Subject: Re: UBC breaks my IDE disk!?!
To: Urban Boquist <urban@boquist.net>
From: Chuck Silvers <chuq@chuq.com>
List: current-users
Date: 02/03/2001 23:29:01
hi urban,

yea, this certainly sounds like a UBC problem given the date when it started
happening.  it sorta sounds like the pages are being unmapped before the i/o
completes, but I would expect it to crash once it downgraded to PIO mode
if that were the case.

are the sector numbers in the error messages within the bounds of the partition?
what are you running to generate load on the system?

manuel, can you shed any light on this?

-Chuck


On Sat, Feb 03, 2001 at 12:02:38PM +0100, Urban Boquist wrote:
> Hi all,
> 
> on my laptop, a Toshiba Portege 7020CT the following holds:
> 
> - a kernel from 2000-11-25, i.e. right before UBC, works just great.
> 
> - a kernel from 2000-11-28, i.e. with UBC, fails miserably when
>   putting some pressure on the hard disk. I get lots of the infamous:
> 
>     pciide0:0:0: lost interrupt
>             type: ata tc_bcount: 65536 tc_skip: 0
>     pciide0:0:0: bus-master DMA error: status=0x22
>     pciide0:0:0: bus-master DMA error: missing interrupt, status=0x22
>     wd0e: DMA error reading fsbn 7200512 of 7200512-7200639 (wd0 bn
>     9986372; cn 10567 tn 8 sn 53), retrying
> 
>   even when the driver downgrades to not using DMA at all the problem
>   continues.
> 
>   Moreover, all the errors are reported as "soft error (corrected)"
>   but they still result in corrupted files... ;-(
> 
> - more recent -current kernels also fail in the same way, the last one
>   I tried was 2000-02-01.
> 
> Although one should never say this in the i386 world, I really don't
> think the hardware is the problem here. With the 2000-11-25 kernel I
> never see any IDE errors or corrupted files, and I have pounded on the
> machine very hard...
> 
> Any ideas would be greatly appreciated! I'm including full dmesg below
> for the working and a non-working kernel.
> 
> Thanks,
> 
>         -- Urban