From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Korsgaard Date: Thu, 05 May 2016 20:40:42 +0200 Subject: [Buildroot] Xorg crash In-Reply-To: <572B84E6.1000503@zacarias.com.ar> (Gustavo Zacarias's message of "Thu, 5 May 2016 14:37:42 -0300") References: <1274ad69-0abb-26b0-b623-67de78c0274f@brickedbrain.com> <572B59A9.2070602@zacarias.com.ar> <20160505193148.721ba349@free-electrons.com> <572B84E6.1000503@zacarias.com.ar> Message-ID: <87vb2smjpx.fsf@dell.be.48ers.dk> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net >>>>> "Gustavo" == Gustavo Zacarias writes: > On 05/05/16 14:31, Thomas Petazzoni wrote: >> Hello, >> >> On Thu, 5 May 2016 11:33:13 -0300, Gustavo Zacarias wrote: >> >>> Try using BR2_OPTIMIZE_2=y and rebuild everything - the X server is >>> notoriously fragile when optimizing for size with certain toolchain >>> combinations/architectures. >> >> Should we: >> >> (1) Switch to -O2 as the default optimization level ? >> >> (2) Force -O2 for the X.org server ? >> >> Thomas > Hi. > From the little testing i've done it seems that switching the xorg > server alone to -O2 is enough. > I'm not a fan of "dark magic" scenarios, however it seems to be a > recurring problem, at least for ARM & MIPS, doesn't seem to be the > case for x86_64. > I could make a couple test configs to see on which combinations it's > failing by extending the Qemu ones, but it's probably more hassle than > anything since there seems to be a wide spectrum of gcc versions > affected. > Going for -O2 might be the quickest solution, at least for xorg-server > itself. Yes, I agree. Simply forcing -O2 for xorg-server (with a sensible comment explaining why) is good enough. People including Xorg are most likely not as tight on space that -Os / -O2 matters much. -- Venlig hilsen, Peter Korsgaard