Skip to contentSkip to conversation

suspend-to-ram broken on X61 thinkpads in 2.6.30-rc1?

DISCUSSION Theodore Ts'o 4 participants 13 messages Apr 10, 2009 – Apr 12, 2009

lore
Share & tools
lore
#0 lore raw

Hi, has anyone else noticed that suspend-to-ram seems to have gotten broken in 2.6.30-rc1, sometime during last half of the merge window? I've been at a conference all of this past week, so I haven't had time to bisect it yet. I was wondering if anyone else had noticed and figured out what was going wrong. Symptoms are that the suspend "moon" starts flashing and the system never manages to put itself to sleep.

- Ted

#1 lore raw
Quoted 6 lines · Theodore Ts'o

Hi, has anyone else noticed that suspend-to-ram seems to have gotten broken in 2.6.30-rc1, sometime during last half of the merge window? I've been at a conference all of this past week, so I haven't had time to bisect it yet. I was wondering if anyone else had noticed and figured out what was going wrong. Symptoms are that the suspend "moon" starts flashing and the system never manages to put itself to sleep.

There was one similar report, but that issue should be fixed in 2.6.30-rc1 if I remember correctly.

Can you test commit 0a0c5168df270a50e3518e4f12bddb31f8f5f38f
(PM: Introduce functions for suspending and resuming device interrupts), plaese?

Rafael

#2 lore raw

2009/4/10 Rafael J. Wysocki <[email protected]>:

Quoted 3 lines · Theodore Ts'o
Quoted 2 lines

Hi, has anyone else noticed that suspend-to-ram seems to have gotten broken in 2.6.30-rc1, sometime during last half of the merge window?

My X61 STR is broken as well.

Quoted 5 lines

There was one similar report, but that issue should be fixed in 2.6.30-rc1 if I remember correctly.

Can you test commit 0a0c5168df270a50e3518e4f12bddb31f8f5f38f
(PM: Introduce functions for suspending and resuming device interrupts), plaese?

This commit is in the latest git, but it's does not seem to fix the problem. 2.6.29 STR works very well. And all the Think "Fn" keys are no longer working (LID close still functioning though ... but that could a configuration issue moving from 2.6.29 to 2.6.30-rc1.

Shall I do the bisecting? I think this is the only one not working for me other than the slow iptables that I reported earlier.

Thanks,
Jeff.

#3 lore raw
Quoted 11 lines · Jeff Chua

2009/4/10 Rafael J. Wysocki <[email protected]>:

Quoted 3 lines · Theodore Ts'o
Quoted 2 lines

Hi, has anyone else noticed that suspend-to-ram seems to have gotten broken in 2.6.30-rc1, sometime during last half of the merge window?

My X61 STR is broken as well.

This commit is in the latest git, but it's does not seem to fix the problem. 2.6.29 STR works very well. And all the Think "Fn" keys are no longer working (LID close still functioning though ... but that could a configuration issue moving from 2.6.29 to 2.6.30-rc1.

I'm seeing this problem as well.

Quoted 2 lines

Shall I do the bisecting? I think this is the only one not working for me other than the slow iptables that I reported earlier.

Well, I can save you at least one or two reboots, based on some test kernels I had been building.

git bisect bad v2.6.30-rc1-git2
git bisect bad v2.6.30-rc1
git bisect bad e0724bf6 (v2.6.29-git14 circa April 6th)
git bisect good 5d80f8e5 (v2.6.29-git3 circa March 27th)

Only 3,074 revisions to go. :-)

- Ted

#4 lore raw
In reply to Jeff Chua
Quoted 10 lines · Jeff Chua

2009/4/10 Rafael J. Wysocki <[email protected]>:

Quoted 3 lines · Theodore Ts'o
Quoted 2 lines

Hi, has anyone else noticed that suspend-to-ram seems to have gotten broken in 2.6.30-rc1, sometime during last half of the merge window?

My X61 STR is broken as well.

Quoted 3 lines

There was one similar report, but that issue should be fixed in 2.6.30-rc1 if I remember correctly.

Quoted 7 lines
Quoted 2 lines

Can you test commit 0a0c5168df270a50e3518e4f12bddb31f8f5f38f
(PM: Introduce functions for suspending and resuming device interrupts), plaese?

This commit is in the latest git, but it's does not seem to fix the problem. 2.6.29 STR works very well. And all the Think "Fn" keys are no longer working (LID close still functioning though ... but that could a configuration issue moving from 2.6.29 to 2.6.30-rc1.

I meant please test the kernel where commit 0a0c5168df270a50e3518e4f12bddb31f8f5f38f is the head. It is possible that one of the commits immediately following it broke things. If that's the case, this kernel will work for you.

Quoted 2 lines

Shall I do the bisecting? I think this is the only one not working for me other than the slow iptables that I reported earlier.

Please first do the test above.

Thanks,
Rafael

#5 lore raw
In reply to Theodore Ts'o
Quoted 6 lines · Theodore Ts'o

Hi, has anyone else noticed that suspend-to-ram seems to have gotten broken in 2.6.30-rc1, sometime during last half of the merge window? I've been at a conference all of this past week, so I haven't had time to bisect it yet. I was wondering if anyone else had noticed and figured out what was going wrong. Symptoms are that the suspend "moon" starts flashing and the system never manages to put itself to sleep.

s2ram seems to work ok for me on thinkpad x60... Maybe it is hw-dependend, mut maybe it is just a kernel config...

