Solved Chromium 43.0.2357.65 crashing in settings

dcbdbis

Active Member

Reaction score: 5
Messages: 172

Good Afternoon all,

FreeBSD 10 Stable. x64.

System and ports built from source. I am aware of the risks of running stable and choose to accept those risks.

Late last week Chromium was updated to the version listed in the title. Since that upgrade, Chromium crashes when attempting to access settings. Previous version worked perfectly. Chromium has been built with it's default settings.

I have gone into .config and have deleted the chromium folder for a clean start. No success, issue remains.

I checked in the bugs section, and no bugs have been filed against this version of Chromium.

Does anyone know of a workaround for this issue?


Sincerely and respectfully,

Dave
 

DutchDaemon

Administrator
Staff member
Administrator
Moderator
Developer

Reaction score: 2,841
Messages: 11,319

I have no actual solution, I will just ack you observation. FreeBSD 10.1-STABLE amd 64, latest Chrome. Crashes in settings and on some media-heavy sites like Google Plus.
 

protocelt

Daemon

Reaction score: 411
Messages: 1,253

Not that this helps any, but this isn't exclusive to FreeBSD. Chromium crashes for me running on Linux as well in the same manner.
 

DutchDaemon

Administrator
Staff member
Administrator
Moderator
Developer

Reaction score: 2,841
Messages: 11,319

Latest console errors were:
Code:
[97123:356541440:0531/201037:ERROR:gles2_cmd_decoder.cc(11539)] [GroupMarkerNotSet(crbug.com/242999)!:A860691508000000]GL ERROR :GL_INVALID_OPERATION : glTexStorage2DEXT: <- error from previous GL command
[97123:356541440:0531/201042:ERROR:gles2_cmd_decoder.cc(11539)] [GroupMarkerNotSet(crbug.com/242999)!:A830FE1C08000000]GL ERROR :GL_INVALID_OPERATION : glTexStorage2DEXT: <- error from previous GL command
[97123:356541440:0531/201110:ERROR:gles2_cmd_decoder.cc(11539)] [GroupMarkerNotSet(crbug.com/242999)!:A84E691508000000]GL ERROR :GL_INVALID_OPERATION : glTexStorage2DEXT: <- error from previous GL command
Segmentation fault
 

palik

New Member

Reaction score: 1
Messages: 13

Hi,
on my Laptop - FreeBSD 11.0 amd 64 - chrome crashes by visiting gmail- or google+ site on Segmentation fault after update to 43.0.2357.65 too.

Any help or workaround greatly appreciated.

BR,
palik
 

protocelt

Daemon

Reaction score: 411
Messages: 1,253

Hi,
on my Laptop - FreeBSD 11.0 amd 64 - chrome crashes by visiting gmail- or google+ site on Segmentation fault after update to 43.0.2357.65 too.

Any help or workaround greatly appreciated.

BR,
palik
I don't know of a work around but there is already a PR 200510 filed for this particular issue.

I also don't know if the segmentation fault when accessing the settings page is related but since I have some spare time, I'll rebuild www/chromium with debugging enabled to get some more useful information and file a PR separately for that issue.

Edit: I rebuilt www/chromium with debugging on successfully but unfortunately it faulted immediately when running it with gdb(1) so there is no information what so ever to use beyond what's known already. :(
 
OP
OP
dcbdbis

dcbdbis

Active Member

Reaction score: 5
Messages: 172

Good Afternoon All,


And thank you for the replies and confirmation of the issue.

I appreciate the update from FreshPorts, and I installed it. Unfortunately, the issue remains.

I guess we'll all have to wait for a patch from upstream.

Thanks again for the replies,

Sincerely and respectfully,

Dave
 
OP
OP
dcbdbis

dcbdbis

Active Member

Reaction score: 5
Messages: 172

Thank you very much.

We'll wait for upstream to fix.

Sincerely and respectfully,

Dave
 

tingo

Daemon

Reaction score: 433
Messages: 2,113

Are we sure that this is a general Chromium (ie. upstream bug)?
 

