Fold a partial checksum
*/
@@ -54,6 +57,7 @@ static inline __sum16 csum_fold(__wsum csum)
sum = (sum & 0xffff) + (sum >> 16);
return (__force __sum16)~sum;
}
+#endif
#ifndef csum_tcpudp_nofold
/*
diff --git a/lib/checksum.c b/lib/checksum.c
index 12dceb2..129775e 100644
--- a/lib/checksum.c
+++ b/lib/checksum.c
@@ -102,6 +102,7 @@ out:
}
#endif
+#ifndef ip_fast_csum
/*
* This is a version of ip_compute_csum() optimized for IP headers,
* which always checksum on 4 octet boundaries.
@@ -111,6 +112,7 @@ __sum16 ip_fast_csum(const void *iph, unsigned int ihl)
return (__force __sum16)~do_csum(iph, ihl*4);
}
EXPORT_SYMBOL(ip_fast_csum);
+#endif
/*
* computes the checksum of a memory block at buff, length len,
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/segment.h | 24 ++
arch/arc/include/asm/uaccess.h | 605 ++++++++++++++++++++++++++++++++++++++++
2 files changed, 629 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/segment.h
create mode 100644 arch/arc/include/asm/uaccess.h
diff --git a/arch/arc/include/asm/segment.h b/arch/arc/include/asm/segment.h
new file mode 100644
index 0000000..da2c459
--- /dev/null
+++ b/arch/arc/include/asm/segment.h
@@ -0,0 +1,24 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASMARC_SEGMENT_H
+#define __ASMARC_SEGMENT_H
+
+#ifndef __ASSEMBLY__
+
+typedef unsigned long mm_segment_t;
+
+#define MAKE_MM_SEG(s) ((mm_segment_t) { (s) })
+
+#define KERNEL_DS MAKE_MM_SEG(0)
+#define USER_DS MAKE_MM_SEG(TASK_SIZE)
+
+#define segment_eq(a, b) ((a) == (b))
+
+#endif /* __ASSEMBLY__ */
+#endif /* __ASMARC_SEGMENT_H */
diff --git a/arch/arc/include/asm/uaccess.h b/arch/arc/include/asm/uaccess.h
new file mode 100644
index 0000000..503c98d
--- /dev/null
+++ b/arch/arc/include/asm/uaccess.h
@@ -0,0 +1,605 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: June 2010
+ * -__clear_user( ) called multiple times during elf load was byte loop
+ * converted to do as much word clear as possible.
+ *
+ * vineetg: Dec 2009
+ * -Hand crafted constant propagation for "constant" copy sizes
+ * -stock kernel shrunk by 33K at -O3
+ *
+ * vineetg: Sept 2009
+ * -Added option to (UN)inline copy_(to|from)_user to reduce code sz
+ * -kernel shrunk by 200K even at -O3 (gcc 4.2.1)
+ * -Enabled when doing -Os
+ *
+ * Amit Bhor, Sameer Dhavale: Codito Technologies 2004
+ */
+
+#ifndef _ASM_ARC_UACCESS_H
+#define _ASM_ARC_UACCESS_H
+
+#include <linux/sched.h>
+#include <asm/errno.h>
+#include <linux/string.h> /* for generic string functions */
+
+
+#define __kernel_ok (segment_eq(get_fs(), KERNEL_DS))
+#define __user_ok(addr, sz) (((sz) <= TASK_SIZE) && \
+ (((addr)+(sz)) <= get_fs()))
+#define __access_ok(addr, sz) (unlikely(__kernel_ok) || \
+ likely(__user_ok((addr), (sz))))
+
+static inline unsigned long
+__arc_copy_from_user(void *to, const void __user *from, unsigned long n)
+{
+ long res = 0;
+ char val;
+ unsigned long tmp1, tmp2, tmp3, tmp4;
+ unsigned long orig_n = n;
+
+ if (n == 0)
+ return 0;
+
+ /* unaligned */
+ if (((unsigned long)to & 0x3) || ((unsigned long)from & 0x3)) {
+
+ unsigned char tmp;
+
+ __asm__ __volatile__ (
+ " mov.f lp_count, %0 \n"
+ " lpnz 2f \n"
+ "1: ldb.ab %1, [%3, 1] \n"
+ " stb.ab %1, [%2, 1] \n"
+ " sub %0,%0,1 \n"
+ "2: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "3: j 2b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 3b \n"
+ " .previous \n"
+
+ : "+r" (n),
+ /*
+ * Note as an '&' earlyclobber operand to make sure the
+ * temporary register inside the loop is not the same as
+ * FROM or TO.
+ */
+ "=&r" (tmp), "+r" (to), "+r" (from)
+ :
+ : "lp_count", "lp_start", "lp_end", "memory");
+
+ return n;
+ }
+
+ /*
+ * Hand-crafted constant propagation to reduce code sz of the
+ * laddered copy 16x,8,4,2,1
+ */
+ if (__builtin_constant_p(orig_n)) {
+ res = orig_n;
+
+ if (orig_n / 16) {
+ orig_n = orig_n % 16;
+
+ __asm__ __volatile__(
+ " lsr lp_count, %7,4 \n"
+ " lp 3f \n"
+ "1: ld.ab %3, [%2, 4] \n"
+ "11: ld.ab %4, [%2, 4] \n"
+ "12: ld.ab %5, [%2, 4] \n"
+ "13: ld.ab %6, [%2, 4] \n"
+ " st.ab %3, [%1, 4] \n"
+ " st.ab %4, [%1, 4] \n"
+ " st.ab %5, [%1, 4] \n"
+ " st.ab %6, [%1, 4] \n"
+ " sub %0,%0,16 \n"
+ "3: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 3b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 4b \n"
+ " .word 11b,4b \n"
+ " .word 12b,4b \n"
+ " .word 13b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from),
+ "=r"(tmp1), "=r"(tmp2), "=r"(tmp3), "=r"(tmp4)
+ : "ir"(n)
+ : "lp_count", "memory");
+ }
+ if (orig_n / 8) {
+ orig_n = orig_n % 8;
+
+ __asm__ __volatile__(
+ "14: ld.ab %3, [%2,4] \n"
+ "15: ld.ab %4, [%2,4] \n"
+ " st.ab %3, [%1,4] \n"
+ " st.ab %4, [%1,4] \n"
+ " sub %0,%0,8 \n"
+ "31: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 31b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 14b,4b \n"
+ " .word 15b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from),
+ "=r"(tmp1), "=r"(tmp2)
+ :
+ : "memory");
+ }
+ if (orig_n / 4) {
+ orig_n = orig_n % 4;
+
+ __asm__ __volatile__(
+ "16: ld.ab %3, [%2,4] \n"
+ " st.ab %3, [%1,4] \n"
+ " sub %0,%0,4 \n"
+ "32: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 32b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 16b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from), "=r"(tmp1)
+ :
+ : "memory");
+ }
+ if (orig_n / 2) {
+ orig_n = orig_n % 2;
+
+ __asm__ __volatile__(
+ "17: ldw.ab %3, [%2,2] \n"
+ " stw.ab %3, [%1,2] \n"
+ " sub %0,%0,2 \n"
+ "33: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 33b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 17b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from), "=r"(tmp1)
+ :
+ : "memory");
+ }
+ if (orig_n & 1) {
+ __asm__ __volatile__(
+ "18: ldb.ab %3, [%2,2] \n"
+ " stb.ab %3, [%1,2] \n"
+ " sub %0,%0,1 \n"
+ "34: ; nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 34b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 18b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from), "=r"(tmp1)
+ :
+ : "memory");
+ }
+ } else { /* n is NOT constant, so laddered copy of 16x,8,4,2,1 */
+
+ __asm__ __volatile__(
+ " mov %0,%3 \n"
+ " lsr.f lp_count, %3,4 \n" /* 16x bytes */
+ " lpnz 3f \n"
+ "1: ld.ab %5, [%2, 4] \n"
+ "11: ld.ab %6, [%2, 4] \n"
+ "12: ld.ab %7, [%2, 4] \n"
+ "13: ld.ab %8, [%2, 4] \n"
+ " st.ab %5, [%1, 4] \n"
+ " st.ab %6, [%1, 4] \n"
+ " st.ab %7, [%1, 4] \n"
+ " st.ab %8, [%1, 4] \n"
+ " sub %0,%0,16 \n"
+ "3: and.f %3,%3,0xf \n" /* stragglers */
+ " bz 34f \n"
+ " bbit0 %3,3,31f \n" /* 8 bytes left */
+ "14: ld.ab %5, [%2,4] \n"
+ "15: ld.ab %6, [%2,4] \n"
+ " st.ab %5, [%1,4] \n"
+ " st.ab %6, [%1,4] \n"
+ " sub.f %0,%0,8 \n"
+ "31: bbit0 %3,2,32f \n" /* 4 bytes left */
+ "16: ld.ab %5, [%2,4] \n"
+ " st.ab %5, [%1,4] \n"
+ " sub.f %0,%0,4 \n"
+ "32: bbit0 %3,1,33f \n" /* 2 bytes left */
+ "17: ldw.ab %5, [%2,2] \n"
+ " stw.ab %5, [%1,2] \n"
+ " sub.f %0,%0,2 \n"
+ "33: bbit0 %3,0,34f \n"
+ "18: ldb.ab %5, [%2,1] \n" /* 1 byte left */
+ " stb.ab %5, [%1,1] \n"
+ " sub.f %0,%0,1 \n"
+ "34: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 34b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 4b \n"
+ " .word 11b,4b \n"
+ " .word 12b,4b \n"
+ " .word 13b,4b \n"
+ " .word 14b,4b \n"
+ " .word 15b,4b \n"
+ " .word 16b,4b \n"
+ " .word 17b,4b \n"
+ " .word 18b,4b \n"
+ " .previous \n"
+ : "=r" (res), "+r"(to), "+r"(from), "+r"(n), "=r"(val),
+ "=r"(tmp1), "=r"(tmp2), "=r"(tmp3), "=r"(tmp4)
+ :
+ : "lp_count", "memory");
+ }
+
+ return res;
+}
+
+extern unsigned long slowpath_copy_to_user(void __user *to, const void *from,
+ unsigned long n);
+
+static inline unsigned long
+__arc_copy_to_user(void __user *to, const void *from, unsigned long n)
+{
+ long res = 0;
+ char val;
+ unsigned long tmp1, tmp2, tmp3, tmp4;
+ unsigned long orig_n = n;
+
+ if (n == 0)
+ return 0;
+
+ /* unaligned */
+ if (((unsigned long)to & 0x3) || ((unsigned long)from & 0x3)) {
+
+ unsigned char tmp;
+
+ __asm__ __volatile__(
+ " mov.f lp_count, %0 \n"
+ " lpnz 3f \n"
+ " ldb.ab %1, [%3, 1] \n"
+ "1: stb.ab %1, [%2, 1] \n"
+ " sub %0, %0, 1 \n"
+ "3: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 3b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 4b \n"
+ " .previous \n"
+
+ : "+r" (n),
+ /* Note as an '&' earlyclobber operand to make sure the
+ * temporary register inside the loop is not the same as
+ * FROM or TO.
+ */
+ "=&r" (tmp), "+r" (to), "+r" (from)
+ :
+ : "lp_count", "lp_start", "lp_end", "memory");
+
+ return n;
+ }
+
+ if (__builtin_constant_p(orig_n)) {
+ res = orig_n;
+
+ if (orig_n / 16) {
+ orig_n = orig_n % 16;
+
+ __asm__ __volatile__(
+ " lsr lp_count, %7,4 \n"
+ " lp 3f \n"
+ " ld.ab %3, [%2, 4] \n"
+ " ld.ab %4, [%2, 4] \n"
+ " ld.ab %5, [%2, 4] \n"
+ " ld.ab %6, [%2, 4] \n"
+ "1: st.ab %3, [%1, 4] \n"
+ "11: st.ab %4, [%1, 4] \n"
+ "12: st.ab %5, [%1, 4] \n"
+ "13: st.ab %6, [%1, 4] \n"
+ " sub %0, %0, 16 \n"
+ "3:;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 3b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 4b \n"
+ " .word 11b,4b \n"
+ " .word 12b,4b \n"
+ " .word 13b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from),
+ "=r"(tmp1), "=r"(tmp2), "=r"(tmp3), "=r"(tmp4)
+ : "ir"(n)
+ : "lp_count", "memory");
+ }
+ if (orig_n / 8) {
+ orig_n = orig_n % 8;
+
+ __asm__ __volatile__(
+ " ld.ab %3, [%2,4] \n"
+ " ld.ab %4, [%2,4] \n"
+ "14: st.ab %3, [%1,4] \n"
+ "15: st.ab %4, [%1,4] \n"
+ " sub %0, %0, 8 \n"
+ "31:;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 31b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 14b,4b \n"
+ " .word 15b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from),
+ "=r"(tmp1), "=r"(tmp2)
+ :
+ : "memory");
+ }
+ if (orig_n / 4) {
+ orig_n = orig_n % 4;
+
+ __asm__ __volatile__(
+ " ld.ab %3, [%2,4] \n"
+ "16: st.ab %3, [%1,4] \n"
+ " sub %0, %0, 4 \n"
+ "32:;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 32b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 16b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from), "=r"(tmp1)
+ :
+ : "memory");
+ }
+ if (orig_n / 2) {
+ orig_n = orig_n % 2;
+
+ __asm__ __volatile__(
+ " ldw.ab %3, [%2,2] \n"
+ "17: stw.ab %3, [%1,2] \n"
+ " sub %0, %0, 2 \n"
+ "33:;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 33b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 17b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from), "=r"(tmp1)
+ :
+ : "memory");
+ }
+ if (orig_n & 1) {
+ __asm__ __volatile__(
+ " ldb.ab %3, [%2,1] \n"
+ "18: stb.ab %3, [%1,1] \n"
+ " sub %0, %0, 1 \n"
+ "34: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 34b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 18b,4b \n"
+ " .previous \n"
+ : "+r" (res), "+r"(to), "+r"(from), "=r"(tmp1)
+ :
+ : "memory");
+ }
+ } else { /* n is NOT constant, so laddered copy of 16x,8,4,2,1 */
+
+ __asm__ __volatile__(
+ " mov %0,%3 \n"
+ " lsr.f lp_count, %3,4 \n" /* 16x bytes */
+ " lpnz 3f \n"
+ " ld.ab %5, [%2, 4] \n"
+ " ld.ab %6, [%2, 4] \n"
+ " ld.ab %7, [%2, 4] \n"
+ " ld.ab %8, [%2, 4] \n"
+ "1: st.ab %5, [%1, 4] \n"
+ "11: st.ab %6, [%1, 4] \n"
+ "12: st.ab %7, [%1, 4] \n"
+ "13: st.ab %8, [%1, 4] \n"
+ " sub %0, %0, 16 \n"
+ "3: and.f %3,%3,0xf \n" /* stragglers */
+ " bz 34f \n"
+ " bbit0 %3,3,31f \n" /* 8 bytes left */
+ " ld.ab %5, [%2,4] \n"
+ " ld.ab %6, [%2,4] \n"
+ "14: st.ab %5, [%1,4] \n"
+ "15: st.ab %6, [%1,4] \n"
+ " sub.f %0, %0, 8 \n"
+ "31: bbit0 %3,2,32f \n" /* 4 bytes left */
+ " ld.ab %5, [%2,4] \n"
+ "16: st.ab %5, [%1,4] \n"
+ " sub.f %0, %0, 4 \n"
+ "32: bbit0 %3,1,33f \n" /* 2 bytes left */
+ " ldw.ab %5, [%2,2] \n"
+ "17: stw.ab %5, [%1,2] \n"
+ " sub.f %0, %0, 2 \n"
+ "33: bbit0 %3,0,34f \n"
+ " ldb.ab %5, [%2,1] \n" /* 1 byte left */
+ "18: stb.ab %5, [%1,1] \n"
+ " sub.f %0, %0, 1 \n"
+ "34: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: j 34b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 4b \n"
+ " .word 11b,4b \n"
+ " .word 12b,4b \n"
+ " .word 13b,4b \n"
+ " .word 14b,4b \n"
+ " .word 15b,4b \n"
+ " .word 16b,4b \n"
+ " .word 17b,4b \n"
+ " .word 18b,4b \n"
+ " .previous \n"
+ : "=r" (res), "+r"(to), "+r"(from), "+r"(n), "=r"(val),
+ "=r"(tmp1), "=r"(tmp2), "=r"(tmp3), "=r"(tmp4)
+ :
+ : "lp_count", "memory");
+ }
+
+ return res;
+}
+
+static inline unsigned long __arc_clear_user(void __user *to, unsigned long n)
+{
+ long res = n;
+ unsigned char *d_char = to;
+
+ __asm__ __volatile__(
+ " bbit0 %0, 0, 1f \n"
+ "75: stb.ab %2, [%0,1] \n"
+ " sub %1, %1, 1 \n"
+ "1: bbit0 %0, 1, 2f \n"
+ "76: stw.ab %2, [%0,2] \n"
+ " sub %1, %1, 2 \n"
+ "2: asr.f lp_count, %1, 2 \n"
+ " lpnz 3f \n"
+ "77: st.ab %2, [%0,4] \n"
+ " sub %1, %1, 4 \n"
+ "3: bbit0 %1, 1, 4f \n"
+ "78: stw.ab %2, [%0,2] \n"
+ " sub %1, %1, 2 \n"
+ "4: bbit0 %1, 0, 5f \n"
+ "79: stb.ab %2, [%0,1] \n"
+ " sub %1, %1, 1 \n"
+ "5: \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "3: j 5b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 75b, 3b \n"
+ " .word 76b, 3b \n"
+ " .word 77b, 3b \n"
+ " .word 78b, 3b \n"
+ " .word 79b, 3b \n"
+ " .previous \n"
+ : "+r"(d_char), "+r"(res)
+ : "i"(0)
+ : "lp_count", "lp_start", "lp_end", "memory");
+
+ return res;
+}
+
+static inline long
+__arc_strncpy_from_user(char *dst, const char __user *src, long count)
+{
+ long res = count;
+ char val;
+ unsigned int hw_count;
+
+ if (count == 0)
+ return 0;
+
+ __asm__ __volatile__(
+ " lp 2f \n"
+ "1: ldb.ab %3, [%2, 1] \n"
+ " breq.d %3, 0, 2f \n"
+ " stb.ab %3, [%1, 1] \n"
+ "2: sub %0, %6, %4 \n"
+ "3: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: mov %0, %5 \n"
+ " j 3b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 4b \n"
+ " .previous \n"
+ : "=r"(res), "+r"(dst), "+r"(src), "=&r"(val), "=l"(hw_count)
+ : "g"(-EFAULT), "ir"(count), "4"(count) /* this "4" seeds lp_count */
+ : "memory");
+
+ return res;
+}
+
+static inline long __arc_strnlen_user(const char __user *s, long n)
+{
+ long res, tmp1, cnt;
+ char val;
+
+ __asm__ __volatile__(
+ " mov %2, %1 \n"
+ "1: ldb.ab %3, [%0, 1] \n"
+ " breq.d %3, 0, 2f \n"
+ " sub.f %2, %2, 1 \n"
+ " bnz 1b \n"
+ " sub %2, %2, 1 \n"
+ "2: sub %0, %1, %2 \n"
+ "3: ;nop \n"
+ " .section .fixup, \"ax\" \n"
+ " .align 4 \n"
+ "4: mov %0, 0 \n"
+ " j 3b \n"
+ " .previous \n"
+ " .section __ex_table, \"a\" \n"
+ " .align 4 \n"
+ " .word 1b, 4b \n"
+ " .previous \n"
+ : "=r"(res), "=r"(tmp1), "=r"(cnt), "=r"(val)
+ : "0"(s), "1"(n)
+ : "memory");
+
+ return res;
+}
+
+#define __copy_from_user(t, f, n) __arc_copy_from_user(t, f, n)
+#define __copy_to_user(t, f, n) __arc_copy_to_user(t, f, n)
+#define __clear_user(d, n) __arc_clear_user(d, n)
+#define __strncpy_from_user(d, s, n) __arc_strncpy_from_user(d, s, n)
+#define __strnlen_user(s, n) __arc_strnlen_user(s, n)
+
+#include <asm-generic/uaccess.h>
+
+extern int fixup_exception(struct pt_regs *regs);
+
+#endif
--
1.7.4.1
This covers the UP / SMP (with no hardware assist for atomic r-m-w) as
well as ARC700 LLOCK/SCOND insns based.
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/atomic.h | 232 ++++++++++++++++++
arch/arc/include/asm/barrier.h | 42 ++++
arch/arc/include/asm/bitops.h | 507 ++++++++++++++++++++++++++++++++++++++++
arch/arc/include/asm/cmpxchg.h | 143 +++++++++++
arch/arc/include/asm/smp.h | 34 +++
5 files changed, 958 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/atomic.h
create mode 100644 arch/arc/include/asm/barrier.h
create mode 100644 arch/arc/include/asm/bitops.h
create mode 100644 arch/arc/include/asm/cmpxchg.h
create mode 100644 arch/arc/include/asm/smp.h
diff --git a/arch/arc/include/asm/atomic.h b/arch/arc/include/asm/atomic.h
new file mode 100644
index 0000000..543d496
--- /dev/null
+++ b/arch/arc/include/asm/atomic.h
@@ -0,0 +1,232 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_ATOMIC_H
+#define _ASM_ARC_ATOMIC_H
+
+#if defined(__KERNEL__)
+
+#if !defined(__ASSEMBLY__)
+
+#include <linux/types.h>
+#include <linux/compiler.h>
+#include <asm/cmpxchg.h>
+#include <asm/barrier.h>
+#include <asm/smp.h>
+
+#define atomic_read(v) ((v)->counter)
+
+#ifdef CONFIG_ARC_HAS_LLSC
+
+#define atomic_set(v, i) (((v)->counter) = (i))
+
+static inline void atomic_add(int i, atomic_t *v)
+{
+ unsigned int temp;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " add %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp) /* Early clobber, to prevent reg reuse */
+ : "r"(&v->counter), "ir"(i)
+ : "cc");
+}
+
+static inline void atomic_sub(int i, atomic_t *v)
+{
+ unsigned int temp;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " sub %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp)
+ : "r"(&v->counter), "ir"(i)
+ : "cc");
+}
+
+/* add and also return the new value */
+static inline int atomic_add_return(int i, atomic_t *v)
+{
+ unsigned int temp;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " add %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp)
+ : "r"(&v->counter), "ir"(i)
+ : "cc");
+
+ return temp;
+}
+
+static inline int atomic_sub_return(int i, atomic_t *v)
+{
+ unsigned int temp;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " sub %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp)
+ : "r"(&v->counter), "ir"(i)
+ : "cc");
+
+ return temp;
+}
+
+static inline void atomic_clear_mask(unsigned long mask, unsigned long *addr)
+{
+ unsigned int temp;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " bic %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp)
+ : "r"(addr), "ir"(mask)
+ : "cc");
+}
+
+#else /* !CONFIG_ARC_HAS_LLSC */
+
+#ifndef CONFIG_SMP
+
+ /* violating atomic_xxx API locking protocol in UP for optimization sake */
+#define atomic_set(v, i) (((v)->counter) = (i))
+
+#else
+
+static inline void atomic_set(atomic_t *v, int i)
+{
+ /*
+ * Independent of hardware support, all of the atomic_xxx() APIs need
+ * to follow the same locking rules to make sure that a "hardware"
+ * atomic insn (e.g. LD) doesn't clobber an "emulated" atomic insn
+ * sequence
+ *
+ * Thus atomic_set() despite being 1 insn (and seemingly atomic)
+ * requires the locking.
+ */
+ unsigned long flags;
+
+ atomic_ops_lock(flags);
+ v->counter = i;
+ atomic_ops_unlock(flags);
+}
+#endif
+
+/*
+ * Non hardware assisted Atomic-R-M-W
+ * Locking would change to irq-disabling only (UP) and spinlocks (SMP)
+ */
+
+static inline void atomic_add(int i, atomic_t *v)
+{
+ unsigned long flags;
+
+ atomic_ops_lock(flags);
+ v->counter += i;
+ atomic_ops_unlock(flags);
+}
+
+static inline void atomic_sub(int i, atomic_t *v)
+{
+ unsigned long flags;
+
+ atomic_ops_lock(flags);
+ v->counter -= i;
+ atomic_ops_unlock(flags);
+}
+
+static inline int atomic_add_return(int i, atomic_t *v)
+{
+ unsigned long flags;
+ unsigned long temp;
+
+ atomic_ops_lock(flags);
+ temp = v->counter;
+ temp += i;
+ v->counter = temp;
+ atomic_ops_unlock(flags);
+
+ return temp;
+}
+
+static inline int atomic_sub_return(int i, atomic_t *v)
+{
+ unsigned long flags;
+ unsigned long temp;
+
+ atomic_ops_lock(flags);
+ temp = v->counter;
+ temp -= i;
+ v->counter = temp;
+ atomic_ops_unlock(flags);
+
+ return temp;
+}
+
+static inline void atomic_clear_mask(unsigned long mask, unsigned long *addr)
+{
+ unsigned long flags;
+
+ atomic_ops_lock(flags);
+ *addr &= ~mask;
+ atomic_ops_unlock(flags);
+}
+
+#endif /* !CONFIG_ARC_HAS_LLSC */
+
+/**
+ * __atomic_add_unless - add unless the number is a given value
+ * @v: pointer of type atomic_t
+ * @a: the amount to add to v...
+ * @u: ...unless v is equal to u.
+ *
+ * Atomically adds @a to @v, so long as it was not @u.
+ * Returns the old value of @v
+ */
+#define __atomic_add_unless(v, a, u) \
+({ \
+ int c, old; \
+ c = atomic_read(v); \
+ while (c != (u) && (old = atomic_cmpxchg((v), c, c + (a))) != c)\
+ c = old; \
+ c; \
+})
+
+#define atomic_inc_not_zero(v) atomic_add_unless((v), 1, 0)
+
+#define atomic_inc(v) atomic_add(1, v)
+#define atomic_dec(v) atomic_sub(1, v)
+
+#define atomic_inc_and_test(v) (atomic_add_return(1, v) == 0)
+#define atomic_dec_and_test(v) (atomic_sub_return(1, v) == 0)
+#define atomic_inc_return(v) atomic_add_return(1, (v))
+#define atomic_dec_return(v) atomic_sub_return(1, (v))
+#define atomic_sub_and_test(i, v) (atomic_sub_return(i, v) == 0)
+
+#define atomic_add_negative(i, v) (atomic_add_return(i, v) < 0)
+
+#define ATOMIC_INIT(i) { (i) }
+
+#include <asm-generic/atomic64.h>
+
+#endif
+
+#endif
+
+#endif
diff --git a/arch/arc/include/asm/barrier.h b/arch/arc/include/asm/barrier.h
new file mode 100644
index 0000000..f6cb7c4
--- /dev/null
+++ b/arch/arc/include/asm/barrier.h
@@ -0,0 +1,42 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_BARRIER_H
+#define __ASM_BARRIER_H
+
+#ifndef __ASSEMBLY__
+
+/* TODO-vineetg: Need to see what this does, don't we need sync anywhere */
+#define mb() __asm__ __volatile__ ("" : : : "memory")
+#define rmb() mb()
+#define wmb() mb()
+#define set_mb(var, value) do { var = value; mb(); } while (0)
+#define set_wmb(var, value) do { var = value; wmb(); } while (0)
+#define read_barrier_depends() mb()
+
+/* TODO-vineetg verify the correctness of macros here */
+#ifdef CONFIG_SMP
+#define smp_mb() mb()
+#define smp_rmb() rmb()
+#define smp_wmb() wmb()
+#else
+#define smp_mb() barrier()
+#define smp_rmb() barrier()
+#define smp_wmb() barrier()
+#endif
+
+#define smp_mb__before_atomic_dec() barrier()
+#define smp_mb__after_atomic_dec() barrier()
+#define smp_mb__before_atomic_inc() barrier()
+#define smp_mb__after_atomic_inc() barrier()
+
+#define smp_read_barrier_depends() do { } while (0)
+
+#endif
+
+#endif
diff --git a/arch/arc/include/asm/bitops.h b/arch/arc/include/asm/bitops.h
new file mode 100644
index 0000000..51cffcc
--- /dev/null
+++ b/arch/arc/include/asm/bitops.h
@@ -0,0 +1,507 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_BITOPS_H
+#define _ASM_BITOPS_H
+
+#ifndef _LINUX_BITOPS_H
+#error only <linux/bitops.h> can be included directly
+#endif
+
+#if defined(__KERNEL__)
+
+#if !defined(__ASSEMBLY__)
+
+#include <linux/types.h>
+#include <linux/compiler.h>
+
+/*
+ * Hardware assisted read-modify-write using ARC700 LLOCK/SCOND insns.
