2002-03-14 20:39:00

by David Forrest

[permalink] [raw]
Subject: K7S5A SIS735 ext2fs corruption

I know this has been on the list before, but I wanted to whine a bit, and
collect what I've seen because I havent seen it resolved

lkml: 2001-12-30 UMOUNTING in 2.4.17 / Ext2 Partitions destroyed (3x)
lkml: 2001-10-25 Repeatable File Corruption (ECS K7S5A w/SIS735)

http://www.geocities.com/mrathlon2000/ and
http://pub65.ezboard.com/fk7s5amotherboardforumfrm10.showMessage?topicID=2.topic
suggests a couple hardware fixes, and blames corruptions on high speeds
and a bad resistor choice.

My ECS k7s5a with the SIS735 Athalon 1.4 chipset has corrupted a couple of
my disks

I was using 2.4.17, and the dma modes kicked in automatically. I thought
I had it managed with the boot parameter of 'ide=nodma', since that
eliminatated errors like
status=0x51 { DriveReady SeekComplete Error }
error=0x84 { BadCRC DriveStatusError }

These were with the drives in the hdparm output below.

I don't have this board up and running right now, but could throw some old
stuff into it and see how it goes.

Dave,
--
Dave Forrest [email protected]
(434)296-7283h 924-3954w http://mug.sys.virginia.edu/~drf5n/


{root@mug:~}# /usr/sbin/hdparm -i /dev/hd[ad]

/dev/hda:

Model=IBM-DHEA-38451, FwRev=HP8OA20C, SerialNo=SH0SH0S4378
Config={ HardSect NotMFM HdSw>15uSec Fixed DTR>10Mbs }
RawCHS=16383/16/63, TrkSize=0, SectSize=0, ECCbytes=28
BuffType=DualPortCache, BuffSize=472kB, MaxMultSect=16, MultSect=off
CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=16514064
IORDY=on/off, tPIO={min:240,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: sdma0 sdma1 sdma2 mdma0 mdma1 mdma2 udma0 udma1 *udma2
AdvancedPM=no
Drive Supports : ATA-3 X3T10 2008D revision 1 : ATA-1 ATA-2 ATA-3


/dev/hdd:

Model=WDC WD205BA, FwRev=16.13M16, SerialNo=WD-WM9490019722
Config={ HardSect NotMFM HdSw>15uSec SpinMotCtl Fixed DTR>5Mbs FmtGapReq
}
RawCHS=16383/16/63, TrkSize=57600, SectSize=600, ECCbytes=40
BuffType=DualPortCache, BuffSize=2048kB, MaxMultSect=16, MultSect=off
CurCHS=16383/16/63, CurSects=-66060037, LBA=yes, LBAsects=40088160
IORDY=on/off, tPIO={min:120,w/IORDY:120}, tDMA={min:120,rec:120}
PIO modes: pio0 pio1 pio2 pio3 pio4
DMA modes: mdma0 mdma1 mdma2 udma0 udma1 *udma2 udma3 udma4
AdvancedPM=no
Drive Supports : Reserved : ATA-1 ATA-2 ATA-3 ATA-4




2002-03-14 21:27:38

by Alan

[permalink] [raw]
Subject: Re: K7S5A SIS735 ext2fs corruption

> lkml: 2001-12-30 UMOUNTING in 2.4.17 / Ext2 Partitions destroyed (3x)
> lkml: 2001-10-25 Repeatable File Corruption (ECS K7S5A w/SIS735)

Please try 2.4.19pre3 - Lionel has been doing sterling work updating all the
SIS drivers. If you still have a problem with the update we *really* want
to know about it

2002-03-15 10:56:43

by Lionel Bouton

[permalink] [raw]
Subject: Re: K7S5A SIS735 ext2fs corruption

On Thu, Mar 14, 2002 at 08:38:36PM -0500, David Forrest wrote:
> I know this has been on the list before, but I wanted to whine a bit, and
> collect what I've seen because I havent seen it resolved
>
> lkml: 2001-12-30 UMOUNTING in 2.4.17 / Ext2 Partitions destroyed (3x)
> lkml: 2001-10-25 Repeatable File Corruption (ECS K7S5A w/SIS735)
>
> http://www.geocities.com/mrathlon2000/ and
> http://pub65.ezboard.com/fk7s5amotherboardforumfrm10.showMessage?topicID=2.topic
> suggests a couple hardware fixes, and blames corruptions on high speeds
> and a bad resistor choice.
>
> My ECS k7s5a with the SIS735 Athalon 1.4 chipset has corrupted a couple of
> my disks
>

Try latest ac kernel or my patches
at http://inet6.dyn.dhs.org/sponsoring/sis5513/index.html
(stock kernel doesn't have correct timings yet).

Look at your board revision too. I had a k7s5al (k7s5a + lan) rev 4+ which did corrupt
data. The new one I own is a rev 7 and doesn't exhibit a single problem.

You can find the revision black printed on a white sticker near the last PCI slot
(the one farest from the AGP one).

If you still have data corruption with latest ac kernel or with my
latest patch applied you definitely have a faulty board.

LB.