| Message ID | 20190802160614.8089-1-vbabka@suse.cz (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | [STABLE,4.9] x86, mm, gup: prevent get_page() race with munmap in paravirt guest | expand |
On Fri, Aug 02, 2019 at 06:06:14PM +0200, Vlastimil Babka wrote: > The x86 version of get_user_pages_fast() relies on disabled interrupts to > synchronize gup_pte_range() between gup_get_pte(ptep); and get_page() against > a parallel munmap. The munmap side nulls the pte, then flushes TLBs, then > releases the page. As TLB flush is done synchronously via IPI disabling > interrupts blocks the page release, and get_page(), which assumes existing > reference on page, is thus safe. > However when TLB flush is done by a hypercall, e.g. in a Xen PV guest, there is > no blocking thanks to disabled interrupts, and get_page() can succeed on a page > that was already freed or even reused. > > We have recently seen this happen with our 4.4 and 4.12 based kernels, with > userspace (java) that exits a thread, where mm_release() performs a futex_wake() > on tsk->clear_child_tid, and another thread in parallel unmaps the page where > tsk->clear_child_tid points to. The spurious get_page() succeeds, but futex code > immediately releases the page again, while it's already on a freelist. Symptoms > include a bad page state warning, general protection faults acessing a poisoned > list prev/next pointer in the freelist, or free page pcplists of two cpus joined > together in a single list. Oscar has also reproduced this scenario, with a > patch inserting delays before the get_page() to make the race window larger. > > Fix this by removing the dependency on TLB flush interrupts the same way as the > generic get_user_pages_fast() code by using page_cache_add_speculative() and > revalidating the PTE contents after pinning the page. Mainline is safe since > 4.13 where the x86 gup code was removed in favor of the common code. Accessing > the page table itself safely also relies on disabled interrupts and TLB flush > IPIs that don't happen with hypercalls, which was acknowledged in commit > 9e52fc2b50de ("x86/mm: Enable RCU based page table freeing > (CONFIG_HAVE_RCU_TABLE_FREE=y)"). That commit with follups should also be > backported for full safety, although our reproducer didn't hit a problem > without that backport. > > Reproduced-by: Oscar Salvador <osalvador@suse.de> > Signed-off-by: Vlastimil Babka <vbabka@suse.cz> > Cc: Thomas Gleixner <tglx@linutronix.de> > Cc: Ingo Molnar <mingo@redhat.com> > Cc: Peter Zijlstra <peterz@infradead.org> > Cc: Juergen Gross <jgross@suse.com> > Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> > Cc: Vitaly Kuznetsov <vkuznets@redhat.com> > Cc: Linus Torvalds <torvalds@linux-foundation.org> > Cc: Borislav Petkov <bp@alien8.de> > Cc: Dave Hansen <dave.hansen@linux.intel.com> > Cc: Andy Lutomirski <luto@kernel.org> > --- > > Hi, I'm sending this stable-only patch for consideration because it's probably > unrealistic to backport the 4.13 switch to generic GUP. I can look at 4.4 and > 3.16 if accepted. The RCU page table freeing could be also considered. > Note the patch also includes page refcount protection. I found out that > 8fde12ca79af ("mm: prevent get_user_pages() from overflowing page refcount") > backport to 4.9 missed the arch-specific gup implementations: > https://lore.kernel.org/lkml/6650323f-dbc9-f069-000b-f6b0f941a065@suse.cz/ This looks sane to me, thank you for the backport. I've queued it up now, and if anyone has any objections, please let me know. thanks, greg k-h
On Mon, 2019-08-19 at 18:58 +0100, Vlastimil Babka wrote: [...] > Hi, I'm sending this stable-only patch for consideration because it's probably > unrealistic to backport the 4.13 switch to generic GUP. I can look at 4.4 and > 3.16 if accepted. The RCU page table freeing could be also considered. I would be interested in backports for 3.16 and 4.4. > Note the patch also includes page refcount protection. I found out that > 8fde12ca79af ("mm: prevent get_user_pages() from overflowing page refcount") > backport to 4.9 missed the arch-specific gup implementations: > https://lore.kernel.org/lkml/6650323f-dbc9-f069-000b-f6b0f941a065@suse.cz/ [...] I suppose that still needs to be addressed for 4.9, right? Ben.
