On Wed, Jun 01, 2005 at 11:48:20AM -0700, Andrew Vasquez wrote: > On Mon, 30 May 2005, Andrew Morton wrote: > > > Ake.Sandgren@xxxxxxxxxxxx (Ake) wrote: > > > > > > Hi! > > > I got the following when i pulled the FC cable from one of my qla2312 > > > cards. > > > This is a 2.6.12-rc5 kernel (with udm1 patches for 2.6.12-rc2) > > > > > > > > > May 30 14:12:45 ican-i kernel: [ 1109.864830] qla2300 0000:03:09.0: LOOP DOWN detected. > > > May 30 14:12:45 ican-i kernel: [ 1109.875948] Debug: sleeping function called from invalid context at include/linux/rwsem.h:43 > > > May 30 14:12:45 ican-i kernel: [ 1109.894492] in_atomic():1, irqs_disabled():1 > > > May 30 14:12:45 ican-i kernel: [ 1109.903864] [dump_stack+30/48] dump_stack+0x1e/0x30 > > > May 30 14:12:45 ican-i kernel: [ 1109.913648] [__might_sleep+167/176] __might_sleep+0xa7/0xb0 > > > May 30 14:12:45 ican-i kernel: [ 1109.924003] [device_for_each_child+35/144] device_for_each_child+0x23/0x90 > > > May 30 14:12:45 ican-i kernel: [ 1109.935865] [pg0+944279047/1069622272] scsi_target_block+0x57/0x60 [scsi_mod] > > > May 30 14:12:45 ican-i kernel: [ 1109.949142] [pg0+944996009/1069622272] fc_remote_port_block+0x39/0x60 [scsi_transport_fc] > ... > > > > This was reported (and anaylsed) a month ago: > > > > http://lkml.org/lkml/2005/4/29/82 > > > > I don't know if anything has been done about it yet though? > > Greg, > > Is there some other interrupt-safe alternative: > > http://lkml.org/lkml/2005/4/29/129 > > to using device_for_each_child()? The scsi_target_[block/unblock]() > functions need to iterate over all the child-scsi_device's belonging > to an fc_rport. Did you look at the code in the -mm tree? That's what's going into 2.6.13, so you might have some chance there. Otherwise you need to rewrite your logic to not do that :) Good luck, greg k-h - : 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
