On Sat, 19 Feb 2005, Samuel Colin wrote: > > This looks like another device problem (not a SCSI problem). > > > Do you mean it's the usb routines of the kernel which seem not to be well > supported by the via usb2 chipset ?
I mean the disk or its USB interface is misbehaving, even though the kernel and drivers all seem to be acting correctly. > What surprises me, in a sense, is that the error is very very particular and > specific, i.e. : operations with the other disk (from the same constructor) > doesn't cause any problem (copy from and to, fsck), except for the scsi > firmware (solved, thanks to you), a reading/writing to lacieroot is OK, and > a fsck on this particular disk, with the particular via usb2 chipset causes > hangs. I feel, er, confused, to say the least. Who knows? Maybe it has to do with what portion of the disk is being used. Maybe it's some very sensitive timing issue that's highly hardware-dependent. > > I noticed > > that "delay.patch" in your patches directory is meant for 2.6.10; did > > you make the corresponding change for 2.6.11 by hand? And is the udelay > > length set to 110 instead of 100? If either of those isn't true, it's the > > first thing you should try. > > > Yes, the udelay was 110 and I had commented the "if". So this trick didn't > solve the problem. I wonder if increasing it more, say to 200, would have an effect? Perhaps I'm fixated on that delay, but it's the only factor I've ever seen that was related to the kind of problem you have. > > As a partial workaround, there is an experimental patch below. It was > > written a couple of months ago, but it should still apply with some > > offsets to your kernel. It won't prevent the errors from occurring, but > > it ought to do a better job of recovering from them. (Technically, it > > tries a USB port reset before doing a class-specific device reset.) > > > I patched, I compiled, I installed, it worked. > I did several fsck (fsck.ext3 -C 0 -f -p -t -t /dev/sda?) on the disk, I > saw of course several usb resets, but fsck did not complain and did its job. > So I think the problem is (with this workaround) solved. I will put the > machine in a more "production-like" state to see how all of this evolves. > By the way, the patch applies with no offset at all (on a 2.6.11-rc4-bk6). That's good to know. I'll submit a version of that patch, after some other changes already in the pipeline have been accepted. Alan Stern ------------------------------------------------------- SF email is sponsored by - The IT Product Guide Read honest & candid reviews on hundreds of IT Products from real users. Discover which products truly live up to the hype. Start reading now. http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click _______________________________________________ linux-usb-devel@lists.sourceforge.net To unsubscribe, use the last form field at: https://lists.sourceforge.net/lists/listinfo/linux-usb-devel
