| Message ID | 20180522235049.23400-1-labbott@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Laura, On Tue, May 22, 2018 at 04:50:49PM -0700, Laura Abbott wrote: > Commit 15122ee2c515 ("arm64: Enforce BBM for huge IO/VMAP mappings") > disallowed block mappings for ioremap since that code does not honor > break-before-make. The same APIs are also used for permission updating > though and the extra checks prevent the permission updates from happening, > even though this should be permitted. This results in read-only permissions > not being fully applied. Visibly, this can occasionaly be seen as a failure > on the built in rodata test when the test data ends up in a section or > as an odd RW gap on the page table dump. Fix this by keeping the check > for the top level p*d_set_huge APIs but using separate functions for the > update APIs. > > Reported-by: Peter Robinson <pbrobinson@gmail.com> > Fixes: 15122ee2c515 ("arm64: Enforce BBM for huge IO/VMAP mappings") > Signed-off-by: Laura Abbott <labbott@redhat.com> > --- > arch/arm64/mm/mmu.c | 28 ++++++++++++++++++++-------- > 1 file changed, 20 insertions(+), 8 deletions(-) Thanks for sending the fix. One thing below... > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > index 2dbb2c9f1ec1..57517ad86910 100644 > --- a/arch/arm64/mm/mmu.c > +++ b/arch/arm64/mm/mmu.c > @@ -66,6 +66,9 @@ static pte_t bm_pte[PTRS_PER_PTE] __page_aligned_bss; > static pmd_t bm_pmd[PTRS_PER_PMD] __page_aligned_bss __maybe_unused; > static pud_t bm_pud[PTRS_PER_PUD] __page_aligned_bss __maybe_unused; > > +static void __pmd_set_huge(pmd_t *pmdp, phys_addr_t phys, pgprot_t prot); > +static void __pud_set_huge(pud_t *pudp, phys_addr_t phys, pgprot_t prot); > + > pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn, > unsigned long size, pgprot_t vma_prot) > { > @@ -200,7 +203,7 @@ static void init_pmd(pud_t *pudp, unsigned long addr, unsigned long end, > /* try section mapping first */ > if (((addr | next | phys) & ~SECTION_MASK) == 0 && > (flags & NO_BLOCK_MAPPINGS) == 0) { > - pmd_set_huge(pmdp, phys, prot); > + __pmd_set_huge(pmdp, phys, prot); Given that there is ongoing work to fix the core ioremap code, it would be nice to avoid adding '__' versions if we can help it. Would it work if we replaced the pXd_present check with a call to pgattr_change_is_safe instead? Will
diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index 2dbb2c9f1ec1..57517ad86910 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -66,6 +66,9 @@ static pte_t bm_pte[PTRS_PER_PTE] __page_aligned_bss; static pmd_t bm_pmd[PTRS_PER_PMD] __page_aligned_bss __maybe_unused; static pud_t bm_pud[PTRS_PER_PUD] __page_aligned_bss __maybe_unused; +static void __pmd_set_huge(pmd_t *pmdp, phys_addr_t phys, pgprot_t prot); +static void __pud_set_huge(pud_t *pudp, phys_addr_t phys, pgprot_t prot); + pgprot_t phys_mem_access_prot(struct file *file, unsigned long pfn, unsigned long size, pgprot_t vma_prot) { @@ -200,7 +203,7 @@ static void init_pmd(pud_t *pudp, unsigned long addr, unsigned long end, /* try section mapping first */ if (((addr | next | phys) & ~SECTION_MASK) == 0 && (flags & NO_BLOCK_MAPPINGS) == 0) { - pmd_set_huge(pmdp, phys, prot); + __pmd_set_huge(pmdp, phys, prot); /* * After the PMD entry has been populated once, we @@ -299,7 +302,7 @@ static void alloc_init_pud(pgd_t *pgdp, unsigned long addr, unsigned long end, */ if (use_1G_block(addr, next, phys) && (flags & NO_BLOCK_MAPPINGS) == 0) { - pud_set_huge(pudp, phys, prot); + __pud_set_huge(pudp, phys, prot); /* * After the PUD entry has been populated once, we @@ -929,31 +932,40 @@ int __init arch_ioremap_pmd_supported(void) return 1; } -int pud_set_huge(pud_t *pudp, phys_addr_t phys, pgprot_t prot) +void __pud_set_huge(pud_t *pudp, phys_addr_t phys, pgprot_t prot) { pgprot_t sect_prot = __pgprot(PUD_TYPE_SECT | pgprot_val(mk_sect_prot(prot))); + BUG_ON(phys & ~PUD_MASK); + set_pud(pudp, pfn_pud(__phys_to_pfn(phys), sect_prot)); +} + +int pud_set_huge(pud_t *pudp, phys_addr_t phys, pgprot_t prot) +{ /* ioremap_page_range doesn't honour BBM */ if (pud_present(READ_ONCE(*pudp))) return 0; - BUG_ON(phys & ~PUD_MASK); - set_pud(pudp, pfn_pud(__phys_to_pfn(phys), sect_prot)); + __pud_set_huge(pudp, phys, prot); return 1; } -int pmd_set_huge(pmd_t *pmdp, phys_addr_t phys, pgprot_t prot) +static void __pmd_set_huge(pmd_t *pmdp, phys_addr_t phys, pgprot_t prot) { pgprot_t sect_prot = __pgprot(PMD_TYPE_SECT | pgprot_val(mk_sect_prot(prot))); + BUG_ON(phys & ~PMD_MASK); + set_pmd(pmdp, pfn_pmd(__phys_to_pfn(phys), sect_prot)); +} +int pmd_set_huge(pmd_t *pmdp, phys_addr_t phys, pgprot_t prot) +{ /* ioremap_page_range doesn't honour BBM */ if (pmd_present(READ_ONCE(*pmdp))) return 0; - BUG_ON(phys & ~PMD_MASK); - set_pmd(pmdp, pfn_pmd(__phys_to_pfn(phys), sect_prot)); + __pmd_set_huge(pmdp, phys, prot); return 1; }
Commit 15122ee2c515 ("arm64: Enforce BBM for huge IO/VMAP mappings") disallowed block mappings for ioremap since that code does not honor break-before-make. The same APIs are also used for permission updating though and the extra checks prevent the permission updates from happening, even though this should be permitted. This results in read-only permissions not being fully applied. Visibly, this can occasionaly be seen as a failure on the built in rodata test when the test data ends up in a section or as an odd RW gap on the page table dump. Fix this by keeping the check for the top level p*d_set_huge APIs but using separate functions for the update APIs. Reported-by: Peter Robinson <pbrobinson@gmail.com> Fixes: 15122ee2c515 ("arm64: Enforce BBM for huge IO/VMAP mappings") Signed-off-by: Laura Abbott <labbott@redhat.com> --- arch/arm64/mm/mmu.c | 28 ++++++++++++++++++++-------- 1 file changed, 20 insertions(+), 8 deletions(-)