The MIPS remote processor driver allows non-Linux firmware to take
control of and execute on one of the systems VPEs. The CPU must be
offlined from Linux first. A sysfs interface is created which allows
firmware to be loaded and changed at runtime. A full description is
available at [1]. An example firmware that can be used with this driver
is available at [2].
This is useful to allow running bare metal code, or an RTOS, on one or
more CPUs while allowing Linux to continue running on those remaining.
The remote processor framework allows for firmwares to provide any
virtio device for communication between the firmware running on the
remote VP and Linux. For example [1] demonstrates a simple firmware
which provides a virtual serial port. Any string sent to the port is
case inverted and returned.
This is conceptually similar to the VPE loader functionality, but is
more standard as it fits into the remoteproc subsystem.
The first patches in this series lay the groundwork for the driver
before it is added. The last series deprecates the VPE loader.
This functionality is supported on:
- MIPS32r2 devices implementing the MIPS MT ASE for multithreading, such
as interAptiv.
- MIPS32r6 devices implementing VPs, such as I6400.
Limitations:
- The remoteproc core supports only 32bit ELFs. Therefore it is only
possible to run 32bit firmware on the remote processor. Also, for
virtio communication, pointers are passed from the kernel to firmware.
There can be no mismatch in pointer sizes between the kernel and
firmware, so this limits the host kernel to 32bit as well.
The functionality has been tested on the Ci40 board which has a 2 core 2
thread interAptiv.
This series is based on v4.8
Depends on James Hogan's ebase series:
MIPS: traps: 64bit kernels should read CP0_EBase 64bit
MIPS: traps: Convert ebase to KSeg0
MIPS: traps: Ensure full EBase is written
Without these patches, if firmware modifies ebase to allow handling
exceptions / interrupts, then when the VPE is returned to Linux the
kernel exception handlers won't be reinstated properly.
[1]
http://wiki.prplfoundation.org/w/images/d/df/MIPS_OS_Remote_Processor_Driver_Whitepaper_1.0.9.pdf
[2] https://github.com/MIPS/mips-rproc-example
Changes in v3:
Update GIC context saving to use CPU hotplug state machine
Update MIPS remoteproc driver to use CPU hotplug state machine
Remove sysfs interface from MIPS rproc driver, now provided by the core.
Drop patches that Ralf has already merged to mips-next
Changes in v2:
Add dependence on additional patches to mips-gic in commit log
Incorporate changes from Marc Zynger's review:
- Remove CONTEXT_SAVING define.
- Make saved local state a per-cpu variable
- Make gic_save_* static functions when enabled, and do { } while(0)
otherwise
Lisa Parratt (1):
MIPS: CPS: Add VP(E) stealing
Matt Redfearn (3):
irqchip: mips-gic: Add context saving for MIPS_REMOTEPROC
remoteproc/MIPS: Add a remoteproc driver for MIPS
MIPS: Deprecate VPE Loader
arch/mips/Kconfig | 12 +-
arch/mips/include/asm/smp-cps.h | 8 +
arch/mips/include/asm/smp.h | 1 +
arch/mips/kernel/smp-cps.c | 162 +++++++++-
arch/mips/kernel/smp.c | 12 +
drivers/irqchip/irq-mips-gic.c | 208 ++++++++++++-
drivers/remoteproc/Kconfig | 11 +
drivers/remoteproc/Makefile | 1 +
drivers/remoteproc/mips_remoteproc.c | 566 +++++++++++++++++++++++++++++++++++
9 files changed, 966 insertions(+), 15 deletions(-)
create mode 100644 drivers/remoteproc/mips_remoteproc.c
--
2.7.4
From: Lisa Parratt <[email protected]>
VP(E) stealing provides a mechanism for removing an offline Virtual
Processor from the Linux kernel such that it is available to run bare
metal code.
Once the CPU has been offlined from Linux, the CPU can be given a task
to run via mips_cps_steal_cpu_and_execute(). The CPU is removed from the
cpu_present mask and is set up to execute from address entry_fn. Stack
space is assigned via the tsk task_struct so that C initialisation code
may be used.
To return the CPU back to Linux control, mips_cps_halt_and_return_cpu
will arrange to halt the CPU and return it to the cpu_present mask. It
is then available to be brought online again via CPU hotplug.
This mechanism is used by the MIPS remote processor driver to allow
CPUs within the system to execute bare metal code, not under control of
the kernel.
Signed-off-by: Lisa Parratt <[email protected]>
Signed-off-by: Matt Redfearn <[email protected]>
---
Changes in v3: None
Changes in v2: None
arch/mips/Kconfig | 7 ++
arch/mips/include/asm/smp-cps.h | 8 ++
arch/mips/include/asm/smp.h | 1 +
arch/mips/kernel/smp-cps.c | 162 ++++++++++++++++++++++++++++++++++++++--
arch/mips/kernel/smp.c | 12 +++
5 files changed, 183 insertions(+), 7 deletions(-)
diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index 212ff92920d2..fd2468e71e63 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -2342,6 +2342,13 @@ config MIPS_CPS
no external assistance. It is safe to enable this when hardware
support is unavailable.
+config MIPS_STEAL
+ bool "VPE stealing"
+ depends on HOTPLUG_CPU && MIPS_CPS
+ help
+ Select this is you wish to be able to run bare metal code on offline
+ VPEs.
+
config MIPS_CPS_PM
depends on MIPS_CPS
select MIPS_CPC
diff --git a/arch/mips/include/asm/smp-cps.h b/arch/mips/include/asm/smp-cps.h
index 2ae1f61a4a95..4f6cd5b14185 100644
--- a/arch/mips/include/asm/smp-cps.h
+++ b/arch/mips/include/asm/smp-cps.h
@@ -34,6 +34,14 @@ extern void mips_cps_boot_vpes(struct core_boot_config *cfg, unsigned vpe);
extern void mips_cps_pm_save(void);
extern void mips_cps_pm_restore(void);
+#ifdef CONFIG_MIPS_STEAL
+
+extern int mips_cps_steal_cpu_and_execute(unsigned int cpu, void *entry_fn,
+ struct task_struct *tsk);
+extern int mips_cps_halt_and_return_cpu(unsigned int cpu);
+
+#endif /* CONFIG_MIPS_STEAL */
+
#ifdef CONFIG_MIPS_CPS
extern bool mips_cps_smp_in_use(void);
diff --git a/arch/mips/include/asm/smp.h b/arch/mips/include/asm/smp.h
index 8bc6c70a4030..0ec332d6ce4e 100644
--- a/arch/mips/include/asm/smp.h
+++ b/arch/mips/include/asm/smp.h
@@ -103,4 +103,5 @@ static inline void arch_send_call_function_ipi_mask(const struct cpumask *mask)
extern void (*dump_ipi_function_ptr)(void *);
void dump_send_ipi(void (*dump_ipi_callback)(void *));
#endif
+
#endif /* __ASM_SMP_H */
diff --git a/arch/mips/kernel/smp-cps.c b/arch/mips/kernel/smp-cps.c
index 6183ad84cc73..9b7a7a5412c9 100644
--- a/arch/mips/kernel/smp-cps.c
+++ b/arch/mips/kernel/smp-cps.c
@@ -8,6 +8,7 @@
* option) any later version.
