| Message ID | 20240528205323.20439-1-osalvador@suse.de (mailing list archive) |
|---|---|
| State | New |
| Headers | show |
| Series | mm/hugetlb: Do not call vma_add_reservation upon ENOMEM | expand |
On Tue, May 28, 2024 at 10:53:23PM +0200, Oscar Salvador wrote: > sysbot reported a splat [1] on __unmap_hugepage_range(). > This is because vma_needs_reservation() can return -ENOMEM if > allocate_file_region_entries() fails to allocate the file_region struct for > the reservation. > Check for that and do not call vma_add_reservation() if that is the case, > otherwise region_abort() and region_del() will see that we do not have any > file_regions. > > If we detect that vma_needs_reservation() returned -ENOMEM, we clear the > hugetlb_restore_reserve flag as if this reservation was still consumed, > so free_huge_folio() will not increment the resv count. > > [1] https://lore.kernel.org/linux-mm/0000000000004096100617c58d54@google.com/T/#ma5983bc1ab18a54910da83416b3f89f3c7ee43aa > > Fixes: df7a6d1f6405 ("mm/hugetlb: restore the reservation if needed") > Reported-and-tested-by: syzbot+d3fe2dc5ffe9380b714b@syzkaller.appspotmail.com > Closes: https://lore.kernel.org/linux-mm/0000000000004096100617c58d54@google.com/ > Signed-off-by: Oscar Salvador <osalvador@suse.de> > --- > mm/hugetlb.c | 16 ++++++++++++++-- > 1 file changed, 14 insertions(+), 2 deletions(-) > > diff --git a/mm/hugetlb.c b/mm/hugetlb.c > index 6be78e7d4f6e..f35abff8be60 100644 > --- a/mm/hugetlb.c > +++ b/mm/hugetlb.c > @@ -5768,8 +5768,20 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma, > * do_exit() will not see it, and will keep the reservation > * forever. > */ > - if (adjust_reservation && vma_needs_reservation(h, vma, address)) > - vma_add_reservation(h, vma, address); > + if (adjust_reservation) { > + int rc = vma_needs_reservation(h, vma, address); vma_needs_reservation returns a long. static long vma_needs_reservation(struct hstate *h, struct vm_area_struct *vma, unsigned long addr) Other than that, this is reviewed. Reviewed-by: Breno Leitao <leitao@debian.org> Thanks for fixing it.
On Wed, May 29, 2024 at 08:43:06AM -0700, Breno Leitao wrote: > vma_needs_reservation returns a long. > > static long vma_needs_reservation(struct hstate *h, > struct vm_area_struct *vma, unsigned long addr) Yeah, I missed that, thanks. @Andrew: Can you fold the following in? Thanks diff --git a/mm/hugetlb.c b/mm/hugetlb.c index f35abff8be60..0b9e64aa279b 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -5769,7 +5769,7 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma, * forever. */ if (adjust_reservation) { - int rc = vma_needs_reservation(h, vma, address); + long rc = vma_needs_reservation(h, vma, address); if (rc < 0) /* Pressumably allocate_file_region_entries failed > Reviewed-by: Breno Leitao <leitao@debian.org> Thanks
diff --git a/mm/hugetlb.c b/mm/hugetlb.c index 6be78e7d4f6e..f35abff8be60 100644 --- a/mm/hugetlb.c +++ b/mm/hugetlb.c @@ -5768,8 +5768,20 @@ void __unmap_hugepage_range(struct mmu_gather *tlb, struct vm_area_struct *vma, * do_exit() will not see it, and will keep the reservation * forever. */ - if (adjust_reservation && vma_needs_reservation(h, vma, address)) - vma_add_reservation(h, vma, address); + if (adjust_reservation) { + int rc = vma_needs_reservation(h, vma, address); + + if (rc < 0) + /* Pressumably allocate_file_region_entries failed + * to allocate a file_region struct. Clear + * hugetlb_restore_reserve so that global reserve + * count will not be incremented by free_huge_folio. + * Act as if we consumed the reservation. + */ + folio_clear_hugetlb_restore_reserve(page_folio(page)); + else if (rc) + vma_add_reservation(h, vma, address); + } tlb_remove_page_size(tlb, page, huge_page_size(h)); /*
sysbot reported a splat [1] on __unmap_hugepage_range(). This is because vma_needs_reservation() can return -ENOMEM if allocate_file_region_entries() fails to allocate the file_region struct for the reservation. Check for that and do not call vma_add_reservation() if that is the case, otherwise region_abort() and region_del() will see that we do not have any file_regions. If we detect that vma_needs_reservation() returned -ENOMEM, we clear the hugetlb_restore_reserve flag as if this reservation was still consumed, so free_huge_folio() will not increment the resv count. [1] https://lore.kernel.org/linux-mm/0000000000004096100617c58d54@google.com/T/#ma5983bc1ab18a54910da83416b3f89f3c7ee43aa Fixes: df7a6d1f6405 ("mm/hugetlb: restore the reservation if needed") Reported-and-tested-by: syzbot+d3fe2dc5ffe9380b714b@syzkaller.appspotmail.com Closes: https://lore.kernel.org/linux-mm/0000000000004096100617c58d54@google.com/ Signed-off-by: Oscar Salvador <osalvador@suse.de> --- mm/hugetlb.c | 16 ++++++++++++++-- 1 file changed, 14 insertions(+), 2 deletions(-)