CODEWITHSUNDEEP
clinuxsignals
Alex
Alex

Reputation: 43

Using sigaction for SIGCHLD to know when children terminate but unable to tell when the child SIGSEVs

According to the man page for sigaction, if we use it with SIGCHLD then in our handler function - this is if we use void (*sa_sigaction)(int, siginfo_t *, void *);

the si_code member of the siginfo_t structure should contain the reason the signal was sent.

I am interested in these reasons...

CLD_EXITED
CLD_KILLED
CLD_DUMPED

Now my question is... even if the child exited abnormally (i.e by executing buggy code to cause a crash) i still get CLD_EXITED. I assumed I should get CLD_DUMPED instead. As for CLD_KILLED, if I kill a child process, I get this one correctly.

What am I missing? In my parent process, I need to know if any of the child processes terminate abruptly.

I am causing the abrupt error by trying to write to a null pointer but I need to know of any kind of abnormal termination. It is not necessary to know exactly what or how it happened.

When is the CLD_DUMPED generated?

Kind Regards.

Upvotes: 2

Views: 1772

Answers (2)

R.. GitHub STOP HELPING ICE
R.. GitHub STOP HELPING ICE

Reputation: 215637

Rather than the si_code given to the signal handler, you should simply be looking at the exit status from waitpid (or one of the other wait-family functions) to determine the cause of exit. The macros W* from sys/wait.h, when applied to the status int, will tell you the cause of termination.

Upvotes: 1

Rumple Stiltskin
Rumple Stiltskin

Reputation: 10435

AFAIK, CLD_DUMPED is caused when the process dumps core.

Upvotes: 0

Related Questions