+ * The Kconfig glue ensures that in SMP, this is only set if the container
+ * SoC/platform has cross-core coherent LLOCK/SCOND
+ */
+#if defined(CONFIG_ARC_HAS_LLSC)
+
+static inline void set_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned int temp;
+
+ m += nr >> 5;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " bset %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp)
+ : "r"(m), "ir"(nr)
+ : "cc");
+}
+
+static inline void clear_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned int temp;
+
+ m += nr >> 5;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " bclr %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp)
+ : "r"(m), "ir"(nr)
+ : "cc");
+}
+
+static inline void change_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned int temp;
+
+ m += nr >> 5;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " bxor %0, %0, %2 \n"
+ " scond %0, [%1] \n"
+ " bnz 1b \n"
+ : "=&r"(temp)
+ : "r"(m), "ir"(nr)
+ : "cc");
+}
+
+/*
+ * Semantically:
+ * Test the bit
+ * if clear
+ * set it and return 0 (old value)
+ * else
+ * return 1 (old value).
+ *
+ * Since ARC lacks a equivalent h/w primitive, the bit is set unconditionally
+ * and the old value of bit is returned
+ */
+static inline int test_and_set_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long old, temp;
+
+ m += nr >> 5;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%2] \n"
+ " bset %1, %0, %3 \n"
+ " scond %1, [%2] \n"
+ " bnz 1b \n"
+ : "=&r"(old), "=&r"(temp)
+ : "r"(m), "ir"(nr)
+ : "cc");
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ return (old & (1 << nr)) != 0;
+}
+
+static inline int
+test_and_clear_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned int old, temp;
+
+ m += nr >> 5;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%2] \n"
+ " bclr %1, %0, %3 \n"
+ " scond %1, [%2] \n"
+ " bnz 1b \n"
+ : "=&r"(old), "=&r"(temp)
+ : "r"(m), "ir"(nr)
+ : "cc");
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ return (old & (1 << nr)) != 0;
+}
+
+static inline int
+test_and_change_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned int old, temp;
+
+ m += nr >> 5;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%2] \n"
+ " bxor %1, %0, %3 \n"
+ " scond %1, [%2] \n"
+ " bnz 1b \n"
+ : "=&r"(old), "=&r"(temp)
+ : "r"(m), "ir"(nr)
+ : "cc");
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ return (old & (1 << nr)) != 0;
+}
+
+#else /* !CONFIG_ARC_HAS_LLSC */
+
+#include <asm/smp.h>
+
+/*
+ * Non hardware assisted Atomic-R-M-W
+ * Locking would change to irq-disabling only (UP) and spinlocks (SMP)
+ *
+ * There's "significant" micro-optimization in writing our own variants of
+ * bitops (over generic variants)
+ *
+ * (1) The generic APIs have "signed" @nr while we have it "unsigned"
+ * This avoids extra code to be generated for pointer arithmatic, since
+ * is "not sure" that index is NOT -ve
+ * (2) Utilize the fact that ARCompact bit fidding insn (BSET/BCLR/ASL) etc
+ * only consider bottom 5 bits of @nr, so NO need to mask them off.
+ * (GCC Quirk: however for constant @nr we still need to do the masking
+ * at compile time)
+ */
+
+static inline void set_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long temp, flags;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ bitops_lock(flags);
+
+ temp = *m;
+ *m = temp | (1UL << nr);
+
+ bitops_unlock(flags);
+}
+
+static inline void clear_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long temp, flags;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ bitops_lock(flags);
+
+ temp = *m;
+ *m = temp & ~(1UL << nr);
+
+ bitops_unlock(flags);
+}
+
+static inline void change_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long temp, flags;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ bitops_lock(flags);
+
+ temp = *m;
+ *m = temp ^ (1UL << nr);
+
+ bitops_unlock(flags);
+}
+
+static inline int test_and_set_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long old, flags;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ bitops_lock(flags);
+
+ old = *m;
+ *m = old | (1 << nr);
+
+ bitops_unlock(flags);
+
+ return (old & (1 << nr)) != 0;
+}
+
+static inline int
+test_and_clear_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long old, flags;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ bitops_lock(flags);
+
+ old = *m;
+ *m = old | (1 << nr);
+
+ bitops_unlock(flags);
+
+ return (old & (1 << nr)) != 0;
+}
+
+static inline int
+test_and_change_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long old, flags;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ bitops_lock(flags);
+
+ old = *m;
+ *m = old ^ (1 << nr);
+
+ bitops_unlock(flags);
+
+ return (old & (1 << nr)) != 0;
+}
+
+#endif /* CONFIG_ARC_HAS_LLSC */
+
+/***************************************
+ * Non atomic variants
+ **************************************/
+
+static inline void __set_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long temp;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ temp = *m;
+ *m = temp | (1UL << nr);
+}
+
+static inline void __clear_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long temp;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ temp = *m;
+ *m = temp & ~(1UL << nr);
+}
+
+static inline void __change_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long temp;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ temp = *m;
+ *m = temp ^ (1UL << nr);
+}
+
+static inline int
+__test_and_set_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long old;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ old = *m;
+ *m = old | (1 << nr);
+
+ return (old & (1 << nr)) != 0;
+}
+
+static inline int
+__test_and_clear_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long old;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ old = *m;
+ *m = old & ~(1 << nr);
+
+ return (old & (1 << nr)) != 0;
+}
+
+static inline int
+__test_and_change_bit(unsigned long nr, volatile unsigned long *m)
+{
+ unsigned long old;
+ m += nr >> 5;
+
+ if (__builtin_constant_p(nr))
+ nr &= 0x1f;
+
+ old = *m;
+ *m = old ^ (1 << nr);
+
+ return (old & (1 << nr)) != 0;
+}
+
+/*
+ * This routine doesn't need to be atomic.
+ */
+static inline int
+__constant_test_bit(unsigned int nr, const volatile unsigned long *addr)
+{
+ return ((1UL << (nr & 31)) &
+ (((const volatile unsigned int *)addr)[nr >> 5])) != 0;
+}
+
+static inline int
+__test_bit(unsigned int nr, const volatile unsigned long *addr)
+{
+ unsigned long mask;
+
+ addr += nr >> 5;
+
+ /* ARC700 only considers 5 bits in bit-fiddling insn */
+ mask = 1 << nr;
+
+ return ((mask & *addr) != 0);
+}
+
+#define test_bit(nr, addr) (__builtin_constant_p(nr) ? \
+ __constant_test_bit((nr), (addr)) : \
+ __test_bit((nr), (addr)))
+
+/*
+ * Count the number of zeros, starting from MSB
+ * Helper for fls( ) friends
+ * This is a pure count, so (1-32) or (0-31) doesn't apply
+ * It could be 0 to 32, based on num of 0's in there
+ * clz(0x8000_0000) = 0, clz(0xFFFF_FFFF)=0, clz(0) = 32, clz(1) = 31
+ */
+static inline __attribute__ ((const)) int clz(unsigned int x)
+{
+ unsigned int res;
+
+ __asm__ __volatile__(
+ " norm.f %0, %1 \n"
+ " mov.n %0, 0 \n"
+ " add.p %0, %0, 1 \n"
+ : "=r"(res)
+ : "r"(x)
+ : "cc");
+
+ return res;
+}
+
+static inline int constant_fls(int x)
+{
+ int r = 32;
+
+ if (!x)
+ return 0;
+ if (!(x & 0xffff0000u)) {
+ x <<= 16;
+ r -= 16;
+ }
+ if (!(x & 0xff000000u)) {
+ x <<= 8;
+ r -= 8;
+ }
+ if (!(x & 0xf0000000u)) {
+ x <<= 4;
+ r -= 4;
+ }
+ if (!(x & 0xc0000000u)) {
+ x <<= 2;
+ r -= 2;
+ }
+ if (!(x & 0x80000000u)) {
+ x <<= 1;
+ r -= 1;
+ }
+ return r;
+}
+
+/*
+ * fls = Find Last Set in word
+ * @result: [1-32]
+ * fls(1) = 1, fls(0x80000000) = 32, fls(0) = 0
+ */
+static inline __attribute__ ((const)) int fls(unsigned long x)
+{
+ if (__builtin_constant_p(x))
+ return constant_fls(x);
+
+ return 32 - clz(x);
+}
+
+/*
+ * __fls: Similar to fls, but zero based (0-31)
+ */
+static inline __attribute__ ((const)) int __fls(unsigned long x)
+{
+ if (!x)
+ return 0;
+ else
+ return fls(x) - 1;
+}
+
+/*
+ * ffs = Find First Set in word (LSB to MSB)
+ * @result: [1-32], 0 if all 0's
+ */
+#define ffs(x) ({ unsigned long __t = (x); fls(__t & -__t); })
+
+/*
+ * __ffs: Similar to ffs, but zero based (0-31)
+ */
+static inline __attribute__ ((const)) int __ffs(unsigned long word)
+{
+ if (!word)
+ return word;
+
+ return ffs(word) - 1;
+}
+
+/*
+ * ffz = Find First Zero in word.
+ * @return:[0-31], 32 if all 1's
+ */
+#define ffz(x) __ffs(~(x))
+
+/* TODO does this affect uni-processor code */
+#define smp_mb__before_clear_bit() barrier()
+#define smp_mb__after_clear_bit() barrier()
+
+#include <asm-generic/bitops/hweight.h>
+#include <asm-generic/bitops/fls64.h>
+#include <asm-generic/bitops/sched.h>
+#include <asm-generic/bitops/lock.h>
+
+#include <asm-generic/bitops/find.h>
+#include <asm-generic/bitops/le.h>
+#include <asm-generic/bitops/ext2-atomic-setbit.h>
+
+#endif /* !__ASSEMBLY__ */
+
+#endif /* __KERNEL__ */
+
+#endif
diff --git a/arch/arc/include/asm/cmpxchg.h b/arch/arc/include/asm/cmpxchg.h
new file mode 100644
index 0000000..03cd689
--- /dev/null
+++ b/arch/arc/include/asm/cmpxchg.h
@@ -0,0 +1,143 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_CMPXCHG_H
+#define __ASM_ARC_CMPXCHG_H
+
+#include <linux/types.h>
+#include <asm/smp.h>
+
+#ifdef CONFIG_ARC_HAS_LLSC
+
+static inline unsigned long
+__cmpxchg(volatile void *ptr, unsigned long expected, unsigned long new)
+{
+ unsigned long prev;
+
+ __asm__ __volatile__(
+ "1: llock %0, [%1] \n"
+ " brne %0, %2, 2f \n"
+ " scond %3, [%1] \n"
+ " bnz 1b \n"
+ "2: \n"
+ : "=&r"(prev)
+ : "r"(ptr), "ir"(expected),
+ "r"(new) /* can't be "ir". scond can't take limm for "b" */
+ : "cc");
+
+ return prev;
+}
+
+#else
+
+static inline unsigned long
+__cmpxchg(volatile void *ptr, unsigned long expected, unsigned long new)
+{
+ unsigned long flags;
+ int prev;
+ volatile unsigned long *p = ptr;
+
+ atomic_ops_lock(flags);
+ prev = *p;
+ if (prev == expected)
+ *p = new;
+ atomic_ops_unlock(flags);
+ return prev;
+}
+
+#endif /* CONFIG_ARC_HAS_LLSC */
+
+#define cmpxchg(ptr, o, n) ((typeof(*(ptr)))__cmpxchg((ptr), \
+ (unsigned long)(o), (unsigned long)(n)))
+
+/*
+ * Since not supported natively, ARC cmpxchg() uses atomic_ops_lock (UP/SMP)
+ * just to gaurantee semantics.
+ * atomic_cmpxchg() needs to use the same locks as it's other atomic siblings
+ * which also happens to be atomic_ops_lock.
+ *
+ * Thus despite semantically being different, implementation of atomic_cmpxchg()
+ * is same as cmpxchg().
+ */
+#define atomic_cmpxchg(v, o, n) ((int)cmpxchg(&((v)->counter), (o), (n)))
+
+
+/*
+ * xchg (reg with memory) based on "Native atomic" EX insn
+ */
+static inline unsigned long __xchg(unsigned long val, volatile void *ptr,
+ int size)
+{
+ extern unsigned long __xchg_bad_pointer(void);
+
+ switch (size) {
+ case 4:
+ __asm__ __volatile__(
+ " ex %0, [%1] \n"
+ : "+r"(val)
+ : "r"(ptr)
+ : "memory");
+
+ return val;
+ }
+ return __xchg_bad_pointer();
+}
+
+#define _xchg(ptr, with) ((typeof(*(ptr)))__xchg((unsigned long)(with), (ptr), \
+ sizeof(*(ptr))))
+
+/*
+ * On ARC700, EX insn is inherently atomic, so by default "vanilla" xchg() need
+ * not require any locking. However there's a quirk.
+ * ARC lacks native CMPXCHG, thus emulated (see above), using external locking -
+ * incidently it "reuses" the same atomic_ops_lock used by atomic APIs.
+ * Now, llist code uses cmpxchg() and xchg() on same data, so xchg() needs to
+ * abide by same serializing rules, thus ends up using atomic_ops_lock as well.
+ *
+ * This however is only relevant if SMP and/or ARC lacks LLSC
+ * if (UP or LLSC)
+ * xchg doesn't need serialization
+ * else <==> !(UP or LLSC) <==> (!UP and !LLSC) <==> (SMP and !LLSC)
+ * xchg needs serialization
+ */
+
+#if !defined(CONFIG_ARC_HAS_LLSC) && defined(CONFIG_SMP)
+
+#define xchg(ptr, with) \
+({ \
+ unsigned long flags; \
+ typeof(*(ptr)) old_val; \
+ \
+ atomic_ops_lock(flags); \
+ old_val = _xchg(ptr, with); \
+ atomic_ops_unlock(flags); \
+ old_val; \
+})
+
+#else
+
+#define xchg(ptr, with) _xchg(ptr, with)
+
+#endif
+
+/*
+ * "atomic" variant of xchg()
+ * REQ: It needs to follow the same serialization rules as other atomic_xxx()
+ * Since xchg() doesn't always do that, it would seem that following defintion
+ * is incorrect. But here's the rationale:
+ * SMP : Even xchg() takes the atomic_ops_lock, so OK.
+ * LLSC: atomic_ops_lock are not relevent at all (even if SMP, since LLSC
+ * is natively "SMP safe", no serialization required).
+ * UP : other atomics disable IRQ, so no way a difft ctxt atomic_xchg()
+ * could clobber them. atomic_xchg() itself would be 1 insn, so it
+ * can't be clobbered by others. Thus no serialization required when
+ * atomic_xchg is involved.
+ */
+#define atomic_xchg(v, new) (xchg(&((v)->counter), new))
+
+#endif
diff --git a/arch/arc/include/asm/smp.h b/arch/arc/include/asm/smp.h
new file mode 100644
index 0000000..4341f3b
--- /dev/null
+++ b/arch/arc/include/asm/smp.h
@@ -0,0 +1,34 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_SMP_H
+#define __ASM_ARC_SMP_H
+
+/*
+ * ARC700 doesn't support atomic Read-Modify-Write ops.
+ * Originally Interrupts had to be disabled around code to gaurantee atomicity.
+ * The LLOCK/SCOND insns allow writing interrupt-hassle-free based atomic ops
+ * based on retry-if-irq-in-atomic (with hardware assist).
+ * However despite these, we provide the IRQ disabling variant
+ *
+ * (1) These insn were introduced only in 4.10 release. So for older released
+ * support needed.
+ */
+#ifndef CONFIG_ARC_HAS_LLSC
+
+#include <linux/irqflags.h>
+
+#define atomic_ops_lock(flags) local_irq_save(flags)
+#define atomic_ops_unlock(flags) local_irq_restore(flags)
+
+#define bitops_lock(flags) local_irq_save(flags)
+#define bitops_unlock(flags) local_irq_restore(flags)
+
+#endif /* !CONFIG_ARC_HAS_LLSC */
+
+#endif
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/Kbuild | 57 ++++++++++++++++++++++++++++++++++++
arch/arc/include/asm/asm-offsets.h | 9 ++++++
2 files changed, 66 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/Kbuild
create mode 100644 arch/arc/include/asm/asm-offsets.h
diff --git a/arch/arc/include/asm/Kbuild b/arch/arc/include/asm/Kbuild
new file mode 100644
index 0000000..d6c037a
--- /dev/null
+++ b/arch/arc/include/asm/Kbuild
@@ -0,0 +1,57 @@
+include include/asm-generic/Kbuild.asm
+
+# 7-Oct-12: Jeremy Bennett <[email protected]>. Some of these
+# headers, beyond those specified in the generic set are needed by user code.
+
+header-y += page.h
+header-y += cachectl.h
+
+generic-y += auxvec.h
+generic-y += bugs.h
+generic-y += bitsperlong.h
+generic-y += clkdev.h
+generic-y += cputime.h
+generic-y += current.h
+generic-y += device.h
+generic-y += div64.h
+generic-y += emergency-restart.h
+generic-y += errno.h
+generic-y += fcntl.h
+generic-y += fb.h
+generic-y += ftrace.h
+generic-y += hardirq.h
+generic-y += hw_irq.h
+generic-y += ioctl.h
+generic-y += ioctls.h
+generic-y += ipcbuf.h
+generic-y += irq_regs.h
+generic-y += kmap_types.h
+generic-y += kvm_para.h
+generic-y += local.h
+generic-y += local64.h
+generic-y += mman.h
+generic-y += msgbuf.h
+generic-y += param.h
+generic-y += parport.h
+generic-y += percpu.h
+generic-y += poll.h
+generic-y += posix_types.h
+generic-y += resource.h
+generic-y += scatterlist.h
+generic-y += sembuf.h
+generic-y += shmbuf.h
+generic-y += shmparam.h
+generic-y += siginfo.h
+generic-y += socket.h
+generic-y += sockios.h
+generic-y += stat.h
+generic-y += statfs.h
+generic-y += termbits.h
+generic-y += termios.h
+generic-y += topology.h
+generic-y += types.h
+generic-y += ucontext.h
+generic-y += unaligned.h
+generic-y += user.h
+generic-y += vga.h
+generic-y += xor.h
diff --git a/arch/arc/include/asm/asm-offsets.h b/arch/arc/include/asm/asm-offsets.h
new file mode 100644
index 0000000..dad1876
--- /dev/null
+++ b/arch/arc/include/asm/asm-offsets.h
@@ -0,0 +1,9 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <generated/asm-offsets.h>
--
1.7.4.1
* MMU I-TLB / D-TLB Miss Exceptions
- Fast Path TLB Refill Handler
- slowpath TLB creation via do_page_fault() -> update_mmu_cache()
* Duplicate PD Exception Handler
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/arcregs.h | 90 ++++++++++
arch/arc/include/asm/tlb-mmu1.h | 104 ++++++++++++
arch/arc/include/asm/tlb.h | 41 +++++
arch/arc/mm/tlb.c | 252 ++++++++++++++++++++++++++++
arch/arc/mm/tlbex.S | 351 +++++++++++++++++++++++++++++++++++++++
5 files changed, 838 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/tlb-mmu1.h
create mode 100644 arch/arc/include/asm/tlb.h
create mode 100644 arch/arc/mm/tlbex.S
diff --git a/arch/arc/include/asm/arcregs.h b/arch/arc/include/asm/arcregs.h
index c12eb9b..aedfbf1 100644
--- a/arch/arc/include/asm/arcregs.h
+++ b/arch/arc/include/asm/arcregs.h
@@ -13,6 +13,7 @@
/* Build Configuration Registers */
#define ARC_REG_VECBASE_BCR 0x68
+#define ARC_REG_MMU_BCR 0x6f
/* status32 Bits Positions */
#define STATUS_H_BIT 0 /* CPU Halted */
@@ -36,6 +37,34 @@
#define STATUS_U_MASK (1<<STATUS_U_BIT)
#define STATUS_L_MASK (1<<STATUS_L_BIT)
+/*
+ * ECR: Exception Cause Reg bits-n-pieces
+ * [23:16] = Exception Vector
+ * [15: 8] = Exception Cause Code
+ * [ 7: 0] = Exception Parameters (for certain types only)
+ */
+#define ECR_VEC_MASK 0xff0000
+#define ECR_CODE_MASK 0x00ff00
+#define ECR_PARAM_MASK 0x0000ff
+
+/* Exception Cause Vector Values */
+#define ECR_V_MACH_CHK 0x20
+#define ECR_V_ITLB_MISS 0x21
+#define ECR_V_DTLB_MISS 0x22
+#define ECR_V_PROTV 0x23
+
+/* Protection Violation Exception Cause Code Values */
+#define ECR_C_PROTV_INST_FETCH 0x00
+#define ECR_C_PROTV_LOAD 0x01
+#define ECR_C_PROTV_STORE 0x02
+#define ECR_C_PROTV_XCHG 0x03
+#define ECR_C_PROTV_MISALIG_DATA 0x04
+
+/* DTLB Miss Exception Cause Code Values */
+#define ECR_C_BIT_DTLB_LD_MISS 8
+#define ECR_C_BIT_DTLB_ST_MISS 9
+
+
/* Auxiliary registers */
#define AUX_IDENTITY 4
#define AUX_INTR_VEC_BASE 0x25
@@ -58,6 +87,44 @@
#define TIMER_CTRL_IE (1 << 0) /* Interupt when Count reachs limit */
#define TIMER_CTRL_NH (1 << 1) /* Count only when CPU NOT halted */
+#if defined(CONFIG_ARC_MMU_V1)
+#define CONFIG_ARC_MMU_VER 1
+#elif defined(CONFIG_ARC_MMU_V2)
+#define CONFIG_ARC_MMU_VER 2
+#elif defined(CONFIG_ARC_MMU_V3)
+#define CONFIG_ARC_MMU_VER 3
+#else
+#error "Error: MMU ver"
+#endif
+
+/* MMU Management regs */
+#define ARC_REG_TLBPD0 0x405
+#define ARC_REG_TLBPD1 0x406
+#define ARC_REG_TLBINDEX 0x407
+#define ARC_REG_TLBCOMMAND 0x408
+#define ARC_REG_PID 0x409
+#define ARC_REG_SCRATCH_DATA0 0x418
+
+/* Bits in MMU PID register */
+#define MMU_ENABLE (1 << 31) /* Enable MMU for process */
+
+/* Error code if probe fails */
+#define TLB_LKUP_ERR 0x80000000
+
+/* TLB Commands */
+#define TLBWrite 0x1
+#define TLBRead 0x2
+#define TLBGetIndex 0x3
+#define TLBProbe 0x4
+
+#if (CONFIG_ARC_MMU_VER >= 2)
+#define TLBWriteNI 0x5 /* write JTLB without inv uTLBs */
+#define TLBIVUTLB 0x6 /* explicitly inv uTLBs */
+#else
+#undef TLBWriteNI /* These cmds don't exist on older MMU */
+#undef TLBIVUTLB
+#endif
+
/* Instruction cache related Auxiliary registers */
#define ARC_REG_IC_BCR 0x77 /* Build Config reg */
#define ARC_REG_IC_IVIC 0x10
@@ -205,6 +272,24 @@ struct arc_fpu {
* Build Configuration Registers, with encoded hardware config
*/
+struct bcr_mmu_1_2 {
+#ifdef CONFIG_CPU_BIG_ENDIAN
+ unsigned int ver:8, ways:4, sets:4, u_itlb:8, u_dtlb:8;
+#else
+ unsigned int u_dtlb:8, u_itlb:8, sets:4, ways:4, ver:8;
+#endif
+};
+
+struct bcr_mmu_3 {
+#ifdef CONFIG_CPU_BIG_ENDIAN
+ unsigned int ver:8, ways:4, sets:4, osm:1, reserv:3, pg_sz:4,
+ u_itlb:4, u_dtlb:4;
+#else
+ unsigned int u_dtlb:4, u_itlb:4, pg_sz:4, reserv:3, osm:1, sets:4,
+ ways:4, ver:8;
+#endif
+};
+
struct bcr_cache {
#ifdef CONFIG_CPU_BIG_ENDIAN
unsigned int pad:12, line_len:4, sz:4, config:4, ver:8;
@@ -218,12 +303,17 @@ struct bcr_cache {
* Generic structures to hold build configuration used at runtime
*/
+struct cpuinfo_arc_mmu {
+ unsigned int ver, pg_sz, sets, ways, u_dtlb, u_itlb, num_tlb;
+};
+
struct cpuinfo_arc_cache {
unsigned int has_aliasing, sz, line_len, assoc, ver;
};
struct cpuinfo_arc {
struct cpuinfo_arc_cache icache, dcache;
+ struct cpuinfo_arc_mmu mmu;
};
extern struct cpuinfo_arc cpuinfo_arc700[];
diff --git a/arch/arc/include/asm/tlb-mmu1.h b/arch/arc/include/asm/tlb-mmu1.h
new file mode 100644
index 0000000..a5ff961
--- /dev/null
+++ b/arch/arc/include/asm/tlb-mmu1.h
@@ -0,0 +1,104 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_TLB_MMU_V1_H__
+#define __ASM_TLB_MMU_V1_H__
+
+#if defined(__ASSEMBLY__) && defined(CONFIG_ARC_MMU_VER == 1)
+
+#include <asm/tlb.h>
+
+.macro TLB_WRITE_HEURISTICS
+
+#define JH_HACK1
+#undef JH_HACK2
+#undef JH_HACK3
+
+#ifdef JH_HACK3
+; Calculate set index for 2-way MMU
+; -avoiding use of GetIndex from MMU
+; and its unpleasant LFSR pseudo-random sequence
+;
+; r1 = TLBPD0 from TLB_RELOAD above
+;
+; -- jh_ex_way_set not cleared on startup
+; didn't want to change setup.c
+; hence extra instruction to clean
+;
+; -- should be in cache since in same line
+; as r0/r1 saves above
+;
+ld r0,[jh_ex_way_sel] ; victim pointer
+and r0,r0,1 ; clean
+xor.f r0,r0,1 ; flip
+st r0,[jh_ex_way_sel] ; store back
+asr r0,r1,12 ; get set # <<1, note bit 12=R=0
+or.nz r0,r0,1 ; set way bit
+and r0,r0,0xff ; clean
+sr r0,[ARC_REG_TLBINDEX]
+#endif
+
+#ifdef JH_HACK2
+; JH hack #2
+; Faster than hack #1 in non-thrash case, but hard-coded for 2-way MMU
+; Slower in thrash case (where it matters) because more code is executed
+; Inefficient due to two-register paradigm of this miss handler
+;
+/* r1 = data TLBPD0 at this point */
+lr r0,[eret] /* instruction address */
+xor r0,r0,r1 /* compare set # */
+and.f r0,r0,0x000fe000 /* 2-way MMU mask */
+bne 88f /* not in same set - no need to probe */
+
+lr r0,[eret] /* instruction address */
+and r0,r0,PAGE_MASK /* VPN of instruction address */
+; lr r1,[ARC_REG_TLBPD0] /* Data VPN+ASID - already in r1 from TLB_RELOAD*/
+and r1,r1,0xff /* Data ASID */
+or r0,r0,r1 /* Instruction address + Data ASID */
+
+lr r1,[ARC_REG_TLBPD0] /* save TLBPD0 containing data TLB*/
+sr r0,[ARC_REG_TLBPD0] /* write instruction address to TLBPD0 */
+sr TLBProbe, [ARC_REG_TLBCOMMAND] /* Look for instruction */
+lr r0,[ARC_REG_TLBINDEX] /* r0 = index where instruction is, if at all */
+sr r1,[ARC_REG_TLBPD0] /* restore TLBPD0 */
+
+xor r0,r0,1 /* flip bottom bit of data index */
+b.d 89f
+sr r0,[ARC_REG_TLBINDEX] /* and put it back */
+88:
+sr TLBGetIndex, [ARC_REG_TLBCOMMAND]
+89:
+#endif
+
+#ifdef JH_HACK1
+;
+; Always checks whether instruction will be kicked out by dtlb miss
+;
+mov_s r3, r1 ; save PD0 prepared by TLB_RELOAD in r3
+lr r0,[eret] /* instruction address */
+and r0,r0,PAGE_MASK /* VPN of instruction address */
+bmsk r1,r3,7 /* Data ASID, bits 7-0 */
+or_s r0,r0,r1 /* Instruction address + Data ASID */
+
+sr r0,[ARC_REG_TLBPD0] /* write instruction address to TLBPD0 */
+sr TLBProbe, [ARC_REG_TLBCOMMAND] /* Look for instruction */
+lr r0,[ARC_REG_TLBINDEX] /* r0 = index where instruction is, if at all */
+sr r3,[ARC_REG_TLBPD0] /* restore TLBPD0 */
+
+sr TLBGetIndex, [ARC_REG_TLBCOMMAND]
+lr r1,[ARC_REG_TLBINDEX] /* r1 = index where MMU wants to put data */
+cmp r0,r1 /* if no match on indices, go around */
+xor.eq r1,r1,1 /* flip bottom bit of data index */
+sr r1,[ARC_REG_TLBINDEX] /* and put it back */
+#endif
+
+.endm
+
+#endif
+
+#endif
diff --git a/arch/arc/include/asm/tlb.h b/arch/arc/include/asm/tlb.h
new file mode 100644
index 0000000..b571e12
--- /dev/null
+++ b/arch/arc/include/asm/tlb.h
@@ -0,0 +1,41 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_TLB_H
+#define _ASM_ARC_TLB_H
+
+#ifdef __KERNEL__
+
+#include <asm/pgtable.h>
+
+/* Masks for actual TLB "PD"s */
+#define PTE_BITS_IN_PD0 (_PAGE_GLOBAL | _PAGE_PRESENT)
+#define PTE_BITS_IN_PD1 (PAGE_MASK | _PAGE_CACHEABLE | \
+ _PAGE_EXECUTE | _PAGE_WRITE | _PAGE_READ | \
+ _PAGE_K_EXECUTE | _PAGE_K_WRITE | _PAGE_K_READ)
+
+#ifndef __ASSEMBLY__
+
+#include <linux/pagemap.h>
+#include <asm-generic/tlb.h>
+
+#ifdef CONFIG_ARC_DBG_TLB_PARANOIA
+void tlb_paranoid_check(unsigned int pid_sw, unsigned long address);
+#else
+#define tlb_paranoid_check(a, b)
+#endif
+
+void arc_mmu_init(void);
+extern char *arc_mmu_mumbojumbo(int cpu_id, char *buf, int len);
+void __init read_decode_mmu_bcr(void);
+
+#endif /* __ASSEMBLY__ */
+
+#endif /* __KERNEL__ */
+
+#endif /* _ASM_ARC_TLB_H */
diff --git a/arch/arc/mm/tlb.c b/arch/arc/mm/tlb.c
index f1edae2..41afc84 100644
--- a/arch/arc/mm/tlb.c
+++ b/arch/arc/mm/tlb.c
@@ -21,3 +21,255 @@ int asid_cache = FIRST_ASID;
* see get_new_mmu_context (asm-arc/mmu_context.h)
*/
struct mm_struct *asid_mm_map[NUM_ASID + 1];
+
+
+/*
+ * Routine to create a TLB entry
+ */
+void create_tlb(struct vm_area_struct *vma, unsigned long address, pte_t *ptep)
+{
+ unsigned long flags;
+ unsigned int idx, asid_or_sasid;
+ unsigned long pd0_flags;
+
+ local_irq_save(flags);
+
+ tlb_paranoid_check(vma->vm_mm->context.asid, address);
+
+ address &= PAGE_MASK;
+
+ /* update this PTE credentials */
+ pte_val(*ptep) |= (_PAGE_PRESENT | _PAGE_ACCESSED);
+
+ /* Create HW TLB entry Flags (in PD0) from PTE Flags */
+#if (CONFIG_ARC_MMU_VER <= 2)
+ pd0_flags = ((pte_val(*ptep) & PTE_BITS_IN_PD0) >> 1);
+#else
+ pd0_flags = ((pte_val(*ptep) & PTE_BITS_IN_PD0));
+#endif
+
+ /* ASID for this task */
+ asid_or_sasid = read_aux_reg(ARC_REG_PID) & 0xff;
+
+ write_aux_reg(ARC_REG_TLBPD0, address | pd0_flags | asid_or_sasid);
+
+ /* Load remaining info in PD1 (Page Frame Addr and Kx/Kw/Kr Flags) */
+ write_aux_reg(ARC_REG_TLBPD1, (pte_val(*ptep) & PTE_BITS_IN_PD1));
+
+ /* First verify if entry for this vaddr+ASID already exists */
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBProbe);
+ idx = read_aux_reg(ARC_REG_TLBINDEX);
+
+ /*
+ * If Not already present get a free slot from MMU.
