Bug#696909: chromium segfaults on startup on armhf
On Mon, 14 Jan 2013 20:52:00 +, peter green wrote: > Anyway I have some bad news. When I try to do an armel build with bfd on > my imx board I get. > LINK(target) out/Release/chrome > /usr/bin/ld.bfd.real: failed to set dynamic section sizes: Memory exhausted > collect2: ld returned 1 exit status This is an upstream bug on binutils 2.22 [0], not yet fixed. It would be nice to reproduce it on a debian/wheezy armel box. Jérémy [0] http://sourceware.org/bugzilla/show_bug.cgi?id=14718 -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#696909: chromium segfaults on startup on armhf
shawnland...@gmail.com wrote: It works when built with gold 2.23 Is anyone working on either getting that into wheezy (if it's a bugfix release that is likely to pass the release teams requirements) or to backport the required fixes? -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#696909: chromium segfaults on startup on armhf
It works when built with gold 2.23 on an armv6+ hardware host (bug 696284). This bug is also a dup, and was njotriced shortely after chromium 22 was made to build again on arm (ftbfs since chromium 18) due to my patches. peter green wrote: >Note: adding debian-arm to cc for feedback from other porters. > >Jonathan Nieder wrote: >> Hi Peter, >> >> peter green wrote: >> >> >>> Patch to make the package use bfd rather than gold on armel and >armhf is >>> attached. I may or may not upload this as a NMU. >>> >> >> If you'll have time to continue working on chromium:arm in the >future, >> it would probably be better to just add yourself to pkg-chromium on >> alioth >Like any other package if chromium gives us problems over in raspbian I > >will try to get those problems fixed and will contribute what I >discover >back to debian where appropriate but I don't have any real knowlege of >chromium's internals or build system nor do I have the time or >inclination to learn them. > >Anyway I have some bad news. When I try to do an armel build with bfd >on >my imx board I get. > LINK(target) out/Release/chrome >/usr/bin/ld.bfd.real: failed to set dynamic section sizes: Memory >exhausted >collect2: ld returned 1 exit status > >I have plenty of swap so i'm pretty sure the real problem is not memory > >per-se but address space. I had the same problem with the raspbian >build >(but not with the debian armhf build) and for raspbian I worked arround > >it by building with user mode qemu on an x64 host (which provides >significantly more virtual address space than any real arm system) but >i'm not sure that is acceptable for debian. > >Thoughts on what to do for chromium on armel? leave it broken? drop it >completely? upload packages built with bfd under user mode qemu >(assuming that succeeds and produces usable packages) to the archive? > > >-- >To UNSUBSCRIBE, email to debian-arm-requ...@lists.debian.org >with a subject of "unsubscribe". Trouble? Contact >listmas...@lists.debian.org >Archive: http://lists.debian.org/50f46ff0.6080...@p10link.net -- Sent from my Android phone with K-9 Mail. Please excuse my brevity.
