Marc-Christian Petersen <m....@gmx.de> writes: > On Tue, 23 Nov 2010 15:25:44 +0100, Bjørn Mork <bj...@mork.no> wrote: > >> Please see if this is fixed by the patch I recently posted to the >> linux-scsi list: http://permalink.gmane.org/gmane.linux.ide/47847 >> >> I believe you are seeing another symptom of the same bug: The driver >> happily forwards a 0 length from userspace to dma_alloc_coherent(). >> smartctl will trigger this when trying to enable the self tests. >> >> I've attached the patch for simplicity. Note that I haven't yet >> received any ack on this, but it did fix the problems for the other user >> reporting similar issues. > > first: sorry for the late answer! > > That patch also fixes my problems. Very cool :-D
Does the selftest actually work? Artem Bokhan reports that the patch fixes the visible error, but that the selftest does not start. I really don't understand why, as AFAICS the patch does not change or modify any part of the command from smartctl. It only ensures that a command with dxfer_len == 0 doesn't cause an attempt to dma_alloc_coherent zero bytes (which will fail). > Many thanks! > > now, how do we get that patch into the official Debian Kernel? ;) I believe that is as easy as getting it into an upstream kernel, which will trigger stable, which will be included in Debian automatically. But I haven't received any feedback from the driver maintainers yet. I don't know if I should expect any... They seem to collect a number of changes in internal releases, which are then pushed upstream. Bjørn -- To UNSUBSCRIBE, email to debian-bugs-dist-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
