On Tue, Jul 17, 2001 at 05:23:07PM +0200, Andrea Arcangeli wrote:
> [ sorry for the late reply ]
>
> On Thu, Jul 05, 2001 at 11:49:15AM +0100, Tim Waugh wrote:
> > On Tue, Jun 26, 2001 at 11:35:30PM +0300, Sergij Kolisnyk wrote:
> >
> > > Hello! The BUG: Author of patch in
> > > http://www.torque.net/parport/archive/current/2458.html have just
> > > confused max_segments with max_sectors. In result driver built
> > > clustered ATA requests exceeding the maximum of 256 sectors. My
> > > driver failed on that error, and I fixed this confusing (after
> > > reading 2.4. ll_rw_blk.c). (May be pd.c has some features, that
> > > prevent it failing on this bug, but I think such confuse should be
> > > fixed anyway ;-) )
> >
> > I don't really know that code well enough to comment. Andrea?
>
> the patch looks correct (maybe it is just included, I didn't checekd).
>
> It probably started to trigger around 2.4.2 because the otherwise
> default MAX_SECTORS grown. Infact I've no idea why it is been decreased
> to 256 in 2.4.2, I definitely overlooked that change.
btw, I'd also avoid to left a dangling pointer after the module is
deleted, this is an incremental fix:
--- paride/drivers/block/paride/pd.c.~1~ Mon Jul 23 03:26:53 2001
+++ paride/drivers/block/paride/pd.c Mon Jul 23 03:27:20 2001
@@ -599,6 +599,8 @@
for (unit=0;unit<PD_UNITS;unit++)
if (PD.present) pi_release(PI);
+
+ max_sectors[MAJOR_NR] = NULL;
}
#endif
>
> Andrea
>
> -- To unsubscribe, send mail to: linux-parport-request@torque.net --
> -- with the single word "unsubscribe" in the body of the message. --
Andrea
-- To unsubscribe, send mail to: linux-parport-request@torque.net --
-- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2b29 : Mon Jul 23 2001 - 02:27:35 EDT