+ * Otherwise, Probe would have located the entry and set INDEX Reg
+ * with existing location. This will cause Write CMD to over-write
+ * existing entry with new PD0 and PD1
+ */
+ if (likely(idx & TLB_LKUP_ERR))
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBGetIndex);
+
+ /*
+ * Commit the Entry to MMU
+ * It doesnt sound safe to use the TLBWriteNI cmd here
+ * which doesn't flush uTLBs. I'd rather be safe than sorry.
+ */
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBWrite);
+
+ local_irq_restore(flags);
+}
+
+/* arch hook called by core VM at the end of handle_mm_fault( ),
+ * when a new PTE is entered in Page Tables or an existing one
+ * is modified. We aggresively pre-install a TLB entry
+ */
+
+void update_mmu_cache(struct vm_area_struct *vma, unsigned long vaddress,
+ pte_t *ptep)
+{
+ /* XXX: This definitely seems useless */
+#if 0
+ BUG_ON(!pfn_valid(pte_pfn(*ptep)));
+#endif
+
+ /* XXX: This is useful - but only once during execve - check why?
+ * handle_mm_fault()
+ * __get_user_pages
+ * copy_strings()
+ * do_execve()
+ */
+ if (current->active_mm != vma->vm_mm)
+ return;
+
+ create_tlb(vma, vaddress, ptep);
+}
+
+/* Read the Cache Build Confuration Registers, Decode them and save into
+ * the cpuinfo structure for later use.
+ * No Validation is done here, simply read/convert the BCRs
+ */
+void __init read_decode_mmu_bcr(void)
+{
+ unsigned int tmp;
+ struct bcr_mmu_1_2 *mmu2; /* encoded MMU2 attr */
+ struct bcr_mmu_3 *mmu3; /* encoded MMU3 attr */
+ struct cpuinfo_arc_mmu *mmu = &cpuinfo_arc700[smp_processor_id()].mmu;
+
+ tmp = read_aux_reg(ARC_REG_MMU_BCR);
+ mmu->ver = (tmp >> 24);
+
+ if (mmu->ver <= 2) {
+ mmu2 = (struct bcr_mmu_1_2 *)&tmp;
+ mmu->pg_sz = PAGE_SIZE;
+ mmu->sets = 1 << mmu2->sets;
+ mmu->ways = 1 << mmu2->ways;
+ mmu->u_dtlb = mmu2->u_dtlb;
+ mmu->u_itlb = mmu2->u_itlb;
+ } else {
+ mmu3 = (struct bcr_mmu_3 *)&tmp;
+ mmu->pg_sz = 512 << mmu3->pg_sz;
+ mmu->sets = 1 << mmu3->sets;
+ mmu->ways = 1 << mmu3->ways;
+ mmu->u_dtlb = mmu3->u_dtlb;
+ mmu->u_itlb = mmu3->u_itlb;
+ }
+
+ mmu->num_tlb = mmu->sets * mmu->ways;
+}
+
+void __init arc_mmu_init(void)
+{
+ /*
+ * ASID mgmt data structures are compile time init
+ * asid_cache = FIRST_ASID and asid_mm_map[] all zeroes
+ */
+
+ local_flush_tlb_all();
+
+ /* Enable the MMU */
+ write_aux_reg(ARC_REG_PID, MMU_ENABLE);
+}
+
+/*
+ * TLB Programmer's Model uses Linear Indexes: 0 to {255, 511} for 128 x {2,4}
+ * The mapping is Column-first.
+ * --------------------- -----------
+ * |way0|way1|way2|way3| |way0|way1|
+ * --------------------- -----------
+ * [set0] | 0 | 1 | 2 | 3 | | 0 | 1 |
+ * [set1] | 4 | 5 | 6 | 7 | | 2 | 3 |
+ * ~ ~ ~ ~
+ * [set127] | 508| 509| 510| 511| | 254| 255|
+ * --------------------- -----------
+ * For normal operations we don't(must not) care how above works since
+ * MMU cmd getIndex(vaddr) abstracts that out.
+ * However for walking WAYS of a SET, we need to know this
+ */
+#define SET_WAY_TO_IDX(mmu, set, way) ((set) * mmu->ways + (way))
+
+/* Handling of Duplicate PD (TLB entry) in MMU.
+ * -Could be due to buggy customer tapeouts or obscure kernel bugs
+ * -MMU complaints not at the time of duplicate PD installation, but at the
+ * time of lookup matching multiple ways.
+ * -Ideally these should never happen - but if they do - workaround by deleting
+ * the duplicate one.
+ * -Knob to be verbose abt it.(TODO: hook them up to debugfs)
+ */
+volatile int dup_pd_verbose = 1;/* Be slient abt it or complain (default) */
+
+void do_tlb_overlap_fault(unsigned long cause, unsigned long address,
+ struct pt_regs *regs)
+{
+ int set, way, n;
+ unsigned int pd0[4], pd1[4]; /* assume max 4 ways */
+ unsigned long flags, is_valid;
+ struct cpuinfo_arc_mmu *mmu = &cpuinfo_arc700[smp_processor_id()].mmu;
+
+ local_irq_save(flags);
+
+ /* re-enable the MMU */
+ write_aux_reg(ARC_REG_PID, MMU_ENABLE | read_aux_reg(ARC_REG_PID));
+
+ /* loop thru all sets of TLB */
+ for (set = 0; set < mmu->sets; set++) {
+
+ /* read out all the ways of current set */
+ for (way = 0, is_valid = 0; way < mmu->ways; way++) {
+ write_aux_reg(ARC_REG_TLBINDEX,
+ SET_WAY_TO_IDX(mmu, set, way));
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBRead);
+ pd0[way] = read_aux_reg(ARC_REG_TLBPD0);
+ pd1[way] = read_aux_reg(ARC_REG_TLBPD1);
+ is_valid |= pd0[way] & _PAGE_PRESENT;
+ }
+
+ /* If all the WAYS in SET are empty, skip to next SET */
+ if (!is_valid)
+ continue;
+
+ /* Scan the set for duplicate ways: needs a nested loop */
+ for (way = 0; way < mmu->ways; way++) {
+ if (!pd0[way])
+ continue;
+
+ for (n = way + 1; n < mmu->ways; n++) {
+ if ((pd0[way] & PAGE_MASK) ==
+ (pd0[n] & PAGE_MASK)) {
+
+ if (dup_pd_verbose) {
+ pr_info("Duplicate PD's @"
+ "[%d:%d]/[%d:%d]\n",
+ set, way, set, n);
+ pr_info("TLBPD0[%u]: %08x\n",
+ way, pd0[way]);
+ }
+
+ /*
+ * clear entry @way and not @n. This is
+ * critical to our optimised loop
+ */
+ pd0[way] = pd1[way] = 0;
+ write_aux_reg(ARC_REG_TLBINDEX,
+ SET_WAY_TO_IDX(mmu, set, way));
+ __tlb_entry_erase();
+ }
+ }
+ }
+ }
+
+ local_irq_restore(flags);
+}
+
+/***********************************************************************
+ * Diagnostic Routines
+ * -Called from Low Level TLB Hanlders if things don;t look good
+ **********************************************************************/
+
+#ifdef CONFIG_ARC_DBG_TLB_PARANOIA
+
+/*
+ * Low Level ASM TLB handler calls this if it finds that HW and SW ASIDS
+ * don't match
+ */
+void print_asid_mismatch(int is_fast_path)
+{
+ int pid_sw, pid_hw;
+ pid_sw = current->active_mm->context.asid;
+ pid_hw = read_aux_reg(ARC_REG_PID) & 0xff;
+
+ pr_emerg("ASID Mismatch in %s Path Handler: sw-pid=0x%x hw-pid=0x%x\n",
+ is_fast_path ? "Fast" : "Slow", pid_sw, pid_hw);
+
+ __asm__ __volatile__("flag 1");
+}
+
+void tlb_paranoid_check(unsigned int pid_sw, unsigned long addr)
+{
+ unsigned int pid_hw;
+
+ pid_hw = read_aux_reg(ARC_REG_PID) & 0xff;
+
+ if (addr < 0x70000000 && ((pid_hw != pid_sw) || (pid_sw == NO_ASID)))
+ print_asid_mismatch(0);
+}
+#endif
diff --git a/arch/arc/mm/tlbex.S b/arch/arc/mm/tlbex.S
new file mode 100644
index 0000000..fc5b971
--- /dev/null
+++ b/arch/arc/mm/tlbex.S
@@ -0,0 +1,351 @@
+/*
+ * TLB Exception Handling for ARC
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Vineetg: April 2011 :
+ * -MMU v1: moved out legacy code into a seperate file
+ * -MMU v3: PD{0,1} bits layout changed: They don't overlap anymore,
+ * helps avoid a shift when preparing PD0 from PTE
+ *
+ * Vineetg: July 2009
+ * -For MMU V2, we need not do heuristics at the time of commiting a D-TLB
+ * entry, so that it doesn't knock out it's I-TLB entry
+ * -Some more fine tuning:
+ * bmsk instead of add, asl.cc instead of branch, delay slot utilise etc
+ *
+ * Vineetg: July 2009
+ * -Practically rewrote the I/D TLB Miss handlers
+ * Now 40 and 135 instructions a peice as compared to 131 and 449 resp.
+ * Hence Leaner by 1.5 K
+ * Used Conditional arithmetic to replace excessive branching
+ * Also used short instructions wherever possible
+ *
+ * Vineetg: Aug 13th 2008
+ * -Passing ECR (Exception Cause REG) to do_page_fault( ) for printing
+ * more information in case of a Fatality
+ *
+ * Vineetg: March 25th Bug #92690
+ * -Added Debug Code to check if sw-ASID == hw-ASID
+
+ * Rahul Trivedi, Amit Bhor: Codito Technologies 2004
+ */
+
+ .cpu A7
+
+#include <linux/linkage.h>
+#include <asm/entry.h>
+#include <asm/tlb.h>
+#include <asm/pgtable.h>
+#include <asm/arcregs.h>
+#include <asm/cache.h>
+#include <asm/processor.h>
+#if (CONFIG_ARC_MMU_VER == 1)
+#include <asm/tlb-mmu1.h>
+#endif
+
+;--------------------------------------------------------------------------
+; scratch memory to save the registers (r0-r3) used to code TLB refill Handler
+; For details refer to comments before TLBMISS_FREEUP_REGS below
+;--------------------------------------------------------------------------
+
+ .section .data
+ .global ex_saved_reg1
+ .align 1 << L1_CACHE_SHIFT ; IMP: Must be Cache Line aligned
+ .type ex_saved_reg1, @object
+ .size ex_saved_reg1, 16
+ex_saved_reg1:
+ .zero 16
+
+;============================================================================
+; Troubleshooting Stuff
+;============================================================================
+
+; Linux keeps ASID (Address Space ID) in task->active_mm->context.asid
+; When Creating TLB Entries, instead of doing 3 dependent loads from memory,
+; we use the MMU PID Reg to get current ASID.
+; In bizzare scenrios SW and HW ASID can get out-of-sync which is trouble.
+; So we try to detect this in TLB Mis shandler
+
+
+.macro DBG_ASID_MISMATCH
+
+#ifdef CONFIG_ARC_DBG_TLB_PARANOIA
+
+ ; make sure h/w ASID is same as s/w ASID
+
+ GET_CURR_TASK_ON_CPU r3
+ ld r0, [r3, TASK_ACT_MM]
+ ld r0, [r0, MM_CTXT+MM_CTXT_ASID]
+
+ lr r1, [ARC_REG_PID]
+ and r1, r1, 0xFF
+ breq r1, r0, 5f
+
+ ; Error if H/w and S/w ASID don't match, but NOT if in kernel mode
+ lr r0, [erstatus]
+ bbit0 r0, STATUS_U_BIT, 5f
+
+ ; We sure are in troubled waters, Flag the error, but to do so
+ ; need to switch to kernel mode stack to call error routine
+ GET_TSK_STACK_BASE r3, sp
+
+ ; Call printk to shoutout aloud
+ mov r0, 1
+ j print_asid_mismatch
+
+5: ; ASIDs match so proceed normally
+ nop
+
+#endif
+
+.endm
+
+;============================================================================
+;TLB Miss handling Code
+;============================================================================
+
+;-----------------------------------------------------------------------------
+; This macro does the page-table lookup for the faulting address.
+; OUT: r0 = PTE faulted on, r1 = ptr to PTE, r2 = Faulting V-address
+.macro LOAD_FAULT_PTE
+
+ lr r2, [efa]
+
+ lr r1, [ARC_REG_SCRATCH_DATA0] ; current pgd
+
+ lsr r0, r2, PGDIR_SHIFT ; Bits for indexing into PGD
+ ld.as r1, [r1, r0] ; PGD entry corresp to faulting addr
+ and.f r1, r1, PAGE_MASK ; Ignoring protection and other flags
+ ; contains Ptr to Page Table
+ bz.d do_slow_path_pf ; if no Page Table, do page fault
+
+ ; Get the PTE entry: The idea is
+ ; (1) x = addr >> PAGE_SHIFT -> masks page-off bits from @fault-addr
+ ; (2) y = x & (PTRS_PER_PTE - 1) -> to get index
+ ; (3) z = pgtbl[y]
+ ; To avoid the multiply by in end, we do the -2, <<2 below
+
+ lsr r0, r2, (PAGE_SHIFT - 2)
+ and r0, r0, ( (PTRS_PER_PTE - 1) << 2)
+ ld.aw r0, [r1, r0] ; get PTE and PTE ptr for fault addr
+
+.endm
+
+;-----------------------------------------------------------------
+; Convert Linux PTE entry into TLB entry
+; A one-word PTE entry is programmed as two-word TLB Entry [PD0:PD1] in mmu
+; IN: r0 = PTE, r1 = ptr to PTE
+
+.macro CONV_PTE_TO_TLB
+ and r3, r0, PTE_BITS_IN_PD1 ; Extract permission flags+PFN from PTE
+ sr r3, [ARC_REG_TLBPD1] ; these go in PD1
+
+ and r2, r0, PTE_BITS_IN_PD0 ; Extract other PTE flags: (V)alid, (G)lb
+#if (CONFIG_ARC_MMU_VER <= 2) /* Neednot be done with v3 onwards */
+ lsr r2, r2 ; shift PTE flags to match layout in PD0
+#endif
+
+ lr r3,[ARC_REG_TLBPD0] ; MMU prepares PD0 with vaddr and asid
+
+ or r3, r3, r2 ; S | vaddr | {sasid|asid}
+ sr r3,[ARC_REG_TLBPD0] ; rewrite PD0
+.endm
+
+;-----------------------------------------------------------------
+; Commit the TLB entry into MMU
+
+.macro COMMIT_ENTRY_TO_MMU
+
+ /* Get free TLB slot: Set = computed from vaddr, way = random */
+ sr TLBGetIndex, [ARC_REG_TLBCOMMAND]
+
+ /* Commit the Write */
+#if (CONFIG_ARC_MMU_VER >= 2) /* introduced in v2 */
+ sr TLBWriteNI, [ARC_REG_TLBCOMMAND]
+#else
+ sr TLBWrite, [ARC_REG_TLBCOMMAND]
+#endif
+.endm
+
+;-----------------------------------------------------------------
+; ARC700 Exception Handling doesn't auto-switch stack and it only provides
+; ONE scratch AUX reg "ARC_REG_SCRATCH_DATA0"
+;
+; For Non-SMP, the scratch AUX reg is repurposed to cache task PGD, so a
+; "global" is used to free-up FIRST core reg to be able to code the rest of
+; exception prologue (IRQ auto-disabled on Exceptions, so it's IRQ-safe).
+; Since the Fast Path TLB Miss handler is coded with 4 regs, the remaining 3
+; need to be saved as well by extending the "global" to be 4 words. Hence
+; ".size ex_saved_reg1, 16"
+; [All of this dance is to avoid stack switching for each TLB Miss, since we
+; only need to save only a handful of regs, as opposed to complete reg file]
+
+; As simple as that....
+
+.macro TLBMISS_FREEUP_REGS
+ st r0, [@ex_saved_reg1]
+ mov_s r0, @ex_saved_reg1
+ st_s r1, [r0, 4]
+ st_s r2, [r0, 8]
+ st_s r3, [r0, 12]
+
+ ; VERIFY if the ASID in MMU-PID Reg is same as
+ ; one in Linux data structures
+
+ DBG_ASID_MISMATCH
+.endm
+
+;-----------------------------------------------------------------
+.macro TLBMISS_RESTORE_REGS
+ mov_s r0, @ex_saved_reg1
+ ld_s r3, [r0,12]
+ ld_s r2, [r0, 8]
+ ld_s r1, [r0, 4]
+ ld_s r0, [r0]
+.endm
+
+.section .text, "ax",@progbits ;Fast Path Code, candidate for ICCM
+
+;-----------------------------------------------------------------------------
+; I-TLB Miss Exception Handler
+;-----------------------------------------------------------------------------
+
+ARC_ENTRY EV_TLBMissI
+
+ TLBMISS_FREEUP_REGS
+
+ ;----------------------------------------------------------------
+ ; Get the PTE corresponding to V-addr accessed
+ LOAD_FAULT_PTE
+
+ ;----------------------------------------------------------------
+ ; VERIFY_PTE: Check if PTE permissions approp for executing code
+ cmp_s r2, VMALLOC_START
+ mov.lo r2, (_PAGE_PRESENT | _PAGE_READ | _PAGE_EXECUTE)
+ mov.hs r2, (_PAGE_PRESENT | _PAGE_K_READ | _PAGE_K_EXECUTE)
+
+ and r3, r0, r2 ; Mask out NON Flag bits from PTE
+ xor.f r3, r3, r2 ; check ( ( pte & flags_test ) == flags_test )
+ bnz do_slow_path_pf
+
+ ; Let Linux VM know that the page was accessed
+ or r0, r0, (_PAGE_PRESENT | _PAGE_ACCESSED) ; set Accessed Bit
+ st_s r0, [r1] ; Write back PTE
+
+ CONV_PTE_TO_TLB
+ COMMIT_ENTRY_TO_MMU
+ TLBMISS_RESTORE_REGS
+ rtie
+
+ARC_EXIT EV_TLBMissI
+
+;-----------------------------------------------------------------------------
+; D-TLB Miss Exception Handler
+;-----------------------------------------------------------------------------
+
+ARC_ENTRY EV_TLBMissD
+
+ TLBMISS_FREEUP_REGS
+
+ ;----------------------------------------------------------------
+ ; Get the PTE corresponding to V-addr accessed
+ ; If PTE exists, it will setup, r0 = PTE, r1 = Ptr to PTE
+ LOAD_FAULT_PTE
+
+ ;----------------------------------------------------------------
+ ; VERIFY_PTE: Chk if PTE permissions approp for data access (R/W/R+W)
+
+ mov_s r2, 0
+ lr r3, [ecr]
+ btst_s r3, ECR_C_BIT_DTLB_LD_MISS ; Read Access
+ or.nz r2, r2, _PAGE_READ ; chk for Read flag in PTE
+ btst_s r3, ECR_C_BIT_DTLB_ST_MISS ; Write Access
+ or.nz r2, r2, _PAGE_WRITE ; chk for Write flag in PTE
+ ; Above laddering takes care of XCHG access
+ ; which is both Read and Write
+
+ ; If kernel mode access, ; make _PAGE_xx flags as _PAGE_K_xx
+ ; For copy_(to|from)_user, despite exception taken in kernel mode,
+ ; this code is not hit, because EFA would still be the user mode
+ ; address (EFA < 0x6000_0000).
+ ; This code is for legit kernel mode faults, vmalloc specifically
+ ; (EFA: 0x7000_0000 to 0x7FFF_FFFF)
+
+ lr r3, [efa]
+ cmp r3, VMALLOC_START - 1 ; If kernel mode access
+ asl.hi r2, r2, 3 ; make _PAGE_xx flags as _PAGE_K_xx
+ or r2, r2, _PAGE_PRESENT ; Common flag for K/U mode
+
+ ; By now, r2 setup with all the Flags we need to check in PTE
+ and r3, r0, r2 ; Mask out NON Flag bits from PTE
+ brne.d r3, r2, do_slow_path_pf ; is ((pte & flags_test) == flags_test)
+
+ ;----------------------------------------------------------------
+ ; UPDATE_PTE: Let Linux VM know that page was accessed/dirty
+ lr r3, [ecr]
+ or r0, r0, (_PAGE_PRESENT | _PAGE_ACCESSED) ; Accessed bit always
+ btst_s r3, ECR_C_BIT_DTLB_ST_MISS ; See if it was a Write Access ?
+ or.nz r0, r0, _PAGE_MODIFIED ; if Write, set Dirty bit as well
+ st_s r0, [r1] ; Write back PTE
+
+ CONV_PTE_TO_TLB
+
+#if (CONFIG_ARC_MMU_VER == 1)
+ ; MMU with 2 way set assoc J-TLB, needs some help in pathetic case of
+ ; memcpy where 3 parties contend for 2 ways, ensuing a livelock.
+ ; But only for old MMU or one with Metal Fix
+ TLB_WRITE_HEURISTICS
+#endif
+
+ COMMIT_ENTRY_TO_MMU
+ TLBMISS_RESTORE_REGS
+ rtie
+
+;-------- Common routine to call Linux Page Fault Handler -----------
+do_slow_path_pf:
+
+ ; Restore the 4-scratch regs saved by fast path miss handler
+ TLBMISS_RESTORE_REGS
+
+ ; Slow path TLB Miss handled as a regular ARC Exception
+ ; (stack switching / save the complete reg-file).
+ ; That requires freeing up r9
+ EXCPN_PROLOG_FREEUP_REG r9
+
+ lr r9, [erstatus]
+
+ SWITCH_TO_KERNEL_STK
+ SAVE_ALL_SYS
+
+ ; ------- setup args for Linux Page fault Hanlder ---------
+ mov_s r0, sp
+ lr r2, [efa]
+ lr r3, [ecr]
+
+ ; Both st and ex imply WRITE access of some sort, hence do_page_fault( )
+ ; invoked with write=1 for DTLB-st/ex Miss and write=0 for ITLB miss or
+ ; DTLB-ld Miss
+ ; DTLB Miss Cause code is ld = 0x01 , st = 0x02, ex = 0x03
+ ; Following code uses that fact that st/ex have one bit in common
+
+ btst_s r3, ECR_C_BIT_DTLB_ST_MISS
+ mov.z r1, 0
+ mov.nz r1, 1
+
+ ; We don't want exceptions to be disabled while the fault is handled.
+ ; Now that we have saved the context we return from exception hence
+ ; exceptions get re-enable
+
+ FAKE_RET_FROM_EXCPN r9
+
+ bl do_page_fault
+ b ret_from_exception
+
+ARC_EXIT EV_TLBMissD
+
+ARC_ENTRY EV_TLBMissB ; Bogus entry to measure sz of DTLBMiss hdlr
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/kernel/entry.S | 8 ++++++++
arch/arc/kernel/process.c | 32 +++++++++++++-------------------
2 files changed, 21 insertions(+), 19 deletions(-)
diff --git a/arch/arc/kernel/entry.S b/arch/arc/kernel/entry.S
index fe9a17c..46d0280 100644
--- a/arch/arc/kernel/entry.S
+++ b/arch/arc/kernel/entry.S
@@ -580,6 +580,14 @@ ARC_ENTRY ret_from_fork
b @ret_from_exception
ARC_EXIT ret_from_fork
+ARC_ENTRY ret_from_kernel_thread
+ bl @schedule_tail
+ ld r1, [sp, PT_r1]
+ j.d [r1]
+ ld r0, [sp, PT_r0]
+ j @sys_exit
+ARC_EXIT ret_from_kernel_thread
+
;################### Special Sys Call Wrappers ##########################
ARC_ENTRY sys_execve_wrapper
diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c
index c116fa5..fee3e11 100644
--- a/arch/arc/kernel/process.c
+++ b/arch/arc/kernel/process.c
@@ -142,24 +142,13 @@ void cpu_idle(void)
}
}
-void kernel_thread_helper(void)
-{
- __asm__ __volatile__(
- "mov r0, r2 \n\t"
- "mov r1, r3 \n\t"
- "j [r1] \n\t");
-}
-
int kernel_thread(int (*fn) (void *), void *arg, unsigned long flags)
{
- struct pt_regs regs;
+ struct pt_regs regs = {
+ .r0 = (unsigned long)arg,
+ .r1 = (unsigned long)fn
+ };
- memset(®s, 0, sizeof(regs));
-
- regs.r2 = (unsigned long)arg;
- regs.r3 = (unsigned long)fn;
- regs.blink = (unsigned long)do_exit;
- regs.ret = (unsigned long)kernel_thread_helper;
regs.status32 = read_aux_reg(0xa);
/* Ok, create the new process.. */
@@ -170,6 +159,7 @@ int kernel_thread(int (*fn) (void *), void *arg, unsigned long flags)
EXPORT_SYMBOL(kernel_thread);
asmlinkage void ret_from_fork(void);
+asmlinkage void ret_from_kernel_thread(void) __attribute__((noreturn));
/* Layout of Child kernel mode stack as setup at the end of this function is
*
@@ -226,18 +216,22 @@ int copy_thread(unsigned long clone_flags,
/* Copy parents pt regs on child's kernel mode stack */
*c_regs = *regs;
- /* __switch_to expects FP(0), BLINK(return addr) at top of stack */
- childksp[0] = 0; /* for POP fp */
- childksp[1] = (unsigned long)ret_from_fork; /* for POP blink */
-
if (!(user_mode(regs))) {
c_regs->sp =
(unsigned long)task_thread_info(p) + (THREAD_SIZE - 4);
+
+ /* __switch_to expects FP(0), BLINK(return addr) at top */
+ childksp[0] = 0; /* fp */
+ childksp[1] = (unsigned long)ret_from_kernel_thread; /* blink */
return 0;
}
/*--------- User Task Only --------------*/
+ /* __switch_to expects FP(0), BLINK(return addr) at top of stack */
+ childksp[0] = 0; /* for POP fp */
+ childksp[1] = (unsigned long)ret_from_fork; /* for POP blink */
+
c_regs->sp = usp;
c_regs->r0 = 0; /* fork returns 0 in child */
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/ptrace.h | 8 +++++++
arch/arc/include/asm/unistd.h | 2 +
arch/arc/kernel/entry.S | 17 ++++++++-------
arch/arc/kernel/process.c | 44 -----------------------------------------
arch/arc/kernel/sys.c | 1 -
5 files changed, 19 insertions(+), 53 deletions(-)
diff --git a/arch/arc/include/asm/ptrace.h b/arch/arc/include/asm/ptrace.h
index ee31d8a..bc2229f 100644
--- a/arch/arc/include/asm/ptrace.h
+++ b/arch/arc/include/asm/ptrace.h
@@ -120,6 +120,14 @@ struct user_regs_struct {
#define in_syscall(regs) (((regs->orig_r8) >= 0 && \
(regs->orig_r8 <= NR_syscalls)) ? 1 : 0)
+#define current_pt_regs() \
+({ \
+ /* open-coded current_thread_info() */ \
+ register unsigned long sp asm ("sp"); \
+ unsigned long pg_start = (sp & ~(THREAD_SIZE - 1)); \
+ (struct pt_regs *)(pg_start + THREAD_SIZE - 4) - 1; \
+})
+
#endif /* __ASSEMBLY__ */
#endif /* __ASM_PTRACE_H */
diff --git a/arch/arc/include/asm/unistd.h b/arch/arc/include/asm/unistd.h
index 754d588..09144a8 100644
--- a/arch/arc/include/asm/unistd.h
+++ b/arch/arc/include/asm/unistd.h
@@ -22,6 +22,8 @@
#define __ARCH_WANT_SYSCALL_NO_FLAGS
#define __ARCH_WANT_SYSCALL_OFF_T
#define __ARCH_WANT_SYSCALL_DEPRECATED
+#define __ARCH_WANT_KERNEL_EXECVE
+#define __ARCH_WANT_SYS_EXECVE
#define sys_mmap2 sys_mmap_pgoff
diff --git a/arch/arc/kernel/entry.S b/arch/arc/kernel/entry.S
index 46d0280..deb2652 100644
--- a/arch/arc/kernel/entry.S
+++ b/arch/arc/kernel/entry.S
@@ -588,15 +588,16 @@ ARC_ENTRY ret_from_kernel_thread
j @sys_exit
ARC_EXIT ret_from_kernel_thread
-;################### Special Sys Call Wrappers ##########################
-
-ARC_ENTRY sys_execve_wrapper
- ; copy pointer to pt_regs as a parameter
- mov r3, sp
- bl @sys_execve
+; When we land here, pt_regs have already been updated in-place correctly
+; A pointer to them is also passed by kernel_execve, we just need to make sure
+; that SP is set to point to them.
+ARC_ENTRY ret_from_kernel_execve
+ ; Force SP to "normal" pt_regs just populated.
+ b.d ret_from_system_call
+ mov sp, r0
+ARC_EXIT ret_from_kernel_execve
- b ret_from_system_call
-ARC_EXIT sys_execve_wrapper
+;################### Special Sys Call Wrappers ##########################
; TBD: call do_fork directly from here
ARC_ENTRY sys_fork_wrapper
diff --git a/arch/arc/kernel/process.c b/arch/arc/kernel/process.c
index d2bb27f..65dfa89 100644
--- a/arch/arc/kernel/process.c
+++ b/arch/arc/kernel/process.c
@@ -53,50 +53,6 @@ asmlinkage int sys_clone(unsigned long clone_flags, unsigned long newsp,
child_tidptr);
}
-int sys_execve(const char __user *filenamei, const char __user *__user *argv,
- const char __user *__user *envp, struct pt_regs *regs)
-{
- long error;
- struct filename *filename;
-
- filename = getname(filenamei);
- error = PTR_ERR(filename);
- if (IS_ERR(filename))
- goto out;
-
- error = do_execve(filename->name, argv, envp, regs);
- putname(filename);
-out:
- return error;
-}
-
-int kernel_execve(const char *filename, const char *const argv[],
- const char *const envp[])
-{
- /*
- * Although the arguments (order, number) to this function are
- * same as sys call, we don't need to setup args in regs again.
- * However in case mainline kernel changes the order of args to
- * kernel_execve, that assumtion will break.
- * So to be safe, let gcc know the args for sys call.
- * If they match no extra code will be generated
- */
- register int arg2 asm("r1") = (int)argv;
- register int arg3 asm("r2") = (int)envp;
-
- register int filenm_n_ret asm("r0") = (int)filename;
-
- __asm__ __volatile__(
- "mov r8, %1 \n\t"
- "trap0 \n\t"
- : "+r"(filenm_n_ret)
- : "i"(__NR_execve), "r"(arg2), "r"(arg3)
- : "r8", "memory");
-
- return filenm_n_ret;
-}
-EXPORT_SYMBOL(kernel_execve);
-
SYSCALL_DEFINE1(arc_settls, void *, user_tls_data_ptr)
{
task_thread_info(current)->thr_ptr = (unsigned int)user_tls_data_ptr;
diff --git a/arch/arc/kernel/sys.c b/arch/arc/kernel/sys.c
index 4c30345..f6bdd07 100644
--- a/arch/arc/kernel/sys.c
+++ b/arch/arc/kernel/sys.c
@@ -5,7 +5,6 @@
#include <asm/syscalls.h>
-#define sys_execve sys_execve_wrapper
#define sys_clone sys_clone_wrapper
#define sys_fork sys_fork_wrapper
#define sys_vfork sys_vfork_wrapper
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/Kbuild | 2 +
arch/arc/Kconfig | 337 ++++++++++++++++++++++++++++++++++++
arch/arc/Kconfig.debug | 34 ++++
arch/arc/Makefile | 115 ++++++++++++
arch/arc/boot/Makefile | 27 +++
arch/arc/include/asm/Kbuild | 1 +
arch/arc/include/asm/vmlinux.lds.h | 12 ++
arch/arc/kernel/Makefile | 16 ++
arch/arc/kernel/arcksyms.c | 56 ++++++
arch/arc/kernel/asm-offsets.c | 46 +++++
arch/arc/kernel/vmlinux.lds.S | 117 +++++++++++++
arch/arc/lib/Makefile | 9 +
arch/arc/mm/Makefile | 10 +
arch/arc/plat-arcfpga/Kconfig | 33 ++++
arch/arc/plat-arcfpga/Makefile | 9 +
15 files changed, 824 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/Kbuild
create mode 100644 arch/arc/Kconfig
create mode 100644 arch/arc/Kconfig.debug
create mode 100644 arch/arc/Makefile
create mode 100644 arch/arc/boot/Makefile
create mode 100644 arch/arc/include/asm/vmlinux.lds.h
create mode 100644 arch/arc/kernel/Makefile
create mode 100644 arch/arc/kernel/arcksyms.c
create mode 100644 arch/arc/kernel/asm-offsets.c
create mode 100644 arch/arc/kernel/vmlinux.lds.S
create mode 100644 arch/arc/lib/Makefile
create mode 100644 arch/arc/mm/Makefile
create mode 100644 arch/arc/plat-arcfpga/Kconfig
create mode 100644 arch/arc/plat-arcfpga/Makefile
diff --git a/arch/arc/Kbuild b/arch/arc/Kbuild
new file mode 100644
index 0000000..082d329
--- /dev/null
+++ b/arch/arc/Kbuild
@@ -0,0 +1,2 @@
+obj-y += kernel/
+obj-y += mm/
diff --git a/arch/arc/Kconfig b/arch/arc/Kconfig
new file mode 100644
index 0000000..d6983afa
--- /dev/null
+++ b/arch/arc/Kconfig
@@ -0,0 +1,337 @@
+#
+# Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+#
+
+config ARC
+ def_bool y
+ select ARCH_WANT_IPC_PARSE_VERSION
+ # ARC Busybox based initramfs absolutely relies on DEVTMPFS for /dev
+ # DEVTMPS in turn needs HOTPLUG
+ select DEVTMPFS if !INITRAMFS_SOURCE=""
+ select GENERIC_ATOMIC64
+ select GENERIC_CLOCKEVENTS
+ select GENERIC_FIND_FIRST_BIT
+ # for now, we don't need GENERIC_IRQ_PROBE, CONFIG_GENERIC_IRQ_CHIP
+ select GENERIC_IRQ_SHOW
+ select GENERIC_PENDING_IRQ if SMP
+ select GENERIC_SMP_IDLE_THREAD
+ select HAVE_GENERIC_HARDIRQS
+ select HOTPLUG if !INITRAMFS_SOURCE=""
+ select MODULES_USE_ELF_RELA
+
+config SCHED_OMIT_FRAME_POINTER
+ def_bool y
+
+config GENERIC_CSUM
+ def_bool y
+
+config RWSEM_GENERIC_SPINLOCK
+ def_bool y
+
+# for old_[ug]id_t - do we need this ?
+config UID16
+ def_bool y
+
+config ARCH_FLATMEM_ENABLE
+ def_bool y
+
+config MMU
+ def_bool y
+
+config NO_IOPORT
+ def_bool y
+
+config GENERIC_CALIBRATE_DELAY
+ def_bool y
+
+config GENERIC_HWEIGHT
+ def_bool y
+
+config BINFMT_ELF
+ def_bool y
+
+config STACKTRACE_SUPPORT
+ def_bool y
+ select STACKTRACE
+
+config HAVE_LATENCYTOP_SUPPORT
+ def_bool y
+
+config NO_DMA
+ def_bool n
+
+source "init/Kconfig"
+source "kernel/Kconfig.freezer"
+
+menu "ARC Architecture Configuration"
+
+choice
+ prompt "ARC Platform"
+ default ARC_PLAT_FPGA_LEGACY
+
+config ARC_PLAT_FPGA_LEGACY
+ bool "\"Legacy\" ARC FPGA dev platform"
+ help
+ Support for ARC development platforms, provided by Synopsys.
+ These are based on FPGA or ISS. e.g.
+ - ARCAngel4
+ - ML509
+ - MetaWare ISS
+
+#New platform adds here
+endchoice
+
+menu "ARC CPU Configuration"
+
+choice
+ prompt "ARC Core"
+ default ARC_CPU_770
+
+config ARC_CPU_750D
+ bool "ARC750D"
+ help
+ Support for ARC750 core
+
+config ARC_CPU_770
+ bool "ARC770"
+ select ARC_CPU_REL_4_10
+ help
+ Support for ARC770 core introduced with Rel 4.10 (Summer 2011)
+ This core has a bunch of cool new features:
+ -MMU-v3: Variable Page Sz (4k, 8k, 16k), bigger J-TLB (128x4)
+ Shared Address Spaces (for sharing TLB entires in MMU)
+ -Caches: New Prog Model, Region Flush
+ -Insns: endian swap, load-locked/store-conditional, time-stamp-ctr
+
+endchoice
+
+config CPU_BIG_ENDIAN
+ bool "Enable Big Endian Mode"
+ default n
+ help
+ Build kernel for Big Endian Mode of ARC CPU
+
+menuconfig ARC_CACHE
+ bool "Enable Cache Support"
+ default y
+
+if ARC_CACHE
+
+config ARC_CACHE_LINE_SHIFT
+ int "Cache Line Length (as power of 2)"
+ range 5 7
+ default "6"
+ help
+ Starting with ARC700 4.9, Cache line length is configurable,
+ This option specifies "N", with Line-len = 2 power N
+ So line lengths of 32, 64, 128 are specified by 5,6,7, respectively
+ Linux only supports same line lengths for I and D caches.
+
+config ARC_HAS_ICACHE
+ bool "Use Instruction Cache"
+ default y
+
+config ARC_HAS_DCACHE
+ bool "Use Data Cache"
+ default y
+
+config ARC_CACHE_PAGES
+ bool "Per Page Cache Control"
+ default y
+ depends on ARC_HAS_ICACHE || ARC_HAS_DCACHE
+ help
+ This can be used to over-ride the global I/D Cache Enable on a
+ per-page basis (but only for pages accessed via MMU such as
+ Kernel Virtual address or User Virtual Address)
+ TLB entries have a per-page Cache Enable Bit.
+ Note that Global I/D ENABLE + Per Page DISABLE works but corollary
+ Global DISABLE + Per Page ENABLE won't work
+
+endif #ARC_CACHE
+
+config ARC_HAS_HW_MPY
+ bool "Use Hardware Multiplier (Normal or Faster XMAC)"
+ default y
+ help
+ Influences how gcc generates code for MPY operations.
+ If enabled, MPYxx insns are generated, provided by Standard/XMAC
+ Multipler. Otherwise software multipy lib is used
+
+choice
+ prompt "ARC700 MMU Version"
+ default ARC_MMU_V3 if ARC_CPU_770
+ default ARC_MMU_V2 if ARC_CPU_750D
+
+config ARC_MMU_V1
+ bool "MMU v1"
+ help
+ Orig ARC700 MMU
+
+config ARC_MMU_V2
+ bool "MMU v2"
+ help
+ Fixed the deficiency of v1 - possible thrashing in memcpy sceanrio
+ when 2 D-TLB and 1 I-TLB entries index into same 2way set.
+
+config ARC_MMU_V3
+ bool "MMU v3"
+ depends on ARC_CPU_770
+ help
+ Introduced with ARC700 4.10: New Features
+ Variable Page size (1k-16k), var JTLB size 128 x (2 or 4)
+ Shared Address Spaces (SASID)
+
+endchoice
+
+
+choice
+ prompt "MMU Page Size"
+ default ARC_PAGE_SIZE_8K
+
+config ARC_PAGE_SIZE_8K
+ bool "8KB"
+ help
+ Choose between 8k vs 16k
+
+config ARC_PAGE_SIZE_16K
+ bool "16KB"
+ depends on ARC_MMU_V3
+
+config ARC_PAGE_SIZE_4K
+ bool "4KB"
+ depends on ARC_MMU_V3
+
+endchoice
+
+config ARC_FPU_SAVE_RESTORE
+ bool "Enable FPU state persistence across context switch"
+ default n
+ help
+ Double Precision Floating Point unit had dedictaed regs which
+ need to be saved/restored across context-switch.
+ Note that ARC FPU is overly simplistic, unlike say x86, which has
+ hardware pieces to allow software to conditionally save/restore,
+ based on actual usage of FPU by a task. Thus our implemn does
+ this for all tasks in system.
+
+menuconfig ARC_CPU_REL_4_10
+ bool "Enable support for Rel 4.10 features"
+ default n
+ help
+ -ARC770 (and dependent features) enabled
+ -ARC750 also shares some of the new features with 770
+
+config ARC_HAS_LLSC
+ bool "Insn: LLOCK/SCOND (efficient atomic ops)"
+ default y
+ depends on ARC_CPU_770
+ # if SMP, enable LLSC ONLY if ARC implementation has coherent atomics
+ depends on !SMP || ARC_HAS_COH_LLSC
+
+config ARC_HAS_SWAPE
+ bool "Insn: SWAPE (endian-swap)"
+ default y
+ depends on ARC_CPU_REL_4_10
+
+config ARC_HAS_RTSC
+ bool "Insn: RTSC (64-bit r/o cycle counter)"
+ default y
+ depends on ARC_CPU_REL_4_10
+
+endmenu # "ARC CPU Configuration"
+
+menu "Platform Board Configuration"
+
+source "arch/arc/plat-arcfpga/Kconfig"
+#New platform adds here
+
+config ARC_PLAT_CLK
+ int "Clk speed in Hz"
+ default "80000000"
+
+config LINUX_LINK_BASE
+ hex "Linux Link Address"
+ default "0x80000000"
+ help
+ ARC700 divides the 32 bit phy address space into two equal halves
+ -Lower 2G (0 - 0x7FFF_FFFF ) is user virtual, translated by MMU
+ -Upper 2G (0x8000_0000 onwards) is untranslated, for kernel
+ Typically Linux kernel is linked at the start of untransalted addr,
+ hence the default value of 0x8zs.
+ However some customers have peripherals mapped at this addr, so
+ Linux needs to be scooted a bit.
+ If you don't know what the above means, leave this setting alone.
+
+config ARC_PLAT_SDRAM_SIZE
+ hex "SD RAM Size"
+ default "0x10000000"
+ help
+ Implies the amount of SDRAM/DRAM Linux is going to claim/own.
+ The actual memory itself could be larger than this number. But for
+ all software purposes, this is the amt of memory.
+
+endmenu # "Platform Board Configuration"
+
+config ARC_STACK_NONEXEC
+ bool "Make stack non-executable"
+ default n
+ help
+ To disable the execute permissions of stack/heap of processes
+ which are enabled by default.
+
+config HZ
+ int "Timer Frequency"
+ default 100
+
+menuconfig ARC_DBG
+ bool "ARC debugging"
+ default y
+
+config ARC_DBG_TLB_PARANOIA
+ bool "Paranoia Checks in Low Level TLB Handlers"
+ depends on ARC_DBG
+ default n
+
+config ARC_DBG_TLB_MISS_COUNT
+ bool "Profile TLB Misses"
+ default n
+ select DEBUG_FS
+ depends on ARC_DBG
+ help
+ Counts number of I and D TLB Misses and exports them via Debugfs
+ The counters can be cleared via Debugfs as well
+
+config CMDLINE
+ string "Kernel command line to built-in"
+ default "print-fatal-signals=1"
+ help
+ The default command line which will be appended to the optional
+ u-boot provided command line (see below)
+
+config CMDLINE_UBOOT
+ bool "Support U-boot kernel command line passing"
+ default n
+ help
+ If you are using U-boot (http://www.denx.de) and wish to pass the kernel
+ command line from the U-boot environment to the Linux kernel then
+ switch this option on.
+ ARC U-boot will setup the cmdline in RAM/flash and set r2 to point
+ to it. kernel startup code will copy the string into cmdline buffer
+ and also append CONFIG_CMDLINE.
+
+source "kernel/Kconfig.preempt"
+
+endmenu # "ARC Architecture Configuration"
+
+source "mm/Kconfig"
+source "net/Kconfig"
+source "drivers/Kconfig"
+source "fs/Kconfig"
+source "arch/arc/Kconfig.debug"
+source "security/Kconfig"
+source "crypto/Kconfig"
+source "lib/Kconfig"
diff --git a/arch/arc/Kconfig.debug b/arch/arc/Kconfig.debug
new file mode 100644
index 0000000..962c609
--- /dev/null
+++ b/arch/arc/Kconfig.debug
@@ -0,0 +1,34 @@
+menu "Kernel hacking"
+
+source "lib/Kconfig.debug"
+
+config EARLY_PRINTK
+ bool "Early printk" if EMBEDDED
+ default y
+ help
+ Write kernel log output directly into the VGA buffer or to a serial
+ port.
+
+ This is useful for kernel debugging when your machine crashes very
+ early before the console code is initialized. For normal operation
+ it is not recommended because it looks ugly and doesn't cooperate
+ with klogd/syslogd or the X server. You should normally N here,
+ unless you want to debug such a crash.
+
+config DEBUG_STACKOVERFLOW
+ bool "Check for stack overflows"
+ depends on DEBUG_KERNEL
+ help
+ This option will cause messages to be printed if free stack space
+ drops below a certain limit.
+
+config 16KSTACKS
+ bool "Use 16Kb for kernel stacks instead of 8Kb"
+ help
+ If you say Y here the kernel will use a 16Kb stacksize for the
+ kernel stack attached to each process/thread. The default is 8K.
+ This increases the resident kernel footprint and will cause less
+ threads to run on the system and also increase the pressure
+ on the VM subsystem for higher order allocations.
+
+endmenu
diff --git a/arch/arc/Makefile b/arch/arc/Makefile
new file mode 100644
index 0000000..4d52a3b
--- /dev/null
+++ b/arch/arc/Makefile
@@ -0,0 +1,115 @@
+#
+# Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+#
+
+UTS_MACHINE := arc
+
+KBUILD_DEFCONFIG := fpga_defconfig
+
+# For ARC FPGA Platforms
+platform-$(CONFIG_ARC_PLAT_FPGA_LEGACY) := arcfpga
+#New platform adds here
+
+PLATFORM := $(platform-y)
+export PLATFORM
+
+cflags-y += -Iarch/arc/plat-$(PLATFORM)/include
+cflags-y += -mA7 -fno-common -pipe -fno-builtin -D__linux__
+
+atleast_gcc44 := $(call cc-ifversion, -gt, 0402, y)
+cflags-$(atleast_gcc44) += -fsection-anchors
+
+cflags-$(CONFIG_ARC_HAS_LLSC) += -mlock
+cflags-$(CONFIG_ARC_HAS_SWAPE) += -mswape
+cflags-$(CONFIG_ARC_HAS_RTSC) += -mrtsc
+cflags-$(CONFIG_ARC_DW2_UNWIND) += -fasynchronous-unwind-tables
+
+ifndef CONFIG_CC_OPTIMIZE_FOR_SIZE
+# Generic build system uses -O2, we want -O3
+cflags-y += -O3
+endif
+
+# small data is default for elf32 tool-chain. If not usable, disable it
+# This also allows repurposing GP as scratch reg to gcc reg allocator
+disable_small_data := y
+cflags-$(disable_small_data) += -mno-sdata -fcall-used-gp
+
+cflags-$(CONFIG_CPU_BIG_ENDIAN) += -mbig-endian
+ldflags-$(CONFIG_CPU_BIG_ENDIAN) += -EB
+
+# STAR 9000518362:
+# arc-linux-uclibc-ld (buildroot) or arceb-elf32-ld (EZChip) don't accept
+# --build-id w/o "-marclinux".
+# Default arc-elf32-ld is OK
+ldflags-y += -marclinux
+
+ARC_LIBGCC := -mA7
+cflags-$(CONFIG_ARC_HAS_HW_MPY) += -multcost=16
+
+ifndef CONFIG_ARC_HAS_HW_MPY
+ cflags-y += -mno-mpy
+
+# newlib for ARC700 assumes MPY to be always present, which is generally true
+# However, if someone really doesn't want MPY, we need to use the 600 ver
+# which coupled with -mno-mpy will use mpy emulation
+# With gcc 4.4.7, -mno-mpy is enough to make any other related adjustments,
+# e.g. increased cost of MPY. With gcc 4.2.1 this had to be explicitly hinted
+
+ ARC_LIBGCC := -marc600
+ ifneq ($(atleast_gcc44),y)
+ cflags-y += -multcost=30
+ endif
+endif
+
+LIBGCC := $(shell $(CC) $(ARC_LIBGCC) $(cflags-y) --print-libgcc-file-name)
+
+# Modules with short calls might break for calls into builtin-kernel
+KBUILD_CFLAGS_MODULE += -mlong-calls
+
+# Finally dump eveything into kernel build system
+KBUILD_CFLAGS += $(cflags-y)
+KBUILD_AFLAGS += $(KBUILD_CFLAGS)
+LDFLAGS += $(ldflags-y)
+
+# Needed for Linker script preprocessing
+KBUILD_CPPFLAGS += -Iarch/arc/plat-$(PLATFORM)/include
+
+head-y := arch/arc/kernel/head.o
+
+# See arch/arc/Kbuild for content of core part of the kernel
+core-y += arch/arc/
+
+# w/o this ifneq, make ARCH=arc clean was crapping out
+ifneq ($(platform-y),)
+core-y += arch/arc/plat-$(PLATFORM)/
+endif
+
+libs-y += arch/arc/lib/ $(LIBGCC)
+
+#default target for make without any arguements.
+KBUILD_IMAGE := bootpImage
+
+all: $(KBUILD_IMAGE)
+boot := arch/arc/boot
+
+bootpImage: vmlinux
+
+uImage: vmlinux
+ $(Q)$(MAKE) $(build)=$(boot) $(boot)/$@
+
+archclean:
+ $(Q)$(MAKE) $(clean)=$(boot)
+
+# Hacks to enable final link due to absence of link-time branch relexation
+# and gcc choosing optimal(shorter) branches at -O3
+#
+# vineetg Feb 2010: -mlong-calls switched off for overall kernel build
+# However lib/decompress_inflate.o (.init.text) calls
+# zlib_inflate_workspacesize (.text) causing relocation errors.
