| Message ID | 20230305065112.1932255-2-jiaqiyan@google.com (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | Memory poison recovery in khugepaged collapsing | expand |
ping for review :) On Sat, Mar 4, 2023 at 10:51 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > Make __collapse_huge_page_copy return whether copying anonymous pages > succeeded, and make collapse_huge_page handle the return status. > > Break existing PTE scan loop into two for-loops. The first loop copies > source pages into target huge page, and can fail gracefully when running > into memory errors in source pages. If copying all pages succeeds, the > second loop releases and clears up these normal pages. Otherwise, the > second loop rolls back the page table and page states by: > - re-establishing the original PTEs-to-PMD connection. > - releasing source pages back to their LRU list. > > Tested manually: > 0. Enable khugepaged on system under test. > 1. Start a two-thread application. Each thread allocates a chunk of > non-huge anonymous memory buffer. > 2. Pick 4 random buffer locations (2 in each thread) and inject > uncorrectable memory errors at corresponding physical addresses. > 3. Signal both threads to make their memory buffer collapsible, i.e. > calling madvise(MADV_HUGEPAGE). > 4. Wait and check kernel log: khugepaged is able to recover from poisoned > pages and skips collapsing them. > 5. Signal both threads to inspect their buffer contents and make sure no > data corruption. > > Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> > --- > include/trace/events/huge_memory.h | 3 +- > mm/khugepaged.c | 148 ++++++++++++++++++++++++----- > 2 files changed, 128 insertions(+), 23 deletions(-) > > diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h > index 3e6fb05852f9a..46cce509957ba 100644 > --- a/include/trace/events/huge_memory.h > +++ b/include/trace/events/huge_memory.h > @@ -36,7 +36,8 @@ > EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ > EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ > EM( SCAN_TRUNCATED, "truncated") \ > - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ > > #undef EM > #undef EMe > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > index 27956d4404134..c3c217f6ebc6e 100644 > --- a/mm/khugepaged.c > +++ b/mm/khugepaged.c > @@ -19,6 +19,7 @@ > #include <linux/page_table_check.h> > #include <linux/swapops.h> > #include <linux/shmem_fs.h> > +#include <linux/kmsan.h> > > #include <asm/tlb.h> > #include <asm/pgalloc.h> > @@ -55,6 +56,7 @@ enum scan_result { > SCAN_CGROUP_CHARGE_FAIL, > SCAN_TRUNCATED, > SCAN_PAGE_HAS_PRIVATE, > + SCAN_COPY_MC, > }; > > #define CREATE_TRACE_POINTS > @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, > return result; > } > > -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > - struct vm_area_struct *vma, > - unsigned long address, > - spinlock_t *ptl, > - struct list_head *compound_pagelist) > +static void __collapse_huge_page_copy_succeeded(pte_t *pte, > + pmd_t *pmd, > + struct vm_area_struct *vma, > + unsigned long address, > + spinlock_t *pte_ptl, > + struct list_head *compound_pagelist) > { > struct page *src_page, *tmp; > pte_t *_pte; > - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; > - _pte++, page++, address += PAGE_SIZE) { > - pte_t pteval = *_pte; > + pte_t pteval; > + unsigned long _address; > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > + _pte++, _address += PAGE_SIZE) { > + pteval = *_pte; > if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > - clear_user_highpage(page, address); > add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); > if (is_zero_pfn(pte_pfn(pteval))) { > /* > - * ptl mostly unnecessary. > + * pte_ptl mostly unnecessary. > */ > - spin_lock(ptl); > - ptep_clear(vma->vm_mm, address, _pte); > - spin_unlock(ptl); > + spin_lock(pte_ptl); > + pte_clear(vma->vm_mm, _address, _pte); > + spin_unlock(pte_ptl); > } > } else { > src_page = pte_page(pteval); > - copy_user_highpage(page, src_page, address, vma); > if (!PageCompound(src_page)) > release_pte_page(src_page); > /* > - * ptl mostly unnecessary, but preempt has to > - * be disabled to update the per-cpu stats > + * pte_ptl mostly unnecessary, but preempt has > + * to be disabled to update the per-cpu stats > * inside page_remove_rmap(). > */ > - spin_lock(ptl); > - ptep_clear(vma->vm_mm, address, _pte); > + spin_lock(pte_ptl); > + ptep_clear(vma->vm_mm, _address, _pte); > page_remove_rmap(src_page, vma, false); > - spin_unlock(ptl); > + spin_unlock(pte_ptl); > free_page_and_swap_cache(src_page); > } > } > - > list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > list_del(&src_page->lru); > mod_node_page_state(page_pgdat(src_page), > @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > } > } > > +static void __collapse_huge_page_copy_failed(pte_t *pte, > + pmd_t *pmd, > + pmd_t orig_pmd, > + struct vm_area_struct *vma, > + unsigned long address, > + struct list_head *compound_pagelist) > +{ > + struct page *src_page, *tmp; > + pte_t *_pte; > + pte_t pteval; > + unsigned long _address; > + spinlock_t *pmd_ptl; > + > + /* > + * Re-establish the PMD to point to the original page table > + * entry. Restoring PMD needs to be done prior to releasing > + * pages. Since pages are still isolated and locked here, > + * acquiring anon_vma_lock_write is unnecessary. > + */ > + pmd_ptl = pmd_lock(vma->vm_mm, pmd); > + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); > + spin_unlock(pmd_ptl); > + /* > + * Release both raw and compound pages isolated > + * in __collapse_huge_page_isolate. > + */ > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > + _pte++, _address += PAGE_SIZE) { > + pteval = *_pte; > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) > + continue; > + src_page = pte_page(pteval); > + if (!PageCompound(src_page)) > + release_pte_page(src_page); > + } > + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > + list_del(&src_page->lru); > + release_pte_page(src_page); > + } > +} > + > +/* > + * __collapse_huge_page_copy - attempts to copy memory contents from raw > + * pages to a hugepage. Cleans up the raw pages if copying succeeds; > + * otherwise restores the original page table and releases isolated raw pages. > + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. > + * > + * @pte: starting of the PTEs to copy from > + * @page: the new hugepage to copy contents to > + * @pmd: pointer to the new hugepage's PMD > + * @orig_pmd: the original raw pages' PMD > + * @vma: the original raw pages' virtual memory area > + * @address: starting address to copy > + * @pte_ptl: lock on raw pages' PTEs > + * @compound_pagelist: list that stores compound pages > + */ > +static int __collapse_huge_page_copy(pte_t *pte, > + struct page *page, > + pmd_t *pmd, > + pmd_t orig_pmd, > + struct vm_area_struct *vma, > + unsigned long address, > + spinlock_t *pte_ptl, > + struct list_head *compound_pagelist) > +{ > + struct page *src_page; > + pte_t *_pte; > + pte_t pteval; > + unsigned long _address; > + int result = SCAN_SUCCEED; > + > + /* > + * Copying pages' contents is subject to memory poison at any iteration. > + */ > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > + _pte++, page++, _address += PAGE_SIZE) { > + pteval = *_pte; > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > + clear_user_highpage(page, _address); > + continue; > + } > + src_page = pte_page(pteval); > + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { > + result = SCAN_COPY_MC; > + break; > + } > + } > + > + if (likely(result == SCAN_SUCCEED)) > + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, > + pte_ptl, compound_pagelist); > + else > + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, > + address, compound_pagelist); > + > + return result; > +} > + > static void khugepaged_alloc_sleep(void) > { > DEFINE_WAIT(wait); > @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, > */ > anon_vma_unlock_write(vma->anon_vma); > > - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, > - &compound_pagelist); > + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, > + vma, address, pte_ptl, > + &compound_pagelist); > pte_unmap(pte); > + if (unlikely(result != SCAN_SUCCEED)) > + goto out_up_write; > + > /* > * spin_lock() below is not the equivalent of smp_wmb(), but > * the smp_wmb() inside __SetPageUptodate() can be reused to > -- > 2.40.0.rc0.216.gc4246ad0f0-goog >
