[opensuse-gnome] GS3.6 SegFaults After Update
After upgrading two boxes to the current GNOME3.6 both land at a spinning-pizza-of-death on a black background. Messages in dmesg like - [ 39.715139] eth0: no IPv6 routers present [ 60.071102] show_signal_msg: 66 callbacks suppressed [ 60.071105] gnome-shell[1618]: segfault at 30 ip 00007f79f3e37116 sp 00007fff88a3a590 error 4 in libgdk-3.so.0.400.4[7f79f3de4000+a4000] [ 63.594961] EXT4-fs (sda2): re-mounted. Opts: acl,user_xattr,commit=0 [ 63.621990] EXT4-fs (sda3): re-mounted. Opts: acl,user_xattr,commit=0 -- To unsubscribe, e-mail: opensuse-gnome+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-gnome+owner@opensuse.org
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
After upgrading two boxes to the current GNOME3.6 both land at a spinning-pizza-of-death on a black background.
Messages in dmesg like -
[ 39.715139] eth0: no IPv6 routers present [ 60.071102] show_signal_msg: 66 callbacks suppressed [ 60.071105] gnome-shell[1618]: segfault at 30 ip 00007f79f3e37116 sp 00007fff88a3a590 error 4 in libgdk-3.so.0.400.4[7f79f3de4000+a4000] [ 63.594961] EXT4-fs (sda2): re-mounted. Opts: acl,user_xattr,commit=0 [ 63.621990] EXT4-fs (sda3): re-mounted. Opts: acl,user_xattr,commit=0
That does not sound good... First, make sure you have ALL updates from GS35: zypper lu -a -r GS36 (if GS36 is the name you gave the repo) Also: can you manage to get a backtrace? An easy way is to setup your system to create coredumps in /cores for you; as starting stuff in gdb is typically running after it and not being able to reproduce. How to setup the system: in /etc/security/limits.conf add a line * soft core unlimited in /etc/sysctl.conf add a line kernel.core_pattern = /cores/%e-%u-%t.core create a folder /cores mkdir /cores chmod 777 /cores -> warning! This is NOT for production! But it can help a great deal in debugging... Later on, using gdb, you can load the coredumps (and redo a bt as often as you need, even installing debuginfo packages afterwards is still possible). With this, we should be able to track this down... (Please also create a bug report about this... to have it tracked). -- To unsubscribe, e-mail: opensuse-gnome+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-gnome+owner@opensuse.org
On Tue, 2012-11-27 at 14:47 +0100, Dominique Leuenberger a.k.a DimStar wrote:
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
After upgrading two boxes to the current GNOME3.6 both land at a spinning-pizza-of-death on a black background. Messages in dmesg like - [ 39.715139] eth0: no IPv6 routers present [ 60.071102] show_signal_msg: 66 callbacks suppressed [ 60.071105] gnome-shell[1618]: segfault at 30 ip 00007f79f3e37116 sp 00007fff88a3a590 error 4 in libgdk-3.so.0.400.4[7f79f3de4000+a4000] [ 63.594961] EXT4-fs (sda2): re-mounted. Opts: acl,user_xattr,commit=0 [ 63.621990] EXT4-fs (sda3): re-mounted. Opts: acl,user_xattr,commit=0 That does not sound good... First, make sure you have ALL updates from GS35: zypper lu -a -r GS36 (if GS36 is the name you gave the repo)
One one workstations I just had to zypper dup back to the mainline GNOME to get it working right away - so I don't know about that one. On my test box [ where I can break things :) ] the "zypper lu -a -r GS36" and a reboot *DID* resolve the problem. I suspect the repository priorities were not adjusted and somehow mainline GTK packages got mixed in with the packages from GS36. Both boxes were showing segfaults from either libgdk or libgtk.
Also: can you manage to get a backtrace? An easy way is to setup your system to create coredumps in /cores for you; as starting stuff in gdb is typically running after it and not being able to reproduce. How to setup the system: in /etc/security/limits.conf add a line * soft core unlimited in /etc/sysctl.conf add a line kernel.core_pattern = /cores/%e-%u-%t.core create a folder /cores mkdir /cores chmod 777 /cores -> warning! This is NOT for production! But it can help a great deal in debugging... Later on, using gdb, you can load the coredumps (and redo a bt as often as you need, even installing debuginfo packages afterwards is still possible).
I'll do this on my test box in any case, for future breakage reporting.
With this, we should be able to track this down... (Please also create a bug report about this... to have it tracked).
