[PATCH 1/7] Merge async and sync code paths some more
Yao Qi
qiyaoltc@gmail.com
Thu Aug 27 13:26:00 GMT 2015
Pedro Alves <palves@redhat.com> writes:
> Nope, not yet at least, though centralizing sigint handling in the event
> loop would probably be a good goal.
>
> It's actually still handled by set_sigint_trap -> inflow.c:pass_signal.
> The difference is that after this patch, linux_nat_terminal_inferior
> also calls set_sigint_trap itself in sync mode, so these call in
> linux_nat_wait_1 are no longer necessary. They actually become harmful,
> because set_sigint_trap/clear_sigint_trap are not idempotent.
> A sequence of calls like set_sigint_trap -> set_sigint_trap -> clear_sigint_trap
> ends up with the wrong SIGINT handler set...
OK, I see.
>
> Thanks, here's the patch that I plan to push, unless you have
> further comments.
I don't have comments, please go ahead.
--
Yao (齐尧)
More information about the Gdb-patches
mailing list