| Message ID | 20180522194430.426688-1-songliubraving@fb.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, May 22, 2018 at 12:44:30PM -0700, Song Liu wrote: > khugepaged_enter_vma_merge() is using a different approach to check > whether a vma is valid for khugepaged_enter(): > > if (!vma->anon_vma) > /* > * Not yet faulted in so we will register later in the > * page fault if needed. > */ > return 0; > if (vma->vm_ops || (vm_flags & VM_NO_KHUGEPAGED)) > /* khugepaged not yet working on file or special mappings */ > return 0; > > This check has some problems. One of the obvious problems is that > it doesn't check shmem_file(), so that vma backed with shmem files > will not call khugepaged_enter(). Here is an example of failed madvise(): > > /* mount /dev/shm with huge=advise: > * mount -o remount,huge=advise /dev/shm */ > /* create file /dev/shm/huge */ > #define HUGE_FILE "/dev/shm/huge" > > fd = open(HUGE_FILE, O_RDONLY); > ptr = mmap(NULL, FILE_SIZE, PROT_READ, MAP_PRIVATE, fd, 0); > ret = madvise(ptr, FILE_SIZE, MADV_HUGEPAGE); > > madvise() will return 0, but this memory region is never put in huge > page (check from /proc/meminfo: ShmemHugePages). > > This patch fixes these problems by reusing hugepage_vma_check() in > khugepaged_enter_vma_merge(). > > vma->vm_flags is not yet updated in khugepaged_enter_vma_merge(), > so we need to pass the new vm_flags to hugepage_vma_check() through > a separate argument. > > Signed-off-by: Song Liu <songliubraving@fb.com> > --- > mm/khugepaged.c | 26 ++++++++++++-------------- > 1 file changed, 12 insertions(+), 14 deletions(-) > > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > index d7b2a4b..9f74e51 100644 > --- a/mm/khugepaged.c > +++ b/mm/khugepaged.c > @@ -430,18 +430,15 @@ int __khugepaged_enter(struct mm_struct *mm) > return 0; > } > > +static bool hugepage_vma_check(struct vm_area_struct *vma, > + unsigned long vm_flags); > + The patch looks good to me. But can we move hugepage_vma_check() here to avoid forward declaration of the function?
> On May 28, 2018, at 3:57 AM, Kirill A. Shutemov <kirill@shutemov.name> wrote: > > On Tue, May 22, 2018 at 12:44:30PM -0700, Song Liu wrote: >> khugepaged_enter_vma_merge() is using a different approach to check >> whether a vma is valid for khugepaged_enter(): >> >> if (!vma->anon_vma) >> /* >> * Not yet faulted in so we will register later in the >> * page fault if needed. >> */ >> return 0; >> if (vma->vm_ops || (vm_flags & VM_NO_KHUGEPAGED)) >> /* khugepaged not yet working on file or special mappings */ >> return 0; >> >> This check has some problems. One of the obvious problems is that >> it doesn't check shmem_file(), so that vma backed with shmem files >> will not call khugepaged_enter(). Here is an example of failed madvise(): >> >> /* mount /dev/shm with huge=advise: >> * mount -o remount,huge=advise /dev/shm */ >> /* create file /dev/shm/huge */ >> #define HUGE_FILE "/dev/shm/huge" >> >> fd = open(HUGE_FILE, O_RDONLY); >> ptr = mmap(NULL, FILE_SIZE, PROT_READ, MAP_PRIVATE, fd, 0); >> ret = madvise(ptr, FILE_SIZE, MADV_HUGEPAGE); >> >> madvise() will return 0, but this memory region is never put in huge >> page (check from /proc/meminfo: ShmemHugePages). >> >> This patch fixes these problems by reusing hugepage_vma_check() in >> khugepaged_enter_vma_merge(). >> >> vma->vm_flags is not yet updated in khugepaged_enter_vma_merge(), >> so we need to pass the new vm_flags to hugepage_vma_check() through >> a separate argument. >> >> Signed-off-by: Song Liu <songliubraving@fb.com> >> --- >> mm/khugepaged.c | 26 ++++++++++++-------------- >> 1 file changed, 12 insertions(+), 14 deletions(-) >> >> diff --git a/mm/khugepaged.c b/mm/khugepaged.c >> index d7b2a4b..9f74e51 100644 >> --- a/mm/khugepaged.c >> +++ b/mm/khugepaged.c >> @@ -430,18 +430,15 @@ int __khugepaged_enter(struct mm_struct *mm) >> return 0; >> } >> >> +static bool hugepage_vma_check(struct vm_area_struct *vma, >> + unsigned long vm_flags); >> + > > The patch looks good to me. > > But can we move hugepage_vma_check() here to avoid forward declaration of > the function? Thanks for the feedback! I will send v3 with this change. Song
