2016-11-01 17:11:14

by Christopher Covington

[permalink] [raw]
Subject: [RFC v2 1/7] mm: Provide generic VDSO unmap and remap functions

When Address Space Layout Randomization (ASLR, randmaps) is enabled, the
address of the VDSO fluctuates from one process to the next. If
Checkpoint/Restore In Userspace (CRIU) is to fully replicate the memory map
of a previous process, it must be able to remap the VDSO of a new process
to the address used by the previous process. Historically this has been
implemented in architecture-specific code for PowerPC and x86. In order to
support 32-bit and 64-bit ARM without further duplication of code, copy
Laurent Dufour's implementation for PowerPC with slight modifications to a
generic location. This is hopefully the beginning of a long process of VDSO
code de-duplication between architectures.

Signed-off-by: Christopher Covington <[email protected]>
---
include/asm-generic/mm_hooks.h | 35 ++++++++++++++++++++++++++++++++---
1 file changed, 32 insertions(+), 3 deletions(-)

diff --git a/include/asm-generic/mm_hooks.h b/include/asm-generic/mm_hooks.h
index cc5d9a1..73f09f1 100644
--- a/include/asm-generic/mm_hooks.h
+++ b/include/asm-generic/mm_hooks.h
@@ -1,7 +1,17 @@
/*
- * Define generic no-op hooks for arch_dup_mmap, arch_exit_mmap
- * and arch_unmap to be included in asm-FOO/mmu_context.h for any
- * arch FOO which doesn't need to hook these.
+ * Define generic hooks for arch_dup_mmap, arch_exit_mmap and arch_unmap to be
+ * included in asm-FOO/mmu_context.h for any arch FOO which doesn't need to
+ * specially hook these.
+ *
+ * arch_remap originally from include/linux-mm-arch-hooks.h
+ * arch_unmap originally from arch/powerpc/include/asm/mmu_context.h
+ * Copyright (C) 2015, IBM Corporation
+ * Author: Laurent Dufour <[email protected]>
+ *
+ * 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_GENERIC_MM_HOOKS_H
#define _ASM_GENERIC_MM_HOOKS_H
@@ -19,6 +29,25 @@ static inline void arch_unmap(struct mm_struct *mm,
struct vm_area_struct *vma,
unsigned long start, unsigned long end)
{
+#ifdef CONFIG_GENERIC_VDSO
+ if (start <= mm->context.vdso && mm->context.vdso < end)
+ mm->context.vdso = 0;
+#endif /* CONFIG_GENERIC_VDSO */
+}
+
+static inline void arch_remap(struct mm_struct *mm,
+ unsigned long old_start, unsigned long old_end,
+ unsigned long new_start, unsigned long new_end)
+{
+#ifdef CONFIG_GENERIC_VDSO
+ /*
+ * mremap() doesn't allow moving multiple vmas so we can limit the
+ * check to old_addr == vdso.
+ */
+ if (old_addr == mm->context.vdso)
+ mm->context.vdso = new_addr;
+
+#endif /* CONFIG_GENERIC_VDSO */
}

static inline void arch_bprm_mm_init(struct mm_struct *mm,
--
Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.


2016-11-01 17:11:15

by Christopher Covington

[permalink] [raw]
Subject: [RFC v2 2/7] arm: Use generic VDSO unmap and remap

Checkpoint/Restore In Userspace (CRIU) needs to be able to unmap and remap
the VDSO to successfully checkpoint and restore applications in the face of
changing VDSO addresses due to Address Space Layout Randomization (ASLR,
randmaps). Previously, this was implemented in architecture-specific code
for PowerPC and x86. However, a generic version based on Laurent Dufour's
PowerPC implementation is now available, so begin using it on ARM.

Signed-off-by: Christopher Covington <[email protected]>
---
arch/arm/mm/Kconfig | 1 +
1 file changed, 1 insertion(+)

diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
index c1799dd..1d3312b 100644
--- a/arch/arm/mm/Kconfig
+++ b/arch/arm/mm/Kconfig
@@ -845,6 +845,7 @@ config VDSO
depends on AEABI && MMU && CPU_V7
default y if ARM_ARCH_TIMER
select GENERIC_TIME_VSYSCALL
+ select GENERIC_VDSO
help
Place in the process address space an ELF shared object
providing fast implementations of gettimeofday and
--
Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

2016-11-01 17:11:21

by Christopher Covington

[permalink] [raw]
Subject: [RFC v2 5/7] powerpc: Rename context.vdso_base to context.vdso

Checkpoint/Restore In Userspace (CRIU) needs to be able to unmap and remap
the VDSO to successfully checkpoint and restore applications in the face of
changing VDSO addresses due to Address Space Layout Randomization (ASLR,
randmaps). x86 and PowerPC have had architecture-specific code to support
this. In order to expand the architectures that support this without
unnecessary duplication of code, a generic version based on the PowerPC code
was created. It differs slightly, based on the results of an informal
survey of all architectures that indicated

unsigned long vdso;

is popular (and it's also concise). Therefore, change the variable name in
powerpc from mm->context.vdso_base to mm->context.vdso.

