| Message ID | 1426849972-19606-1-git-send-email-a.ryabinin@samsung.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On 03/20/2015 02:12 PM, Andrey Ryabinin wrote: > Usually ELF_ET_DYN_BASE is 2/3 of TASK_SIZE. With 3G/1G user/kernel > split this is not so, because 2*TASK_SIZE overflows 32 bits, > so the actual value of ELF_ET_DYN_BASE is: > (2 * TASK_SIZE / 3) = 0x2a000000 AFAIK on most platforms (e.g. Intel) that's (TASK_SIZE / 3 * 2) so ARM is kind of special here. > > When ASLR is disabled PIE binaries will load at ELF_ET_DYN_BASE address. > On 32bit platforms AddressSanitzer uses addresses [0x20000000 - 0x40000000] > for shadow memory [1]. So ASan doesn't work for PIE binaries when ASLR disabled > as it fails to map shadow memory. > Also after Kees's 'split ET_DYN ASLR from mmap ASLR' patchset PIE binaries > has a high chance of loading somewhere in between [0x2a000000 - 0x40000000] > even if ASLR enabled. This makes ASan with PIE absolutely incompatible. > > Fix overflow by dividing TASK_SIZE prior to multiplying. > After this patch ELF_ET_DYN_BASE equals to (for CONFIG_VMSPLIT_3G=y): > (TASK_SIZE / 3 * 2) = 0x7f555554 > > [1] https://code.google.com/p/address-sanitizer/wiki/AddressSanitizerAlgorithm#Mapping Perhaps we should fix other platforms as well? -Y
On 03/20/2015 02:31 PM, Yury Gribov wrote: > On 03/20/2015 02:12 PM, Andrey Ryabinin wrote: >> Usually ELF_ET_DYN_BASE is 2/3 of TASK_SIZE. With 3G/1G user/kernel >> split this is not so, because 2*TASK_SIZE overflows 32 bits, >> so the actual value of ELF_ET_DYN_BASE is: >> (2 * TASK_SIZE / 3) = 0x2a000000 > > AFAIK on most platforms (e.g. Intel) that's (TASK_SIZE / 3 * 2) so ARM is kind of special here. > >> >> When ASLR is disabled PIE binaries will load at ELF_ET_DYN_BASE address. >> On 32bit platforms AddressSanitzer uses addresses [0x20000000 - 0x40000000] >> for shadow memory [1]. So ASan doesn't work for PIE binaries when ASLR disabled >> as it fails to map shadow memory. >> Also after Kees's 'split ET_DYN ASLR from mmap ASLR' patchset PIE binaries >> has a high chance of loading somewhere in between [0x2a000000 - 0x40000000] >> even if ASLR enabled. This makes ASan with PIE absolutely incompatible. >> >> Fix overflow by dividing TASK_SIZE prior to multiplying. >> After this patch ELF_ET_DYN_BASE equals to (for CONFIG_VMSPLIT_3G=y): >> (TASK_SIZE / 3 * 2) = 0x7f555554 >> >> [1] https://code.google.com/p/address-sanitizer/wiki/AddressSanitizerAlgorithm#Mapping > > Perhaps we should fix other platforms as well? > Seems only avr32 and cris platforms have the same problem. All other 32bit platforms where ELF_ET_DYB_BASE = (2 * TASK_SIZE / 3) have TASK_SIZE < 2G, so there is no overflow.
On Fri, Mar 20, 2015 at 4:12 AM, Andrey Ryabinin <a.ryabinin@samsung.com> wrote: > Usually ELF_ET_DYN_BASE is 2/3 of TASK_SIZE. With 3G/1G user/kernel > split this is not so, because 2*TASK_SIZE overflows 32 bits, > so the actual value of ELF_ET_DYN_BASE is: > (2 * TASK_SIZE / 3) = 0x2a000000 > > When ASLR is disabled PIE binaries will load at ELF_ET_DYN_BASE address. > On 32bit platforms AddressSanitzer uses addresses [0x20000000 - 0x40000000] > for shadow memory [1]. So ASan doesn't work for PIE binaries when ASLR disabled > as it fails to map shadow memory. > Also after Kees's 'split ET_DYN ASLR from mmap ASLR' patchset PIE binaries > has a high chance of loading somewhere in between [0x2a000000 - 0x40000000] > even if ASLR enabled. This makes ASan with PIE absolutely incompatible. > > Fix overflow by dividing TASK_SIZE prior to multiplying. > After this patch ELF_ET_DYN_BASE equals to (for CONFIG_VMSPLIT_3G=y): > (TASK_SIZE / 3 * 2) = 0x7f555554 > > [1] https://code.google.com/p/address-sanitizer/wiki/AddressSanitizerAlgorithm#Mapping > > Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com> > Reported-by: Maria Guseva <m.guseva@samsung.com> > Cc: stable@vger.kernel.org > --- > arch/arm/include/asm/elf.h | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/arch/arm/include/asm/elf.h b/arch/arm/include/asm/elf.h > index c1ff8ab..1984a92 100644 > --- a/arch/arm/include/asm/elf.h > +++ b/arch/arm/include/asm/elf.h > @@ -115,7 +115,7 @@ int dump_task_regs(struct task_struct *t, elf_gregset_t *elfregs); > the loader. We need to make sure that it is out of the way of the program > that it will "exec", and that there is sufficient room for the brk. */ > > -#define ELF_ET_DYN_BASE (2 * TASK_SIZE / 3) > +#define ELF_ET_DYN_BASE (TASK_SIZE / 3 * 2) > > /* When the program starts, a1 contains a pointer to a function to be > registered with atexit, as per the SVR4 ABI. A value of 0 means we Acked-by: Kees Cook <keescook@chromium.org> Though, yes, as mentioned in the other replies, please send fixes for the other architectures too. Thanks for catching this! -Kees
On Fri, Mar 20, 2015 at 02:12:52PM +0300, Andrey Ryabinin wrote: > Usually ELF_ET_DYN_BASE is 2/3 of TASK_SIZE. With 3G/1G user/kernel > split this is not so, because 2*TASK_SIZE overflows 32 bits, > so the actual value of ELF_ET_DYN_BASE is: > (2 * TASK_SIZE / 3) = 0x2a000000 > > When ASLR is disabled PIE binaries will load at ELF_ET_DYN_BASE address. > On 32bit platforms AddressSanitzer uses addresses [0x20000000 - 0x40000000] > for shadow memory [1]. So ASan doesn't work for PIE binaries when ASLR disabled > as it fails to map shadow memory. > Also after Kees's 'split ET_DYN ASLR from mmap ASLR' patchset PIE binaries > has a high chance of loading somewhere in between [0x2a000000 - 0x40000000] > even if ASLR enabled. This makes ASan with PIE absolutely incompatible. > > Fix overflow by dividing TASK_SIZE prior to multiplying. > After this patch ELF_ET_DYN_BASE equals to (for CONFIG_VMSPLIT_3G=y): > (TASK_SIZE / 3 * 2) = 0x7f555554 > > [1] https://code.google.com/p/address-sanitizer/wiki/AddressSanitizerAlgorithm#Mapping > > Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com> > Reported-by: Maria Guseva <m.guseva@samsung.com> > Cc: stable@vger.kernel.org Who's handling this patch? I'm guessing it should be me, so if it could find its way into my patch system for when I next apply a bunch of patches, that'd be good. Thanks.
On 03/26/2015 06:05 PM, Russell King - ARM Linux wrote: > On Fri, Mar 20, 2015 at 02:12:52PM +0300, Andrey Ryabinin wrote: >> Usually ELF_ET_DYN_BASE is 2/3 of TASK_SIZE. With 3G/1G user/kernel >> split this is not so, because 2*TASK_SIZE overflows 32 bits, >> so the actual value of ELF_ET_DYN_BASE is: >> (2 * TASK_SIZE / 3) = 0x2a000000 >> >> When ASLR is disabled PIE binaries will load at ELF_ET_DYN_BASE address. >> On 32bit platforms AddressSanitzer uses addresses [0x20000000 - 0x40000000] >> for shadow memory [1]. So ASan doesn't work for PIE binaries when ASLR disabled >> as it fails to map shadow memory. >> Also after Kees's 'split ET_DYN ASLR from mmap ASLR' patchset PIE binaries >> has a high chance of loading somewhere in between [0x2a000000 - 0x40000000] >> even if ASLR enabled. This makes ASan with PIE absolutely incompatible. >> >> Fix overflow by dividing TASK_SIZE prior to multiplying. >> After this patch ELF_ET_DYN_BASE equals to (for CONFIG_VMSPLIT_3G=y): >> (TASK_SIZE / 3 * 2) = 0x7f555554 >> >> [1] https://code.google.com/p/address-sanitizer/wiki/AddressSanitizerAlgorithm#Mapping >> >> Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com> >> Reported-by: Maria Guseva <m.guseva@samsung.com> >> Cc: stable@vger.kernel.org > > Who's handling this patch? I'm guessing it should be me, so if it could > find its way into my patch system for when I next apply a bunch of patches, > that'd be good. > It's already there. Patch number - 8320/1.
diff --git a/arch/arm/include/asm/elf.h b/arch/arm/include/asm/elf.h index c1ff8ab..1984a92 100644 --- a/arch/arm/include/asm/elf.h +++ b/arch/arm/include/asm/elf.h @@ -115,7 +115,7 @@ int dump_task_regs(struct task_struct *t, elf_gregset_t *elfregs); the loader. We need to make sure that it is out of the way of the program that it will "exec", and that there is sufficient room for the brk. */ -#define ELF_ET_DYN_BASE (2 * TASK_SIZE / 3) +#define ELF_ET_DYN_BASE (TASK_SIZE / 3 * 2) /* When the program starts, a1 contains a pointer to a function to be registered with atexit, as per the SVR4 ABI. A value of 0 means we
Usually ELF_ET_DYN_BASE is 2/3 of TASK_SIZE. With 3G/1G user/kernel split this is not so, because 2*TASK_SIZE overflows 32 bits, so the actual value of ELF_ET_DYN_BASE is: (2 * TASK_SIZE / 3) = 0x2a000000 When ASLR is disabled PIE binaries will load at ELF_ET_DYN_BASE address. On 32bit platforms AddressSanitzer uses addresses [0x20000000 - 0x40000000] for shadow memory [1]. So ASan doesn't work for PIE binaries when ASLR disabled as it fails to map shadow memory. Also after Kees's 'split ET_DYN ASLR from mmap ASLR' patchset PIE binaries has a high chance of loading somewhere in between [0x2a000000 - 0x40000000] even if ASLR enabled. This makes ASan with PIE absolutely incompatible. Fix overflow by dividing TASK_SIZE prior to multiplying. After this patch ELF_ET_DYN_BASE equals to (for CONFIG_VMSPLIT_3G=y): (TASK_SIZE / 3 * 2) = 0x7f555554 [1] https://code.google.com/p/address-sanitizer/wiki/AddressSanitizerAlgorithm#Mapping Signed-off-by: Andrey Ryabinin <a.ryabinin@samsung.com> Reported-by: Maria Guseva <m.guseva@samsung.com> Cc: stable@vger.kernel.org --- arch/arm/include/asm/elf.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)