[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

Lock up after resume from S3 suspend when ide dma is enabled (FC5)



Hello,

This follows on from a post on the fedora-list...
https://www.redhat.com/archives/fedora-list/2006-April/msg00581.html

Quick summary -- S3 suspend works, system resumes ok, then
after about 20-60 secs it locks up completely.

I've tracked this down to ide dma - booting the kernel with
ide=nodma removes the problem.

I appreciate this is is more than likely an upstream kernel
issue, but hopefully people here could help me confirm this,
and identify the problem in more detail, before taking
things there.

I assume this is somewhat hardware specific, since it
doesn't seem to have been reported already.
I've got a Samsung SP1614N (160GB) PATA disk, running off an
nForce 2 chipset with integrated graphics (GF4MX). (Its a
Soltek SL7-BA-F mainboard with an Athlon XP Barton CPU).


Test case to reproduce on stock FC5 (both 2.6.15-1.2054 and
2.6.16-1.2080 kernels)

1. Reboot, and hit 'a' at the grub prompt to edit the kernel
parameters
2. Add 'apm=off ide=nodma init=/bin/bash', and then boot
(I'm not sure if the apm=off is really necessary)
3. Enter 'echo "mem" > /sys/power/state'
...system should suspend...
4. Wake system
(At this point I have no video (see note below), so I have
to type blind, and toggle the caps lock key to check the
system is still responding)
5. Type some commands, toggle caps lock, wait a few mins.
Confirm the system is still alive.

6. Repeat as above, but remove 'ide=nodma' from the kernel
parameters, and confirm that the system locks up after a
minute or two.
(For me, with dma enabled, once I hit the power button to
resume, the system disk activity LED also comes on and stays
on indefinitely)

Note on video --- Without the nvidia driver loaded, the
video doesn't come back for me after a resume. I'm bypassing
the loading of this, and pretty much everything else using
the init=/bin/bash kernel parameter to give a minimal system
to help isolate this issue. But just to confirm - using the
Nvidia driver rpms from atrpms as described in the
fedora-list post, the system suspends and resumes from X
successfully.


I'd appreciate it if anyone who has a Samsung PATA disk
and/or an nForce 2 chipset could try out this test case.And
any suggestions for getting more information on the lock-up
would be helpful.

There are some kernel messages showing up before it hangs,
but they don't get written to the log because of the ide
issues. I'm going to see if I can dump them to a usb flash
drive.
I've tried sysrq without success, but this is probably
because the video is locked up. Maybe sysrq over a serial
console - is this possible with the stock Fedora kernels?

Even with DMA disabled, some ide error messages show up in
the log...I don't know if they are significant or not
---------------------- 
hda: status error: status=0x51 { DriveReady SeekComplete
Error }
hda: status error: error=0x04 { DriveStatusError }
ide: failed opcode was: unknown
hda: no DRQ after issuing MULTWRITE_EXT
hda: status error: status=0x51 { DriveReady SeekComplete
Error }
hda: status error: error=0x04 { DriveStatusError }
ide: failed opcode was: unknown
hda: no DRQ after issuing MULTWRITE_EXT
 done
hda: status error: status=0x51 { DriveReady SeekComplete
Error }
hda: status error: error=0x04 { DriveStatusError }
ide: failed opcode was: unknown
hda: no DRQ after issuing MULTWRITE_EXT
hda: status error: status=0x51 { DriveReady SeekComplete
Error }
hda: status error: error=0x04 { DriveStatusError }
ide: failed opcode was: unknown
hda: no DRQ after issuing MULTWRITE_EXT
ide0: reset: success
---------------------- 

Regards,
Ken


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]