GDB (not) handling SIGINT...?

[Pedro Alves]

On 11/15/2018 05:12 PM, Paul Smith wrote:

> Oh, I understand now.  The SIGINT is passed to the child to get it to
> stop.  Somehow I assumed that there was a magical ptrace way to do
> this, rather than having to send normal signals.
> 

Exactly.

> I guess there is now: using PTRACE_SEIZE instead of PTRACE_ATTACH,
> which will allow you to later send PTRACE_INTERRUPT: presumably if this
> were used would solve the problem for attached processes anyway...?
Yup, assuming we're OK with "ctrl-c" in the GDB console not
resulting in SIGINT.

Thanks,
Pedro Alves