powerpc/mm: Fix UBSAN warning reported on hugetlb
Subsystems (3)
PATCH 1/1 powerpc/mm: Fix UBSAN warning reported on hugetlb
Powerpc architecture supports 16GB hugetlb pages with hash translation. For 4K page size, this is implemented as a hugepage directory entry at PGD level and for 64K it is implemented as a huge page pte at PUD level
With 16GB hugetlb size, offset within a page is greater than 32 bits. Hence switch to use unsigned long type when using hugepd_shift.
Inorder to keep things simpler, we make sure we always use unsigned long type when using hugepd_shift() even though all the hugetlb page size won't require that.
The walk_hugepd_range change won't have any impact because we don't use that for the hugetlb walk. That code is needed to support hugepd on init_mm with PPC_8XX and 8M page. Even though 8M page size won't result in any real issue, we update that to keep it simpler.
The hugetlb_free_p*d_range changes are all related to nohash usage where we can have multiple pgd entries pointing to the same hugepd entries. Hence on book3s64 where we can have > 4GB hugetlb page size we will always find more < next even if we compute the value of more correctly.
Hence there is no functional change in this patch except that it fixes the below warning.
UBSAN: shift-out-of-bounds in arch/powerpc/mm/hugetlbpage.c:499:21 shift exponent 34 is too large for 32-bit type 'int' CPU: 39 PID: 1673 Comm: a.out Not tainted 6.0.0-rc2-00327-gee88a56e8517-dirty #1 Call Trace: [c00000002ccb3720] [c000000000cb21e4] dump_stack_lvl+0x98/0xe0 (unreliable) [c00000002ccb3760] [c000000000cacf60] ubsan_epilogue+0x18/0x70 [c00000002ccb37c0] [c000000000cac44c] __ubsan_handle_shift_out_of_bounds+0x1bc/0x390 [c00000002ccb38c0] [c0000000000d6f78] hugetlb_free_pgd_range+0x5d8/0x600 [c00000002ccb39f0] [c000000000550e94] free_pgtables+0x114/0x290 [c00000002ccb3ac0] [c00000000056cbe0] exit_mmap+0x150/0x550 [c00000002ccb3be0] [c00000000017bf0c] mmput+0xcc/0x210 [c00000002ccb3c20] [c00000000018f180] do_exit+0x420/0xdd0 [c00000002ccb3cf0] [c00000000018fcdc] do_group_exit+0x4c/0xd0 [c00000002ccb3d30] [c00000000018fd84] sys_exit_group+0x24/0x30 [c00000002ccb3d50] [c00000000003cde0] system_call_exception+0x250/0x600 [c00000002ccb3e10] [c00000000000c3bc] system_call_common+0xec/0x250
--- arch/powerpc/mm/hugetlbpage.c | 6 +++--- mm/pagewalk.c | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-)
arch/powerpc/mm/hugetlbpage.c +3 -3
@@ -392,7 +392,7 @@ static void hugetlb_free_pmd_range(struct mmu_gather *tlb, pud_t *pud,* single hugepage, but all of them point to * the same kmem cache that holds the hugepte. */ - more = addr + (1 <<hugepd_shift (*(hugepd_t *)pmd)); + more = addr + (1UL <<hugepd_shift (*(hugepd_t *)pmd));if (more > next) next = more;
@@ -434,7 +434,7 @@ static void hugetlb_free_pud_range(struct mmu_gather *tlb, p4d_t *p4d,* single hugepage, but all of them point to * the same kmem cache that holds the hugepte. */ - more = addr + (1 <<hugepd_shift (*(hugepd_t *)pud)); + more = addr + (1UL <<hugepd_shift (*(hugepd_t *)pud));if (more > next) next = more;
@@ -496,7 +496,7 @@ void hugetlb_free_pgd_range(struct mmu_gather *tlb,* for a single hugepage, but all of them point to the * same kmem cache that holds the hugepte. */ - more = addr + (1 <<hugepd_shift (*(hugepd_t *)pgd)); + more = addr + (1ULL <<hugepd_shift (*(hugepd_t *)pgd));if (more > next) next = more;
mm/pagewalk.c +1 -1
@@ -65,7 +65,7 @@ static int walk_hugepd_range(hugepd_t *phpd, unsigned long addr,int err =0 ;const struct mm_walk_ops *ops = walk->ops;int shift =hugepd_shift (*phpd); -int page_size =1 << shift; +long page_size =1UL << shift;if (!ops->pte_entry)return 0 ;
Signature · 1 line
2.37.3
Le 08/09/2022 à 09:24, Aneesh Kumar K.V a écrit :
Quoted 52 lines
Powerpc architecture supports 16GB hugetlb pages with hash translation. For 4K page size, this is implemented as a hugepage directory entry at PGD level and for 64K it is implemented as a huge page pte at PUD level
With 16GB hugetlb size, offset within a page is greater than 32 bits. Hence switch to use unsigned long type when using
hugepd_shift.Inorder to keep things simpler, we make sure we always use unsigned long type when using
hugepd_shift()even though all the hugetlb page size won't require that.The
walk_hugepd_rangechange won't have any impact because we don't use that for the hugetlb walk. That code is needed to support hugepd oninit_mmwithPPC_8XXand 8M page. Even though 8M page size won't result in any real issue, we update that to keep it simpler.The hugetlb_free_p*d_range changes are all related to nohash usage where we can have multiple pgd entries pointing to the same hugepd entries. Hence on book3s64 where we can have > 4GB hugetlb page size we will always find more < next even if we compute the value of more correctly.