-- To unsubscribe, e-mail: opensuse-gnome+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-gnome+owner@opensuse.org
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
After upgrading two boxes to the current GNOME3.6 both land at a spinning-pizza-of-death on a black background. Messages in dmesg like - [ 39.715139] eth0: no IPv6 routers present [ 60.071102] show_signal_msg: 66 callbacks suppressed [ 60.071105] gnome-shell[1618]: segfault at 30 ip 00007f79f3e37116 sp 00007fff88a3a590 error 4 in libgdk-3.so.0.400.4[7f79f3de4000+a4000] [ 63.594961] EXT4-fs (sda2): re-mounted. Opts: acl,user_xattr,commit=0 [ 63.621990] EXT4-fs (sda3): re-mounted. Opts: acl,user_xattr,commit=0 That does not sound good... First, make sure you have ALL updates from GS35: zypper lu -a -r GS36 (if GS36 is the name you gave the repo) One one workstations I just had to zypper dup back to the mainline GNOME to get it working right away - so I don't know about that one. On my test box [ where I can break things :) ] the "zypper lu -a -r GS36" and a reboot *DID* resolve the problem. I suspect the repository
On Tue, 2012-11-27 at 14:47 +0100, Dominique Leuenberger a.k.a DimStar wrote: priorities were not adjusted and somehow mainline GTK packages got mixed in with the packages from GS36. Both boxes were showing segfaults from either libgdk or libgtk.
Also: can you manage to get a backtrace? An easy way is to setup your system to create coredumps in /cores for you; as starting stuff in gdb is typically running after it and not being able to reproduce. How to setup the system: in /etc/security/limits.conf add a line * soft core unlimited in /etc/sysctl.conf add a line kernel.core_pattern = /cores/%e-%u-%t.core create a folder /cores mkdir /cores chmod 777 /cores -> warning! This is NOT for production! But it can help a great deal in debugging... Later on, using gdb, you can load the coredumps (and redo a bt as often as you need, even installing debuginfo packages afterwards is still possible). I'll do this on my test box in any case, for future breakage reporting. With this, we should be able to track this down... (Please also create a bug report about this... to have it tracked).
Now it appears to have happened to my primary laptop; boot-up starts X and leaves it at a spinning-pizza-of-death. Cores are enabled but nothing is being dropped in "/cores". Argh!!! So I'm dupp'ing again to see if that fixes it. I had an oddity this morning. I did a zypper refresh, and then tried to install GVFS debuginfo packages. It said none matching my versions (from GS36) were available. I have the priority of the GS36 repo at 80, higher than the openSUSE repo-oss (at 99). So I thought it was odd, I changed the priority back to 99 on GS36, so they were all the same. And then the debuginfo packages were found. But now after reboot things seem out-of-whack. It seems odd. I was at GVFS 1.14 like in the GS36 repor, but then "zypper lu -a -r GS36" showed older packages installed for other stuff. Scratches head... We'll see if dup'ing fixes me. -- To unsubscribe, e-mail: opensuse-gnome+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-gnome+owner@opensuse.org
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
After upgrading two boxes to the current GNOME3.6 both land at a spinning-pizza-of-death on a black background. Messages in dmesg like - [ 39.715139] eth0: no IPv6 routers present [ 60.071102] show_signal_msg: 66 callbacks suppressed [ 60.071105] gnome-shell[1618]: segfault at 30 ip 00007f79f3e37116 sp 00007fff88a3a590 error 4 in libgdk-3.so.0.400.4[7f79f3de4000+a4000] [ 63.594961] EXT4-fs (sda2): re-mounted. Opts: acl,user_xattr,commit=0 [ 63.621990] EXT4-fs (sda3): re-mounted. Opts: acl,user_xattr,commit=0 That does not sound good... First, make sure you have ALL updates from GS35: zypper lu -a -r GS36 (if GS36 is the name you gave the repo) One one workstations I just had to zypper dup back to the mainline GNOME to get it working right away - so I don't know about that one. On my test box [ where I can break things :) ] the "zypper lu -a -r GS36" and a reboot *DID* resolve the problem. I suspect the repository
Also: can you manage to get a backtrace? An easy way is to setup your system to create coredumps in /cores for you; as starting stuff in gdb is typically running after it and not being able to reproduce. How to setup the system: in /etc/security/limits.conf add a line * soft core unlimited in /etc/sysctl.conf add a line kernel.core_pattern = /cores/%e-%u-%t.core create a folder /cores mkdir /cores chmod 777 /cores -> warning! This is NOT for production! But it can help a great deal in debugging... Later on, using gdb, you can load the coredumps (and redo a bt as often as you need, even installing debuginfo packages afterwards is still possible). I'll do this on my test box in any case, for future breakage reporting. With this, we should be able to track this down... (Please also create a bug report about this... to have it tracked). Now it appears to have happened to my primary laptop; boot-up starts X and leaves it at a spinning-pizza-of-death. Cores are enabled but nothing is being dropped in "/cores". Argh!!! So I'm dupp'ing again to see if that fixes it. I had an oddity this morning. I did a zypper refresh, and then tried to install GVFS debuginfo
On Tue, 2012-11-27 at 14:47 +0100, Dominique Leuenberger a.k.a DimStar wrote: priorities were not adjusted and somehow mainline GTK packages got mixed in with the packages from GS36. Both boxes were showing segfaults from either libgdk or libgtk. packages. It said none matching my versions (from GS36) were available. I have the priority of the GS36 repo at 80, higher than the openSUSE repo-oss (at 99). So I thought it was odd, I changed the priority back to 99 on GS36, so they were all the same. And then the debuginfo packages were found. But now after reboot things seem out-of-whack. It seems odd. I was at GVFS 1.14 like in the GS36 repor, but then "zypper lu -a -r GS36" showed older packages installed for other stuff. Scratches head... We'll see if dup'ing fixes me.