Signed-off-by: Christopher Covington <[email protected]>
---
arch/powerpc/include/asm/book3s/32/mmu-hash.h | 2 +-
arch/powerpc/include/asm/book3s/64/mmu.h | 2 +-
arch/powerpc/include/asm/mm-arch-hooks.h | 6 +++---
arch/powerpc/include/asm/mmu-40x.h | 2 +-
arch/powerpc/include/asm/mmu-44x.h | 2 +-
arch/powerpc/include/asm/mmu-8xx.h | 2 +-
arch/powerpc/include/asm/mmu-book3e.h | 2 +-
arch/powerpc/include/asm/mmu_context.h | 4 ++--
arch/powerpc/include/asm/vdso.h | 2 +-
arch/powerpc/include/uapi/asm/elf.h | 2 +-
arch/powerpc/kernel/signal_32.c | 8 ++++----
arch/powerpc/kernel/signal_64.c | 4 ++--
arch/powerpc/kernel/vdso.c | 8 ++++----
arch/powerpc/perf/callchain.c | 12 ++++++------
14 files changed, 29 insertions(+), 29 deletions(-)

diff --git a/arch/powerpc/include/asm/book3s/32/mmu-hash.h b/arch/powerpc/include/asm/book3s/32/mmu-hash.h
index b82e063..75738bb 100644
--- a/arch/powerpc/include/asm/book3s/32/mmu-hash.h
+++ b/arch/powerpc/include/asm/book3s/32/mmu-hash.h
@@ -79,7 +79,7 @@ struct hash_pte {

typedef struct {
unsigned long id;
- unsigned long vdso_base;
+ unsigned long vdso;
} mm_context_t;

#endif /* !__ASSEMBLY__ */
diff --git a/arch/powerpc/include/asm/book3s/64/mmu.h b/arch/powerpc/include/asm/book3s/64/mmu.h
index 8afb0e0..8486a10 100644
--- a/arch/powerpc/include/asm/book3s/64/mmu.h
+++ b/arch/powerpc/include/asm/book3s/64/mmu.h
@@ -72,7 +72,7 @@ typedef struct {
#else
u16 sllp; /* SLB page size encoding */
#endif
- unsigned long vdso_base;
+ unsigned long vdso;
#ifdef CONFIG_PPC_SUBPAGE_PROT
struct subpage_prot_table spt;
#endif /* CONFIG_PPC_SUBPAGE_PROT */
diff --git a/arch/powerpc/include/asm/mm-arch-hooks.h b/arch/powerpc/include/asm/mm-arch-hooks.h
index f2a2da8..ea6da89 100644
--- a/arch/powerpc/include/asm/mm-arch-hooks.h
+++ b/arch/powerpc/include/asm/mm-arch-hooks.h
@@ -18,10 +18,10 @@ static inline void arch_remap(struct mm_struct *mm,
{
/*
* mremap() doesn't allow moving multiple vmas so we can limit the
- * check to old_start == vdso_base.
+ * check to old_start == vdso.
*/
- if (old_start == mm->context.vdso_base)
- mm->context.vdso_base = new_start;
+ if (old_start == mm->context.vdso)
+ mm->context.vdso = new_start;
}
#define arch_remap arch_remap

diff --git a/arch/powerpc/include/asm/mmu-40x.h b/arch/powerpc/include/asm/mmu-40x.h
index 3491686..e8e57b7 100644
--- a/arch/powerpc/include/asm/mmu-40x.h
+++ b/arch/powerpc/include/asm/mmu-40x.h
@@ -56,7 +56,7 @@
typedef struct {
unsigned int id;
unsigned int active;
- unsigned long vdso_base;
+ unsigned long vdso;
} mm_context_t;

#endif /* !__ASSEMBLY__ */
diff --git a/arch/powerpc/include/asm/mmu-44x.h b/arch/powerpc/include/asm/mmu-44x.h
index bf52d70..471891c 100644
--- a/arch/powerpc/include/asm/mmu-44x.h
+++ b/arch/powerpc/include/asm/mmu-44x.h
@@ -107,7 +107,7 @@ extern unsigned int tlb_44x_index;
typedef struct {
unsigned int id;
unsigned int active;
- unsigned long vdso_base;
+ unsigned long vdso;
} mm_context_t;

#endif /* !__ASSEMBLY__ */
diff --git a/arch/powerpc/include/asm/mmu-8xx.h b/arch/powerpc/include/asm/mmu-8xx.h
index 3e0e492..2834af0 100644
--- a/arch/powerpc/include/asm/mmu-8xx.h
+++ b/arch/powerpc/include/asm/mmu-8xx.h
@@ -167,7 +167,7 @@
typedef struct {
unsigned int id;
unsigned int active;
- unsigned long vdso_base;
+ unsigned long vdso;
} mm_context_t;

#define PHYS_IMMR_BASE (mfspr(SPRN_IMMR) & 0xfff80000)
diff --git a/arch/powerpc/include/asm/mmu-book3e.h b/arch/powerpc/include/asm/mmu-book3e.h
index b62a8d4..28dc4e0 100644
--- a/arch/powerpc/include/asm/mmu-book3e.h
+++ b/arch/powerpc/include/asm/mmu-book3e.h
@@ -228,7 +228,7 @@ extern unsigned int tlbcam_index;
typedef struct {
unsigned int id;
unsigned int active;
- unsigned long vdso_base;
+ unsigned long vdso;
#ifdef CONFIG_PPC_MM_SLICES
u64 low_slices_psize; /* SLB page size encodings */
u64 high_slices_psize; /* 4 bits per slice for now */
diff --git a/arch/powerpc/include/asm/mmu_context.h b/arch/powerpc/include/asm/mmu_context.h
index 5c45114..c907478 100644
--- a/arch/powerpc/include/asm/mmu_context.h
+++ b/arch/powerpc/include/asm/mmu_context.h
@@ -146,8 +146,8 @@ static inline void arch_unmap(struct mm_struct *mm,
struct vm_area_struct *vma,
unsigned long start, unsigned long end)
{
- if (start <= mm->context.vdso_base && mm->context.vdso_base < end)
- mm->context.vdso_base = 0;
+ if (start <= mm->context.vdso && mm->context.vdso < end)
+ mm->context.vdso = 0;
}

static inline void arch_bprm_mm_init(struct mm_struct *mm,
diff --git a/arch/powerpc/include/asm/vdso.h b/arch/powerpc/include/asm/vdso.h
index c53f5f6..fc90971 100644
--- a/arch/powerpc/include/asm/vdso.h
+++ b/arch/powerpc/include/asm/vdso.h
@@ -17,7 +17,7 @@

#ifndef __ASSEMBLY__

-/* Offsets relative to thread->vdso_base */
+/* Offsets relative to mm->context.vdso */
extern unsigned long vdso64_rt_sigtramp;
extern unsigned long vdso32_sigtramp;
extern unsigned long vdso32_rt_sigtramp;
diff --git a/arch/powerpc/include/uapi/asm/elf.h b/arch/powerpc/include/uapi/asm/elf.h
index 3a9e44c..d7c81ae 100644
--- a/arch/powerpc/include/uapi/asm/elf.h
+++ b/arch/powerpc/include/uapi/asm/elf.h
@@ -182,7 +182,7 @@ do { \
NEW_AUX_ENT(AT_DCACHEBSIZE, dcache_bsize); \
NEW_AUX_ENT(AT_ICACHEBSIZE, icache_bsize); \
NEW_AUX_ENT(AT_UCACHEBSIZE, ucache_bsize); \
- VDSO_AUX_ENT(AT_SYSINFO_EHDR, current->mm->context.vdso_base); \
+ VDSO_AUX_ENT(AT_SYSINFO_EHDR, current->mm->context.vdso); \
} while (0)

/* PowerPC64 relocations defined by the ABIs */
diff --git a/arch/powerpc/kernel/signal_32.c b/arch/powerpc/kernel/signal_32.c
index 27aa913..7bb0882 100644
--- a/arch/powerpc/kernel/signal_32.c
+++ b/arch/powerpc/kernel/signal_32.c
@@ -1006,9 +1006,9 @@ int handle_rt_signal32(struct ksignal *ksig, sigset_t *oldset,
/* Save user registers on the stack */
frame = &rt_sf->uc.uc_mcontext;
addr = frame;
- if (vdso32_rt_sigtramp && tsk->mm->context.vdso_base) {
+ if (vdso32_rt_sigtramp && tsk->mm->context.vdso) {
sigret = 0;
- tramp = tsk->mm->context.vdso_base + vdso32_rt_sigtramp;
+ tramp = tsk->mm->context.vdso + vdso32_rt_sigtramp;
} else {
sigret = __NR_rt_sigreturn;
tramp = (unsigned long) frame->tramp;
@@ -1449,9 +1449,9 @@ int handle_signal32(struct ksignal *ksig, sigset_t *oldset,
|| __put_user(ksig->sig, &sc->signal))
goto badframe;

- if (vdso32_sigtramp && tsk->mm->context.vdso_base) {
+ if (vdso32_sigtramp && tsk->mm->context.vdso) {
sigret = 0;
- tramp = tsk->mm->context.vdso_base + vdso32_sigtramp;
+ tramp = tsk->mm->context.vdso + vdso32_sigtramp;
} else {
sigret = __NR_sigreturn;
tramp = (unsigned long) frame->mctx.tramp;
diff --git a/arch/powerpc/kernel/signal_64.c b/arch/powerpc/kernel/signal_64.c
index 96698fd..608a919 100644
--- a/arch/powerpc/kernel/signal_64.c
+++ b/arch/powerpc/kernel/signal_64.c
@@ -791,8 +791,8 @@ int handle_rt_signal64(struct ksignal *ksig, sigset_t *set,
tsk->thread.fp_state.fpscr = 0;

/* Set up to return from userspace. */
- if (vdso64_rt_sigtramp && tsk->mm->context.vdso_base) {
- regs->link = tsk->mm->context.vdso_base + vdso64_rt_sigtramp;
+ if (vdso64_rt_sigtramp && tsk->mm->context.vdso) {
+ regs->link = tsk->mm->context.vdso + vdso64_rt_sigtramp;
} else {
err |= setup_trampoline(__NR_rt_sigreturn, &frame->tramp[0]);
if (err)
diff --git a/arch/powerpc/kernel/vdso.c b/arch/powerpc/kernel/vdso.c
index 4111d30..33ea0f8 100644
--- a/arch/powerpc/kernel/vdso.c
+++ b/arch/powerpc/kernel/vdso.c
@@ -180,7 +180,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
vdso_base = VDSO32_MBASE;
#endif

- current->mm->context.vdso_base = 0;
+ current->mm->context.vdso = 0;

/* vDSO has a problem and was disabled, just don't "enable" it for the
* process
@@ -215,7 +215,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
* install_special_mapping or the perf counter mmap tracking code
* will fail to recognise it as a vDSO (since arch_vma_name fails).
*/
- current->mm->context.vdso_base = vdso_base;
+ current->mm->context.vdso = vdso_base;

/*
* our vma flags don't have VM_WRITE so by default, the process isn't
@@ -232,7 +232,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC,
vdso_pagelist);
if (rc) {
- current->mm->context.vdso_base = 0;
+ current->mm->context.vdso = 0;
goto fail_mmapsem;
}

@@ -246,7 +246,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)

const char *arch_vma_name(struct vm_area_struct *vma)
{
- if (vma->vm_mm && vma->vm_start == vma->vm_mm->context.vdso_base)
+ if (vma->vm_mm && vma->vm_start == vma->vm_mm->context.vdso)
return "[vdso]";
return NULL;
}
diff --git a/arch/powerpc/perf/callchain.c b/arch/powerpc/perf/callchain.c
index 0fc2671..5c893a2 100644
--- a/arch/powerpc/perf/callchain.c
+++ b/arch/powerpc/perf/callchain.c
@@ -209,8 +209,8 @@ static int is_sigreturn_64_address(unsigned long nip, unsigned long fp)
{
if (nip == fp + offsetof(struct signal_frame_64, tramp))
return 1;
- if (vdso64_rt_sigtramp && current->mm->context.vdso_base &&
- nip == current->mm->context.vdso_base + vdso64_rt_sigtramp)
+ if (vdso64_rt_sigtramp && current->mm->context.vdso &&
+ nip == current->mm->context.vdso + vdso64_rt_sigtramp)
return 1;
return 0;
}
@@ -368,8 +368,8 @@ static int is_sigreturn_32_address(unsigned int nip, unsigned int fp)
{
if (nip == fp + offsetof(struct signal_frame_32, mctx.mc_pad))
return 1;
- if (vdso32_sigtramp && current->mm->context.vdso_base &&
- nip == current->mm->context.vdso_base + vdso32_sigtramp)
+ if (vdso32_sigtramp && current->mm->context.vdso &&
+ nip == current->mm->context.vdso + vdso32_sigtramp)
return 1;
return 0;
}
@@ -379,8 +379,8 @@ static int is_rt_sigreturn_32_address(unsigned int nip, unsigned int fp)
if (nip == fp + offsetof(struct rt_signal_frame_32,
uc.uc_mcontext.mc_pad))
return 1;
- if (vdso32_rt_sigtramp && current->mm->context.vdso_base &&
- nip == current->mm->context.vdso_base + vdso32_rt_sigtramp)
+ if (vdso32_rt_sigtramp && current->mm->context.vdso &&
+ nip == current->mm->context.vdso + vdso32_rt_sigtramp)
return 1;
return 0;
}
--
Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

2016-11-01 17:12:01

by Christopher Covington

[permalink] [raw]
Subject: [RFC v2 3/7] arm64: Use unsigned long for VDSO

Use an unsigned long type for the base address of the VDSO in order to be
compatible with the new generic VDSO remap and unmap functions originating
from PowerPC and now also used by 32-bit ARM.

Signed-off-by: Christopher Covington <[email protected]>
---
arch/arm64/include/asm/mmu.h | 2 +-
arch/arm64/kernel/vdso.c | 6 +++---
2 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/arch/arm64/include/asm/mmu.h b/arch/arm64/include/asm/mmu.h
index 8d9fce0..5b00198 100644
--- a/arch/arm64/include/asm/mmu.h
+++ b/arch/arm64/include/asm/mmu.h
@@ -18,7 +18,7 @@

typedef struct {
atomic64_t id;
- void *vdso;
+ unsigned long vdso;
} mm_context_t;

/*
diff --git a/arch/arm64/kernel/vdso.c b/arch/arm64/kernel/vdso.c
index a2c2478..4b10e72 100644
--- a/arch/arm64/kernel/vdso.c
+++ b/arch/arm64/kernel/vdso.c
@@ -97,7 +97,7 @@ int aarch32_setup_vectors_page(struct linux_binprm *bprm, int uses_interp)

if (down_write_killable(&mm->mmap_sem))
return -EINTR;
- current->mm->context.vdso = (void *)addr;
+ current->mm->context.vdso = addr;

/* Map vectors page at the high address. */
ret = _install_special_mapping(mm, addr, PAGE_SIZE,
@@ -178,7 +178,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm,
goto up_fail;

vdso_base += PAGE_SIZE;
- mm->context.vdso = (void *)vdso_base;
+ mm->context.vdso = vdso_base;
ret = _install_special_mapping(mm, vdso_base, vdso_text_len,
VM_READ|VM_EXEC|
VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC,
@@ -191,7 +191,7 @@ int arch_setup_additional_pages(struct linux_binprm *bprm,
return 0;

up_fail:
- mm->context.vdso = NULL;
+ mm->context.vdso = 0;
up_write(&mm->mmap_sem);
return PTR_ERR(ret);
}
--
Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

2016-11-01 17:11:29

by Christopher Covington

[permalink] [raw]
Subject: [RFC v2 6/7] mm/powerpc: Use generic VDSO remap and unmap functions

The PowerPC VDSO remap and unmap code was copied to a generic location,
only modifying the variable name expected in mm->context (vdso instead of
vdso_base) to match most other architectures. Having adopted this generic
naming, drop the code in arch/powerpc and use the generic version.

Signed-off-by: Christopher Covington <[email protected]>
---
arch/powerpc/Kconfig | 1 +
arch/powerpc/include/asm/Kbuild | 1 +
arch/powerpc/include/asm/mm-arch-hooks.h | 28 -------------------------
arch/powerpc/include/asm/mmu_context.h | 35 +-------------------------------
4 files changed, 3 insertions(+), 62 deletions(-)
delete mode 100644 arch/powerpc/include/asm/mm-arch-hooks.h

diff --git a/arch/powerpc/Kconfig b/arch/powerpc/Kconfig
index 65fba4c..f4a1cb9 100644
--- a/arch/powerpc/Kconfig
+++ b/arch/powerpc/Kconfig
@@ -134,6 +134,7 @@ config PPC
select ARCH_HAS_TICK_BROADCAST if GENERIC_CLOCKEVENTS_BROADCAST
select GENERIC_STRNCPY_FROM_USER
select GENERIC_STRNLEN_USER
+ select GENERIC_VDSO
select HAVE_MOD_ARCH_SPECIFIC
select MODULES_USE_ELF_RELA
select CLONE_BACKWARDS
diff --git a/arch/powerpc/include/asm/Kbuild b/arch/powerpc/include/asm/Kbuild
index 5c4fbc8..4d89ec6 100644
--- a/arch/powerpc/include/asm/Kbuild
+++ b/arch/powerpc/include/asm/Kbuild
@@ -8,3 +8,4 @@ generic-y += mcs_spinlock.h
generic-y += preempt.h
generic-y += rwsem.h
generic-y += vtime.h
+generic-y += mm-arch-hooks.h
diff --git a/arch/powerpc/include/asm/mm-arch-hooks.h b/arch/powerpc/include/asm/mm-arch-hooks.h
deleted file mode 100644
index ea6da89..0000000
--- a/arch/powerpc/include/asm/mm-arch-hooks.h
+++ /dev/null
@@ -1,28 +0,0 @@
-/*
- * Architecture specific mm hooks
- *
- * Copyright (C) 2015, IBM Corporation
- * Author: Laurent Dufour <[email protected]>
- *
- * 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_POWERPC_MM_ARCH_HOOKS_H
-#define _ASM_POWERPC_MM_ARCH_HOOKS_H
-
-static inline void arch_remap(struct mm_struct *mm,
- unsigned long old_start, unsigned long old_end,
- unsigned long new_start, unsigned long new_end)
-{
- /*
- * mremap() doesn't allow moving multiple vmas so we can limit the
- * check to old_start == vdso.
- */
- if (old_start == mm->context.vdso)
- mm->context.vdso = new_start;
-}
-#define arch_remap arch_remap
-
-#endif /* _ASM_POWERPC_MM_ARCH_HOOKS_H */
diff --git a/arch/powerpc/include/asm/mmu_context.h b/arch/powerpc/include/asm/mmu_context.h
index c907478..d8dcf45 100644
--- a/arch/powerpc/include/asm/mmu_context.h
+++ b/arch/powerpc/include/asm/mmu_context.h
@@ -8,6 +8,7 @@
#include <linux/spinlock.h>
#include <asm/mmu.h>
#include <asm/cputable.h>
+#include <asm-generic/mm_hooks.h>
#include <asm/cputhreads.h>

/*
@@ -133,39 +134,5 @@ static inline void enter_lazy_tlb(struct mm_struct *mm,
#endif
}

