Sir, This is from Areca researching and developing SAS/SATA RAID adapter, http://www.areca.com.tw/. Currently, our customers report an issue which happened on the xfs filesystem with dm-crypt. We have verified it and found it does exist. We found, 1. XFS with Areca works fine, 2. EXT3 with Areca works fine, 3. dm-crypt with EXT3 over Areca works fine. But once the configuration is xfs filesystem plus dm-crypt, the scsi commands aborted happened. We find that even though the driver claims the maximum SG entry number, ARCMSR_MAX_SG_ENTRIES, to the upper layer. But we still got the scsi commands with sg number over our limit. That is why arcmsr, which is Areca RAID driver, would issue "abort scsi command" messages constantly. The process is as below, 1. cryptsetup luksFormat /dev/sdb 2. cryptsetup luksOpen /dev/sdb export 3. mkfs -t xfs /dev/mapper/export 4. mount /dev/mapper/export /export 5. then start copying data to /export. Shortly the process hangs with messages of arcmsr "abort device command of scsi". Would you please check it out? Thank you, P.S. The following is one of our customers' report. ----- Original Message ----- From: "Dr. Uwe Meyer-Gruhl" <fragen@xxxxxxxxxxx> To: "Erich Chen" <erich@xxxxxxxxxxxx> Sent: Tuesday, November 13, 2007 8:33 AM Subject: Areca controller problem with Linux XFS over DM-CRYPT > Hi, > > > since you are the maintainer of the ARCMSR driver, maybe you can tell what > the problem is here. I am using an ARC 1261ML under SuSE Linux 10.3 > (kernel 2.6.22) on a dual-core machine. The harddisks are configured as > RAID6 as /dev/sdb. > > When I do: > > cryptsetup luksFormat /dev/sdb > cryptsetup luksOpen /dev/sdb export > mkfs -t xfs /dev/mapper/export > mount /dev/mapper/export /export > > And then I start copying to the mountpoint /export. Shortly thereafter, > the process hangs and I see this on the console: > > Nov 12 23:35:08 blob kernel: arcmsr6: abort device command of scsi id = 0 > lun = 1 > Nov 12 23:35:42 blob syslog-ng[2628]: last message repeated 15 times > > This goes on forever, I have to reboot the machine. The web interface of > the controller says everything is fine. > > The same thing works with ext3 instead of xfs and of course when I do not > use dm-crypt but /dev/sdb directly. > > I have read that barrier support seems to be somewhat crucial for xfs > (http://www.linux-magazin.de/heft_abo/sonderheft/2006/04/beschraenktes_schre iben/(offset)/4), > which seems unsupported - at least over dm-crypt: > > Nov 12 23:20:37 blob kernel: Filesystem "dm-2": Disabling barriers, not > supported by the underlying device > Nov 12 23:20:37 blob kernel: XFS mounting filesystem dm-2 > Nov 12 23:20:37 blob kernel: Ending clean XFS mount for filesystem: dm-2 > > > I tried to disable write cache on the harddisks and TCQ support for the > volume, alas, both to no avail. > > Also, I am not alone with that problem: > http://forums.storagereview.net/index.php?showtopic=25513 > > I think this is a crucial question since ext3 features only 16 TByte for a > filesystem and this volume can be reached by Areca controllers quite > easily nowadays... and I do not want to give up encryption as well. > > > Do you have any ideas or solution for this? > > > -- > Dr. Uwe Meyer-Gruhl (Geschaftsfuhrer) > Tel.: +49 89 82908139, Fax.: +49 89 8205273, mg@xxxxxxxxxxx > > congenio Gesellschaft fur innovative Softwareprojekte mbH > Herzog-Johann-Str. 22, D-81245 Munchen, GERMANY > HRB-Nummer 118403 (Amtsgericht Munchen), http://www.congenio.de > --------------------------------------------------------------------- dm-crypt mailing list - http://www.saout.de/misc/dm-crypt/ To unsubscribe, e-mail: dm-crypt-unsubscribe@xxxxxxxx For additional commands, e-mail: dm-crypt-help@xxxxxxxx
