| Message ID | 1446587315-18145-2-git-send-email-labbott@fedoraproject.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Hi Laura, On 3 November 2015 at 22:48, Laura Abbott <labbott@fedoraproject.org> wrote: > > Rather than always putting the least restrictived permissions restrictive > (PAGE_KERNEL_EXEC) when spliting a pmd into pages, use > the existing permissions from the pmd for the page. > > Signed-off-by: Laura Abbott <labbott@fedoraproject.org> > --- > arch/arm64/mm/mmu.c | 9 ++++----- > 1 file changed, 4 insertions(+), 5 deletions(-) > > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > index 9211b85..ff41efa 100644 > --- a/arch/arm64/mm/mmu.c > +++ b/arch/arm64/mm/mmu.c > @@ -75,14 +75,13 @@ static void __init *early_alloc(unsigned long sz) > static void split_pmd(pmd_t *pmd, pte_t *pte) > { > unsigned long pfn = pmd_pfn(*pmd); > + unsigned long addr = pfn << PAGE_SHIFT; > + pgprot_t prot = __pgprot(pmd_val(*pmd) ^ addr) | PTE_TYPE_PAGE; pgprot_t is a struct type when STRICT_MM_TYPECHECKS is in effect, so __pgprot() should cover the entire expression. With the above fixed: Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org> Thanks, > + > int i = 0; > > do { > - /* > - * Need to have the least restrictive permissions available > - * permissions will be fixed up later > - */ > - set_pte(pte, pfn_pte(pfn, PAGE_KERNEL_EXEC)); > + set_pte(pte, pfn_pte(pfn, prot)); > pfn++; > } while (pte++, i++, i < PTRS_PER_PTE); > } > -- > 2.4.3 >
On 2015/11/4 5:48, Laura Abbott wrote: > > Rather than always putting the least restrictived permissions > (PAGE_KERNEL_EXEC) when spliting a pmd into pages, use > the existing permissions from the pmd for the page. > > Signed-off-by: Laura Abbott <labbott@fedoraproject.org> > --- > arch/arm64/mm/mmu.c | 9 ++++----- > 1 file changed, 4 insertions(+), 5 deletions(-) > > diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c > index 9211b85..ff41efa 100644 > --- a/arch/arm64/mm/mmu.c > +++ b/arch/arm64/mm/mmu.c > @@ -75,14 +75,13 @@ static void __init *early_alloc(unsigned long sz) > static void split_pmd(pmd_t *pmd, pte_t *pte) > { > unsigned long pfn = pmd_pfn(*pmd); > + unsigned long addr = pfn << PAGE_SHIFT; > + pgprot_t prot = __pgprot(pmd_val(*pmd) ^ addr) | PTE_TYPE_PAGE; > + Hi Laura, I'm not quite understand, I find split_pud() doesn't set the flag PMD_TYPE_TABLE. If we clear xx_TABLE_BIT, does that means the page is large page? And what is the different from the flag xx_TYPE_SECT? Thanks, Xishi Qiu > int i = 0; > > do { > - /* > - * Need to have the least restrictive permissions available > - * permissions will be fixed up later > - */ > - set_pte(pte, pfn_pte(pfn, PAGE_KERNEL_EXEC)); > + set_pte(pte, pfn_pte(pfn, prot)); > pfn++; > } while (pte++, i++, i < PTRS_PER_PTE); > }
On 11/05/2015 02:15 AM, Xishi Qiu wrote: > On 2015/11/4 5:48, Laura Abbott wrote: > >> >> Rather than always putting the least restrictived permissions >> (PAGE_KERNEL_EXEC) when spliting a pmd into pages, use >> the existing permissions from the pmd for the page. >> >> Signed-off-by: Laura Abbott <labbott@fedoraproject.org> >> --- >> arch/arm64/mm/mmu.c | 9 ++++----- >> 1 file changed, 4 insertions(+), 5 deletions(-) >> >> diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c >> index 9211b85..ff41efa 100644 >> --- a/arch/arm64/mm/mmu.c >> +++ b/arch/arm64/mm/mmu.c >> @@ -75,14 +75,13 @@ static void __init *early_alloc(unsigned long sz) >> static void split_pmd(pmd_t *pmd, pte_t *pte) >> { >> unsigned long pfn = pmd_pfn(*pmd); >> + unsigned long addr = pfn << PAGE_SHIFT; >> + pgprot_t prot = __pgprot(pmd_val(*pmd) ^ addr) | PTE_TYPE_PAGE; >> + > > Hi Laura, > > I'm not quite understand, I find split_pud() doesn't set the flag > PMD_TYPE_TABLE. If we clear xx_TABLE_BIT, does that means the page > is large page? I'm assuming by large page you mean a block mapping. Yes, without that entry the kernel treats this as a block mapping > And what is the different from the flag xx_TYPE_SECT? That would mark it this as a block mapping which is not what we want here. > > Thanks, > Xishi Qiu >
diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index 9211b85..ff41efa 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -75,14 +75,13 @@ static void __init *early_alloc(unsigned long sz) static void split_pmd(pmd_t *pmd, pte_t *pte) { unsigned long pfn = pmd_pfn(*pmd); + unsigned long addr = pfn << PAGE_SHIFT; + pgprot_t prot = __pgprot(pmd_val(*pmd) ^ addr) | PTE_TYPE_PAGE; + int i = 0; do { - /* - * Need to have the least restrictive permissions available - * permissions will be fixed up later - */ - set_pte(pte, pfn_pte(pfn, PAGE_KERNEL_EXEC)); + set_pte(pte, pfn_pte(pfn, prot)); pfn++; } while (pte++, i++, i < PTRS_PER_PTE); }
Rather than always putting the least restrictived permissions (PAGE_KERNEL_EXEC) when spliting a pmd into pages, use the existing permissions from the pmd for the page. Signed-off-by: Laura Abbott <labbott@fedoraproject.org> --- arch/arm64/mm/mmu.c | 9 ++++----- 1 file changed, 4 insertions(+), 5 deletions(-)