-static inline void arch_dup_mmap(struct mm_struct *oldmm,
- struct mm_struct *mm)
-{
-}
-
-static inline void arch_exit_mmap(struct mm_struct *mm)
-{
-}
-
-static inline void arch_unmap(struct mm_struct *mm,
- struct vm_area_struct *vma,
- unsigned long start, unsigned long end)
-{
- if (start <= mm->context.vdso && mm->context.vdso < end)
- mm->context.vdso = 0;
-}
-
-static inline void arch_bprm_mm_init(struct mm_struct *mm,
- struct vm_area_struct *vma)
-{
-}
-
-static inline bool arch_vma_access_permitted(struct vm_area_struct *vma,
- bool write, bool execute, bool foreign)
-{
- /* by default, allow everything */
- return true;
-}
-
-static inline bool arch_pte_access_permitted(pte_t pte, bool write)
-{
- /* by default, allow everything */
- return true;
-}
#endif /* __KERNEL__ */
#endif /* __ASM_POWERPC_MMU_CONTEXT_H */
--
Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

2016-11-01 17:11:59

by Christopher Covington

[permalink] [raw]
Subject: [RFC v2 4/7] arm64: Use generic VDSO unmap and remap functions

Checkpoint/Restore In Userspace (CRIU) must be able to remap and unmap the
Virtual Dynamic Shared Object (VDSO) to be able to handle the changing
addresses that result from address space layout randomization. Now that
generic support is available and arm64 has adopted unsigned long for the
type of mm->context.vdso, opt-in to VDSO unmap and remap support.

Signed-off-by: Christopher Covington <[email protected]>
---
arch/arm64/Kconfig | 1 +
1 file changed, 1 insertion(+)

diff --git a/arch/arm64/Kconfig b/arch/arm64/Kconfig
index 969ef88..534df3f 100644
--- a/arch/arm64/Kconfig
+++ b/arch/arm64/Kconfig
@@ -50,6 +50,7 @@ config ARM64
select GENERIC_STRNCPY_FROM_USER
select GENERIC_STRNLEN_USER
select GENERIC_TIME_VSYSCALL
+ select GENERIC_VDSO
select HANDLE_DOMAIN_IRQ
select HARDIRQS_SW_RESEND
select HAVE_ALIGNED_STRUCT_PAGE if SLUB
--
Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the
Code Aurora Forum, a Linux Foundation Collaborative Project.

2016-11-01 17:20:54

by Russell King (Oracle)

[permalink] [raw]
Subject: Re: [RFC v2 2/7] arm: Use generic VDSO unmap and remap

You know, on its own, this patch is totally meaningless. Sorry, there's
nothing more I can say about this.

On Tue, Nov 01, 2016 at 11:10:56AM -0600, Christopher Covington wrote:
> Checkpoint/Restore In Userspace (CRIU) needs to be able to unmap and remap
> the VDSO to successfully checkpoint and restore applications in the face of
> changing VDSO addresses due to Address Space Layout Randomization (ASLR,
> randmaps). Previously, this was implemented in architecture-specific code
> for PowerPC and x86. However, a generic version based on Laurent Dufour's
> PowerPC implementation is now available, so begin using it on ARM.
>
> Signed-off-by: Christopher Covington <[email protected]>
> ---
> arch/arm/mm/Kconfig | 1 +
> 1 file changed, 1 insertion(+)
>
> diff --git a/arch/arm/mm/Kconfig b/arch/arm/mm/Kconfig
> index c1799dd..1d3312b 100644
> --- a/arch/arm/mm/Kconfig
> +++ b/arch/arm/mm/Kconfig
> @@ -845,6 +845,7 @@ config VDSO
> depends on AEABI && MMU && CPU_V7
> default y if ARM_ARCH_TIMER
> select GENERIC_TIME_VSYSCALL
> + select GENERIC_VDSO
> help
> Place in the process address space an ELF shared object
> providing fast implementations of gettimeofday and
> --
> Qualcomm Datacenter Technologies as an affiliate of Qualcomm Technologies, Inc.
> Qualcomm Technologies, Inc. is a member of the
> Code Aurora Forum, a Linux Foundation Collaborative Project.
>

--
RMK's Patch system: http://www.armlinux.org.uk/developer/patches/
FTTC broadband for 0.8mile line: currently at 9.6Mbps down 400kbps up
according to speedtest.net.

2016-11-01 17:24:18

by Dmitry Safonov

[permalink] [raw]
Subject: Re: [RFC v2 1/7] mm: Provide generic VDSO unmap and remap functions

Hi Christopher,

by this moment I got another patch for this. I hope, you don't mind
if I send it concurrently. I haven't sent it yet as I was testing it in qemu.

Thanks,
Dmitry

2016-11-02 00:23:10

by Christopher Covington

[permalink] [raw]
Subject: Re: [RFC v2 1/7] mm: Provide generic VDSO unmap and remap functions



On November 1, 2016 11:23:54 AM MDT, Dmitry Safonov <[email protected]> wrote:
>Hi Christopher,
>
> by this moment I got another patch for this. I hope, you don't mind
>if I send it concurrently. I haven't sent it yet as I was testing it in
> qemu.