LeFroid

Member

Reaction score: 1
Messages: 40

Hello,

I patched chromium for the 42 -> 43 updates and can confirm the crashes on chrome://settings , gmail.com and the other crashes on misc. websites.
Since gdb isn't useful with debug builds, I've been looking through the commits between chromium 42.0.2311.153 and 43.0.2357.81 to see if anything relevant pops up. If anyone has the time and willingness to help, please take a look at the following link: https://chromium.googlesource.com/chromium/src/+log/42.0.2311.153..43.0.2357.81?pretty=fuller&n=10000
 

Samy Mahmoudi

New Member


Messages: 4

Hello all,

Same issue on FreeBSD 10.1-RELEASE since an execution of:

Code:
#!/bin/sh

zfs destroy zfsroot@stable
zfs destroy zfsroot/usr@stable
zfs destroy zfsroot/usr/ports@stable

zfs rename zfsroot@testing stable
zfs rename zfsroot/usr@testing stable
zfs rename zfsroot/usr/ports@testing stable

pkg upgrade &&
portsnap fetch update &&
zfs snapshot zfsroot@testing &&
zfs snapshot zfsroot/usr@testing &&
zfs snapshot zfsroot/usr/ports@testing
Two different systems (amd64) were affected, so rolled back to the "stable" state...

Just out of curiosity, is there a filesystem-independant way of upgrading your packages with the possibility to set the previous packages back (e.g. by installing an outdated version of a package)?

Samy

P.S. Possible workaround for palik: if you use ZFS, then always snapshot your filesystems before upgrading so you could possibly rollback to the previous state.
 

protocelt

Daemon

Reaction score: 411
Messages: 1,253

Just out of curiosity, is there a filesystem-independant way of upgrading your packages with the possibility to set the previous packages back (e.g. by installing an outdated version of a package)?
Not sure off hand if pkg(8) has that option itself however the default behavior of portmaster(8) is to create a backup of the installed package in /var/cache/pkg when updating it unless you tell it otherwise. You can just use pkg(8) to delete the updated package and point it to the older package to re-install it. Keep in mind dependencies need to be the same between the packages or you may run into problems. Rolling back to a working file system snapshot is probably more full proof.
 

kpa

Beastie's Twin

Reaction score: 1,810
Messages: 6,318

"Cherry picking" updates or reverts back to old packages is absolutely not supported, this has been repeated many times on the ports mailing list. The only way you can hope to revert back to a working set of ports/packages is to use the snapshot/rollback functionality of the filesystem used, zfs snapshot for ZFS and for UFS I'm not sure how to actually do a rollback using an UFS snapshot.
 

Samy Mahmoudi

New Member


Messages: 4

Also note that some chrome processes stay alive after the segmentation fault, resulting in very high CPU consumption.
 

daeron

Member

Reaction score: 2
Messages: 48

Also happening on chromium-43.0.2357.81 on new build of FreeBSD 10.1 stable AMD64
Happy to open the History, Extension, and About pages but the "Settings" page causes a Segmentation fault.
Haven't found any other pages causing it yet.
 

mrpickles

New Member


Messages: 1

Is there another way to modify the settings. I looked for a master_preferences file but was unable to find it. I was under the impression one was created the first time chromium was launched
 

Paul Dokas

New Member


Messages: 1

I've also been experiencing this issue, however my experience is slightly different. I have a fairly complex setup with lots of extensions and many tabs that open at startup and chrome crashes shortly after the tabs start loading. Settings is not one of these tabs. If I move my chrome related configurations out of the way, then I get the same behavior that others are seeing where it crashes when attempting to enter settings.

Looking at www/chromium in the repo, I notice that there are a bunch of patches that were added on May 25. This is coincidentally around the same time that these crashes started. This combined with the fact that it's only FreeBSD people who are seeing this behavior makes me suspect that these patches are somehow the cause of these crashes.
 

drhowarddrfine

Son of Beastie

Reaction score: 1,289
Messages: 3,352

There was a similar issue some time back and the maintainers were saying Chromium has gotten so complicated that it's hard for them to find bugs or avoid them. If this is a similar instance, it might take about three weeks for a fix to come out.
 
Top