*/
+#include <linux/cpu.h>
#include <linux/delay.h>
#include <linux/io.h>
#include <linux/irqchip/mips-gic.h>
@@ -39,6 +40,31 @@ static int __init setup_nothreads(char *s)
}
early_param("nothreads", setup_nothreads);
+#ifdef CONFIG_MIPS_STEAL
+struct cpumask cpu_stolen_mask;
+
+static inline bool cpu_stolen(int cpu)
+{
+ return cpumask_test_cpu(cpu, &cpu_stolen_mask);
+}
+
+static inline void set_cpu_stolen(int cpu, bool state)
+{
+ if (state)
+ cpumask_set_cpu(cpu, &cpu_stolen_mask);
+ else
+ cpumask_clear_cpu(cpu, &cpu_stolen_mask);
+}
+#else
+static inline bool cpu_stolen(int cpu)
+{
+ return false;
+}
+
+static inline void set_cpu_stolen(int cpu, bool state) { }
+
+#endif /* CONFIG_MIPS_STEAL */
+
static unsigned core_vpe_count(unsigned core)
{
unsigned cfg;
@@ -109,6 +135,10 @@ static void __init cps_smp_setup(void)
write_gcr_bev_base(core_entry);
}
+#ifdef CONFIG_MIPS_STEAL
+ cpumask_clear(&cpu_stolen_mask);
+#endif /* CONFIG_MIPS_STEAL */
+
#ifdef CONFIG_MIPS_MT_FPAFF
/* If we have an FPU, enroll ourselves in the FPU-full mask */
if (cpu_has_fpu)
@@ -287,7 +317,7 @@ static void remote_vpe_boot(void *dummy)
mips_cps_boot_vpes(core_cfg, cpu_vpe_id(¤t_cpu_data));
}
-static void cps_boot_secondary(int cpu, struct task_struct *idle)
+static void cps_start_secondary(int cpu, void *entry_fn, struct task_struct *tsk)
{
unsigned core = cpu_data[cpu].core;
unsigned vpe_id = cpu_vpe_id(&cpu_data[cpu]);
@@ -297,9 +327,9 @@ static void cps_boot_secondary(int cpu, struct task_struct *idle)
unsigned int remote;
int err;
- vpe_cfg->pc = (unsigned long)&smp_bootstrap;
- vpe_cfg->sp = __KSTK_TOS(idle);
- vpe_cfg->gp = (unsigned long)task_thread_info(idle);
+ vpe_cfg->pc = (unsigned long)entry_fn;
+ vpe_cfg->sp = __KSTK_TOS(tsk);
+ vpe_cfg->gp = (unsigned long)task_thread_info(tsk);
atomic_or(1 << cpu_vpe_id(&cpu_data[cpu]), &core_cfg->vpe_mask);
@@ -343,6 +373,11 @@ out:
preempt_enable();
}
+static void cps_boot_secondary(int cpu, struct task_struct *idle)
+{
+ cps_start_secondary(cpu, &smp_bootstrap, idle);
+}
+
static void cps_init_secondary(void)
{
/* Disable MT - we only want to run 1 TC per VPE */
@@ -394,6 +429,28 @@ static int cps_cpu_disable(void)
if (!cps_pm_support_state(CPS_PM_POWER_GATED))
return -EINVAL;
+#ifdef CONFIG_MIPS_STEAL
+ /*
+ * With the MT ASE only VPEs in the same core may read / write the
+ * control registers of other VPEs. Therefore to maintain control of
+ * any stolen VPEs at least one sibling VPE must be kept online.
+ */
+ if (cpu_has_mipsmt) {
+ int stolen, siblings = 0;
+
+ for_each_cpu((stolen), &cpu_stolen_mask)
+ if (cpu_data[stolen].core == cpu_data[cpu].core)
+ siblings++;
+
+ if (siblings == 1)
+ /*
+ * When a VPE has been stolen, keep at least one of it's
+ * siblings around in order to control it.
+ */
+ return -EBUSY;
+ }
+#endif /* CONFIG_MIPS_STEAL */
+
core_cfg = &mips_cps_core_bootcfg[current_cpu_data.core];
atomic_sub(1 << cpu_vpe_id(¤t_cpu_data), &core_cfg->vpe_mask);
smp_mb__after_atomic();
@@ -426,7 +483,7 @@ void play_dead(void)
core = cpu_data[cpu].core;
/* Look for another online VPE within the core */
- for_each_online_cpu(cpu_death_sibling) {
+ for_each_possible_cpu(cpu_death_sibling) {
if (cpu_data[cpu_death_sibling].core != core)
continue;
@@ -434,8 +491,11 @@ void play_dead(void)
* There is an online VPE within the core. Just halt
* this TC and leave the core alone.
*/
- cpu_death = CPU_DEATH_HALT;
- break;
+ if (cpu_online(cpu_death_sibling) ||
+ cpu_stolen(cpu_death_sibling))
+ cpu_death = CPU_DEATH_HALT;
+ if (cpu_online(cpu_death_sibling))
+ break;
}
}
@@ -466,6 +526,94 @@ void play_dead(void)
panic("Failed to offline CPU %u", cpu);
}
+#ifdef CONFIG_MIPS_STEAL
+
+/* Find an online sibling CPU (another VPE in the same core) */
+static inline int mips_cps_get_online_sibling(unsigned int cpu)
+{
+ int sibling;
+
+ for_each_online_cpu(sibling)
+ if (cpu_data[sibling].core == cpu_data[cpu].core)
+ return sibling;
+
+ return -1;
+}
+
+int mips_cps_steal_cpu_and_execute(unsigned int cpu, void *entry_fn,
+ struct task_struct *tsk)
+{
+ int err = -EINVAL;
+
+ preempt_disable();
+
+ if (!cpu_present(cpu) || cpu_online(cpu) || cpu_stolen(cpu))
+ goto out;
+
+ if (cpu_has_mipsmt && (mips_cps_get_online_sibling(cpu) < 0))
+ pr_warn("CPU%d has no online siblings to control it\n", cpu);
+ else {
+ set_cpu_present(cpu, false);
+ set_cpu_stolen(cpu, true);
+
+ cps_start_secondary(cpu, entry_fn, tsk);
+ err = 0;
+ }
+out:
+ preempt_enable();
+ return err;
+}
+
+static void mips_cps_halt_sibling(void *ptr_cpu)
+{
+ unsigned int cpu = (unsigned long)ptr_cpu;
+ unsigned int vpe_id = cpu_vpe_id(&cpu_data[cpu]);
+ unsigned long flags;
+ int vpflags;
+
+ local_irq_save(flags);
+ vpflags = dvpe();
+ settc(vpe_id);
+ write_tc_c0_tchalt(TCHALT_H);
+ evpe(vpflags);
+ local_irq_restore(flags);
+}
+
+int mips_cps_halt_and_return_cpu(unsigned int cpu)
+{
+ unsigned int core = cpu_data[cpu].core;
+ unsigned int vpe_id = cpu_vpe_id(&cpu_data[cpu]);
+
+ if (!cpu_stolen(cpu))
+ return -EINVAL;
+
+ if (cpu_has_mipsmt && (core == cpu_data[smp_processor_id()].core))
+ mips_cps_halt_sibling((void *)(unsigned long)cpu);
+ else if (cpu_has_mipsmt) {
+ int sibling = mips_cps_get_online_sibling(cpu);
+
+ if (sibling < 0) {
+ pr_warn("CPU%d has no online siblings\n", cpu);
+ return -EINVAL;
+ }
+
+ if (smp_call_function_single(sibling, mips_cps_halt_sibling,
+ (void *)(unsigned long)cpu, 1))
+ panic("Failed to call sibling CPU\n");
+
+ } else if (cpu_has_vp) {
+ mips_cm_lock_other(core, vpe_id);
+ write_cpc_co_vp_stop(1 << vpe_id);
+ mips_cm_unlock_other();
+ }
+
+ set_cpu_stolen(cpu, false);
+ set_cpu_present(cpu, true);
+ return 0;
+}
+
+#endif /* CONFIG_MIPS_STEAL */
+
static void wait_for_sibling_halt(void *ptr_cpu)
{
unsigned cpu = (unsigned long)ptr_cpu;
diff --git a/arch/mips/kernel/smp.c b/arch/mips/kernel/smp.c
index b0baf48951fa..047322eaa0b0 100644
--- a/arch/mips/kernel/smp.c
+++ b/arch/mips/kernel/smp.c
@@ -233,6 +233,18 @@ static __init void smp_ipi_init_one(unsigned int virq,
struct irqaction *action)
{
int ret;
+#ifdef CONFIG_MIPS_STEAL
+ struct irq_data *data;
+ /*
+ * A bit of a hack to ensure that the ipi_offset is 0.
