Re: [gentoo-user] Latest chromium-40 on ~x86
Hi, On Sun, 25 Jan 2015 19:43:12 +0100 Nils Holland wrote: On Sun, Jan 25, 2015 at 02:03:48PM +0300, Andrew Savchenko wrote: I gave up on chromium starting from chromium-36, where they dropped pre-SSE2 x86 support (and I use such system: Athlon-XP). I tried to re-add this stuff with partial success (works, but still SIGILLs sometimes) and it's very hard to clean all pieces. Looks like they're slowly abandoning x86 and older hardware at all. Actually, I can't say that I'm too much of a fan of chromium either. I'm more than happy using vimb, sometimes also midori and firefox. Chromium mostly only sits here as a last resort when some site doesn't seem to work right in one of the other browsers (which, fortunatly, only happens with a frequency that is rapidly approaching never). There are many things I don't like in Chromium, but there is one feature so important, that I can overlook all disadvantages of chromium: this feature is security. Unlike other browsers (I don't consider chrome or chromium forks here as a separate browsers) chromium is secure by design: it isolates tabs and plugins, supports various namespaces, seccomp sandboxing, yama framefork. Other browsers don't: tabs are not isolated, plugins have poor isolation (it seems firefox is working in this direction at least), thogh they work fine on my yama-enabled systems. So the fact that I don't care about chromium too much, with the added fact that my main machine is a lower-range laptop and takes quite a while to build chromium, is also the reason why I can't be bothered right now to re-build in order to try out various things or otherwise try to collect debugging information. distcc will help you here, that is the way how I maintain older boxes. Of course, you should use ccache too. Best regards, Andrew Savchenko pgpS22Eu7owem.pgp Description: PGP signature
Re: [gentoo-user] Latest chromium-40 on ~x86
On Sat, 24 Jan 2015 16:43:41 +0100 Nils Holland wrote: Hi folks, I've been using chromium successfully on my ~x86 system for quite a long time, but starting with the last two updates that came in during the last few days (namely, chromium-40.0.2214.85 and chromium-40.0.2214.91), I started having problems. Both of these versions build just fine, but upon trying to launch them, the browser's interface comes up just fine, but will only display a Something went wrong... page. I can try typing in and accessing URLs, but all I will ever get is this error page. That's not all, though, I also get to see error messages, namely the following in my terminal: ../../sandbox/linux/seccomp-bpf-helpers/sigsys_handlers.cc:**CRASHING**:seccomp-bpf failure in syscall 0265 And this here in dmesg: chrome[5274]: segfault at e806109 ip b5c5c945 sp ac280980 error 6 in chrome[b1864000+5eed000] Great, I thought, something wrong with the sandbox stuff. So I tried to launch chromium without it (chromium --no-sandbox), and indeed: The browser works absolutely fine this way - I get none of the problems or messages mentioned above. Of course, I tried to find a related entry in both the Gentoo as well as the chromium bug trackers, but I couldn't find anything in either. I'm a bit reluctand to report my own bug as I wouldn't be 100% sure that I'm not causing the problem (after all, I'm building my chromium with USE=custom-cflags, which is not officially supported, but has always produced nicely working builds for me in the past), so I thought I'd ask here first if I'm the only one observing this behavior. The question, thus, would probably be: Anyone using one of the recent chromium-40 versions on ~x86 or anywhere else and seeing something similar? Or probably someone who has experienced something like that before and could offer a guess what might be wrong here - a real bug, custom-cflags, or something entirely different? I gave up on chromium starting from chromium-36, where they dropped pre-SSE2 x86 support (and I use such system: Athlon-XP). I tried to re-add this stuff with partial success (works, but still SIGILLs sometimes) and it's very hard to clean all pieces. Looks like they're slowly abandoning x86 and older hardware at all. Best regards, Andrew Savchenko pgpkjjYmvwwIi.pgp Description: PGP signature
