Date: Tue, 26 Feb 2019 10:53:26 -0800 From: bob prohaska <fbsd@www.zefox.net> To: freebsd-arm@freebsd.org Subject: Re: Clang segfault on rpi3 workaround using -O2 Message-ID: <20190226185326.GA58325@www.zefox.net> In-Reply-To: <20190219180757.GA13504@www.zefox.net> References: <20190219180757.GA13504@www.zefox.net>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, Feb 19, 2019 at 10:07:57AM -0800, bob prohaska wrote: > The segfaults in clang during buildworld on rpi3 persisted past > r339376, but a hint from > https://github.com/android-ndk/ndk/issues/386 > suggested trying CFLAGS=-O2 as a long-shot workaround. > > To my surprise, it worked. Three passes were required to > clear the segfaults, but now a -j4 buildworld/buildkernel > cycle runs without error to completion. A test of www/chromium > is underway now, and about half finished with no errors yet, > using r344113. > > The github report is more than a year old and notes that > clang was fixed around that time. Is it possible the fix > didn't find its way into FreeBSD? > The CFLAGS=-O2 workaround also seems to (mostly) work building www/chromium. There was only one segfault compiling chromium, restarting make permitted a successful installation. Previously make segfaulted several times per day, seemingly independent of swap usage, so it doesn't seem related to VM pressure. Unfortunately that version of chromium had a (now corrected) libray bug so it wouldn't run. www/chromium is now up to date and make is trying again. There's a bit more detail at http://www.zefox.net/~fbsd/rpi3/clang_trouble/r344113/readme in case anybody's interested. Thanks for reading, bob prohaska
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20190226185326.GA58325>