Hello,
This patch series implements a new syscall, process_memwatch. Currently, only the support to watch soft-dirty PTE bit is added. This syscall is generic to watch the memory of the process. There is enough room to add more operations like this to watch memory in the future.
Soft-dirty PTE bit of the memory pages can be viewed by using pagemap procfs file. The soft-dirty PTE bit for the memory in a process can be cleared by writing to the clear_refs file. This series adds features that weren't possible through the Proc FS interface. - There is no atomic get soft-dirty PTE bit status and clear operation possible. - The soft-dirty PTE bit of only a part of memory cannot be cleared.
Historically, soft-dirty PTE bit tracking has been used in the CRIU project. The Proc FS interface is enough for that as I think the process is frozen. We have the use case where we need to track the soft-dirty PTE bit for running processes. We need this tracking and clear mechanism of a region of memory while the process is running to emulate the getWriteWatch() syscall of Windows. This syscall is used by games to keep track of dirty pages and keep processing only the dirty pages. This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information.
As in the current kernel there is no way to clear a part of memory (instead of clearing the Soft-Dirty bits for the entire processi) and get+clear operation cannot be performed atomically, there are other methods to mimic this information entirely in userspace with poor performance: - The mprotect syscall and SIGSEGV handler for bookkeeping - The userfaultfd syscall with the handler for bookkeeping
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);
This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall: - Get the pages that are soft-dirty. - Clear the pages which are soft-dirty. - The optional flag to ignore the VM_SOFTDIRTY and only track per page soft-dirty PTE bit
There are two decisions which have been taken about how to get the output from the syscall. - Return offsets of the pages from the start in the vec - Stop execution when vec is filled with dirty pages These two arguments doesn't follow the mincore() philosophy where the output array corresponds to the address range in one to one fashion, hence the output buffer length isn't passed and only a flag is set if the page is present. This makes mincore() easy to use with less control. We are passing the size of the output array and putting return data consecutively which is offset of dirty pages from the start. The user can convert these offsets back into the dirty page addresses easily. Suppose, the user want to get first 10 dirty pages from a total memory of 100 pages. He'll allocate output buffer of size 10 and process_memwatch() syscall will abort after finding the 10 pages. This behaviour is needed to support Windows' getWriteWatch(). The behaviour like mincore() can be achieved by passing output buffer of 100 size. This interface can be used for any desired behaviour.
Regards, Muhammad Usama Anjum
Muhammad Usama Anjum (5): fs/proc/task_mmu: make functions global to be used in other files mm: Implement process_memwatch syscall mm: wire up process_memwatch syscall for x86 selftests: vm: add process_memwatch syscall tests mm: add process_memwatch syscall documentation
Documentation/admin-guide/mm/soft-dirty.rst | 48 +- arch/x86/entry/syscalls/syscall_32.tbl | 1 + arch/x86/entry/syscalls/syscall_64.tbl | 1 + fs/proc/task_mmu.c | 84 +-- include/linux/mm_inline.h | 99 +++ include/linux/syscalls.h | 3 +- include/uapi/asm-generic/unistd.h | 5 +- include/uapi/linux/memwatch.h | 12 + kernel/sys_ni.c | 1 + mm/Makefile | 2 +- mm/memwatch.c | 285 ++++++++ tools/include/uapi/asm-generic/unistd.h | 5 +- .../arch/x86/entry/syscalls/syscall_64.tbl | 1 + tools/testing/selftests/vm/.gitignore | 1 + tools/testing/selftests/vm/Makefile | 2 + tools/testing/selftests/vm/memwatch_test.c | 635 ++++++++++++++++++ 16 files changed, 1098 insertions(+), 87 deletions(-) create mode 100644 include/uapi/linux/memwatch.h create mode 100644 mm/memwatch.c create mode 100644 tools/testing/selftests/vm/memwatch_test.c
Update the clear_soft_dirty() and clear_soft_dirty_pmd() to optionally clear and return the status if page is dirty.
Signed-off-by: Muhammad Usama Anjum usama.anjum@collabora.com --- fs/proc/task_mmu.c | 84 +-------------------------------- include/linux/mm_inline.h | 99 +++++++++++++++++++++++++++++++++++++++ 2 files changed, 101 insertions(+), 82 deletions(-)
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c index f8cd58846a28..94d5761cc369 100644 --- a/fs/proc/task_mmu.c +++ b/fs/proc/task_mmu.c @@ -1076,86 +1076,6 @@ struct clear_refs_private { enum clear_refs_types type; };
-#ifdef CONFIG_MEM_SOFT_DIRTY - -static inline bool pte_is_pinned(struct vm_area_struct *vma, unsigned long addr, pte_t pte) -{ - struct page *page; - - if (!pte_write(pte)) - return false; - if (!is_cow_mapping(vma->vm_flags)) - return false; - if (likely(!test_bit(MMF_HAS_PINNED, &vma->vm_mm->flags))) - return false; - page = vm_normal_page(vma, addr, pte); - if (!page) - return false; - return page_maybe_dma_pinned(page); -} - -static inline void clear_soft_dirty(struct vm_area_struct *vma, - unsigned long addr, pte_t *pte) -{ - /* - * The soft-dirty tracker uses #PF-s to catch writes - * to pages, so write-protect the pte as well. See the - * Documentation/admin-guide/mm/soft-dirty.rst for full description - * of how soft-dirty works. - */ - pte_t ptent = *pte; - - if (pte_present(ptent)) { - pte_t old_pte; - - if (pte_is_pinned(vma, addr, ptent)) - return; - old_pte = ptep_modify_prot_start(vma, addr, pte); - ptent = pte_wrprotect(old_pte); - ptent = pte_clear_soft_dirty(ptent); - ptep_modify_prot_commit(vma, addr, pte, old_pte, ptent); - } else if (is_swap_pte(ptent)) { - ptent = pte_swp_clear_soft_dirty(ptent); - set_pte_at(vma->vm_mm, addr, pte, ptent); - } -} -#else -static inline void clear_soft_dirty(struct vm_area_struct *vma, - unsigned long addr, pte_t *pte) -{ -} -#endif - -#if defined(CONFIG_MEM_SOFT_DIRTY) && defined(CONFIG_TRANSPARENT_HUGEPAGE) -static inline void clear_soft_dirty_pmd(struct vm_area_struct *vma, - unsigned long addr, pmd_t *pmdp) -{ - pmd_t old, pmd = *pmdp; - - if (pmd_present(pmd)) { - /* See comment in change_huge_pmd() */ - old = pmdp_invalidate(vma, addr, pmdp); - if (pmd_dirty(old)) - pmd = pmd_mkdirty(pmd); - if (pmd_young(old)) - pmd = pmd_mkyoung(pmd); - - pmd = pmd_wrprotect(pmd); - pmd = pmd_clear_soft_dirty(pmd); - - set_pmd_at(vma->vm_mm, addr, pmdp, pmd); - } else if (is_migration_entry(pmd_to_swp_entry(pmd))) { - pmd = pmd_swp_clear_soft_dirty(pmd); - set_pmd_at(vma->vm_mm, addr, pmdp, pmd); - } -} -#else -static inline void clear_soft_dirty_pmd(struct vm_area_struct *vma, - unsigned long addr, pmd_t *pmdp) -{ -} -#endif - static int clear_refs_pte_range(pmd_t *pmd, unsigned long addr, unsigned long end, struct mm_walk *walk) { @@ -1168,7 +1088,7 @@ static int clear_refs_pte_range(pmd_t *pmd, unsigned long addr, ptl = pmd_trans_huge_lock(pmd, vma); if (ptl) { if (cp->type == CLEAR_REFS_SOFT_DIRTY) { - clear_soft_dirty_pmd(vma, addr, pmd); + check_soft_dirty_pmd(vma, addr, pmd, true); goto out; }
@@ -1194,7 +1114,7 @@ static int clear_refs_pte_range(pmd_t *pmd, unsigned long addr, ptent = *pte;
if (cp->type == CLEAR_REFS_SOFT_DIRTY) { - clear_soft_dirty(vma, addr, pte); + check_soft_dirty(vma, addr, pte, true); continue; }
diff --git a/include/linux/mm_inline.h b/include/linux/mm_inline.h index 7b25b53c474a..65014c347a94 100644 --- a/include/linux/mm_inline.h +++ b/include/linux/mm_inline.h @@ -360,4 +360,103 @@ pte_install_uffd_wp_if_needed(struct vm_area_struct *vma, unsigned long addr, #endif }
+#ifdef CONFIG_MEM_SOFT_DIRTY +static inline bool pte_is_pinned(struct vm_area_struct *vma, unsigned long addr, pte_t pte) +{ + struct page *page; + + if (!pte_write(pte)) + return false; + if (!is_cow_mapping(vma->vm_flags)) + return false; + if (likely(!test_bit(MMF_HAS_PINNED, &vma->vm_mm->flags))) + return false; + page = vm_normal_page(vma, addr, pte); + if (!page) + return false; + return page_maybe_dma_pinned(page); +} + +static inline bool check_soft_dirty(struct vm_area_struct *vma, + unsigned long addr, pte_t *pte, bool clear) +{ + /* + * The soft-dirty tracker uses #PF-s to catch writes + * to pages, so write-protect the pte as well. See the + * Documentation/admin-guide/mm/soft-dirty.rst for full description + * of how soft-dirty works. + */ + pte_t ptent = *pte; + int dirty = 0; + + if (pte_present(ptent)) { + pte_t old_pte; + + dirty = pte_soft_dirty(ptent); + + if (dirty && clear && !pte_is_pinned(vma, addr, ptent)) { + old_pte = ptep_modify_prot_start(vma, addr, pte); + ptent = pte_wrprotect(old_pte); + ptent = pte_clear_soft_dirty(ptent); + ptep_modify_prot_commit(vma, addr, pte, old_pte, ptent); + } + } else if (is_swap_pte(ptent)) { + dirty = pte_swp_soft_dirty(ptent); + + if (dirty && clear) { + ptent = pte_swp_clear_soft_dirty(ptent); + set_pte_at(vma->vm_mm, addr, pte, ptent); + } + } + + return !!dirty; +} +#else +static inline bool check_soft_dirty(struct vm_area_struct *vma, + unsigned long addr, pte_t *pte, bool clear) +{ + return false; +} +#endif + +#if defined(CONFIG_MEM_SOFT_DIRTY) && defined(CONFIG_TRANSPARENT_HUGEPAGE) +static inline bool check_soft_dirty_pmd(struct vm_area_struct *vma, + unsigned long addr, pmd_t *pmdp, bool clear) +{ + pmd_t old, pmd = *pmdp; + int dirty = 0; + + if (pmd_present(pmd)) { + dirty = pmd_soft_dirty(pmd); + if (dirty && clear) { + /* See comment in change_huge_pmd() */ + old = pmdp_invalidate(vma, addr, pmdp); + if (pmd_dirty(old)) + pmd = pmd_mkdirty(pmd); + if (pmd_young(old)) + pmd = pmd_mkyoung(pmd); + + pmd = pmd_wrprotect(pmd); + pmd = pmd_clear_soft_dirty(pmd); + + set_pmd_at(vma->vm_mm, addr, pmdp, pmd); + } + } else if (is_migration_entry(pmd_to_swp_entry(pmd))) { + dirty = pmd_swp_soft_dirty(pmd); + + if (dirty && clear) { + pmd = pmd_swp_clear_soft_dirty(pmd); + set_pmd_at(vma->vm_mm, addr, pmdp, pmd); + } + } + return !!dirty; +} +#else +static inline bool check_soft_dirty_pmd(struct vm_area_struct *vma, + unsigned long addr, pmd_t *pmdp, bool clear) +{ + return false; +} +#endif + #endif
This syscall can be used to watch the process's memory and perform atomic operations which aren't possible through procfs. Two operations have been implemented. MEMWATCH_SD_GET is used to get the soft dirty pages. MEMWATCH_SD_CLEAR clears the soft dirty bit from dirty pages. MEMWATCH_SD_IGNORE_VMA can be specified to ignore VMA dirty flags. These operations can be used collectively in one operation as well.
