From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Seymour, Shane M" Subject: RE: What partition should the MTMKPART argument specify? Was: Re: st driver doesn't seem to grok LTO partitioning Date: Thu, 4 Feb 2016 01:43:24 +0000 Message-ID: References: <20151218170644.24167419@harpe.intellique.com> <688812570.3359420.1452030904683.JavaMail.zimbra@redhat.com> <20160106161049.34f29973@harpe.intellique.com> <1993563395.3556606.1452093814452.JavaMail.zimbra@redhat.com> <20160106170720.61db9e86@harpe.intellique.com> <1371374871.6203837.1452802373256.JavaMail.zimbra@redhat.com> <0840949A-970D-43FB-A691-E2F8AC2A0804@kolumbus.fi> <6685EAA5-B880-40FB-951D-BA94883049B3@kolumbus.fi> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 8BIT Return-path: Received: from g9t5009.houston.hp.com ([15.240.92.67]:46849 "EHLO g9t5009.houston.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S965028AbcBDBnr convert rfc822-to-8bit (ORCPT ); Wed, 3 Feb 2016 20:43:47 -0500 In-Reply-To: Content-Language: en-US Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Kai Makisara Cc: Laurence Oberman , Emmanuel Florac , Laurence Oberman , "linux-scsi@vger.kernel.org" Hi Kai, Tested with patched kernel 4.5.0-rc2-next-20160202+. It's looking good everything partition related passed with DDS5 and LTO6. You can definitely add me as a tested-by. I did find one issue below but it's not related to the partitioning changes. # ./mt -f /dev/st0 stsetoption debug # ./mt -f /dev/st0 stsetoption can-partitions # ./mt -f /dev/st0 mkpartition 10000 [ 1871.160395] st 3:0:0:0: [st0] Block limits 1 - 16777215 bytes. [ 1871.160955] st 3:0:0:0: [st0] Mode sense. Length 11, medium 0, WBS 10, BLL 8 [ 1871.160958] st 3:0:0:0: [st0] Density 5a, tape length: 0, drv buffer: 1 [ 1871.160962] st 3:0:0:0: [st0] Block size: 0, buffer size: 4096 (1 blocks). [ 1871.160963] st 3:0:0:0: [st0] Updating partition number in status. [ 1871.161915] st 3:0:0:0: [st0] Got tape pos. blk 0 part 0. [ 1871.161942] st 3:0:0:0: [st0] Loading tape. [ 1871.176784] st 3:0:0:0: [st0] Block limits 1 - 16777215 bytes. [ 1871.177284] st 3:0:0:0: [st0] Mode sense. Length 11, medium 0, WBS 10, BLL 8 [ 1871.177289] st 3:0:0:0: [st0] Density 5a, tape length: 0, drv buffer: 1 [ 1871.177293] st 3:0:0:0: [st0] Block size: 0, buffer size: 4096 (1 blocks). [ 1871.178632] st 3:0:0:0: [st0] Partition page length is 16 bytes. [ 1871.178637] st 3:0:0:0: [st0] PP: max 3, add 1, xdp 1, psum 03, pofmetc 4, rec 03, units 09, sizes: 38 2543 [ 1871.178642] st 3:0:0:0: [st0] MP: 11 0e 03 01 3c 03 09 00 00 26 09 ef [ 1871.178645] st 3:0:0:0: [st0] psd_cnt 2, max.parts 3, nbr_parts 1 [ 1871.178648] st 3:0:0:0: [st0] Formatting tape with two partitions (1 = 10000 MB). [ 1871.178651] st 3:0:0:0: [st0] Sent partition page length is 12 bytes. needs_format: 1 [ 1871.178655] st 3:0:0:0: [st0] PP: max 3, add 1, xdp 1, psum 03, pofmetc 4, rec 03, units 09, sizes: 65535 10 [ 1871.178659] st 3:0:0:0: [st0] MP: 11 0a 03 01 3c 03 09 00 ff ff 00 0a [ 1871.179655] st 3:0:0:0: [st0] Sending FORMAT MEDIUM [ 1883.702742] st 3:0:0:0: [st0] Rewinding tape. And we can change to partition 1 and back again. I ran this (I used the tell option but that failed more on that below): # ./mt -f /dev/st0 setpartition 1 # ./mt -f /dev/st0 status SCSI 2 tape drive: File number=0, block number=0, partition=1. Tape block size 0 bytes. Density code 0x5a (no translation). Soft error count since last status=0 General status bits on (41010000): BOT ONLINE IM_REP_EN Then wrote a tar archive to the drive and changed back to partition 0 and confirmed that I couldn't read the archive. # ./mt -f /dev/st0 mkpartition -10000 [ 3813.227898] st 3:0:0:0: [st0] Block limits 1 - 16777215 bytes. [ 3813.228399] st 3:0:0:0: [st0] Mode sense. Length 11, medium 0, WBS 10, BLL 8 [ 3813.228404] st 3:0:0:0: [st0] Density 5a, tape length: 0, drv buffer: 1 [ 3813.228408] st 3:0:0:0: [st0] Block size: 0, buffer size: 4096 (1 blocks). [ 3813.228427] st 3:0:0:0: [st0] Loading tape. [ 