+ * This is to deal with removing / reallocating IPIs
+ * to subsets of the possible CPUs, where the IPI IRQ domain
+ * will set ipi_offset to the first cpu in the cpumask when the
+ * IPI is reallocated.
+ */
+ data = irq_get_irq_data(virq);
+ data->common->ipi_offset = 0;
+#endif /* CONFIG_MIPS_STEAL */
irq_set_handler(virq, handle_percpu_irq);
ret = setup_irq(virq, action);
--
2.7.4
The MIPS remote processor driver (CONFIG_MIPS_RPROC) provides a more
standard mechanism for using one or more VPs as coprocessors running
separate firmware.
Here we deprecate this mechanism before it is removed.
Signed-off-by: Matt Redfearn <[email protected]>
---
Changes in v3: None
Changes in v2: None
arch/mips/Kconfig | 5 ++++-
1 file changed, 4 insertions(+), 1 deletion(-)
diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
index fd2468e71e63..3e767c373c7b 100644
--- a/arch/mips/Kconfig
+++ b/arch/mips/Kconfig
@@ -2264,7 +2264,7 @@ comment "MIPS R2-to-R6 emulator is only available for UP kernels"
depends on SMP && CPU_MIPSR6
config MIPS_VPE_LOADER
- bool "VPE loader support."
+ bool "VPE loader support (DEPRECATED)"
depends on SYS_SUPPORTS_MULTITHREADING && MODULES
select CPU_MIPSR2_IRQ_VI
select CPU_MIPSR2_IRQ_EI
@@ -2273,6 +2273,9 @@ config MIPS_VPE_LOADER
Includes a loader for loading an elf relocatable object
onto another VPE and running it.
+ Unless you have a specific need, you should use CONFIG_MIPS_RPROC
+ instead of this.
+
config MIPS_VPE_LOADER_CMP
bool
default "y"
--
2.7.4
This driver allows a MIPS processor offlined from Linux to be used as a
remote processor. Firmware may be loaded via the sysfs interface and
changed at runtime, allowing the processor to handle real-time tasks or
perform coprocessing while remaining processors are available to Linux.
Coprocessor firmware must abide by the remoteproc standard, i.e.
implement the resource table containing memory layouts and virtio device
descriptions, and additionally abide by the MIPS UHI coprocessor boot
protocol in the startup code.
Signed-off-by: Lisa Parratt <[email protected]>
Signed-off-by: Matt Redfearn <[email protected]>
---
Changes in v3:
Update MIPS remoteproc driver to use CPU hotplug state machine
Remove sysfs interface from MIPS rproc driver, now provided by the core.
Drop patches that Ralf has already merged to mips-next
Changes in v2: None
drivers/remoteproc/Kconfig | 11 +
drivers/remoteproc/Makefile | 1 +
drivers/remoteproc/mips_remoteproc.c | 566 +++++++++++++++++++++++++++++++++++
3 files changed, 578 insertions(+)
create mode 100644 drivers/remoteproc/mips_remoteproc.c
diff --git a/drivers/remoteproc/Kconfig b/drivers/remoteproc/Kconfig
index 1a8bf76a925f..05db52e0e668 100644
--- a/drivers/remoteproc/Kconfig
+++ b/drivers/remoteproc/Kconfig
@@ -100,4 +100,15 @@ config ST_REMOTEPROC
processor framework.
This can be either built-in or a loadable module.
+config MIPS_REMOTEPROC
+ tristate "MIPS remoteproc support"
+ depends on MIPS_CPS && HAS_DMA
+ select CMA
+ select REMOTEPROC
+ select MIPS_STEAL
+ help
+ Say y here to support using offline cores/VPEs as remote processors
+ via the remote processor framework.
+ If unsure say N.
+
endmenu
diff --git a/drivers/remoteproc/Makefile b/drivers/remoteproc/Makefile
index 92d3758bd15c..de19cd320f3a 100644
--- a/drivers/remoteproc/Makefile
+++ b/drivers/remoteproc/Makefile
@@ -14,3 +14,4 @@ obj-$(CONFIG_DA8XX_REMOTEPROC) += da8xx_remoteproc.o
obj-$(CONFIG_QCOM_MDT_LOADER) += qcom_mdt_loader.o
obj-$(CONFIG_QCOM_Q6V5_PIL) += qcom_q6v5_pil.o
obj-$(CONFIG_ST_REMOTEPROC) += st_remoteproc.o
+obj-$(CONFIG_MIPS_REMOTEPROC) += mips_remoteproc.o
diff --git a/drivers/remoteproc/mips_remoteproc.c b/drivers/remoteproc/mips_remoteproc.c
new file mode 100644
index 000000000000..95aea67416b7
--- /dev/null
+++ b/drivers/remoteproc/mips_remoteproc.c
@@ -0,0 +1,566 @@
+/*
+ * MIPS Remote Processor driver
+ *
+ * Copyright (C) 2016 Imagination Technologies
+ * Lisa Parratt <[email protected]>
+ * Matt Redfearn <[email protected]>
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation; either version 2 of the License, or (at your
+ * option) any later version.
+ */
+
+#include <linux/cpu.h>
+#include <linux/dma-mapping.h>
+#include <linux/interrupt.h>
+#include <linux/io.h>
+#include <linux/irq.h>
+#include <linux/module.h>
+#include <linux/of_irq.h>
+#include <linux/platform_device.h>
+#include <linux/remoteproc.h>
+
+#include <asm/smp-cps.h>
+#include <asm/tlbflush.h>
+#include <asm/tlbmisc.h>
+
+#include "remoteproc_internal.h"
+
+struct mips_rproc {
+ struct rproc *rproc;
+ struct task_struct *tsk;
+ struct device dev;
+ unsigned int cpu;
+ int ipi_linux;
+ int ipi_remote;
+};
+
+static struct rproc *mips_rprocs[NR_CPUS];
+
+static int cpuhp_state;
+
+#define to_mips_rproc(d) container_of(d, struct mips_rproc, dev)
+
+
+/* Compute the largest page mask a physical address can be mapped with */
+static unsigned long mips_rproc_largest_pm(unsigned long pa,
+ unsigned long maxmask)
+{
+ unsigned long mask;
+ /* Find address bits limiting alignment */
+ unsigned long shift = ffs(pa);
+
+ /* Obey MIPS restrictions on page sizes */
+ if (pa) {
+ if (shift & 1)
+ shift -= 2;
+ else
+ shift--;
+ }
+ mask = ULONG_MAX << shift;
+ return maxmask & ~mask;
+}
+
+/* Compute the next largest page mask for a given page mask */
+static unsigned long mips_rproc_next_pm(unsigned long pm, unsigned long maxmask)
+{
+ if (pm != PM_4K)
+ return ((pm << 2) | pm) & maxmask;
+ else
+ return PM_16K;
+}
+
+static void mips_map_page(unsigned long da, unsigned long pa, int c,
+ unsigned long pagemask, unsigned long pagesize)
+{
+ unsigned long pa2 = pa + (pagesize / 2);
+ unsigned long entryhi, entrylo0, entrylo1;
+
+ /* Compute the mapping */
+ pa = (pa >> 6) & (ULONG_MAX << MIPS_ENTRYLO_PFN_SHIFT);
+ pa2 = (pa2 >> 6) & (ULONG_MAX << MIPS_ENTRYLO_PFN_SHIFT);
+ entryhi = da & 0xfffffe000;
+ entrylo0 = (c << ENTRYLO_C_SHIFT) | ENTRYLO_D | ENTRYLO_V | pa;
+ entrylo1 = (c << ENTRYLO_C_SHIFT) | ENTRYLO_D | ENTRYLO_V | pa2;
+
+ pr_debug("Create wired entry %d, CCA %d\n", read_c0_wired(), c);
+ pr_debug(" EntryHi: 0x%016lx\n", entryhi);
+ pr_debug(" EntryLo0: 0x%016lx\n", entrylo0);
+ pr_debug(" EntryLo1: 0x%016lx\n", entrylo1);
+ pr_debug(" Pagemask: 0x%016lx\n", pagemask);
+ pr_debug("\n");
+
+ add_wired_entry(entrylo0, entrylo1, entryhi, pagemask);
+}
+
+/*
+ * Compute the page required to fulfill a mapping. Escapes alignment derived
+ * page size limitations before using biggest fit to map the remainder.