+# Thus forcing all exten calls in this file to be long calls
+export CFLAGS_decompress_inflate.o = -mmedium-calls
+export CFLAGS_initramfs.o = -mmedium-calls
diff --git a/arch/arc/boot/Makefile b/arch/arc/boot/Makefile
new file mode 100644
index 0000000..4ff734d
--- /dev/null
+++ b/arch/arc/boot/Makefile
@@ -0,0 +1,27 @@
+targets := vmlinux.bin vmlinux.bin.gz uImage
+
+# uImage build relies on mkimage being availble on your host for ARC target
+# You will need to build u-boot for ARC, rename mkimage to arc-elf32-mkimage
+# and make sure it's reacable from your PATH
+MKIMAGE := $(srctree)/scripts/mkuboot.sh
+
+OBJCOPYFLAGS= -O binary -R .note -R .note.gnu.build-id -R .comment -S
+
+LINUX_START_TEXT = $$(readelf -h vmlinux | \
+ grep "Entry point address" | grep -o 0x.*)
+
+quiet_cmd_uimage = UIMAGE $@
+ cmd_uimage = $(CONFIG_SHELL) $(MKIMAGE) -A arc -O linux -T kernel \
+ -C gzip -a $(CONFIG_LINUX_LINK_BASE) \
+ -e $(LINUX_START_TEXT) -d $< $@
+
+$(obj)/vmlinux.bin: vmlinux FORCE
+ $(call if_changed,objcopy)
+
+$(obj)/vmlinux.bin.gz: $(obj)/vmlinux.bin FORCE
+ $(call if_changed,gzip)
+
+$(obj)/uImage: $(obj)/vmlinux.bin.gz FORCE
+ $(call if_changed,uimage)
+
+PHONY += FORCE
diff --git a/arch/arc/include/asm/Kbuild b/arch/arc/include/asm/Kbuild
index d6c037a..148dfcf 100644
--- a/arch/arc/include/asm/Kbuild
+++ b/arch/arc/include/asm/Kbuild
@@ -33,6 +33,7 @@ generic-y += mman.h
generic-y += msgbuf.h
generic-y += param.h
generic-y += parport.h
+generic-y += pci.h
generic-y += percpu.h
generic-y += poll.h
generic-y += posix_types.h
diff --git a/arch/arc/include/asm/vmlinux.lds.h b/arch/arc/include/asm/vmlinux.lds.h
new file mode 100644
index 0000000..12c8e42
--- /dev/null
+++ b/arch/arc/include/asm/vmlinux.lds.h
@@ -0,0 +1,12 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_VMLINUX_LDS_H__
+#define __ASM_VMLINUX_LDS_H__
+
+#endif
diff --git a/arch/arc/kernel/Makefile b/arch/arc/kernel/Makefile
new file mode 100644
index 0000000..47f6b33
--- /dev/null
+++ b/arch/arc/kernel/Makefile
@@ -0,0 +1,16 @@
+#
+# Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+
+obj-y := arcksyms.o setup.o irq.o time.o reset.o ptrace.o entry.o process.o \
+ signal.o traps.o sys.o troubleshoot.o stacktrace.o
+
+obj-$(CONFIG_ARC_FPU_SAVE_RESTORE) += fpu.o
+CFLAGS_fpu.o += -mdpfp
+
+obj-y += ctx_sw_asm.o
+
+extra-y := vmlinux.lds head.o
diff --git a/arch/arc/kernel/arcksyms.c b/arch/arc/kernel/arcksyms.c
new file mode 100644
index 0000000..4d9e777
--- /dev/null
+++ b/arch/arc/kernel/arcksyms.c
@@ -0,0 +1,56 @@
+/*
+ * arcksyms.c - Exporting symbols not exportable from their own sources
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ */
+
+#include <linux/module.h>
+
+/* libgcc functions, not part of kernel sources */
+extern void __ashldi3(void);
+extern void __ashrdi3(void);
+extern void __divsi3(void);
+extern void __divsf3(void);
+extern void __lshrdi3(void);
+extern void __modsi3(void);
+extern void __muldi3(void);
+extern void __ucmpdi2(void);
+extern void __udivsi3(void);
+extern void __umodsi3(void);
+extern void __cmpdi2(void);
+extern void __fixunsdfsi(void);
+extern void __muldf3(void);
+extern void __divdf3(void);
+extern void __floatunsidf(void);
+extern void __floatunsisf(void);
+
+EXPORT_SYMBOL(__ashldi3);
+EXPORT_SYMBOL(__ashrdi3);
+EXPORT_SYMBOL(__divsi3);
+EXPORT_SYMBOL(__divsf3);
+EXPORT_SYMBOL(__lshrdi3);
+EXPORT_SYMBOL(__modsi3);
+EXPORT_SYMBOL(__muldi3);
+EXPORT_SYMBOL(__ucmpdi2);
+EXPORT_SYMBOL(__udivsi3);
+EXPORT_SYMBOL(__umodsi3);
+EXPORT_SYMBOL(__cmpdi2);
+EXPORT_SYMBOL(__fixunsdfsi);
+EXPORT_SYMBOL(__muldf3);
+EXPORT_SYMBOL(__divdf3);
+EXPORT_SYMBOL(__floatunsidf);
+EXPORT_SYMBOL(__floatunsisf);
+
+/* ARC optimised assembler routines */
+EXPORT_SYMBOL(memset);
+EXPORT_SYMBOL(memcpy);
+EXPORT_SYMBOL(memcmp);
+EXPORT_SYMBOL(strchr);
+EXPORT_SYMBOL(strcpy);
+EXPORT_SYMBOL(strcmp);
+EXPORT_SYMBOL(strlen);
diff --git a/arch/arc/kernel/asm-offsets.c b/arch/arc/kernel/asm-offsets.c
new file mode 100644
index 0000000..7f3f611
--- /dev/null
+++ b/arch/arc/kernel/asm-offsets.c
@@ -0,0 +1,46 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/sched.h>
+#include <linux/mm.h>
+#include <linux/interrupt.h>
+#include <asm/hardirq.h>
+#include <linux/thread_info.h>
+#include <asm/page.h>
+#include <linux/kbuild.h>
+
+int main(void)
+{
+ DEFINE(TASK_THREAD, offsetof(struct task_struct, thread));
+ DEFINE(TASK_THREAD_INFO, offsetof(struct task_struct, stack));
+
+ BLANK();
+
+ DEFINE(THREAD_KSP, offsetof(struct thread_struct, ksp));
+ DEFINE(THREAD_CALLEE_REG, offsetof(struct thread_struct, callee_reg));
+ DEFINE(THREAD_FAULT_ADDR,
+ offsetof(struct thread_struct, fault_address));
+
+ BLANK();
+
+ DEFINE(THREAD_INFO_FLAGS, offsetof(struct thread_info, flags));
+ DEFINE(THREAD_INFO_PREEMPT_COUNT,
+ offsetof(struct thread_info, preempt_count));
+
+ BLANK();
+
+ DEFINE(TASK_ACT_MM, offsetof(struct task_struct, active_mm));
+ DEFINE(TASK_TGID, offsetof(struct task_struct, tgid));
+
+ DEFINE(MM_CTXT, offsetof(struct mm_struct, context));
+ DEFINE(MM_PGD, offsetof(struct mm_struct, pgd));
+
+ DEFINE(MM_CTXT_ASID, offsetof(mm_context_t, asid));
+
+ return 0;
+}
diff --git a/arch/arc/kernel/vmlinux.lds.S b/arch/arc/kernel/vmlinux.lds.S
new file mode 100644
index 0000000..67c35bd
--- /dev/null
+++ b/arch/arc/kernel/vmlinux.lds.S
@@ -0,0 +1,117 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <asm-generic/vmlinux.lds.h>
+#include <asm/vmlinux.lds.h>
+#include <asm/cache.h>
+#include <asm/page.h>
+#include <asm/thread_info.h>
+#include <plat/memmap.h>
+
+OUTPUT_ARCH(arc)
+ENTRY(_stext)
+
+#ifdef CONFIG_CPU_BIG_ENDIAN
+jiffies = jiffies_64 + 4;
+#else
+jiffies = jiffies_64;
+#endif
+
+SECTIONS
+{
+ . = CONFIG_LINUX_LINK_BASE;
+
+ _int_vec_base_lds = .;
+ .vector : {
+ *(.vector)
+ . = ALIGN(PAGE_SIZE);
+ }
+
+ /*
+ * The reason for having a seperate subsection .init.ramfs is to
+ * prevent objump from including it in kernel dumps
+ *
+ * Reason for having .init.ramfs above .init is to make sure that the
+ * binary blob is tucked away to one side, reducing the displacement
+ * between .init.text and .text, avoiding any possible relocation
+ * errors because of calls from .init.text to .text
+ * Yes such calls do exist. e.g.
+ * decompress_inflate.c:gunzip( ) -> zlib_inflate_workspace( )
+ */
+
+ __init_begin = .;
+
+ .init.ramfs : { INIT_RAM_FS }
+
+ . = ALIGN(PAGE_SIZE);
+ _stext = .;
+
+ HEAD_TEXT_SECTION
+ INIT_TEXT_SECTION(L1_CACHE_BYTES)
+
+ /* INIT_DATA_SECTION open-coded: special INIT_RAM_FS handling */
+ .init.data : {
+ INIT_DATA
+ INIT_SETUP(L1_CACHE_BYTES)
+ INIT_CALLS
+ CON_INITCALL
+ SECURITY_INITCALL
+ }
+
+ PERCPU_SECTION(L1_CACHE_BYTES)
+
+ /*
+ * .exit.text is discard at runtime, not link time, to deal with
+ * references from .debug_frame
+ * It will be init freed, being inside [__init_start : __init_end]
+ */
+ .exit.text : { EXIT_TEXT }
+ .exit.data : { EXIT_DATA }
+
+ . = ALIGN(PAGE_SIZE);
+ __init_end = .;
+
+ .text : {
+ _text = .;
+ TEXT_TEXT
+ SCHED_TEXT
+ LOCK_TEXT
+ KPROBES_TEXT
+ *(.fixup)
+ *(.gnu.warning)
+ }
+ EXCEPTION_TABLE(L1_CACHE_BYTES)
+ _etext = .;
+
+ _sdata = .;
+ RO_DATA_SECTION(PAGE_SIZE)
+
+ /*
+ * 1. this is .data essentially
+ * 2. THREAD_SIZE for init.task, must be kernel-stk sz aligned
+ */
+ RW_DATA_SECTION(L1_CACHE_BYTES, PAGE_SIZE, THREAD_SIZE)
+
+ _edata = .;
+
+ BSS_SECTION(0, 0, 0)
+
+ NOTES
+
+ . = ALIGN(PAGE_SIZE);
+ _end = . ;
+
+ STABS_DEBUG
+ DWARF_DEBUG
+ DISCARDS
+
+ .arcextmap 0 : {
+ *(.gnu.linkonce.arcextmap.*)
+ *(.arcextmap.*)
+ }
+}
diff --git a/arch/arc/lib/Makefile b/arch/arc/lib/Makefile
new file mode 100644
index 0000000..f4eaaf4
--- /dev/null
+++ b/arch/arc/lib/Makefile
@@ -0,0 +1,9 @@
+#
+# Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+
+lib-y := strchr-700.o strcmp.o strcpy-700.o strlen.o \
+ memcmp.o memcpy-700.o memset.o
diff --git a/arch/arc/mm/Makefile b/arch/arc/mm/Makefile
new file mode 100644
index 0000000..303f567
--- /dev/null
+++ b/arch/arc/mm/Makefile
@@ -0,0 +1,10 @@
+#
+# Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+#
+
+obj-y := extable.o ioremap.o dma.o fault.o init.o mmap.o \
+ tlb.o tlbex.o cache_arc700.o
diff --git a/arch/arc/plat-arcfpga/Kconfig b/arch/arc/plat-arcfpga/Kconfig
new file mode 100644
index 0000000..3b41a0e
--- /dev/null
+++ b/arch/arc/plat-arcfpga/Kconfig
@@ -0,0 +1,33 @@
+#
+# Copyright (C) 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+#
+
+if ARC_PLAT_FPGA_LEGACY
+
+choice
+ prompt "FPGA Board"
+
+config ARC_BOARD_ANGEL4
+ bool "ARC Angel4"
+ help
+ ARC Angel4 FPGA Ref Platform (Xilinx Virtex Based)
+
+config ARC_BOARD_ML509
+ bool "ML509"
+ help
+ ARC ML509 FPGA Ref Platform (Xilinx Virtex-5 Based)
+
+endchoice
+
+config ARC_SERIAL_BAUD
+ int "UART Baud rate"
+ default "115200"
+ depends on SERIAL_ARC
+ help
+ Baud rate for the ARC UART
+
+endif
diff --git a/arch/arc/plat-arcfpga/Makefile b/arch/arc/plat-arcfpga/Makefile
new file mode 100644
index 0000000..385eb9d
--- /dev/null
+++ b/arch/arc/plat-arcfpga/Makefile
@@ -0,0 +1,9 @@
+#
+# Copyright (C) 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+#
+# This program is free software; you can redistribute it and/or modify
+# it under the terms of the GNU General Public License version 2 as
+# published by the Free Software Foundation.
+#
+
+obj-y := platform.o irq.o
--
1.7.4.1
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/tlb.h | 17 ++
arch/arc/include/asm/tlbflush.h | 28 ++++
arch/arc/mm/tlb.c | 311 +++++++++++++++++++++++++++++++++++++++
3 files changed, 356 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/tlbflush.h
diff --git a/arch/arc/include/asm/tlb.h b/arch/arc/include/asm/tlb.h
index b571e12..3eb2ce0 100644
--- a/arch/arc/include/asm/tlb.h
+++ b/arch/arc/include/asm/tlb.h
@@ -21,6 +21,23 @@
#ifndef __ASSEMBLY__
+#define tlb_flush(tlb) local_flush_tlb_mm((tlb)->mm)
+
+/*
+ * This pair is called at time of munmap/exit to flush cache and TLB entries
+ * for mappings being torn down.
+ * 1) cache-flush part -implemented via tlb_start_vma( ) can be NOP (for now)
+ * as we don't support aliasing configs in our VIPT D$.
+ * 2) tlb-flush part - implemted via tlb_end_vma( ) can be NOP as well-
+ * albiet for difft reasons - its better handled by moving to new ASID
+ *
+ * Note, read http://lkml.org/lkml/2004/1/15/6
+ */
+#define tlb_start_vma(tlb, vma)
+#define tlb_end_vma(tlb, vma)
+
+#define __tlb_remove_tlb_entry(tlb, ptep, address)
+
#include <linux/pagemap.h>
#include <asm-generic/tlb.h>
diff --git a/arch/arc/include/asm/tlbflush.h b/arch/arc/include/asm/tlbflush.h
new file mode 100644
index 0000000..b2f9bc7
--- /dev/null
+++ b/arch/arc/include/asm/tlbflush.h
@@ -0,0 +1,28 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef __ASM_ARC_TLBFLUSH__
+#define __ASM_ARC_TLBFLUSH__
+
+#include <linux/mm.h>
+
+void local_flush_tlb_all(void);
+void local_flush_tlb_mm(struct mm_struct *mm);
+void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long page);
+void local_flush_tlb_kernel_range(unsigned long start, unsigned long end);
+void local_flush_tlb_range(struct vm_area_struct *vma,
+ unsigned long start, unsigned long end);
+
+/* XXX: Revisit for SMP */
+#define flush_tlb_range(vma, s, e) local_flush_tlb_range(vma, s, e)
+#define flush_tlb_page(vma, page) local_flush_tlb_page(vma, page)
+#define flush_tlb_kernel_range(s, e) local_flush_tlb_kernel_range(s, e)
+#define flush_tlb_all() local_flush_tlb_all()
+#define flush_tlb_mm(mm) local_flush_tlb_mm(mm)
+
+#endif
diff --git a/arch/arc/mm/tlb.c b/arch/arc/mm/tlb.c
index 41afc84..c10111d 100644
--- a/arch/arc/mm/tlb.c
+++ b/arch/arc/mm/tlb.c
@@ -6,13 +6,97 @@
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License version 2 as
* published by the Free Software Foundation.
+ *
+ * vineetg: Aug 2011
+ * -Reintroduce duplicate PD fixup - some customer chips still have the issue
+ *
+ * vineetg: May 2011
+ * -No need to flush_cache_page( ) for each call to update_mmu_cache()
+ * some of the LMBench tests improved amazingly
+ * = page-fault thrice as fast (75 usec to 28 usec)
+ * = mmap twice as fast (9.6 msec to 4.6 msec),
+ * = fork (5.3 msec to 3.7 msec)
+ *
+ * vineetg: April 2011 :
+ * -MMU v3: PD{0,1} bits layout changed: They don't overlap anymore,
+ * helps avoid a shift when preparing PD0 from PTE
+ *
+ * vineetg: April 2011 : Preparing for MMU V3
+ * -MMU v2/v3 BCRs decoded differently
+ * -Remove TLB_SIZE hardcoding as it's variable now: 256 or 512
+ * -tlb_entry_erase( ) can be void
+ * -local_flush_tlb_range( ):
+ * = need not "ceil" @end
+ * = walks MMU only if range spans < 32 entries, as opposed to 256
+ *
+ * Vineetg: Sept 10th 2008
+ * -Changes related to MMU v2 (Rel 4.8)
+ *
+ * Vineetg: Aug 29th 2008
+ * -In TLB Flush operations (Metal Fix MMU) there is a explict command to
+ * flush Micro-TLBS. If TLB Index Reg is invalid prior to TLBIVUTLB cmd,
+ * it fails. Thus need to load it with ANY valid value before invoking
+ * TLBIVUTLB cmd
+ *
+ * Vineetg: Aug 21th 2008:
+ * -Reduced the duration of IRQ lockouts in TLB Flush routines
+ * -Multiple copies of TLB erase code seperated into a "single" function
+ * -In TLB Flush routines, interrupt disabling moved UP to retrieve ASID
+ * in interrupt-safe region.
+ *
+ * Vineetg: April 23rd Bug #93131
+ * Problem: tlb_flush_kernel_range() doesnt do anything if the range to
+ * flush is more than the size of TLB itself.
+ *
+ * Rahul Trivedi : Codito Technologies 2004
*/
#include <linux/module.h>
#include <asm/arcregs.h>
+#include <asm/setup.h>
#include <asm/mmu_context.h>
#include <asm/tlb.h>
+/* Need for ARC MMU v2
+ *
+ * ARC700 MMU-v1 had a Joint-TLB for Code and Data and is 2 way set-assoc.
+ * For a memcpy operation with 3 players (src/dst/code) such that all 3 pages
+ * map into same set, there would be contention for the 2 ways causing severe
+ * Thrashing.
+ *
+ * Although J-TLB is 2 way set assoc, ARC700 caches J-TLB into uTLBS which has
+ * much higher associativity. u-D-TLB is 8 ways, u-I-TLB is 4 ways.
+ * Given this, the thrasing problem should never happen because once the 3
+ * J-TLB entries are created (even though 3rd will knock out one of the prev
+ * two), the u-D-TLB and u-I-TLB will have what is required to accomplish memcpy
+ *
+ * Yet we still see the Thrashing because a J-TLB Write cause flush of u-TLBs.
+ * This is a simple design for keeping them in sync. So what do we do?
+ * The solution which James came up was pretty neat. It utilised the assoc
+ * of uTLBs by not invalidating always but only when absolutely necessary.
+ *
+ * - Existing TLB commands work as before
+ * - New command (TLBWriteNI) for TLB write without clearing uTLBs
+ * - New command (TLBIVUTLB) to invalidate uTLBs.
+ *
+ * The uTLBs need only be invalidated when pages are being removed from the
+ * OS page table. If a 'victim' TLB entry is being overwritten in the main TLB
+ * as a result of a miss, the removed entry is still allowed to exist in the
+ * uTLBs as it is still valid and present in the OS page table. This allows the
+ * full associativity of the uTLBs to hide the limited associativity of the main
+ * TLB.
+ *
+ * During a miss handler, the new "TLBWriteNI" command is used to load
+ * entries without clearing the uTLBs.
+ *
+ * When the OS page table is updated, TLB entries that may be associated with a
+ * removed page are removed (flushed) from the TLB using TLBWrite. In this
+ * circumstance, the uTLBs must also be cleared. This is done by using the
+ * existing TLBWrite command. An explicit IVUTLB is also required for those
+ * corner cases when TLBWrite was not executed at all because the corresp
+ * J-TLB entry got evicted/replaced.
+ */
+
/* A copy of the ASID from the PID reg is kept in asid_cache */
int asid_cache = FIRST_ASID;
@@ -22,6 +106,233 @@ int asid_cache = FIRST_ASID;
*/
struct mm_struct *asid_mm_map[NUM_ASID + 1];
+/*
+ * Utility Routine to erase a J-TLB entry
+ * The procedure is to look it up in the MMU. If found, ERASE it by
+ * issuing a TlbWrite CMD with PD0 = PD1 = 0
+ */
+
+static void __tlb_entry_erase(void)
+{
+ write_aux_reg(ARC_REG_TLBPD1, 0);
+ write_aux_reg(ARC_REG_TLBPD0, 0);
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBWrite);
+}
+
+static void tlb_entry_erase(unsigned int vaddr_n_asid)
+{
+ unsigned int idx;
+
+ /* Locate the TLB entry for this vaddr + ASID */
+ write_aux_reg(ARC_REG_TLBPD0, vaddr_n_asid);
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBProbe);
+ idx = read_aux_reg(ARC_REG_TLBINDEX);
+
+ /* No error means entry found, zero it out */
+ if (likely(!(idx & TLB_LKUP_ERR))) {
+ __tlb_entry_erase();
+ } else { /* Some sort of Error */
+
+ /* Duplicate entry error */
+ if (idx & 0x1) {
+ /* TODO we need to handle this case too */
+ pr_emerg("unhandled Duplicate flush for %x\n",
+ vaddr_n_asid);
+ }
+ /* else entry not found so nothing to do */
+ }
+}
+
+/****************************************************************************
+ * ARC700 MMU caches recently used J-TLB entries (RAM) as uTLBs (FLOPs)
+ *
+ * New IVUTLB cmd in MMU v2 explictly invalidates the uTLB
+ *
+ * utlb_invalidate ( )
+ * -For v2 MMU calls Flush uTLB Cmd
+ * -For v1 MMU does nothing (except for Metal Fix v1 MMU)
+ * This is because in v1 TLBWrite itself invalidate uTLBs
+ ***************************************************************************/
+
+static void utlb_invalidate(void)
+{
+#if (CONFIG_ARC_MMU_VER >= 2)
+
+#if (CONFIG_ARC_MMU_VER < 3)
+ /* MMU v2 introduced the uTLB Flush command.
+ * There was however an obscure hardware bug, where uTLB flush would
+ * fail when a prior probe for J-TLB (both totally unrelated) would
+ * return lkup err - because the entry didnt exist in MMU.
+ * The Workround was to set Index reg with some valid value, prior to
+ * flush. This was fixed in MMU v3 hence not needed any more
+ */
+ unsigned int idx;
+
+ /* make sure INDEX Reg is valid */
+ idx = read_aux_reg(ARC_REG_TLBINDEX);
+
+ /* If not write some dummy val */
+ if (unlikely(idx & TLB_LKUP_ERR))
+ write_aux_reg(ARC_REG_TLBINDEX, 0xa);
+#endif
+
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBIVUTLB);
+#endif
+
+}
+
+/*
+ * Un-conditionally (without lookup) erase the entire MMU contents
+ */
+
+noinline void local_flush_tlb_all(void)
+{
+ unsigned long flags;
+ unsigned int entry;
+ struct cpuinfo_arc_mmu *mmu = &cpuinfo_arc700[smp_processor_id()].mmu;
+
+ local_irq_save(flags);
+
+ /* Load PD0 and PD1 with template for a Blank Entry */
+ write_aux_reg(ARC_REG_TLBPD1, 0);
+ write_aux_reg(ARC_REG_TLBPD0, 0);
+
+ for (entry = 0; entry < mmu->num_tlb; entry++) {
+ /* write this entry to the TLB */
+ write_aux_reg(ARC_REG_TLBINDEX, entry);
+ write_aux_reg(ARC_REG_TLBCOMMAND, TLBWrite);
+ }
+
+ utlb_invalidate();
+
+ local_irq_restore(flags);
+}
+
+/*
+ * Flush the entrie MM for userland. The fastest way is to move to Next ASID
+ */
+noinline void local_flush_tlb_mm(struct mm_struct *mm)
+{
+ /*
+ * Small optimisation courtesy IA64
+ * flush_mm called during fork,exit,munmap etc, multiple times as well.
+ * Only for fork( ) do we need to move parent to a new MMU ctxt,
+ * all other cases are NOPs, hence this check.
+ */
+ if (atomic_read(&mm->mm_users) == 0)
+ return;
+
+ /*
+ * Workaround for Android weirdism:
+ * A binder VMA could end up in a task such that vma->mm != tsk->mm
+ * old code would cause h/w - s/w ASID to get out of sync
+ */
+ if (current->mm != mm)
+ destroy_context(mm);
+ else
+ get_new_mmu_context(mm);
+}
+
+/*
+ * Flush a Range of TLB entries for userland.
+ * @start is inclusive, while @end is exclusive
+ * Difference between this and Kernel Range Flush is
+ * -Here the fastest way (if range is too large) is to move to next ASID
+ * without doing any explicit Shootdown
+ * -In case of kernel Flush, entry has to be shot down explictly
+ */
+void local_flush_tlb_range(struct vm_area_struct *vma, unsigned long start,
+ unsigned long end)
+{
+ unsigned long flags;
+ unsigned int asid;
+
+ /* If range @start to @end is more than 32 TLB entries deep,
+ * its better to move to a new ASID rather than searching for
+ * individual entries and then shooting them down
+ *
+ * The calc above is rough, doesn't account for unaligned parts,
+ * since this is heuristics based anyways
+ */
+ if (unlikely((end - start) >= PAGE_SIZE * 32)) {
+ local_flush_tlb_mm(vma->vm_mm);
+ return;
+ }
+
+ /*
+ * @start moved to page start: this alone suffices for checking
+ * loop end condition below, w/o need for aligning @end to end
+ * e.g. 2000 to 4001 will anyhow loop twice
+ */
+ start &= PAGE_MASK;
+
+ local_irq_save(flags);
+ asid = vma->vm_mm->context.asid;
+
+ if (asid != NO_ASID) {
+ while (start < end) {
+ tlb_entry_erase(start | (asid & 0xff));
+ start += PAGE_SIZE;
+ }
+ }
+
+ utlb_invalidate();
+
+ local_irq_restore(flags);
+}
+
+/* Flush the kernel TLB entries - vmalloc/modules (Global from MMU perspective)
+ * @start, @end interpreted as kvaddr
+ * Interestingly, shared TLB entries can also be flushed using just
+ * @start,@end alone (interpreted as user vaddr), although technically SASID
+ * is also needed. However our smart TLbProbe lookup takes care of that.
+ */
+void local_flush_tlb_kernel_range(unsigned long start, unsigned long end)
+{
+ unsigned long flags;
+
+ /* exactly same as above, except for TLB entry not taking ASID */
+
+ if (unlikely((end - start) >= PAGE_SIZE * 32)) {
+ local_flush_tlb_all();
+ return;
+ }
+
+ start &= PAGE_MASK;
+
+ local_irq_save(flags);
+ while (start < end) {
+ tlb_entry_erase(start);
+ start += PAGE_SIZE;
+ }
+
+ utlb_invalidate();
+
+ local_irq_restore(flags);
+}
+
+/*
+ * Delete TLB entry in MMU for a given page (??? address)
+ * NOTE One TLB entry contains translation for single PAGE
+ */
+
+void local_flush_tlb_page(struct vm_area_struct *vma, unsigned long page)
+{
+ unsigned long flags;
+
+ /* Note that it is critical that interrupts are DISABLED between
+ * checking the ASID and using it flush the TLB entry
+ */
+ local_irq_save(flags);
+
+ if (vma->vm_mm->context.asid != NO_ASID) {
+ tlb_entry_erase((page & PAGE_MASK) |
+ (vma->vm_mm->context.asid & 0xff));
+ utlb_invalidate();
+ }
+
+ local_irq_restore(flags);
+}
/*
* Routine to create a TLB entry
--
1.7.4.1
TBD: Do we need early ioremap support like openrisc
Signed-off-by: Vineet Gupta <[email protected]>
---
arch/arc/include/asm/dma-mapping.h | 176 +++++++++++++++++++++++++
arch/arc/include/asm/dma.h | 14 ++
arch/arc/include/asm/io.h | 26 ++++
arch/arc/mm/dma.c | 91 +++++++++++++
arch/arc/mm/ioremap.c | 62 +++++++++
arch/arc/plat-arcfpga/include/plat/dma_addr.h | 45 +++++++
6 files changed, 414 insertions(+), 0 deletions(-)
create mode 100644 arch/arc/include/asm/dma-mapping.h
create mode 100644 arch/arc/include/asm/dma.h
create mode 100644 arch/arc/include/asm/io.h
create mode 100644 arch/arc/mm/dma.c
create mode 100644 arch/arc/mm/ioremap.c
create mode 100644 arch/arc/plat-arcfpga/include/plat/dma_addr.h
diff --git a/arch/arc/include/asm/dma-mapping.h b/arch/arc/include/asm/dma-mapping.h
new file mode 100644
index 0000000..ea0f56f
--- /dev/null
+++ b/arch/arc/include/asm/dma-mapping.h
@@ -0,0 +1,176 @@
+/*
+ * DMA Mapping glue for ARC
+ *
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef ASM_ARC_DMA_MAPPING_H
+#define ASM_ARC_DMA_MAPPING_H
+
+#include <asm-generic/dma-coherent.h>
+#include <plat/dma_addr.h>
+
+void *dma_alloc_noncoherent(struct device *dev, size_t size,
+ dma_addr_t *dma_handle, gfp_t gfp);
+
+void dma_free_noncoherent(struct device *dev, size_t size, void *vaddr,
+ dma_addr_t dma_handle);
+
+void *dma_alloc_coherent(struct device *dev, size_t size,
+ dma_addr_t *dma_handle, gfp_t gfp);
+
+void dma_free_coherent(struct device *dev, size_t size, void *kvaddr,
+ dma_addr_t dma_handle);
+
+/*
+ * streaming DMA Mapping API...
+ * CPU accesses page via normal paddr, thus needs to explicitly made
+ * consistent before each use
+ */
+
+static inline void __inline_dma_cache_sync(unsigned long paddr, size_t size,
+ enum dma_data_direction dir)
+{
+ switch (dir) {
+ case DMA_FROM_DEVICE:
+ dma_cache_inv(paddr, size);
+ break;
+ case DMA_TO_DEVICE:
+ dma_cache_wback(paddr, size);
+ break;
+ case DMA_BIDIRECTIONAL:
+ dma_cache_wback_inv(paddr, size);
+ break;
+ default:
+ pr_err("Invalid DMA dir [%d] for OP @ %lx\n", dir, paddr);
+ }
+}
+
+void __arc_dma_cache_sync(unsigned long paddr, size_t size,
+ enum dma_data_direction dir);
+
+#define _dma_cache_sync(addr, sz, dir) \
+do { \
+ if (__builtin_constant_p(dir)) \
+ __inline_dma_cache_sync(addr, sz, dir); \
+ else \
+ __arc_dma_cache_sync(addr, sz, dir); \
+} \
+while (0);
+
+static inline dma_addr_t
+dma_map_single(struct device *dev, void *cpu_addr, size_t size,
+ enum dma_data_direction dir)
+{
+ _dma_cache_sync((unsigned long)cpu_addr, size, dir);
+ return plat_kernel_addr_to_dma(dev, cpu_addr);
+}
+
+static inline void
+dma_unmap_single(struct device *dev, dma_addr_t dma_addr,
+ size_t size, enum dma_data_direction dir)
+{
+}
+
+static inline dma_addr_t
+dma_map_page(struct device *dev, struct page *page,
+ unsigned long offset, size_t size,
+ enum dma_data_direction dir)
+{
+ unsigned long paddr = page_to_phys(page) + offset;
+ return dma_map_single(dev, (void *)paddr, size, dir);
+}
+
+static inline void
+dma_unmap_page(struct device *dev, dma_addr_t dma_handle,
+ size_t size, enum dma_data_direction dir)
+{
+}
+
+static inline int
+dma_map_sg(struct device *dev, struct scatterlist *sg,
+ int nents, enum dma_data_direction dir)
+{
+ struct scatterlist *s;
+ int i;
+
+ for_each_sg(sg, s, nents, i)
+ sg->dma_address = dma_map_page(dev, sg_page(s), s->offset,
+ s->length, dir);
+
+ return nents;
+}
+
+static inline void
+dma_unmap_sg(struct device *dev, struct scatterlist *sg,
+ int nents, enum dma_data_direction dir)
+{
+ struct scatterlist *s;
+ int i;
+
+ for_each_sg(sg, s, nents, i)
+ dma_unmap_page(dev, sg_dma_address(s), sg_dma_len(s), dir);
+}
+
+static inline void
+dma_sync_single_for_cpu(struct device *dev, dma_addr_t dma_handle,
+ size_t size, enum dma_data_direction dir)
+{
+ _dma_cache_sync(plat_dma_addr_to_kernel(dev, dma_handle), size,
+ DMA_FROM_DEVICE);
+}
+
+static inline void
+dma_sync_single_for_device(struct device *dev, dma_addr_t dma_handle,
+ size_t size, enum dma_data_direction dir)
+{
+ _dma_cache_sync(plat_dma_addr_to_kernel(dev, dma_handle), size,
+ DMA_TO_DEVICE);
+}
+
+static inline void
+dma_sync_sg_for_cpu(struct device *dev, struct scatterlist *sg, int nelems,
+ enum dma_data_direction dir)
+{
+ int i;
+
+ for (i = 0; i < nelems; i++, sg++)
+ _dma_cache_sync((unsigned int)sg_virt(sg), sg->length, dir);
+}
+
+static inline void
+dma_sync_sg_for_device(struct device *dev, struct scatterlist *sg, int nelems,
+ enum dma_data_direction dir)
+{
+ int i;
+
+ for (i = 0; i < nelems; i++, sg++)
+ _dma_cache_sync((unsigned int)sg_virt(sg), sg->length, dir);
+}
+
+static inline int dma_supported(struct device *dev, u64 dma_mask)
+{
+ /* Support 32 bit DMA mask exclusively */
+ return dma_mask == DMA_BIT_MASK(32);
+}
+
+static inline int dma_mapping_error(struct device *dev, dma_addr_t dma_addr)
+{
+ return 0;
+}
+
+static inline int dma_set_mask(struct device *dev, u64 dma_mask)
+{
+ if (!dev->dma_mask || !dma_supported(dev, dma_mask))
+ return -EIO;
+
+ *dev->dma_mask = dma_mask;
+
+ return 0;
+}
+
+#endif
diff --git a/arch/arc/include/asm/dma.h b/arch/arc/include/asm/dma.h
new file mode 100644
index 0000000..ca7c451
--- /dev/null
+++ b/arch/arc/include/asm/dma.h
@@ -0,0 +1,14 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef ASM_ARC_DMA_H
+#define ASM_ARC_DMA_H
+
+#define MAX_DMA_ADDRESS 0xC0000000
+
+#endif
diff --git a/arch/arc/include/asm/io.h b/arch/arc/include/asm/io.h
new file mode 100644
index 0000000..b2221f5
--- /dev/null
+++ b/arch/arc/include/asm/io.h
@@ -0,0 +1,26 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#ifndef _ASM_ARC_IO_H
+#define _ASM_ARC_IO_H
+
+#include <asm/byteorder.h>
+#include <asm/page.h>
+
+extern void __iomem *ioremap(unsigned long physaddr, unsigned long size);
+extern void iounmap(const void __iomem *addr);
+
+#define ioremap_nocache(phy, sz) ioremap(phy, sz)
+#define ioremap_wc(phy, sz) ioremap(phy, sz)
+
+/* Change struct page to physical address */
+#define page_to_phys(page) (page_to_pfn(page) << PAGE_SHIFT)
+
+#include <asm-generic/io.h>
+
+#endif /* _ASM_ARC_IO_H */
diff --git a/arch/arc/mm/dma.c b/arch/arc/mm/dma.c
new file mode 100644
index 0000000..91fc4ef
--- /dev/null
+++ b/arch/arc/mm/dma.c
@@ -0,0 +1,91 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+/*
+ * Helpers for Coherent DMA API.
+ *
+ * Allocate pages(s) but the CPU always accesses them using a V-P mapping
+ * which has Cached Bit off
+ *
+ * If NON_CONSISTENT request, then CPU accesses page with normal paddr
+ *
+ * For both the cases above, actual DMA handle gen by platform as some
+ * platforms are OK with 0x8000_0000 baed addresses while some are NOT.
+ */
+
+#include <linux/dma-mapping.h>
+#include <linux/dma-debug.h>
+#include <linux/export.h>
+#include <asm/cacheflush.h>
+
+void *dma_alloc_noncoherent(struct device *dev, size_t size,
+ dma_addr_t *dma_handle, gfp_t gfp)
+{
+ void *paddr;
+
+ /* This is linear addr (0x8000_0000 based) */
+ paddr = alloc_pages_exact(size, gfp);
+ if (!paddr)
+ return NULL;
+
+ /* This is bus address, platform dependent */
+ *dma_handle = plat_kernel_addr_to_dma(dev, paddr);
+
+ return paddr;
+}
+EXPORT_SYMBOL(dma_alloc_noncoherent);
+
+void dma_free_noncoherent(struct device *dev, size_t size, void *vaddr,
+ dma_addr_t dma_handle)
+{
+ free_pages_exact((void *)plat_dma_addr_to_kernel(dev, dma_handle),
+ size);
+}
+EXPORT_SYMBOL(dma_free_noncoherent);
+
+void *dma_alloc_coherent(struct device *dev, size_t size,
+ dma_addr_t *dma_handle, gfp_t gfp)
+{
+ void *paddr, *kvaddr;
+
+ /* This is linear addr (0x8000_0000 based) */
+ paddr = alloc_pages_exact(size, gfp);
+ if (!paddr)
+ return NULL;
+
+ /* This is kernel Virtual address (0x7000_0000 based) */
+ kvaddr = ioremap_nocache((unsigned long)paddr, size);
+ if (kvaddr != NULL)
+ memset(kvaddr, 0, size);
+
+ /* This is bus address, platform dependent */
+ *dma_handle = plat_kernel_addr_to_dma(dev, paddr);
+
+ return kvaddr;
+}
+EXPORT_SYMBOL(dma_alloc_coherent);
+
+void dma_free_coherent(struct device *dev, size_t size, void *kvaddr,
+ dma_addr_t dma_handle)
+{
+ iounmap((void __force __iomem *)kvaddr);
+
+ free_pages_exact((void *)plat_dma_addr_to_kernel(dev, dma_handle),
+ size);
+}
+EXPORT_SYMBOL(dma_free_coherent);
+
+/*
+ * Helper for streaming DMA...
+ */
+void __arc_dma_cache_sync(unsigned long paddr, size_t size,
+ enum dma_data_direction dir)
+{
+ __inline_dma_cache_sync(paddr, size, dir);
+}
+EXPORT_SYMBOL(__arc_dma_cache_sync);
diff --git a/arch/arc/mm/ioremap.c b/arch/arc/mm/ioremap.c
new file mode 100644
index 0000000..a82ec3a
--- /dev/null
+++ b/arch/arc/mm/ioremap.c
@@ -0,0 +1,62 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ */
+
+#include <linux/vmalloc.h>
+#include <linux/init.h>
+#include <linux/module.h>
+#include <linux/io.h>
+#include <linux/mm.h>
+#include <asm/cache.h>
+
+void __iomem *ioremap(unsigned long paddr, unsigned long size)
+{
+ unsigned long vaddr;
+ struct vm_struct *area;
+ unsigned long off, end;
+ const pgprot_t prot = PAGE_KERNEL_NO_CACHE;
+
+ /* Don't allow wraparound or zero size */
+ end = paddr + size - 1;
+ if (!size || (end < paddr))
+ return NULL;
+
+ /* If the region is h/w uncached, nothing special needed */
+ if (paddr >= ARC_UNCACHED_ADDR_SPACE)
+ return (void __iomem *)paddr;
+
+ /* Mappings have to be page-aligned, page-sized */
+ off = paddr & ~PAGE_MASK;
+ paddr &= PAGE_MASK;
+ size = PAGE_ALIGN(end + 1) - paddr;
+
+ /*
+ * Ok, go for it..
+ */
+ area = get_vm_area(size, VM_IOREMAP);
+ if (!area)
+ return NULL;
+
+ area->phys_addr = paddr;
+ vaddr = (unsigned long)area->addr;
+ if (ioremap_page_range(vaddr, vaddr + size, paddr, prot)) {
+ vfree(area->addr);
+ return NULL;
+ }
+
+ return (void __iomem *)(off + (char __iomem *)vaddr);
+}
+EXPORT_SYMBOL(ioremap);
+
+void iounmap(const void __iomem *addr)
+{
+ if (addr >= (void __force __iomem *)ARC_UNCACHED_ADDR_SPACE)
+ return;
+
+ vfree((void *)(PAGE_MASK & (unsigned long __force)addr));
+}
+EXPORT_SYMBOL(iounmap);
diff --git a/arch/arc/plat-arcfpga/include/plat/dma_addr.h b/arch/arc/plat-arcfpga/include/plat/dma_addr.h
new file mode 100644
index 0000000..0e96343
--- /dev/null
+++ b/arch/arc/plat-arcfpga/include/plat/dma_addr.h
@@ -0,0 +1,45 @@
+/*
+ * Copyright (C) 2004, 2007-2010, 2011-2012 Synopsys, Inc. (http://www.synopsys.com)
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * vineetg: Feb 2009
+ * -For AA4 board, kernel to DMA address APIs
+ */
+
+/*
+ * kernel addresses are 0x800_000 based, while Bus addr are 0 based
+ */
+
+#ifndef __PLAT_DMA_ADDR_H
+#define __PLAT_DMA_ADDR_H
+
+#include <linux/device.h>
+
+static inline unsigned long plat_dma_addr_to_kernel(struct device *dev,
+ dma_addr_t dma_addr)
+{
+ return dma_addr + PAGE_OFFSET;
+}
+
+static inline dma_addr_t plat_kernel_addr_to_dma(struct device *dev, void *ptr)
+{
+ unsigned long addr = (unsigned long)ptr;
+ /*
+ * To Catch buggy drivers which can call DMA map API with kernel vaddr
+ * i.e. for buffers alloc via vmalloc or ioremap which are not
+ * gaurnateed to be PHY contiguous and hence unfit for DMA anyways.
+ * On ARC kernel virtual address is 0x7000_0000 to 0x7FFF_FFFF, so
+ * ideally we want to check this range here, but our implementation is
+ * better as it checks for even worse user virtual address as well.
+ */
+ if (likely(addr >= PAGE_OFFSET))
+ return addr - PAGE_OFFSET;
+
+ BUG();
+ return addr;
+}
+
+#endif
--
1.7.4.1
On Wednesday 07 November 2012, Vineet Gupta wrote:
> Signed-off-by: Vineet Gupta <[email protected]>
> ---
> arch/arc/configs/fpga_defconfig | 607 +++++++++++++++++++++++++++++++++++++++
> 1 files changed, 607 insertions(+), 0 deletions(-)
> create mode 100644 arch/arc/configs/fpga_defconfig
>
> diff --git a/arch/arc/configs/fpga_defconfig b/arch/arc/configs/fpga_defconfig
> new file mode 100644
> index 0000000..cfdca3c
> --- /dev/null
> +++ b/arch/arc/configs/fpga_defconfig
> @@ -0,0 +1,607 @@
> +#
> +# Automatically generated file; DO NOT EDIT.
> +# Linux/arc 3.7.0-rc3 Kernel Configuration
> +#
Please use "make defconfig" to generate minimal defconfig files, instead of
copying the entire file here.
Arnd
On Wednesday 07 November 2012, Vineet Gupta wrote:
> +
> +config ARC
> + def_bool y
> + select ARCH_WANT_IPC_PARSE_VERSION
> + # ARC Busybox based initramfs absolutely relies on DEVTMPFS for /dev
> + # DEVTMPS in turn needs HOTPLUG
> + select DEVTMPFS if !INITRAMFS_SOURCE=""
> + select GENERIC_ATOMIC64
> + select GENERIC_CLOCKEVENTS
> + select GENERIC_FIND_FIRST_BIT
> + # for now, we don't need GENERIC_IRQ_PROBE, CONFIG_GENERIC_IRQ_CHIP
> + select GENERIC_IRQ_SHOW
> + select GENERIC_PENDING_IRQ if SMP
> + select GENERIC_SMP_IDLE_THREAD
> + select HAVE_GENERIC_HARDIRQS
> + select HOTPLUG if !INITRAMFS_SOURCE=""
> + select MODULES_USE_ELF_RELA
You should not need ARCH_WANT_IPC_PARSE_VERSION
> +# for old_[ug]id_t - do we need this ?
> +config UID16
> + def_bool y
No, not needed either.
> +menu "ARC Architecture Configuration"
> +
> +choice
> + prompt "ARC Platform"
> + default ARC_PLAT_FPGA_LEGACY
> +
> +config ARC_PLAT_FPGA_LEGACY
> + bool "\"Legacy\" ARC FPGA dev platform"
> + help
> + Support for ARC development platforms, provided by Synopsys.
> + These are based on FPGA or ISS. e.g.
> + - ARCAngel4
> + - ML509
> + - MetaWare ISS
> +
> +#New platform adds here
> +endchoice
Platform selection should generally be non-exclusive. Just remove the
"choice" statement here, and make sure that any platforms can be enabled
together to build a single kernel.
> +menu "ARC CPU Configuration"
> +
> +choice
> + prompt "ARC Core"
> + default ARC_CPU_770
> +
> +config ARC_CPU_750D
> + bool "ARC750D"
> + help
> + Support for ARC750 core
> +
> +config ARC_CPU_770
> + bool "ARC770"
> + select ARC_CPU_REL_4_10
> + help
> + Support for ARC770 core introduced with Rel 4.10 (Summer 2011)
> + This core has a bunch of cool new features:
> + -MMU-v3: Variable Page Sz (4k, 8k, 16k), bigger J-TLB (128x4)
> + Shared Address Spaces (for sharing TLB entires in MMU)
> + -Caches: New Prog Model, Region Flush
> + -Insns: endian swap, load-locked/store-conditional, time-stamp-ctr
> +
> +endchoice
Same thing here: If the different CPUs can in theory run the same kernel
code, they should allow that. It doesn't stop you from making the default
to enable only one of them and optimize for that case.
> +
> +choice
> + prompt "ARC700 MMU Version"
> + default ARC_MMU_V3 if ARC_CPU_770
> + default ARC_MMU_V2 if ARC_CPU_750D
> +
> +endchoice
> +choice
> + prompt "MMU Page Size"
> + default ARC_PAGE_SIZE_8K
> +
>
> +
> +endchoice
For MMU, page size and endianess, I would not expect that to work though,
so they probably need to remain mutually exclusive.
> +menu "Platform Board Configuration"
> +
> +source "arch/arc/plat-arcfpga/Kconfig"
> +#New platform adds here
> +
> +config ARC_PLAT_CLK
> + int "Clk speed in Hz"
> + default "80000000"
> +
> +config LINUX_LINK_BASE
> + hex "Linux Link Address"
> + default "0x80000000"
> + help
> + ARC700 divides the 32 bit phy address space into two equal halves
> + -Lower 2G (0 - 0x7FFF_FFFF ) is user virtual, translated by MMU
> + -Upper 2G (0x8000_0000 onwards) is untranslated, for kernel
> + Typically Linux kernel is linked at the start of untransalted addr,
> + hence the default value of 0x8zs.
> + However some customers have peripherals mapped at this addr, so
> + Linux needs to be scooted a bit.
> + If you don't know what the above means, leave this setting alone.
> +
> +config ARC_PLAT_SDRAM_SIZE
> + hex "SD RAM Size"
> + default "0x10000000"
> + help
> + Implies the amount of SDRAM/DRAM Linux is going to claim/own.
> + The actual memory itself could be larger than this number. But for
> + all software purposes, this is the amt of memory.
> +
> +endmenu # "Platform Board Configuration"
Clock speed and memory size are things that the kernel should normally
be able to find out itself, if not from the hardware then from looking
at the device tree binary that gets passed from the boot loader, or
some other boot protocol that is already established.
> +if ARC_PLAT_FPGA_LEGACY
> +
> +choice
> + prompt "FPGA Board"
> +
> +config ARC_BOARD_ANGEL4
> + bool "ARC Angel4"
> + help
> + ARC Angel4 FPGA Ref Platform (Xilinx Virtex Based)
> +
> +config ARC_BOARD_ML509
> + bool "ML509"
> + help
> + ARC ML509 FPGA Ref Platform (Xilinx Virtex-5 Based)
> +
> +endchoice
Board files again should not be mutually exclusive.
Arnd
On Wednesday 07 November 2012, Vineet Gupta wrote:
> +static struct platform_device arc_uart##n##_dev = { \
> + .name = "arc-uart", \
> + .id = n, \
> + .num_resources = ARRAY_SIZE(arc_uart##n##_res), \
> + .resource = arc_uart##n##_res, \
> + .dev = { \
> + .platform_data = &arc_uart_info, \
> + }, \
> +}
> +
> +ARC_UART_DEV(0);
> +#if CONFIG_SERIAL_ARC_NR_PORTS > 1
> +ARC_UART_DEV(1);
> +#endif
> +
> +static struct platform_device *fpga_early_devs[] __initdata = {
> +#if defined(CONFIG_SERIAL_ARC_CONSOLE)
> + &arc_uart0_dev,
> +#endif
> +};
statically defining platform devices like this is considered very
bad style, especially since it prevents you from doing proper
boot-time configuration. Please get the available devices from
the boot loader. Normally this is done using a flattened device
tree blob that gets passed, unless you can probe the hardware
directly.
Arnd
On Wednesday 07 November 2012, Vineet Gupta wrote:
> + * Being uClibc based we need some of the deprecated syscalls:
> + * -Not emulated by uClibc at all
> + * unlink, mkdir,... (needed by Busybox, LTP etc)
> + * times (needed by LTP pan test harness)
> + * -Not emulated efficiently
> + * select: emulated using pselect (but extra code to chk usec > 1sec)
> + *
> + * some (send/recv) correctly emulated using (recfrom/sendto) and
> + * some arch specific ones (fork/vfork)can easily be emulated using clone but
> + * thats the price of using common-denominator....
> + */
> +#define __ARCH_WANT_SYSCALL_NO_AT
> +#define __ARCH_WANT_SYSCALL_NO_FLAGS
> +#define __ARCH_WANT_SYSCALL_OFF_T
> +#define __ARCH_WANT_SYSCALL_DEPRECATED
I'm pretty sure that this has been solved before, best get in contact with
the maintainers of the openrisc/c6x/hexagon platforms, that probably all
use uClibc without needing these.
You have to remove the legacy calls here.
Arnd
On Wednesday 07 November 2012, Vineet Gupta wrote:
> This patchset based off-of 3.7-rc3, introduces the Linux kernel port to
> ARC700 processor family (750D and 770D) from Synopsys.
>
> ARC700 is highly configurable and power efficient 32-bit RISC core with MMU.
> It is embedded in SoCs deployed in TV Set Top boxes, Digital Media Players,
> all the way to Network-on-Chips.
Hi Vineet,
I'm largely quite happy with how the series has turned out, having looked
at earlier versions of the ARC kernel a couple of years ago. I've commented
already on individual patches, but overall, I see two main issues that
need to be resolved in order to merge the port:
* You have to use the generic syscall interface, and that means not using
any of the legacy system calls that have since been replaced by newer
versions of the same.
* You are missing dynamic hardware detection. Rather than building a kernel
with everything known about the hardware at compile time, new ports these
days are normally able to run on all kinds of hardware and detect the
differences by looking at configuration registers (e.g. PCI), asking
firmware (Open Firmware, ACPI, ...) or by interpreting a device tree
that is passed by the boot loader (most embedded systems).
I assume that device tree is the right solution for you because a lot of
the hardware you use is likely shared with ARM, PowerPC or MIPS based
SoCs and they are all using (or getting migrated to) DT now instead of
static platform devices,
Arnd
On Wed, Nov 7, 2012 at 11:47 AM, Vineet Gupta
<[email protected]> wrote:
> Hi,
>
> This patchset based off-of 3.7-rc3, introduces the Linux kernel port to
> ARC700 processor family (750D and 770D) from Synopsys.
We've been using the port for 6 month now at Ezchip running on top of
FPGA in 4 way SMP configuration (support for which will no doubt
follow later) and have subjected it to various stress tests (hackbench
and such) and it is working up quite well over all.
I realize some code style or organization changes might be in order
but at the same time wanted to attest the nature of its good and
behavior performance on real hardware; so for what it's worth:
Tested-by: Gilad Ben-Yossef <[email protected]>
Cheers,
Gilad
--
Gilad Ben-Yossef
Chief Coffee Drinker
[email protected]
Israel Cell: +972-52-8260388
US Cell: +1-973-8260388
http://benyossef.com
"If you take a class in large-scale robotics, can you end up in a
situation where the homework eats your dog?"
-- Jean-Baptiste Queru
On 7 November 2012 10:47, Vineet Gupta <[email protected]> wrote:
> diff --git a/arch/arc/include/asm/ptrace.h b/arch/arc/include/asm/ptrace.h
> new file mode 100644
> index 0000000..c178357
> --- /dev/null
> +++ b/arch/arc/include/asm/ptrace.h
> @@ -0,0 +1,120 @@
> +/* THE pt_regs: Defines how regs are saved during entry into kernel */
> +
> +struct pt_regs {
I'd recommend not exporting the pt_regs structure to userspace. This
struct is used heavily within the kernel and it's nice to have the
leeway to be able to modify it as things evolve. GDB doesn't need to
know about this struct anymore as it should be using regset's for any
recent architectures anyway, so it's the regset layout that should be
ABI stable instead of pt_regs.
You can check the openrisc architecture as an example of how to do this.
/Jonas
On 7 November 2012 10:47, Vineet Gupta <[email protected]> wrote:
>> diff --git a/arch/arc/include/asm/ptrace.h b/arch/arc/include/asm/ptrace.h
>> new file mode 100644
>> index 0000000..c178357
>> --- /dev/null
>> +++ b/arch/arc/include/asm/ptrace.h
>> @@ -0,0 +1,120 @@
>> +/* THE pt_regs: Defines how regs are saved during entry into kernel */
>> +
>> +struct pt_regs {
> I'd recommend not exporting the pt_regs structure to userspace. This
> struct is used heavily within the kernel and it's nice to have the
> leeway to be able to modify it as things evolve. GDB doesn't need to
> know about this struct anymore as it should be using regset's for any
> recent architectures anyway, so it's the regset layout that should be
> ABI stable instead of pt_regs.
>
> You can check the openrisc architecture as an example of how to do this.
Hi Jonas,
thanks for your review comment. I completely agree that pt_regs should not be exported and very recently a change in pt_regs forced a gdbserver change too (ABI incompatibility) which goes along what you are recommending. Infact the ptrace patch for kernel (which will follow in series #2) provides a stable regset ABI - extracting information from pt_regs.
However our current gdb/gdbserver is 6.8 based and making it switch to regset interface might not be possible for this release of tools. Since customers are already using our stuff, we can not have a broken ABI. We do have ABI versioning, so in next release we can fix gdb and remove this.
Please note that an additional reason for exporting pt_regs is due to the fact that it is part of sigcontext. Keeping it exactly same as pt_regs helps us do batch save/restore of user context in signal handling (please look at my signal handling patch) but the flip side is that userspace SA_SIGINFO needs to be able to have access to sigcontext and hence we explicitly need pt_regs. We could arguably opencode pt_regs there - but that won't be clean IMHO.
What do you think !
-Vineet-
Hi Arnd,
Thanks for your valuable comments I will rework the port.
P.S. Sorry for top-posting.
Thanks,
Vineet
On Wednesday 07 November 2012, Vineet Gupta wrote:
> This patchset based off-of 3.7-rc3, introduces the Linux kernel port to
> ARC700 processor family (750D and 770D) from Synopsys.
>
> ARC700 is highly configurable and power efficient 32-bit RISC core with MMU.
> It is embedded in SoCs deployed in TV Set Top boxes, Digital Media Players,
> all the way to Network-on-Chips.
Hi Vineet,
I'm largely quite happy with how the series has turned out, having looked
at earlier versions of the ARC kernel a couple of years ago. I've commented
already on individual patches, but overall, I see two main issues that
need to be resolved in order to merge the port:
* You have to use the generic syscall interface, and that means not using
any of the legacy system calls that have since been replaced by newer
versions of the same.
* You are missing dynamic hardware detection. Rather than building a kernel
with everything known about the hardware at compile time, new ports these
days are normally able to run on all kinds of hardware and detect the
differences by looking at configuration registers (e.g. PCI), asking
firmware (Open Firmware, ACPI, ...) or by interpreting a device tree
that is passed by the boot loader (most embedded systems).