On 9/19/19 8:26 PM, Ben Hutchings wrote: > On Mon, 2019-08-19 at 18:58 +0100, Vlastimil Babka wrote: > [...] >> Hi, I'm sending this stable-only patch for consideration because it's probably >> unrealistic to backport the 4.13 switch to generic GUP. I can look at 4.4 and >> 3.16 if accepted. The RCU page table freeing could be also considered. > > I would be interested in backports for 3.16 and 4.4. > >> Note the patch also includes page refcount protection. I found out that >> 8fde12ca79af ("mm: prevent get_user_pages() from overflowing page refcount") >> backport to 4.9 missed the arch-specific gup implementations: >> https://lore.kernel.org/lkml/6650323f-dbc9-f069-000b-f6b0f941a065@suse.cz/ > [...] > > I suppose that still needs to be addressed for 4.9, right? Yeah, I'll take a look, thanks for reminding. > Ben. >
On 9/19/19 8:26 PM, Ben Hutchings wrote: > On Mon, 2019-08-19 at 18:58 +0100, Vlastimil Babka wrote: > [...] >> Hi, I'm sending this stable-only patch for consideration because it's probably >> unrealistic to backport the 4.13 switch to generic GUP. I can look at 4.4 and >> 3.16 if accepted. The RCU page table freeing could be also considered. > > I would be interested in backports for 3.16 and 4.4. > >> Note the patch also includes page refcount protection. I found out that >> 8fde12ca79af ("mm: prevent get_user_pages() from overflowing page refcount") >> backport to 4.9 missed the arch-specific gup implementations: >> https://lore.kernel.org/lkml/6650323f-dbc9-f069-000b-f6b0f941a065@suse.cz/ > [...] > > I suppose that still needs to be addressed for 4.9, right? Here's what is AFAIK missing for 4.9 for x86 and s390. ----8<---- From d981bbf770ca41e999115cf3b0f27dde57479df0 Mon Sep 17 00:00:00 2001 From: Vlastimil Babka <vbabka@suse.cz> Date: Wed, 6 Nov 2019 16:32:57 +0100 Subject: [PATCH STABLE 4.9] mm, gup: add missing refcount overflow checks on x86 and s390 The mainline commit 8fde12ca79af ("mm: prevent get_user_pages() from overflowing page refcount") was backported to 4.9.y stable as commit 2ed768cfd895. The backport however missed that in 4.9, there are several arch-specific gup.c versions with fast gup implementations, so these do not prevent refcount overflow. This is partially fixed for x86 in stable-only commit d73af79742e7 ("x86, mm, gup: prevent get_page() race with munmap in paravirt guest"). This stable-only commit adds missing parts to x86 version, as well as s390 version, both taken from the SUSE SLES/openSUSE 4.12-based kernels. The remaining architectures with own gup.c are sparc, mips, sh. It's unlikely the known overflow scenario based on FUSE, which needs 140GB of RAM, is a problem for those architectures, and I don't feel confident enough to patch them. Signed-off-by: Vlastimil Babka <vbabka@suse.cz> --- arch/s390/mm/gup.c | 9 ++++++--- arch/x86/mm/gup.c | 10 ++++++++-- 2 files changed, 14 insertions(+), 5 deletions(-) diff --git a/arch/s390/mm/gup.c b/arch/s390/mm/gup.c index 97fc449a7470..33a940389a6d 100644 --- a/arch/s390/mm/gup.c +++ b/arch/s390/mm/gup.c @@ -38,7 +38,8 @@ static inline int gup_pte_range(pmd_t *pmdp, pmd_t pmd, unsigned long addr, VM_BUG_ON(!pfn_valid(pte_pfn(pte))); page = pte_page(pte); head = compound_head(page); - if (!page_cache_get_speculative(head)) + if (unlikely(WARN_ON_ONCE(page_ref_count(head) < 0) + || !page_cache_get_speculative(head))) return 0; if (unlikely(pte_val(pte) != pte_val(*ptep))) { put_page(head); @@ -76,7 +77,8 @@ static inline int gup_huge_pmd(pmd_t *pmdp, pmd_t pmd, unsigned long addr, refs++; } while (addr += PAGE_SIZE, addr != end); - if (!page_cache_add_speculative(head, refs)) { + if (unlikely(WARN_ON_ONCE(page_ref_count(head) < 0) + || !page_cache_add_speculative(head, refs))) { *nr -= refs; return 0; } @@ -150,7 +152,8 @@ static int gup_huge_pud(pud_t *pudp, pud_t pud, unsigned long addr, refs++; } while (addr += PAGE_SIZE, addr != end); - if (!page_cache_add_speculative(head, refs)) { + if (unlikely(WARN_ON_ONCE(page_ref_count(head) < 0) + || !page_cache_add_speculative(head, refs))) { *nr -= refs; return 0; } diff --git a/arch/x86/mm/gup.c b/arch/x86/mm/gup.c index d7db45bdfb3b..551fc7fea046 100644 --- a/arch/x86/mm/gup.c +++ b/arch/x86/mm/gup.c @@ -202,10 +202,12 @@ static int __gup_device_huge_pmd(pmd_t pmd, unsigned long addr, undo_dev_pagemap(nr, nr_start, pages); return 0; } + if (unlikely(!try_get_page(page))) { + put_dev_pagemap(pgmap); + return 0; + } SetPageReferenced(page); pages[*nr] = page; - get_page(page); - put_dev_pagemap(pgmap); (*nr)++; pfn++; } while (addr += PAGE_SIZE, addr != end); @@ -230,6 +232,8 @@ static noinline int gup_huge_pmd(pmd_t pmd, unsigned long addr, refs = 0; head = pmd_page(pmd); + if (WARN_ON_ONCE(page_ref_count(head) <= 0)) + return 0; page = head + ((addr & ~PMD_MASK) >> PAGE_SHIFT); do { VM_BUG_ON_PAGE(compound_head(page) != head, page); @@ -289,6 +293,8 @@ static noinline int gup_huge_pud(pud_t pud, unsigned long addr, refs = 0; head = pud_page(pud); + if (WARN_ON_ONCE(page_ref_count(head) <= 0)) + return 0; page = head + ((addr & ~PUD_MASK) >> PAGE_SHIFT); do { VM_BUG_ON_PAGE(compound_head(page) != head, page);
diff --git a/arch/x86/mm/gup.c b/arch/x86/mm/gup.c index 1680768d392c..d7db45bdfb3b 100644 --- a/arch/x86/mm/gup.c +++ b/arch/x86/mm/gup.c @@ -97,6 +97,20 @@ static inline int pte_allows_gup(unsigned long pteval, int write) return 1; } +/* + * Return the compund head page with ref appropriately incremented, + * or NULL if that failed. + */ +static inline struct page *try_get_compound_head(struct page *page, int refs) +{ + struct page *head = compound_head(page); + if (WARN_ON_ONCE(page_ref_count(head) < 0)) + return NULL; + if (unlikely(!page_cache_add_speculative(head, refs))) + return NULL; + return head; +} + /* * The performance critical leaf functions are made noinline otherwise gcc * inlines everything into a single function which results in too much @@ -112,7 +126,7 @@ static noinline int gup_pte_range(pmd_t pmd, unsigned long addr, ptep = pte_offset_map(&pmd, addr); do { pte_t pte = gup_get_pte(ptep); - struct page *page; + struct page *head, *page; /* Similar to the PMD case, NUMA hinting must take slow path */ if (pte_protnone(pte)) { @@ -138,7 +152,21 @@ static noinline int gup_pte_range(pmd_t pmd, unsigned long addr, } VM_BUG_ON(!pfn_valid(pte_pfn(pte))); page = pte_page(pte); - get_page(page); + + head = try_get_compound_head(page, 1); + if (!head) { + put_dev_pagemap(pgmap); + pte_unmap(ptep); + return 0; + } + + if (unlikely(pte_val(pte) != pte_val(*ptep))) { + put_page(head); + put_dev_pagemap(pgmap); + pte_unmap(ptep); + return 0; + } + put_dev_pagemap(pgmap); SetPageReferenced(page); pages[*nr] = page;
