Ubuntu
linux-source-2.6.15 package

XFS corruption post-suspend-to-RAM in 2.6.15-25

Bug #50275 reported by Sebastian Kapfer
6
Affects Status Importance Assigned to Milestone
linux-source-2.6.15 (Ubuntu)
Invalid
High
Unassigned
linux-source-2.6.17 (Ubuntu)
Fix Released
Undecided
Unassigned

Bug Description

Binary package hint: linux-source-2.6.15

Laptop (IBM R40) with installation of dapper current as of 2006-06-19

Root filesystem is XFS on internal IDE (hda2), details below

Steps to reproduce:

1. Put laptop to sleep in GNOME session (special function keys on the laptop)
2. Resume
3. FS is corrupted. The system may freeze suddenly. Verify by rebooting to single-user, run xfs_check.
4. Running xfs_repair kills a few dozen files which were not being written to (including gnomevfs libs, evolution binaries...) but which obviously were in memory. Critical things like "sed" may also disappear, causing initscripts to fail at next boot.

The system is fine when not being suspended but properly shut down.
Suspending worked fine for months in breezy. I did not test suspending in the dapper development versions because I wasn't using the laptop much then.

Sorry for the vague report. I am happy to provide further input if there are any questions.

Hardware details:

System is a IBM R40 2722, original config but replaced RAM.

RAM is working (two days of memtest86)

IDE chipset:

0000:00:1f.1 IDE interface: Intel Corporation 82801DBM (ICH4-M) IDE Controller (rev 01) (prog-if 8a [Master SecP PriP])
        Subsystem: IBM: Unknown device 052d
        Flags: bus master, medium devsel, latency 0, IRQ 11
        I/O ports at <unassigned>
        I/O ports at <unassigned>
        I/O ports at <unassigned>
        I/O ports at <unassigned>
        I/O ports at 1860 [size=16]
        Memory at 50000000 (32-bit, non-prefetchable) [size=1K]

hard drive:

[17179573.392000] hda: HITACHI_DK23EA-60B, ATA DISK drive

(60 GB PATA, DMA mode)

Tags: dapper
Revision history for this message
Daniel T Chen (crimsun) wrote :

Are you positive that the ThinkPad actually suspends-to-RAM completely? These symptoms read eerily similar to bug 50031 ...

Changed in linux-source-2.6.15:
importance: Untriaged → High
status: Unconfirmed → Needs Info
Revision history for this message
Sebastian Kapfer (caci) wrote :

Hi,

no, I think my Thinkpad does sleep completely. I can do none of the things the people in the other bug can do, like toggling capslock, magic sysrq, Thinklight etc.

/etc/default/acpi-support:

ACPI_SLEEP=true
ACPI_SLEEP_MODE=mem
ENABLE_LAPTOP_MODE=true

otherwise unchanged.

Revision history for this message
Michael W. (michael-w00) wrote :

I had the same problem a few weeks ago on my Samsung P35. I too am sure that the laptop suspends completely because sometimes I was able to resume without problem, but tow or three times the filesystem was corrupted and since then I haven't suspended the laptop. I tested suspending in dapper/testing and it worked without any issues.

Revision history for this message
Olivier Cortès (olive) wrote :

I don't have XFS but I suffer exactly a similar behaviour with ext3. Nearly the same symptoms. see bugs #53102 and possibly #16610...

Revision history for this message
Ben Collins (ben-collins) wrote :

Perhaps this doesn't have anything to do with xfs. Please submit your outputs for these commands:

lspci -vv
lspci -vvn

Let's see what sort of block hardware we have.

Also, if anyone has dmesg output after this corruption has occured, I'd appreciate it.

Revision history for this message
Ben Collins (ben-collins) wrote :

This bug may be fixed in the current edgy developement system. In an effort to track this bug to our latest distribution, it is being targeted for edgy.

Please confirm whether this bug exists in edgy. If not, then please re-attach all related output (e.g. dmesg, oops output) while under edgy.

Latest Edgy CD's can be downloaded from:

http://cdimage.ubuntu.com/releases/edgy/knot-1/

If this image does not boot fom you, you can also download a current daily build from:

http://cdimage.ubuntu.com/daily-live/current/

Note that it may only be necessary to boot the LiveCD to see if your bug is fixed. In this case, you do not need to upgrade your installed system at all to confirm it. Certain bugs may require that you do an actual installation.

Changed in linux-source-2.6.17:
status: Unconfirmed → Needs Info
Revision history for this message
Sebastian Kapfer (caci) wrote :

Just tried 10 suspend-resume cycles with Dapper, and it didn't break. Now I'm puzzled. The XFS still breaks sometimes, but it's not directly related to suspending. *scratches head*

Maybe it's laptop-mode. I'll try that next, and a Edgy disc after that.

Revision history for this message
Sebastian Kapfer (caci) wrote :

Sorry but neither Knot-1 nor Knot-2 start on my machine. The CD is fine, I verified MD5 from my hard disk installation. The CD however does not work. Probably something with the graphics (Radeon M7).

Revision history for this message
Sebastian Kapfer (caci) wrote :

Not sure what to do. A clean edgy works fine on the same machine. Probably we can close this for lack of facts (?)

Revision history for this message
Sebastian Kapfer (caci) wrote :

Since XFS has had tons of bugs fixed, and this particular issue is no longer reproducible, I'm closing my own bug.

Changed in linux-source-2.6.17:
status: Needs Info → Fix Released
Revision history for this message
Launchpad Janitor (janitor) wrote :

[Expired for linux-source-2.6.15 (Ubuntu) because there has been no activity for 60 days.]

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.
Subscribing...

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.