mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* libata/iswraid DMA Timeout
@ 2004-02-23 23:51 jabbera
  2004-02-24  4:00 ` Eric D. Mudama
  2004-02-24 17:20 ` Martins Krikis
  0 siblings, 2 replies; 5+ messages in thread
From: jabbera @ 2004-02-23 23:51 UTC (permalink / raw)
  To: linux-kernel

Sorry never posted to this list so bear with me. 
I am currently using kernel version 2.4.25 i386 with libata and iswraid. 
Note: iswraid is for 2.4.22, but it applied to 2.4.25 without issue. 
 
I am not sure which patch this problem is in relation to, but here is the 
information. 
 
I was using a php script to move a bunch of my files around the file system 
when my system locks up and I see: 
ata1: DMA timeout, stat 0x21 
ATA: abnormal status 0x59 on port 0xC407 
 
I try to do an emegency sync but I fails, and I am forced to reboot. 
This problem is reproducable. If any additional information is needed please 
email me as I am not a member of the mailing list. 
 
Thanks. 

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: libata/iswraid DMA Timeout
  2004-02-23 23:51 libata/iswraid DMA Timeout jabbera
@ 2004-02-24  4:00 ` Eric D. Mudama
  2004-02-24 17:20 ` Martins Krikis
  1 sibling, 0 replies; 5+ messages in thread
From: Eric D. Mudama @ 2004-02-24  4:00 UTC (permalink / raw)
  To: jabbera; +Cc: linux-kernel


On Mon, Feb 23 at 18:51, jabbera@student.umass.edu wrote:
>Sorry never posted to this list so bear with me. 
>I am currently using kernel version 2.4.25 i386 with libata and iswraid. 
>Note: iswraid is for 2.4.22, but it applied to 2.4.25 without issue. 
> 
>I am not sure which patch this problem is in relation to, but here is the 
>information. 
> 
>I was using a php script to move a bunch of my files around the file system 
>when my system locks up and I see: 
>ata1: DMA timeout, stat 0x21 
>ATA: abnormal status 0x59 on port 0xC407 

0x59 is the status that the drive sends when it has a block in error
and wants to transmit said block.  The LBA in the task file after the
bad block is transferred is the location of the error.

I think that the drive you're copying from had a hard ECC error.
Maybe see if you can try to read all the files on that drive and see
if you can find a bad block.


-- 
Eric D. Mudama
edmudama@mail.bounceswoosh.org


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: libata/iswraid DMA Timeout
  2004-02-23 23:51 libata/iswraid DMA Timeout jabbera
  2004-02-24  4:00 ` Eric D. Mudama
@ 2004-02-24 17:20 ` Martins Krikis
  2004-02-24 20:32   ` jabbera
  1 sibling, 1 reply; 5+ messages in thread
From: Martins Krikis @ 2004-02-24 17:20 UTC (permalink / raw)
  To: jabbera; +Cc: linux-kernel

--- jabbera@student.umass.edu wrote:

> Sorry never posted to this list so bear with me. 
> I am currently using kernel version 2.4.25 i386 with libata and
> iswraid. 
> Note: iswraid is for 2.4.22, but it applied to 2.4.25 without issue. 

Please try the new iswraid that was posted to this list on
Friday, Feb. 20th. It is completely redone and has a lot more
features than what you were using, including /proc fs support,
ICH6R support, updating RAID metadata on errors, ability to
work with many volumes and various configurations of them,
etc., etc. Yes, it's very new and new bugs could have been
introduced but I think you'll be better off with the latest
anyway.
 
> I was using a php script to move a bunch of my files around the file
> system 
> when my system locks up and I see: 
> ata1: DMA timeout, stat 0x21 
> ATA: abnormal status 0x59 on port 0xC407 

I had a bad SATA cable once and was seeing the same thing.
I reported it to libata1 author via personal email.
I don't think iswraid has anything to do with it as this
is a lower level driver issue. Iswraid, both old and new,
are fairly high level drivers that don't deal directly with
the SATA controllers or their interrupts. In fact, iswraid
would work with any SCSI disks that have the right metadata
format on them.

> I try to do an emegency sync but I fails, and I am forced to reboot.

In my experience the system was still usable, except the userland
application was hanging. Adding another SATA transaction (to any
disk, I believe), was what made it hang completely. So I think
we've seen the same problem.

I certainly don't understand enough of libata1 to fix this,
but you could perhaps start by changing your SATA cables or
at least swapping them around to see whether the timeout follows
the cable, perhaps, as it was in my case.

Good luck, and please report back any iswraid bugs/wishes/questions.

  Martins Krikis
  Storage Components Division (SCD)
  Intel Massachusetts


__________________________________
Do you Yahoo!?
Yahoo! Mail SpamGuard - Read only the mail you want.
http://antispam.yahoo.com/tools

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: libata/iswraid DMA Timeout
  2004-02-24 17:20 ` Martins Krikis
@ 2004-02-24 20:32   ` jabbera
  2004-02-24 20:43     ` Martins Krikis
  0 siblings, 1 reply; 5+ messages in thread
From: jabbera @ 2004-02-24 20:32 UTC (permalink / raw)
  To: Martins Krikis; +Cc: jabbera, linux-kernel

Quoting Martins Krikis <mkrikis@yahoo.com>:
[snip]
> > I try to do an emegency sync but I fails, and I am forced to reboot.
> 
> In my experience the system was still usable, except the userland
> application was hanging. Adding another SATA transaction (to any
> disk, I believe), was what made it hang completely. So I think
> we've seen the same problem.
> 
> I certainly don't understand enough of libata1 to fix this,
> but you could perhaps start by changing your SATA cables or
> at least swapping them around to see whether the timeout follows
> the cable, perhaps, as it was in my case.
[snip]

Was your fault a random occurance or could you always reproduce it. I have only 
seen mine doing this one particualar operation, and nowhere else, and it is 
perfectly reproducable every time.



^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: libata/iswraid DMA Timeout
  2004-02-24 20:32   ` jabbera
@ 2004-02-24 20:43     ` Martins Krikis
  0 siblings, 0 replies; 5+ messages in thread
From: Martins Krikis @ 2004-02-24 20:43 UTC (permalink / raw)
  To: jabbera; +Cc: linux-kernel

--- jabbera@student.umass.edu wrote:

> Was your fault a random occurance or could you always reproduce it. I
> have only 
> seen mine doing this one particualar operation, and nowhere else, and
> it is 
> perfectly reproducable every time.

Reproducible. Always happened on mount (not on fdisk, not on mke2fs,
I never got to anything beyond mount). I started swapping cables
and thus discovered the faulty cable. Of course, your problem
could be different.

  Martins


__________________________________
Do you Yahoo!?
Yahoo! Mail SpamGuard - Read only the mail you want.
http://antispam.yahoo.com/tools

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2004-02-24 20:44 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-02-23 23:51 libata/iswraid DMA Timeout jabbera
2004-02-24  4:00 ` Eric D. Mudama
2004-02-24 17:20 ` Martins Krikis
2004-02-24 20:32   ` jabbera
2004-02-24 20:43     ` Martins Krikis

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®