Hence there is no functional change in this patch except that it fixes the below warning.
UBSAN: shift-out-of-bounds in arch/powerpc/mm/hugetlbpage.c:499:21 shift exponent 34 is too large for 32-bit type 'int' CPU: 39 PID: 1673 Comm: a.out Not tainted 6.0.0-rc2-00327-gee88a56e8517-dirty #1 Call Trace: [c00000002ccb3720] [c000000000cb21e4] dump_stack_lvl+0x98/0xe0 (unreliable) [c00000002ccb3760] [c000000000cacf60] ubsan_epilogue+0x18/0x70 [c00000002ccb37c0] [c000000000cac44c] __ubsan_handle_shift_out_of_bounds+0x1bc/0x390 [c00000002ccb38c0] [c0000000000d6f78] hugetlb_free_pgd_range+0x5d8/0x600 [c00000002ccb39f0] [c000000000550e94] free_pgtables+0x114/0x290 [c00000002ccb3ac0] [c00000000056cbe0] exit_mmap+0x150/0x550 [c00000002ccb3be0] [c00000000017bf0c] mmput+0xcc/0x210 [c00000002ccb3c20] [c00000000018f180] do_exit+0x420/0xdd0 [c00000002ccb3cf0] [c00000000018fcdc] do_group_exit+0x4c/0xd0 [c00000002ccb3d30] [c00000000018fd84] sys_exit_group+0x24/0x30 [c00000002ccb3d50] [c00000000003cde0] system_call_exception+0x250/0x600 [c00000002ccb3e10] [c00000000000c3bc] system_call_common+0xec/0x250Signed-off-by Aneesh Kumar K.V <[email protected]>--- arch/powerpc/mm/hugetlbpage.c | 6 +++--- mm/pagewalk.c | 2 +- 2 files changed, 4 insertions(+), 4 deletions(-)mm/pagewalk.c +0
-0
mm/pagewalk.c static int walk_hugepd_range(hugepd_t *phpd, unsigned long addr, · line 68
Quoted 5 lines
mm/pagewalk.c +1
-1@@ -65,7 +65,7 @@ static int walk_hugepd_range(hugepd_t *phpd, unsigned long addr,int err =0 ;const struct mm_walk_ops *ops = walk->ops;int shift =hugepd_shift (*phpd); -int page_size =1 << shift; +long page_size =1UL << shift;
1UL means _unsigned_ long. Should page_size be unsigned ?
Quoted 3 lines
mm/pagewalk.c +0
-0if (!ops->pte_entry)return 0 ;
Michael Ellerman
AppliedQuoted 8 lines · Aneesh Kumar K.V
Powerpc architecture supports 16GB hugetlb pages with hash translation. For 4K page size, this is implemented as a hugepage directory entry at PGD level and for 64K it is implemented as a huge page pte at PUD level
With 16GB hugetlb size, offset within a page is greater than 32 bits. Hence switch to use unsigned long type when using
hugepd_shift.[...]
Applied to powerpc/next.
[1/1] powerpc/mm: Fix UBSAN warning reported on hugetlb
https://git.kernel.org/powerpc/c/7dd3a7b90bca2c12e2146a47d63cf69a2f5d7e89
cheers