Red Hat Bugzilla – Bug 159708
Tape configuration does not work correctly
Last modified: 2012-06-20 09:17:52 EDT
Description of problem: After configuring a Tape LUN, iscsi-ls -l shows the device mapped to. But it also gives some error message as well. If any IO is executed on this tape, the linux host panics and crashes. Version-Release number of selected component (if applicable): - kernel-smp-2.6.9-11.EL - kernel-smp-devel-2.6.9-11.EL How reproducible: Always reproducible Steps to Reproduce: 1. Configured a tape (Quantum ATL1500) LUN on MDS 9216 2. Execution of iscsi-ls -l gives the following output: [root@cnitin-linux-6 ~]# iscsi-ls -l ******************************************************************************* SFNet iSCSI Driver Version ...4:0.1.11(12-Jan-2005) ******************************************************************************* TARGET NAME : iqn.2004-04.com.cisco.cnitin-tape-drive-2 TARGET ALIAS : HOST ID : 0 BUS ID : 0 TARGET ID : 0 TARGET ADDRESS : 10.1.1.80:3260,12416 SESSION STATUS : ESTABLISHED AT Wed Jun 1 16:00:00 IST 2005 SESSION ID : ISID 00023d000001 TSIH 3080 DEVICE DETAILS: --------------- LUN ID : 0 Vendor: QUANTUM Model: SuperDLT1 Rev: 2323 Type: Sequential-Access ANSI SCSI revision: 04 open class /sys/block/st0 failed: No such file or directory page83 type0: open class /sys/block/st0 failed: No such file or directory page80: Device: /dev/st0 ******************************************************************************* Actual results: iscsi-ls -l gives errors when a tape LUN is configured. Also, an IO on tape device displayed panics the system. Expected results: iscsi-ls -l should display the tape details without errors. An IO on tape should not panic the system. Additional info: (1) At one point of time when the crash happened, could gather the following information from the system console: Call Trace: [<e883e47d>] scsi_device_unbusy+0x46/0x5b [scsi_mod] [<e883ab15>] scsi_finish_command+0xad/0xb [scsi_mod] [<e883aa3a>] scsi_softirq+0xb6/0xbe [scsi_mod] [<c0125a5c>] __do_softirq+0x4c/0xb1 [<c010806d>] do_softirq+0x4f/0x56 ======================================= ======================================= [<e8a8fd6b>] iscsi_recv_pdu+0x178/0x249 [iscsi_sfnet] [<c01173fd>] smp_apic_timer_interrupt+0xd9/0xdd [<c02c7db6>] apic_timer_interrupt+0x1/0x20 [<e8a8c839>] iscsi_rx_thread+0x0/0x1e7 [iscsi_sfnet] [<e8a8c900>] iscsi_rx_thread+0xc7/0x1e7 [iscsi_sfnet] [<c0132e31>] kthread+0x73/0x9b [<c0132dbe>] kthread+0x0/0x9b [<c01041f1>] kernel_thread_helper+0x5/0xb <0> Kernel panic: not syncing: Fatal Exception interrupt (2) At another ocassion when crash happened, captured the following from the log messages: Jun 1 15:51:14 cnitin-linux-6 kernel: scsi0 : SFNet iSCSI driver Jun 1 15:51:14 cnitin-linux-6 kernel: Vendor: QUANTUM Model: SuperDLT1 Rev: 2323 Jun 1 15:51:14 cnitin-linux-6 kernel: Type: Sequential-Access ANSI SCSI revision: 04 Jun 1 15:51:15 cnitin-linux-6 scsi.agent[4082]: tape at /devices/platform/host0/target0:0:0/0:0:0:0 Jun 1 15:51:15 cnitin-linux-6 kernel: st: Version 20040403, fixed bufsize 32768, s/g segs 256 Jun 1 15:51:15 cnitin-linux-6 kernel: Attached scsi tape st0 at scsi0, channel 0, id 0, lun 0 Jun 1 15:51:15 cnitin-linux-6 kernel: st0: try direct i/o: yes (alignment 512 B), max page reachable by HBA 1048575 Jun 1 15:51:17 cnitin-linux-6 hald[3501]: Timed out waiting for hotplug event 271. Rebasing to 272 Jun 1 15:52:10 cnitin-linux-6 kernel: st0: Block limits 4 - 16777212 bytes. Jun 1 15:52:10 cnitin-linux-6 kernel: st0: Write not multiple of tape block size. Jun 1 15:52:25 cnitin-linux-6 kernel: st0: Write not multiple of