The fix qla2xxx: Add dev_loss_tmo_callbk/terminate_rport_io callback support owned by Seokmann Ju which is on top of 2.6.26.2 solve this issue. Thanks everybody. Gal. On Tue, 2008-12-09 at 14:31 +0300, Vladislav Bolkhovitin wrote: > Gal Rosen wrote: > > On Tue, 2008-12-09 at 14:04 +0300, Vladislav Bolkhovitin wrote: > >> Gal Rosen wrote: > >>> Ok, now the picture is more clear. > >>> It happened only with the virtual ports. > >>> With physical port and LUNs that defined with write cache enable rmmod > >>> succeeded, and on the messages file I see: > >>> [57972.480234] sd 40:0:0:1: [sdb] Synchronizing SCSI cache > >>> [57972.530263] sd 40:0:0:1: [sdb] Result: hostbyte=DID_NO_CONNECT > >>> driverbyte=DRIVER_OK,SUGGEST_OK > >>> > >>> While with virtual port I see only: > >>> [57972.480234] sd 40:0:0:1: [sdb] Synchronizing SCSI cache > >>> and never get the Result. > >>> > >>> Anyone has an idea? > >> I can guess that the remote device was disconnected before > >> SYNCHRONIZE_CACHE command was sent to it, so it either doesn't go to > >> anywhere, or goes to nowhere. Then SCSI mid-layer is waiting for it to > >> finish, maybe, infinitely (unlikely) or until some timeout. > >> > > That's right I see that blk_execute_rq() put the SCSI command > > SYNCHRONIZE_CACHE on queue and then wait_for_completion but the request > > never return. > > But why it is happened on the IBM with the LSI controller and not > > happening with CLARiiON ? > > Some race somewhere, I guess, when IBM/LSI are too fast to loose. > > > Gal. > >>> Thanks, > >>> Gal. > >>> > >>> On Mon, 2008-12-08 at 19:48 +0200, Gal Rosen wrote: > >>>> On Mon, 2008-12-08 at 09:23 -0800, Seokmann Ju wrote: > >>>>> On Dec 8, 2008, at 3:33 AM, Gal Rosen wrote: > >>>>> > >>>>>> Hi, > >>>>>> > >>>>>> I am working against IBM Total Storage DS4000 with LSI INF-01-00 Rev: > >>>>>> 0534 controller. First I don't understand why this controller > >>>>>> present 32 > >>>>>> access LUNs instead of one. > >>>>>> > >>>>>> But my main issue is when I create virtual port and then try to rmmod > >>>>>> the qlogic driver, it stuck. I must say that I made changes to the > >>>>>> qlogic driver which add possibility to create virtual port in FC-AL > >>>>>> topology. The base driver is from kernel 2.6.26.2. > >>>>>> > >>>>>> Here is a part of the messages file: > >>>>>> > >>>>>> [ 1814.053285] scsi(7): DPC handler exiting > >>>>> [snip] > >>>>>> host5 is the physical port, and host8 is the virtual port that created > >>>>>> on host5. > >>>>>> This issue does not occur against CLARiiON for instanse. > >>>>> As you may know, the NPIV feature implemented in such a way to work only > >>>>> in switched fabric topology. > >>>>> It requires certain switch with NPIV capable to be able to make the > >>>>> feature fully working. > >>>> Yes, I know that, but we use the Multi ID feature in Qlogic 24xx > >>>> firmware, to create the virtual port in LOOP topology. > >>>> > >>>> I found that if I configure the LUN in the array to write cache disable > >>>> this problem does not occur; does this tell something to anyone ? > >>>> I don't think that it related to the vport, it will happened also with > >>>> physical port. > >>>> > >>>> Gal. > >>>>> Thank you, > >>>>> Seokmann > >>>>> > >>>>>> Please help !!! > >>>>>> > >>>>>> Thanks, > >>>>>> Gal Rosen. > >>>>>> > >>>>>> > >>>>>> -- > >>>>>> To unsubscribe from this list: send the line "unsubscribe linux- > >>>>>> scsi" in > >>>>>> the body of a message to majordomo@xxxxxxxxxxxxxxx > >>>>>> More majordomo info at http://vger.kernel.org/majordomo-info.html > >>> -- > >>> To unsubscribe from this list: send the line "unsubscribe linux-scsi" in > >>> the body of a message to majordomo@xxxxxxxxxxxxxxx > >>> More majordomo info at http://vger.kernel.org/majordomo-info.html > >>> > > > -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
