On Wed, 2011-10-05 at 14:23 +0200, Paolo Bonzini wrote: > When I do this: > Hey Paolo, > modprobe tcm_loop > mount -t configfs configfs /sys/kernel/config/ > > cd /sys/kernel/config/target/core > mkdir fileio_0 > mkdir fileio_0/fileio > echo 'fd_dev_name='$HOME/test2.img',fd_dev_size=5368709120' > fileio_0/fileio/control > echo 1 > fileio_0/fileio/enable > > cd /sys/kernel/config/target > mkdir -p loopback/naa.600140554cf3a18e/tpgt_1 > mkdir -p loopback/naa.600140554cf3a18e/tpgt_1/lun/lun_1 > cd loopback/naa.600140554cf3a18e/tpgt_1 > # "Obvious" error here: need to set up nexus before virtual_scsi_port! > ln -sf ../../../core/fileio_0/fileio lun/lun_1/virtual_scsi_port > echo naa.60014053226f0388 > nexus > echo '- - -' > /sys/bus/tcm_loop_bus/drivers/tcm_loop/tcm_loop_adapter_0/host*/scsi_host/host*/scan > > ... I get the following two OOPSes: > > [ 264.135309] BUG: unable to handle kernel NULL pointer dereference at 0000000000000090 > [ 264.136279] IP: [<ffffffffa0510498>] transport_processing_thread+0x61b/0xc8b [target_core_mod] <SNIP> > > Reproduced with both 3.0.3 and 3.1.0-rc8. > I've been able to track this down to an issue with the explict rescan sending down scsi_cmnd->device->id == 0 values referencing an tcm_loop_tpg endpoint from tcm_loop_hba->tl_hba_tpgs[] within tcm_loop_queuecommand() that has not been configured. The issue is not directly related to the ordering of the nexus vs. Port/LUNs, and (unfortuately) can be triggered directly using the explict rescan. Anyways, I'm sending out a bugfix patch shortly, and will make sure this is included in the v3.2-rc1 target patches w/ a Cc to stable@xxxxxxxxxx to ensure it's picked up by v3.1.1 and v3.0.5 Thanks alot for reporting! --nab -- 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