Please do, that'd be great.

Thanks,
Cov

--
Qualcomm Datacenter Technologies, Inc. as an affiliate of Qualcomm Technologies, Inc.
Qualcomm Technologies, Inc. is a member of the Code Aurora Forum, a Linux Foundation Collaborative Project.

Sent from my Snapdragon powered Android device with K-9 Mail. Please excuse my brevity.

2016-11-04 04:59:46

by Michael Ellerman

[permalink] [raw]
Subject: Re: [RFC v2 6/7] mm/powerpc: Use generic VDSO remap and unmap functions

Christopher Covington <[email protected]> writes:

> The PowerPC VDSO remap and unmap code was copied to a generic location,
> only modifying the variable name expected in mm->context (vdso instead of
> vdso_base) to match most other architectures. Having adopted this generic
> naming, drop the code in arch/powerpc and use the generic version.
>
> Signed-off-by: Christopher Covington <[email protected]>
> ---
> arch/powerpc/Kconfig | 1 +
> arch/powerpc/include/asm/Kbuild | 1 +
> arch/powerpc/include/asm/mm-arch-hooks.h | 28 -------------------------
> arch/powerpc/include/asm/mmu_context.h | 35 +-------------------------------
> 4 files changed, 3 insertions(+), 62 deletions(-)
> delete mode 100644 arch/powerpc/include/asm/mm-arch-hooks.h

This looks OK.

Have you tested it on powerpc? I could but I don't know how to actually
trigger these paths, I assume I need a CRIU setup?

Can you flip the subject to "powerpc/mm: ...".

cheers

2016-11-04 20:31:18

by Will Deacon

[permalink] [raw]
Subject: Re: [RFC v2 5/7] powerpc: Rename context.vdso_base to context.vdso

[fixing akpm's email address]

On Fri, Nov 04, 2016 at 03:58:22PM +1100, Michael Ellerman wrote:
> Christopher Covington <[email protected]> writes:
>
> > Checkpoint/Restore In Userspace (CRIU) needs to be able to unmap and remap
> > the VDSO to successfully checkpoint and restore applications in the face of
> > changing VDSO addresses due to Address Space Layout Randomization (ASLR,
> > randmaps). x86 and PowerPC have had architecture-specific code to support
> > this. In order to expand the architectures that support this without
> > unnecessary duplication of code, a generic version based on the PowerPC code
> > was created. It differs slightly, based on the results of an informal
> > survey of all architectures that indicated
> >
> > unsigned long vdso;
> >
> > is popular (and it's also concise). Therefore, change the variable name in
> > powerpc from mm->context.vdso_base to mm->context.vdso.
> >
> > Signed-off-by: Christopher Covington <[email protected]>
> > ---
> > arch/powerpc/include/asm/book3s/32/mmu-hash.h | 2 +-
> > arch/powerpc/include/asm/book3s/64/mmu.h | 2 +-
> > arch/powerpc/include/asm/mm-arch-hooks.h | 6 +++---
> > arch/powerpc/include/asm/mmu-40x.h | 2 +-
> > arch/powerpc/include/asm/mmu-44x.h | 2 +-
> > arch/powerpc/include/asm/mmu-8xx.h | 2 +-
> > arch/powerpc/include/asm/mmu-book3e.h | 2 +-
> > arch/powerpc/include/asm/mmu_context.h | 4 ++--
> > arch/powerpc/include/asm/vdso.h | 2 +-
> > arch/powerpc/include/uapi/asm/elf.h | 2 +-
> > arch/powerpc/kernel/signal_32.c | 8 ++++----
> > arch/powerpc/kernel/signal_64.c | 4 ++--
> > arch/powerpc/kernel/vdso.c | 8 ++++----
> > arch/powerpc/perf/callchain.c | 12 ++++++------
> > 14 files changed, 29 insertions(+), 29 deletions(-)
>
> This is kind of annoying, but I guess it's worth doing.
>
> It's going to conflict like hell though. Who were you thinking would
> merge this series? I think it should go via Andrew Morton's tree, as
> that way if we get bad conflicts we can pull it out and redo it.

The other thing you can do is generate the patch towards the end of the
merge window and send it as a separate pull request. The disadvantage of
that is that it can't spend any time in -next, but that might be ok for a
mechanical rename.

Will

2016-11-07 08:01:11

by Michael Ellerman

[permalink] [raw]
Subject: Re: [RFC v2 5/7] powerpc: Rename context.vdso_base to context.vdso