I assume that device tree is the right solution for you because a lot of
the hardware you use is likely shared with ARM, PowerPC or MIPS based
SoCs and they are all using (or getting migrated to) DT now instead of
static platform devices,
Arnd
On 8 November 2012 19:52, Vineet Gupta <[email protected]> wrote:
> On 7 November 2012 10:47, Vineet Gupta <[email protected]> wrote:
>> I'd recommend not exporting the pt_regs structure to userspace. This
>> struct is used heavily within the kernel and it's nice to have the
>> leeway to be able to modify it as things evolve. GDB doesn't need to
>> know about this struct anymore as it should be using regset's for any
>> recent architectures anyway, so it's the regset layout that should be
>> ABI stable instead of pt_regs.
>>
>> You can check the openrisc architecture as an example of how to do this.
>
> Hi Jonas,
>
> thanks for your review comment. I completely agree that pt_regs should not be exported and very recently a change in pt_regs forced a gdbserver change too (ABI incompatibility) which goes along what you are recommending. Infact the ptrace patch for kernel (which will follow in series #2) provides a stable regset ABI - extracting information from pt_regs.
>
> However our current gdb/gdbserver is 6.8 based and making it switch to regset interface might not be possible for this release of tools. Since customers are already using our stuff, we can not have a broken ABI. We do have ABI versioning, so in next release we can fix gdb and remove this.
For generic syscalls it was generally decided that the upstream kernel
would drop support for legacy syscalls, _even though_ userspace bits
like uClibc required them. For _new_ architectures, userspace would
need to adapt to the _mainline_ kernel; and as a stop-gap for existing
software, external patches could be carried in an external git
repository where _non-mainline_ kernel patches could be carried to
provide a working kernel until userspace could catch up.
I'd say the same applies here: GDB 6.8 may continue work with a
mainline kernel plus a patch that restores the export of pt_regs, but
GDB 7.2 (or whatever version you move on to) would be the one that
fully supports an unpatched mainline.
>
> Please note that an additional reason for exporting pt_regs is due to the fact that it is part of sigcontext. Keeping it exactly same as pt_regs helps us do batch save/restore of user context in signal handling (please look at my signal handling patch) but the flip side is that userspace SA_SIGINFO needs to be able to have access to sigcontext and hence we explicitly need pt_regs. We could arguably opencode pt_regs there - but that won't be clean IMHO.
It's not really pt_regs you need here, but the userspace equivalent
thereof. On OpenRISC, for example, we have user_regs_struct which is
an ABI stable version of pt_regs. These may even be identical to
begin with, but at least the distinction can be made now between the
kernel-internal struct and the ABI stable one.
/Jonas
On 07/11/12 14:21, Arnd Bergmann wrote:
> On Wednesday 07 November 2012, Vineet Gupta wrote:
>> + * Being uClibc based we need some of the deprecated syscalls:
>> + * -Not emulated by uClibc at all
>> + * unlink, mkdir,... (needed by Busybox, LTP etc)
>> + * times (needed by LTP pan test harness)
>> + * -Not emulated efficiently
>> + * select: emulated using pselect (but extra code to chk usec > 1sec)
>> + *
>> + * some (send/recv) correctly emulated using (recfrom/sendto) and
>> + * some arch specific ones (fork/vfork)can easily be emulated using clone but
>> + * thats the price of using common-denominator....
>> + */
>> +#define __ARCH_WANT_SYSCALL_NO_AT
>> +#define __ARCH_WANT_SYSCALL_NO_FLAGS
>> +#define __ARCH_WANT_SYSCALL_OFF_T
>> +#define __ARCH_WANT_SYSCALL_DEPRECATED
>
> I'm pretty sure that this has been solved before, best get in contact with
> the maintainers of the openrisc/c6x/hexagon platforms, that probably all
> use uClibc without needing these.
>
> You have to remove the legacy calls here.
Hi Vineet,
FYI, we'll soon be making some uClibc patches available so that it can
work without these deprecated syscalls and with the newer IPC interface
(the work was needed for metag too). It's based roughly on the OpenRisc
patches from last year, but with a fair few changes.
Thanks
James
On Friday 09 November 2012 02:06 AM, Jonas Bonn wrote:
> On 8 November 2012 19:52, Vineet Gupta <[email protected]> wrote:
>> On 7 November 2012 10:47, Vineet Gupta <[email protected]> wrote:
>>> I'd recommend not exporting the pt_regs structure to userspace. This
>>> struct is used heavily within the kernel and it's nice to have the
>>> leeway to be able to modify it as things evolve. GDB doesn't need to
>>> know about this struct anymore as it should be using regset's for any
>>> recent architectures anyway, so it's the regset layout that should be
>>> ABI stable instead of pt_regs.
>>>
>>> You can check the openrisc architecture as an example of how to do this.
>> Hi Jonas,
>>
>> thanks for your review comment. I completely agree that pt_regs should not be exported and very recently a change in pt_regs forced a gdbserver change too (ABI incompatibility) which goes along what you are recommending. Infact the ptrace patch for kernel (which will follow in series #2) provides a stable regset ABI - extracting information from pt_regs.
>>
>> However our current gdb/gdbserver is 6.8 based and making it switch to regset interface might not be possible for this release of tools. Since customers are already using our stuff, we can not have a broken ABI. We do have ABI versioning, so in next release we can fix gdb and remove this.
> For generic syscalls it was generally decided that the upstream kernel
> would drop support for legacy syscalls, _even though_ userspace bits
> like uClibc required them. For _new_ architectures, userspace would
> need to adapt to the _mainline_ kernel; and as a stop-gap for existing
> software, external patches could be carried in an external git
> repository where _non-mainline_ kernel patches could be carried to
> provide a working kernel until userspace could catch up.
>
> I'd say the same applies here: GDB 6.8 may continue work with a
> mainline kernel plus a patch that restores the export of pt_regs, but
> GDB 7.2 (or whatever version you move on to) would be the one that
> fully supports an unpatched mainline.
So there are 2 points/issues to discuss here.
1. Not exporting pt_regs
2. Asking customers to refer to a non mainline tree.
Will adding a #ifdef __KERNEL__ guard over pt_regs (which my ptrace
patch posted today as 33/55 already does), help serve the cause. So we
don't export any real "C" structs (pt_regs and/or user_regs_struct....).
All we export now is just a bunch of #defs called PT_xxx, to refer to
offsets with user_regs_struct.
And it doesn't matter if user_regs_struct has pt_regs as an element -
since we don't export the struct at all.
But just to clarify, gdbserver will still use ARC specific
ptrace(PEEKUSR) as opposed to regset interface. The switch to new gdb
version is non trivial and I can't ask our active customers to go to
some out-of-mainline tree until that gets fixed.
>> Please note that an additional reason for exporting pt_regs is due to the fact that it is part of sigcontext. Keeping it exactly same as pt_regs helps us do batch save/restore of user context in signal handling (please look at my signal handling patch) but the flip side is that userspace SA_SIGINFO needs to be able to have access to sigcontext and hence we explicitly need pt_regs. We could arguably opencode pt_regs there - but that won't be clean IMHO.
> It's not really pt_regs you need here, but the userspace equivalent
> thereof. On OpenRISC, for example, we have user_regs_struct which is
> an ABI stable version of pt_regs. These may even be identical to
> begin with, but at least the distinction can be made now between the
> kernel-internal struct and the ABI stable one.
>
> /Jonas
On Monday 12 November 2012, Vineet Gupta wrote:
> But just to clarify, gdbserver will still use ARC specific
> ptrace(PEEKUSR) as opposed to regset interface. The switch to new gdb
> version is non trivial and I can't ask our active customers to go to
> some out-of-mainline tree until that gets fixed.
But your customers could use an out-of-tree kernel patch to put the
implementation of PEEKUSR back into their kernels, right?
Over time, they could just drop that patch once they move to a
newer gdb version.
Arnd
On 07/11/12 14:06, Arnd Bergmann wrote:
> On Wednesday 07 November 2012, Vineet Gupta wrote:
>> Signed-off-by: Vineet Gupta <[email protected]>
>> ---
>> arch/arc/configs/fpga_defconfig | 607 +++++++++++++++++++++++++++++++++++++++
>> 1 files changed, 607 insertions(+), 0 deletions(-)
>> create mode 100644 arch/arc/configs/fpga_defconfig
>>
>> diff --git a/arch/arc/configs/fpga_defconfig b/arch/arc/configs/fpga_defconfig
>> new file mode 100644
>> index 0000000..cfdca3c
>> --- /dev/null
>> +++ b/arch/arc/configs/fpga_defconfig
>> @@ -0,0 +1,607 @@
>> +#
>> +# Automatically generated file; DO NOT EDIT.
>> +# Linux/arc 3.7.0-rc3 Kernel Configuration
>> +#
>
> Please use "make defconfig" to generate minimal defconfig files, instead of
> copying the entire file here.
Hi Arnd,
I think you mean "make savedefconfig"? (I think "make defconfig" loads
the default defconfig).
Cheers
James
On Monday 12 November 2012, James Hogan wrote:
> >
> > Please use "make defconfig" to generate minimal defconfig files, instead of
> > copying the entire file here.
>
> I think you mean "make savedefconfig"? (I think "make defconfig" loads
> the default defconfig).
Yes, that's right. Sorry for the typo.
Arnd
On Wed, 7 Nov 2012, Vineet Gupta wrote:
> + ******************************************************************
> + * Inline ASM macros to read/write AUX Regs
> + * Essentially invocation of lr/sr insns from "C"
> + */
> +
> +#if 1
Leftover ???
> +#define read_aux_reg(reg) __builtin_arc_lr(reg)
> +
> +/* gcc builtin sr needs reg param to be long immediate */
> +#define write_aux_reg(reg_immed, val) \
> + __builtin_arc_sr((unsigned int)val, reg_immed)
> +
> +#else
> +/*
> + * Conditionally Enable IRQs
Unconditionally methinks
The following two functions are related to irq chips I guess. So why
would you want them here ?
> +static inline void arch_mask_irq(unsigned int irq)
> +{
> + unsigned int ienb;
> +
> + ienb = read_aux_reg(AUX_IENABLE);
> + ienb &= ~(1 << irq);
> + write_aux_reg(AUX_IENABLE, ienb);
> +}
> +
> +static inline void arch_unmask_irq(unsigned int irq)
> +{
> + unsigned int ienb;
> +
> + ienb = read_aux_reg(AUX_IENABLE);
> + ienb |= (1 << irq);
> + write_aux_reg(AUX_IENABLE, ienb);
> +}
The only user is the interrupt controller code, right?
> diff --git a/arch/arc/kernel/irq.c b/arch/arc/kernel/irq.c
> new file mode 100644
> index 0000000..16fcbe8
> --- /dev/null
> +++ b/arch/arc/kernel/irq.c
> @@ -0,0 +1,32 @@
> +/*
> + * Copyright (C) 2011-12 Synopsys, Inc. (http://www.synopsys.com)
> + *
> + * This program is free software; you can redistribute it and/or modify
> + * it under the terms of the GNU General Public License version 2 as
> + * published by the Free Software Foundation.
> + *
> + */
> +
> +#include <linux/interrupt.h>
> +#include <linux/module.h>
> +#include <asm/irqflags.h>
> +#include <asm/arcregs.h>
> +
> +void arch_local_irq_enable(void)
> +{
> +
> + unsigned long flags;
> + flags = arch_local_save_flags();
> + flags |= (STATUS_E1_MASK | STATUS_E2_MASK);
> +
> + /*
> + * If called from hard ISR (between irq_enter and irq_exit)
> + * don't allow Level 1. In Soft ISR we allow further Level 1s
> + */
> +
> + if (in_irq())
> + flags &= ~(STATUS_E1_MASK | STATUS_E2_MASK);
Hmm. This looks weird and the comment is not very helpful. So using my
crystal ball you want to enforce, that nothing enables interrupts
while a hard interrupt handler is running, right?
Is there a chip limitation which you have to enforce here? If yes,
then please explain it.
Btw, all hard interrupt handlers in Linux run with interrupts disabled and
they are not supposed to reenable interrupts, which is true for almost
all drivers except for a few archaic IDE drivers. In fact you might
even WARN about it at least once, so the offending code gets fixed.
Also the code flow is backwards. What about:
unsigned long flags;
if (in_irq())
return;
flags = ....
> + arch_local_irq_restore(flags);
> +}
Thanks,
tglx
On Wed, 7 Nov 2012, Vineet Gupta wrote:
> +void __init init_IRQ(void)
> +{
> + const int irq = TIMER0_IRQ;
> +
> + /*
> + * Each CPU needs to register irq of it's private TIMER0.
> + * The APIs request_percpu_irq()/enable_percpu_irq() will not be
> + * functional, if we don't "prep" the generic IRQ sub-system with
> + * the following:
> + * -Ensure that devid passed to request_percpu_irq() is indeed per cpu
> + * -disable NOAUTOEN, w/o which the device handler never gets called
What sets NOAUTOEN in the first place? The core code definitely does
not.
> + */
> + irq_set_percpu_devid(irq);
> + irq_modify_status(irq, IRQ_NOAUTOEN, 0);
Aside of that we have irq_clear_status_flags() for this.
> + plat_init_IRQ();
> +}
> +int __init get_hw_config_num_irq(void)
How is that function used ?
> +{
> + uint32_t val = read_aux_reg(ARC_REG_VECBASE_BCR);
> +
> + switch (val & 0x03) {
> + case 0:
> + return 16;
> + case 1:
> + return 32;
> + case 2:
> + return 8;
> + default:
> + return 0;
> + }
> +
> + return 0;
> +}
Thanks,
tglx
On Wed, 7 Nov 2012, Vineet Gupta wrote:
> +void cpu_idle(void)
> +{
> + /* Since we SLEEP in idle loop, TIF_POLLING_NRFLAG can't be set */
> +
> + /* endless idle loop with no priority at all */
> + while (1) {
> + tick_nohz_idle_enter();
> +
> + while (!need_resched())
> + arch_idle();
> +
> + tick_nohz_idle_exit();
> +
> + preempt_enable_no_resched();
> + schedule();
> + preempt_disable();
schedule_preempt_disabled() please
> + }
> diff --git a/arch/arc/kernel/time.c b/arch/arc/kernel/time.c
>
> +static void arc_periodic_timer_setup(unsigned int limit)
> +{
> + /* setup start and end markers */
> + write_aux_reg(ARC_REG_TIMER0_LIMIT, limit);
> + write_aux_reg(ARC_REG_TIMER0_CNT, 0); /* start from 0 */
> +
> + /* IE: Interrupt on count = limit,
> + * NH: Count cycles only when CPU running (NOT Halted)
> + */
> + write_aux_reg(ARC_REG_TIMER0_CTRL, TIMER_CTRL_IE | TIMER_CTRL_NH);
> +}
> +
> +/*
> + * Acknowledge the interrupt & enable/disable the interrupt
> + */
> +static void arc_periodic_timer_ack(unsigned int irq_reenable)
> +{
> + /* 1. Ack the interrupt by writing to CTRL reg.
> + * Any write will cause intr to be ack, however it has to be one of
> + * writable bits (NH: Count when not halted)
> + * 2. If required by caller, re-arm timer to Interrupt at the end of
> + * next cycle.
> + *
> + * Small optimisation:
> + * Normal code would have been
> + * if (irq_reenable) CTRL_REG = (IE | NH); else CTRL_REG = NH;
> + * However since IE is BIT0 we can fold the branch
> + */
> + write_aux_reg(ARC_REG_TIMER0_CTRL, irq_reenable | TIMER_CTRL_NH);
> +}
....
> +/********** Clock Event Device *********/
> +
> +static int arc_clkevent_set_next_event(unsigned long delta,
> + struct clock_event_device *dev)
> +{
> + arc_periodic_timer_setup(delta);
This is confusing. Is arc_periodic_timer_setup() setting up a periodic
timer or a oneshot timer? It looks you use it for both and the
differentiation happens in arc_periodic_timer_ack(). So I assume the
timer only knows about periodic mode, but you trick it into oneshot
with the ack function, right ? So it's just me being confused about
the function names, but that could do with some explanatory comments.
> + return 0;
> +}
> +
> +static void arc_clkevent_set_mode(enum clock_event_mode mode,
> + struct clock_event_device *dev)
> +{
> + pr_info("Device [%s] clockevent mode now [%d]\n", dev->name, mode);
Please remove the debug leftover.
> + switch (mode) {
> + case CLOCK_EVT_MODE_PERIODIC:
> + arc_periodic_timer_setup(CONFIG_ARC_PLAT_CLK / HZ);
> + break;
> + case CLOCK_EVT_MODE_ONESHOT:
> + break;
> + default:
> + break;
> + }
> +
> + return;
> +}
> +
> +static DEFINE_PER_CPU(struct clock_event_device, arc_clockevent_device) = {
> + .name = "ARC Timer0",
> + .features = CLOCK_EVT_FEAT_ONESHOT | CLOCK_EVT_FEAT_PERIODIC,
> + .mode = CLOCK_EVT_MODE_UNUSED,
> + .rating = 300,
> + .irq = TIMER0_IRQ, /* hardwired, no need for resources */
> + .set_next_event = arc_clkevent_set_next_event,
> + .set_mode = arc_clkevent_set_mode,
> +};
> +
> +irqreturn_t timer_irq_handler(int irq, void *dev_id)
static please
> +static int arc_finished_booting;
> +
> +/*
> + * Scheduler clock - returns current time in nanosec units.
> + * It's return value must NOT wrap around.
> + *
> + * Although the return value is nanosec units based, what's more important
> + * is whats the "source" of this value. The orig jiffies based computation
> + * was only as granular as jiffies itself (10ms on ARC).
> + * We need something that is more granular, so use the same mechanism as
> + * gettimeofday(), which uses ARC Timer T1 wrapped as a clocksource.
> + * Unfortunately the first call to sched_clock( ) is way before that subsys
> + * is initialiased, thus use the jiffies based value in the interim.
> + */
> +unsigned long long sched_clock(void)
> +{
> + if (!arc_finished_booting) {
> + return (unsigned long long)(jiffies - INITIAL_JIFFIES)
> + * (NSEC_PER_SEC / HZ);
> + } else {
> + struct timespec ts;
> + getrawmonotonic(&ts);
This can live lock. sched_clock() is used by the tracer. So assume you
are function tracing and you trace a function called from within the
timekeeping seqcount write "locked" region. You spin forever in
getrawmonotonic(). Not what you want, right ?
Thanks,
tglx
On Wed, Nov 7, 2012 at 4:21 PM, Arnd Bergmann <[email protected]> wrote:
> On Wednesday 07 November 2012, Vineet Gupta wrote:
>> + * Being uClibc based we need some of the deprecated syscalls:
>> + * -Not emulated by uClibc at all
>> + * unlink, mkdir,... (needed by Busybox, LTP etc)
>> + * times (needed by LTP pan test harness)
>> + * -Not emulated efficiently
>> + * select: emulated using pselect (but extra code to chk usec > 1sec)
>> + *
> ...
> I'm pretty sure that this has been solved before, best get in contact with
> the maintainers of the openrisc/c6x/hexagon platforms, that probably all
> use uClibc without needing these.
>
> You have to remove the legacy calls here.
So, I completely agree about not adding more deprecated system call or
ABIs (thinking about the ptrace regset issues in another patch in the
same patchset), but on the other hand I have to wonder if having a
port in the tree that doesn't have a working C library or a debugger
makes sense.
I mean, it is not quite the same thing as saying: "well, users of the
old versions of the user tools will need to maintain out of tree
patches". That makes sense - it puts the burden of maintenance on
people clinging to new versions when newer one exists, but this is not
what is happening with Arc. Right now, there are no working version of
the tools for Arc, so everyone will need to use the out of tree
patches.
I wonder what is worse - having an in tree port that no one (can) use
or adding some deprecated crap (sorry...), clearly marked for deletion
the minute a version of the relevant user tools exists that can be
used with the new mechanisms?
Just my 2c as an Arc kernel port user,
Gilad
--
Gilad Ben-Yossef
Chief Coffee Drinker
[email protected]
Israel Cell: +972-52-8260388
US Cell: +1-973-8260388
http://benyossef.com
"If you take a class in large-scale robotics, can you end up in a
situation where the homework eats your dog?"
-- Jean-Baptiste Queru
On Tuesday 13 November 2012, Gilad Ben-Yossef wrote:
> So, I completely agree about not adding more deprecated system call or
> ABIs (thinking about the ptrace regset issues in another patch in the
> same patchset), but on the other hand I have to wonder if having a
> port in the tree that doesn't have a working C library or a debugger
> makes sense.
>
> I mean, it is not quite the same thing as saying: "well, users of the
> old versions of the user tools will need to maintain out of tree
> patches". That makes sense - it puts the burden of maintenance on
> people clinging to new versions when newer one exists, but this is not
> what is happening with Arc. Right now, there are no working version of
> the tools for Arc, so everyone will need to use the out of tree
> patches.
>
> I wonder what is worse - having an in tree port that no one (can) use
> or adding some deprecated crap (sorry...), clearly marked for deletion
> the minute a version of the relevant user tools exists that can be
> used with the new mechanisms?
The point is that all existing users already need to rebuild all their
user space since the upstream version is using the generic system call
numbers. What I want to avoid is breaking everything twice, and the most
logical point to do that is when moving from an out-of-tree kernel fork
to the mainline version.
If mainline doesn't work for you yet, the most logical choice is to
stay on whatever kernel you have working right now, and only change
over to the upstream version once it works with an ABI that we want
to maintain in the long term. Obviously I can't stop from using a
mix of the two while you are waiting for (or working on) getting
gdb and uclibc supported with the new interface, but my recommendation
is not to ship that in products to end-users that would suffer
from another ABI change later on.
What I'm trying to enforce here is that the upstream version follows
the exact same rules that we apply to all other ports, which is
that we don't break existing user space that was running with an
older upstream kernel.
Arnd
On 09/11/12 09:50, James Hogan wrote:
> On 07/11/12 14:21, Arnd Bergmann wrote:
>> On Wednesday 07 November 2012, Vineet Gupta wrote:
>>> + * Being uClibc based we need some of the deprecated syscalls:
>>> + * -Not emulated by uClibc at all
>>> + * unlink, mkdir,... (needed by Busybox, LTP etc)
>>> + * times (needed by LTP pan test harness)
>>> + * -Not emulated efficiently
>>> + * select: emulated using pselect (but extra code to chk usec > 1sec)
>>> + *
>>> + * some (send/recv) correctly emulated using (recfrom/sendto) and
>>> + * some arch specific ones (fork/vfork)can easily be emulated using clone but
>>> + * thats the price of using common-denominator....
>>> + */
>>> +#define __ARCH_WANT_SYSCALL_NO_AT
>>> +#define __ARCH_WANT_SYSCALL_NO_FLAGS
>>> +#define __ARCH_WANT_SYSCALL_OFF_T
>>> +#define __ARCH_WANT_SYSCALL_DEPRECATED
>>
>> I'm pretty sure that this has been solved before, best get in contact with
>> the maintainers of the openrisc/c6x/hexagon platforms, that probably all
>> use uClibc without needing these.
>>
>> You have to remove the legacy calls here.
> FYI, we'll soon be making some uClibc patches available so that it can
> work without these deprecated syscalls and with the newer IPC interface
> (the work was needed for metag too). It's based roughly on the OpenRisc
> patches from last year, but with a fair few changes.
Hi Vineet,
The uClibc patches I mentioned have been posted, see here:
http://lists.busybox.net/pipermail/uclibc/2012-November/047110.html
Please do try them out and provide any feedback.
Cheers
James
On 13 November 2012 12:41, James Hogan <[email protected]> wrote:
> The uClibc patches I mentioned have been posted, see here:
>
> http://lists.busybox.net/pipermail/uclibc/2012-November/047110.html
>
> Please do try them out and provide any feedback.
>
Hi James,
Many thanks for picking this up...
This is the third time around for trying to get this into uClibc as
Mark Salter also pushed a patch set which got about as much feedback
(almost none) on the list as mine did. His patchset, like yours,
preferred the "old" syscalls when present.
...and that gets me around to the question: why should the "old"
syscalls be preferred? I'm specifically asking this here because I'm
curious what the kernel developers' take on this is.
Regards,
Jonas
On 13/11/12 12:01, Jonas Bonn wrote:
> On 13 November 2012 12:41, James Hogan <[email protected]> wrote:
>> The uClibc patches I mentioned have been posted, see here:
>>
>> http://lists.busybox.net/pipermail/uclibc/2012-November/047110.html
>>
>> Please do try them out and provide any feedback.
>>
>
> Hi James,
>
> Many thanks for picking this up...
>
> This is the third time around for trying to get this into uClibc as
> Mark Salter also pushed a patch set which got about as much feedback
> (almost none) on the list as mine did. His patchset, like yours,
> preferred the "old" syscalls when present.
>
> ...and that gets me around to the question: why should the "old"
> syscalls be preferred? I'm specifically asking this here because I'm
> curious what the kernel developers' take on this is.
Hi Jonas,
Hopefully with several architecture maintainers asking for this it might
get somewhere, but indeed we're aware of the feedback problem on that list.
The points that I've considered for defaulting to old syscalls:
* doesn't change existing behaviour of other architectures, so is simply
less risk of breaking them.
* could possibly make uClibc slightly smaller if it doesn't have to add
extra arguments.
against:
* if defaulted to new syscalls, could give customers a chance to rebuild
their libc so that when deprecated syscalls are removed they don't notice.
Cheers
James
On Tuesday 13 November 2012, James Hogan wrote:
> Hopefully with several architecture maintainers asking for this it might
> get somewhere, but indeed we're aware of the feedback problem on that list.
>
> The points that I've considered for defaulting to old syscalls:
> * doesn't change existing behaviour of other architectures, so is simply
> less risk of breaking them.
> * could possibly make uClibc slightly smaller if it doesn't have to add
> extra arguments.
The second argument is not very good when you consider that it means adding
the same wrapper into the kernel rather than into uClibc, where it can
be stripped out when unused (e.g. for static compilation) and doesn't
have to be present in memory since the library is in pageable user space
memory, while the kernel side implementation would always have to
present.
Arnd
On 14/11/12 12:23, Arnd Bergmann wrote:
> On Tuesday 13 November 2012, James Hogan wrote:
>> Hopefully with several architecture maintainers asking for this it might
>> get somewhere, but indeed we're aware of the feedback problem on that list.
>>
>> The points that I've considered for defaulting to old syscalls:
>> * doesn't change existing behaviour of other architectures, so is simply
>> less risk of breaking them.
>> * could possibly make uClibc slightly smaller if it doesn't have to add
>> extra arguments.
>
> The second argument is not very good when you consider that it means adding
> the same wrapper into the kernel rather than into uClibc, where it can
> be stripped out when unused (e.g. for static compilation) and doesn't
> have to be present in memory since the library is in pageable user space
> memory, while the kernel side implementation would always have to
> present.
The context is existing architectures which already have to support the
deprecated syscalls pretty much forever (in addition to the new ones),
and therefore can't strip out the kernel side implementation anyway.
I don't see any of these arguments (either way) as holding too much
weight to be honest though.
Cheers
James
On Tuesday 13 November 2012 04:07 PM, Arnd Bergmann wrote:
> On Tuesday 13 November 2012, Gilad Ben-Yossef wrote:
>> So, I completely agree about not adding more deprecated system call or
>> ABIs (thinking about the ptrace regset issues in another patch in the
>> same patchset), but on the other hand I have to wonder if having a
>> port in the tree that doesn't have a working C library or a debugger
>> makes sense.
>>
>> I mean, it is not quite the same thing as saying: "well, users of the
>> old versions of the user tools will need to maintain out of tree
>> patches". That makes sense - it puts the burden of maintenance on
>> people clinging to new versions when newer one exists, but this is not
>> what is happening with Arc. Right now, there are no working version of
>> the tools for Arc, so everyone will need to use the out of tree
>> patches.
>>
>> I wonder what is worse - having an in tree port that no one (can) use
>> or adding some deprecated crap (sorry...), clearly marked for deletion
>> the minute a version of the relevant user tools exists that can be
>> used with the new mechanisms?