Yes, the "zypper dup --from GS36" restored the laptop to working. Scratches head again (I might be balding). -- To unsubscribe, e-mail: opensuse-gnome+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-gnome+owner@opensuse.org
On Fri, 2012-12-14 at 14:54 -0500, Adam Tauno Williams wrote:
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
Quoting Adam Tauno Williams <awilliam@whitemice.org>:
After upgrading two boxes to the current GNOME3.6 both land at a spinning-pizza-of-death on a black background. Messages in dmesg like - [ 39.715139] eth0: no IPv6 routers present [ 60.071102] show_signal_msg: 66 callbacks suppressed [ 60.071105] gnome-shell[1618]: segfault at 30 ip 00007f79f3e37116 sp 00007fff88a3a590 error 4 in libgdk-3.so.0.400.4[7f79f3de4000+a4000] [ 63.594961] EXT4-fs (sda2): re-mounted. Opts: acl,user_xattr,commit=0 [ 63.621990] EXT4-fs (sda3): re-mounted. Opts: acl,user_xattr,commit=0 That does not sound good... First, make sure you have ALL updates from GS35: zypper lu -a -r GS36 (if GS36 is the name you gave the repo) One one workstations I just had to zypper dup back to the mainline GNOME to get it working right away - so I don't know about that one. On my test box [ where I can break things :) ] the "zypper lu -a -r GS36" and a reboot *DID* resolve the problem. I suspect the repository
Also: can you manage to get a backtrace? An easy way is to setup your system to create coredumps in /cores for you; as starting stuff in gdb is typically running after it and not being able to reproduce. How to setup the system: in /etc/security/limits.conf add a line * soft core unlimited in /etc/sysctl.conf add a line kernel.core_pattern = /cores/%e-%u-%t.core create a folder /cores mkdir /cores chmod 777 /cores -> warning! This is NOT for production! But it can help a great deal in debugging... Later on, using gdb, you can load the coredumps (and redo a bt as often as you need, even installing debuginfo packages afterwards is still possible). I'll do this on my test box in any case, for future breakage reporting. With this, we should be able to track this down... (Please also create a bug report about this... to have it tracked). Now it appears to have happened to my primary laptop; boot-up starts X and leaves it at a spinning-pizza-of-death. Cores are enabled but nothing is being dropped in "/cores". Argh!!! So I'm dupp'ing again to see if that fixes it. I had an oddity this morning. I did a zypper refresh, and then tried to install GVFS debuginfo
On Tue, 2012-11-27 at 14:47 +0100, Dominique Leuenberger a.k.a DimStar wrote: priorities were not adjusted and somehow mainline GTK packages got mixed in with the packages from GS36. Both boxes were showing segfaults from either libgdk or libgtk. packages. It said none matching my versions (from GS36) were available. I have the priority of the GS36 repo at 80, higher than the openSUSE repo-oss (at 99). So I thought it was odd, I changed the priority back to 99 on GS36, so they were all the same. And then the debuginfo packages were found. But now after reboot things seem out-of-whack. It seems odd. I was at GVFS 1.14 like in the GS36 repor, but then "zypper lu -a -r GS36" showed older packages installed for other stuff. Scratches head... We'll see if dup'ing fixes me. Yes, the "zypper dup --from GS36" restored the laptop to working. Scratches head again (I might be balding).
FYI, GNOME 3.6 has been stable on my laptop and both workstations ever since this issue. And I did a fresh from DVD install and subsequent dup to GNOME 3.6 on a test laptop yesterday [2013-02-25]... and it is working perfectly. -- Adam Tauno Williams GPG D95ED383 Systems Administrator, Python Developer, LPI / NCLA -- To unsubscribe, e-mail: opensuse-gnome+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-gnome+owner@opensuse.org
participants (2)
-
Adam Tauno Williams -
Dominique Leuenberger a.k.a DimStar