NAME process_memwatch - get process's memory information
SYNOPSIS #include <linux/memwatch.h> /* Definition of MEMWATCH_* constants */
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);
Note: Glibc does not provide a wrapper for this system call; call it using syscall(2).
DESCRIPTION process_memwatch() system call is used to get information about the memory of the process.
Arguments pidfd specifies the pidfd of process whose memory needs to be watched. The calling process must have PTRACE_MODE_ATTACH_FS‐ CREDS capabilities over the process whose pidfd has been specified. It can be zero which means that the process wants to watch its own memory. The operation is determined by flags. The start argument must be a multiple of the system page size. The len argument need not be a multiple of the page size, but since the information is returned for the whole pages, len is effectively rounded up to the next multi‐ ple of the page size.
vec is an output array in which the offsets of the pages are returned. Offset is calculated from start address. User lets the kernel know about the size of the vec by passing size in vec_len. The system call returns when the whole range has been searched or vec is completely filled. The whole range isn't cleared if vec fills up completely.
Operations The flags argument specifies the operation to be performed. The MEMWATCH_SD_GET and MEMWATCH_SD_CLEAR operations can be used separately or together to perform MEMWATCH_SD_GET and MEMWATCH_SD_CLEAR atomically as one operation.
MEMWATCH_SD_GET Get the page offsets which are soft dirty.
MEMWATCH_SD_CLEAR Clear the pages which are soft dirty.
MEMWATCH_SD_NO_REUSED_REGIONS This optional flag can be specified in combination with other flags. VM_SOFTDIRTY is ignored for the VMAs for performance reasons. This flag shows only those pages dirty which have been written by the user ex‐ plicitly. All new allocations are not be returned as dirty.
RETURN VALUE The 0 or positive value is returned on success. Positive value when returned shows the number of dirty pages filled in vec. In the event of an error (and assuming that process_memwatch() was invoked via syscall(2)), all opera‐ tions return -1 and set errno to indicate the error.
ERRORS EINVAL invalid arguments.
ESRCH Cannot access the process.
EIO I/O error.
This is based on a patch from Gabriel Krisman Bertazi.
Signed-off-by: Muhammad Usama Anjum usama.anjum@collabora.com --- include/uapi/linux/memwatch.h | 12 ++ mm/Makefile | 2 +- mm/memwatch.c | 285 ++++++++++++++++++++++++++++++++++ 3 files changed, 298 insertions(+), 1 deletion(-) create mode 100644 include/uapi/linux/memwatch.h create mode 100644 mm/memwatch.c
diff --git a/include/uapi/linux/memwatch.h b/include/uapi/linux/memwatch.h new file mode 100644 index 000000000000..7e86ffdc10f5 --- /dev/null +++ b/include/uapi/linux/memwatch.h @@ -0,0 +1,12 @@ +/* SPDX-License-Identifier: GPL-2.0 WITH Linux-syscall-note */ + +#ifndef _MEMWATCH_H +#define _MEMWATCH_H + +/* memwatch operations */ +#define MEMWATCH_SD_GET 0x1 +#define MEMWATCH_SD_CLEAR 0x2 +#define MEMWATCH_SD_NO_REUSED_REGIONS 0x4 + +#endif + diff --git a/mm/Makefile b/mm/Makefile index 8083fa85a348..aa72e4ced1f3 100644 --- a/mm/Makefile +++ b/mm/Makefile @@ -37,7 +37,7 @@ CFLAGS_init-mm.o += $(call cc-disable-warning, override-init) CFLAGS_init-mm.o += $(call cc-disable-warning, initializer-overrides)
mmu-y := nommu.o -mmu-$(CONFIG_MMU) := highmem.o memory.o mincore.o \ +mmu-$(CONFIG_MMU) := highmem.o memory.o memwatch.o mincore.o \ mlock.o mmap.o mmu_gather.o mprotect.o mremap.o \ msync.o page_vma_mapped.o pagewalk.o \ pgtable-generic.o rmap.o vmalloc.o diff --git a/mm/memwatch.c b/mm/memwatch.c new file mode 100644 index 000000000000..9be09bc431d2 --- /dev/null +++ b/mm/memwatch.c @@ -0,0 +1,285 @@ +// SPDX-License-Identifier: GPL-2.0 +/* + * Copyright 2020 Collabora Ltd. + */ +#include <linux/pagewalk.h> +#include <linux/vmalloc.h> +#include <linux/syscalls.h> +#include <asm/tlb.h> +#include <asm/tlbflush.h> +#include <linux/sched/mm.h> +#include <linux/mm_inline.h> +#include <uapi/linux/memwatch.h> +#include <uapi/asm-generic/errno-base.h> +#include <linux/compat.h> +#include <linux/minmax.h> + +#ifdef CONFIG_MEM_SOFT_DIRTY +#define MEMWATCH_SD_OPS_MASK (MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR | \ + MEMWATCH_SD_NO_REUSED_REGIONS) + +struct memwatch_sd_private { + unsigned long start; + unsigned int flags; + unsigned int index; + unsigned int vec_len; + unsigned long *vec; +}; + +static int memwatch_pmd_entry(pmd_t *pmd, unsigned long addr, + unsigned long end, struct mm_walk *walk) +{ + struct memwatch_sd_private *p = walk->private; + struct vm_area_struct *vma = walk->vma; + unsigned long start = addr; + spinlock_t *ptl; + pte_t *pte; + int dirty; + bool dirty_vma = (p->flags & MEMWATCH_SD_NO_REUSED_REGIONS) ? 0 : + (vma->vm_flags & VM_SOFTDIRTY); + + end = min(end, walk->vma->vm_end); + ptl = pmd_trans_huge_lock(pmd, vma); + if (ptl) { + if (dirty_vma || check_soft_dirty_pmd(vma, addr, pmd, false)) { + /* + * Break huge page into small pages if operation needs to be performed is + * on a portion of the huge page or the return buffer cannot store complete + * data. Then process this PMD as having normal pages. + */ + if (((p->flags & MEMWATCH_SD_CLEAR) && (end - addr < HPAGE_SIZE)) || + ((p->flags & MEMWATCH_SD_GET) && + (p->index + HPAGE_SIZE/PAGE_SIZE > p->vec_len))) { + spin_unlock(ptl); + split_huge_pmd(vma, pmd, addr); + goto process_pages; + } else { + dirty = check_soft_dirty_pmd(vma, addr, pmd, + p->flags & MEMWATCH_SD_CLEAR); + if ((p->flags & MEMWATCH_SD_GET) && (dirty_vma || dirty)) { + for (; addr != end && p->index < p->vec_len; + addr += PAGE_SIZE) + p->vec[p->index++] = addr - p->start; + } + } + } + spin_unlock(ptl); + return 0; + } + +process_pages: + if (pmd_trans_unstable(pmd)) + return 0; + + pte = pte_offset_map_lock(vma->vm_mm, pmd, addr, &ptl); + for (; addr != end; pte++, addr += PAGE_SIZE) { + dirty = check_soft_dirty(vma, addr, pte, p->flags & MEMWATCH_SD_CLEAR); + + if ((p->flags & MEMWATCH_SD_GET) && (dirty_vma || dirty)) { + p->vec[p->index++] = addr - p->start; + WARN_ON(p->index > p->vec_len); + } + } + pte_unmap_unlock(pte - 1, ptl); + cond_resched(); + + if (p->flags & MEMWATCH_SD_CLEAR) + flush_tlb_mm_range(vma->vm_mm, start, end, PAGE_SHIFT, false); + + return 0; +} + +static int memwatch_pte_hole(unsigned long addr, unsigned long end, int depth, + struct mm_walk *walk) +{ + struct memwatch_sd_private *p = walk->private; + struct vm_area_struct *vma = walk->vma; + + if (p->flags & MEMWATCH_SD_NO_REUSED_REGIONS) + return 0; + + if (vma && (vma->vm_flags & VM_SOFTDIRTY) && (p->flags & MEMWATCH_SD_GET)) { + for (; addr != end && p->index < p->vec_len; addr += PAGE_SIZE) + p->vec[p->index++] = addr - p->start; + } + + return 0; +} + +static int memwatch_pre_vma(unsigned long start, unsigned long end, struct mm_walk *walk) +{ + struct memwatch_sd_private *p = walk->private; + struct vm_area_struct *vma = walk->vma; + int ret; + unsigned long end_cut = end; + + if (p->flags & MEMWATCH_SD_NO_REUSED_REGIONS) + return 0; + + if ((p->flags & MEMWATCH_SD_CLEAR) && (vma->vm_flags & VM_SOFTDIRTY)) { + if (vma->vm_start < start) { + ret = split_vma(vma->vm_mm, vma, start, 1); + if (ret) + return ret; + } + + if (p->flags & MEMWATCH_SD_GET) + end_cut = min(start + p->vec_len * PAGE_SIZE, end); + + if (vma->vm_end > end_cut) { + ret = split_vma(vma->vm_mm, vma, end_cut, 0); + if (ret) + return ret; + } + } + + return 0; +} + +static void memwatch_post_vma(struct mm_walk *walk) +{ + struct memwatch_sd_private *p = walk->private; + struct vm_area_struct *vma = walk->vma; + + if (p->flags & MEMWATCH_SD_NO_REUSED_REGIONS) + return; + + if ((p->flags & MEMWATCH_SD_CLEAR) && (vma->vm_flags & VM_SOFTDIRTY)) { + vma->vm_flags &= ~VM_SOFTDIRTY; + vma_set_page_prot(vma); + } +} + +static int memwatch_pmd_test_walk(unsigned long start, unsigned long end, + struct mm_walk *walk) +{ + struct memwatch_sd_private *p = walk->private; + struct vm_area_struct *vma = walk->vma; + + if ((p->flags & MEMWATCH_SD_GET) && (p->index == p->vec_len)) + return -1; + + if (vma->vm_flags & VM_PFNMAP) + return 1; + + return 0; +} + +static const struct mm_walk_ops memwatch_ops = { + .test_walk = memwatch_pmd_test_walk, + .pre_vma = memwatch_pre_vma, + .pmd_entry = memwatch_pmd_entry, + .pte_hole = memwatch_pte_hole, + .post_vma = memwatch_post_vma, +}; + +static long do_process_memwatch(int pidfd, void __user *start_addr, int len, + unsigned int flags, loff_t __user *vec, int vec_len) +{ + struct memwatch_sd_private watch; + struct mmu_notifier_range range; + unsigned long start, end; + struct task_struct *task; + struct mm_struct *mm; + unsigned int f_flags; + int ret; + + start = (unsigned long)untagged_addr(start_addr); + if ((!IS_ALIGNED(start, PAGE_SIZE)) || !access_ok((void __user *)start, len)) + return -EINVAL; + + if ((flags == 0) || (flags == MEMWATCH_SD_NO_REUSED_REGIONS) || + (flags & ~MEMWATCH_SD_OPS_MASK)) + return -EINVAL; + + if ((flags & MEMWATCH_SD_GET) && ((vec_len == 0) || (!vec) || + !access_ok(vec, vec_len))) + return -EINVAL; + + end = start + len; + watch.start = start; + watch.flags = flags; + watch.index = 0; + watch.vec_len = vec_len; + + if (pidfd) { + task = pidfd_get_task(pidfd, &f_flags); + if (IS_ERR(task)) + return PTR_ERR(task); + } else { + task = current; + } + + if (flags & MEMWATCH_SD_GET) { + watch.vec = vzalloc(vec_len * sizeof(loff_t)); + if (!watch.vec) { + ret = -ENOMEM; + goto put_task; + } + } + + mm = mm_access(task, PTRACE_MODE_ATTACH_FSCREDS); + if (IS_ERR_OR_NULL(mm)) { + ret = mm ? PTR_ERR(mm) : -ESRCH; + goto free_watch; + } + + if (flags & MEMWATCH_SD_CLEAR) { + mmap_write_lock(mm); + + mmu_notifier_range_init(&range, MMU_NOTIFY_SOFT_DIRTY, 0, NULL, + mm, start, end); + mmu_notifier_invalidate_range_start(&range); + inc_tlb_flush_pending(mm); + } else { + mmap_read_lock(mm); + } + + ret = walk_page_range(mm, start, end, &memwatch_ops, &watch); + + if (flags & MEMWATCH_SD_CLEAR) { + mmu_notifier_invalidate_range_end(&range); + dec_tlb_flush_pending(mm); + + mmap_write_unlock(mm); + } else { + mmap_read_unlock(mm); + } + + mmput(mm); + + if (ret < 0) + goto free_watch; + + if (flags & MEMWATCH_SD_GET) { + ret = copy_to_user(vec, watch.vec, watch.index * sizeof(loff_t)); + if (ret) { + ret = -EIO; + goto free_watch; + } + ret = watch.index; + } else { + ret = 0; + } + +free_watch: + if (flags & MEMWATCH_SD_GET) + vfree(watch.vec); +put_task: + if (pidfd) + put_task_struct(task); + + return ret; +} +#endif + +SYSCALL_DEFINE6(process_memwatch, int, pidfd, void __user*, start, + int, len, unsigned int, flags, loff_t __user *, vec, int, vec_len) +{ + int ret = -EPERM; + +#ifdef CONFIG_MEM_SOFT_DIRTY + ret = do_process_memwatch(pidfd, start, len, flags, vec, vec_len); +#endif + return ret; +}
Wire up syscall entry point for both i386 and x86_64 architectures.
Signed-off-by: Muhammad Usama Anjum usama.anjum@collabora.com --- arch/x86/entry/syscalls/syscall_32.tbl | 1 + arch/x86/entry/syscalls/syscall_64.tbl | 1 + include/linux/syscalls.h | 3 ++- include/uapi/asm-generic/unistd.h | 5 ++++- kernel/sys_ni.c | 1 + tools/include/uapi/asm-generic/unistd.h | 5 ++++- tools/perf/arch/x86/entry/syscalls/syscall_64.tbl | 1 + 7 files changed, 14 insertions(+), 3 deletions(-)
diff --git a/arch/x86/entry/syscalls/syscall_32.tbl b/arch/x86/entry/syscalls/syscall_32.tbl index 320480a8db4f..601d33909880 100644 --- a/arch/x86/entry/syscalls/syscall_32.tbl +++ b/arch/x86/entry/syscalls/syscall_32.tbl @@ -455,3 +455,4 @@ 448 i386 process_mrelease sys_process_mrelease 449 i386 futex_waitv sys_futex_waitv 450 i386 set_mempolicy_home_node sys_set_mempolicy_home_node +451 i386 process_memwatch sys_process_memwatch diff --git a/arch/x86/entry/syscalls/syscall_64.tbl b/arch/x86/entry/syscalls/syscall_64.tbl index c84d12608cd2..3bddea588ce7 100644 --- a/arch/x86/entry/syscalls/syscall_64.tbl +++ b/arch/x86/entry/syscalls/syscall_64.tbl @@ -372,6 +372,7 @@ 448 common process_mrelease sys_process_mrelease 449 common futex_waitv sys_futex_waitv 450 common set_mempolicy_home_node sys_set_mempolicy_home_node +451 common process_memwatch sys_process_memwatch
# # Due to a historical design error, certain syscalls are numbered differently diff --git a/include/linux/syscalls.h b/include/linux/syscalls.h index a34b0f9a9972..efa240510e4c 100644 --- a/include/linux/syscalls.h +++ b/include/linux/syscalls.h @@ -939,7 +939,6 @@ asmlinkage long sys_move_pages(pid_t pid, unsigned long nr_pages, const int __user *nodes, int __user *status, int flags); - asmlinkage long sys_rt_tgsigqueueinfo(pid_t tgid, pid_t pid, int sig, siginfo_t __user *uinfo); asmlinkage long sys_perf_event_open( @@ -1056,6 +1055,8 @@ asmlinkage long sys_memfd_secret(unsigned int flags); asmlinkage long sys_set_mempolicy_home_node(unsigned long start, unsigned long len, unsigned long home_node, unsigned long flags); +asmlinkage long sys_process_memwatch(int pidfd, void __user *addr, int len, + unsigned int flags, loff_t __user *vec, int vec_len);
/* * Architecture-specific system calls diff --git a/include/uapi/asm-generic/unistd.h b/include/uapi/asm-generic/unistd.h index 45fa180cc56a..805a8d5cf0c4 100644 --- a/include/uapi/asm-generic/unistd.h +++ b/include/uapi/asm-generic/unistd.h @@ -886,8 +886,11 @@ __SYSCALL(__NR_futex_waitv, sys_futex_waitv) #define __NR_set_mempolicy_home_node 450 __SYSCALL(__NR_set_mempolicy_home_node, sys_set_mempolicy_home_node)
+#define __NR_process_memwatch 451 +__SC_COMP(__NR_process_memwatch, sys_process_memwatch, compat_sys_process_memwatch) + #undef __NR_syscalls -#define __NR_syscalls 451 +#define __NR_syscalls 452
/* * 32 bit systems traditionally used different diff --git a/kernel/sys_ni.c b/kernel/sys_ni.c index a492f159624f..74f31317481a 100644 --- a/kernel/sys_ni.c +++ b/kernel/sys_ni.c @@ -298,6 +298,7 @@ COND_SYSCALL(set_mempolicy); COND_SYSCALL(migrate_pages); COND_SYSCALL(move_pages); COND_SYSCALL(set_mempolicy_home_node); +COND_SYSCALL(process_memwatch);
COND_SYSCALL(perf_event_open); COND_SYSCALL(accept4); diff --git a/tools/include/uapi/asm-generic/unistd.h b/tools/include/uapi/asm-generic/unistd.h index 45fa180cc56a..805a8d5cf0c4 100644 --- a/tools/include/uapi/asm-generic/unistd.h +++ b/tools/include/uapi/asm-generic/unistd.h @@ -886,8 +886,11 @@ __SYSCALL(__NR_futex_waitv, sys_futex_waitv) #define __NR_set_mempolicy_home_node 450 __SYSCALL(__NR_set_mempolicy_home_node, sys_set_mempolicy_home_node)
+#define __NR_process_memwatch 451 +__SC_COMP(__NR_process_memwatch, sys_process_memwatch, compat_sys_process_memwatch) + #undef __NR_syscalls -#define __NR_syscalls 451 +#define __NR_syscalls 452
/* * 32 bit systems traditionally used different diff --git a/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl b/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl index c84d12608cd2..3bddea588ce7 100644 --- a/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl +++ b/tools/perf/arch/x86/entry/syscalls/syscall_64.tbl @@ -372,6 +372,7 @@ 448 common process_mrelease sys_process_mrelease 449 common futex_waitv sys_futex_waitv 450 common set_mempolicy_home_node sys_set_mempolicy_home_node +451 common process_memwatch sys_process_memwatch
# # Due to a historical design error, certain syscalls are numbered differently
Several unit tests and functionality tests are included.