The x86 version of get_user_pages_fast() relies on disabled interrupts to synchronize gup_pte_range() between gup_get_pte(ptep); and get_page() against a parallel munmap. The munmap side nulls the pte, then flushes TLBs, then releases the page. As TLB flush is done synchronously via IPI disabling interrupts blocks the page release, and get_page(), which assumes existing reference on page, is thus safe. However when TLB flush is done by a hypercall, e.g. in a Xen PV guest, there is no blocking thanks to disabled interrupts, and get_page() can succeed on a page that was already freed or even reused. We have recently seen this happen with our 4.4 and 4.12 based kernels, with userspace (java) that exits a thread, where mm_release() performs a futex_wake() on tsk->clear_child_tid, and another thread in parallel unmaps the page where tsk->clear_child_tid points to. The spurious get_page() succeeds, but futex code immediately releases the page again, while it's already on a freelist. Symptoms include a bad page state warning, general protection faults acessing a poisoned list prev/next pointer in the freelist, or free page pcplists of two cpus joined together in a single list. Oscar has also reproduced this scenario, with a patch inserting delays before the get_page() to make the race window larger. Fix this by removing the dependency on TLB flush interrupts the same way as the generic get_user_pages_fast() code by using page_cache_add_speculative() and revalidating the PTE contents after pinning the page. Mainline is safe since 4.13 where the x86 gup code was removed in favor of the common code. Accessing the page table itself safely also relies on disabled interrupts and TLB flush IPIs that don't happen with hypercalls, which was acknowledged in commit 9e52fc2b50de ("x86/mm: Enable RCU based page table freeing (CONFIG_HAVE_RCU_TABLE_FREE=y)"). That commit with follups should also be backported for full safety, although our reproducer didn't hit a problem without that backport. Reproduced-by: Oscar Salvador <osalvador@suse.de> Signed-off-by: Vlastimil Babka <vbabka@suse.cz> Cc: Thomas Gleixner <tglx@linutronix.de> Cc: Ingo Molnar <mingo@redhat.com> Cc: Peter Zijlstra <peterz@infradead.org> Cc: Juergen Gross <jgross@suse.com> Cc: Kirill A. Shutemov <kirill.shutemov@linux.intel.com> Cc: Vitaly Kuznetsov <vkuznets@redhat.com> Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Borislav Petkov <bp@alien8.de> Cc: Dave Hansen <dave.hansen@linux.intel.com> Cc: Andy Lutomirski <luto@kernel.org> --- Hi, I'm sending this stable-only patch for consideration because it's probably unrealistic to backport the 4.13 switch to generic GUP. I can look at 4.4 and 3.16 if accepted. The RCU page table freeing could be also considered. Note the patch also includes page refcount protection. I found out that 8fde12ca79af ("mm: prevent get_user_pages() from overflowing page refcount") backport to 4.9 missed the arch-specific gup implementations: https://lore.kernel.org/lkml/6650323f-dbc9-f069-000b-f6b0f941a065@suse.cz/ arch/x86/mm/gup.c | 32 ++++++++++++++++++++++++++++++-- 1 file changed, 30 insertions(+), 2 deletions(-)