3813.242855] st 3:0:0:0: [st0] Block limits 1 - 16777215 bytes. [ 3813.243304] st 3:0:0:0: [st0] Mode sense. Length 11, medium 0, WBS 10, BLL 8 [ 3813.243310] st 3:0:0:0: [st0] Density 5a, tape length: 0, drv buffer: 1 [ 3813.243313] st 3:0:0:0: [st0] Block size: 0, buffer size: 4096 (1 blocks). [ 3813.244552] st 3:0:0:0: [st0] Partition page length is 16 bytes. [ 3813.244559] st 3:0:0:0: [st0] PP: max 3, add 1, xdp 1, psum 03, pofmetc 4, rec 03, units 09, sizes: 2543 38 [ 3813.244564] st 3:0:0:0: [st0] MP: 11 0e 03 01 3c 03 09 00 09 ef 00 26 [ 3813.244567] st 3:0:0:0: [st0] psd_cnt 2, max.parts 3, nbr_parts 1 [ 3813.244570] st 3:0:0:0: [st0] Formatting tape with two partitions (0 = 10000 MB). [ 3813.244573] st 3:0:0:0: [st0] Sent partition page length is 12 bytes. needs_format: 1 [ 3813.244578] st 3:0:0:0: [st0] PP: max 3, add 1, xdp 1, psum 03, pofmetc 4, rec 03, units 09, sizes: 10 65535 [ 3813.244582] st 3:0:0:0: [st0] MP: 11 0a 03 01 3c 03 09 00 00 0a ff ff [ 3813.245907] st 3:0:0:0: [st0] Sending FORMAT MEDIUM [ 3821.916760] st 3:0:0:0: [st0] Rewinding tape. That worked and I did the same tar test after changing partitions to make sure that I couldn't read it back after changing back to partition 0. I retested the DDS5 drive it still works as expected. I did find one issue in testing unrelated to the changes, the tell option didn't work with my LTO-6 drive: # ./mt -f /dev/st0 tell /dev/st0: Input/output error [ 2045.974642] st 3:0:0:0: [st0] Block limits 1 - 16777215 bytes. [ 2045.975221] st 3:0:0:0: [st0] Mode sense. Length 11, medium 0, WBS 10, BLL 8 [ 2045.975224] st 3:0:0:0: [st0] Density 5a, tape length: 0, drv buffer: 1 [ 2045.975226] st 3:0:0:0: [st0] Block size: 0, buffer size: 4096 (1 blocks). [ 2045.975718] st 3:0:0:0: [st0] Error: 8000002, cmd: 34 1 0 0 0 0 [ 2045.975723] st 3:0:0:0: [st0] Sense Key : Illegal Request [current] [ 2045.975726] st 3:0:0:0: [st0] Add. Sense: Invalid field in cdb [ 2045.975729] st 3:0:0:0: [st0] Can't read tape position. [ 2045.975857] st 3:0:0:0: [st0] Rewinding tape. I believe that in get_location() we're doing this: static int get_location(struct scsi_tape *STp, unsigned int *block, int *partition, int logical) { int result; unsigned char scmd[MAX_COMMAND_SIZE]; struct st_request *SRpnt; if (STp->ready != ST_READY) return (-EIO); memset(scmd, 0, MAX_COMMAND_SIZE); if ((STp->device)->scsi_level < SCSI_2) { scmd[0] = QFA_REQUEST_BLOCK; scmd[4] = 3; } else { scmd[0] = READ_POSITION; if (!logical && !STp->scsi2_logical) scmd[1] = 1; <<<<<<<<<<<<<< } When called from the ioctl that the tell option uses the variable logical is passed in as 0 (from what I could see everything else sets it to 1). For a READ_POSITION the drive I'm using only supports 0, 6, or 8 in the service action field of the second byte: https://docs.oracle.com/cd/E38452_01/en/LTO6_Vol3_E1_D20/LTO6_Vol3_E1_D20.pdf For SSC-3 (e.g. http://www.13thmonkey.org/documentation/SCSI/ssc3r01c.pdf) it looks like service action 1 is vendor specific and optional so it may not be implemented. SSC-2 defines that bit separately (http://www.staff.uni-mainz.de/tacke/scsi/SCSI2-10.html) as "A block address type (BT) bit of one requests the target to return its current first block location and last block location as a device-specific value.". It may need a test on scsi3 like in partition_tape so it would look like: static int get_location(struct scsi_tape *STp, unsigned int *block, int *partition, int logical) { int result; unsigned char scmd[MAX_COMMAND_SIZE]; struct st_request *SRpnt; + bool scsi3 = STp->device->scsi_level >= SCSI_3; if (STp->ready != ST_READY) return (-EIO); memset(scmd, 0, MAX_COMMAND_SIZE); if ((STp->device)->scsi_level < SCSI_2) { scmd[0] = QFA_REQUEST_BLOCK; scmd[4] = 3; } else { scmd[0] = READ_POSITION; - if (!logical && !STp->scsi2_logical) + if (!logical && !STp->scsi2_logical && !scsi3) scmd[1] = 1; } If you're fine with that and you don't believe it will have any side effects I can post a patch for that. Thanks Shane