On Mon, Mar 20, 2023 at 7:42 AM Jiaqi Yan <jiaqiyan@google.com> wrote: > > ping for review :) Quite busy recently so I didn't spend too much time on the recent revisions. Hopefully I can find some time this week. > > On Sat, Mar 4, 2023 at 10:51 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > > > Make __collapse_huge_page_copy return whether copying anonymous pages > > succeeded, and make collapse_huge_page handle the return status. > > > > Break existing PTE scan loop into two for-loops. The first loop copies > > source pages into target huge page, and can fail gracefully when running > > into memory errors in source pages. If copying all pages succeeds, the > > second loop releases and clears up these normal pages. Otherwise, the > > second loop rolls back the page table and page states by: > > - re-establishing the original PTEs-to-PMD connection. > > - releasing source pages back to their LRU list. > > > > Tested manually: > > 0. Enable khugepaged on system under test. > > 1. Start a two-thread application. Each thread allocates a chunk of > > non-huge anonymous memory buffer. > > 2. Pick 4 random buffer locations (2 in each thread) and inject > > uncorrectable memory errors at corresponding physical addresses. > > 3. Signal both threads to make their memory buffer collapsible, i.e. > > calling madvise(MADV_HUGEPAGE). > > 4. Wait and check kernel log: khugepaged is able to recover from poisoned > > pages and skips collapsing them. > > 5. Signal both threads to inspect their buffer contents and make sure no > > data corruption. > > > > Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> > > --- > > include/trace/events/huge_memory.h | 3 +- > > mm/khugepaged.c | 148 ++++++++++++++++++++++++----- > > 2 files changed, 128 insertions(+), 23 deletions(-) > > > > diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h > > index 3e6fb05852f9a..46cce509957ba 100644 > > --- a/include/trace/events/huge_memory.h > > +++ b/include/trace/events/huge_memory.h > > @@ -36,7 +36,8 @@ > > EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ > > EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ > > EM( SCAN_TRUNCATED, "truncated") \ > > - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ > > > > #undef EM > > #undef EMe > > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > > index 27956d4404134..c3c217f6ebc6e 100644 > > --- a/mm/khugepaged.c > > +++ b/mm/khugepaged.c > > @@ -19,6 +19,7 @@ > > #include <linux/page_table_check.h> > > #include <linux/swapops.h> > > #include <linux/shmem_fs.h> > > +#include <linux/kmsan.h> > > > > #include <asm/tlb.h> > > #include <asm/pgalloc.h> > > @@ -55,6 +56,7 @@ enum scan_result { > > SCAN_CGROUP_CHARGE_FAIL, > > SCAN_TRUNCATED, > > SCAN_PAGE_HAS_PRIVATE, > > + SCAN_COPY_MC, > > }; > > > > #define CREATE_TRACE_POINTS > > @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, > > return result; > > } > > > > -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > - struct vm_area_struct *vma, > > - unsigned long address, > > - spinlock_t *ptl, > > - struct list_head *compound_pagelist) > > +static void __collapse_huge_page_copy_succeeded(pte_t *pte, > > + pmd_t *pmd, > > + struct vm_area_struct *vma, > > + unsigned long address, > > + spinlock_t *pte_ptl, > > + struct list_head *compound_pagelist) > > { > > struct page *src_page, *tmp; > > pte_t *_pte; > > - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; > > - _pte++, page++, address += PAGE_SIZE) { > > - pte_t pteval = *_pte; > > + pte_t pteval; > > + unsigned long _address; > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > + _pte++, _address += PAGE_SIZE) { > > + pteval = *_pte; > > if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > - clear_user_highpage(page, address); > > add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); > > if (is_zero_pfn(pte_pfn(pteval))) { > > /* > > - * ptl mostly unnecessary. > > + * pte_ptl mostly unnecessary. > > */ > > - spin_lock(ptl); > > - ptep_clear(vma->vm_mm, address, _pte); > > - spin_unlock(ptl); > > + spin_lock(pte_ptl); > > + pte_clear(vma->vm_mm, _address, _pte); > > + spin_unlock(pte_ptl); > > } > > } else { > > src_page = pte_page(pteval); > > - copy_user_highpage(page, src_page, address, vma); > > if (!PageCompound(src_page)) > > release_pte_page(src_page); > > /* > > - * ptl mostly unnecessary, but preempt has to > > - * be disabled to update the per-cpu stats > > + * pte_ptl mostly unnecessary, but preempt has > > + * to be disabled to update the per-cpu stats > > * inside page_remove_rmap(). > > */ > > - spin_lock(ptl); > > - ptep_clear(vma->vm_mm, address, _pte); > > + spin_lock(pte_ptl); > > + ptep_clear(vma->vm_mm, _address, _pte); > > page_remove_rmap(src_page, vma, false); > > - spin_unlock(ptl); > > + spin_unlock(pte_ptl); > > free_page_and_swap_cache(src_page); > > } > > } > > - > > list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > list_del(&src_page->lru); > > mod_node_page_state(page_pgdat(src_page), > > @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > } > > } > > > > +static void __collapse_huge_page_copy_failed(pte_t *pte, > > + pmd_t *pmd, > > + pmd_t orig_pmd, > > + struct vm_area_struct *vma, > > + unsigned long address, > > + struct list_head *compound_pagelist) > > +{ > > + struct page *src_page, *tmp; > > + pte_t *_pte; > > + pte_t pteval; > > + unsigned long _address; > > + spinlock_t *pmd_ptl; > > + > > + /* > > + * Re-establish the PMD to point to the original page table > > + * entry. Restoring PMD needs to be done prior to releasing > > + * pages. Since pages are still isolated and locked here, > > + * acquiring anon_vma_lock_write is unnecessary. > > + */ > > + pmd_ptl = pmd_lock(vma->vm_mm, pmd); > > + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); > > + spin_unlock(pmd_ptl); > > + /* > > + * Release both raw and compound pages isolated > > + * in __collapse_huge_page_isolate. > > + */ > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > + _pte++, _address += PAGE_SIZE) { > > + pteval = *_pte; > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) > > + continue; > > + src_page = pte_page(pteval); > > + if (!PageCompound(src_page)) > > + release_pte_page(src_page); > > + } > > + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > + list_del(&src_page->lru); > > + release_pte_page(src_page); > > + } > > +} > > + > > +/* > > + * __collapse_huge_page_copy - attempts to copy memory contents from raw > > + * pages to a hugepage. Cleans up the raw pages if copying succeeds; > > + * otherwise restores the original page table and releases isolated raw pages. > > + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. > > + * > > + * @pte: starting of the PTEs to copy from > > + * @page: the new hugepage to copy contents to > > + * @pmd: pointer to the new hugepage's PMD > > + * @orig_pmd: the original raw pages' PMD > > + * @vma: the original raw pages' virtual memory area > > + * @address: starting address to copy > > + * @pte_ptl: lock on raw pages' PTEs > > + * @compound_pagelist: list that stores compound pages > > + */ > > +static int __collapse_huge_page_copy(pte_t *pte, > > + struct page *page, > > + pmd_t *pmd, > > + pmd_t orig_pmd, > > + struct vm_area_struct *vma, > > + unsigned long address, > > + spinlock_t *pte_ptl, > > + struct list_head *compound_pagelist) > > +{ > > + struct page *src_page; > > + pte_t *_pte; > > + pte_t pteval; > > + unsigned long _address; > > + int result = SCAN_SUCCEED; > > + > > + /* > > + * Copying pages' contents is subject to memory poison at any iteration. > > + */ > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > + _pte++, page++, _address += PAGE_SIZE) { > > + pteval = *_pte; > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > + clear_user_highpage(page, _address); > > + continue; > > + } > > + src_page = pte_page(pteval); > > + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { > > + result = SCAN_COPY_MC; > > + break; > > + } > > + } > > + > > + if (likely(result == SCAN_SUCCEED)) > > + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, > > + pte_ptl, compound_pagelist); > > + else > > + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, > > + address, compound_pagelist); > > + > > + return result; > > +} > > + > > static void khugepaged_alloc_sleep(void) > > { > > DEFINE_WAIT(wait); > > @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, > > */ > > anon_vma_unlock_write(vma->anon_vma); > > > > - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, > > - &compound_pagelist); > > + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, > > + vma, address, pte_ptl, > > + &compound_pagelist); > > pte_unmap(pte); > > + if (unlikely(result != SCAN_SUCCEED)) > > + goto out_up_write; > > + > > /* > > * spin_lock() below is not the equivalent of smp_wmb(), but > > * the smp_wmb() inside __SetPageUptodate() can be reused to > > -- > > 2.40.0.rc0.216.gc4246ad0f0-goog > >
On Sat, Mar 4, 2023 at 10:51 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > Make __collapse_huge_page_copy return whether copying anonymous pages > succeeded, and make collapse_huge_page handle the return status. > > Break existing PTE scan loop into two for-loops. The first loop copies > source pages into target huge page, and can fail gracefully when running > into memory errors in source pages. If copying all pages succeeds, the > second loop releases and clears up these normal pages. Otherwise, the > second loop rolls back the page table and page states by: > - re-establishing the original PTEs-to-PMD connection. > - releasing source pages back to their LRU list. > > Tested manually: > 0. Enable khugepaged on system under test. > 1. Start a two-thread application. Each thread allocates a chunk of > non-huge anonymous memory buffer. > 2. Pick 4 random buffer locations (2 in each thread) and inject > uncorrectable memory errors at corresponding physical addresses. > 3. Signal both threads to make their memory buffer collapsible, i.e. > calling madvise(MADV_HUGEPAGE). > 4. Wait and check kernel log: khugepaged is able to recover from poisoned > pages and skips collapsing them. > 5. Signal both threads to inspect their buffer contents and make sure no > data corruption. > > Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> > --- > include/trace/events/huge_memory.h | 3 +- > mm/khugepaged.c | 148 ++++++++++++++++++++++++----- > 2 files changed, 128 insertions(+), 23 deletions(-) > > diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h > index 3e6fb05852f9a..46cce509957ba 100644 > --- a/include/trace/events/huge_memory.h > +++ b/include/trace/events/huge_memory.h > @@ -36,7 +36,8 @@ > EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ > EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ > EM( SCAN_TRUNCATED, "truncated") \ > - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ > > #undef EM > #undef EMe > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > index 27956d4404134..c3c217f6ebc6e 100644 > --- a/mm/khugepaged.c > +++ b/mm/khugepaged.c > @@ -19,6 +19,7 @@ > #include <linux/page_table_check.h> > #include <linux/swapops.h> > #include <linux/shmem_fs.h> > +#include <linux/kmsan.h> > > #include <asm/tlb.h> > #include <asm/pgalloc.h> > @@ -55,6 +56,7 @@ enum scan_result { > SCAN_CGROUP_CHARGE_FAIL, > SCAN_TRUNCATED, > SCAN_PAGE_HAS_PRIVATE, > + SCAN_COPY_MC, > }; > > #define CREATE_TRACE_POINTS > @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, > return result; > } > > -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > - struct vm_area_struct *vma, > - unsigned long address, > - spinlock_t *ptl, > - struct list_head *compound_pagelist) > +static void __collapse_huge_page_copy_succeeded(pte_t *pte, > + pmd_t *pmd, > + struct vm_area_struct *vma, > + unsigned long address, > + spinlock_t *pte_ptl, > + struct list_head *compound_pagelist) > { > struct page *src_page, *tmp; > pte_t *_pte; > - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; > - _pte++, page++, address += PAGE_SIZE) { > - pte_t pteval = *_pte; > + pte_t pteval; > + unsigned long _address; > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > + _pte++, _address += PAGE_SIZE) { > + pteval = *_pte; > if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > - clear_user_highpage(page, address); > add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); > if (is_zero_pfn(pte_pfn(pteval))) { > /* > - * ptl mostly unnecessary. > + * pte_ptl mostly unnecessary. > */ > - spin_lock(ptl); > - ptep_clear(vma->vm_mm, address, _pte); > - spin_unlock(ptl); > + spin_lock(pte_ptl); Why did you have to rename ptl to pte_ptl? It seems unnecessary. > + pte_clear(vma->vm_mm, _address, _pte); > + spin_unlock(pte_ptl); > } > } else { > src_page = pte_page(pteval); > - copy_user_highpage(page, src_page, address, vma); > if (!PageCompound(src_page)) > release_pte_page(src_page); > /* > - * ptl mostly unnecessary, but preempt has to > - * be disabled to update the per-cpu stats > + * pte_ptl mostly unnecessary, but preempt has > + * to be disabled to update the per-cpu stats > * inside page_remove_rmap(). > */ > - spin_lock(ptl); > - ptep_clear(vma->vm_mm, address, _pte); > + spin_lock(pte_ptl); > + ptep_clear(vma->vm_mm, _address, _pte); > page_remove_rmap(src_page, vma, false); > - spin_unlock(ptl); > + spin_unlock(pte_ptl); > free_page_and_swap_cache(src_page); > } > } > - > list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > list_del(&src_page->lru); > mod_node_page_state(page_pgdat(src_page), > @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > } > } > > +static void __collapse_huge_page_copy_failed(pte_t *pte, > + pmd_t *pmd, > + pmd_t orig_pmd, > + struct vm_area_struct *vma, > + unsigned long address, > + struct list_head *compound_pagelist) > +{ > + struct page *src_page, *tmp; > + pte_t *_pte; > + pte_t pteval; > + unsigned long _address; > + spinlock_t *pmd_ptl; > + > + /* > + * Re-establish the PMD to point to the original page table > + * entry. Restoring PMD needs to be done prior to releasing > + * pages. Since pages are still isolated and locked here, > + * acquiring anon_vma_lock_write is unnecessary. > + */ > + pmd_ptl = pmd_lock(vma->vm_mm, pmd); > + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); > + spin_unlock(pmd_ptl); > + /* > + * Release both raw and compound pages isolated > + * in __collapse_huge_page_isolate. > + */ It looks like the below code could be replaced by release_pte_pages() with advancing _pte to (pte + HPAGE_PMD_NR - 1). > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > + _pte++, _address += PAGE_SIZE) { > + pteval = *_pte; > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) > + continue; > + src_page = pte_page(pteval); > + if (!PageCompound(src_page)) > + release_pte_page(src_page); > + } > + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > + list_del(&src_page->lru); > + release_pte_page(src_page); > + } > +} > + > +/* > + * __collapse_huge_page_copy - attempts to copy memory contents from raw > + * pages to a hugepage. Cleans up the raw pages if copying succeeds; > + * otherwise restores the original page table and releases isolated raw pages. > + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. > + * > + * @pte: starting of the PTEs to copy from > + * @page: the new hugepage to copy contents to > + * @pmd: pointer to the new hugepage's PMD > + * @orig_pmd: the original raw pages' PMD > + * @vma: the original raw pages' virtual memory area > + * @address: starting address to copy > + * @pte_ptl: lock on raw pages' PTEs > + * @compound_pagelist: list that stores compound pages > + */ > +static int __collapse_huge_page_copy(pte_t *pte, > + struct page *page, > + pmd_t *pmd, > + pmd_t orig_pmd, > + struct vm_area_struct *vma, > + unsigned long address, > + spinlock_t *pte_ptl, > + struct list_head *compound_pagelist) > +{ > + struct page *src_page; > + pte_t *_pte; > + pte_t pteval; > + unsigned long _address; > + int result = SCAN_SUCCEED; > + > + /* > + * Copying pages' contents is subject to memory poison at any iteration. > + */ > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > + _pte++, page++, _address += PAGE_SIZE) { > + pteval = *_pte; > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > + clear_user_highpage(page, _address); > + continue; > + } > + src_page = pte_page(pteval); > + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { > + result = SCAN_COPY_MC; > + break; > + } > + } > + > + if (likely(result == SCAN_SUCCEED)) > + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, > + pte_ptl, compound_pagelist); > + else > + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, > + address, compound_pagelist); > + > + return result; > +} > + > static void khugepaged_alloc_sleep(void) > { > DEFINE_WAIT(wait); > @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, > */ > anon_vma_unlock_write(vma->anon_vma); > > - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, > - &compound_pagelist); > + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, > + vma, address, pte_ptl, > + &compound_pagelist); > pte_unmap(pte); > + if (unlikely(result != SCAN_SUCCEED)) > + goto out_up_write; > + > /* > * spin_lock() below is not the equivalent of smp_wmb(), but > * the smp_wmb() inside __SetPageUptodate() can be reused to > -- > 2.40.0.rc0.216.gc4246ad0f0-goog >
On Thu, Mar 23, 2023 at 2:38 PM Yang Shi <shy828301@gmail.com> wrote: > > On Sat, Mar 4, 2023 at 10:51 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > > > Make __collapse_huge_page_copy return whether copying anonymous pages > > succeeded, and make collapse_huge_page handle the return status. > > > > Break existing PTE scan loop into two for-loops. The first loop copies > > source pages into target huge page, and can fail gracefully when running > > into memory errors in source pages. If copying all pages succeeds, the > > second loop releases and clears up these normal pages. Otherwise, the > > second loop rolls back the page table and page states by: > > - re-establishing the original PTEs-to-PMD connection. > > - releasing source pages back to their LRU list. > > > > Tested manually: > > 0. Enable khugepaged on system under test. > > 1. Start a two-thread application. Each thread allocates a chunk of > > non-huge anonymous memory buffer. > > 2. Pick 4 random buffer locations (2 in each thread) and inject > > uncorrectable memory errors at corresponding physical addresses. > > 3. Signal both threads to make their memory buffer collapsible, i.e. > > calling madvise(MADV_HUGEPAGE). > > 4. Wait and check kernel log: khugepaged is able to recover from poisoned > > pages and skips collapsing them. > > 5. Signal both threads to inspect their buffer contents and make sure no > > data corruption. > > > > Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> > > --- > > include/trace/events/huge_memory.h | 3 +- > > mm/khugepaged.c | 148 ++++++++++++++++++++++++----- > > 2 files changed, 128 insertions(+), 23 deletions(-) > > > > diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h > > index 3e6fb05852f9a..46cce509957ba 100644 > > --- a/include/trace/events/huge_memory.h > > +++ b/include/trace/events/huge_memory.h > > @@ -36,7 +36,8 @@ > > EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ > > EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ > > EM( SCAN_TRUNCATED, "truncated") \ > > - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ > > > > #undef EM > > #undef EMe > > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > > index 27956d4404134..c3c217f6ebc6e 100644 > > --- a/mm/khugepaged.c > > +++ b/mm/khugepaged.c > > @@ -19,6 +19,7 @@ > > #include <linux/page_table_check.h> > > #include <linux/swapops.h> > > #include <linux/shmem_fs.h> > > +#include <linux/kmsan.h> > > > > #include <asm/tlb.h> > > #include <asm/pgalloc.h> > > @@ -55,6 +56,7 @@ enum scan_result { > > SCAN_CGROUP_CHARGE_FAIL, > > SCAN_TRUNCATED, > > SCAN_PAGE_HAS_PRIVATE, > > + SCAN_COPY_MC, > > }; > > > > #define CREATE_TRACE_POINTS > > @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, > > return