[SOLVED] Regression: Suspend fails with 4.4.115 kernel in 14.2?

upnort · 02-07-2018, 08:04 PM

I updated Slackware 14.2 64-bit on my Thinkpad T400. I no longer can suspend the system.

Looking into the logs indicates the regression is at line 301 in /usr/lib64/pm-utils/pm-functions:

Code:

echo mem > /sys/power/state

Trying to execute that same command in a terminal results in:

Code:

echo: write error: Resource temporarily unavailable

Permissions of /sys/power/state are 644 with the contents of freeze mem disk.

Suspend works fine when booting with 4.4.111, but that has a regression with the e1000e kernel module.

Anybody else seeing this regression?

Edit: Rebuilding acpid and pm-utils against the new kernel and gcc did not help. Compiling acpid 2.0.28 did not help.

Edit: Suspend works with the 4.4.113 and 4.4.114 kernels. Looks like the regression appeared in 4.4.115. I compiled the 113 and 114 kernels with the new gcc.

Edit: The bug affects all three of my Intel systems: 1) Quad core Intel Core i5-6400, 2) Intel Celeron 1037U, 3) T400 Dual core Intel Core2 Duo P8400

magicm · 02-08-2018, 09:16 AM

I am not seeing this regression. Suspend/Resume seems to work fine for me:

System: Host: igloo.magicm.net Kernel: 4.4.115 x86_64 bits: 64 Desktop: Xfce 4.12.3 Distro: Slackware 14.2
Machine: Device: laptop System: LENOVO product: 4170CTO v: ThinkPad T420s serial: xxxxxx
Mobo: LENOVO model: 4170CTO serial: xxxxxxx
UEFI [Legacy]: LENOVO v: 8CET57WW (1.37 ) date: 04/11/2013

Did no custom rebuilding (unless you count regenerating initrd)

Lysander666 · 02-08-2018, 04:23 PM

Suspend/resume work OK for me as well on 32bit Slack.

Code:

System:    Host: lysurfer_viii.netstark Kernel: 4.4.115-smp i686 bits: 32
           Desktop: LXDE (Openbox 3.6.1) Distro: Slackware 14.2
Machine:   Device: laptop System: SAMSUNG product: NC10/N110 serial: N/A
           Mobo: SAMSUNG model: NC10/N110 serial: N/A
           BIOS: Phoenix v: 07D0.M012.20090908.RHU date: 09/08/2009
CPU:       Single core Intel Atom N270 (-MT-) cache: 512 KB
           clock speeds: max: 1600 MHz 1: 1600 MHz 2: 1067 MHz
Info:      Processes: 162 Uptime: 1:38 Memory: 506.9/2006.3MB
           Client: Shell (bash) inxi: 2.3.53

upnort · 02-08-2018, 05:22 PM

I updated my original post with CPU information. Circumstantially seems like I have configured something in common with all three Intel systems that 4.4.115 does not like. Still looking.

Open to any ideas!

upnort · 02-08-2018, 09:41 PM

Well! This is a good one!

The culprit is /etc/rc.d/rc.alsa-oss.

I had the file chmod +x. Been that way, oh, years.

After disabling the script and rebooting, suspend again works just like previously.

I can suspend multiple times, manually execute rc.alsa-oss, and boom that blocks the suspend.

Can anybody else confirm this odd behavior?

I believe the problem child is the snd_seq_oss module. I can manually rmmod the other two modules loaded by rc.alsa-oss. I am unable to remove the snd_seq_oss module. I suspect that is why the suspend times out and fails.

Before I tag this thread as solved, I will leave open for a few days with the hope that somebody has an explanation.