Signed-off-by: Muhammad Usama Anjum usama.anjum@collabora.com --- TAP version 13 1..44 ok 1 sanity_tests no flag specified ok 2 sanity_tests wrong flag specified ok 3 sanity_tests mixture of correct and wrong flags ok 4 sanity_tests wrong pidfd ok 5 sanity_tests pidfd of process with over which no capabilities ok 6 sanity_tests Clear area with larger vec size ok 7 Page testing: all new pages must be soft dirty ok 8 Page testing: all pages must not be soft dirty ok 9 Page testing: all pages dirty other than first and the last one ok 10 Page testing: only middle page dirty ok 11 Page testing: only two middle pages dirty ok 12 Page testing: only get 2 dirty pages and clear them as well ok 13 Page testing: Range clear only ok 14 Large Page testing: all new pages must be soft dirty ok 15 Large Page testing: all pages must not be soft dirty ok 16 Large Page testing: all pages dirty other than first and the last one ok 17 Large Page testing: only middle page dirty ok 18 Large Page testing: only two middle pages dirty ok 19 Large Page testing: only get 2 dirty pages and clear them as well ok 20 Large Page testing: Range clear only ok 21 Huge page testing: all new pages must be soft dirty ok 22 Huge page testing: all pages must not be soft dirty ok 23 Huge page testing: all pages dirty other than first and the last one ok 24 Huge page testing: only middle page dirty ok 25 Huge page testing: only two middle pages dirty ok 26 Huge page testing: only get 2 dirty pages and clear them as well ok 27 Huge page testing: Range clear only ok 28 Performance Page testing: page isn't dirty ok 29 Performance Page testing: all pages must not be soft dirty ok 30 Performance Page testing: all pages dirty other than first and the last one ok 31 Performance Page testing: only middle page dirty ok 32 Performance Page testing: only two middle pages dirty ok 33 Performance Page testing: only get 2 dirty pages and clear them as well ok 34 Performance Page testing: Range clear only ok 35 hpage_unit_tests all new huge page must be dirty ok 36 hpage_unit_tests all the huge page must not be dirty ok 37 hpage_unit_tests all the huge page must be dirty and clear ok 38 hpage_unit_tests only middle page dirty ok 39 hpage_unit_tests clear first half of huge page ok 40 hpage_unit_tests clear first half of huge page with limited buffer ok 41 hpage_unit_tests clear second half huge page ok 42 unmapped_region_tests Get dirty pages ok 43 unmapped_region_tests Get dirty pages ok 44 Test test_simple # Totals: pass:44 fail:0 xfail:0 xpass:0 skip:0 error:0 --- tools/testing/selftests/vm/.gitignore | 1 + tools/testing/selftests/vm/Makefile | 2 + tools/testing/selftests/vm/memwatch_test.c | 635 +++++++++++++++++++++ 3 files changed, 638 insertions(+) create mode 100644 tools/testing/selftests/vm/memwatch_test.c
diff --git a/tools/testing/selftests/vm/.gitignore b/tools/testing/selftests/vm/.gitignore index 31e5eea2a9b9..462cff7e23bb 100644 --- a/tools/testing/selftests/vm/.gitignore +++ b/tools/testing/selftests/vm/.gitignore @@ -14,6 +14,7 @@ mlock2-tests mrelease_test mremap_dontunmap mremap_test +memwatch_test on-fault-limit transhuge-stress protection_keys diff --git a/tools/testing/selftests/vm/Makefile b/tools/testing/selftests/vm/Makefile index d9fa6a9ea584..65b8c94b104d 100644 --- a/tools/testing/selftests/vm/Makefile +++ b/tools/testing/selftests/vm/Makefile @@ -41,6 +41,7 @@ TEST_GEN_FILES += map_fixed_noreplace TEST_GEN_FILES += map_hugetlb TEST_GEN_FILES += map_populate TEST_GEN_FILES += memfd_secret +TEST_GEN_PROGS += memwatch_test TEST_GEN_FILES += migration TEST_GEN_FILES += mlock-random-test TEST_GEN_FILES += mlock2-tests @@ -98,6 +99,7 @@ TEST_FILES += va_128TBswitch.sh include ../lib.mk
$(OUTPUT)/madv_populate: vm_util.c +$(OUTPUT)/memwatch_test: vm_util.c $(OUTPUT)/soft-dirty: vm_util.c $(OUTPUT)/split_huge_page_test: vm_util.c
diff --git a/tools/testing/selftests/vm/memwatch_test.c b/tools/testing/selftests/vm/memwatch_test.c new file mode 100644 index 000000000000..a109eff5d807 --- /dev/null +++ b/tools/testing/selftests/vm/memwatch_test.c @@ -0,0 +1,635 @@ +// SPDX-License-Identifier: GPL-2.0 +#include <stdio.h> +#include <fcntl.h> +#include <unistd.h> +#include <string.h> +#include <sys/mman.h> +#include <errno.h> +#include <malloc.h> +#include <asm-generic/unistd.h> +#include <linux/memwatch.h> +#include "vm_util.h" +#include "../kselftest.h" +#include <linux/types.h> + +#define TEST_ITERATIONS 10000 + +static long process_memwatch(pid_t pidfd, void *start, int len, + unsigned int flags, loff_t *vec, int vec_len) +{ + return syscall(__NR_process_memwatch, pidfd, start, len, flags, vec, vec_len); +} + +int sanity_tests(int page_size) +{ + char *mem; + int mem_size, vec_size, ret; + loff_t *vec; + + /* 1. wrong operation */ + vec_size = 100; + mem_size = page_size; + + vec = malloc(sizeof(loff_t) * vec_size); + mem = mmap(NULL, mem_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, -1, 0); + if (!mem || !vec) + ksft_exit_fail_msg("error nomem\n"); + + ksft_test_result(process_memwatch(0, mem, mem_size, 0, vec, vec_size) < 0, + "%s no flag specified\n", __func__); + ksft_test_result(process_memwatch(0, mem, mem_size, 0x01000000, vec, vec_size) < 0, + "%s wrong flag specified\n", __func__); + ksft_test_result(process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET | 0xFF, + vec, vec_size) < 0, + "%s mixture of correct and wrong flags\n", __func__); + ksft_test_result(process_memwatch(-1, mem, mem_size, MEMWATCH_SD_GET, vec, vec_size) < 0, + "%s wrong pidfd\n", __func__); + ksft_test_result(process_memwatch(1, mem, mem_size, MEMWATCH_SD_GET, vec, vec_size) < 0, + "%s pidfd of process with over which no capabilities\n", __func__); + + /* 2. Clear area with larger vec size */ + ret = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR, + vec, vec_size); + ksft_test_result(ret >= 0, "%s Clear area with larger vec size\n", __func__); + + free(vec); + munmap(mem, mem_size); + return 0; +} + +void *gethugepage(int map_size) +{ + int ret; + char *map; + size_t hpage_len = read_pmd_pagesize(); + + map = memalign(hpage_len, map_size); + if (!map) + ksft_exit_fail_msg("memalign failed %d %s\n", errno, strerror(errno)); + + ret = madvise(map, map_size, MADV_HUGEPAGE); + if (ret) + ksft_exit_fail_msg("madvise failed %d %d %s\n", ret, errno, strerror(errno)); + + memset(map, 0, map_size); + + if (check_huge(map)) + return map; + + free(map); + return NULL; + +} + +int hpage_unit_tests(int page_size) +{ + char *map; + int i, ret; + size_t hpage_len = read_pmd_pagesize(); + size_t num_pages = 1; + int map_size = hpage_len * num_pages; + int vec_size = map_size/page_size; + loff_t *vec, *vec2; + + vec = malloc(sizeof(loff_t) * vec_size); + vec2 = malloc(sizeof(loff_t) * vec_size); + if (!vec || !vec2) + ksft_exit_fail_msg("malloc failed\n"); + + map = gethugepage(map_size); + if (map) { + // 1. all new huge page must be dirty + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR, + vec, vec_size); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + for (i = 0; i < vec_size; i++) + if (vec[i] != i * page_size) + break; + + ksft_test_result(i == vec_size, "%s all new huge page must be dirty\n", __func__); + + // 2. all the huge page must not be dirty + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_GET, + vec, vec_size); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + ksft_test_result(ret == 0, "%s all the huge page must not be dirty\n", __func__); + + // 3. all the huge page must be dirty and clear dirty as well + memset(map, -1, map_size); + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR, + vec, vec_size); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + for (i = 0; i < vec_size; i++) + if (vec[i] != i * page_size) + break; + + ksft_test_result(ret == vec_size && i == vec_size, + "%s all the huge page must be dirty and clear\n", __func__); + + // 4. only middle page dirty + free(map); + map = gethugepage(map_size); + clear_softdirty(); + map[vec_size/2 * page_size]++; + + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_GET, vec, vec_size); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + for (i = 0; i < vec_size; i++) { + if (vec[i] == vec_size/2 * page_size) + break; + } + ksft_test_result(vec[i] == vec_size/2 * page_size, + "%s only middle page dirty\n", __func__); + + free(map); + } else { + ksft_test_result_skip("all new huge page must be dirty\n"); + ksft_test_result_skip("all the huge page must not be dirty\n"); + ksft_test_result_skip("all the huge page must be dirty and clear\n"); + ksft_test_result_skip("only middle page dirty\n"); + } + + // 5. clear first half of huge page + map = gethugepage(map_size); + if (map) { + ret = process_memwatch(0, map, map_size/2, MEMWATCH_SD_CLEAR, NULL, 0); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_GET, vec, vec_size); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + for (i = 0; i < vec_size/2; i++) + if (vec[i] != (i + vec_size/2) * page_size) + break; + + ksft_test_result(i == vec_size/2 && ret == vec_size/2, + "%s clear first half of huge page\n", __func__); + free(map); + } else { + ksft_test_result_skip("clear first half of huge