Re: [gentoo-user] Latest chromium-40 on ~x86
On 01/24/2015 07:44 PM, Peter Humphrey wrote: On Saturday 24 January 2015 16:43:41 Nils Holland wrote: I've been using chromium successfully on my ~x86 system for quite a long time, but starting with the last two updates that came in during the last few days (namely, chromium-40.0.2214.85 and chromium-40.0.2214.91), I started having problems. ---8 The question, thus, would probably be: Anyone using one of the recent chromium-40 versions on ~x86 or anywhere else and seeing something similar? Or probably someone who has experienced something like that before and could offer a guess what might be wrong here - a real bug, custom-cflags, or something entirely different? This is and amd64 box, not ~x86, but chromium-40.0.2214.91 is working fine here. It's not been running more than a few hours since today's upgrade, but at least it does run. Also amd64: I observed another issue with chromium-40.0.2214.91 -- it breaks X11 with compositing wm (i'm running compiz) when I activate any simple effect, e.g switch to another viewpoint or something like that. At now catched that with Nvidia; have not tried with i915 (yet). The X11 `breakage' freezes the screen, so I need to kill/restart X11 to proceed. The syslog writes lots of: Jan 24 20:30:41 gentoo kernel: [1829180.480065] NVRM: Xid (PCI::05:00): 13, Graphics Exception: ChID 0003, Class 8297, Offset 17b4, Data 0001 Jan 24 20:30:46 gentoo kernel: [1829185.546349] NVRM: Xid (PCI::05:00): 13, Graphics Exception: ChID 0003, Class 8297, Offset 17b4, Data 0001 The Xorg.0.log also shows an error (in nvidia driver): (EE) [mi] EQ overflow continuing. 200 events have been dropped. (EE) (EE) Backtrace: (EE) 0: /usr/bin/X (xorg_backtrace+0x48) [0x580338] (EE) 1: /usr/bin/X (QueuePointerEvents+0x52) [0x44c432] (EE) 2: /usr/lib64/xorg/modules/input/evdev_drv.so (0x7fcbb28c2000+0x571d) [0x7fcbb28c771d] (EE) 3: /usr/bin/X (0x40+0x71eb8) [0x471eb8] (EE) 4: /usr/bin/X (0x40+0x99dfa) [0x499dfa] (EE) 5: /lib64/libpthread.so.0 (0x7fcbb9eca000+0xfd30) [0x7fcbb9ed9d30] (EE) 6: /lib64/libc.so.6 (ioctl+0x7) [0x7fcbb8c1e237] (EE) 7: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x116bec) [0x7fcbb3900bec] (EE) 8: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x116ca7) [0x7fcbb3900ca7] (EE) 9: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x1196e9) [0x7fcbb39036e9] (EE) 10: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x12e181) [0x7fcbb3918181] (EE) 11: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x13af8d) [0x7fcbb3924f8d] (EE) 12: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x12c2a2) [0x7fcbb39162a2] (EE) 13: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x54b567) [0x7fcbb3d35567] (EE) 14: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x54bf30) [0x7fcbb3d35f30] (EE) 15: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x54e2f8) [0x7fcbb3d382f8] (EE) 16: /usr/lib64/xorg/modules/drivers/nvidia_drv.so (0x7fcbb37ea000+0x55ccc5) [0x7fcbb3d46cc5] (EE) 17: /usr/bin/X (0x40+0x357fe) [0x4357fe] (EE) 18: /usr/bin/X (0x40+0x395da) [0x4395da] (EE) 19: /lib64/libc.so.6 (__libc_start_main+0xf5) [0x7fcbb8b62aa5] (EE) 20: /usr/bin/X (0x40+0x24d7e) [0x424d7e] (EE) [1829865.192] [mi] Increasing EQ size to 1024 to prevent dropped events. [1829865.193] [mi] EQ processing has resumed after 239 dropped events. [1829865.212] [mi] This may be caused my a misbehaving driver monopolizing the server's resources.
