Diagnostic patch: add stack trace on SIGQUIT
rea at codelabs.ru
Sat Feb 2 12:30:45 GMT 2013
Mon, Jan 28, 2013 at 11:53:19PM +0400, Eygene Ryabinkin wrote:
> Today at the morning I had faced a weird lockup that most likely was
> due to some locking problem: python was waiting in the 'usem' state.
> I had no tools to trace the threading state, so I had left this issue
> for later investigations, though it isn't reliably reproducible for me.
> After some thinking, I had came up with the patch that prints the call
> stack for all active threads on SIGQUIT (and tries to dump core, as usual
> for SIGQUIT),
> This patch isn't the universal tool in its current form, since if the
> main thread is busy with waiting for some event, it won't respond to
> the signal immediately. However, if it is not blocked, handler will
> print the proper trace that can be useful for own analysis and for
> sending such diagnostics to other developers.
Pushed this patch to 'next'.
More information about the OfflineIMAP-project