Bug 7004 - sis5513 dma corrupts filesystem
Summary: sis5513 dma corrupts filesystem
Status: REJECTED INSUFFICIENT_DATA
Alias: None
Product: IO/Storage
Classification: Unclassified
Component: IDE (show other bugs)
Hardware: i386 Linux
: P2 normal
Assignee: Bartlomiej Zolnierkiewicz
URL:
Keywords:
Depends on:
Blocks:
 
Reported: 2006-08-14 07:47 UTC by Marijn Schouten
Modified: 2008-02-13 08:47 UTC (History)
3 users (show)

See Also:
Kernel Version: 2.6.16.26
Subsystem:
Regression: ---
Bisected commit-id:


Attachments
dmesg of 2.6.17.11 kernel from cd with ide=nodma (11.05 KB, text/plain)
2006-08-27 15:39 UTC, Marijn Schouten
Details
the kernel config corresponding to the previous dmesg log (38.31 KB, text/plain)
2006-08-27 15:41 UTC, Marijn Schouten
Details

Description Marijn Schouten 2006-08-14 07:47:12 UTC
Most recent kernel where this bug did not occur: unknown
Distribution: lots
Hardware Environment:

0000:00:00.0 Host bridge: Silicon Integrated Systems [SiS] 650/M650 Host (rev 
01)
0000:00:01.0 PCI bridge: Silicon Integrated Systems [SiS] Virtual PCI-to-PCI 
bridge (AGP)
0000:00:02.0 ISA bridge: Silicon Integrated Systems [SiS] SiS961 [MuTIOL Media 
IO]
0000:00:02.1 SMBus: Silicon Integrated Systems [SiS] SiS961/2 SMBus Controller
0000:00:02.2 USB Controller: Silicon Integrated Systems [SiS] USB 1.0 
Controller (rev 07)
0000:00:02.3 USB Controller: Silicon Integrated Systems [SiS] USB 1.0 
Controller (rev 07)
0000:00:02.5 IDE interface: Silicon Integrated Systems [SiS] 5513 [IDE] (rev 
d0)
0000:00:02.7 Multimedia audio controller: Silicon Integrated Systems [SiS] 
Sound Controller (rev a0)
0000:00:06.0 Communication controller: Agere Systems LT WinModem
0000:00:08.0 Ethernet controller: Realtek Semiconductor Co., Ltd. 
RTL-8139/8139C/8139C+ (rev 10)
0000:00:0c.0 CardBus bridge: O2 Micro, Inc. OZ601/6912/711E0 
CardBus/SmartCardBus Controller
0000:00:0d.0 FireWire (IEEE 1394): Agere Systems FW323 (rev 04)
0000:01:00.0 VGA compatible controller: Silicon Integrated Systems [SiS] 
65x/M650/740 PCI/AGP VGA Display Adapter
0000:02:00.0 Network controller: RaLink Ralink RT2500 802.11 Cardbus Reference 
Card (rev 01)

Software Environment:

Problem Description: 
Booting a kernel without turning off dma, with kernel option ide=nodma causes 
corruption of the root filesystem which causes booting to fail.

I've tried with lots of kernels including 2.6.16.26, 2.6.18_rc3, 
2.6.17-gentoo-r4 and (x)ubuntu-2.6.15-26-386.

An ubuntu bug including dmesg logs is here: 
https://launchpad.net/distros/ubuntu/+source/linux-source-2.6.15/+bug/56233
Comment 1 Alan 2006-08-14 11:15:22 UTC
[17179717.108000] ltmodem: module license 'Proprietary' taints kernel.
[17179717.108000] Loading Lucent Modem Controller driver version 8.26-alk-8

To start with can you duplicate the problem without this loaded.

Other things that look relevant given the mess spewed forth when booting would
be acpi=off (as the ACPI appears terminally broken in the BIOS).

Alan
Comment 2 Marijn Schouten 2006-08-16 04:41:50 UTC
I've duplicated this problem without that proprietary driver.

acpi=off doesn't seem to help.

I'll be reinstalling gentoo, so i can more easily try any kernel you want me to 
with any boot options and provide a .config if that is helpful.
Comment 3 Marijn Schouten 2006-08-27 15:39:06 UTC
Created attachment 8884 [details]
dmesg of 2.6.17.11 kernel from cd with ide=nodma

I don't know if it is useful at all, but here is a log _with_ ide=nodma. I was
able to boot this also without ide=nodma, but then I can only get into an
extremely limited shell(ash) within busybox. And in that case the first half of
dmesg output is simply not there and I cannot get it off in any case. I might
try setting up a netconsole link.
Comment 4 Marijn Schouten 2006-08-27 15:41:15 UTC
Created attachment 8885 [details]
the kernel config corresponding to the previous dmesg log
Comment 5 Alan 2006-12-01 12:31:45 UTC
Reassigning to IDE  maintainer since he's back
Comment 6 Bartlomiej Zolnierkiewicz 2006-12-16 16:35:59 UTC
Added to the big queue...

If somebody has time to work on it (Alan?) please do so.
Comment 7 Marijn Schouten 2007-01-03 04:39:36 UTC
When booting vectorlinux 5.8 with a kernel based on 2.6.18.5 I still need 
ide=nodma.

I've also tried Sabayon 3.25c with a kernel based on 2.6.19.1 and get a kernel 
panic there (see http://www.sabayonlinux.org/forum/viewtopic.php?t=3086)

EIP: [<c05afcb1>] sis_init_one+0x153/0x354 SS:ESP 0068:cef01ee4
Comment 8 colin redfern 2007-03-01 19:19:33 UTC
It appears that I also am uffering from this bug.
Hardware is an SiS 630 based motherboard with a controller recognised as a sis5513.
I have been running Linux on this system for years (I have been both upgrading
and reinstalling it since Redhat 7.1)

The only change has been replacing the original ancient CD ROM with a CD Rewriter.

I formatted the Harddrive and tried to install FC6  but there is corruption even
with the kernel option ide=nodma or ide0=nodma and ide1=nodma 

If I ever get FC6 installed I can post dmesg and lspci.

Comment 9 Marijn Schouten 2007-03-02 02:26:21 UTC
I heard ide=nodma was broken around kernels 2.6.19, which might explain your 
inability to install FC6.
Comment 10 Bartlomiej Zolnierkiewicz 2007-06-02 16:04:45 UTC
Marijin/Colin, is this still with 2.6.22-rc3
(it contains a *lot* of sis5513 tuning fixes).
Comment 11 Adrian Bunk 2007-07-13 21:00:30 UTC
Please reopen this bug if it's still present with kernel 2.6.22.
Comment 12 colin redfern 2007-07-16 05:16:29 UTC
(In reply to comment #9)
> I heard ide=nodma was broken around kernels 2.6.19, which might explain your 
> inability to install FC6.

Yup it would appear that the problem may have been exacerbated by the the particular kernel used by the installer. I eventually replaced the harddrive in case it was failing. It was not- that drive now working fine in another machine.
Now on 2.6.20 hdparm is working properly once again and once again I get >22MB sec. :-)
Comment 13 Marijn Schouten 2007-12-25 06:19:56 UTC
I tried to boot with Ubuntu Gutsy 7.10 i386, but ide=nodma doesn't let me do that anymore, nor did "acpi=off noapic ide=nodma". I get a lot of:

ata1.00: configured for UDMA/33
ata1: EH complete
ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen

then later:

ata1.00: configured for UDMA/25
ata1: EH complete
ata1.00: exception Emask 0x0 SAct 0x0 SErr 0x0 action 0x2 frozen

This last bit keeps on repeating longer than my patience.

The installed XUbuntu 6.06 still works fine with ide=nodma, thus my harddrive is still fine.
Comment 14 Marijn Schouten 2007-12-25 06:38:47 UTC
(In reply to comment #10)
> Marijin/Colin, is this still with 2.6.22-rc3
> (it contains a *lot* of sis5513 tuning fixes).

Ubuntu Gutsy 7.10 i386 seems to come with 2.6.22 something... but I will try out some more cd's.
Comment 15 Bartlomiej Zolnierkiewicz 2007-12-25 06:53:14 UTC
Ubuntu Gutsy 7.10 uses SCSI/libata instead of IDE subsystem by default...
Comment 16 Marijn Schouten 2008-01-08 04:11:24 UTC
SCSI/libata is supposed to replace the IDE subsystem, right? 
Or is it the other way around?
Comment 17 Bartlomiej Zolnierkiewicz 2008-01-08 04:29:17 UTC
The most distributions have switched to SCSI/libata, OTOH IDE has been through rapid improvements lately... so I guess the answer depends on who you ask ;)
Comment 18 Bartlomiej Zolnierkiewicz 2008-02-13 08:47:26 UTC
I'm closing this bug assuming that it was fixed somewhere along the way... 
(please re-open if it still happens with some recent kernel, i.e. 2.6.24).

Note You need to log in before you can comment on or make changes to this bug.