Re: [gentoo-user] Latest chromium-40 on ~x86
On Sun, Jan 25, 2015 at 02:03:48PM +0300, Andrew Savchenko wrote: I gave up on chromium starting from chromium-36, where they dropped pre-SSE2 x86 support (and I use such system: Athlon-XP). I tried to re-add this stuff with partial success (works, but still SIGILLs sometimes) and it's very hard to clean all pieces. Looks like they're slowly abandoning x86 and older hardware at all. Actually, I can't say that I'm too much of a fan of chromium either. I'm more than happy using vimb, sometimes also midori and firefox. Chromium mostly only sits here as a last resort when some site doesn't seem to work right in one of the other browsers (which, fortunatly, only happens with a frequency that is rapidly approaching never). So the fact that I don't care about chromium too much, with the added fact that my main machine is a lower-range laptop and takes quite a while to build chromium, is also the reason why I can't be bothered right now to re-build in order to try out various things or otherwise try to collect debugging information. That's why I thought I'd ask if I'm the only one seeing this. It's entirely possible that I'll eventually decide to just unmerge chromium completely, as I haven't really had a *real* reason to use it in quite a long time. ;-) Greetings, Nils
Re: [gentoo-user] Latest chromium-40 on ~x86
Am 25.01.2015 um 11:08 schrieb victor romanchuk: On 01/24/2015 07:44 PM, Peter Humphrey wrote: On Saturday 24 January 2015 16:43:41 Nils Holland wrote: I've been using chromium successfully on my ~x86 system for quite a long time, but starting with the last two updates that came in during the last few days (namely, chromium-40.0.2214.85 and chromium-40.0.2214.91), I started having problems. ---8 The question, thus, would probably be: Anyone using one of the recent chromium-40 versions on ~x86 or anywhere else and seeing something similar? Or probably someone who has experienced something like that before and could offer a guess what might be wrong here - a real bug, custom-cflags, or something entirely different? This is and amd64 box, not ~x86, but chromium-40.0.2214.91 is working fine here. It's not been running more than a few hours since today's upgrade, but at least it does run. Also amd64: I observed another issue with chromium-40.0.2214.91 -- it breaks X11 with compositing wm (i'm running compiz) when I activate any simple effect, e.g switch to another viewpoint or something like that. At now catched that with Nvidia; have not tried with i915 (yet). works fine with KDE and desktop effects turned on. Using xorg's amd drivers. Runs on for days without problems. www-client/chromium Available versions: 40.0.2214.91 Installed versions: 40.0.2214.85^d{tbz2}(17:00:07 19.01.2015)
Re: [gentoo-user] Latest chromium-40 on ~x86
On 01/25/2015 04:29 PM, Volker Armin Hemmann wrote: Am 25.01.2015 um 11:08 schrieb victor romanchuk: On 01/24/2015 07:44 PM, Peter Humphrey wrote: On Saturday 24 January 2015 16:43:41 Nils Holland wrote: The question, thus, would probably be: Anyone using one of the recent chromium-40 versions on ~x86 or anywhere else and seeing something similar? Or probably someone who has experienced something like that before and could offer a guess what might be wrong here - a real bug, custom-cflags, or something entirely different? This is and amd64 box, not ~x86, but chromium-40.0.2214.91 is working fine here. It's not been running more than a few hours since today's upgrade, but at least it does run. Also amd64: I observed another issue with chromium-40.0.2214.91 -- it breaks X11 with compositing wm (i'm running compiz) when I activate any simple effect, e.g switch to another viewpoint or something like that. At now catched that with Nvidia; have not tried with i915 (yet). works fine with KDE and desktop effects turned on. Using xorg's amd drivers. Runs on for days without problems. Just emerged `subject' on notebook [adm64] with i915: behaves normally despite of compiz. Most likely something gets wrong with x11-drivers/nvidia-drivers-340.65 (the card is too old and incompatible with latest driver release) thanks
Re: [gentoo-user] Latest chromium-40 on ~x86
On Saturday 24 January 2015 16:43:41 Nils Holland wrote: I've been using chromium successfully on my ~x86 system for quite a long time, but starting with the last two updates that came in during the last few days (namely, chromium-40.0.2214.85 and chromium-40.0.2214.91), I started having problems. ---8 The question, thus, would probably be: Anyone using one of the recent chromium-40 versions on ~x86 or anywhere else and seeing something similar? Or probably someone who has experienced something like that before and could offer a guess what might be wrong here - a real bug, custom-cflags, or something entirely different? This is and amd64 box, not ~x86, but chromium-40.0.2214.91 is working fine here. It's not been running more than a few hours since today's upgrade, but at least it does run. -- Rgds Peter.