sa_mask should probably be initialised before calling sigaction
I was going to file this bug for chrome but then I realised that breakpad
is a separate project ... I spotted this by running the chrome test_shell
under valgrind btw.
The change is here:
http://codereview.chromium.org/13653/show
but I'm not claiming that's 100% correct. In particular:
1. I'm not sure if any signals should actually be blocked during the signal
setup, and
2. Recompiling chrome with this fix works but for some inexplicable reason,
valgrind still claims act->sa_mask is uninitialised. I haven't dug into
that in detail.
Comments please!
Thank you
Original issue reported on code.google.com by craig.sc...@gmail.com on 9 Dec 2008 at 11:31
Original issue reported on code.google.com by
craig.sc...@gmail.com
on 9 Dec 2008 at 11:31Attachments: