Skip to contentSkip to conversation

s2disk hangs with kernel >=2.6.29, SATA, Gigabyte EG45M-DS2H (bisected)

DISCUSSION Richard Atterer 3 participants 7 messages May 16, 2009 – May 18, 2009 libata

lore
Share & tools
lore
#0 lore raw

Hello,

I have had problems using s2disk starting sometime before 2.6.29 (and continuing up until now, 2.6.30-rc6), but only got around to bisecting the issue now.

The symptom is as follows: When I start s2disk (by running the hibernate command), the system switches to the console and prints "s2disk:
Snapshotting system". The cursor freezes for a moment, I hear the noise that my DVD writer makes when it is reset, the lights of my USB hub go off and on again. Then the cursor continues to blink and nothing more happens, the system just hangs.

I have a Gigabyte EG45M-DS2H motherboard with two disks attached: One SATA disk containing Linux, and one older PATA disk with data. AHCI is enabled in the BIOS.

The patch that causes the hang is 295f00042aaf6b553b5f37348f89bab463d4a469:
ide: don't execute the next queued command from the hard-IRQ context (v2)

Any help with this problem would be very much appreciated!

Cheers,

Richard

Signature · 4 lines
  __   _
  |_) /|  Richard Atterer     |  GnuPG key: 888354F7
  | \/¯|  http://atterer.net  |  08A9 7B7D 3D13 3EF2 3D25  D157 79E6 F6DC 8883 54F7
  ¯ '` ¯
#1 lore raw
Quoted 4 lines · Richard Atterer

Hello,

I have had problems using s2disk starting sometime before 2.6.29 (and continuing up until now, 2.6.30-rc6), but only got around to bisecting the issue now.

Quoted 8 lines

The symptom is as follows: When I start s2disk (by running the hibernate command), the system switches to the console and prints "s2disk: Snapshotting system". The cursor freezes for a moment, I hear the noise that my DVD writer makes when it is reset, the lights of my USB hub go off and on again. Then the cursor continues to blink and nothing more happens, the system just hangs.

I have a Gigabyte EG45M-DS2H motherboard with two disks attached: One SATA disk containing Linux, and one older PATA disk with data. AHCI is enabled in the BIOS.

The patch that causes the hang is 295f00042aaf6b553b5f37348f89bab463d4a469:
ide: don't execute the next queued command from the hard-IRQ context (v2)

Any help with this problem would be very much appreciated!

Does it work if you disable CONFIG_IDE, and use only CONFIG_ATA for PATA?

Jeff

#2 lore raw
Quoted 1 line · Jeff Garzik

Does it work if you disable CONFIG_IDE, and use only CONFIG_ATA for PATA?

Yes, it works fine then!

I had to disconnect my PATA data drive to avoid the disks from being renumbered before I could test without CONFIG_IDE.

(But I then double-checked with the broken CONFIG_IDE=m kernel: The failure happens independently of whether the second PATA drive is connected or not.)

Cheers,

Richard

Signature · 4 lines
  __   _
  |_) /|  Richard Atterer     |  GnuPG key: 888354F7
  | \/¯|  http://atterer.net  |  08A9 7B7D 3D13 3EF2 3D25  D157 79E6 F6DC 8883 54F7
  ¯ '` ¯
Tested-by Simon Holm Thøgersen
#3 lore raw
In reply to Richard Atterer
Quoted 13 lines · On Sunday 17 May 2009 00:51:53 Richard Atterer wrote:

Hello,

I have had problems using s2disk starting sometime before 2.6.29 (and continuing up until now, 2.6.30-rc6), but only got around to bisecting the issue now.

The symptom is as follows: When I start s2disk (by running the hibernate command), the system switches to the console and prints "s2disk:
Snapshotting system". The cursor freezes for a moment, I hear the noise that my DVD writer makes when it is reset, the lights of my USB hub go off and on again. Then the cursor continues to blink and nothing more happens, the system just hangs.

Quoted 6 lines

I have a Gigabyte EG45M-DS2H motherboard with two disks attached: One SATA disk containing Linux, and one older PATA disk with data. AHCI is enabled in the BIOS.