Bug#696909: chromium segfaults on startup on armhf
Note: adding debian-arm to cc for feedback from other porters. Jonathan Nieder wrote: Hi Peter, peter green wrote: Patch to make the package use bfd rather than gold on armel and armhf is attached. I may or may not upload this as a NMU. If you'll have time to continue working on chromium:arm in the future, it would probably be better to just add yourself to pkg-chromium on alioth Like any other package if chromium gives us problems over in raspbian I will try to get those problems fixed and will contribute what I discover back to debian where appropriate but I don't have any real knowlege of chromium's internals or build system nor do I have the time or inclination to learn them. Anyway I have some bad news. When I try to do an armel build with bfd on my imx board I get. LINK(target) out/Release/chrome /usr/bin/ld.bfd.real: failed to set dynamic section sizes: Memory exhausted collect2: ld returned 1 exit status I have plenty of swap so i'm pretty sure the real problem is not memory per-se but address space. I had the same problem with the raspbian build (but not with the debian armhf build) and for raspbian I worked arround it by building with user mode qemu on an x64 host (which provides significantly more virtual address space than any real arm system) but i'm not sure that is acceptable for debian. Thoughts on what to do for chromium on armel? leave it broken? drop it completely? upload packages built with bfd under user mode qemu (assuming that succeeds and produces usable packages) to the archive? -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#696909: chromium segfaults on startup on armhf
Hi Peter, peter green wrote: > Patch to make the package use bfd rather than gold on armel and armhf is > attached. I may or may not upload this as a NMU. If you'll have time to continue working on chromium:arm in the future, it would probably be better to just add yourself to pkg-chromium on alioth. Thanks for your work, Jonathan -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#696909: chromium segfaults on startup on armhf
I can confim that using bfd instead of gold fixes the issue on armhf. armel seems to be sufering from the same issue but I have not yet tried a build with bfd on armel. I will do an armel (but on armv7 hardware, I don't have any armv4t or armv5 hardware to test on and all my armv6 stuff is running raspbian) build and test once I have finished doing the one for raspbian, unfortunately each build takes days on my imx53 (I have an odriod u2 on order which should speed things up) Patch to make the package use bfd rather than gold on armel and armhf is attached. I may or may not upload this as a NMU. diff -Nru chromium-browser-22.0.1229.94~r161065+dfsg/debian/changelog chromium-browser-22.0.1229.94~r161065+dfsg/debian/changelog --- chromium-browser-22.0.1229.94~r161065+dfsg/debian/changelog 2012-12-31 20:02:00.0 + +++ chromium-browser-22.0.1229.94~r161065+dfsg/debian/changelog 2013-01-05 12:54:20.0 + @@ -1,3 +1,10 @@ +chromium-browser (22.0.1229.94~r161065+dfsg-0.2) unstable; urgency=low + + * Change build-depends/build-conflicts to avoid gold linker on arm* as it +seems to be causing segfaults (Closes: 696909) + + -- Peter Michael Green Sat, 05 Jan 2013 12:52:12 + + chromium-browser (22.0.1229.94~r161065+dfsg-0.1) unstable; urgency=low * Non-maintainer upload. diff -Nru chromium-browser-22.0.1229.94~r161065+dfsg/debian/control chromium-browser-22.0.1229.94~r161065+dfsg/debian/control --- chromium-browser-22.0.1229.94~r161065+dfsg/debian/control 2012-10-17 23:19:01.0 + +++ chromium-browser-22.0.1229.94~r161065+dfsg/debian/control 2013-01-05 12:56:07.0 + @@ -63,7 +63,7 @@ libxt-dev, libxtst-dev, libpam0g-dev, - binutils-gold, + binutils-gold [!armel !armhf], libflac-dev, libwebp-dev, autotools-dev, @@ -80,6 +80,8 @@ subversion, libudev-dev, libssl-dev +Build-Conflicts: + binutils-gold [armel armhf] Standards-Version: 3.9.2 Package: chromium-browser
Bug#696909: chromium segfaults on startup on armhf
>I searched for the bug on google an according to a chromium engineer >is a problem related to the linker. I hope it helps. > >the problem is cause by a gold bug -> you need to use gold from >binutils 2.22 or later We already have binutils 2.22 in wheezy/sid. Still i'm tempted to try a build with bfd to see if it makes any difference (it will mean a lot more swap grinding but meh) -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
Bug#696909: chromium segfaults on startup on armhf
package: chromium severity: grave version: 22.0.1229.94~r161065-3 Chromium segfaults on startup on armhf. I got a backtrace but it isn't a very informative one :( Reading symbols from /usr/lib/chromium/chromium...Reading symbols from /usr/lib/debug/usr/lib/chromium/chromium...done. done. (gdb) run Starting program: /usr/lib/chromium/chromium [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/arm-linux-gnueabihf/libthread_db.so.1". Program received signal SIGSEGV, Segmentation fault. 0x001fad20 in ?? () (gdb) bt #0 0x001fad20 in ?? () #1 0x2bea9c8e in __libc_csu_init () #2 0xb609dcb8 in __libc_start_main () from /lib/arm-linux-gnueabihf/libc.so.6 #3 0x2a20225a in _start () (gdb) -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org