On Tue, 1 Feb 2000, Manfred Spraul wrote:
> Linux Lists wrote:
> >
> > - Why does it happen only in 2.3.x kernels??
> * new network code [although that one was added in 2.3.41?]
> * new block cache
> * ...
>
> > - Why does it happen only when there are concurrent IDE accesses (i.e., if
> > the incoming file is directed to /dev/null, the problem does NOT
> > happen)??
> it's always possible that you trigger a bug in the filesystem code [ops.
> I forgot that the file system cache is new as well...]
Actually, wouldn't it be more of an IDE bug than a filesystem one
(considering that the problem doesn't happen in a SCSI-based system)??
Who should I contact in this case??
> > - Why does it happen only with concurrent _IDE_ accesses (i.e., it does
> > NOT happen when I use a SCSI HD instead, being everything else in the
> > system still the same)??
>
> Perhaps you use more than 7 kB stack? Perhaps the IDE driver needs more
> stack space than the SCSI HD?
I don't think I use more than 7kB stack ... actually, I don't think I use
even 1kB ... :)
> Or perhaps IRQ sharing, or just timing?
IRQ sharing is definitely _NOT_ the case (no boards / devices are sharing
IRQ's in the system). Timing is definitely an option (actually, it's my
best bet up to now).
> [i.e. the bottom half handler runs just after/before/during a certain
> hardware interrupt]?
There are no bottom half handlers in my driver, and the only bottom half
that is called from within it is NET_BH (which is called by a zillion
other drivers BTW ... ;). And NET_BH has to be running _after_ the
interrupt handler, because:
1. It's scheduled from within the interrupt handler;
2. The problem happens at an UP system;
> You could also try to use a debugger, or the ikd patches
>
> [ftp://e-mind.com/..., or perhaps
> ftp.xx.kernel.org/pub/linux/people/andrea]
Ok, I'll check that as well.
Thanks for all your help so far!!
Regards,
Ivan
-
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/
This archive was generated by hypermail 2b29 : Mon Feb 07 2000 - 21:00:07 EST