result; > > } > > > > -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > - struct vm_area_struct *vma, > > - unsigned long address, > > - spinlock_t *ptl, > > - struct list_head *compound_pagelist) > > +static void __collapse_huge_page_copy_succeeded(pte_t *pte, > > + pmd_t *pmd, > > + struct vm_area_struct *vma, > > + unsigned long address, > > + spinlock_t *pte_ptl, > > + struct list_head *compound_pagelist) > > { > > struct page *src_page, *tmp; > > pte_t *_pte; > > - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; > > - _pte++, page++, address += PAGE_SIZE) { > > - pte_t pteval = *_pte; > > + pte_t pteval; > > + unsigned long _address; > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > + _pte++, _address += PAGE_SIZE) { > > + pteval = *_pte; > > if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > - clear_user_highpage(page, address); > > add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); > > if (is_zero_pfn(pte_pfn(pteval))) { > > /* > > - * ptl mostly unnecessary. > > + * pte_ptl mostly unnecessary. > > */ > > - spin_lock(ptl); > > - ptep_clear(vma->vm_mm, address, _pte); > > - spin_unlock(ptl); > > + spin_lock(pte_ptl); > > Why did you have to rename ptl to pte_ptl? It seems unnecessary. Thanks, I will use `ptl` in the next version. > > > + pte_clear(vma->vm_mm, _address, _pte); > > + spin_unlock(pte_ptl); > > } > > } else { > > src_page = pte_page(pteval); > > - copy_user_highpage(page, src_page, address, vma); > > if (!PageCompound(src_page)) > > release_pte_page(src_page); > > /* > > - * ptl mostly unnecessary, but preempt has to > > - * be disabled to update the per-cpu stats > > + * pte_ptl mostly unnecessary, but preempt has > > + * to be disabled to update the per-cpu stats > > * inside page_remove_rmap(). > > */ > > - spin_lock(ptl); > > - ptep_clear(vma->vm_mm, address, _pte); > > + spin_lock(pte_ptl); > > + ptep_clear(vma->vm_mm, _address, _pte); > > page_remove_rmap(src_page, vma, false); > > - spin_unlock(ptl); > > + spin_unlock(pte_ptl); > > free_page_and_swap_cache(src_page); > > } > > } > > - > > list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > list_del(&src_page->lru); > > mod_node_page_state(page_pgdat(src_page), > > @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > } > > } > > > > +static void __collapse_huge_page_copy_failed(pte_t *pte, > > + pmd_t *pmd, > > + pmd_t orig_pmd, > > + struct vm_area_struct *vma, > > + unsigned long address, > > + struct list_head *compound_pagelist) > > +{ > > + struct page *src_page, *tmp; > > + pte_t *_pte; > > + pte_t pteval; > > + unsigned long _address; > > + spinlock_t *pmd_ptl; > > + > > + /* > > + * Re-establish the PMD to point to the original page table > > + * entry. Restoring PMD needs to be done prior to releasing > > + * pages. Since pages are still isolated and locked here, > > + * acquiring anon_vma_lock_write is unnecessary. > > + */ > > + pmd_ptl = pmd_lock(vma->vm_mm, pmd); > > + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); > > + spin_unlock(pmd_ptl); > > + /* > > + * Release both raw and compound pages isolated > > + * in __collapse_huge_page_isolate. > > + */ > > It looks like the below code could be replaced by release_pte_pages() > with advancing _pte to (pte + HPAGE_PMD_NR - 1). > Yeah, but I think _pte should be (pte + HPAGE_PMR_NR) because _pte is decremented before comparison in release_pte_pages(pte, _pte, compound_pagelist): while (--_pte >= pte) {...} Advancing _pte to (pte + HPAGE_PMD_NR - 1) may leave the last page not released. > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > + _pte++, _address += PAGE_SIZE) { > > + pteval = *_pte; > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) > > + continue; > > + src_page = pte_page(pteval); > > + if (!PageCompound(src_page)) > > + release_pte_page(src_page); > > + } > > + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > + list_del(&src_page->lru); > > + release_pte_page(src_page); > > + } > > +} > > + > > +/* > > + * __collapse_huge_page_copy - attempts to copy memory contents from raw > > + * pages to a hugepage. Cleans up the raw pages if copying succeeds; > > + * otherwise restores the original page table and releases isolated raw pages. > > + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. > > + * > > + * @pte: starting of the PTEs to copy from > > + * @page: the new hugepage to copy contents to > > + * @pmd: pointer to the new hugepage's PMD > > + * @orig_pmd: the original raw pages' PMD > > + * @vma: the original raw pages' virtual memory area > > + * @address: starting address to copy > > + * @pte_ptl: lock on raw pages' PTEs > > + * @compound_pagelist: list that stores compound pages > > + */ > > +static int __collapse_huge_page_copy(pte_t *pte, > > + struct page *page, > > + pmd_t *pmd, > > + pmd_t orig_pmd, > > + struct vm_area_struct *vma, > > + unsigned long address, > > + spinlock_t *pte_ptl, > > + struct list_head *compound_pagelist) > > +{ > > + struct page *src_page; > > + pte_t *_pte; > > + pte_t pteval; > > + unsigned long _address; > > + int result = SCAN_SUCCEED; > > + > > + /* > > + * Copying pages' contents is subject to memory poison at any iteration. > > + */ > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > + _pte++, page++, _address += PAGE_SIZE) { > > + pteval = *_pte; > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > + clear_user_highpage(page, _address); > > + continue; > > + } > > + src_page = pte_page(pteval); > > + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { > > + result = SCAN_COPY_MC; > > + break; > > + } > > + } > > + > > + if (likely(result == SCAN_SUCCEED)) > > + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, > > + pte_ptl, compound_pagelist); > > + else > > + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, > > + address, compound_pagelist); > > + > > + return result; > > +} > > + > > static void khugepaged_alloc_sleep(void) > > { > > DEFINE_WAIT(wait); > > @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, > > */ > > anon_vma_unlock_write(vma->anon_vma); > > > > - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, > > - &compound_pagelist); > > + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, > > + vma, address, pte_ptl, > > + &compound_pagelist); > > pte_unmap(pte); > > + if (unlikely(result != SCAN_SUCCEED)) > > + goto out_up_write; > > + > > /* > > * spin_lock() below is not the equivalent of smp_wmb(), but > > * the smp_wmb() inside __SetPageUptodate() can be reused to > > -- > > 2.40.0.rc0.216.gc4246ad0f0-goog > >
On Fri, Mar 24, 2023 at 8:34 AM Jiaqi Yan <jiaqiyan@google.com> wrote: > > On Thu, Mar 23, 2023 at 2:38 PM Yang Shi <shy828301@gmail.com> wrote: > > > > On Sat, Mar 4, 2023 at 10:51 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > > > > > Make __collapse_huge_page_copy return whether copying anonymous pages > > > succeeded, and make collapse_huge_page handle the return status. > > > > > > Break existing PTE scan loop into two for-loops. The first loop copies > > > source pages into target huge page, and can fail gracefully when running > > > into memory errors in source pages. If copying all pages succeeds, the > > > second loop releases and clears up these normal pages. Otherwise, the > > > second loop rolls back the page table and page states by: > > > - re-establishing the original PTEs-to-PMD connection. > > > - releasing source pages back to their LRU list. > > > > > > Tested manually: > > > 0. Enable khugepaged on system under test. > > > 1. Start a two-thread application. Each thread allocates a chunk of > > > non-huge anonymous memory buffer. > > > 2. Pick 4 random buffer locations (2 in each thread) and inject > > > uncorrectable memory errors at corresponding physical addresses. > > > 3. Signal both threads to make their memory buffer collapsible, i.e. > > > calling madvise(MADV_HUGEPAGE). > > > 4. Wait and check kernel log: khugepaged is able to recover from poisoned > > > pages and skips collapsing them. > > > 5. Signal both threads to inspect their buffer contents and make sure no > > > data corruption. > > > > > > Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> > > > --- > > > include/trace/events/huge_memory.h | 3 +- > > > mm/khugepaged.c | 148 ++++++++++++++++++++++++----- > > > 2 files changed, 128 insertions(+), 23 deletions(-) > > > > > > diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h > > > index 3e6fb05852f9a..46cce509957ba 100644 > > > --- a/include/trace/events/huge_memory.h > > > +++ b/include/trace/events/huge_memory.h > > > @@ -36,7 +36,8 @@ > > > EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ > > > EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ > > > EM( SCAN_TRUNCATED, "truncated") \ > > > - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > > + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > > + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ > > > > > > #undef EM > > > #undef EMe > > > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > > > index 27956d4404134..c3c217f6ebc6e 100644 > > > --- a/mm/khugepaged.c > > > +++ b/mm/khugepaged.c > > > @@ -19,6 +19,7 @@ > > > #include <linux/page_table_check.h> > > > #include <linux/swapops.h> > > > #include <linux/shmem_fs.h> > > > +#include <linux/kmsan.h> > > > > > > #include <asm/tlb.h> > > > #include <asm/pgalloc.h> > > > @@ -55,6 +56,7 @@ enum scan_result { > > > SCAN_CGROUP_CHARGE_FAIL, > > > SCAN_TRUNCATED, > > > SCAN_PAGE_HAS_PRIVATE, > > > + SCAN_COPY_MC, > > > }; > > > > > > #define