+ */
+static inline void mips_rproc_fit_page(unsigned long da, unsigned long pa,
+ int c, unsigned long size,
+ unsigned long maxmask)
+{
+ unsigned long bigmask, nextmask;
+ unsigned long pagemask, pagesize;
+ unsigned long distance, target;
+
+ do {
+ /* Compute the current largest page mask */
+ bigmask = mips_rproc_largest_pm(pa, maxmask);
+ /* Compute the next largest pagesize */
+ nextmask = mips_rproc_next_pm(bigmask, maxmask);
+ /*
+ * Compute the distance from our current physical address to
+ * the next page boundary.
+ */
+ distance = (nextmask + 0x2000) - (pa & nextmask);
+ /*
+ * Decide between searching to get to the next highest page
+ * boundary or finishing.
+ */
+ target = distance < size ? distance : size;
+ /* Fit */
+ while (target) {
+ /* Find the largest supported page size that will fit */
+ for (pagesize = maxmask + 0x2000;
+ (pagesize > 0x2000) && (pagesize > target);
+ pagesize /= 4) {
+ }
+ /* Convert it to a page mask */
+ pagemask = pagesize - 0x2000;
+ /* Emit it */
+ mips_map_page(da, pa, c, pagemask, pagesize);
+ /* Move to next step */
+ size -= pagesize;
+ da += pagesize;
+ pa += pagesize;
+ target -= pagesize;
+ }
+ } while (size);
+}
+
+static int mips_rproc_carveouts(struct rproc *rproc, int max_pagemask)
+{
+ struct rproc_mem_entry *carveout;
+
+ list_for_each_entry(carveout, &rproc->carveouts, node) {
+ int c = CONF_CM_CACHABLE_COW;
+
+ dev_dbg(&rproc->dev,
+ "carveout mapping da 0x%x -> %pad length 0x%x, CCA %d",
+ carveout->da, &carveout->dma, carveout->len, c);
+
+ mips_rproc_fit_page(carveout->da, carveout->dma, c,
+ carveout->len, max_pagemask);
+ }
+ return 0;
+}
+
+static int mips_rproc_vdevs(struct rproc *rproc, int max_pagemask)
+{
+ struct rproc_vdev *rvdev;
+
+ list_for_each_entry(rvdev, &rproc->rvdevs, node) {
+ int i, size;
+
+ for (i = 0; i < ARRAY_SIZE(rvdev->vring); i++) {
+ struct rproc_vring *vring = &rvdev->vring[i];
+ unsigned long pa = vring->dma;
+ int c;
+
+ if (hw_coherentio) {
+ /*
+ * The DMA API will allocate cacheable buffers
+ * for shared resources, so the firmware should
+ * also access those buffers cached
+ */
+ c = (_page_cachable_default >> _CACHE_SHIFT);
+ } else {
+ /*
+ * Otherwise, shared buffers should be accessed
+ * uncached
+ */
+ c = CONF_CM_UNCACHED;
+ }
+
+ /* actual size of vring (in bytes) */
+ size = PAGE_ALIGN(vring_size(vring->len, vring->align));
+
+ dev_dbg(&rproc->dev,
+ "vring mapping da %pad -> %pad length 0x%x, CCA %d",
+ &vring->dma, &vring->dma, size, c);
+
+ mips_rproc_fit_page(pa, pa, c, size, max_pagemask);
+ }
+ }
+ return 0;
+}
+
+static void mips_rproc_cpu_entry(void)
+{
+ struct rproc *rproc = mips_rprocs[smp_processor_id()];
+ struct mips_rproc *mproc = *(struct mips_rproc **)rproc->priv;
+ int ipi_to_remote = ipi_get_hwirq(mproc->ipi_remote, mproc->cpu);
+ int ipi_from_remote = ipi_get_hwirq(mproc->ipi_linux, 0);
+ unsigned long old_pagemask, max_pagemask;
+
+ if (!rproc)
+ return;
+
+ dev_info(&rproc->dev, "Starting %s on MIPS CPU%d\n",
+ rproc->firmware, mproc->cpu);
+
+ /* Get the maximum pagemask supported on this CPU */
+ old_pagemask = read_c0_pagemask();
+ write_c0_pagemask(PM_HUGE_MASK);
+ mtc0_tlbw_hazard();
+ max_pagemask = read_c0_pagemask();
+ write_c0_pagemask(old_pagemask);
+ mtc0_tlbw_hazard();
+
+ /* Start with no wired entries */
+ write_c0_wired(0);
+
+ /* Flush all previous TLB entries */
+ local_flush_tlb_all();
+
+ /* Map firmware resources into virtual memory */
+ mips_rproc_carveouts(rproc, max_pagemask);
+ mips_rproc_vdevs(rproc, max_pagemask);
+
+ dev_dbg(&rproc->dev, "IPI to remote: %d\n", ipi_to_remote);
+ dev_dbg(&rproc->dev, "IPI from remote: %d\n", ipi_from_remote);
+
+ /* Hand off the CPU to the firmware */
+ dev_dbg(&rproc->dev, "Jumping to firmware at 0x%x\n", rproc->bootaddr);
+
+ write_c0_entryhi(0); /* Set ASID 0 */
+ tlbw_use_hazard();
+
+ /* Firmware protocol */
+ __asm__("addiu $a0, $zero, -3");
+ __asm__("move $a1, %0" :: "r" (ipi_to_remote));
+ __asm__("move $a2, %0" :: "r" (ipi_from_remote));
+ __asm__("move $a3, $zero");
+ __asm__("jr %0" :: "r" (rproc->bootaddr));
+}
+
+static irqreturn_t mips_rproc_ipi_handler(int irq, void *dev_id)
+{
+ /* Synthetic interrupts shouldn't need acking */
+ return IRQ_WAKE_THREAD;
+}
+
+static irqreturn_t mips_rproc_vq_int(int irq, void *p)
+{
+ struct rproc *rproc = (struct rproc *)p;
+ void *entry;
+ int id;
+
+ /* We don't have a mailbox, so iterate over all vqs and kick them. */
+ idr_for_each_entry(&rproc->notifyids, entry, id)
+ rproc_vq_interrupt(rproc, id);
+
+ return IRQ_HANDLED;
+}
+
+/* Helper function to find the IPI domain */
+static struct irq_domain *ipi_domain(void)
+{
+ struct device_node *node = of_irq_find_parent(of_root);
+ struct irq_domain *ipidomain;
+
+ ipidomain = irq_find_matching_host(node, DOMAIN_BUS_IPI);
+ /*
+ * Some platforms have half DT setup. So if we found irq node but
+ * didn't find an ipidomain, try to search for one that is not in the
+ * DT.
+ */
+ if (node && !ipidomain)
+ ipidomain = irq_find_matching_host(NULL, DOMAIN_BUS_IPI);
+
+ return ipidomain;
+}
+
+int mips_rproc_op_start(struct rproc *rproc)
+{
+ struct mips_rproc *mproc = *(struct mips_rproc **)rproc->priv;
+ int err;
+ int cpu = mproc->cpu;
+
+ pr_info("%s\n", __func__);
+
+ /* Create task for the CPU to use before handing off to firmware */
+ mproc->tsk = fork_idle(cpu);
+ if (IS_ERR(mproc->tsk)) {
+ dev_err(&rproc->dev, "fork_idle() failed for CPU%d\n", cpu);
+ return -ENOMEM;
+ }
+
+ /* We won't be needing the Linux IPIs anymore */
+ if (mips_smp_ipi_free(get_cpu_mask(cpu))) {
+ dev_err(&mproc->dev, "Failed to reserve incoming kick\n");
+ goto exit_free_tsk;
+ }
+
+ /*
+ * Direct IPIs from the remote processor to CPU0 since that can't be
+ * offlined while the remote CPU is running.
