Date: Sun, 5 Oct 2003 13:42:04 -0700 From: moron <moron@industrial.org> To: freebsd-stable@FreeBSD.org Subject: DMA related disc issues Message-ID: <200310051342.04864.moron@industrial.org>
next in thread | raw e-mail | index | archive | help
Howdy. I apologize for vagueness of this report but I am not immediately sure how to proceed with this or what details will be relevant. I look after a community server which has been running just fine for the last 6 months or so with no disk issues at all. The box has an all in one Shuttle motherboard which uses the VIA 8233 ATA133 disk controller according to what I see from dmesg. The main drive is a 30 gig Maxtor and I very recently added an 80 gig Seagate to the second IDE bus. There is also a CDROM sharing the primary IDE bus with the main drive. Up until quite recently the machine has had a single drive in it and been performing flawlessly, no disk problems or anything of that nature (nothing untoward in messages, not spontaneous reboots, etc.). I have performed a few CVSups and make worlds over the last month to deal with security issues and such and about two weeks ago I added the second Seagate drive to the system. Shortly after adding the new drive (but also around the time of a CVSup), the machine suddenly locked up. This was after about a day and a half of stable operation, with some serious disk I/O (busy mailing lists, a cpdup from one drive to the other, etc.). The lockup happened at a time where no obviously heavy disk I/O was happening, just general usage. I initially assumed that it was a loose cable and so opened the box up and re-seated the cables. After this the machine worked fine again for maybe 12 hours but then encountered the same type of errors which start with stuff along the lines of: Sep 26 05:59:35 www /kernel: ad0: READ command timeout tag=0 serv=0 - resetting Sep 26 05:59:35 www /kernel: ata0: resetting devices .. ad0: DMA limited to UDMA33, non-ATA66 cable or device After a few of these the machine stops being able to access its drives and eventually either reboots or hangs. After searching for a bit I saw that disabling DMA access altogether might help the situation out so I have done that at startup which now gives me the following initial complaints upon boot but then no further obvious problems other than sub-standard performance (which I do need to fix): Sep 27 13:02:19 www /kernel: ad0: READ command timeout tag=0 serv=0 - resetting Sep 27 13:02:20 www /kernel: ata0: resetting devices .. done Sep 27 13:02:51 www /kernel: ad0: WRITE command timeout tag=0 serv=0 - resetting Sep 27 13:02:51 www /kernel: ata0: resetting devices .. done I am currently running the FreeBSD 4.9-PRERELEASE that I CVSup'd earlier in the day on the 27th. Now I have not had any disk issues that I am aware of since going to PIO mode and the machine worked awesome up until around the addition of the second drive / OS update. I am wondering if there are any known issues that could be affecting DMA access with recent kernels or whether anyone has any other suggestions on what might be going wrong here? The fact that the machine runs fine without DMA, ran awesome up until recently and can perform intensive disk I/O for a period with no problems makes me think that this isn't a hardware problem but I am open to any suggestions that might help to nail down the culprit. If you need more specifics to offer suggestions please let me know. Cheers! -- ---> (culture) http://industrial.org : (label) http://deterrent.net ---> (community) http://ampfea.org : (hire me) http://codegrunt.com ---> (send EEEI news to) infosuck@industrial.org ---> Whomever dies with the most URLs wins!!!!!!!!!!!!!
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200310051342.04864.moron>