segmentation fault handler crashes with segmentation fault

Question

I am using arm machine and segmentation handler sometimes crashes while obtaining crash context address:

void crit_err_hdlr(int sig_num, siginfo_t * info, void * ucontext)
{
 uintptr_t fault_address = NULL;
 struct sigcontext *ctx = &(((ucontext_t*)ucontext)->uc_mcontext);
 fault_address = ctx->arm_pc;

 //after reading ctx->arm_pc it crashes
}

This is how handler is setup:

struct sigaction sigact = { 0 };
sigact.sa_sigaction = crit_err_hdlr;
sigact.sa_flags = SA_SIGINFO |  SA_RESTART;
if (sigaction(SIGSEGV, &sigact, (struct sigaction *)NULL) != 0)
{
 //error, bail out
}

Is there any possible way to check if arm_pc filed in sigcontext is readable, so I can avoid the crash in crash handler?

Handling `SIGSEGV` is critical, as more or less everything of the process's memory management might already be trashed, in the moment the handler is called. — alk, Jun 15 '13 at 15:17
What OS are you talking about? Linux, Posix, etc.? See [isbadreadptr on Unix](http://stackoverflow.com/questions/4611776/isbadreadptr-analogue-on-unix) or [IsBadReadPtr search](http://stackoverflow.com/search?q=IsBadReadptr+is%3Aquestion). — artless noise, Jun 17 '13 at 01:18

score 1 · Accepted Answer · answered Jun 17 '13 at 15:42

1

The problem was the memory corruption performing operations inside the handler

answered Jun 17 '13 at 15:42

Ulterior

2,786
3
30
58

segmentation fault handler crashes with segmentation fault

1 Answers1