diff options
| author | Nobuyoshi Nakada <[email protected]> | 2024-04-03 23:14:52 +0900 |
|---|---|---|
| committer | Nobuyoshi Nakada <[email protected]> | 2024-04-03 23:14:52 +0900 |
| commit | f928e60d41314a9aad2d9629114bb3ae57f51096 (patch) | |
| tree | ac575311fb153385fa3f8ad470c648dd044cbb40 /signal.c | |
| parent | 4960a598d6bc42adcfd488c7f455b2defcdc8161 (diff) | |
Check if a fatal signal is from kernel
On Linux, `siginfo_t` uses a union for each `si_code`, and the field
corresponding to `si_pid` does not belong to the `_sigfault` field for
SIGSEGV. It actually overlaps the `si_addr` field, which is usually
non-zero on stack overflow.
https://github.com/ruby/ruby/pull/10201#issuecomment-2034723244
Diffstat (limited to 'signal.c')
| -rw-r--r-- | signal.c | 7 |
1 files changed, 5 insertions, 2 deletions
@@ -867,16 +867,19 @@ check_stack_overflow(int sig, const void *addr) } } # endif + # ifdef _WIN32 # define CHECK_STACK_OVERFLOW() check_stack_overflow(sig, 0) # else # define FAULT_ADDRESS info->si_addr # ifdef USE_UCONTEXT_REG -# define CHECK_STACK_OVERFLOW() (info->si_pid ? (void)0 : check_stack_overflow(sig, (uintptr_t)FAULT_ADDRESS, ctx)) +# define CHECK_STACK_OVERFLOW_() check_stack_overflow(sig, (uintptr_t)FAULT_ADDRESS, ctx) # else -# define CHECK_STACK_OVERFLOW() (info->si_pid ? (void)0 : check_stack_overflow(sig, FAULT_ADDRESS)) +# define CHECK_STACK_OVERFLOW_() check_stack_overflow(sig, FAULT_ADDRESS) # endif # define MESSAGE_FAULT_ADDRESS " at %p", FAULT_ADDRESS +# define SIGNAL_FROM_USER_P() ((info)->si_code == SI_USER) +# define CHECK_STACK_OVERFLOW() (SIGNAL_FROM_USER_P() ? (void)0 : CHECK_STACK_OVERFLOW_()) # endif #else # define CHECK_STACK_OVERFLOW() (void)0 |