Description of problem: iSCSI service restart on RHEL4 U4 system leads to Kernel panic (#service iscsi restart). Issue observed couple of times on the same system and also on different configurations. Issue is not very consistent and some time the service restart comes up without the kernel panic. Refer the system logs for P8-PE2850 and P10-PE2900 for more details on kernel panic. ======================================== from Jason Jolly: This is the same "badness" as seen in CR124579. Alden was able to repro. this "badness" w/o MPP. Not sure if the following "oops" is caused by the first issue or not. Below are the "badness" and "oops" from the system log from the P8-PE2850 system. Jul 2 19:19:42 localhost kernel: Device 'iscsi0' does not have a release() function, it is broken and must be fixed. Jul 2 19:19:42 localhost kernel: Badness in device_release at drivers/base/core.c:85 Jul 2 19:19:42 localhost kernel: [c01c07d3>] kobject_cleanup+0x40/0x60 Jul 2 19:19:42 localhost kernel: [c01c07f3>] kobject_release+0x0/0x8 Jul 2 19:19:42 localhost kernel: [c01c0a99>] kref_put+0x42/0x45 Jul 2 19:19:42 localhost kernel: [f8a2bb78>] iscsi_cleanup+0x28/0x41 [iscsi_sfnet] Jul 2 19:19:42 localhost kernel: [c01369fe>] sys_delete_module+0x137/0x180 Jul 2 19:19:42 localhost kernel: [c01507f5>] unmap_vma_list+0xe/0x17 Jul 2 19:19:42 localhost kernel: [c0150b2f>] do_munmap+0x129/0x137 Jul 2 19:19:42 localhost kernel: [c011ae55>] do_page_fault+0x0/0x5c6 Jul 2 19:19:42 localhost kernel: [c02d4703>] syscall_call+0x7/0xb Jul 2 19:19:42 localhost iscsi: removing driver succeeded Oops: 0000 [#1] Jul 2 19:19:50 localhost kernel: SMP Jul 2 19:19:50 localhost kernel: Modules linked in: iscsi_sfnet(U) mppVhba(U) dell_rbu dcdbas parport_pc lp parport autofs4 i2c_dev i2c_core sunrpc crc32c libcrc32c scsi_transport_iscsi dm_mirror dm_multipath dm_mod button battery ac joydev md5 ipv6 uhci_hcd ehci_hcd hw_random e1000 floppy ext3 jbd megaraid_mbox megaraid_mm mppUpper(U) sg sd_mod scsi_mod Jul 2 19:19:50 localhost kernel: CPU: 3 Jul 2 19:19:50 localhost kernel: EIP: 0060:[f8888cb7>] Not tainted VLI Jul 2 19:19:50 localhost kernel: EFLAGS: 00010006 (2.6.9-42.ELsmp) Jul 2 19:19:50 localhost kernel: EIP is at mppCmn_NewSelectPath+0x9d/0x267 [mppUpper] Jul 2 19:19:50 localhost kernel: eax: 00000000 ebx: f6215c80 ecx: 00000000 edx: 00001b00 Jul 2 19:19:50 localhost kernel: esi: f37863c8 edi: f4e38000 ebp: f4e38000 esp: f3673f50 Jul 2 19:19:50 localhost kernel: ds: 007b es: 007b ss: 0068 Jul 2 19:19:50 localhost kernel: Process mpp_dcr (pid: 8212, threadinfo=f3673000 task=f425cb30) Jul 2 19:19:50 localhost kernel: Stack: 00000286 00000000 00000000 00000000 00000000 00000080 00000000 00000000 Jul 2 19:19:50 localhost kernel: f3673fc2 00000100 f6215c80 f37863c8 f4e38000 00000000 f88bad90 f3673fc3 Jul 2 19:19:50 localhost kernel: c02d1d16 00000000 00000001 f425cb30 c011e794 00100100 00200200 f37e0000 Jul 2 19:19:50 localhost kernel: Call Trace: Jul 2 19:19:50 localhost kernel: [f88bad90>] mppLnx_do_queuecommand+0x110/0x6ce [mppVhba] Jul 2 19:19:50 localhost kernel: [c02d1d16>] __down_interruptible+0x103/0x115 Jul 2 19:19:50 localhost kernel: [c011e794>] default_wake_function+0x0/0xc Jul 2 19:19:50 localhost kernel: [f88be4b8>] mppLnx_dpc_handler+0x110/0x210 [mppVhba] Jul 2 19:19:50 localhost kernel: [f88be3a8>] mppLnx_dpc_handler+0x0/0x210 [mppVhba] Jul 2 19:19:50 localhost kernel: [c01041f5>] kernel_thread_helper+0x5/0xb Jul 2 19:19:50 localhost kernel: Code: 15 01 c7 44 24 04 00 00 00 00 31 c0 3b 05 5c ae 89 f8 0f 83 0d 01 00 00 6b 54 24 24 1b 0f b6 44 24 15 01 c2 8b 8c 95 94 00 00 00 0f> b6 11 0f b6 da 88 54 24 14 6b d3 4c 8b 7c 85 34 8d 04 3a 8b Jul 2 19:19:50 localhost kernel: 0>Fatal exception: panic in 5 seconds Version-Release number of selected component (if applicable): How reproducible: Seldom. Steps to recreate: 1) Configure the setup as shown in attached configuration diagram 2) Restart iSCSI service from RHEL4 U4 system console(#service iscsi restart) 3) Verify RHEL4 U4 gives kernel panic and the server hangs. Actual results: Oops. Expected results: No Oops. Additional info:
The iscsi badness is from a LSI patch which they are using. Closing bugzilla.