Re: 2.1.94 still locks up hard after a few minutes
mlord (mlord@pobox.com)
Thu, 09 Apr 1998 22:10:10 -0400
Linus Torvalds wrote:
> On Thu, 9 Apr 1998, Adam J. Richter wrote:
> > In case there is a fix that you were trying in 2.1.94, I
> > I must report that this problem persists in 2.1.94. I can also
> > confirm that the problem occurs within about five minutes of
> > booting even if I do not start X windows. I hope this
> > data point is helpful.
>
> 2.1.94 does not even try to fix this problem: it definitely _is_ a real
> problem, and I can reproduce it myself but not nearly as easily as you
> seem to be able to (I have to run some fairly big things overnight, and
> then the machine tends to be dead in the morning).
>
> This bug is definitely due to the changes in device locking. The machine
> doesn't actually really die, and the system works fine _except_ that all
> IO to a certain controller will block forever (which, when the disk is
> your main filesystem, will obviously appear as a complete lockup).
In a separate message, I have just released a large IDE driver update
which fixes a number of problems I have noticed recently, and which
also tries (untested) to make the IDE driver more SMP-friendly.
While testing, I noticed that the block device request queue "plug"
does NOT always seem to have (rq_status == RQ_INACTIVE)..
Dunno where the problem is, but it sure confused my driver while
testing.
I now compare for (blk_dev[].current_request == blk_dev[].plug)
instead of (blk_dev[].current_request.rq_status == RQ_INACTIVE).
Works much better, is faster, but shouldn't have been needed
???
--
mlord@pobox.com
The Linux IDE guy
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu