Skip to contentSkip to conversation

Data corruption during read on VIA vt8235

DISCUSSION Marcin Garski 2 participants 2 messages Mar 5, 2004 – Mar 6, 2004

lore
Share & tools
lore
#0 lore raw

[Please CC me on replies, I am not subscribed to the list, thanks] Hi,

I've Soltek SL-75FRV mainboard (VIA KT400 and vt8235 chipsets). Also i've two IDE disk both runing on UDMA(100) (DMA enabled). I'm using 2.4.22 kernel from Fedora Core 1 + patch for XFS suppport.

Several checking md5 sum of big file (650MB) give different results
(e.g: first, second and third* *file check give good md5 sum, but fourth check give bad sum). Also if i copy big file through network (ethernet), file have bits difference, the same
thing happen during file copy (also big file) betwen two disks.
Usually there are from 1 to 3 differneces in file, each difference is one bit (e.g good file - 4B, bad file - 4A).

That is not a memory problem because memtest86 shows no errors.

I found some old message:
http://www.ussg.iu.edu/hypermail/linux/kernel/0111.0/0914.html
where author had similar problem to mine.

Could you give me some hints how to more deeply diagnose this problem.

Signature · 2 lines
Best Regards
Marcin Garski
#1 lore raw
Quoted 7 lines · Marcin Garski

[Please CC me on replies, I am not subscribed to the list, thanks] Hi,

I've Soltek SL-75FRV mainboard (VIA KT400 and vt8235 chipsets). Also i've two IDE disk both runing on UDMA(100) (DMA enabled). I'm using 2.4.22 kernel from Fedora Core 1 + patch for XFS suppport.

Quoted 8 lines

Several checking md5 sum of big file (650MB) give different results (e.g: first, second and third* *file check give good md5 sum, but fourth check give bad sum).
Also if i copy big file through network (ethernet), file have bits difference, the same
thing happen during file copy (also big file) betwen two disks. Usually there are from 1 to 3 differneces in file, each difference is one bit (e.g good file - 4B, bad file - 4A).

If md5sum fails, those too will fail.

Quoted 7 lines

That is not a memory problem because memtest86 shows no errors.

I found some old message:
http://www.ussg.iu.edu/hypermail/linux/kernel/0111.0/0914.html
where author had similar problem to mine.

Could you give me some hints how to more deeply diagnose this problem.

Does it happen with UDMA66? UDMA33? etc
Kernel .config?
--
vda