public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* dma problems with Serverworks CSB5 chipset
@ 2004-07-28 17:38 Richard Wohlstadter
  2004-08-03 18:08 ` Marcelo Tosatti
  0 siblings, 1 reply; 6+ messages in thread
From: Richard Wohlstadter @ 2004-07-28 17:38 UTC (permalink / raw)
  To: linux-kernel

Hello,

I have 200 servers in a cluster running vanilla kernel 2.4.26(not 
tainted).  Under heavy I/O activity I have various servers completely 
lose access to their IDE bus.  Logs show the same error every time:

hda: dma_timer_expiry: dma status == 0x61

The kernel resets the IDE bus at this point.  Sometimes things start 
working again but mostly all ide access is lost and I have to reboot the 
server.  The chipset is:

  00:0f.1 IDE interface: ServerWorks CSB5 IDE Controller (rev 93)

I have searched archives for problems with this chipset and I have seen 
other users with this same issue, but no resolution to the problem.  Is 
there a known problem with this chipset version or could there be some 
issues still with the serverworks driver?  Any help would be much 
appreciated.  Thanks.

Rich Wohlstadter
modi@swbell.net

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: dma problems with Serverworks CSB5 chipset
  2004-08-03 18:08 ` Marcelo Tosatti
@ 2004-08-03 17:31   ` Alan Cox
  2004-08-03 18:47     ` Rich Wohlstadter
  2004-08-08 18:09     ` Torbjörn Olander
  0 siblings, 2 replies; 6+ messages in thread
From: Alan Cox @ 2004-08-03 17:31 UTC (permalink / raw)
  To: Marcelo Tosatti
  Cc: Richard Wohlstadter, Linux Kernel Mailing List,
	Bartlomiej Zolnierkiewicz

On Maw, 2004-08-03 at 19:08, Marcelo Tosatti wrote:
> ServerWorks OSB4/5 chipsets are known to not work reliably with the Linux
> IDE code. AFAIK its a hardware problem which we dont correctly work around.
> 
> Have you tried disabling DMA?
> 
> Bart and Alan are IDE experts, they can probably give you more useful
> information.

CSB5 is reliable, rock solidly so in my experience. OSB4 was the older
interface with problems. Are these systems SMP, what disks are you using
and in what IDE mode ?


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: dma problems with Serverworks CSB5 chipset
  2004-07-28 17:38 dma problems with Serverworks CSB5 chipset Richard Wohlstadter
@ 2004-08-03 18:08 ` Marcelo Tosatti
  2004-08-03 17:31   ` Alan Cox
  0 siblings, 1 reply; 6+ messages in thread
From: Marcelo Tosatti @ 2004-08-03 18:08 UTC (permalink / raw)
  To: Richard Wohlstadter; +Cc: linux-kernel, Bartlomiej Zolnierkiewicz, Alan Cox

On Wed, Jul 28, 2004 at 12:38:59PM -0500, Richard Wohlstadter wrote:
> Hello,
> 
> I have 200 servers in a cluster running vanilla kernel 2.4.26(not 
> tainted).  Under heavy I/O activity I have various servers completely 
> lose access to their IDE bus.  Logs show the same error every time:
> 
> hda: dma_timer_expiry: dma status == 0x61
> 
> The kernel resets the IDE bus at this point.  Sometimes things start 
> working again but mostly all ide access is lost and I have to reboot the 
> server.  The chipset is:
> 
>  00:0f.1 IDE interface: ServerWorks CSB5 IDE Controller (rev 93)
> 
> I have searched archives for problems with this chipset and I have seen 
> other users with this same issue, but no resolution to the problem.  Is 
> there a known problem with this chipset version or could there be some 
> issues still with the serverworks driver?  Any help would be much 
> appreciated.  Thanks.

Richard,

ServerWorks OSB4/5 chipsets are known to not work reliably with the Linux
IDE code. AFAIK its a hardware problem which we dont correctly work around.

Have you tried disabling DMA?