Pavel

Signature · 2 lines
(english) http://www.livejournal.com/~pavelmachek
(cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html
#6 lore raw
Quoted 10 lines · Pavel Machek
Quoted 6 lines · Theodore Ts'o

Hi, has anyone else noticed that suspend-to-ram seems to have gotten broken in 2.6.30-rc1, sometime during last half of the merge window? I've been at a conference all of this past week, so I haven't had time to bisect it yet. I was wondering if anyone else had noticed and figured out what was going wrong. Symptoms are that the suspend "moon" starts flashing and the system never manages to put itself to sleep.

s2ram seems to work ok for me on thinkpad x60... Maybe it is hw-dependend, mut maybe it is just a kernel config...

Can you attach yours, so that we could compare, please?

Rafael

#7 lore raw
Quoted 1 line · Rafael J. Wysocki

Can you attach yours, so that we could compare, please?

Attached is my .config.

Quoted 4 lines

I meant please test the kernel where commit 0a0c5168df270a50e3518e4f12bddb31f8f5f38f is the head. It is possible that one of the commits immediately following it broke things. If that's the case, this kernel will work for you.

I'll do that now.

Thanks,
Jeff.

#8 lore raw

So I tried to do a bisect using a reduced config, and I wasn't able to reproduce. So I went back to 2.6.30-rc1 and started playing with configurations. I was able to finally get a reproduction, and I found that the af_packet module seems to be responsible. If I move the modules directory out of the way and reboot, then manually insmod the af_packet module, then suspend, it fails by hanging with the flashing moon LED. If I reboot and don't load the af_packet module, the suspend works. Needless to say this falls into the WTF category. I will do more testing to see what's up...

- Ted

-----Original Message-----
From: Jeff Chua <[email protected]>

Date: Sat, 11 Apr 2009 09:12:28
To: Rafael J. Wysocki<[email protected]>
Cc: Pavel Machek<[email protected]>; Theodore Ts'o<[email protected]>; <[email protected]>
Subject: Re: suspend-to-ram broken on X61 thinkpads in 2.6.30-rc1?

Quoted 1 line · Rafael J. Wysocki

Can you attach yours, so that we could compare, please?

Attached is my .config.

Quoted 4 lines

I meant please test the kernel where commit 0a0c5168df270a50e3518e4f12bddb31f8f5f38f is the head. It is possible that one of the commits immediately following it broke things. If that's the case, this kernel will work for you.

I'll do that now.

Thanks,
Jeff.

���{.n�+�������+%�����ݶ��w��{.n�+����{��G�����{ay�ʇڙ�,j��f���h��������z_�(�階�ݢj"���m������G������?����&���~��iO��z��v�^�m���� ������?�I�

#9 lore raw
Quoted 1 line · Theodore Tso

So I tried to do a bisect using a reduced config, and I wasn't able to reproduce.  So I went back to 2.6.30-rc1 and started playing with configurations.   I was able to finally get a reproduction, and I found that the af_packet module seems to be responsible.   If I move the modules directory out of the way and reboot, then manually insmod the af_packet module, then suspend, it fails by hanging with the flashing moon LED.  If I reboot and don't load the af_packet module, the suspend works.   Needless to say this falls into the WTF category.   I will do more testing to see what's up...

Latest linux git-pull seems to fix the problem. Suspend-to-ram and resume now works. I've tried a few times and it works. Reverting the patch makes it hang during suspend.

Latest git pull from 62b8e680e61d3f48f2a12ee248ca03ea8f376926 to d848223808c5d21e1b3cea090047e34722c6254b

Thanks,
Jeff.

#10 lore raw
Quoted 9 lines · Jeff Chua
Quoted 1 line · Theodore Tso

So I tried to do a bisect using a reduced config, and I wasn't able to reproduce. So I went back to 2.6.30-rc1 and started playing with configurations. I was able to finally get a reproduction, and I found that the af_packet module seems to be responsible. If I move the modules directory out of the way and reboot, then manually insmod the af_packet module, then suspend, it fails by hanging with the flashing moon LED. If I reboot and don't load the af_packet module, the suspend works. Needless to say this falls into the WTF category. I will do more testing to see what's up...

Latest linux git-pull seems to fix the problem. Suspend-to-ram and resume now works. I've tried a few times and it works. Reverting the patch makes it hang during suspend.

Latest git pull from 62b8e680e61d3f48f2a12ee248ca03ea8f376926 to d848223808c5d21e1b3cea090047e34722c6254b

Great, thanks for checking this.

Ted, can you please see if the current -git fixed the problem for you too?

Rafael

#11 lore raw
Quoted 5 lines · Rafael J. Wysocki

Great, thanks for checking this.

Ted, can you please see if the current -git fixed the problem for you too?

Confirmed, it seems to be fixed.

- Ted

#12 lore raw
Quoted 8 lines · Theodore Tso
Quoted 5 lines · Rafael J. Wysocki

Great, thanks for checking this.

Ted, can you please see if the current -git fixed the problem for you too?

Confirmed, it seems to be fixed.

Great, thanks!

Rafael