CREATE_TRACE_POINTS > > > @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, > > > return result; > > > } > > > > > > -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > > - struct vm_area_struct *vma, > > > - unsigned long address, > > > - spinlock_t *ptl, > > > - struct list_head *compound_pagelist) > > > +static void __collapse_huge_page_copy_succeeded(pte_t *pte, > > > + pmd_t *pmd, > > > + struct vm_area_struct *vma, > > > + unsigned long address, > > > + spinlock_t *pte_ptl, > > > + struct list_head *compound_pagelist) > > > { > > > struct page *src_page, *tmp; > > > pte_t *_pte; > > > - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; > > > - _pte++, page++, address += PAGE_SIZE) { > > > - pte_t pteval = *_pte; > > > + pte_t pteval; > > > + unsigned long _address; > > > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > + _pte++, _address += PAGE_SIZE) { > > > + pteval = *_pte; > > > if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > > - clear_user_highpage(page, address); > > > add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); > > > if (is_zero_pfn(pte_pfn(pteval))) { > > > /* > > > - * ptl mostly unnecessary. > > > + * pte_ptl mostly unnecessary. > > > */ > > > - spin_lock(ptl); > > > - ptep_clear(vma->vm_mm, address, _pte); > > > - spin_unlock(ptl); > > > + spin_lock(pte_ptl); > > > > Why did you have to rename ptl to pte_ptl? It seems unnecessary. > > Thanks, I will use `ptl` in the next version. > > > > > > + pte_clear(vma->vm_mm, _address, _pte); > > > + spin_unlock(pte_ptl); > > > } > > > } else { > > > src_page = pte_page(pteval); > > > - copy_user_highpage(page, src_page, address, vma); > > > if (!PageCompound(src_page)) > > > release_pte_page(src_page); > > > /* > > > - * ptl mostly unnecessary, but preempt has to > > > - * be disabled to update the per-cpu stats > > > + * pte_ptl mostly unnecessary, but preempt has > > > + * to be disabled to update the per-cpu stats > > > * inside page_remove_rmap(). > > > */ > > > - spin_lock(ptl); > > > - ptep_clear(vma->vm_mm, address, _pte); > > > + spin_lock(pte_ptl); > > > + ptep_clear(vma->vm_mm, _address, _pte); > > > page_remove_rmap(src_page, vma, false); > > > - spin_unlock(ptl); > > > + spin_unlock(pte_ptl); > > > free_page_and_swap_cache(src_page); > > > } > > > } > > > - > > > list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > > list_del(&src_page->lru); > > > mod_node_page_state(page_pgdat(src_page), > > > @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > > } > > > } > > > > > > +static void __collapse_huge_page_copy_failed(pte_t *pte, > > > + pmd_t *pmd, > > > + pmd_t orig_pmd, > > > + struct vm_area_struct *vma, > > > + unsigned long address, > > > + struct list_head *compound_pagelist) > > > +{ > > > + struct page *src_page, *tmp; > > > + pte_t *_pte; > > > + pte_t pteval; > > > + unsigned long _address; > > > + spinlock_t *pmd_ptl; > > > + > > > + /* > > > + * Re-establish the PMD to point to the original page table > > > + * entry. Restoring PMD needs to be done prior to releasing > > > + * pages. Since pages are still isolated and locked here, > > > + * acquiring anon_vma_lock_write is unnecessary. > > > + */ > > > + pmd_ptl = pmd_lock(vma->vm_mm, pmd); > > > + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); > > > + spin_unlock(pmd_ptl); > > > + /* > > > + * Release both raw and compound pages isolated > > > + * in __collapse_huge_page_isolate. > > > + */ > > > > It looks like the below code could be replaced by release_pte_pages() > > with advancing _pte to (pte + HPAGE_PMD_NR - 1). > > > > Yeah, but I think _pte should be (pte + HPAGE_PMR_NR) because _pte is > decremented before comparison in release_pte_pages(pte, _pte, > compound_pagelist): > > while (--_pte >= pte) {...} > > Advancing _pte to (pte + HPAGE_PMD_NR - 1) may leave the last page not released. Yeah, good catch. I think it is because the only user of release_pte_pages() is __collapse_huge_page_isolate(). Once the loop in it is done _pte is pte + HPAGE_PMD_NR. > > > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > + _pte++, _address += PAGE_SIZE) { > > > + pteval = *_pte; > > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) > > > + continue; > > > + src_page = pte_page(pteval); > > > + if (!PageCompound(src_page)) > > > + release_pte_page(src_page); > > > + } > > > + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > > + list_del(&src_page->lru); > > > + release_pte_page(src_page); > > > + } > > > +} > > > + > > > +/* > > > + * __collapse_huge_page_copy - attempts to copy memory contents from raw > > > + * pages to a hugepage. Cleans up the raw pages if copying succeeds; > > > + * otherwise restores the original page table and releases isolated raw pages. > > > + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. > > > + * > > > + * @pte: starting of the PTEs to copy from > > > + * @page: the new hugepage to copy contents to > > > + * @pmd: pointer to the new hugepage's PMD > > > + * @orig_pmd: the original raw pages' PMD > > > + * @vma: the original raw pages' virtual memory area > > > + * @address: starting address to copy > > > + * @pte_ptl: lock on raw pages' PTEs > > > + * @compound_pagelist: list that stores compound pages > > > + */ > > > +static int __collapse_huge_page_copy(pte_t *pte, > > > + struct page *page, > > > + pmd_t *pmd, > > > + pmd_t orig_pmd, > > > + struct vm_area_struct *vma, > > > + unsigned long address, > > > + spinlock_t *pte_ptl, > > > + struct list_head *compound_pagelist) > > > +{ > > > + struct page *src_page; > > > + pte_t *_pte; > > > + pte_t pteval; > > > + unsigned long _address; > > > + int result = SCAN_SUCCEED; > > > + > > > + /* > > > + * Copying pages' contents is subject to memory poison at any iteration. > > > + */ > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > + _pte++, page++, _address += PAGE_SIZE) { > > > + pteval = *_pte; > > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > > + clear_user_highpage(page, _address); > > > + continue; > > > + } > > > + src_page = pte_page(pteval); > > > + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { > > > + result = SCAN_COPY_MC; > > > + break; > > > + } > > > + } > > > + > > > + if (likely(result == SCAN_SUCCEED)) > > > + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, > > > + pte_ptl, compound_pagelist); > > > + else > > > + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, > > > + address, compound_pagelist); > > > + > > > + return result; > > > +} > > > + > > > static void khugepaged_alloc_sleep(void) > > > { > > > DEFINE_WAIT(wait); > > > @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, > > > */ > > > anon_vma_unlock_write(vma->anon_vma); > > > > > > - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, > > > - &compound_pagelist); > > > + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, > > > + vma, address, pte_ptl, > > > + &compound_pagelist); > > > pte_unmap(pte); > > > + if (unlikely(result != SCAN_SUCCEED)) > > > + goto out_up_write; > > > + > > > /* > > > * spin_lock() below is not the equivalent of smp_wmb(), but > > > * the smp_wmb() inside __SetPageUptodate() can be reused to > > > -- > > > 2.40.0.rc0.216.gc4246ad0f0-goog > > >
On Fri, Mar 24, 2023 at 1:11 PM Yang Shi <shy828301@gmail.com> wrote: > > On Fri, Mar 24, 2023 at 8:34 AM Jiaqi Yan <jiaqiyan@google.com> wrote: > > > > On Thu, Mar 23, 2023 at 2:38 PM Yang Shi <shy828301@gmail.com> wrote: > > > > > > On Sat, Mar 4, 2023 at 10:51 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > > > > > > > Make __collapse_huge_page_copy return whether copying anonymous pages > > > > succeeded, and make collapse_huge_page handle the return status. > > > > > > > > Break existing PTE scan loop into two for-loops. The first loop copies > > > > source pages into target huge page, and can fail gracefully when running > > > > into memory errors in source pages. If copying all pages succeeds, the > > > > second loop releases and clears up these normal pages. Otherwise, the > > > > second loop rolls back the page table and page states by: > > > > - re-establishing the original PTEs-to-PMD connection. > > > > - releasing source pages back to their LRU list. > > > > > > > > Tested manually: > > > > 0. Enable khugepaged on system under test. > > > > 1. Start a two-thread application. Each thread allocates a chunk of > > > > non-huge anonymous memory buffer. > > > > 2. Pick 4 random buffer locations (2 in each thread) and inject > > > > uncorrectable memory errors at corresponding physical addresses. > > > > 3. Signal both threads to make their memory buffer collapsible, i.e. > > > > calling madvise(MADV_HUGEPAGE). > > > > 4. Wait and check kernel log: khugepaged is able to recover from poisoned > > > > pages and skips collapsing them. > > > > 5. Signal both threads to inspect their buffer contents and make sure no > > > > data corruption. > > > > > > > > Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> > > > > --- > > > > include/trace/events/huge_memory.h | 3 +- > > > > mm/khugepaged.c | 148 ++++++++++++++++++++++++----- > > > > 2 files changed, 128 insertions(+), 23 deletions(-) > > > > > > > > diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h > > > > index 3e6fb05852f9a..46cce509957ba 100644 > > > > --- a/include/trace/events/huge_memory.h > > > > +++ b/include/trace/events/huge_memory.h > > > > @@ -36,7 +36,8 @@ > > > > EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ > > > > EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ > > > > EM( SCAN_TRUNCATED, "truncated") \ > > > > - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > > > + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > > > + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ > > > > > > > > #undef EM > > > > #undef EMe > > > > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > > > > index 27956d4404134..c3c217f6ebc6e 100644 > > > > --- a/mm/khugepaged.c > > > > +++ b/mm/khugepaged.c > > > > @@ -19,6 +19,7 @@ > > > > #include <linux/page_table_check.h> > > > > #include <linux/swapops.h> > > > > #include <linux/shmem_fs.h> > > > > +#include <linux/kmsan.h> > > > > > > > > #include <asm/tlb.h> > > > > #include <asm/pgalloc.h> > > > > @@ -55,6 +56,7 @@ enum scan_result { > > > > SCAN_CGROUP_CHARGE_FAIL, > > > > SCAN_TRUNCATED, > > > > SCAN_PAGE_HAS_PRIVATE, > > > > + SCAN_COPY_MC, > > > > }; > > > > > > > > #define CREATE_TRACE_POINTS > > > > @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, > > > > return result; > > > > } > > > > > > > > -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > > > - struct vm_area_struct *vma, > > > > - unsigned long address, > > > > - spinlock_t *ptl, > > > > - struct list_head *compound_pagelist) > > > > +static void __collapse_huge_page_copy_succeeded(pte_t *pte, > > > > + pmd_t *pmd, > > > > + struct vm_area_struct *vma, > > > > + unsigned long address, > > > > + spinlock_t *pte_ptl, > > > > + struct list_head *compound_pagelist) > > > > { > > > > struct page *src_page, *tmp; > > > > pte_t *_pte; > > > > - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; > > > > - _pte++, page++, address += PAGE_SIZE) { > > > > - pte_t pteval = *_pte; > > > > + pte_t pteval; > > > > + unsigned long _address; > > > > > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > > + _pte++, _address += PAGE_SIZE) { > > > > + pteval = *_pte; > > > > if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > > > - clear_user_highpage(page, address); > > > > add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); > > > > if (is_zero_pfn(pte_pfn(pteval))) { > > > > /* > > > > - * ptl mostly unnecessary. > > > > + * pte_ptl mostly unnecessary. > > > > */ > > > > - spin_lock(ptl); > > > > - ptep_clear(vma->vm_mm, address, _pte); > > > > - spin_unlock(ptl); > > > > + spin_lock(pte_ptl); > > > > > > Why did you have to rename ptl to pte_ptl? It seems unnecessary. > > > > Thanks, I will use `ptl` in the next version. > > > > > > > > > + pte_clear(vma->vm_mm, _address, _pte); > > > > + spin_unlock(pte_ptl); > > > > } > > > > } else { > > > > src_page = pte_page(pteval); > > > > - copy_user_highpage(page, src_page, address, vma); > > > > if (!PageCompound(src_page)) > > > > release_pte_page(src_page); > > > > /* > > > > - * ptl mostly unnecessary, but preempt has to > > > > - * be disabled to update the per-cpu stats > > > > + * pte_ptl mostly unnecessary, but preempt has > > > > + * to be disabled to update the per-cpu stats > > > > * inside page_remove_rmap(). > > > > */ > > > > - spin_lock(ptl); > > > > - ptep_clear(vma->vm_mm, address, _pte); > > > > + spin_lock(pte_ptl); > > > > + ptep_clear(vma->vm_mm, _address, _pte); > > > > page_remove_rmap(src_page, vma, false); > > > > - spin_unlock(ptl); > > > > + spin_unlock(pte_ptl); > > > > free_page_and_swap_cache(src_page); > > > > } > > > > } > > > > - > > > > list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > > > list_del(&src_page->lru); > > > > mod_node_page_state(page_pgdat(src_page), > > > > @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > > > } > > > > } > > > > > > > > +static void __collapse_huge_page_copy_failed(pte_t *pte, > > > > + pmd_t *pmd, > > > > + pmd_t orig_pmd, > > > > + struct vm_area_struct *vma, > > > > + unsigned long address, > > > > + struct list_head *compound_pagelist) > > > > +{ > > > > + struct page *src_page, *tmp; > > > > + pte_t *_pte; > > > > + pte_t pteval; > > > > + unsigned long _address; > > > > + spinlock_t *pmd_ptl; > > > > + > > > > + /* > > > > + * Re-establish the PMD to point to the original page table > > > > + * entry. Restoring PMD needs to be done prior to releasing > > > > + * pages. Since pages are still isolated and locked here, > > > > + * acquiring anon_vma_lock_write is unnecessary. > > > > + */ > > > > + pmd_ptl = pmd_lock(vma->vm_mm, pmd); > > > > + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); > > > > + spin_unlock(pmd_ptl); > > > > + /* > > > > + * Release both raw and compound pages isolated > > > > + * in __collapse_huge_page_isolate. > > > > + */ > > > > > > It looks like the below code could be replaced by release_pte_pages() > > > with advancing _pte to (pte + HPAGE_PMD_NR - 1). > > > > > > > Yeah, but I think _pte should be (pte + HPAGE_PMR_NR) because _pte is > > decremented before comparison in release_pte_pages(pte, _pte, > > compound_pagelist): > > > > while (--_pte >= pte) {...} > > > > Advancing _pte to (pte + HPAGE_PMD_NR - 1) may leave the last page not released. > > Yeah, good catch. I think it is because the only user of > release_pte_pages() is __collapse_huge_page_isolate(). Once the loop > in it is done _pte is pte + HPAGE_PMD_NR. > Thanks for confirming ;) Since you have reviewed other 2 commits, I will soon send out V11 for your comments on this commit. > > > > > > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > > + _pte++, _address += PAGE_SIZE) { > > > > + pteval = *_pte; > > > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) > > > > + continue; > > > > + src_page = pte_page(pteval); > > > > + if (!PageCompound(src_page)) > > > > + release_pte_page(src_page); > > > > + } > > > > + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > > > + list_del(&src_page->lru); > > > > + release_pte_page(src_page); > > > > + } > > > > +} > > > > + > > > > +/* > > > > + * __collapse_huge_page_copy - attempts to copy memory contents from raw > > > > + * pages to a hugepage. Cleans up the raw pages if copying succeeds; > > > > + * otherwise restores the original page table and releases isolated raw pages. > > > > + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. > > > > + * > > > > + * @pte: starting of the PTEs to copy from > > > > + * @page: the new hugepage to copy contents to > > > > + * @pmd: pointer to the new hugepage's PMD > > > > + * @orig_pmd: the original raw pages' PMD > > > > + * @vma: the original raw pages' virtual memory area > > > > + * @address: starting address to copy > > > > + * @pte_ptl: lock on raw pages' PTEs > > > > + * @compound_pagelist: list that stores compound pages > > > > + */ > > > > +static int __collapse_huge_page_copy(pte_t *pte, > > > > + struct page *page, > > > > + pmd_t *pmd, > > > > + pmd_t orig_pmd, > > > > + struct vm_area_struct *vma, > > > > + unsigned long address, > > > > + spinlock_t *pte_ptl, > > > > + struct list_head *compound_pagelist) > > > > +{ > > > > + struct page *src_page; > > > > + pte_t *_pte; > > > > + pte_t pteval; > > > > + unsigned long _address; > > > > + int result = SCAN_SUCCEED; > > > > + > > > > + /* > > > > + * Copying pages' contents is subject to memory poison at any iteration. > > > > + */ > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > > + _pte++, page++, _address += PAGE_SIZE) { > > > > + pteval = *_pte; > > > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > > > + clear_user_highpage(page, _address); > > > > + continue; > > > > + } > > > > + src_page = pte_page(pteval); > > > > + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { > > > > + result = SCAN_COPY_MC; > > > > + break; > > > > + } > > > > + } > > > > + > > > > + if (likely(result == SCAN_SUCCEED)) > > > > + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, > > > > + pte_ptl, compound_pagelist); > > > > + else > > > > + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, > > > > + address, compound_pagelist); > > > > + > > > > + return result; > > > > +} > > > > + > > > > static void khugepaged_alloc_sleep(void) > > > > { > > > > DEFINE_WAIT(wait); > > > > @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, > > > > */ > > > > anon_vma_unlock_write(vma->anon_vma); > > > > > > > > - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, > > > > - &compound_pagelist); > > > > + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, > > > > + vma, address, pte_ptl, > > > > + &compound_pagelist); > > > > pte_unmap(pte); > > > > + if (unlikely(result != SCAN_SUCCEED)) > > > > + goto out_up_write; > > > > + > > > > /* > > > > * spin_lock() below is not the equivalent of smp_wmb(), but > > > > * the smp_wmb() inside __SetPageUptodate() can be reused to > > > > -- > > > > 2.40.0.rc0.216.gc4246ad0f0-goog > > > >
On Fri, Mar 24, 2023 at 3:31 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > On Fri, Mar 24, 2023 at 1:11 PM Yang Shi <shy828301@gmail.com> wrote: > > > > On Fri, Mar 24, 2023 at 8:34 AM Jiaqi Yan <jiaqiyan@google.com> wrote: > > > > > > On Thu, Mar 23, 2023 at 2:38 PM Yang Shi <shy828301@gmail.com> wrote: > > > > > > > > On Sat, Mar 4, 2023 at 10:51 PM Jiaqi Yan <jiaqiyan@google.com> wrote: > > > > > > > > > > Make __collapse_huge_page_copy return whether copying anonymous pages > > > > > succeeded, and make collapse_huge_page handle the return status. > > > > > > > > > > Break existing PTE scan loop into two for-loops. The first loop copies > > > > > source pages into target huge page, and can fail gracefully when running > > > > > into memory errors in source pages. If copying all pages succeeds, the > > > > > second loop releases and clears up these normal pages. Otherwise, the > > > > > second loop rolls back the page table and page states by: > > > > > - re-establishing the original PTEs-to-PMD connection. > > > > > - releasing source pages back to their LRU list. > > > > > > > > > > Tested manually: > > > > > 0. Enable khugepaged on system under test. > > > > > 1. Start a two-thread application. Each thread allocates a chunk of > > > > > non-huge anonymous memory buffer. > > > > > 2. Pick 4 random buffer locations (2 in each thread) and inject > > > > > uncorrectable memory errors at corresponding physical addresses. > > > > > 3. Signal both threads to make their memory buffer collapsible, i.e. > > > > > calling madvise(MADV_HUGEPAGE). > > > > > 4. Wait and check kernel log: khugepaged is able to recover from poisoned > > > > > pages and skips collapsing them. > > > > > 5. Signal both threads to inspect their buffer contents and make sure no > > > > > data corruption. > > > > > > > > > > Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> > > > > > --- > > > > > include/trace/events/huge_memory.h | 3 +- > > > > > mm/khugepaged.c | 148 ++++++++++++++++++++++++----- > > > > > 2 files changed, 128 insertions(+), 23 deletions(-) > > > > > > > > > > diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h > > > > > index 3e6fb05852f9a..46cce509957ba 100644 > > > > > --- a/include/trace/events/huge_memory.h > > > > > +++ b/include/trace/events/huge_memory.h > > > > > @@ -36,7 +36,8 @@ > > > > > EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ > > > > > EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ > > > > > EM( SCAN_TRUNCATED, "truncated") \ > > > > > - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > > > > + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ > > > > > + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ > > > > > > > > > > #undef EM > > > > > #undef EMe > > > > > diff --git a/mm/khugepaged.c b/mm/khugepaged.c > > > > > index 27956d4404134..c3c217f6ebc6e 100644 > > > > > --- a/mm/khugepaged.c > > > > > +++ b/mm/khugepaged.c > > > > > @@ -19,6 +19,7 @@ > > > > > #include <linux/page_table_check.h> > > > > > #include <linux/swapops.h> > > > > > #include <linux/shmem_fs.h> > > > > > +#include <linux/kmsan.h> > > > > > > > > > > #include <asm/tlb.h> > > > > > #include <asm/pgalloc.h> > > > > > @@ -55,6 +56,7 @@ enum scan_result { > > > > > SCAN_CGROUP_CHARGE_FAIL, > > > > > SCAN_TRUNCATED, > > > > > SCAN_PAGE_HAS_PRIVATE, > > > > > + SCAN_COPY_MC, > > > > > }; > > > > > > > > > > #define CREATE_TRACE_POINTS > > > > > @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, > > > > > return result; > > > > > } > > > > > > > > > > -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > > > > - struct vm_area_struct *vma, > > > > > - unsigned long address, > > > > > - spinlock_t *ptl, > > > > > - struct list_head *compound_pagelist) > > > > > +static void __collapse_huge_page_copy_succeeded(pte_t *pte, > > > > > + pmd_t *pmd, > > > > > + struct vm_area_struct *vma, > > > > > + unsigned long address, > > > > > + spinlock_t *pte_ptl, > > > > > + struct list_head *compound_pagelist) > > > > > { > > > > > struct page *src_page, *tmp; > > > > > pte_t *_pte; > > > > > - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; > > > > > - _pte++, page++, address += PAGE_SIZE) { > > > > > - pte_t pteval = *_pte; > > > > > + pte_t pteval; > > > > > + unsigned long _address; > > > > > > > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > > > + _pte++, _address += PAGE_SIZE) { > > > > > + pteval = *_pte; > > > > > if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > > > > - clear_user_highpage(page, address); > > > > > add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); > > > > > if (is_zero_pfn(pte_pfn(pteval))) { > > > > > /* > > > > > - * ptl mostly unnecessary. > > > > > + * pte_ptl mostly unnecessary. > > > > > */ > > > > > - spin_lock(ptl); > > > > > - ptep_clear(vma->vm_mm, address, _pte); > > > > > - spin_unlock(ptl); > > > > > + spin_lock(pte_ptl); > > > > > > > > Why did you have to rename ptl to pte_ptl? It seems unnecessary. > > > > > > Thanks, I will use `ptl` in the next version. > > > > > > > > > > > > + pte_clear(vma->vm_mm, _address, _pte); A bug here: calling pte_clear is wrong; we should use ptep_clear. V11 will make sure it is fixed. BTW, __collapse_huge_page_copy_succeeded can drop _address and use address directly. > > > > > + spin_unlock(pte_ptl); > > > > > } > > > > > } else { > > > > > src_page = pte_page(pteval); > > > > > - copy_user_highpage(page, src_page, address, vma); > > > > > if (!PageCompound(src_page)) > > > > > release_pte_page(src_page); > > > > > /* > > > > > - * ptl mostly unnecessary, but preempt has to > > > > > - * be disabled to update the per-cpu stats > > > > > + * pte_ptl mostly unnecessary, but preempt has > > > > > + * to be disabled to update the per-cpu stats > > > > > * inside page_remove_rmap(). > > > > > */ > > > > > - spin_lock(ptl); > > > > > - ptep_clear(vma->vm_mm, address, _pte); > > > > > + spin_lock(pte_ptl); > > > > > + ptep_clear(vma->vm_mm, _address, _pte); > > > > > page_remove_rmap(src_page, vma, false); > > > > > - spin_unlock(ptl); > > > > > + spin_unlock(pte_ptl); > > > > > free_page_and_swap_cache(src_page); > > > > > } > > > > > } > > > > > - > > > > > list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > > > > list_del(&src_page->lru); > > > > > mod_node_page_state(page_pgdat(src_page), > > > > > @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, > > > > > } > > > > > } > > > > > > > > > > +static void __collapse_huge_page_copy_failed(pte_t *pte, > > > > > + pmd_t *pmd, > > > > > + pmd_t orig_pmd, > > > > > + struct vm_area_struct *vma, > > > > > + unsigned long address, > > > > > + struct list_head *compound_pagelist) > > > > > +{ > > > > > + struct page *src_page, *tmp; > > > > > + pte_t *_pte; > > > > > + pte_t pteval; > > > > > + unsigned long _address; > > > > > + spinlock_t *pmd_ptl; > > > > > + > > > > > + /* > > > > > + * Re-establish the PMD to point to the original page table > > > > > + * entry. Restoring PMD needs to be done prior to releasing > > > > > + * pages. Since pages are still isolated and locked here, > > > > > + * acquiring anon_vma_lock_write is unnecessary. > > > > > + */ > > > > > + pmd_ptl = pmd_lock(vma->vm_mm, pmd); > > > > > + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); > > > > > + spin_unlock(pmd_ptl); > > > > > + /* > > > > > + * Release both raw and compound pages isolated > > > > > + * in __collapse_huge_page_isolate. > > > > > + */ > > > > > > > > It looks like the below code could be replaced by release_pte_pages() > > > > with advancing _pte to (pte + HPAGE_PMD_NR - 1). > > > > > > > > > > Yeah, but I think _pte should be (pte + HPAGE_PMR_NR) because _pte is > > > decremented before comparison in release_pte_pages(pte, _pte, > > > compound_pagelist): > > > > > > while (--_pte >= pte) {...} > > > > > > Advancing _pte to (pte + HPAGE_PMD_NR - 1) may leave the last page not released. > > > > Yeah, good catch. I think it is because the only user of > > release_pte_pages() is __collapse_huge_page_isolate(). Once the loop > > in it is done _pte is pte + HPAGE_PMD_NR. > > > > Thanks for confirming ;) > Since you have reviewed other 2 commits, I will soon send out V11 for > your comments on this commit. > > > > > > > > > > > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > > > + _pte++, _address += PAGE_SIZE) { > > > > > + pteval = *_pte; > > > > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) > > > > > + continue; > > > > > + src_page = pte_page(pteval); > > > > > + if (!PageCompound(src_page)) > > > > > + release_pte_page(src_page); > > > > > + } > > > > > + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { > > > > > + list_del(&src_page->lru); > > > > > + release_pte_page(src_page); > > > > > + } > > > > > +} > > > > > + > > > > > +/* > > > > > + * __collapse_huge_page_copy - attempts to copy memory contents from raw > > > > > + * pages to a hugepage. Cleans up the raw pages if copying succeeds; > > > > > + * otherwise restores the original page table and releases isolated raw pages. > > > > > + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. > > > > > + * > > > > > + * @pte: starting of the PTEs to copy from > > > > > + * @page: the new hugepage to copy contents to > > > > > + * @pmd: pointer to the new hugepage's PMD > > > > > + * @orig_pmd: the original raw pages' PMD > > > > > + * @vma: the original raw pages' virtual memory area > > > > > + * @address: starting address to copy > > > > > + * @pte_ptl: lock on raw pages' PTEs > > > > > + * @compound_pagelist: list that stores compound pages > > > > > + */ > > > > > +static int __collapse_huge_page_copy(pte_t *pte, > > > > > + struct page *page, > > > > > + pmd_t *pmd, > > > > > + pmd_t orig_pmd, > > > > > + struct vm_area_struct *vma, > > > > > + unsigned long address, > > > > > + spinlock_t *pte_ptl, > > > > > + struct list_head *compound_pagelist) > > > > > +{ > > > > > + struct page *src_page; > > > > > + pte_t *_pte; > > > > > + pte_t pteval; > > > > > + unsigned long _address; > > > > > + int result = SCAN_SUCCEED; > > > > > + > > > > > + /* > > > > > + * Copying pages' contents is subject to memory poison at any iteration. > > > > > + */ > > > > > + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; > > > > > + _pte++, page++, _address += PAGE_SIZE) { > > > > > + pteval = *_pte; > > > > > + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { > > > > > + clear_user_highpage(page, _address); > > > > > + continue; > > > > > + } > > > > > + src_page = pte_page(pteval); > > > > > + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { > > > > > + result = SCAN_COPY_MC; > > > > > + break; > > > > > + } > > > > > + } > > > > > + > > > > > + if (likely(result == SCAN_SUCCEED)) > > > > > + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, > > > > > + pte_ptl, compound_pagelist); > > > > > + else > > > > > + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, > > > > > + address, compound_pagelist); > > > > > + > > > > > + return result; > > > > > +} > > > > > + > > > > > static void khugepaged_alloc_sleep(void) > > > > > { > > > > > DEFINE_WAIT(wait); > > > > > @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, > > > > > */ > > > > > anon_vma_unlock_write(vma->anon_vma); > > > > > > > > > > - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, > > > > > - &compound_pagelist); > > > > > + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, > > > > > + vma, address, pte_ptl, > > > > > + &compound_pagelist); > > > > > pte_unmap(pte); > > > > > + if (unlikely(result != SCAN_SUCCEED)) > > > > > + goto out_up_write; > > > > > + > > > > > /* > > > > > * spin_lock() below is not the equivalent of smp_wmb(), but > > > > > * the smp_wmb() inside __SetPageUptodate() can be reused to > > > > > -- > > > > > 2.40.0.rc0.216.gc4246ad0f0-goog > > > > >
diff --git a/include/trace/events/huge_memory.h b/include/trace/events/huge_memory.h index 3e6fb05852f9a..46cce509957ba 100644 --- a/include/trace/events/huge_memory.h +++ b/include/trace/events/huge_memory.h @@ -36,7 +36,8 @@ EM( SCAN_ALLOC_HUGE_PAGE_FAIL, "alloc_huge_page_failed") \ EM( SCAN_CGROUP_CHARGE_FAIL, "ccgroup_charge_failed") \ EM( SCAN_TRUNCATED, "truncated") \ - EMe(SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ + EM( SCAN_PAGE_HAS_PRIVATE, "page_has_private") \ + EMe(SCAN_COPY_MC, "copy_poisoned_page") \ #undef EM #undef EMe diff --git a/mm/khugepaged.c b/mm/khugepaged.c index 27956d4404134..c3c217f6ebc6e 100644 --- a/mm/khugepaged.c +++ b/mm/khugepaged.c @@ -19,6 +19,7 @@ #include <linux/page_table_check.h> #include <linux/swapops.h> #include <linux/shmem_fs.h> +#include <linux/kmsan.h> #include <asm/tlb.h> #include <asm/pgalloc.h> @@ -55,6 +56,7 @@ enum scan_result { SCAN_CGROUP_CHARGE_FAIL, SCAN_TRUNCATED, SCAN_PAGE_HAS_PRIVATE, + SCAN_COPY_MC, }; #define CREATE_TRACE_POINTS @@ -681,47 +683,47 @@ static int __collapse_huge_page_isolate(struct vm_area_struct *vma, return result; } -static void __collapse_huge_page_copy(pte_t *pte, struct page *page, - struct vm_area_struct *vma, - unsigned long address, - spinlock_t *ptl, - struct list_head *compound_pagelist) +static void __collapse_huge_page_copy_succeeded(pte_t *pte, + pmd_t *pmd, + struct vm_area_struct *vma, + unsigned long address, + spinlock_t *pte_ptl, + struct list_head *compound_pagelist) { struct page *src_page, *tmp; pte_t *_pte; - for (_pte = pte; _pte < pte + HPAGE_PMD_NR; - _pte++, page++, address += PAGE_SIZE) { - pte_t pteval = *_pte; + pte_t pteval; + unsigned long _address; + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; + _pte++, _address += PAGE_SIZE) { + pteval = *_pte; if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { - clear_user_highpage(page, address); add_mm_counter(vma->vm_mm, MM_ANONPAGES, 1); if (is_zero_pfn(pte_pfn(pteval))) { /* - * ptl mostly unnecessary. + * pte_ptl mostly unnecessary. */ - spin_lock(ptl); - ptep_clear(vma->vm_mm, address, _pte); - spin_unlock(ptl); + spin_lock(pte_ptl); + pte_clear(vma->vm_mm, _address, _pte); + spin_unlock(pte_ptl); } } else { src_page = pte_page(pteval); - copy_user_highpage(page, src_page, address, vma); if (!PageCompound(src_page)) release_pte_page(src_page); /* - * ptl mostly unnecessary, but preempt has to - * be disabled to update the per-cpu stats + * pte_ptl mostly unnecessary, but preempt has + * to be disabled to update the per-cpu stats * inside page_remove_rmap(). */ - spin_lock(ptl); - ptep_clear(vma->vm_mm, address, _pte); + spin_lock(pte_ptl); + ptep_clear(vma->vm_mm, _address, _pte); page_remove_rmap(src_page, vma, false); - spin_unlock(ptl); + spin_unlock(pte_ptl); free_page_and_swap_cache(src_page); } } - list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { list_del(&src_page->lru); mod_node_page_state(page_pgdat(src_page), @@ -733,6 +735,104 @@ static void __collapse_huge_page_copy(pte_t *pte, struct page *page, } } +static void __collapse_huge_page_copy_failed(pte_t *pte, + pmd_t *pmd, + pmd_t orig_pmd, + struct vm_area_struct *vma, + unsigned long address, + struct list_head *compound_pagelist) +{ + struct page *src_page, *tmp; + pte_t *_pte; + pte_t pteval; + unsigned long _address; + spinlock_t *pmd_ptl; + + /* + * Re-establish the PMD to point to the original page table + * entry. Restoring PMD needs to be done prior to releasing + * pages. Since pages are still isolated and locked here, + * acquiring anon_vma_lock_write is unnecessary. + */ + pmd_ptl = pmd_lock(vma->vm_mm, pmd); + pmd_populate(vma->vm_mm, pmd, pmd_pgtable(orig_pmd)); + spin_unlock(pmd_ptl); + /* + * Release both raw and compound pages isolated + * in __collapse_huge_page_isolate. + */ + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; + _pte++, _address += PAGE_SIZE) { + pteval = *_pte; + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) + continue; + src_page = pte_page(pteval); + if (!PageCompound(src_page)) + release_pte_page(src_page); + } + list_for_each_entry_safe(src_page, tmp, compound_pagelist, lru) { + list_del(&src_page->lru); + release_pte_page(src_page); + } +} + +/* + * __collapse_huge_page_copy - attempts to copy memory contents from raw + * pages to a hugepage. Cleans up the raw pages if copying succeeds; + * otherwise restores the original page table and releases isolated raw pages. + * Returns SCAN_SUCCEED if copying succeeds, otherwise returns SCAN_COPY_MC. + * + * @pte: starting of the PTEs to copy from + * @page: the new hugepage to copy contents to + * @pmd: pointer to the new hugepage's PMD + * @orig_pmd: the original raw pages' PMD + * @vma: the original raw pages' virtual memory area + * @address: starting address to copy + * @pte_ptl: lock on raw pages' PTEs + * @compound_pagelist: list that stores compound pages + */ +static int __collapse_huge_page_copy(pte_t *pte, + struct page *page, + pmd_t *pmd, + pmd_t orig_pmd, + struct vm_area_struct *vma, + unsigned long address, + spinlock_t *pte_ptl, + struct list_head *compound_pagelist) +{ + struct page *src_page; + pte_t *_pte; + pte_t pteval; + unsigned long _address; + int result = SCAN_SUCCEED; + + /* + * Copying pages' contents is subject to memory poison at any iteration. + */ + for (_pte = pte, _address = address; _pte < pte + HPAGE_PMD_NR; + _pte++, page++, _address += PAGE_SIZE) { + pteval = *_pte; + if (pte_none(pteval) || is_zero_pfn(pte_pfn(pteval))) { + clear_user_highpage(page, _address); + continue; + } + src_page = pte_page(pteval); + if (copy_mc_user_highpage(page, src_page, _address, vma) > 0) { + result = SCAN_COPY_MC; + break; + } + } + + if (likely(result == SCAN_SUCCEED)) + __collapse_huge_page_copy_succeeded(pte, pmd, vma, address, + pte_ptl, compound_pagelist); + else + __collapse_huge_page_copy_failed(pte, pmd, orig_pmd, vma, + address, compound_pagelist); + + return result; +} + static void khugepaged_alloc_sleep(void) { DEFINE_WAIT(wait); @@ -1106,9 +1206,13 @@ static int collapse_huge_page(struct mm_struct *mm, unsigned long address, */ anon_vma_unlock_write(vma->anon_vma); - __collapse_huge_page_copy(pte, hpage, vma, address, pte_ptl, - &compound_pagelist); + result = __collapse_huge_page_copy(pte, hpage, pmd, _pmd, + vma, address, pte_ptl, + &compound_pagelist); pte_unmap(pte); + if (unlikely(result != SCAN_SUCCEED)) + goto out_up_write; + /* * spin_lock() below is not the equivalent of smp_wmb(), but * the smp_wmb() inside __SetPageUptodate() can be reused to
Make __collapse_huge_page_copy return whether copying anonymous pages succeeded, and make collapse_huge_page handle the return status. Break existing PTE scan loop into two for-loops. The first loop copies source pages into target huge page, and can fail gracefully when running into memory errors in source pages. If copying all pages succeeds, the second loop releases and clears up these normal pages. Otherwise, the second loop rolls back the page table and page states by: - re-establishing the original PTEs-to-PMD connection. - releasing source pages back to their LRU list. Tested manually: 0. Enable khugepaged on system under test. 1. Start a two-thread application. Each thread allocates a chunk of non-huge anonymous memory buffer. 2. Pick 4 random buffer locations (2 in each thread) and inject uncorrectable memory errors at corresponding physical addresses. 3. Signal both threads to make their memory buffer collapsible, i.e. calling madvise(MADV_HUGEPAGE). 4. Wait and check kernel log: khugepaged is able to recover from poisoned pages and skips collapsing them. 5. Signal both threads to inspect their buffer contents and make sure no data corruption. Signed-off-by: Jiaqi Yan <jiaqiyan@google.com> --- include/trace/events/huge_memory.h | 3 +- mm/khugepaged.c | 148 ++++++++++++++++++++++++----- 2 files changed, 128 insertions(+), 23 deletions(-)