| Message ID | 1372780685-2528-1-git-send-email-dserrg@gmail.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Tue, Jul 02, 2013 at 04:58:05PM +0100, Sergey Dyasly wrote: > With LPAE enabled, physical address space is larger than 4GB. Allow mapping any > part of it via /dev/mem by utilizing the same check as in arm64. > > PHYS_MASK covers 40 bits with LPAE enabled and 32 bits otherwise. > > Signed-off-by: Sergey Dyasly <dserrg@gmail.com> > --- > > v2: > - Removed #ifdef and changed original function as suggested by Catalin Marinas > > arch/arm/mm/mmap.c | 6 ++---- > 1 file changed, 2 insertions(+), 4 deletions(-) > > diff --git a/arch/arm/mm/mmap.c b/arch/arm/mm/mmap.c > index 10062ce..09bab92 100644 > --- a/arch/arm/mm/mmap.c > +++ b/arch/arm/mm/mmap.c > @@ -204,13 +204,11 @@ int valid_phys_addr_range(phys_addr_t addr, size_t size) > } > > /* > - * We don't use supersection mappings for mmap() on /dev/mem, which > - * means that we can't map the memory area above the 4G barrier into > - * userspace. > + * Do not allow /dev/mem mappings beyond the supported physical range. > */ > int valid_mmap_phys_addr_range(unsigned long pfn, size_t size) > { > - return !(pfn + (size >> PAGE_SHIFT) > 0x00100000); > + return !(((pfn << PAGE_SHIFT) + size) & ~PHYS_MASK); I don't think that's what I suggested. For classic MMU ~PHYS_MASK is 0.
On Tue, 2 Jul 2013 18:15:31 +0100 Catalin Marinas <catalin.marinas@arm.com> wrote: > > int valid_mmap_phys_addr_range(unsigned long pfn, size_t size) > > { > > - return !(pfn + (size >> PAGE_SHIFT) > 0x00100000); > > + return !(((pfn << PAGE_SHIFT) + size) & ~PHYS_MASK); > > I don't think that's what I suggested. For classic MMU ~PHYS_MASK is 0. > Thanks for your comments! Indeed, I did some testing today and v2 check doesn't work at all because of 32-bit overflow in case of using '<<' operator. Also, it turned out that v1 doesn't reject mappings > 40 bit in case of LPAE (the same overflow). I didn't want to additionally define PHYS_MASK_SHIFT, so I came up with something like: return (pfn + (size >> PAGE_SHIFT)) <= (1 + (PHYS_MASK >> PAGE_SHIFT)) I will post v3 tomorrow. Best regards, Sergey Dyasly
diff --git a/arch/arm/mm/mmap.c b/arch/arm/mm/mmap.c index 10062ce..09bab92 100644 --- a/arch/arm/mm/mmap.c +++ b/arch/arm/mm/mmap.c @@ -204,13 +204,11 @@ int valid_phys_addr_range(phys_addr_t addr, size_t size) } /* - * We don't use supersection mappings for mmap() on /dev/mem, which - * means that we can't map the memory area above the 4G barrier into - * userspace. + * Do not allow /dev/mem mappings beyond the supported physical range. */ int valid_mmap_phys_addr_range(unsigned long pfn, size_t size) { - return !(pfn + (size >> PAGE_SHIFT) > 0x00100000); + return !(((pfn << PAGE_SHIFT) + size) & ~PHYS_MASK); } #ifdef CONFIG_STRICT_DEVMEM
With LPAE enabled, physical address space is larger than 4GB. Allow mapping any part of it via /dev/mem by utilizing the same check as in arm64. PHYS_MASK covers 40 bits with LPAE enabled and 32 bits otherwise. Signed-off-by: Sergey Dyasly <dserrg@gmail.com> --- v2: - Removed #ifdef and changed original function as suggested by Catalin Marinas arch/arm/mm/mmap.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-)