3.20-rc0: GUI dies on resume, then systemd gots sigill

3.20-rc0: GUI dies on resume, then systemd gots sigill

[Pavel Machek]

Hi!

In 3.19, suspend/resume was rock solid.

3.20-rc0, after resume, I could see X windows dying. So I tried to log
in on text console, but at that point got a note about SIGILL from
systemd (and dead system). syslog ilustrates it nicely.

Any ideas?
								Pavel

Feb 18 00:33:40 duo nm-dispatcher: Dispatching action 'down' for wlan0
Feb 18 00:33:41 duo systemd[1]: Started Network Manager Script
Dispatcher Service.
Feb 18 00:33:41 duo systemd[1]: Starting Sleep.
Feb 18 00:33:41 duo systemd[1]: Reached target Sleep.
Feb 18 00:33:41 duo systemd[1]: Starting Suspend...
Feb 18 00:33:41 duo systemd-sleep[5114]: Suspending system...
Feb 18 00:33:42 duo kernel: PM: Syncing filesystems ... done.
Feb 18 00:33:42 duo kernel: PM: Preparing system for mem sleep
Feb 18 10:35:50 duo rsyslogd: [origin software="rsyslogd"
swVersion="8.4.2" x-pid="5230" x-info="http://www.rsyslog.com"] start
Feb 18 10:35:50 duo kernel: Restarting tasks ... done.
Feb 18 10:35:50 duo kernel: video LNXVIDEO:00: Restoring backlight
state
Feb 18 10:35:50 duo kernel: sdhci-pci 0000:15:00.2: Will use DMA mode
even though HW doesn't fully claim to support it.
Feb 18 10:35:50 duo kernel: sdhci-pci 0000:15:00.2: Will use DMA mode
even though HW doesn't fully claim to support it.
Feb 18 10:35:50 duo kernel: sdhci-pci 0000:15:00.2: Will use DMA mode
even though HW doesn't fully claim to support it.
Feb 18 10:35:50 duo kernel: sdhci-pci 0000:15:00.2: Will use DMA mode
even though HW doesn't fully claim to support it.
Feb 18 10:35:47 duo nm-dispatcher: Disconnected from the system bus,
exiting.
Feb 18 10:35:47 duo mate-session[2957]: WARNING: Child process 3084
was already dead.
Feb 18 10:35:47 duo ModemManager[2409]: <warn>  Could not acquire the
'org.freedesktop.ModemManager1' service name
Feb 18 10:35:47 duo systemd[1]: NetworkManager.service: main process
exited, code=killed, status=4/ILL
Feb 18 10:35:47 duo systemd-sleep[5114]: System resumed.
Feb 18 10:35:47 duo systemd-sleep[5114]: /lib/systemd/system-sleep
terminated by signal ILL.
Feb 18 10:35:47 duo systemd[1]: Unit NetworkManager.service entered
failed state.
Feb 18 10:35:47 duo systemd[1]: accounts-daemon.service: main process
exited, code=killed, status=4/ILL
Feb 18 10:35:50 duo ModemManager[2409]: <info>  ModemManager is shut
down
Feb 18 10:35:51 duo org.a11y.Bus[2897]: g_dbus_connection_real_closed:
Remote peer vanished with error: Underlying GIOStream returned 0 bytes
on an async read (g-io-error-quark, 0). Exiting.
Feb 18 10:35:51 duo NetworkManager[5232]: <info> NetworkManager
(version 0.9.10.0) is starting...
Feb 18 10:35:51 duo NetworkManager[5232]: <info> Read config:
/etc/NetworkManager/NetworkManager.conf
Feb 18 10:35:51 duo NetworkManager[5232]: <info> WEXT support is
enabled