page\n"); + } + + // 6. clear first half of huge page with limited buffer + map = gethugepage(map_size); + if (map) { + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_CLEAR | MEMWATCH_SD_GET, + vec, vec_size/2); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_GET, vec, vec_size); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + for (i = 0; i < vec_size/2; i++) + if (vec[i] != (i + vec_size/2) * page_size) + break; + + ksft_test_result(i == vec_size/2 && ret == vec_size/2, + "%s clear first half of huge page with limited buffer\n", + __func__); + free(map); + } else { + ksft_test_result_skip("clear first half of huge page with limited buffer\n"); + } + + // 7. clear second half of huge page + map = gethugepage(map_size); + if (map) { + memset(map, -1, map_size); + ret = process_memwatch(0, map + map_size/2, map_size/2, MEMWATCH_SD_CLEAR, NULL, 0); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + ret = process_memwatch(0, map, map_size, MEMWATCH_SD_GET, vec, vec_size); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + for (i = 0; i < vec_size/2; i++) + if (vec[i] != i * page_size) + break; + + ksft_test_result(i == vec_size/2, "%s clear second half huge page\n", __func__); + free(map); + } else { + ksft_test_result_skip("clear second half huge page\n"); + } + + free(vec); + free(vec2); + return 0; +} + +int base_tests(char *prefix, char *mem, int mem_size, int page_size, int skip) +{ + int vec_size, i, j, ret, dirty_pages, dirty_pages2; + loff_t *vec, *vec2; + + if (skip) { + ksft_test_result_skip("%s all new pages must be soft dirty\n", prefix); + ksft_test_result_skip("%s all pages must not be soft dirty\n", prefix); + ksft_test_result_skip("%s all pages dirty other than first and the last one\n", + prefix); + ksft_test_result_skip("%s only middle page dirty\n", prefix); + ksft_test_result_skip("%s only two middle pages dirty\n", prefix); + ksft_test_result_skip("%s only get 2 dirty pages and clear them as well\n", prefix); + ksft_test_result_skip("%s Range clear only\n", prefix); + return 0; + } + + vec_size = mem_size/page_size; + vec = malloc(sizeof(loff_t) * vec_size); + vec2 = malloc(sizeof(loff_t) * vec_size); + + /* 1. all new pages must be soft dirty and clear the range for next test */ + dirty_pages = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR, + vec, vec_size - 2); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + dirty_pages2 = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR, + vec2, vec_size); + if (dirty_pages2 < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages2, errno, strerror(errno)); + + for (i = 0; i < dirty_pages; i++) + if (vec[i] != i * page_size) + break; + for (j = 0; j < dirty_pages2; j++) + if (vec2[j] != (j + vec_size - 2) * page_size) + break; + + ksft_test_result(dirty_pages == vec_size - 2 && i == dirty_pages && + dirty_pages2 == 2 && j == dirty_pages2, + "%s all new pages must be soft dirty\n", prefix); + + // 2. all pages must not be soft dirty + dirty_pages = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET, vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 0, "%s all pages must not be soft dirty\n", prefix); + + // 3. all pages dirty other than first and the last one + memset(mem + page_size, -1, (mem_size - 2 * page_size)); + + dirty_pages = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET, vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + for (i = 0; i < dirty_pages; i++) { + if (vec[i] != (i + 1) * page_size) + break; + } + + ksft_test_result(dirty_pages == vec_size - 2 && i == vec_size - 2, + "%s all pages dirty other than first and the last one\n", prefix); + + // 4. only middle page dirty + clear_softdirty(); + mem[vec_size/2 * page_size]++; + + dirty_pages = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET, vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + for (i = 0; i < vec_size; i++) { + if (vec[i] == vec_size/2 * page_size) + break; + } + ksft_test_result(vec[i] == vec_size/2 * page_size, + "%s only middle page dirty\n", prefix); + + // 5. only two middle pages dirty and walk over only middle pages + clear_softdirty(); + mem[vec_size/2 * page_size]++; + mem[(vec_size/2 + 1) * page_size]++; + + dirty_pages = process_memwatch(0, &mem[vec_size/2 * page_size], 2 * page_size, + MEMWATCH_SD_GET, vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 2 && vec[0] == 0 && vec[1] == page_size, + "%s only two middle pages dirty\n", prefix); + + /* 6. only get 2 dirty pages and clear them as well */ + memset(mem, -1, mem_size); + + /* get and clear second and third pages */ + ret = process_memwatch(0, mem + page_size, 2 * page_size, + MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR, vec, 2); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + dirty_pages = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET, + vec2, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + for (i = 0; i < vec_size - 2; i++) { + if (i == 0 && (vec[i] != 0 || vec2[i] != 0)) + break; + else if (i == 1 && (vec[i] != page_size || vec2[i] != (i + 2) * page_size)) + break; + else if (i > 1 && (vec2[i] != (i + 2) * page_size)) + break; + } + + ksft_test_result(dirty_pages == vec_size - 2 && i == vec_size - 2, + "%s only get 2 dirty pages and clear them as well\n", prefix); + /* 7. Range clear only */ + memset(mem, -1, mem_size); + dirty_pages = process_memwatch(0, mem, mem_size, MEMWATCH_SD_CLEAR, NULL, 0); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + dirty_pages2 = process_memwatch(0, mem, mem_size, MEMWATCH_SD_GET, vec, vec_size); + if (dirty_pages2 < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages2, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 0 && dirty_pages2 == 0, "%s Range clear only\n", + prefix); + + free(vec); + free(vec2); + return 0; +} + +int performance_base_tests(char *prefix, char *mem, int mem_size, int page_size, int skip) +{ + int vec_size, i, ret, dirty_pages, dirty_pages2; + loff_t *vec, *vec2; + + if (skip) { + ksft_test_result_skip("%s all new pages must be soft dirty\n", prefix); + ksft_test_result_skip("%s all pages must not be soft dirty\n", prefix); + ksft_test_result_skip("%s all pages dirty other than first and the last one\n", + prefix); + ksft_test_result_skip("%s only middle page dirty\n", prefix); + ksft_test_result_skip("%s only two middle pages dirty\n", prefix); + ksft_test_result_skip("%s only get 2 dirty pages and clear them as well\n", prefix); + ksft_test_result_skip("%s Range clear only\n", prefix); + return 0; + } + + vec_size = mem_size/page_size; + vec = malloc(sizeof(loff_t) * vec_size); + vec2 = malloc(sizeof(loff_t) * vec_size); + + /* 1. all new pages must be soft dirty and clear the range for next test */ + dirty_pages = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR | + MEMWATCH_SD_NO_REUSED_REGIONS, + vec, vec_size - 2); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + dirty_pages2 = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR | + MEMWATCH_SD_NO_REUSED_REGIONS, + vec2, vec_size); + if (dirty_pages2 < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages2, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 0 && dirty_pages2 == 0, + "%s page isn't dirty\n", prefix); + + // 2. all pages must not be soft dirty + dirty_pages = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_GET | MEMWATCH_SD_NO_REUSED_REGIONS, + vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 0, "%s all pages must not be soft dirty\n", prefix); + + // 3. all pages dirty other than first and the last one + memset(mem + page_size, -1, (mem_size - 2 * page_size)); + + dirty_pages = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_GET | MEMWATCH_SD_NO_REUSED_REGIONS, + vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + for (i = 0; i < dirty_pages; i++) { + if (vec[i] != (i + 1) * page_size) + break; + } + + ksft_test_result(dirty_pages == vec_size - 2 && i == vec_size - 2, + "%s all pages dirty other than first and the last one\n", prefix); + + // 4. only middle page dirty + clear_softdirty(); + mem[vec_size/2 * page_size]++; + + dirty_pages = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_GET | MEMWATCH_SD_NO_REUSED_REGIONS, + vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + for (i = 0; i < vec_size; i++) { + if (vec[i] == vec_size/2 * page_size) + break; + } + ksft_test_result(vec[i] == vec_size/2 * page_size, + "%s only middle page dirty\n", prefix); + + // 5. only two middle pages dirty and walk over only middle pages + clear_softdirty(); + mem[vec_size/2 * page_size]++; + mem[(vec_size/2 + 1) * page_size]++; + + dirty_pages = process_memwatch(0, &mem[vec_size/2 * page_size], 2 * page_size, + MEMWATCH_SD_GET | MEMWATCH_SD_NO_REUSED_REGIONS, + vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 2 && vec[0] == 0 && vec[1] == page_size, + "%s only two middle pages dirty\n", prefix); + + /* 6. only get 2 dirty pages and clear them as well */ + memset(mem, -1, mem_size); + + /* get and clear second and third pages */ + ret = process_memwatch(0, mem + page_size, 2 * page_size, + MEMWATCH_SD_GET | MEMWATCH_SD_CLEAR | MEMWATCH_SD_NO_REUSED_REGIONS, + vec, 2); + if (ret < 0) + ksft_exit_fail_msg("error %d %d %s\n", ret, errno, strerror(errno)); + + dirty_pages = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_GET | MEMWATCH_SD_NO_REUSED_REGIONS, + vec2, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + for (i = 0; i < vec_size - 2; i++) { + if (i == 0 && (vec[i] != 0 || vec2[i] != 0)) + break; + else if (i == 1 && (vec[i] != page_size || vec2[i] != (i + 2) * page_size)) + break; + else if (i > 1 && (vec2[i] != (i + 2) * page_size)) + break; + } + + ksft_test_result(dirty_pages == vec_size - 2 && i == vec_size - 2, + "%s only get 2 dirty pages and clear them as well\n", prefix); + /* 7. Range clear only */ + memset(mem, -1, mem_size); + dirty_pages = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_CLEAR | MEMWATCH_SD_NO_REUSED_REGIONS, + NULL, 0); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + dirty_pages2 = process_memwatch(0, mem, mem_size, + MEMWATCH_SD_GET | MEMWATCH_SD_NO_REUSED_REGIONS, + vec, vec_size); + if (dirty_pages2 < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages2, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 0 && dirty_pages2 == 0, "%s Range clear only\n", + prefix); + + free(vec); + free(vec2); + return 0; +} + +int unmapped_region_tests(int page_size) +{ + void *start = (void *)0x10000000; + int dirty_pages, len = 0x00040000; + int vec_size = len / page_size; + loff_t *vec = malloc(sizeof(loff_t) * vec_size); + + /* 1. Get dirty pages */ + dirty_pages = process_memwatch(0, start, len, MEMWATCH_SD_GET, vec, vec_size); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + ksft_test_result(dirty_pages >= 0, "%s Get dirty pages\n", __func__); + + /* 2. Clear dirty bit of whole address space */ + dirty_pages = process_memwatch(0, 0, 0x7FFFFFFF, MEMWATCH_SD_CLEAR, NULL, 0); + if (dirty_pages < 0) + ksft_exit_fail_msg("error %d %d %s\n", dirty_pages, errno, strerror(errno)); + + ksft_test_result(dirty_pages == 0, "%s Get dirty pages\n", __func__); + + free(vec); + return 0; +} + +static void test_simple(int page_size) +{ + int i; + char *map; + loff_t *vec = NULL; + + map = aligned_alloc(page_size, page_size); + if (!map) + ksft_exit_fail_msg("mmap failed\n"); + + clear_softdirty(); + + for (i = 0 ; i < TEST_ITERATIONS; i++) { + if (process_memwatch(0, map, page_size, MEMWATCH_SD_GET, vec, 1) == 1) { + ksft_print_msg("dirty bit was 1, but should be 0 (i=%d)\n", i); + break; + } + + clear_softdirty(); + // Write something to the page to get the dirty bit enabled on the page + map[0]++; + + if (process_memwatch(0, map, page_size, MEMWATCH_SD_GET, vec, 1) == 0) { + ksft_print_msg("dirty bit was 0, but should be 1 (i=%d)\n", i); + break; + } + + clear_softdirty(); + } + free(map); + + ksft_test_result(i == TEST_ITERATIONS, "Test %s\n", __func__); +} + +int main(int argc, char **argv) +{ + int page_size = getpagesize(); + size_t hpage_len = read_pmd_pagesize(); + char *mem, *map; + int mem_size; + + ksft_print_header(); + ksft_set_plan(44); + + /* 1. Sanity testing */ + sanity_tests(page_size); + + /* 2. Normal page testing */ + mem_size = 10 * page_size; + mem = mmap(NULL, mem_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, -1, 0); + if (!mem) + ksft_exit_fail_msg("error nomem\n"); + + base_tests("Page testing:", mem, mem_size, page_size, 0); + + munmap(mem, mem_size); + + /* 3. Large page testing */ + mem_size = 512 * 10 * page_size; + mem = mmap(NULL, mem_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, -1, 0); + if (!mem) + ksft_exit_fail_msg("error nomem\n"); + + base_tests("Large Page testing:", mem, mem_size, page_size, 0); + + munmap(mem, mem_size); + + /* 4. Huge page testing */ + map = gethugepage(hpage_len); + if (check_huge(map)) + base_tests("Huge page testing:", map, hpage_len, page_size, 0); + else + base_tests("Huge page testing:", NULL, 0, 0, 1); + + free(map); + + /* 5. Normal page testing */ + mem_size = 10 * page_size; + mem = mmap(NULL, mem_size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON, -1, 0); + if (!mem) + ksft_exit_fail_msg("error nomem\n"); + + performance_base_tests("Performance Page testing:", mem, mem_size, page_size, 0); + + munmap(mem, mem_size); + + /* 6. Huge page tests */ + hpage_unit_tests(page_size); + + /* 7. Unmapped address test */ + unmapped_region_tests(page_size); + + /* 8. Iterative test */ + test_simple(page_size); + + return ksft_exit_pass(); +}
Add the syscall with explanation of the operations.
Signed-off-by: Muhammad Usama Anjum usama.anjum@collabora.com --- Documentation/admin-guide/mm/soft-dirty.rst | 48 ++++++++++++++++++++- 1 file changed, 47 insertions(+), 1 deletion(-)
diff --git a/Documentation/admin-guide/mm/soft-dirty.rst b/Documentation/admin-guide/mm/soft-dirty.rst index cb0cfd6672fa..030d75658010 100644 --- a/Documentation/admin-guide/mm/soft-dirty.rst +++ b/Documentation/admin-guide/mm/soft-dirty.rst @@ -5,7 +5,12 @@ Soft-Dirty PTEs ===============
The soft-dirty is a bit on a PTE which helps to track which pages a task -writes to. In order to do this tracking one should +writes to. + +Using Proc FS +------------- + +In order to do this tracking one should
1. Clear soft-dirty bits from the task's PTEs.
@@ -20,6 +25,47 @@ writes to. In order to do this tracking one should 64-bit qword is the soft-dirty one. If set, the respective PTE was written to since step 1.
+Using System Call +----------------- + +process_memwatch system call can be used to find the dirty pages.:: + + long process_memwatch(int pidfd, unsigned long start, int len, + unsigned int flags, void *vec, int vec_len); + +The pidfd specifies the pidfd of process whose memory needs to be watched. +The calling process must have PTRACE_MODE_ATTACH_FSCREDS capabilities over +the process whose pidfd has been specified. It can be zero which means that +the process wants to watch its own memory. The operation is determined by +flags. The start argument must be a multiple of the system page size. The +len argument need not be a multiple of the page size, but since the +information is returned for the whole pages, len is effectively rounded +up to the next multiple of the page size. + +The vec is output array in which the offsets of the pages are returned. +Offset is calculated from start address. User lets the kernel know about the +size of the vec by passing size in vec_len. The system call returns when the +whole range has been searched or vec is completely filled. The whole range +isn't cleared if vec fills up completely. + +The flags argument specifies the operation to be performed. The MEMWATCH_SD_GET +and MEMWATCH_SD_CLEAR operations can be used separately or together to perform +MEMWATCH_SD_GET and MEMWATCH_SD_CLEAR atomically as one operation.:: + + MEMWATCH_SD_GET + Get the page offsets which are soft dirty. + + MEMWATCH_SD_CLEAR + Clear the pages which are soft dirty. + + MEMWATCH_SD_NO_REUSED_REGIONS + This optional flag can be specified in combination with other flags. + VM_SOFTDIRTY is ignored for the VMAs for performances reasons. This + flag shows only those pages dirty which have been written to by the + user. All new allocations aren't returned to be dirty. + +Explanation +-----------
Internally, to do this tracking, the writable bit is cleared from PTEs when the soft-dirty bit is cleared. So, after this, when the task tries to
On 7/26/22 9:18 PM, Muhammad Usama Anjum wrote:
Hello,
This patch series implements a new syscall, process_memwatch. Currently, only the support to watch soft-dirty PTE bit is added. This syscall is generic to watch the memory of the process. There is enough room to add more operations like this to watch memory in the future.
Soft-dirty PTE bit of the memory pages can be viewed by using pagemap procfs file. The soft-dirty PTE bit for the memory in a process can be cleared by writing to the clear_refs file. This series adds features that weren't possible through the Proc FS interface.
- There is no atomic get soft-dirty PTE bit status and clear operation possible.
- The soft-dirty PTE bit of only a part of memory cannot be cleared.
Historically, soft-dirty PTE bit tracking has been used in the CRIU project. The Proc FS interface is enough for that as I think the process is frozen. We have the use case where we need to track the soft-dirty PTE bit for running processes. We need this tracking and clear mechanism of a region of memory while the process is running to emulate the getWriteWatch() syscall of Windows. This syscall is used by games to keep track of dirty pages and keep processing only the dirty pages. This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information.
As in the current kernel there is no way to clear a part of memory (instead of clearing the Soft-Dirty bits for the entire processi) and get+clear operation cannot be performed atomically, there are other methods to mimic this information entirely in userspace with poor performance:
The mprotect syscall and SIGSEGV handler for bookkeeping
The userfaultfd syscall with the handler for bookkeeping
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);
Any thoughts?
This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall:
- Get the pages that are soft-dirty.
- Clear the pages which are soft-dirty.