+ */
+ mproc->ipi_linux = irq_reserve_ipi(ipi_domain(), get_cpu_mask(0));
+ if (!mproc->ipi_linux) {
+ dev_err(&mproc->dev, "Failed to reserve incoming kick\n");
+ goto exit_restore_ipi;
+ }
+
+ mproc->ipi_remote = irq_reserve_ipi(ipi_domain(), get_cpu_mask(cpu));
+ if (!mproc->ipi_remote) {
+ dev_err(&mproc->dev, "Failed to reserve outgoing kick\n");
+ goto exit_destroy_ipi_linux;
+ }
+
+ /* register incoming ipi */
+ err = request_threaded_irq(mproc->ipi_linux, mips_rproc_ipi_handler,
+ mips_rproc_vq_int, 0,
+ "mips-rproc IPI in", rproc);
+ if (err) {
+ dev_err(&mproc->dev, "Failed to register incoming kick: %d\n",
+ err);
+ goto exit_destroy_ipi_remote;
+ }
+
+ if (mips_cps_steal_cpu_and_execute(cpu, &mips_rproc_cpu_entry,
+ mproc->tsk)) {
+ dev_err(&mproc->dev, "Failed to steal CPU%d for remote\n", cpu);
+ goto exit_free_irq;
+ }
+
+ dev_info(&mproc->dev, "CPU%d started\n", cpu);
+ return 0;
+
+exit_free_irq:
+ free_irq(mproc->ipi_linux, rproc);
+exit_destroy_ipi_remote:
+ irq_destroy_ipi(mproc->ipi_remote, get_cpu_mask(cpu));
+exit_destroy_ipi_linux:
+ irq_destroy_ipi(mproc->ipi_linux, get_cpu_mask(0));
+exit_restore_ipi:
+ /* Set up the Linux IPIs again */
+ mips_smp_ipi_allocate(get_cpu_mask(cpu));
+exit_free_tsk:
+ free_task(mproc->tsk);
+
+ return -EINVAL;
+}
+
+int mips_rproc_op_stop(struct rproc *rproc)
+{
+ struct mips_rproc *mproc = *(struct mips_rproc **)rproc->priv;
+
+ free_irq(mproc->ipi_linux, rproc);
+
+ irq_destroy_ipi(mproc->ipi_linux, get_cpu_mask(0));
+ irq_destroy_ipi(mproc->ipi_remote, get_cpu_mask(mproc->cpu));
+
+ /* Set up the Linux IPIs again */
+ mips_smp_ipi_allocate(get_cpu_mask(mproc->cpu));
+
+ free_task(mproc->tsk);
+
+ return mips_cps_halt_and_return_cpu(mproc->cpu);
+}
+
+void mips_rproc_op_kick(struct rproc *rproc, int vqid)
+{
+ struct mips_rproc *mproc = *(struct mips_rproc **)rproc->priv;
+
+ if (rproc->state == RPROC_RUNNING)
+ ipi_send_single(mproc->ipi_remote, mproc->cpu);
+}
+
+struct rproc_ops mips_rproc_proc_ops = {
+ .start = mips_rproc_op_start,
+ .stop = mips_rproc_op_stop,
+ .kick = mips_rproc_op_kick,
+};
+
+
+static int mips_rproc_probe(struct platform_device *pdev)
+{
+ return 0;
+}
+
+static int mips_rproc_remove(struct platform_device *pdev)
+{
+ return 0;
+}
+
+static struct platform_driver mips_rproc_driver = {
+ .probe = mips_rproc_probe,
+ .remove = mips_rproc_remove,
+ .driver = {
+ .name = "mips-rproc"
+ },
+};
+
+static void mips_rproc_release(struct device *dev)
+{
+}
+
+static int mips_rproc_uevent(struct device *dev, struct kobj_uevent_env *env)
+{
+ struct mips_rproc *mproc = to_mips_rproc(dev);
+
+ if (!mproc)
+ return -ENODEV;
+
+ return 0;
+}
+
+static struct device_type mips_rproc_type = {
+ .release = mips_rproc_release,
+ .uevent = mips_rproc_uevent
+};
+
+/* Create an rproc instance in response to CPU down */
+static int mips_rproc_device_register(unsigned int cpu)
+{
+ char name[64], *template = "mips-cpu%u";
+ struct rproc *rproc;
+ struct mips_rproc *mproc;
+ int err;
+
+ pr_debug("Allocating MIPS rproc for cpu%d\n", cpu);
+
+ if (mips_rprocs[cpu]) {
+ dev_err(&rproc->dev, "CPU%d in use\n", cpu);
+ return -EBUSY;
+ }
+
+ mproc = kzalloc(sizeof(struct mips_rproc), GFP_KERNEL);
+ if (!mproc) {
+ err = -ENOMEM;
+ goto exit;
+ }
+
+ mproc->dev.driver = &mips_rproc_driver.driver;
+ mproc->dev.type = &mips_rproc_type;
+ mproc->dev.id = cpu;
+ snprintf(name, sizeof(name), template, cpu);
+ dev_set_name(&mproc->dev, name);
+ mproc->cpu = cpu;
+
+ /* Set device to have coherent DMA ops */
+ arch_setup_dma_ops(&mproc->dev, 0, 0, NULL, 1);
+
+ rproc = rproc_alloc(&mproc->dev, dev_name(&mproc->dev),
+ &mips_rproc_proc_ops, NULL,
+ sizeof(struct mips_rproc *));
+ if (!rproc) {
+ pr_err("Error allocting rproc\n");
+ err = -ENOMEM;
+ goto exit_free_mproc;
+ }
+
+ *(struct mips_rproc **)rproc->priv = mproc;
+
+ err = device_register(&mproc->dev);
+ if (err) {
+ pr_err("device add failed\n");
+ goto exit_free_rproc;
+ }
+
+ err = rproc_add(rproc);
+ if (err) {
+ dev_err(&mproc->dev, "Failed to add rproc: %d\n", err);
+ goto exit_free_rproc;
+ }
+
+ mips_rprocs[cpu] = rproc;
+ return 0;
+
+exit_free_rproc:
+ rproc_put(rproc);
+exit_free_mproc:
+ kfree(mproc);
+exit:
+ return err;
+}
+
+/* Destroy rproc instance in response to CPU up */
+static int mips_rproc_device_unregister(unsigned int cpu)
+{
+ struct rproc *rproc = mips_rprocs[cpu];
+ struct mips_rproc *mproc;
+
+ if (!rproc)
+ /* No rproc class has been created for this CPU */
+ return 0;
+
+ pr_debug("Deallocating MIPS rproc for cpu%d\n", cpu);
+
+ rproc_del(rproc);
+ rproc_put(rproc);
+
+ mproc = *(struct mips_rproc **)rproc->priv;
+ device_unregister(&mproc->dev);
+ kfree(mproc);
+
+ mips_rprocs[cpu] = NULL;
+ return 0;
+}
+
+static int __init mips_rproc_init(void)
+{
+ int cpu;
+
+ /*
+ * Register with the cpu hotplug state machine.
+ * This driver requires opposite sense to "normal" drivers, since the
+ * driver is activated for offline CPUs via the teardown callback and
+ * deactivated via the online callback.
+ */
+ cpuhp_state = cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "MIPS:REMOTEPROC",
+ mips_rproc_device_unregister,
+ mips_rproc_device_register);
+
+ /* Register a device for each offline CPU */
+ get_online_cpus();
+ for_each_possible_cpu(cpu)
+ if (!cpu_online(cpu))
+ mips_rproc_device_register(cpu);
+ put_online_cpus();
+
+ return 0;
+}
+
+static void __exit mips_rproc_exit(void)
+{
+ int cpu;
+
+ if (cpuhp_state) {
+ /*
+ * Unregister with the cpu hotplug state machine, but don't call
+ * the teardown callback, since that would try to start the
+ * remote processor device.