Bart and Alan are IDE experts, they can probably give you more useful
information.

^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: dma problems with Serverworks CSB5 chipset
  2004-08-03 17:31   ` Alan Cox
@ 2004-08-03 18:47     ` Rich Wohlstadter
  2004-08-08 18:09     ` Torbjörn Olander
  1 sibling, 0 replies; 6+ messages in thread
From: Rich Wohlstadter @ 2004-08-03 18:47 UTC (permalink / raw)
  To: Alan Cox
  Cc: Marcelo Tosatti, Linux Kernel Mailing List,
	Bartlomiej Zolnierkiewicz

> On Maw, 2004-08-03 at 19:08, Marcelo Tosatti wrote:
>> ServerWorks OSB4/5 chipsets are known to not work reliably with the
>> Linux
>> IDE code. AFAIK its a hardware problem which we dont correctly work
>> around.
>>
>> Have you tried disabling DMA?
>>
>> Bart and Alan are IDE experts, they can probably give you more useful
>> information.
>
> CSB5 is reliable, rock solidly so in my experience. OSB4 was the older
> interface with problems. Are these systems SMP, what disks are you using
> and in what IDE mode ?
>

Yep, they are SMP(IBM blades with 2 Xeon 2.4Ghz).  The blades use 1
little 40g laptop drive( TOSHIBA MK4019GAXB ).  The drive is running UDMA
mode 4.  Here is the output of /proc/ide/svwks:

                             ServerWorks OSB4/CSB5/CSB6

                            ServerWorks CSB5 Chipset (rev 93)
------------------------------- General Status
---------------------------------
--------------- Primary Channel ---------------- Secondary Channel
-------------
                 enabled                          enabled
--------------- drive0 --------- drive1 -------- drive0 ---------- drive1
------
DMA enabled:    yes              yes             no                no
UDMA enabled:   yes              yes             no                no
UDMA enabled:   4                4               0                 0
DMA enabled:    2                2               2                 2
PIO  enabled:   4                4               ?                 ?

Let me know if you need any other specifics and thanks in advance for any
advice.

Rich


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: dma problems with Serverworks CSB5 chipset
  2004-08-03 17:31   ` Alan Cox
  2004-08-03 18:47     ` Rich Wohlstadter
@ 2004-08-08 18:09     ` Torbjörn Olander
  2004-08-23 18:59       ` Torbjörn Olander
  1 sibling, 1 reply; 6+ messages in thread
From: Torbjörn Olander @ 2004-08-08 18:09 UTC (permalink / raw)
  To: Alan Cox; +Cc: linux-kernel

I'm having the same problems here with linux 2.6.8-rc2 kernel here with
an Intel SAI2 dual SMP P3 motherboard with Serverworks Serverset III LE
chipset and CSB5 southbridge.

I run one WDC WD400BB-00DEA0 as master on each channel with SW RAID1 for
system. To even get DMA working I had to add
append="ide0=ata66 ide1=ata66"
in my lilo config. The system works fine with low load on the disks, but
if I copy greater amounts of data from a local SW RAID5 array I get the
same DMA timeout error and often it locks up completely.

If i disable DMA with hdparm the system works fine.

/proc/ide/svwks output:
(DMA disabled with hdparm)

                             ServerWorks OSB4/CSB5/CSB6

                            ServerWorks CSB5 Chipset (rev 93)
------------------------------- General Status
---------------------------------
--------------- Primary Channel ---------------- Secondary Channel
-------------
                 enabled                          enabled
--------------- drive0 --------- drive1 -------- drive0 ----------
drive1 ------
DMA enabled:    no               no              no                no 
UDMA enabled:   yes              no              yes               no 
UDMA enabled:   2                0               2                 0
DMA enabled:    2                2               2                 2
PIO  enabled:   4                0               4                 0