- The optional flag to ignore the VM_SOFTDIRTY and only track per page
soft-dirty PTE bit
There are two decisions which have been taken about how to get the output from the syscall.
- Return offsets of the pages from the start in the vec
- Stop execution when vec is filled with dirty pages
These two arguments doesn't follow the mincore() philosophy where the output array corresponds to the address range in one to one fashion, hence the output buffer length isn't passed and only a flag is set if the page is present. This makes mincore() easy to use with less control. We are passing the size of the output array and putting return data consecutively which is offset of dirty pages from the start. The user can convert these offsets back into the dirty page addresses easily. Suppose, the user want to get first 10 dirty pages from a total memory of 100 pages. He'll allocate output buffer of size 10 and process_memwatch() syscall will abort after finding the 10 pages. This behaviour is needed to support Windows' getWriteWatch(). The behaviour like mincore() can be achieved by passing output buffer of 100 size. This interface can be used for any desired behaviour.
Regards, Muhammad Usama Anjum
Muhammad Usama Anjum (5): fs/proc/task_mmu: make functions global to be used in other files mm: Implement process_memwatch syscall mm: wire up process_memwatch syscall for x86 selftests: vm: add process_memwatch syscall tests mm: add process_memwatch syscall documentation
Documentation/admin-guide/mm/soft-dirty.rst | 48 +- arch/x86/entry/syscalls/syscall_32.tbl | 1 + arch/x86/entry/syscalls/syscall_64.tbl | 1 + fs/proc/task_mmu.c | 84 +-- include/linux/mm_inline.h | 99 +++ include/linux/syscalls.h | 3 +- include/uapi/asm-generic/unistd.h | 5 +- include/uapi/linux/memwatch.h | 12 + kernel/sys_ni.c | 1 + mm/Makefile | 2 +- mm/memwatch.c | 285 ++++++++ tools/include/uapi/asm-generic/unistd.h | 5 +- .../arch/x86/entry/syscalls/syscall_64.tbl | 1 + tools/testing/selftests/vm/.gitignore | 1 + tools/testing/selftests/vm/Makefile | 2 + tools/testing/selftests/vm/memwatch_test.c | 635 ++++++++++++++++++ 16 files changed, 1098 insertions(+), 87 deletions(-) create mode 100644 include/uapi/linux/memwatch.h create mode 100644 mm/memwatch.c create mode 100644 tools/testing/selftests/vm/memwatch_test.c
On 26.07.22 18:18, Muhammad Usama Anjum wrote:
Hello,
Hi,
This patch series implements a new syscall, process_memwatch. Currently, only the support to watch soft-dirty PTE bit is added. This syscall is generic to watch the memory of the process. There is enough room to add more operations like this to watch memory in the future.
Soft-dirty PTE bit of the memory pages can be viewed by using pagemap procfs file. The soft-dirty PTE bit for the memory in a process can be cleared by writing to the clear_refs file. This series adds features that weren't possible through the Proc FS interface.
- There is no atomic get soft-dirty PTE bit status and clear operation possible.
Such an interface might be easy to add, no?
- The soft-dirty PTE bit of only a part of memory cannot be cleared.
Same.
So I'm curious why we need a new syscall for that.
Historically, soft-dirty PTE bit tracking has been used in the CRIU project. The Proc FS interface is enough for that as I think the process is frozen. We have the use case where we need to track the soft-dirty PTE bit for running processes. We need this tracking and clear mechanism of a region of memory while the process is running to emulate the getWriteWatch() syscall of Windows. This syscall is used by games to keep track of dirty pages and keep processing only the dirty pages. This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information.
As in the current kernel there is no way to clear a part of memory (instead of clearing the Soft-Dirty bits for the entire processi) and get+clear operation cannot be performed atomically, there are other methods to mimic this information entirely in userspace with poor performance:
- The mprotect syscall and SIGSEGV handler for bookkeeping
- The userfaultfd syscall with the handler for bookkeeping
You write "poor performance". Did you actually implement a prototype using userfaultfd-wp? Can you share numbers for comparison?
Adding an new syscall just for handling a corner case feature (soft-dirty, which we all love, of course) needs good justification.
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall:
- Get the pages that are soft-dirty.
- Clear the pages which are soft-dirty.
- The optional flag to ignore the VM_SOFTDIRTY and only track per page
soft-dirty PTE bit
Huh, why? VM_SOFTDIRTY is an internal implementation detail and should remain such.
VM_SOFTDIRTY translates to "all pages in this VMA are soft-dirty".
Hello,
Thank you for reviewing and commenting.
On 8/10/22 2:03 PM, David Hildenbrand wrote:
On 26.07.22 18:18, Muhammad Usama Anjum wrote:
Hello,
Hi,
This patch series implements a new syscall, process_memwatch. Currently, only the support to watch soft-dirty PTE bit is added. This syscall is generic to watch the memory of the process. There is enough room to add more operations like this to watch memory in the future.
Soft-dirty PTE bit of the memory pages can be viewed by using pagemap procfs file. The soft-dirty PTE bit for the memory in a process can be cleared by writing to the clear_refs file. This series adds features that weren't possible through the Proc FS interface.
- There is no atomic get soft-dirty PTE bit status and clear operation possible.
Such an interface might be easy to add, no?
Are you referring to ioctl? I think this syscall can be used in future for adding other operations like soft-dirty. This is why syscall has been added.
If community doesn't agree, I can translate this syscall to the ioctl same as it is.
- The soft-dirty PTE bit of only a part of memory cannot be cleared.
Same.
So I'm curious why we need a new syscall for that.
Historically, soft-dirty PTE bit tracking has been used in the CRIU project. The Proc FS interface is enough for that as I think the process is frozen. We have the use case where we need to track the soft-dirty PTE bit for running processes. We need this tracking and clear mechanism of a region of memory while the process is running to emulate the getWriteWatch() syscall of Windows. This syscall is used by games to keep track of dirty pages and keep processing only the dirty pages. This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information.
As in the current kernel there is no way to clear a part of memory (instead of clearing the Soft-Dirty bits for the entire processi) and get+clear operation cannot be performed atomically, there are other methods to mimic this information entirely in userspace with poor performance:
- The mprotect syscall and SIGSEGV handler for bookkeeping
- The userfaultfd syscall with the handler for bookkeeping
You write "poor performance". Did you actually implement a prototype using userfaultfd-wp? Can you share numbers for comparison?
Adding an new syscall just for handling a corner case feature (soft-dirty, which we all love, of course) needs good justification.
The cycles are given in thousands. 60 means 60k cycles here which have been measured with rdtsc().
| | Region size in Pages | 1 | 10 | 100 | 1000 | 10000 | |---|----------------------|------|------|-------|-------|--------| | 1 | MEMWATCH | 7 | 58 | 281 | 1178 | 17563 | | 2 | MEMWATCH Perf | 4 | 23 | 107 | 1331 | 8924 | | 3 | USERFAULTFD | 5405 | 6550 | 10387 | 55708 | 621522 | | 4 | MPROTECT_SEGV | 35 | 611 | 1060 | 6646 | 60149 |
1. MEMWATCH --> process_memwatch considering VM_SOFTDIRT (splitting is possible) 2. MEMWATCH Perf --> process_memwatch without considering VM_SOFTDIRTY 3. Userafaultfd --> userfaultfd with handling is userspace 4. Mprotect_segv --> mprotect and signal handler in userspace
Note: Implementation of mprotect_segv is very similar to userfaultfd. In both of these, the signal/fault is being handled in the userspace. In mprotect_segv, the memory region is write-protected through mprotect and SEGV signal is received when something is written to this region. This signal's handler is where we do calculations about soft dirty pages. Mprotect_segv mechanism must be lighter than userfaultfd inside kernel.
My benchmark application is purely single threaded to keep effort to a minimum until we decide to spend more time. It has been written to measure the time taken in a serial execution of these statements without locks. If the multi-threaded application is used and randomization is introduced, it should affect `MPROTECT_SEGV` and `userfaultd` implementations more than memwatch. But in this particular setting, memwatch and mprotect_segv perform closely.
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall:
- Get the pages that are soft-dirty.
- Clear the pages which are soft-dirty.
- The optional flag to ignore the VM_SOFTDIRTY and only track per page
soft-dirty PTE bit
Huh, why? VM_SOFTDIRTY is an internal implementation detail and should remain such.
VM_SOFTDIRTY translates to "all pages in this VMA are soft-dirty".
Clearing soft-dirty bit for a range of memory may result in splitting the VMA. Soft-dirty bit of the per page need to be cleared. The VM_SOFTDIRTY flag from this splitted VMA need to be cleared. The kernel may decide to merge this splitted VMA back. Please note that kernel doesn't take into account the VM_SOFTDIRTY flag of the VMAs when it decides to merge the VMAs. This not only gives performance hit, but also the non-dirty pages of the whole VMA start to appear as dirty again after the VMA merging. To avoid this penalty, MEMWATCH_SD_NO_REUSED_REGIONS flag has been added to ignore the VM_SOFTDIRTY and just rely on the soft-dirty bit present on the per page. The user is aware about the constraint that the new regions will not be found dirty if this flag is specified.
David Hildenbrand david@redhat.com writes:
On 26.07.22 18:18, Muhammad Usama Anjum wrote:
Hello,
Hi,
This patch series implements a new syscall, process_memwatch. Currently, only the support to watch soft-dirty PTE bit is added. This syscall is generic to watch the memory of the process. There is enough room to add more operations like this to watch memory in the future.
Soft-dirty PTE bit of the memory pages can be viewed by using pagemap procfs file. The soft-dirty PTE bit for the memory in a process can be cleared by writing to the clear_refs file. This series adds features that weren't possible through the Proc FS interface.
- There is no atomic get soft-dirty PTE bit status and clear operation possible.
Such an interface might be easy to add, no?
- The soft-dirty PTE bit of only a part of memory cannot be cleared.
Same.