diff --git a/mm/khugepaged.c b/mm/khugepaged.c index d7b2a4b..9f74e51 100644 --- a/mm/khugepaged.c +++ b/mm/khugepaged.c @@ -430,18 +430,15 @@ int __khugepaged_enter(struct mm_struct *mm) return 0; } +static bool hugepage_vma_check(struct vm_area_struct *vma, + unsigned long vm_flags); + int khugepaged_enter_vma_merge(struct vm_area_struct *vma, unsigned long vm_flags) { unsigned long hstart, hend; - if (!vma->anon_vma) - /* - * Not yet faulted in so we will register later in the - * page fault if needed. - */ - return 0; - if (vma->vm_ops || (vm_flags & VM_NO_KHUGEPAGED)) - /* khugepaged not yet working on file or special mappings */ + + if (!hugepage_vma_check(vma, vm_flags)) return 0; hstart = (vma->vm_start + ~HPAGE_PMD_MASK) & HPAGE_PMD_MASK; hend = vma->vm_end & HPAGE_PMD_MASK; @@ -819,10 +816,11 @@ khugepaged_alloc_page(struct page **hpage, gfp_t gfp, int node) } #endif -static bool hugepage_vma_check(struct vm_area_struct *vma) +static bool hugepage_vma_check(struct vm_area_struct *vma, + unsigned long vm_flags) { - if ((!(vma->vm_flags & VM_HUGEPAGE) && !khugepaged_always()) || - (vma->vm_flags & VM_NOHUGEPAGE) || + if ((!(vm_flags & VM_HUGEPAGE) && !khugepaged_always()) || + (vm_flags & VM_NOHUGEPAGE) || test_bit(MMF_DISABLE_THP, &vma->vm_mm->flags)) return false; if (shmem_file(vma->vm_file)) { @@ -835,7 +833,7 @@ static bool hugepage_vma_check(struct vm_area_struct *vma) return false; if (is_vma_temporary_stack(vma)) return false; - return !(vma->vm_flags & VM_NO_KHUGEPAGED); + return !(vm_flags & VM_NO_KHUGEPAGED); } /* @@ -862,7 +860,7 @@ static int hugepage_vma_revalidate(struct mm_struct *mm, unsigned long address, hend = vma->vm_end & HPAGE_PMD_MASK; if (address < hstart || address + HPAGE_PMD_SIZE > hend) return SCAN_ADDRESS_RANGE; - if (!hugepage_vma_check(vma)) + if (!hugepage_vma_check(vma, vma->vm_flags)) return SCAN_VMA_CHECK; return 0; } @@ -1694,7 +1692,7 @@ static unsigned int khugepaged_scan_mm_slot(unsigned int pages, progress++; break; } - if (!hugepage_vma_check(vma)) { + if (!hugepage_vma_check(vma, vma->vm_flags)) { skip: progress++; continue;
khugepaged_enter_vma_merge() is using a different approach to check whether a vma is valid for khugepaged_enter(): if (!vma->anon_vma) /* * Not yet faulted in so we will register later in the * page fault if needed. */ return 0; if (vma->vm_ops || (vm_flags & VM_NO_KHUGEPAGED)) /* khugepaged not yet working on file or special mappings */ return 0; This check has some problems. One of the obvious problems is that it doesn't check shmem_file(), so that vma backed with shmem files will not call khugepaged_enter(). Here is an example of failed madvise(): /* mount /dev/shm with huge=advise: * mount -o remount,huge=advise /dev/shm */ /* create file /dev/shm/huge */ #define HUGE_FILE "/dev/shm/huge" fd = open(HUGE_FILE, O_RDONLY); ptr = mmap(NULL, FILE_SIZE, PROT_READ, MAP_PRIVATE, fd, 0); ret = madvise(ptr, FILE_SIZE, MADV_HUGEPAGE); madvise() will return 0, but this memory region is never put in huge page (check from /proc/meminfo: ShmemHugePages). This patch fixes these problems by reusing hugepage_vma_check() in khugepaged_enter_vma_merge(). vma->vm_flags is not yet updated in khugepaged_enter_vma_merge(), so we need to pass the new vm_flags to hugepage_vma_check() through a separate argument. Signed-off-by: Song Liu <songliubraving@fb.com> --- mm/khugepaged.c | 26 ++++++++++++-------------- 1 file changed, 12 insertions(+), 14 deletions(-)