On Tue, 2004-08-03 at 19:31, Alan Cox wrote:
> On Maw, 2004-08-03 at 19:08, Marcelo Tosatti wrote:
> > ServerWorks OSB4/5 chipsets are known to not work reliably with the Linux
> > IDE code. AFAIK its a hardware problem which we dont correctly work around.
> > 
> > Have you tried disabling DMA?
> > 
> > Bart and Alan are IDE experts, they can probably give you more useful
> > information.
> 
> CSB5 is reliable, rock solidly so in my experience. OSB4 was the older
> interface with problems. Are these systems SMP, what disks are you using
> and in what IDE mode ?
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
-- 
Med vänliga hälsningar / Best regards
Torbjörn Olander
tosse@wlug.westbo.se


^ permalink raw reply	[flat|nested] 6+ messages in thread

* Re: dma problems with Serverworks CSB5 chipset
  2004-08-08 18:09     ` Torbjörn Olander
@ 2004-08-23 18:59       ` Torbjörn Olander
  0 siblings, 0 replies; 6+ messages in thread
From: Torbjörn Olander @ 2004-08-23 18:59 UTC (permalink / raw)
  To: linux-kernel

Just want to report that after changing the disks to two Maxtor 6Y160P0
drives it works fine so the problem must be something in the combination
of CSB5 + WD400BB.

On Sun, 2004-08-08 at 20:09, Torbjörn Olander wrote:
> I'm having the same problems here with linux 2.6.8-rc2 kernel here with
> an Intel SAI2 dual SMP P3 motherboard with Serverworks Serverset III LE
> chipset and CSB5 southbridge.
> 
> I run one WDC WD400BB-00DEA0 as master on each channel with SW RAID1 for
> system. To even get DMA working I had to add
> append="ide0=ata66 ide1=ata66"
> in my lilo config. The system works fine with low load on the disks, but
> if I copy greater amounts of data from a local SW RAID5 array I get the
> same DMA timeout error and often it locks up completely.
> 
> If i disable DMA with hdparm the system works fine.
> 
> /proc/ide/svwks output:
> (DMA disabled with hdparm)
> 
>                              ServerWorks OSB4/CSB5/CSB6
> 
>                             ServerWorks CSB5 Chipset (rev 93)
> ------------------------------- General Status
> ---------------------------------
> --------------- Primary Channel ---------------- Secondary Channel
> -------------
>                  enabled                          enabled
> --------------- drive0 --------- drive1 -------- drive0 ----------
> drive1 ------
> DMA enabled:    no               no              no                no 
> UDMA enabled:   yes              no              yes               no 
> UDMA enabled:   2                0               2                 0
> DMA enabled:    2                2               2                 2
> PIO  enabled:   4                0               4                 0
> 
> 
> 
> On Tue, 2004-08-03 at 19:31, Alan Cox wrote:
> > On Maw, 2004-08-03 at 19:08, Marcelo Tosatti wrote:
> > > ServerWorks OSB4/5 chipsets are known to not work reliably with the Linux
> > > IDE code. AFAIK its a hardware problem which we dont correctly work around.
> > > 
> > > Have you tried disabling DMA?
> > > 
> > > Bart and Alan are IDE experts, they can probably give you more useful
> > > information.
> > 
> > CSB5 is reliable, rock solidly so in my experience. OSB4 was the older
> > interface with problems. Are these systems SMP, what disks are you using
> > and in what IDE mode ?
> > 
> > -
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@vger.kernel.org
> > More majordomo info at  http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at  http://www.tux.org/lkml/
-- 
Med vänliga hälsningar / Best regards
Torbjörn Olander
tosse@wlug.westbo.se


^ permalink raw reply	[flat|nested] 6+ messages in thread

end of thread, other threads:[~2004-08-23 20:12 UTC | newest]

Thread overview: 6+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-07-28 17:38 dma problems with Serverworks CSB5 chipset Richard Wohlstadter
2004-08-03 18:08 ` Marcelo Tosatti
2004-08-03 17:31   ` Alan Cox
2004-08-03 18:47     ` Rich Wohlstadter
2004-08-08 18:09     ` Torbjörn Olander
2004-08-23 18:59       ` Torbjörn Olander

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