So I'm curious why we need a new syscall for that.
Hi David,
Yes, sure. Though it has to be through an ioctl since we need both input and output semantics at the same call to keep the atomic semantics.
I answered Peter Enderborg about our concerns when turning this into an ioctl. But they are possible to overcome.
project. The Proc FS interface is enough for that as I think the process is frozen. We have the use case where we need to track the soft-dirty PTE bit for running processes. We need this tracking and clear mechanism of a region of memory while the process is running to emulate the getWriteWatch() syscall of Windows. This syscall is used by games to keep track of dirty pages and keep processing only the dirty pages. This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information.
As in the current kernel there is no way to clear a part of memory (instead of clearing the Soft-Dirty bits for the entire processi) and get+clear operation cannot be performed atomically, there are other methods to mimic this information entirely in userspace with poor performance:
- The mprotect syscall and SIGSEGV handler for bookkeeping
- The userfaultfd syscall with the handler for bookkeeping
You write "poor performance". Did you actually implement a prototype using userfaultfd-wp? Can you share numbers for comparison?
Yes, we did. I think Usama can share some numbers.
The problem with userfaultfd, as far as I understand, is that it will require a second userspace process to be called in order to handle the annotation that a page was touched, before remapping the page to make it accessible to the originating process, every time a page is touched. This context switch is prohibitively expensive to our use case, where Windows applications might invoke it quite often. Soft-dirty bit instead, allows the page tracking to be done entirely in kernelspace.
If I understand correctly, userfaultfd is usefull for VM/container migration, where the cost of the context switch is not a real concern, since there are much bigger costs from the migration itself.
Maybe we're missing some feature about userfaultfd that would allow us to avoid the cost, but from our observations we didn't find a way to overcome it.
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall:
- Get the pages that are soft-dirty.
- Clear the pages which are soft-dirty.
- The optional flag to ignore the VM_SOFTDIRTY and only track per page
soft-dirty PTE bit
Huh, why? VM_SOFTDIRTY is an internal implementation detail and should remain such. VM_SOFTDIRTY translates to "all pages in this VMA are soft-dirty".
That is something very specific about our use case, and we should explain it a bit better. The problem is that VM_SOFTDIRTY modifications introduce the overhead of the mm write lock acquisition, which is very visible in our benchmarks of Windows games running over Wine.
Since the main reason for VM_SOFTDIRTY to exist, as far as we understand it, is to track vma remapping, and this is a use case we don't need to worry about when implementing windows semantics, we'd like to be able to avoid this extra overhead, optionally, iff userspace knows it can be done safely.
VM_SOFTDIRTY is indeed an internal interface. Which is why we are proposing to expose the feature in terms of tracking VMA reuse.
Thanks,
On 7/26/22 18:18, Muhammad Usama Anjum wrote:
Hello,
This patch series implements a new syscall, process_memwatch. Currently, only the support to watch soft-dirty PTE bit is added. This syscall is generic to watch the memory of the process. There is enough room to add more operations like this to watch memory in the future.
Soft-dirty PTE bit of the memory pages can be viewed by using pagemap procfs file. The soft-dirty PTE bit for the memory in a process can be cleared by writing to the clear_refs file. This series adds features that weren't possible through the Proc FS interface.
- There is no atomic get soft-dirty PTE bit status and clear operation possible.
- The soft-dirty PTE bit of only a part of memory cannot be cleared.
Historically, soft-dirty PTE bit tracking has been used in the CRIU project. The Proc FS interface is enough for that as I think the process is frozen. We have the use case where we need to track the soft-dirty PTE bit for running processes. We need this tracking and clear mechanism of a region of memory while the process is running to emulate the getWriteWatch() syscall of Windows. This syscall is used by games to keep track of dirty pages and keep processing only the dirty pages. This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information.
As in the current kernel there is no way to clear a part of memory (instead of clearing the Soft-Dirty bits for the entire processi) and get+clear operation cannot be performed atomically, there are other methods to mimic this information entirely in userspace with poor performance:
The mprotect syscall and SIGSEGV handler for bookkeeping
The userfaultfd syscall with the handler for bookkeeping
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall:
- Get the pages that are soft-dirty.
- Clear the pages which are soft-dirty.
- The optional flag to ignore the VM_SOFTDIRTY and only track per page
soft-dirty PTE bit
Why can it not be done as a IOCTL?
There are two decisions which have been taken about how to get the output from the syscall.
- Return offsets of the pages from the start in the vec
- Stop execution when vec is filled with dirty pages
These two arguments doesn't follow the mincore() philosophy where the output array corresponds to the address range in one to one fashion, hence the output buffer length isn't passed and only a flag is set if the page is present. This makes mincore() easy to use with less control. We are passing the size of the output array and putting return data consecutively which is offset of dirty pages from the start. The user can convert these offsets back into the dirty page addresses easily. Suppose, the user want to get first 10 dirty pages from a total memory of 100 pages. He'll allocate output buffer of size 10 and process_memwatch() syscall will abort after finding the 10 pages. This behaviour is needed to support Windows' getWriteWatch(). The behaviour like mincore() can be achieved by passing output buffer of 100 size. This interface can be used for any desired behaviour.
Regards, Muhammad Usama Anjum
Muhammad Usama Anjum (5): fs/proc/task_mmu: make functions global to be used in other files mm: Implement process_memwatch syscall mm: wire up process_memwatch syscall for x86 selftests: vm: add process_memwatch syscall tests mm: add process_memwatch syscall documentation
Documentation/admin-guide/mm/soft-dirty.rst | 48 +- arch/x86/entry/syscalls/syscall_32.tbl | 1 + arch/x86/entry/syscalls/syscall_64.tbl | 1 + fs/proc/task_mmu.c | 84 +-- include/linux/mm_inline.h | 99 +++ include/linux/syscalls.h | 3 +- include/uapi/asm-generic/unistd.h | 5 +- include/uapi/linux/memwatch.h | 12 + kernel/sys_ni.c | 1 + mm/Makefile | 2 +- mm/memwatch.c | 285 ++++++++ tools/include/uapi/asm-generic/unistd.h | 5 +- .../arch/x86/entry/syscalls/syscall_64.tbl | 1 + tools/testing/selftests/vm/.gitignore | 1 + tools/testing/selftests/vm/Makefile | 2 + tools/testing/selftests/vm/memwatch_test.c | 635 ++++++++++++++++++ 16 files changed, 1098 insertions(+), 87 deletions(-) create mode 100644 include/uapi/linux/memwatch.h create mode 100644 mm/memwatch.c create mode 100644 tools/testing/selftests/vm/memwatch_test.c
On 8/10/22 2:22 PM, Peter.Enderborg@sony.com wrote:
On 7/26/22 18:18, Muhammad Usama Anjum wrote:
Hello,
This patch series implements a new syscall, process_memwatch. Currently, only the support to watch soft-dirty PTE bit is added. This syscall is generic to watch the memory of the process. There is enough room to add more operations like this to watch memory in the future.
Soft-dirty PTE bit of the memory pages can be viewed by using pagemap procfs file. The soft-dirty PTE bit for the memory in a process can be cleared by writing to the clear_refs file. This series adds features that weren't possible through the Proc FS interface.
- There is no atomic get soft-dirty PTE bit status and clear operation possible.
- The soft-dirty PTE bit of only a part of memory cannot be cleared.
Historically, soft-dirty PTE bit tracking has been used in the CRIU project. The Proc FS interface is enough for that as I think the process is frozen. We have the use case where we need to track the soft-dirty PTE bit for running processes. We need this tracking and clear mechanism of a region of memory while the process is running to emulate the getWriteWatch() syscall of Windows. This syscall is used by games to keep track of dirty pages and keep processing only the dirty pages. This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information.
As in the current kernel there is no way to clear a part of memory (instead of clearing the Soft-Dirty bits for the entire processi) and get+clear operation cannot be performed atomically, there are other methods to mimic this information entirely in userspace with poor performance:
The mprotect syscall and SIGSEGV handler for bookkeeping
The userfaultfd syscall with the handler for bookkeeping
long process_memwatch(int pidfd, unsigned long start, int len, unsigned int flags, void *vec, int vec_len);This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall:
- Get the pages that are soft-dirty.
- Clear the pages which are soft-dirty.
- The optional flag to ignore the VM_SOFTDIRTY and only track per page
soft-dirty PTE bit
Why can it not be done as a IOCTL?
It can be done as ioctl. I think this syscall can be used in future for adding other operations like soft-dirty. This is why syscall has been added.
"Peter.Enderborg@sony.com" Peter.Enderborg@sony.com writes:
This syscall can be used by the CRIU project and other applications which require soft-dirty PTE bit information. The following operations are supported in this syscall:
- Get the pages that are soft-dirty.
- Clear the pages which are soft-dirty.
- The optional flag to ignore the VM_SOFTDIRTY and only track per page
soft-dirty PTE bit
Hi Peter,
(For context, I wrote a previous version of this patch and have been working with Usama on the current patch).
Why can it not be done as a IOCTL?
Considering an ioctl is basically a namespaced syscall with extra-steps, surely we can do it :) There are a few reasons we haven't, though:
1) ioctl auditing/controling is much harder than syscall
2) There is a concern for performance, since this might be executed frequently by windows applications running over wine. There is an extra cost with unnecessary copy_[from/to]_user that we wanted to avoid, even though we haven't measured.
3) I originally wrote this at the time process_memadvise was merged. I felt it fits the same kind of interface exposed by process_memadvise/process_mrelease, recently merged.
4) Not obvious whether the ioctl would be against pagemap/clear_refs. Neither file name describes both input and output semantics.
Obviously, all of those reasons can be worked around, and we can turn this into an ioctl.
Thanks,
linux-kselftest-mirror@lists.linaro.org
-
David Hildenbrand -
Gabriel Krisman Bertazi -
Muhammad Usama Anjum -
Peter.Enderborg＠sony.com