lkml.org 
[lkml]   [2000]   [Feb]   [16]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: [2.2.14] Deadlock in ide-tape?
On Tue, 15 Feb 2000, Steve Dodd wrote:

> On Mon, Feb 14, 2000 at 08:31:46PM +0200, Gadi Oxman wrote:
>
> > The trace below shows that the driver is waiting in an ioctl() for
> > a completion of a mt command.
>
> Ah, I did grovel through the source and come to that conclusion as well (I
> wasn't thinking very clearly yesterday <g>)
>
> > Which command were you running? After how much time from the
> > mt command was the panic triggered?
> >
> > If this was a retension command, for example, it can take up to 10
> > minutes in some tape drives. If the default timeout in ikd is lower
> > than that, it explains the panic.
>
> It was an MTEOM command, and without the deadlock detection it just blocks for
> ages (tens of minutes), with no tape drive activity; the problem seems to
> manifest itself when it's executed right after an MTWEOF (which is a bit of a
> silly thing to do, I know...). It's an HP Colorado 8Gb drive (i.e. 4Gb <g>) --
> is it possible that it's a bug / feature in the drive firmware?
>
> Apart from that, the driver seems to work great though..
>
> Cheers,
> Steve
>
> --
> Steve Dodd, Network Adminstrator, WotNot Ltd.

Perhaps it's a feature. In any case, the driver should not die there
no matter what. It does have a timeout, but it's currently set at two
hours, which might be a bit too long...

Try changing:

#define IDETAPE_DSC_MA_TIMEOUT 2*60*60*HZ /* 2 hours */
in drivers/block/ide-tape.c to 30 or 15 minutes, and check if the
driver recovers from it in your case (it should recover, but it has
been a long while since I last tested the timeout mechanism).

I don't know what's the best value for this timeout; there are some
tape drives which search for the end of media or filemarks linearly
and aren't smart enough to optimize it. 2 hours is probably a bit
too much, though.

Gadi


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:56    [from the cache]
©2003-2011 Jasper Spaans