Feb 18 10:35:51 duo dbus[5242]: [system] Activating systemd to
hand-off: service name='org.freedesktop.login1'
unit='dbus-org.freedesktop.login1.service'
Feb 18 10:35:51 duo dbus[5242]: [system] Successfully activated
service 'org.freedesktop.systemd1'
Feb 18 10:35:51 duo systemd[1]: Started Daemon for power management.
Feb 18 10:35:51 duo systemd[1]: Started System Logging Service.
Feb 18 10:35:51 duo systemd[1]: lightdm.service: main process exited,
code=exited, status=1/FAILURE
Feb 18 10:35:51 duo systemd[1]: Unit lightdm.service entered failed
state.
Feb 18 10:35:51 duo systemd[1]: Started Load/Save RF Kill Switch
Status of rfkill3.
Feb 18 10:35:51 duo dbus[5242]: [system] Activating via systemd:
service name='org.freedesktop.PolicyKit1' unit='polkitd.service'
Feb 18 10:35:51 duo dbus[5242]: [system] Successfully activated
service 'org.freedesktop.login1'
Feb 18 10:35:51 duo systemd[1]: Started Login Service.
Feb 18 10:35:51 duo systemd[1]: Starting Authenticate and Authorize
Users to Run Privileged Tasks...
Feb 18 10:35:51 duo systemd[1]: Starting Run anacron jobs...
Feb 18 10:35:51 duo systemd[1]: Started Run anacron jobs.
Feb 18 10:35:51 duo systemd[1]: lightdm.service holdoff time over,
scheduling restart.
Feb 18 10:35:51 duo systemd[1]: Stopping Light Display Manager...
Feb 18 10:35:51 duo systemd[1]: Starting Light Display Manager...
Feb 18 10:35:51 duo anacron[5262]: Anacron 2.3 started on 2015-02-18
Feb 18 10:35:51 duo anacron[5262]: Will run job `cron.daily' in 5 min.
Feb 18 10:35:51 duo anacron[5262]: Will run job `cron.weekly' in 10
min.
Feb 18 10:35:51 duo anacron[5262]: Jobs will be executed sequentially
Feb 18 10:35:52 duo NetworkManager[5232]:
(nm-manager-auth.c:103):pk_authority_get: runtime check failed:
(authority)
Feb 18 10:35:52 duo NetworkManager[5232]: <info> init!
Feb 18 10:35:52 duo NetworkManager[5232]: <info>
update_system_hostname
Feb 18 10:35:52 duo NetworkManager[5232]: <info>
interface-parser: parsing file /etc/network/interfaces
Feb 18 10:35:52 duo systemd[1]: dbus.service: main process exited,
code=killed, status=4/ILL
Feb 18 10:35:52 duo systemd[1]: Unit dbus.service entered failed
state.
Feb 18 10:35:52 duo systemd[1]: polkitd.service: main process exited,
code=killed, status=4/ILL
Feb 18 10:35:52 duo systemd[1]: Failed to start Authenticate and
Authorize Users to Run Privileged Tasks.
Feb 18 10:35:52 duo systemd[1]: Unit polkitd.service entered failed
state.
Feb 18 10:35:52 duo systemd[1]: lightdm.service: main process exited,
code=killed, status=4/ILL
Feb 18 10:35:52 duo systemd[1]: Failed to start Light Display Manager.

...
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 3.20-rc0: GUI dies on resume, then systemd gots sigill

[Rafael J. Wysocki]

On Wednesday, February 18, 2015 10:46:32 AM Pavel Machek wrote:
> Hi!
> 
> In 3.19, suspend/resume was rock solid.
> 
> 3.20-rc0, after resume, I could see X windows dying. So I tried to log
> in on text console, but at that point got a note about SIGILL from
> systemd (and dead system). syslog ilustrates it nicely.
> 
> Any ideas?

The dying X may be a graphics driver problem.

Have you tried to suspend/resume without X?

The systemd failure after the failing resume indicates that something was
stuck somewhere or we corrupted memory.

Rafael

Re: 3.20-rc0: GUI dies on resume, then systemd gots sigill

[Pavel Machek]

Hi!

On Wed 2015-02-18 17:39:08, Rafael J. Wysocki wrote:
> On Wednesday, February 18, 2015 10:46:32 AM Pavel Machek wrote:
> > Hi!
> > 
> > In 3.19, suspend/resume was rock solid.
> > 
> > 3.20-rc0, after resume, I could see X windows dying. So I tried to log
> > in on text console, but at that point got a note about SIGILL from
> > systemd (and dead system). syslog ilustrates it nicely.
> > 
> > Any ideas?
> 
> The dying X may be a graphics driver problem.
> 
> Have you tried to suspend/resume without X?
> 
> The systemd failure after the failing resume indicates that something was
> stuck somewhere or we corrupted memory.

It looks like everyone and their dog gets SIGILL, so... yes, I'll try
without X, no, I don't think it will make difference.

For the record, 796e1c55717e9a6ff5c81b12289ffa1ffd919b6f is bad.

Does it work for you? Do you have any Intel graphics machine nearby?
									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 3.20-rc0: GUI dies on resume, then systemd gots sigill

[Pavel Machek]

On Wed 2015-02-18 20:00:52, Pavel Machek wrote:
> Hi!
> 
> On Wed 2015-02-18 17:39:08, Rafael J. Wysocki wrote:
> > On Wednesday, February 18, 2015 10:46:32 AM Pavel Machek wrote:
> > > Hi!
> > > 
> > > In 3.19, suspend/resume was rock solid.
> > > 
> > > 3.20-rc0, after resume, I could see X windows dying. So I tried to log
> > > in on text console, but at that point got a note about SIGILL from
> > > systemd (and dead system). syslog ilustrates it nicely.
> > > 
> > > Any ideas?
> > 
> > The dying X may be a graphics driver problem.
> > 
> > Have you tried to suspend/resume without X?
> > 
> > The systemd failure after the failing resume indicates that something was
> > stuck somewhere or we corrupted memory.
> 
> It looks like everyone and their dog gets SIGILL, so... yes, I'll try
> without X, no, I don't think it will make difference.
> 
> For the record, 796e1c55717e9a6ff5c81b12289ffa1ffd919b6f is bad.
> 
> Does it work for you? Do you have any Intel graphics machine nearby?