+ */
+ __cpuhp_remove_state(cpuhp_state, false);
+ cpuhp_state = 0;
+ }
+
+ get_online_cpus();
+ /* Unregister devices created for any offline CPUs */
+ for_each_possible_cpu(cpu)
+ mips_rproc_device_unregister(cpu);
+ put_online_cpus();
+}
+
+subsys_initcall(mips_rproc_init);
+module_exit(mips_rproc_exit);
+
+module_platform_driver(mips_rproc_driver);
+
+MODULE_LICENSE("GPL v2");
+MODULE_DESCRIPTION("MIPS Remote Processor control driver");
--
2.7.4
The MIPS remote processor driver allows non-Linux firmware to take
control of and execute on one of the systems VPEs. If that VPE is
brought back under Linux, it is necessary to ensure that all GIC
interrupts are routed and masked as Linux expects them, as the firmware
can have done anything it likes with the GIC configuration (hopefully
just for that VPEs local interrupt sources, but allow for shared
external interrupts as well).
The configuration of shared and local CPU interrupts is maintained and
updated every time a change is made. When a CPU is brought online, the
saved configuration is restored.
These functions will also be useful for restoring GIC context after a
suspend to RAM.
This patch depends on Paul Burton's recent patches:
irqchip: mips-gic: Implement activate op for device domain
irqchip: mips-gic: Cleanup chip & handler setup
Without these patches, the context saving will restore an incorrect map
to VPE value, since there is a route to not having the interrupt
affinity set.
Signed-off-by: Matt Redfearn <[email protected]>
---
Changes in v3:
Update GIC context saving to use CPU hotplug state machine
Changes in v2:
Add dependence on additional patches to mips-gic in commit log
Incorporate changes from Marc Zynger's review:
- Remove CONTEXT_SAVING define.
- Make saved local state a per-cpu variable
- Make gic_save_* static functions when enabled, and do { } while(0)
otherwise
drivers/irqchip/irq-mips-gic.c | 208 +++++++++++++++++++++++++++++++++++++++--
1 file changed, 201 insertions(+), 7 deletions(-)
diff --git a/drivers/irqchip/irq-mips-gic.c b/drivers/irqchip/irq-mips-gic.c
index 6185696405d5..7092399b68de 100644
--- a/drivers/irqchip/irq-mips-gic.c
+++ b/drivers/irqchip/irq-mips-gic.c
@@ -8,6 +8,7 @@
*/
#include <linux/bitmap.h>
#include <linux/clocksource.h>
+#include <linux/cpu.h>
#include <linux/init.h>
#include <linux/interrupt.h>
#include <linux/irq.h>
@@ -56,6 +57,79 @@ static unsigned int timer_cpu_pin;
static struct irq_chip gic_level_irq_controller, gic_edge_irq_controller;
DECLARE_BITMAP(ipi_resrv, GIC_MAX_INTRS);
+#ifdef CONFIG_MIPS_REMOTEPROC
+struct gic_local_state_t {
+ u8 mask;
+};
+
+DEFINE_PER_CPU(struct gic_local_state_t, gic_local_state);
+
+static void gic_save_local_rmask(int vpe, int mask)
+{
+ struct gic_local_state_t *state = per_cpu_ptr(&gic_local_state, vpe);
+
+ state->mask &= mask;
+}
+
+static void gic_save_local_smask(int vpe, int mask)
+{
+ struct gic_local_state_t *state = per_cpu_ptr(&gic_local_state, vpe);
+
+ state->mask |= mask;
+}
+
+static struct {
+ unsigned vpe: 8;
+ unsigned pin: 4;
+
+ unsigned polarity: 1;
+ unsigned trigger: 1;
+ unsigned dual_edge: 1;
+ unsigned mask: 1;
+} gic_shared_state[GIC_MAX_INTRS];
+
+static void gic_save_shared_vpe(int intr, int vpe)
+{
+ gic_shared_state[intr].vpe = vpe;
+}
+
+static void gic_save_shared_pin(int intr, int pin)
+{
+ gic_shared_state[intr].pin = pin;
+}
+
+static void gic_save_shared_polarity(int intr, int polarity)
+{
+ gic_shared_state[intr].polarity = polarity;
+}
+
+static void gic_save_shared_trigger(int intr, int trigger)
+{
+ gic_shared_state[intr].trigger = trigger;
+}
+
+static void gic_save_shared_dual_edge(int intr, int dual_edge)
+{
+ gic_shared_state[intr].dual_edge = dual_edge;
+}
+
+static void gic_save_shared_mask(int intr, int mask)
+{
+ gic_shared_state[intr].mask = mask;
+}
+
+#else
+#define gic_save_local_rmask(vpe, i) do { } while (0)
+#define gic_save_local_smask(vpe, i) do { } while (0)
+
+#define gic_save_shared_vpe(i, v) do { } while (0)
+#define gic_save_shared_pin(i, p) do { } while (0)
+#define gic_save_shared_polarity(i, p) do { } while (0)
+#define gic_save_shared_trigger(i, t) do { } while (0)
+#define gic_save_shared_dual_edge(i, d) do { } while (0)
+#define gic_save_shared_mask(i, m) do { } while (0)
+#endif /* CONFIG_MIPS_REMOTEPROC */
+
static void __gic_irq_dispatch(void);
static inline u32 gic_read32(unsigned int reg)
@@ -105,52 +179,94 @@ static inline void gic_update_bits(unsigned int reg, unsigned long mask,
gic_write(reg, regval);
}
-static inline void gic_reset_mask(unsigned int intr)
+static inline void gic_write_reset_mask(unsigned int intr)
{
gic_write(GIC_REG(SHARED, GIC_SH_RMASK) + GIC_INTR_OFS(intr),
1ul << GIC_INTR_BIT(intr));
}
-static inline void gic_set_mask(unsigned int intr)
+static inline void gic_reset_mask(unsigned int intr)
+{
+ gic_save_shared_mask(intr, 0);
+ gic_write_reset_mask(intr);
+}
+
+static inline void gic_write_set_mask(unsigned int intr)
{
gic_write(GIC_REG(SHARED, GIC_SH_SMASK) + GIC_INTR_OFS(intr),
1ul << GIC_INTR_BIT(intr));
}
-static inline void gic_set_polarity(unsigned int intr, unsigned int pol)
+static inline void gic_set_mask(unsigned int intr)
+{
+ gic_save_shared_mask(intr, 1);
+ gic_write_set_mask(intr);
+}
+
+static inline void gic_write_polarity(unsigned int intr, unsigned int pol)
{
gic_update_bits(GIC_REG(SHARED, GIC_SH_SET_POLARITY) +
GIC_INTR_OFS(intr), 1ul << GIC_INTR_BIT(intr),
(unsigned long)pol << GIC_INTR_BIT(intr));
}
-static inline void gic_set_trigger(unsigned int intr, unsigned int trig)
+static inline void gic_set_polarity(unsigned int intr, unsigned int pol)
+{
+ gic_save_shared_polarity(intr, pol);
+ gic_write_polarity(intr, pol);
+}
+
+static inline void gic_write_trigger(unsigned int intr, unsigned int trig)
{
gic_update_bits(GIC_REG(SHARED, GIC_SH_SET_TRIGGER) +
GIC_INTR_OFS(intr), 1ul << GIC_INTR_BIT(intr),
(unsigned long)trig << GIC_INTR_BIT(intr));
}
-static inline void gic_set_dual_edge(unsigned int intr, unsigned int dual)
+static inline void gic_set_trigger(unsigned int intr, unsigned int trig)
