Julian Coleman --> port-sparc64 (2007-02-09 15:15:45 +0000):
> >                               But then I noticed data corruption while
> > running a GENERIC kernel - and found the following in the GENERIC
> > config file:
> > 
> > ## Disable UDMA 4 which causes data corruption on the Acer Labs
> > ## chipset on Sun Blade 100 and Netra X1 machines.
> > wd*     at atabus? drive ? flags 0x0000
> > 
> > However, IIRC downgrading to UDMA3 did not fix the problems either,
> 
> I've been running a Blade 100 for a while (currently 4.0_BETA) with:
> 
>   # Disable UDMA 5 & 6 for disks
>   wd0     at atabus0 drive 0 flags 0x0c00
>   wd1     at atabus1 drive 0 flags 0x0c00
>   # Disable UDMA and DMA for CD
>   cd0     at atapibus0 drive 1 flags 0x0ff0       # ATAPI CD-ROM drives
> 
> The controller/drives attach as:
[...]
>   raid0: RAID Level 1
>   raid0: Components: /dev/wd0a /dev/wd1a
>   raid0: Total Sectors: 160836352 (78533 MB)
> 
> I've not seen any data corruption problems.  I did have the hard drives
> running faster, but ended up with too many:
> 
>   wd1a: DMA error writing fsbn ...
>   wd1: soft error (corrected)
> 
> messages.  I need to try them with 80-wire cables.
> 
> I couldn't get the CD drive to work with DMA at all, so it's forced to be
> PIO only.
> 
> Do you have a hard drive attached as the slave on either channel?

My Blade 100 is not in use anymore, I can't do any tests now... But I
also had a RAID1 configuration using two disks (as masters each), and
no CD drives attached. And IIRC after forcing UDMA mode 2 or 3 I didn't
see data corruption errors anymore, but still the soft errors you
mention.

Regards, Jukka

-- 
bashian roulette:
$ ((RANDOM%6)) || rm -rf ~