Did init 3, login as root, suspend, and yes, systemd got SIGILL.

									Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 3.20-rc0: GUI dies on resume, then systemd gots sigill

[Pavel Machek]

Hi!

> > > > In 3.19, suspend/resume was rock solid.
> > > > 
> > > > 3.20-rc0, after resume, I could see X windows dying. So I tried to log
> > > > in on text console, but at that point got a note about SIGILL from
> > > > systemd (and dead system). syslog ilustrates it nicely.
> > > > 
> > > > Any ideas?
> > > 
> > > The dying X may be a graphics driver problem.
> > > 
> > > Have you tried to suspend/resume without X?
> > > 
> > > The systemd failure after the failing resume indicates that something was
> > > stuck somewhere or we corrupted memory.
> > 
> > It looks like everyone and their dog gets SIGILL, so... yes, I'll try
> > without X, no, I don't think it will make difference.
> > 
> > For the record, 796e1c55717e9a6ff5c81b12289ffa1ffd919b6f is bad.
> > 
> > Does it work for you? Do you have any Intel graphics machine nearby?
> 
> Did init 3, login as root, suspend, and yes, systemd got SIGILL.

Tried it on desktop machine, when I press the power buton machine
power ups CD rom, but that's pretty much it. Not even power LED turns
on.

								Pavel
-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html

Re: 3.20-rc0: GUI dies on resume, then systemd gots sigill

[Rafael J. Wysocki]

On Wednesday, February 18, 2015 08:00:52 PM Pavel Machek wrote:
> Hi!
> 
> On Wed 2015-02-18 17:39:08, Rafael J. Wysocki wrote:
> > On Wednesday, February 18, 2015 10:46:32 AM Pavel Machek wrote:
> > > Hi!
> > > 
> > > In 3.19, suspend/resume was rock solid.
> > > 
> > > 3.20-rc0, after resume, I could see X windows dying. So I tried to log
> > > in on text console, but at that point got a note about SIGILL from
> > > systemd (and dead system). syslog ilustrates it nicely.
> > > 
> > > Any ideas?
> > 
> > The dying X may be a graphics driver problem.
> > 
> > Have you tried to suspend/resume without X?
> > 
> > The systemd failure after the failing resume indicates that something was
> > stuck somewhere or we corrupted memory.
> 
> It looks like everyone and their dog gets SIGILL, so... yes, I'll try
> without X, no, I don't think it will make difference.
> 
> For the record, 796e1c55717e9a6ff5c81b12289ffa1ffd919b6f is bad.
> 
> Does it work for you? Do you have any Intel graphics machine nearby?

I'm at a conference, so not really.

CCing Daniel though, he may be able to help.

Rafael

Re: 3.20-rc0: GUI dies on resume, then systemd gots sigill

[Pavel Machek]

On Wed 2015-02-18 21:38:58, Pavel Machek wrote:
> Hi!
> 
> > > > > In 3.19, suspend/resume was rock solid.
> > > > > 
> > > > > 3.20-rc0, after resume, I could see X windows dying. So I tried to log
> > > > > in on text console, but at that point got a note about SIGILL from
> > > > > systemd (and dead system). syslog ilustrates it nicely.
> > > > > 
> > > > > Any ideas?
> > > > 
> > > > The dying X may be a graphics driver problem.
> > > > 
> > > > Have you tried to suspend/resume without X?
> > > > 
> > > > The systemd failure after the failing resume indicates that something was
> > > > stuck somewhere or we corrupted memory.
> > > 
> > > It looks like everyone and their dog gets SIGILL, so... yes, I'll try
> > > without X, no, I don't think it will make difference.
> > > 
> > > For the record, 796e1c55717e9a6ff5c81b12289ffa1ffd919b6f is bad.
> > > 
> > > Does it work for you? Do you have any Intel graphics machine nearby?
> > 
> > Did init 3, login as root, suspend, and yes, systemd got SIGILL.
> 
> Tried it on desktop machine, when I press the power buton machine
> power ups CD rom, but that's pretty much it. Not even power LED turns
> on.

On x60, it was broken in 4.0-rc1, but it seems fixed now.
									Pavel

-- 
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html