+{
+ gic_save_shared_trigger(intr, trig);
+ gic_write_trigger(intr, trig);
+}
+
+static inline void gic_write_dual_edge(unsigned int intr, unsigned int dual)
{
gic_update_bits(GIC_REG(SHARED, GIC_SH_SET_DUAL) + GIC_INTR_OFS(intr),
1ul << GIC_INTR_BIT(intr),
(unsigned long)dual << GIC_INTR_BIT(intr));
}
-static inline void gic_map_to_pin(unsigned int intr, unsigned int pin)
+static inline void gic_set_dual_edge(unsigned int intr, unsigned int dual)
+{
+ gic_save_shared_dual_edge(intr, dual);
+ gic_write_dual_edge(intr, dual);
+}
+
+static inline void gic_write_map_to_pin(unsigned int intr, unsigned int pin)
{
gic_write32(GIC_REG(SHARED, GIC_SH_INTR_MAP_TO_PIN_BASE) +
GIC_SH_MAP_TO_PIN(intr), GIC_MAP_TO_PIN_MSK | pin);
}
-static inline void gic_map_to_vpe(unsigned int intr, unsigned int vpe)
+static inline void gic_map_to_pin(unsigned int intr, unsigned int pin)
+{
+ gic_save_shared_pin(intr, pin);
+ gic_write_map_to_pin(intr, pin);
+}
+
+static inline void gic_write_map_to_vpe(unsigned int intr, unsigned int vpe)
{
gic_write(GIC_REG(SHARED, GIC_SH_INTR_MAP_TO_VPE_BASE) +
GIC_SH_MAP_TO_VPE_REG_OFF(intr, vpe),
GIC_SH_MAP_TO_VPE_REG_BIT(vpe));
}
+static inline void gic_map_to_vpe(unsigned int intr, unsigned int vpe)
+{
+ gic_save_shared_vpe(intr, vpe);
+ gic_write_map_to_vpe(intr, vpe);
+}
+
#ifdef CONFIG_CLKSRC_MIPS_GIC
cycle_t gic_read_count(void)
{
@@ -537,6 +653,7 @@ static void gic_mask_local_irq(struct irq_data *d)
{
int intr = GIC_HWIRQ_TO_LOCAL(d->hwirq);
+ gic_save_local_rmask(smp_processor_id(), (1 << intr));
gic_write32(GIC_REG(VPE_LOCAL, GIC_VPE_RMASK), 1 << intr);
}
@@ -544,6 +661,7 @@ static void gic_unmask_local_irq(struct irq_data *d)
{
int intr = GIC_HWIRQ_TO_LOCAL(d->hwirq);
+ gic_save_local_smask(smp_processor_id(), (1 << intr));
gic_write32(GIC_REG(VPE_LOCAL, GIC_VPE_SMASK), 1 << intr);
}
@@ -561,6 +679,7 @@ static void gic_mask_local_irq_all_vpes(struct irq_data *d)
spin_lock_irqsave(&gic_lock, flags);
for (i = 0; i < gic_vpes; i++) {
+ gic_save_local_rmask(mips_cm_vp_id(i), 1 << intr);
gic_write(GIC_REG(VPE_LOCAL, GIC_VPE_OTHER_ADDR),
mips_cm_vp_id(i));
gic_write32(GIC_REG(VPE_OTHER, GIC_VPE_RMASK), 1 << intr);
@@ -576,6 +695,7 @@ static void gic_unmask_local_irq_all_vpes(struct irq_data *d)
spin_lock_irqsave(&gic_lock, flags);
for (i = 0; i < gic_vpes; i++) {
+ gic_save_local_smask(mips_cm_vp_id(i), 1 << intr);
gic_write(GIC_REG(VPE_LOCAL, GIC_VPE_OTHER_ADDR),
mips_cm_vp_id(i));
gic_write32(GIC_REG(VPE_OTHER, GIC_VPE_SMASK), 1 << intr);
@@ -978,6 +1098,75 @@ static struct irq_domain_ops gic_ipi_domain_ops = {
.match = gic_ipi_domain_match,
};
+#ifdef CONFIG_MIPS_REMOTEPROC
+static void gic_restore_shared(void)
+{
+ unsigned long flags;
+ int i;
+
+ spin_lock_irqsave(&gic_lock, flags);
+ for (i = 0; i < gic_shared_intrs; i++) {
+ gic_write_polarity(i, gic_shared_state[i].polarity);
+ gic_write_trigger(i, gic_shared_state[i].trigger);
+ gic_write_dual_edge(i, gic_shared_state[i].dual_edge);
+ gic_write_map_to_vpe(i, gic_shared_state[i].vpe);
+ gic_write_map_to_pin(i, gic_shared_state[i].pin);
+
+ if (gic_shared_state[i].mask)
+ gic_write_set_mask(i);
+ else
+ gic_write_reset_mask(i);
+ }
+ spin_unlock_irqrestore(&gic_lock, flags);
+}
+
+static void gic_restore_local(unsigned int vpe)
+{
+ struct gic_local_state_t state;
+ int hw, virq, intr, mask;
+ unsigned long flags;
+
+ for (hw = 0; hw < GIC_NUM_LOCAL_INTRS; hw++) {
+ intr = GIC_LOCAL_TO_HWIRQ(hw);
+ virq = irq_linear_revmap(gic_irq_domain, intr);
+ gic_local_irq_domain_map(gic_irq_domain, virq, hw);
+ }
+
+ local_irq_save(flags);
+ gic_write(GIC_REG(VPE_LOCAL, GIC_VPE_OTHER_ADDR), vpe);
+
+ /* Enable EIC mode if necessary */
+ gic_write32(GIC_REG(VPE_OTHER, GIC_VPE_CTL), cpu_has_veic);
+
+ /* Restore interrupt masks */
+ state = per_cpu(gic_local_state, vpe);
+ mask = state.mask;
+ gic_write32(GIC_REG(VPE_OTHER, GIC_VPE_RMASK), ~mask);
+ gic_write32(GIC_REG(VPE_OTHER, GIC_VPE_SMASK), mask);
+
+ local_irq_restore(flags);
+}
+
+/*
+ * The MIPS remote processor driver allows non-Linux firmware to take control
+ * of and execute on one of the systems VPEs. If that VPE is brought back under
+ * Linux, it is necessary to ensure that all GIC interrupts are routed and
+ * masked as Linux expects them, as the firmware can have done anything it
+ * likes with the GIC configuration (hopefully just for that VPEs local
+ * interrupt sources, but allow for shared external interrupts as well).
+ */
+static int gic_cpu_online(unsigned int cpu)
+{
+ unsigned int vpe = mips_cm_vp_id(cpu);
+
+ gic_restore_shared();
+ gic_restore_local(vpe);
+
+ return 0;
+}
+
+#endif /* CONFIG_MIPS_REMOTEPROC */
+
static void __init __gic_init(unsigned long gic_base_addr,
unsigned long gic_addrspace_size,
unsigned int cpu_vec, unsigned int irqbase,
@@ -1077,6 +1266,11 @@ static void __init __gic_init(unsigned long gic_base_addr,
}
gic_basic_init();
+
+#ifdef CONFIG_MIPS_REMOTEPROC
+ cpuhp_setup_state(CPUHP_AP_ONLINE_DYN, "MIPS:GIC(REMOTEPROC)",
+ gic_cpu_online, NULL);
+#endif /* CONFIG_MIPS_REMOTEPROC */
}
void __init gic_init(unsigned long gic_base_addr,
--
2.7.4
This interface is currently used by the Intel / Lantiq voice Firmware.
This firmware is used by all Intel / Lantiq MIPS SoCs when they should
support analog voice.
What is the proposed timeline for the removal of this interface?
Hauke
On 10/11/2016 03:42 PM, Matt Redfearn wrote:
> The MIPS remote processor driver (CONFIG_MIPS_RPROC) provides a more
> standard mechanism for using one or more VPs as coprocessors running
> separate firmware.
>
> Here we deprecate this mechanism before it is removed.
>
> Signed-off-by: Matt Redfearn <[email protected]>
> ---
>
> Changes in v3: None
> Changes in v2: None
>
> arch/mips/Kconfig | 5 ++++-
> 1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
> index fd2468e71e63..3e767c373c7b 100644
> --- a/arch/mips/Kconfig
> +++ b/arch/mips/Kconfig
> @@ -2264,7 +2264,7 @@ comment "MIPS R2-to-R6 emulator is only available for UP kernels"
> depends on SMP && CPU_MIPSR6
>
> config MIPS_VPE_LOADER
> - bool "VPE loader support."