> The point is that all existing users already need to rebuild all their
> user space since the upstream version is using the generic system call
> numbers. What I want to avoid is breaking everything twice, and the most
> logical point to do that is when moving from an out-of-tree kernel fork
> to the mainline version.
I completely agree.
> If mainline doesn't work for you yet, the most logical choice is to
> stay on whatever kernel you have working right now, and only change
> over to the upstream version once it works with an ABI that we want
> to maintain in the long term. Obviously I can't stop from using a
> mix of the two while you are waiting for (or working on) getting
> gdb and uclibc supported with the new interface, but my recommendation
> is not to ship that in products to end-users that would suffer
> from another ABI change later on.
>
> What I'm trying to enforce here is that the upstream version follows
> the exact same rules that we apply to all other ports, which is
> that we don't break existing user space that was running with an
> older upstream kernel.
So the primary concern here is not breaking the userspace ABI - right ?
For syscalls I agree that we will indeed need to fix the ABI - by fixing
uClibc. And if uClibc doesn't merge the fixes we can stay out of tree
for uClibc - as we currently already are.
For gdbserver, the kernel provides the complete regset ABI. However it
also provides a very limited version of old ABI - i.e. ptrace with
PEEKUSR/POKEUSR. Note that latter is just a shim layer and it reuses the
regset callbacks. This allows us to support the legacy gdbserver. If and
when gdbserver upgrades it can switch over to new interface. So all
along there will be NO ABI breakage at all. The cost is couple extra
functions in kernel which we might have to maintain for some foreseeable
future. Agree ?
-Vineet
On Thursday 15 November 2012, Vineet Gupta wrote:
> So the primary concern here is not breaking the userspace ABI - right ?
>
> For syscalls I agree that we will indeed need to fix the ABI - by fixing
> uClibc. And if uClibc doesn't merge the fixes we can stay out of tree
> for uClibc - as we currently already are.
I'm sure we can find a solution for uClibc. We already have seven
architectures using the generic syscalls, including at least one
(arm64) that is going to run on a large number of installations
in the future.
> For gdbserver, the kernel provides the complete regset ABI. However it
> also provides a very limited version of old ABI - i.e. ptrace with
> PEEKUSR/POKEUSR. Note that latter is just a shim layer and it reuses the
> regset callbacks. This allows us to support the legacy gdbserver. If and
> when gdbserver upgrades it can switch over to new interface. So all
> along there will be NO ABI breakage at all. The cost is couple extra
> functions in kernel which we might have to maintain for some foreseeable
> future. Agree ?
It's more a question of principle. The rule is that we don't break user
space, and ptrace is just another user space interface is this. If
we merge it now, we will keep it around forever, taking it out later
is not an option IMHO. My preference would be not to merge the backwards
compatibility layer for ptrace at all, but I'm willing to listen to
other people that support your view if you want to keep it. In one
way, ptrace is less critical than the other system calls, and that is
because it's already architecture specific.
Arnd
Hi Vineet,
On 07/11/12 09:47, Vineet Gupta wrote:
> +config ARC
I just came across arch/mips/Kconfig which also defines ARC (and ARC32).
It's only used within arch/mips/, however it's probably more likely that
your ARC/CONFIG_ARC will find it's way into the generic bits of the
kernel which could get hit when the other ARC is defined.
Perhaps it's worth getting the other ARC renamed just in case?
Cheers
James
On Thu, Nov 15, 2012 at 05:49:57PM +0000, James Hogan wrote:
> On 07/11/12 09:47, Vineet Gupta wrote:
> > +config ARC
>
> I just came across arch/mips/Kconfig which also defines ARC (and ARC32).
> It's only used within arch/mips/, however it's probably more likely that
> your ARC/CONFIG_ARC will find it's way into the generic bits of the
> kernel which could get hit when the other ARC is defined.
>
> Perhaps it's worth getting the other ARC renamed just in case?
The MIPS world surely isn't as attached to the CONFIG_ARC config symbol
as Synopsis so I'm going to rename CONFIG_ARC and a few other firmware
related config symbols to use a consistent prefix of CONFIG_FW_.
Ralf
On Wed, Nov 07, 2012 at 10:47:53AM +0100, Vineet Gupta wrote:
> +; When we land here, pt_regs have already been updated in-place correctly
> +; A pointer to them is also passed by kernel_execve, we just need to make sure
> +; that SP is set to point to them.
> +ARC_ENTRY ret_from_kernel_execve
> + ; Force SP to "normal" pt_regs just populated.
> + b.d ret_from_system_call
> + mov sp, r0
won't that splatter crap into regs->r0? IOW, why not branch to
ret_from_exception here?
> +ARC_EXIT ret_from_kernel_execve
Another thing: why not fold that branch to ret_from_exception into the end of
ret_from_kernel_thread() (instead of calling sys_exit()), select
GENERIC_KERNEL_EXECVE and lose __ARCH_WANT_KERNEL_EXECVE.
Actually, now that I look at your ret_from_kernel_thread... How the hell
will it cope with kernel_thread() payload trying to return? AFAICS, this
j.d [r1] will lose the return address, won't it? And while we are at it,
I would suggest passing callback and its argument via callee-saved registers -
makes for simpler life in ret_from_kernel_thread(), since switch_to() itself
will take care of loading those...
> + ; --------- check for signals/restore-sigmask ------------
> + bbit0 r9, TIF_SIGPENDING, chk_next_work
> +
> + ; save CALLEE Regs.
> + ; (i) If this signal causes coredump - full regfile needed
> + ; (ii) If signal is SIGTRAP/SIGSTOP, task is being traced thus
> + ; tracer might call PEEKUSR for a CALLEE reg
> + ;
> + ; NOTE: SP will grow up by size of CALLEE Reg-File
> + SAVE_CALLEE_SAVED_USER ; clobbers r12
> +
> + ; save location of saved Callee Regs @ thread_struct->callee
> + GET_CURR_TASK_FIELD_PTR TASK_THREAD, r10
> + st sp, [r10, THREAD_CALLEE_REG]
> +
> + bl @do_signal
> +
> + ; unwind SP for cheap discard of Callee saved Regs
> + DISCARD_CALLEE_SAVED_USER
Uh-oh... And what if tracer wanted to modify callee-saved regs?
> + b resume_user_mode_begin ; loop back to start of U mode ret
> +
> + ; --- notify_resume ---
> +chk_next_work:
> + btst r9, TIF_NOTIFY_RESUME
> + blnz @do_notify_resume
> +
> + ;--------- All things done, go back to Userland ------
> +
> + b restore_regs
No. After NOTIFY_RESUME stuff you need to recheck SIGPENDING. This should
go to resume_user_mode_begin, not restore regs. Another problem here is
IRQ handling - you hit do_signal()/do_notify_resume() with IRQs disabled,
which is broken - you need to re-enable it before going into either.
> + if (insyscall) {
> + /* No handler for syscall: restart it */
> + if (regs->r0 == -ERESTARTNOHAND ||
> + regs->r0 == -ERESTARTSYS || regs->r0 == -ERESTARTNOINTR) {
> + regs->r0 = regs->orig_r0;
> + regs->ret -= 4;
> + } else if (regs->r0 == -ERESTART_RESTARTBLOCK) {
> + regs->r8 = __NR_restart_syscall;
> + regs->ret -= 4;
> + }
What's to prevent double decrement on ->ret if two signals arrive? Note
that e.g. x86 gets away with similar code only because it uses the same
register for syscall number and return value; since none of -ERESTART...
is a valid syscall number, we either won't get into an analog of that code at
all (-ENOSYS is not restart-worthy) or will revert to a value that is
a valid syscall number, so all subsequent do_signal() calls will not hit
that code. This is subtle and unfortunately not spelled out in the
architectures where it is enough.
You need to make sure that after the first restart in_syscall() will be false.
Same ought to be done in sigreturn(), BTW...
On Friday 16 November 2012 01:00 AM, Ralf Baechle wrote:
> On Thu, Nov 15, 2012 at 05:49:57PM +0000, James Hogan wrote:
>
>> On 07/11/12 09:47, Vineet Gupta wrote:
>>> +config ARC
>> I just came across arch/mips/Kconfig which also defines ARC (and ARC32).
>> It's only used within arch/mips/, however it's probably more likely that
>> your ARC/CONFIG_ARC will find it's way into the generic bits of the
>> kernel which could get hit when the other ARC is defined.
>>
>> Perhaps it's worth getting the other ARC renamed just in case?
> The MIPS world surely isn't as attached to the CONFIG_ARC config symbol
> as Synopsis so I'm going to rename CONFIG_ARC and a few other firmware
> related config symbols to use a consistent prefix of CONFIG_FW_.
Thanks Ralf !
-Vineet
> Ralf
On Friday 16 November 2012 09:38 AM, Al Viro wrote:
> On Wed, Nov 07, 2012 at 10:47:53AM +0100, Vineet Gupta wrote:
>> +; When we land here, pt_regs have already been updated in-place correctly
>> +; A pointer to them is also passed by kernel_execve, we just need to make sure
>> +; that SP is set to point to them.
>> +ARC_ENTRY ret_from_kernel_execve
>> + ; Force SP to "normal" pt_regs just populated.
>> + b.d ret_from_system_call
>> + mov sp, r0
> won't that splatter crap into regs->r0? IOW, why not branch to
> ret_from_exception here?
Yes it does - although I didn't notice the ill-effect since execve as an
API doesn't expect r0 to have a certain value (like fork for child).
Fixed in next revision.
>
>> +ARC_EXIT ret_from_kernel_execve
> Another thing: why not fold that branch to ret_from_exception into the end of
> ret_from_kernel_thread() (instead of calling sys_exit()), select
> GENERIC_KERNEL_EXECVE and lose __ARCH_WANT_KERNEL_EXECVE.
If you noticed I only had 3 patches and didn't have the equivalent of
"switch to saner kernel_execve() semantics" which would achieve above
because the bug you elucidate to below was giving me grief in bring up
(and I couldn't spend enough time debugging it).
> Actually, now that I look at your ret_from_kernel_thread... How the hell
> will it cope with kernel_thread() payload trying to return? AFAICS, this
> j.d [r1] will lose the return address, won't it?
Exactly so. I was missing the equivalent of following
ARC_ENTRY ret_from_kernel_thread
bl @schedule_tail
ld r1, [sp, PT_r1]
- j.d [r1]
+ jl.d [r1]
ld r0, [sp, PT_r0]
j @sys_exit
ARC_EXIT ret_from_kernel_thread
And since none of the bootup kernel threads took the return path I had a
working system - the problem only showed up when kernel_execve( )
started returning to exercise the broken path above.
> And while we are at it,
> I would suggest passing callback and its argument via callee-saved registers -
> makes for simpler life in ret_from_kernel_thread(), since switch_to() itself
> will take care of loading those...
Absolutely - done that too !
Would you prefer to take a look at the updated patches now or will you
rather wait for v2 series which might take a week or two.
Thanks for taking the time to review this stuff.
-Vineet
Hi!
> https://raw.github.com/vineetgarc/publish/master/ELCE-2012-ARC-Linux.pdf
>
> All comments/criticisms are welcome !
Huh. arch/arc is very similar to arch/arm. So similar that it took me
10 minutes to figure out. I guess theres no alternate name arc could
use?
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
On Tuesday 20 November 2012 07:17 PM, Pavel Machek wrote:
> Hi!
>
>> https://raw.github.com/vineetgarc/publish/master/ELCE-2012-ARC-Linux.pdf
>>
>> All comments/criticisms are welcome !
> Huh. arch/arc is very similar to arch/arm. So similar that it took me
> 10 minutes to figure out. I guess theres no alternate name arc could
> use?
Thanks for the review comment - I'll pass this message to our marketing
folks !
>
> Pavel
On Tue 2012-11-20 14:47:37, Pavel Machek wrote:
> Hi!
>
> > https://raw.github.com/vineetgarc/publish/master/ELCE-2012-ARC-Linux.pdf
> >
> > All comments/criticisms are welcome !
>
> Huh. arch/arc is very similar to arch/arm. So similar that it took me
> 10 minutes to figure out. I guess theres no alternate name arc could
> use?
Hmm... "Advanced Risc"... :-(.
Ok, plus there's already computing standard called "arc".
https://www.linux-mips.org/wiki/ARC
What about calling the architecture "sarc" or "syarc" ?
Pavel
--
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
On Tuesday 20 November 2012 07:29 PM, Pavel Machek wrote:
> On Tue 2012-11-20 14:47:37, Pavel Machek wrote:
>> Hi!
>>
>>> https://raw.github.com/vineetgarc/publish/master/ELCE-2012-ARC-Linux.pdf
>>>
>>> All comments/criticisms are welcome !
>> Huh. arch/arc is very similar to arch/arm. So similar that it took me
>> 10 minutes to figure out. I guess theres no alternate name arc could
>> use?
> Hmm... "Advanced Risc"... :-(.
>
> Ok, plus there's already computing standard called "arc".
>
> https://www.linux-mips.org/wiki/ARC
>
> What about calling the architecture "sarc" or "syarc" ?
I think it would be best use of everybody's time if you could focus on
the technical content of the port rather than the name. It has been like
that for +10 years and I don't intend to make the kernel port name
different from THE arch name itself.
>
> Pavel
Hi Arnd,
On Wednesday 07 November 2012 07:43 PM, Arnd Bergmann wrote:
> On Wednesday 07 November 2012, Vineet Gupta wrote:
>
>> +
>> +config ARC
>> + def_bool y
>> + select ARCH_WANT_IPC_PARSE_VERSION
>> + # ARC Busybox based initramfs absolutely relies on DEVTMPFS for /dev
>> + # DEVTMPS in turn needs HOTPLUG
>> + select DEVTMPFS if !INITRAMFS_SOURCE=""
>> + select GENERIC_ATOMIC64
>> + select GENERIC_CLOCKEVENTS
>> + select GENERIC_FIND_FIRST_BIT
>> + # for now, we don't need GENERIC_IRQ_PROBE, CONFIG_GENERIC_IRQ_CHIP
>> + select GENERIC_IRQ_SHOW
>> + select GENERIC_PENDING_IRQ if SMP
>> + select GENERIC_SMP_IDLE_THREAD
>> + select HAVE_GENERIC_HARDIRQS
>> + select HOTPLUG if !INITRAMFS_SOURCE=""
>> + select MODULES_USE_ELF_RELA
> You should not need ARCH_WANT_IPC_PARSE_VERSION
It seems, we have a small bug in sysV IPC.
A lot of LTP tests have started failing on ARC 3.9-rc1 kernel due to absence of
ARCH_WANT_IPC_PARSE_VERSION.
I'm assuming that for newer arches, msgctl(2) and friends need to have need to
have @cmd of type 0x01nn where 0x0100 is IPC_64.
However for ! ARCH_WANT_IPC_PARSE_VERSION, ipc/util.h seems to be defining a dummy
version of ipc_parse_version(), which doesn't strip out the 0x0100 from @cmds,
causing the code to not enter the switch block in SYSCALL_DEFINE3(msgctl) making
it fail.
Adding the above item to arch/arc/Kconfig obviously fixes the breakage.
It seems there is some bogosity in the way ipc_parse_version() is defined. Given
the way it is used, to strip out the 0x0100 bit, it seems we always need the
function variant - the IA-64 ish comment in there seems to be bit-rot.
What do you think ?
-Vineet
Hi Vineet,
On 11/03/13 12:29, Vineet Gupta wrote:
> On Wednesday 07 November 2012 07:43 PM, Arnd Bergmann wrote:
>> On Wednesday 07 November 2012, Vineet Gupta wrote:
>>
>>> +
>>> +config ARC
>>> + def_bool y
>>> + select ARCH_WANT_IPC_PARSE_VERSION
>>> + # ARC Busybox based initramfs absolutely relies on DEVTMPFS for /dev
>>> + # DEVTMPS in turn needs HOTPLUG
>>> + select DEVTMPFS if !INITRAMFS_SOURCE=""
>>> + select GENERIC_ATOMIC64
>>> + select GENERIC_CLOCKEVENTS
>>> + select GENERIC_FIND_FIRST_BIT
>>> + # for now, we don't need GENERIC_IRQ_PROBE, CONFIG_GENERIC_IRQ_CHIP
>>> + select GENERIC_IRQ_SHOW
>>> + select GENERIC_PENDING_IRQ if SMP
>>> + select GENERIC_SMP_IDLE_THREAD
>>> + select HAVE_GENERIC_HARDIRQS
>>> + select HOTPLUG if !INITRAMFS_SOURCE=""
>>> + select MODULES_USE_ELF_RELA
>> You should not need ARCH_WANT_IPC_PARSE_VERSION
>
> It seems, we have a small bug in sysV IPC.
> A lot of LTP tests have started failing on ARC 3.9-rc1 kernel due to absence of
> ARCH_WANT_IPC_PARSE_VERSION.
>
> I'm assuming that for newer arches, msgctl(2) and friends need to have need to
> have @cmd of type 0x01nn where 0x0100 is IPC_64.
> However for ! ARCH_WANT_IPC_PARSE_VERSION, ipc/util.h seems to be defining a dummy
> version of ipc_parse_version(), which doesn't strip out the 0x0100 from @cmds,
> causing the code to not enter the switch block in SYSCALL_DEFINE3(msgctl) making
> it fail.
> Adding the above item to arch/arc/Kconfig obviously fixes the breakage.
>
> It seems there is some bogosity in the way ipc_parse_version() is defined. Given
> the way it is used, to strip out the 0x0100 bit, it seems we always need the
> function variant - the IA-64 ish comment in there seems to be bit-rot.
>
> What do you think ?
FYI, for metag we assumed this was intentional and that IPC_64 shouldn't
be passed since it is the default behaviour for new arches.
You may find the following uClibc commit relevant:
http://git.uclibc.org/uClibc/commit/?id=6c2fa3451f35f8d78a31689044b2d19efa4eaca2
Cheers
James
Hi James,
On Monday 11 March 2013 06:14 PM, James Hogan wrote:
> Hi Vineet,
>
> On 11/03/13 12:29, Vineet Gupta wrote:
>> On Wednesday 07 November 2012 07:43 PM, Arnd Bergmann wrote:
>>> On Wednesday 07 November 2012, Vineet Gupta wrote:
>>>
>>>> +
>>>> +config ARC
>>>> + def_bool y
>>>> + select ARCH_WANT_IPC_PARSE_VERSION
>>>> + # ARC Busybox based initramfs absolutely relies on DEVTMPFS for /dev
>>>> + # DEVTMPS in turn needs HOTPLUG
>>>> + select DEVTMPFS if !INITRAMFS_SOURCE=""
>>>> + select GENERIC_ATOMIC64
>>>> + select GENERIC_CLOCKEVENTS
>>>> + select GENERIC_FIND_FIRST_BIT
>>>> + # for now, we don't need GENERIC_IRQ_PROBE, CONFIG_GENERIC_IRQ_CHIP
>>>> + select GENERIC_IRQ_SHOW
>>>> + select GENERIC_PENDING_IRQ if SMP
>>>> + select GENERIC_SMP_IDLE_THREAD
>>>> + select HAVE_GENERIC_HARDIRQS
>>>> + select HOTPLUG if !INITRAMFS_SOURCE=""
>>>> + select MODULES_USE_ELF_RELA
>>> You should not need ARCH_WANT_IPC_PARSE_VERSION
>> It seems, we have a small bug in sysV IPC.
>> A lot of LTP tests have started failing on ARC 3.9-rc1 kernel due to absence of
>> ARCH_WANT_IPC_PARSE_VERSION.
>>
>> I'm assuming that for newer arches, msgctl(2) and friends need to have need to
>> have @cmd of type 0x01nn where 0x0100 is IPC_64.
>> However for ! ARCH_WANT_IPC_PARSE_VERSION, ipc/util.h seems to be defining a dummy
>> version of ipc_parse_version(), which doesn't strip out the 0x0100 from @cmds,
>> causing the code to not enter the switch block in SYSCALL_DEFINE3(msgctl) making
>> it fail.
>> Adding the above item to arch/arc/Kconfig obviously fixes the breakage.
>>
>> It seems there is some bogosity in the way ipc_parse_version() is defined. Given
>> the way it is used, to strip out the 0x0100 bit, it seems we always need the
>> function variant - the IA-64 ish comment in there seems to be bit-rot.
>>
>> What do you think ?
> FYI, for metag we assumed this was intentional and that IPC_64 shouldn't
> be passed since it is the default behaviour for new arches.
> You may find the following uClibc commit relevant:
> http://git.uclibc.org/uClibc/commit/?id=6c2fa3451f35f8d78a31689044b2d19efa4eaca2
I've been using uClibc trunk - ported to ARC - for my testing, so I already have
that change. Looking at Markos' uClibc branch he is indeed defining __IPC_64 to 0
in metag port.
However, looking at the kernel code, it seems newer style is to do the other way
round - i.e. pass 0x0100 in those cmds - unless the comments in there are bogus.
include/uapi/linux/ipc.h
/*
* Version flags for semctl, msgctl, and shmctl commands
* These are passed as bitflags or-ed with the actual command
*/
#define IPC_OLD 0 /* Old version (no 32-bit UID support on many
architectures) */
#define IPC_64 0x0100 /* New version (support 32-bit UIDs, bigger
message sizes, etc. */
So someone with more ABI wisdom needs to suggest what is the right approach.
-Vineet
On 11/03/13 12:56, Vineet Gupta wrote:
> Hi James,
>
> On Monday 11 March 2013 06:14 PM, James Hogan wrote:
>> Hi Vineet,
>>
>> On 11/03/13 12:29, Vineet Gupta wrote:
>>> On Wednesday 07 November 2012 07:43 PM, Arnd Bergmann wrote:
>>>> On Wednesday 07 November 2012, Vineet Gupta wrote:
>>>>
>>>>> +
>>>>> +config ARC
>>>>> + def_bool y
>>>>> + select ARCH_WANT_IPC_PARSE_VERSION
>>>>> + # ARC Busybox based initramfs absolutely relies on DEVTMPFS for /dev
>>>>> + # DEVTMPS in turn needs HOTPLUG
>>>>> + select DEVTMPFS if !INITRAMFS_SOURCE=""
>>>>> + select GENERIC_ATOMIC64
>>>>> + select GENERIC_CLOCKEVENTS
>>>>> + select GENERIC_FIND_FIRST_BIT
>>>>> + # for now, we don't need GENERIC_IRQ_PROBE, CONFIG_GENERIC_IRQ_CHIP
>>>>> + select GENERIC_IRQ_SHOW
>>>>> + select GENERIC_PENDING_IRQ if SMP
>>>>> + select GENERIC_SMP_IDLE_THREAD
>>>>> + select HAVE_GENERIC_HARDIRQS
>>>>> + select HOTPLUG if !INITRAMFS_SOURCE=""
>>>>> + select MODULES_USE_ELF_RELA
>>>> You should not need ARCH_WANT_IPC_PARSE_VERSION
>>> It seems, we have a small bug in sysV IPC.
>>> A lot of LTP tests have started failing on ARC 3.9-rc1 kernel due to absence of
>>> ARCH_WANT_IPC_PARSE_VERSION.
>>>
>>> I'm assuming that for newer arches, msgctl(2) and friends need to have need to
>>> have @cmd of type 0x01nn where 0x0100 is IPC_64.
>>> However for ! ARCH_WANT_IPC_PARSE_VERSION, ipc/util.h seems to be defining a dummy
>>> version of ipc_parse_version(), which doesn't strip out the 0x0100 from @cmds,
>>> causing the code to not enter the switch block in SYSCALL_DEFINE3(msgctl) making
>>> it fail.
>>> Adding the above item to arch/arc/Kconfig obviously fixes the breakage.
>>>
>>> It seems there is some bogosity in the way ipc_parse_version() is defined. Given
>>> the way it is used, to strip out the 0x0100 bit, it seems we always need the
>>> function variant - the IA-64 ish comment in there seems to be bit-rot.
>>>
>>> What do you think ?
>> FYI, for metag we assumed this was intentional and that IPC_64 shouldn't
>> be passed since it is the default behaviour for new arches.
>> You may find the following uClibc commit relevant:
>> http://git.uclibc.org/uClibc/commit/?id=6c2fa3451f35f8d78a31689044b2d19efa4eaca2
>
> I've been using uClibc trunk - ported to ARC - for my testing, so I already have
> that change. Looking at Markos' uClibc branch he is indeed defining __IPC_64 to 0
> in metag port.
> However, looking at the kernel code, it seems newer style is to do the other way
> round - i.e. pass 0x0100 in those cmds - unless the comments in there are bogus.
>
> include/uapi/linux/ipc.h
>
> /*
> * Version flags for semctl, msgctl, and shmctl commands
> * These are passed as bitflags or-ed with the actual command
> */
> #define IPC_OLD 0 /* Old version (no 32-bit UID support on many
> architectures) */
> #define IPC_64 0x0100 /* New version (support 32-bit UIDs, bigger
> message sizes, etc. */
>
> So someone with more ABI wisdom needs to suggest what is the right approach.
Agreed.
Note that in uClibc, 64bit arches (excl alpha/mips for some reason) were
already defining __IPC_64 as 0, so there's some precedent for it working
the way it does.
Cheers
James
On Monday 11 March 2013, James Hogan wrote:
> > /*
> > * Version flags for semctl, msgctl, and shmctl commands
> > * These are passed as bitflags or-ed with the actual command
> > */
> > #define IPC_OLD 0 /* Old version (no 32-bit UID support on many
> > architectures) */
> > #define IPC_64 0x0100 /* New version (support 32-bit UIDs, bigger
> > message sizes, etc. */
> >
> > So someone with more ABI wisdom needs to suggest what is the right approach.
>
> Agreed.
>
> Note that in uClibc, 64bit arches (excl alpha/mips for some reason) were
> already defining __IPC_64 as 0, so there's some precedent for it working
> the way it does.
Yes, there is no reason to need the distinction. If LTP is checking for
IPC_OLD compatibility on anything but really old architectures, that is
a bug in LTP, or possibly in the libc.
Arnd
Hi Arnd,
On Monday 11 March 2013 07:00 PM, Arnd Bergmann wrote:
> On Monday 11 March 2013, James Hogan wrote:
>>> /*
>>> * Version flags for semctl, msgctl, and shmctl commands
>>> * These are passed as bitflags or-ed with the actual command
>>> */
>>> #define IPC_OLD 0 /* Old version (no 32-bit UID support on many
>>> architectures) */
>>> #define IPC_64 0x0100 /* New version (support 32-bit UIDs, bigger
>>> message sizes, etc. */
>>>
>>> So someone with more ABI wisdom needs to suggest what is the right approach.
>> Agreed.
>>
>> Note that in uClibc, 64bit arches (excl alpha/mips for some reason) were
>> already defining __IPC_64 as 0, so there's some precedent for it working
>> the way it does.
> Yes, there is no reason to need the distinction. If LTP is checking for
> IPC_OLD compatibility on anything but really old architectures, that is
> a bug in LTP, or possibly in the libc.
>
> Arnd
I'm sorrry - I don't quite understand what you mean.
The question is what should the msgctl(2) ABI be w.r.t. @cmd arg. Does it need to
be of type 0x01NN (i.e. IPC_64 ORed) or should it be be 0xNN. If it's 0x01NN,
then ARCH_WANT_IPC_PARSE_VERSION is needed in Kconfig, otherwise, I need to fix
uClibc.
The reason for confusion seems to be a comment in existing uapi/linux/ipc.h which
seems to suggest that IPC_OLD is 0 and IPC_64 is newer version.
Thx,
-Vineet
On Monday 11 March 2013, Vineet Gupta wrote:
> The question is what should the msgctl(2) ABI be w.r.t. @cmd arg. Does it need to
> be of type 0x01NN (i.e. IPC_64 ORed) or should it be be 0xNN. If it's 0x01NN,
> then ARCH_WANT_IPC_PARSE_VERSION is needed in Kconfig, otherwise, I need to fix
> uClibc.
>
> The reason for confusion seems to be a comment in existing uapi/linux/ipc.h which
> seems to suggest that IPC_OLD is 0 and IPC_64 is newer version.
Modern architecture ports should ignore this flag and always use the IPC_64 API,
there is no reason to maintain backwards compatibility with old binaries.
Arnd