Date: Mon, 03 Nov 2014 18:41:23 -0600 From: The BSD Dreamer <beastie@tardisi.com> To: freebsd-chromium@freebsd.org Subject: Re: Still segfault at start Message-ID: <545820B3.2050508@tardisi.com> In-Reply-To: <0242ddfdf4f8ab796793d3f186e60d39@webmail.olejniczak.info> References: <54562C5D.5050603@freebsd.org> <20141102100531.GA5218@lena.kiev> <0242ddfdf4f8ab796793d3f186e60d39@webmail.olejniczak.info>
next in thread | previous in thread | raw e-mail | index | archive | help
I had PR 191465, where I submitted a patch to get -fno-omit-frame-pointer to get injected into the compile (as something makes it disappear from the toplevel Makefile.) But, what was committed wasn't my patch, and the option doesn't show up in my build log of libexecinfo. At home I'm stuck with a working version 35, after I had forced a recompile of all its dependencies with gcc48 (ahead of it becoming default....) Still segfaulted. Then I looked at libexecinfo, and forcing -fno-omit-frame-pointer in ...and suddenly chrome started. So I submitted the PR of what I had done. But, then I got busy, so only work machine was getting updates....and it got up to 37 before it stopped. And, then what used to work is now also segfaulting. But, noticed on the weekend that libexecinfo versions didn't match between work and home. Have been trying to get it built through poudriere at home....along with other patches I've been working on. But, then decided to try a clean poudriere setup on work machine.... in watching it I noticed its libexecinfo build didn't match the build at home (work uses just portsnap, home goes through portshaker to merge in the various things I'm working on....aka https://github.com/TheDreamer/beastie-work...though I haven't push commits up in a while....) Though getting it back in, doesn't seem to have helped...yet. Probably need to backout some of the compiler overrides on chromium? On 11/03/14 12:39, Tomek wrote: > Hi, > > 2 listopada 2014 14:06 - "René Ladan" <rene@freebsd.org>:: >> On 02-11-2014 11:05, Lena@lena.kiev.ua wrote: >> >>> 8.4 i386, one more newer version 38.0.2125.111 from >>> pkg.FreeBSD.org/freebsd:8:x86:32 - at start >>> this time a window frame didn't appear, again segfault: >>> >>> (deleted ~/.config/chromium, ~/.cache/chromium, ~/.pki) >>> ~ $ LD_LIBRARY_PATH=/usr/local/lib/gcc48 chrome --ssl-version-min=tls1 >>> Segmentation fault (core dumped) >>> ~ $ gdb /usr/local/share/chromium/chrome chrome.core >>> ... >>> (gdb) bt >>> #0 0x2dde2f37 in getframeaddr () from /usr/local/lib/libexecinfo.so.1 >>> #1 0x2dddeb21 in backtrace () from /usr/local/lib/libexecinfo.so.1 > > The problem lies in backtrace from execinfo - in stack_trace_posix.cc line 751: > count_ = base::saturated_cast<size_t>(backtrace(trace_, arraysize(trace_))); > > The solution is either comment it out and return 0 or add -fno-omit-frame-pointer > in CFLAGS in port Makefile. > > I've tested it on FreeBSD 8.4 in VirtualBox. Can anybody test it also? > > This is not really a chromium problem - it is broken execinfo... > > -- > Best Wishes > Tomek > _______________________________________________ > freebsd-chromium@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-chromium > To unsubscribe, send any mail to "freebsd-chromium-unsubscribe@freebsd.org" > -- Email: beastie@tardisi.com Name: Lawrence "The Dreamer" Chen GitHub: https://github.com/TheDreamer Pages: http://beastie.tardisi.com Blog: http://lawrencechen.net WWW: http://tardisi.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?545820B3.2050508>