> + bool "VPE loader support (DEPRECATED)"
> depends on SYS_SUPPORTS_MULTITHREADING && MODULES
> select CPU_MIPSR2_IRQ_VI
> select CPU_MIPSR2_IRQ_EI
> @@ -2273,6 +2273,9 @@ config MIPS_VPE_LOADER
> Includes a loader for loading an elf relocatable object
> onto another VPE and running it.
>
> + Unless you have a specific need, you should use CONFIG_MIPS_RPROC
> + instead of this.
> +
> config MIPS_VPE_LOADER_CMP
> bool
> default "y"
>
Hi Hauke,
On 12/10/16 12:09, Hauke Mehrtens wrote:
> This interface is currently used by the Intel / Lantiq voice Firmware.
> This firmware is used by all Intel / Lantiq MIPS SoCs when they should
> support analog voice.
>
> What is the proposed timeline for the removal of this interface?
Ultimately I guess that's Ralf's call, but when I've discussed this with
him previously, he wasn't aware of any users of this feature. But if
there are indeed active users then that will slow down it's removal to
give them sufficient time to migrate or otherwise stop using it.
Thanks,
Matt
>
> Hauke
>
> On 10/11/2016 03:42 PM, Matt Redfearn wrote:
>> The MIPS remote processor driver (CONFIG_MIPS_RPROC) provides a more
>> standard mechanism for using one or more VPs as coprocessors running
>> separate firmware.
>>
>> Here we deprecate this mechanism before it is removed.
>>
>> Signed-off-by: Matt Redfearn <[email protected]>
>> ---
>>
>> Changes in v3: None
>> Changes in v2: None
>>
>> arch/mips/Kconfig | 5 ++++-
>> 1 file changed, 4 insertions(+), 1 deletion(-)
>>
>> diff --git a/arch/mips/Kconfig b/arch/mips/Kconfig
>> index fd2468e71e63..3e767c373c7b 100644
>> --- a/arch/mips/Kconfig
>> +++ b/arch/mips/Kconfig
>> @@ -2264,7 +2264,7 @@ comment "MIPS R2-to-R6 emulator is only available for UP kernels"
>> depends on SMP && CPU_MIPSR6
>>
>> config MIPS_VPE_LOADER
>> - bool "VPE loader support."
>> + bool "VPE loader support (DEPRECATED)"
>> depends on SYS_SUPPORTS_MULTITHREADING && MODULES
>> select CPU_MIPSR2_IRQ_VI
>> select CPU_MIPSR2_IRQ_EI
>> @@ -2273,6 +2273,9 @@ config MIPS_VPE_LOADER
>> Includes a loader for loading an elf relocatable object
>> onto another VPE and running it.
>>
>> + Unless you have a specific need, you should use CONFIG_MIPS_RPROC
>> + instead of this.
>> +
>> config MIPS_VPE_LOADER_CMP
>> bool
>> default "y"
>>
On 10/11/2016 03:42 PM, Matt Redfearn wrote:
>
> The MIPS remote processor driver allows non-Linux firmware to take
> control of and execute on one of the systems VPEs. The CPU must be
> offlined from Linux first. A sysfs interface is created which allows
> firmware to be loaded and changed at runtime. A full description is
> available at [1]. An example firmware that can be used with this driver
> is available at [2].
>
> This is useful to allow running bare metal code, or an RTOS, on one or
> more CPUs while allowing Linux to continue running on those remaining.
>
> The remote processor framework allows for firmwares to provide any
> virtio device for communication between the firmware running on the
> remote VP and Linux. For example [1] demonstrates a simple firmware
> which provides a virtual serial port. Any string sent to the port is
> case inverted and returned.
>
> This is conceptually similar to the VPE loader functionality, but is
> more standard as it fits into the remoteproc subsystem.
>
> The first patches in this series lay the groundwork for the driver
> before it is added. The last series deprecates the VPE loader.
>
> This functionality is supported on:
> - MIPS32r2 devices implementing the MIPS MT ASE for multithreading, such
> as interAptiv.
> - MIPS32r6 devices implementing VPs, such as I6400.
>
> Limitations:
> - The remoteproc core supports only 32bit ELFs. Therefore it is only
> possible to run 32bit firmware on the remote processor. Also, for
> virtio communication, pointers are passed from the kernel to firmware.
> There can be no mismatch in pointer sizes between the kernel and
> firmware, so this limits the host kernel to 32bit as well.
>
> The functionality has been tested on the Ci40 board which has a 2 core 2
> thread interAptiv.
>
Does this also work with big firmware binaries, like 2 MB firmware size?
It looks like dma_alloc_coherent() is used in rproc_handle_carveout() to
allocate the memory which will probably not work after full boot.
Is it possible to configure the cache line in a way that it will not be
removed from the cache to avoid high latency loads from main memory?
This is only needed for small firmware binaries, but allows even shorter
reaction times and better real time behavior.
In general I am in favor of having this as a standard Linux interface.
Hauke
Hi Hauke,
On 17/10/16 22:13, Hauke Mehrtens wrote:
> On 10/11/2016 03:42 PM, Matt Redfearn wrote:
>> The MIPS remote processor driver allows non-Linux firmware to take
>> control of and execute on one of the systems VPEs. The CPU must be
>> offlined from Linux first. A sysfs interface is created which allows
>> firmware to be loaded and changed at runtime. A full description is
>> available at [1]. An example firmware that can be used with this driver
>> is available at [2].
>>
>> This is useful to allow running bare metal code, or an RTOS, on one or
>> more CPUs while allowing Linux to continue running on those remaining.
>>
>> The remote processor framework allows for firmwares to provide any
>> virtio device for communication between the firmware running on the
>> remote VP and Linux. For example [1] demonstrates a simple firmware
>> which provides a virtual serial port. Any string sent to the port is
>> case inverted and returned.
>>
>> This is conceptually similar to the VPE loader functionality, but is
>> more standard as it fits into the remoteproc subsystem.
>>
>> The first patches in this series lay the groundwork for the driver
>> before it is added. The last series deprecates the VPE loader.
>>
>> This functionality is supported on:
>> - MIPS32r2 devices implementing the MIPS MT ASE for multithreading, such
>> as interAptiv.
>> - MIPS32r6 devices implementing VPs, such as I6400.
>>
>> Limitations:
>> - The remoteproc core supports only 32bit ELFs. Therefore it is only
>> possible to run 32bit firmware on the remote processor. Also, for
>> virtio communication, pointers are passed from the kernel to firmware.
>> There can be no mismatch in pointer sizes between the kernel and
>> firmware, so this limits the host kernel to 32bit as well.
>>
>> The functionality has been tested on the Ci40 board which has a 2 core 2
>> thread interAptiv.
>>
> Does this also work with big firmware binaries, like 2 MB firmware size?
> It looks like dma_alloc_coherent() is used in rproc_handle_carveout() to
> allocate the memory which will probably not work after full boot.
In my testing, a 2Mb firmware carveout does work fine on a Creator Ci40
with 256Mb RAM. But like you say, using the DMA api will limit the size
of the firmware. In our driver's case, we end up setting up wired TLB
entries mapping the carveout into the offline CPUs virtual memory, so in
principle there is no reason (for our driver) that the memory has to be
contiguous since we could map smaller pages. But this is code in the
generic remoteproc framework so must work for other processors which
could not deal with that.
>
> Is it possible to configure the cache line in a way that it will not be
> removed from the cache to avoid high latency loads from main memory?
> This is only needed for small firmware binaries, but allows even shorter
> reaction times and better real time behavior.
The firmware running on the offline CPU has full access to the hardware
so could do whatever it likes with it's cache lines - it's not something
that the remote processor driver ought to do as a matter of course
though I don't think.
>
> In general I am in favor of having this as a standard Linux interface.
Great :-)
Thanks,
Matt
> Hauke