CC'ing linux-scsi. Am Freitag, den 12.09.2008, 09:47 -0400 schrieb Konrad Rzeszutek: > On Fri, Sep 12, 2008 at 07:38:26AM +0200, Arne Redlich wrote: > > Am Donnerstag, den 11.09.2008, 22:08 -0400 schrieb Konrad Rzeszutek: > > > Allow more than 255 LUNs to be reported. The logic would > > > make LUNs above 255 start at the 16k number. > > > > > > target_disk.c | 3 +-- > > > 1 file changed, 1 insertion(+), 2 deletions(-) > > > > > > Signed-off-by: Konrad Rzeszutek <ketuzsezr@xxxxxxxxxx> > > > > > > Index: kernel/target_disk.c > > > =================================================================== > > > --- kernel/target_disk.c (revision 171) > > > +++ kernel/target_disk.c (working copy) > > > @@ -279,8 +279,7 @@ > > > if (lun->l_state != IDEV_RUNNING) > > > continue; > > > > > > - *data++ = cpu_to_be32((0x3ff & lun->lun) << 16 | > > > - ((lun->lun > 0xff) ? (0x1 << 30) : 0)); > > > + *data++ = cpu_to_be32((0x3ff & lun->lun) << 16); > > > > Thanks, but I'm afraid this patch isn't correct. > > > > Did you actually see this behaviour or did you stumble across it while > > > iscsiadm -m discovery -t st -p 192.168.100.1 > 192.168.100.1:3260,1 iqn.2009.com.test:konrad > > 09:43:15 # 7 :/etc/iscsi/ > > iscsiadm -m node -L all > Logging in to [iface: default, target: iqn.2009.com.test:konrad, portal: > 192.168.100.1,3260] > Login to [iface: default, target: iqn.2009.com.test:konrad, portal: > 192.168.100.1,3260]: successful > > 09:43:20 # 8 :/etc/iscsi/ > > lsscsi > [1:0:0:0] disk MAXTOR ATLAS10K5_73SCA JNZH /dev/sda > [1:0:6:0] process ESG-SHV SCA HSBP M35 1.06 - > [4:0:6:0] process ESG-SHV SCA HSBP M35 1.06 - > [4:0:10:0] disk MAXTOR ATLAS10K5_73SCA JNZH /dev/sdb > [8:0:0:0] disk IET VIRTUAL-DISK 0 /dev/sdc > [8:0:0:255] disk IET VIRTUAL-DISK 0 /dev/sdd > > 09:43:22 # 9 :/etc/iscsi/ > > /root/sg_luns /dev/sdc > Lun list length = 48 which imples 6 lun entries > Report luns [select_report=0]: > 0000000000000000 > 00ff000000000000 > 4100000000000000 > 4200000000000000 > 4201000000000000 > 4300000000000000 Yes, this is absolutely ok - the highest two bits indicate the address method. Looking at the kernel's scsi_scan.c, scsilun_to_int()'s documentation mentions this as well, but the code doesn't take any measures to handle it. Syncing the code with the comment would be trivial, but then it still would break on the inverse operation or extended logical unit addressing formats. So I guess the issue needs to be fixed differently - in particular the conversion to unsigned int when using extended addressing formats with more than 4 bytes. Is this already on someone's TODO list? Cheers, Arne > 09:43:34 # 10 :/etc/iscsi/ > > dmesg | tail > sdd: Write Protect is off > sdd: Mode Sense: 77 00 00 08 > SCSI device sdd: drive cache: write through > sdd: unknown partition table > sd 8:0:0:255: Attached scsi disk sdd > sd 8:0:0:255: Attached scsi generic sg4 type 0 > scsi: host 8 channel 0 id 0 lun16640 has a LUN larger than allowed by > the host adapter > scsi: host 8 channel 0 id 0 lun16896 has a LUN larger than allowed by > the host adapter > scsi: host 8 channel 0 id 0 lun16897 has a LUN larger than allowed by > the host adapter > scsi: host 8 channel 0 id 0 lun17152 has a LUN larger than allowed by > the host adapter > > And on the 192.168.100.1: > > Target iqn.2009.com.test:konrad > Lun 0 Path=/dev/VolGroupData/a000,Type=fileio > Lun 255 Path=/dev/VolGroupData/a255,Type=fileio > Lun 256 Path=/dev/VolGroupData/a256,Type=fileio > Lun 512 Path=/dev/VolGroupData_1/b256,Type=fileio > Lun 513 Path=/dev/VolGroupData_2/c01,Type=fileio > Lun 768 Path=/dev/VolGroupData_2/c256,Type=fileio -- 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
