Re: [Smartmontools-support]Hard lockup for smartctl and Promise 20265(+ide-taskfile?)

[Georg A. <ac...@in...>]

On Thu, Dec 26, 2002 at 12:14:35AM -0600, Bruce Allen wrote:
> Hi Greg,

It's "Georg" ;-)

> Thanks for your note. I'm sorry to hear that you are having problems with
> smartmontools.  I'm adding this to the WARNINGS file -- if it's resolved
> then I'll remove it.
> 
> Before I act further on this, I need a bit of help to understand what's
> going on.
> 
> > I have a reproduceable hard lockup when using smartctl on a Promise
> > 20265 IDE-controller (pdc202xx-driver, no RAID, just 2 separate
> > drives).
> 
> OK, this is just a 2-drive IDE controller PCI card, right?

It's a Asus K7V mainboard with this controller on-board, the regular chipset
controller is a VIA 82C686. The Promise chip has two additional IDE-ports
(Ultra-whatever), but I'm using only one for the two drives. In total there
are 6 IDE-drives installed.
 
> > It never happend with 2.4.18 or earlier, but always appears with
> > 2.4.19 (and the patch for drives >120GB) or 2.4.20 (with enabled
> > ide-taskfile). So maybe it has something to do with the taskfile...
> > BTW: I use the kernel from the XFS-CVS, but I don't think that it has
> > anything to do with the FS.
> 
> So this card requires its own ide driver (pdc202xx-driver)?  And you're
> building it with ide-taskfile enabled?  It would help if you could be
> specific about any IDE-related config options you are using.

Yes, the pdc202xx-driver is needed. Here's are the selected IDE-options (for
2.4.20):

#define CONFIG_BLK_DEV_IDE 1
#define CONFIG_BLK_DEV_IDEDISK 1
#define CONFIG_IDEDISK_MULTI_MODE 1
#define CONFIG_IDEDISK_STROKE 1
#define CONFIG_BLK_DEV_IDECD 1
#define CONFIG_IDE_TASK_IOCTL 1
#define CONFIG_BLK_DEV_IDEPCI 1
#define CONFIG_IDEPCI_SHARE_IRQ 1
#define CONFIG_BLK_DEV_IDEDMA_PCI 1
#define CONFIG_IDEDMA_PCI_AUTO 1
#define CONFIG_BLK_DEV_IDEDMA 1
#define CONFIG_BLK_DEV_ADMA 1
#define CONFIG_BLK_DEV_PDC202XX 1
#define CONFIG_PDC202XX_BURST 1
#define CONFIG_BLK_DEV_VIA82CXXX 1
#define CONFIG_IDEDMA_AUTO 1
#define CONFIG_BLK_DEV_IDE_MODES 1

> > It happens only for /dev/hde and /dev/hdf, hd[a-d] have no problems
> > with smartctl (V2.1).
> 
> OK, two questions here.  First, when you say it only happens for hde and
> hdf, are these the two disks that are controlled by the pdc202xx-driver,
> whereas hd[a-d] are controlled by the normal ide driver?  Or are hd[a-d]
> also controlled by the pdc202xx-driver?

Yes, only hd[e-f] are controlled by pdc202xx, hd[a-d] are handled by the
normal/VIA-driver.

> Second question: I don't know what is meant by smartctl (V2.1).  Could you
> please send the output of smartctl -V to the mailing list so I can see
> what version you are using?

braindead5:/root> smartctl -V
smartctl version 2.1 - S.M.A.R.T. Control Program
<...>
 
> > The connected disks are Maxtor 4K080H4&98196H8 (each about 80GB).
> > 
> > The lockup happens for all commands I have tried (smartctrl -e /dev/hde is
> > the easiest way).
> > 
> > There is _NO OOPS_, even kdb and alt-sysreq don't work anymore (a rarely
> > seen situation...), but the IDE-LED is on, so the request seems to be
> > started (at least ;-)
> > 
> > There are no other problems (disk access, DMA, etc.) for the drives
> > connected to the Promise IDE-port, it happens _only_ with smartctl.
> 
> I suspect it is related to the ide-taskfile support.  Though I'll have to
> see the output of smartctl -V to see if this is even used in the smartctl
> -e command.
>
> > I don't know if this is the right place to ask, but maybe you know a better
> > one :-)
> 
> The only other place to inquire is with the pdc202xx-driver
> developers/maintainers.  The fact that it works in 2.4.18 and broke in
> 2.4.19 suggests that something may have changed in the ide-taskfile
> support. Have you asked?

Will do... 

-- 
         Georg Acher, ac...@in...         
         http://wwwbode.in.tum.de/~acher
         "Oh no, not again !" The bowl of petunias