Will Deacon <[email protected]> writes:
> On Fri, Nov 04, 2016 at 03:58:22PM +1100, Michael Ellerman wrote:
>> Christopher Covington <[email protected]> writes:
>> > arch/powerpc/include/asm/book3s/32/mmu-hash.h | 2 +-
>> > arch/powerpc/include/asm/book3s/64/mmu.h | 2 +-
>> > arch/powerpc/include/asm/mm-arch-hooks.h | 6 +++---
>> > arch/powerpc/include/asm/mmu-40x.h | 2 +-
>> > arch/powerpc/include/asm/mmu-44x.h | 2 +-
>> > arch/powerpc/include/asm/mmu-8xx.h | 2 +-
>> > arch/powerpc/include/asm/mmu-book3e.h | 2 +-
>> > arch/powerpc/include/asm/mmu_context.h | 4 ++--
>> > arch/powerpc/include/asm/vdso.h | 2 +-
>> > arch/powerpc/include/uapi/asm/elf.h | 2 +-
>> > arch/powerpc/kernel/signal_32.c | 8 ++++----
>> > arch/powerpc/kernel/signal_64.c | 4 ++--
>> > arch/powerpc/kernel/vdso.c | 8 ++++----
>> > arch/powerpc/perf/callchain.c | 12 ++++++------
>> > 14 files changed, 29 insertions(+), 29 deletions(-)
>>
>> This is kind of annoying, but I guess it's worth doing.
>>
>> It's going to conflict like hell though. Who were you thinking would
>> merge this series? I think it should go via Andrew Morton's tree, as
>> that way if we get bad conflicts we can pull it out and redo it.
>
> The other thing you can do is generate the patch towards the end of the
> merge window and send it as a separate pull request. The disadvantage of
> that is that it can't spend any time in -next, but that might be ok for a
> mechanical rename.

True. Though in this case it's a mechanical rename that then allows us
to use the generic code, so I'd prefer we had some -next coverage on the
latter.

The other other option would be to wrap all uses of the arch value in a
macro (or actually two probably, one a getter one a setter). That would
then allow arches to use the generic code regardless of the name and
type of their mm->context.vdso_whatever.

That would allow the basic series to go in, and then each arch could do
a series later that switches it to the "standard" name and type.

cheers

2016-11-07 20:21:52

by Laurent Dufour

[permalink] [raw]
Subject: Re: [RFC v2 6/7] mm/powerpc: Use generic VDSO remap and unmap functions

On 04/11/2016 05:59, Michael Ellerman wrote:
> Christopher Covington <[email protected]> writes:
>
>> The PowerPC VDSO remap and unmap code was copied to a generic location,
>> only modifying the variable name expected in mm->context (vdso instead of
>> vdso_base) to match most other architectures. Having adopted this generic
>> naming, drop the code in arch/powerpc and use the generic version.
>>
>> Signed-off-by: Christopher Covington <[email protected]>
>> ---
>> arch/powerpc/Kconfig | 1 +
>> arch/powerpc/include/asm/Kbuild | 1 +
>> arch/powerpc/include/asm/mm-arch-hooks.h | 28 -------------------------
>> arch/powerpc/include/asm/mmu_context.h | 35 +-------------------------------
>> 4 files changed, 3 insertions(+), 62 deletions(-)
>> delete mode 100644 arch/powerpc/include/asm/mm-arch-hooks.h
>
> This looks OK.
>
> Have you tested it on powerpc? I could but I don't know how to actually
> trigger these paths, I assume I need a CRIU setup?

FWIW, tested on ppc64le using a sample test process moving its VDSO and
then catching a signal on 4.9-rc4 and using CRIU on top of 4.8 with
sightly changes to due minor upstream changes.

Reviewed-by: Laurent Dufour <[email protected]>
Tested-by: Laurent Dufour <[email protected]>

2016-11-07 23:52:03

by Michael Ellerman

[permalink] [raw]
Subject: Re: [RFC v2 6/7] mm/powerpc: Use generic VDSO remap and unmap functions

Laurent Dufour <[email protected]> writes:

> On 04/11/2016 05:59, Michael Ellerman wrote:
>> Christopher Covington <[email protected]> writes:
>>
>>> The PowerPC VDSO remap and unmap code was copied to a generic location,
>>> only modifying the variable name expected in mm->context (vdso instead of
>>> vdso_base) to match most other architectures. Having adopted this generic
>>> naming, drop the code in arch/powerpc and use the generic version.
>>>
>>> Signed-off-by: Christopher Covington <[email protected]>
>>> ---
>>> arch/powerpc/Kconfig | 1 +
>>> arch/powerpc/include/asm/Kbuild | 1 +
>>> arch/powerpc/include/asm/mm-arch-hooks.h | 28 -------------------------
>>> arch/powerpc/include/asm/mmu_context.h | 35 +-------------------------------
>>> 4 files changed, 3 insertions(+), 62 deletions(-)
>>> delete mode 100644 arch/powerpc/include/asm/mm-arch-hooks.h
>>
>> This looks OK.
>>
>> Have you tested it on powerpc? I could but I don't know how to actually
>> trigger these paths, I assume I need a CRIU setup?
>
> FWIW, tested on ppc64le using a sample test process moving its VDSO and
> then catching a signal on 4.9-rc4 and using CRIU on top of 4.8 with
> sightly changes to due minor upstream changes.
>
> Reviewed-by: Laurent Dufour <[email protected]>
> Tested-by: Laurent Dufour <[email protected]>

Thanks, in that case:

Acked-by: Michael Ellerman <[email protected]>

cheers