Suspend/resume issue in LMDE (again)

[mark1mint]

Sorry for bringing this up again,I know this is a well known issue (well,more like an ordeal considering how many people have this issue and the number of fixes one has to try ),anyhow I find myself amongst the unlucky ones who can't have suspend and resume work in LMDE:I have a Sapphire HD 5450 graphic card with proprietary driver from the repo,Asus P8H61-M LE motherboard,package uswsusp is installed:I've just "fixed" the swap issue following this tutorial http://forums.linuxmint.com/viewtopic.php?f=199&t=64071&p=581088&hilit=suspend+resume#p369793, because I actually had the UUID of the swap partition in /etc/initramfs-tools/conf.d/resume not matching the actual UUID of the swap on the disk,so I've edited this line,then during the update of initrams this warning came up cryptsetup: WARNING: failed to detect canonical device of /dev/sdb10 ,however the swap issue as far as the UUID identifier goes is now apparently fixed.
But,suspend doesn't work at all:the computer just shuts down,there's no blinking of the power led as expected,no sign of activity at all,then after pressing the power button (to resume) it first attempts to spin up the disk a couple of times ,then finally wakes up to a blank screen.
This is quite a lot like (if not the exact same thing) what Ubuntu Lucid did on this dual-boot computer with the standard kernel,then unexpectedly the upgrade to the 3.0.0-20-generic kernel from the Ubuntu Kernel Team fixed it for good,now I can suspend and resume Ubuntu at will with no issues,I would have expected LMDE with a newer kernel to work straight out of the box,which it doesn't.
What now ? What should I try to fix this (very annoying,BTW) suspend/resume issue ?
What kind of additional info can I post to troubleshoot this issue ?

[theorizer]

I confirm this issue. I have the same problem. My video card is however an NVIDIA 8400GS.
On update-initramfs -u I get the warning: "cryptsetup: WARNING: failed to detect canonical device of /dev/sda2", which is my swap partition.
The swap partition in file /etc/initramfs-tools/conf.d/resume has been set properly.

[mark1mint]

So what should we try next ? Is the warning message irrelevant,given that the swap partition UUID has been actually corrected ? Maybe the uswsusp package has to be removed,as I've noticed that in Ubuntu (where suspend actually works for me with the 3.0.0-20-generic kernel) it isn't installed?

[theorizer]

So after some googling I found a solution to eliminate the cryptsetup warning:
In /etc/uswsusp.conf set

Code: Select all

resume device = /dev/disk/by-uuid/xxxx

then

Code: Select all

update-initramfs -u

and the warning is away, but resume from suspend just doesn't want to work. By the way, does suspend use the swap-partition? Or is it only used when hibernating?

[mark1mint]

Ok,that will fix the warning message,but unfortunately won't fix the actual resume issue -frankly I do not know if suspend really uses the swap partition,I would have thought that only hibernate writes to that space but the reading of other threads suggests otherwise.
BTW,have you tried the s2ram command in a terminal ? I did try it in the past and didn't work either,it wrote to the disk as it were hibernating instead of suspending to RAM,then the computer simply shut down instead of going into sleep mode,and at the next startup there was no trace of the previously running applications.That however was before fixing the swap partition UUID ,I haven't yet tried if it works now.

[theorizer]

No, s2ram doesn't work either. My biggest problem is that with suspend everything seems to be ok, but the resume just doesn't work, nor generates the system a single letter log about the resume.

[mark1mint]

Yes,as far as I can tell,that's pretty much what happens when suspend doesn't work properly,the computer apparently goes into sleep mode but then doesn't resume or wakes up to a blank screen-and I couldn't find -so far- any useful info about the resume in the log either.
Just a quick question,I have an ATI graphic card,which one do you have ?
I was just having a look at the pm-suspend log to try to locate any useful info at least before the (failed) resume and found this line /usr/lib/pm-utils/sleep.d/99video suspend suspend: disabled,I wonder if this has something to do with the issue,or if any other line in the log containing the word "video" may be an hint in the right direction.

[mark1mint]

I apologize for the bump,folks:does anyone have any idea about why this fix hasn't worked so far ?
Would it be worth to try s2both from a terminal instead of s2ram?

I know it's a widely discussed issue in almost any distro,but this is a major annoyance,a computer that can't be put in any sleep state isn't really usable IMHO.

Edit:just came across this http://wiki.debian.org/Suspend,I'll look into the debug log and see if I can dig out anything useful.

[mark1mint]

So,digging further into this,I've enabled the debugging log for pm-utils and can confirm after forcing standby with sudo pm-suspend (with the usual resume failure) and a quick look at it that no quirks are used for the graphic card according to this line ATI Catalyst driver detected, not using quirks.
Reading the Debian wiki above,this is because the kernel has kernel mode setting (KMS) enabled by default and this doesn't allow to use quirks,so it has to be disabled if there are issues.
Referring to http://wiki.debian.org/KernelModesetting this can be done editing /etc/modprobe.d/i915-kms.conf or /etc/modprobe.d/radeon-kms.conf ,only I have no idea how,both are set to 1 on my system,should that be 0 ?
Another weirdness is that upon issuing the pm-suspend command or pressing Suspend in the Quit menu,the system totally shuts off,like it has actually been shut down,at least apparently:on Ubuntu,when suspend used to fail,the system would still go in sleep mode (only to resume to a blank screen thereafter),as the power led was flashing,which it does not in LMDE.
While on this,I've tried the pm-is-supported command and all exit codes are 0,so,as far this test goes,all suspend modes should be supported on my system.Should I also try pm-suspend-hybrid and see what happens?
Of course I've saved the logs and will keep saving them as I try,so if anyone is interested in having a look I can post/upload them as needed.

[mark1mint]

Below is the last part of the log after enabling debugging for pm-utils and trying to suspend again,does anyone see anything obvious that could help determine the issue ? I'm after the impression that failure to suspend is related somehow to my graphic card and/or the ATI driver,but I cannot be sure if that's actually the case and what countermeasures have to be deployed.

Code: Select all

Running hook /usr/lib/pm-utils/sleep.d/99video suspend suspend:
+ hook_ok /usr/lib/pm-utils/sleep.d/99video
+ local hook=99video
+ [ -f /var/run/pm-utils/pm-suspend/storage/disable_hook:99video ]
+ return 252
+ local status=252
+ log
+ is_set true
+ return 0
+ local fmt=%s\n
+ [  = -n ]
+ printf %s\n

+ log -n /usr/lib/pm-utils/sleep.d/99video suspend suspend:
+ is_set true
+ return 0
+ local fmt=%s\n
+ [ -n = -n ]
+ fmt=%s
+ shift
+ printf %s /usr/lib/pm-utils/sleep.d/99video suspend suspend:
/usr/lib/pm-utils/sleep.d/99video suspend suspend: + hook_exit_status 252
+ log disabled.
+ is_set true
+ return 0
+ local fmt=%s\n
+ [ disabled. = -n ]
+ printf %s\n disabled.
disabled.
+ LAST_HOOK=99video
+ IFS=

+ IFS= 

+ inhibited
+ [ -f /var/run/pm-utils/pm-suspend/storage/inhibit ]
+ return 0
+ date
+ log Mon Jan 28 16:33:53 CET 2013: performing suspend
+ is_set true
: