| Message ID | 1576486038-9899-17-git-send-email-amit.kachhap@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | arm64: return address signing | expand |
On Mon, Dec 16, 2019 at 02:17:18PM +0530, Amit Daniel Kachhap wrote: > This test is specific for arm64. When in-kernel Pointer Authentication > config is enabled, the return address stored in the stack is signed. > This feature helps in ROP kind of attack. If any parameters used to > generate the pac (<key, sp, lr>) is modified then this will fail in > the authentication stage and will lead to abort. > > This test changes the input parameter APIA kernel keys to cause abort. > The pac computed from the new key can be same as last due to hash > collision so this is retried for few times as there is no reliable way > to compare the pacs. Even though this test may fail even after retries > but this may cause authentication failure at a later stage in earlier > function returns. > > This test can be invoked as, > echo CORRUPT_PAC > /sys/kernel/debug/provoke-crash/DIRECT > > or as below if inserted as a module, > insmod lkdtm.ko cpoint_name=DIRECT cpoint_type=CORRUPT_PAC cpoint_count=1 > > [ 13.118166] lkdtm: Performing direct entry CORRUPT_PAC > [ 13.118298] lkdtm: Clearing PAC from the return address > [ 13.118466] Unable to handle kernel paging request at virtual address bfff8000108648ec > [ 13.118626] Mem abort info: > [ 13.118666] ESR = 0x86000004 > [ 13.118866] EC = 0x21: IABT (current EL), IL = 32 bits > [ 13.118966] SET = 0, FnV = 0 > [ 13.119117] EA = 0, S1PTW = 0 > > Cc: Kees Cook <keescook@chromium.org> > Signed-off-by: Amit Daniel Kachhap <amit.kachhap@arm.com> Acked-by: Catalin Marinas <catalin.marinas@arm.com>
diff --git a/drivers/misc/lkdtm/bugs.c b/drivers/misc/lkdtm/bugs.c index a4fdad0..b9ef583 100644 --- a/drivers/misc/lkdtm/bugs.c +++ b/drivers/misc/lkdtm/bugs.c @@ -376,3 +376,39 @@ void lkdtm_DOUBLE_FAULT(void) panic("tried to double fault but didn't die\n"); } #endif + +#ifdef CONFIG_ARM64_PTR_AUTH +static noinline void change_pac_parameters(void) +{ + /* Reset the keys of current task */ + ptrauth_thread_init_kernel(current); + ptrauth_thread_switch_kernel(current); +} + +#define CORRUPT_PAC_ITERATE 10 +noinline void lkdtm_CORRUPT_PAC(void) +{ + int i; + + if (!system_supports_address_auth()) { + pr_err("FAIL: arm64 pointer authentication feature not present\n"); + return; + } + + pr_info("Change the PAC parameters to force function return failure\n"); + /* + * Pac is a hash value computed from input keys, return address and + * stack pointer. As pac has fewer bits so there is a chance of + * collision, so iterate few times to reduce the collision probability. + */ + for (i = 0; i < CORRUPT_PAC_ITERATE; i++) + change_pac_parameters(); + + pr_err("FAIL: %s test failed. Kernel may be unstable from here\n", __func__); +} +#else /* !CONFIG_ARM64_PTR_AUTH */ +noinline void lkdtm_CORRUPT_PAC(void) +{ + pr_err("FAIL: arm64 pointer authentication config disabled\n"); +} +#endif diff --git a/drivers/misc/lkdtm/core.c b/drivers/misc/lkdtm/core.c index ee0d6e7..5ce4ac8 100644 --- a/drivers/misc/lkdtm/core.c +++ b/drivers/misc/lkdtm/core.c @@ -116,6 +116,7 @@ static const struct crashtype crashtypes[] = { CRASHTYPE(STACK_GUARD_PAGE_LEADING), CRASHTYPE(STACK_GUARD_PAGE_TRAILING), CRASHTYPE(UNSET_SMEP), + CRASHTYPE(CORRUPT_PAC), CRASHTYPE(UNALIGNED_LOAD_STORE_WRITE), CRASHTYPE(OVERWRITE_ALLOCATION), CRASHTYPE(WRITE_AFTER_FREE), diff --git a/drivers/misc/lkdtm/lkdtm.h b/drivers/misc/lkdtm/lkdtm.h index c56d23e..8d13d01 100644 --- a/drivers/misc/lkdtm/lkdtm.h +++ b/drivers/misc/lkdtm/lkdtm.h @@ -31,6 +31,7 @@ void lkdtm_UNSET_SMEP(void); #ifdef CONFIG_X86_32 void lkdtm_DOUBLE_FAULT(void); #endif +void lkdtm_CORRUPT_PAC(void); /* lkdtm_heap.c */ void __init lkdtm_heap_init(void);
This test is specific for arm64. When in-kernel Pointer Authentication config is enabled, the return address stored in the stack is signed. This feature helps in ROP kind of attack. If any parameters used to generate the pac (<key, sp, lr>) is modified then this will fail in the authentication stage and will lead to abort. This test changes the input parameter APIA kernel keys to cause abort. The pac computed from the new key can be same as last due to hash collision so this is retried for few times as there is no reliable way to compare the pacs. Even though this test may fail even after retries but this may cause authentication failure at a later stage in earlier function returns. This test can be invoked as, echo CORRUPT_PAC > /sys/kernel/debug/provoke-crash/DIRECT or as below if inserted as a module, insmod lkdtm.ko cpoint_name=DIRECT cpoint_type=CORRUPT_PAC cpoint_count=1 [ 13.118166] lkdtm: Performing direct entry CORRUPT_PAC [ 13.118298] lkdtm: Clearing PAC from the return address [ 13.118466] Unable to handle kernel paging request at virtual address bfff8000108648ec [ 13.118626] Mem abort info: [ 13.118666] ESR = 0x86000004 [ 13.118866] EC = 0x21: IABT (current EL), IL = 32 bits [ 13.118966] SET = 0, FnV = 0 [ 13.119117] EA = 0, S1PTW = 0 Cc: Kees Cook <keescook@chromium.org> Signed-off-by: Amit Daniel Kachhap <amit.kachhap@arm.com> --- Changes since last version: * The pointer authentication error is now caused by changing the kernel keys as suggested by Ard. * Richard [1] suggested to iterate till a new pac value is generated but variables required to compute the pac cannot be acquired in a generic way so this test just retries crashing for some time and then stops. [1]: http://lists.infradead.org/pipermail/linux-arm-kernel/2019-November/696084.html drivers/misc/lkdtm/bugs.c | 36 ++++++++++++++++++++++++++++++++++++ drivers/misc/lkdtm/core.c | 1 + drivers/misc/lkdtm/lkdtm.h | 1 + 3 files changed, 38 insertions(+)