The patch that causes the hang is 295f00042aaf6b553b5f37348f89bab463d4a469:
ide: don't execute the next queued command from the hard-IRQ context (v2)

Hmm, this should have been fixed by:

commit 2ea5521022ac8f4f528dcbae02668e02a3501a5a

Author: Bartlomiej Zolnierkiewicz <[email protected]>
Date:   Wed Jan 14 19:19:04 2009 +0100

ide: fix suspend regression

On Monday 12 January 2009, Simon Holm Thøgersen wrote:
> commit 295f000 ("ide: don't execute the next queued command from the > hard-IRQ context (v2)") breaks suspend to disk for me. On > 'echo disk > /sys/power/state' the systems hangs, letting me switch > virtual consoles, but not responding to Alt+SysRq

Restart the request queue early for REQ_TYPE_PM_RESUME requests (though there is only one resume request for the whole resume sequence it stays in the queue until is fully completed and now depends on kblockd for processing consequential resume states).

What was the last 'good' kernel that you've started bisection with?

Thanks,
Bart

#4 lore raw
Quoted 11 lines · Bartlomiej Zolnierkiewicz
Quoted 2 lines · On Sunday 17 May 2009 00:51:53 Richard Atterer wrote:

The patch that causes the hang is 295f00042aaf6b553b5f37348f89bab463d4a469:
ide: don't execute the next queued command from the hard-IRQ context (v2)

Hmm, this should have been fixed by:

commit 2ea5521022ac8f4f528dcbae02668e02a3501a5a

Author: Bartlomiej Zolnierkiewicz <[email protected]>
Date:   Wed Jan 14 19:19:04 2009 +0100

ide: fix suspend regression

I just tried that revision (with a small patch to get pcf50633-core to compile) and hibernating does not work. :-/

Quoted 1 line

What was the last 'good' kernel that you've started bisection with?

I started with the 2.6.28 release.

In case it makes a difference: The swap partition that I'm hibernating to is encrypted via luks and uses lvm. This is the standard way suggested by the Debian installer when you choose encryption.

Cheers,

Richard

PS: This is my /etc/uswsusp.conf file:
resume device = /dev/mapper/pc-swap
splash = n
compress = y
early writeout = y
image size = 1897884221
RSA key file = /etc/uswsusp.key <-- that file does not exist
shutdown method = platform

Signature · 4 lines
  __   _
  |_) /|  Richard Atterer     |  GnuPG key: 888354F7
  | \/¯|  http://atterer.net  |  08A9 7B7D 3D13 3EF2 3D25  D157 79E6 F6DC 8883 54F7
  ¯ '` ¯
#5 lore raw
Quoted 7 lines · On Monday 18 May 2009 16:16:46 Richard Atterer wrote:
Quoted 6 lines · Bartlomiej Zolnierkiewicz
Quoted 2 lines · On Sunday 17 May 2009 00:51:53 Richard Atterer wrote:

The patch that causes the hang is 295f00042aaf6b553b5f37348f89bab463d4a469:
ide: don't execute the next queued command from the hard-IRQ context (v2)

Hmm, this should have been fixed by:

Quoted 8 lines
Quoted 5 lines

commit 2ea5521022ac8f4f528dcbae02668e02a3501a5a

Author: Bartlomiej Zolnierkiewicz <[email protected]>
Date:   Wed Jan 14 19:19:04 2009 +0100

ide: fix suspend regression

I just tried that revision (with a small patch to get pcf50633-core to compile) and hibernating does not work. :-/

Ok, thanks for testing it, also just to make sure:

Does the one commit before 295f00042aaf6b553b5f37348f89bab463d4a469 (namely ebdab07dad3d3a008e519b0a028e1e1ad5ecaef0) work fine?

Thanks.
Bart

#6 lore raw
Quoted 2 lines · Bartlomiej Zolnierkiewicz

Does the one commit before 295f00042aaf6b553b5f37348f89bab463d4a469 (namely ebdab07dad3d3a008e519b0a028e1e1ad5ecaef0) work fine?

Yes, it does; I tried that one during bisecting, and then re-tried before writing my first mail, just to ensure I hadn't messed something up.

IIRC I also tried to modify my 2.6.30-rc6 checkout to reinstate the calls to spin_unlock() and local_irq_enable_in_hardirq() that were replaced with spin_unlock_irq() by patch 295f000. This also didn't cause the problem to go away. Anyway, I have no clue what all this code does, so that was just a shot in the dark! :-)

Cheers,

Richard

Signature · 4 lines
  __   _
  |_) /|  Richard Atterer
  | \/¯|  http://atterer.net
  ¯ '` ¯