tape block size. Jun 1 15:52:59 cnitin-linux-6 kernel: Unable to handle kernel NULL pointer dereference at virtual address 00000190 Jun 1 15:52:59 cnitin-linux-6 kernel: printing eip: Jun 1 15:52:59 cnitin-linux-6 kernel: e89adc96 Jun 1 15:52:59 cnitin-linux-6 kernel: *pde = 1e65e001 Jun 1 15:52:59 cnitin-linux-6 kernel: Oops: 0000 [#1] Jun 1 15:52:59 cnitin-linux-6 kernel: SMP Jun 1 15:52:59 cnitin-linux-6 kernel: Modules linked in: st crc32c libcrc32c iscsi_sfnet(U) md5 ipv6 parport_pc lp parport autofs4 i2c_dev i2c_core sunrpc scsi_transport_iscsi(U) button battery ac e100 mii e1000 floppy dm_snapshot dm_zero dm_mirror ext3 jbd dm_mod cpqarray sd_mod scsi_mod Jun 1 15:52:59 cnitin-linux-6 kernel: CPU: 1 Jun 1 15:52:59 cnitin-linux-6 kernel: EIP: 0060:[<e89adc96>] Not tainted VLI Jun 1 15:52:59 cnitin-linux-6 kernel: EFLAGS: 00010246 (2.6.9-11.ELsmp) Jun 1 15:52:59 cnitin-linux-6 kernel: EIP is at read_tape+0xfd/0x3c9 [st] Jun 1 15:52:59 cnitin-linux-6 kernel: eax: 00000000 ebx: e7811800 ecx: dfebcf70 edx: e7ce1400 Jun 1 15:52:59 cnitin-linux-6 kernel: esi: 00000000 edi: e7ce1400 ebp: 00000000 esp: dfebcf08 Jun 1 15:52:59 cnitin-linux-6 kernel: ds: 007b es: 007b ss: 0068 Jun 1 15:52:59 cnitin-linux-6 kernel: Process tar (pid: 4270, threadinfo=dfebc000 task=e5f06150) Jun 1 15:52:59 cnitin-linux-6 kernel: Stack: 00000000 00000001 00000000 e78872cc 00000000 dfe2c800 e7ce14c8 dfebcf70 Jun 1 15:52:59 cnitin-linux-6 kernel: e7ce1400 00000108 00000000 00000000 00000000 08217800 00000000 00000000 Jun 1 15:52:59 cnitin-linux-6 kernel: dfebc000 e89ae1fc 00000000 dfe2c800 e7ce14c8 e7ce1400 00000003 00002800 Jun 1 15:52:59 cnitin-linux-6 kernel: Call Trace: Jun 1 15:52:59 cnitin-linux-6 kernel: [<e89ae1fc>] st_read+0x29a/0x3cf [st] Jun 1 15:53:00 cnitin-linux-6 kernel: [<c0156011>] vfs_read+0xb6/0xe2 Jun 1 15:53:00 cnitin-linux-6 kernel: [<c0156224>] sys_read+0x3c/0x62 Jun 1 15:53:00 cnitin-linux-6 kernel: [<c02c7377>] syscall_call+0x7/0xb Jun 1 15:53:00 cnitin-linux-6 kernel: Code: 00 89 e8 88 54 24 20 0f 95 44 24 1d c1 f8 10 88 44 24 1e 89 e8 c1 f8 08 88 44 24 1f 8b 19 6a 01 6a 00 8b 7c 24 20 8b 47 04 89 fa <ff> b0 90 01 00 00 89 d8 6a 02 56 8d 4c 24 30 e8 83 e6 ff ff 89 Jun 1 15:53:00 cnitin-linux-6 kernel: <0>Fatal exception: panic in 5 seconds Jun 1 15:53:00 cnitin-linux-6 kernel: iscsi-sfnet:host0: Session dropped
open class /sys/block/st0 failed: No such file or directory That is a bug in that iscsi-ls should not look for tape in the block dir. It has a gend_disk but it never has add_disk() called on it. Nitin could you have Smitha fix that one since we do not have iSCSI tape here? It is pretty trivial I can send a patch for what I think is right if Smitha has no time and you want to be the test subject. As for the io error. I think I know what this is. I will send something to test later.
ok so Smitha fixed the iscsi-ls bug. I am looking into the oops, but do not have iscsi tape (have been trying to hack virtual iscsi tape support into IET over the weekend).
Arg!. Smitha could you take this? Hacking in tape support to a target to fix a tape bug is too much trouble :( Citin, If you can replicate on a disk let me know and I will help.
Thank you for submitting this issue for consideration in Red Hat Enterprise Linux. The release for which you requested us to review is now End of Life. Please See https://access.redhat.com/support/policy/updates/errata/ If you would like Red Hat to re-consider your feature request for an active release, please re-open the request via appropriate support channels and provide additional supporting details about the importance of this issue.