MHVTL general protection fault

[Bart Van Assche]

This is what I ran into after having built and loaded the mhvtl kernel module via "/etc/init.d/mhvtl start" (mhvtl version 0.18-17). Has anyone seen this before ?

# cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-2.6.38.7-scst-debug root=UUID=c2d91556-8ed3-4a2a-95d9-50d0203bcfcc ro quiet splash slub_debug=FZPU vt.handoff=7

From the kernel log:

mhvtl: vtl_remove_lu: ioctl to remove device <c t l> <00 02 00>, hba: ffff880002fb0448
general protection fault: 0000 [#1] PREEMPT SMP
last sysfs file: /sys/devices/pseudo_0/adapter0/host2/target2:0:8/2:0:8:0/scsi_generic/sg12/uevent
CPU 0
Modules linked in: mhvtl fuse drbd af_packet ocfs2_dlmfs ocfs2_stack_o2cb ocfs2_dlm ocfs2_nodemanager ocfs2_stackglue configfs rtc_cmos rtc_core i2c_piix4 processor thermal_sys rtc_lib i2c_core hwmon intel_agp container ac intel_gtt button ext4 mbcache jbd2 crc16 ehci_hcd uhci_hcd sr_mod sg usbcore e1000 piix cdrom sd_mod ide_core [last unloaded: mhvtl]

Pid: 3633, comm: vtltape Not tainted 2.6.38.7-scst-debug #1 VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform
RIP: 0010:[<ffffffff812ac7d6>]  [<ffffffff812ac7d6>] scsi_device_put+0x16/0x60
RSP: 0018:ffff880002605e28  EFLAGS: 00010286
RAX: 0000000100000000 RBX: 000000000000000c RCX: 0000000000000000
RDX: ffffffff813f0d4e RSI: 0000000000000000 RDI: 6b6b6b6b6b6b6b6b
RBP: ffff880002605e38 R08: ffff88001ea6d250 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000000 R12: ffff880002618a40
R13: ffff880019cda520 R14: ffff880002fb0458 R15: ffff880019cda520
FS:  00007f03a5202720(0000) GS:ffff88001d000000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 00007f8a370447b0 CR3: 000000001985a000 CR4: 00000000000006f0
DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
DR3: 0000000000000000 DR6: 00000000ffff0ff0 DR7: 0000000000000400
Process vtltape (pid: 3633, threadinfo ffff880002604000, task ffff88001b114600)
Stack:
000000000000000c ffff880002618a40 ffff880002605eb8 ffffffffa03005bb
ffff880002fb0448 ffffffff81316cc0 0000000200000000 0000000000000000
0000000000000000 0000000000000002 0000000000000001 0000000000000000
Call Trace:
[<ffffffffa03005bb>] vtl_c_ioctl+0x3ab/0x868 [mhvtl]
[<ffffffff81316cc0>] ? sys_sendto+0x110/0x140
[<ffffffff8113e5f5>] do_vfs_ioctl+0x95/0x580
[<ffffffff81087ef9>] ? trace_hardirqs_off_caller+0x29/0xc0
[<ffffffff81002f5c>] ? sysret_check+0x27/0x62
[<ffffffff8108dee9>] ? trace_hardirqs_on_caller+0x29/0x190
[<ffffffff8113eb2f>] sys_ioctl+0x4f/0x80
[<ffffffff81002f2b>] system_call_fastpath+0x16/0x1b
Code: 48 20 0f 18 0a 48 39 f9 75 db 31 c0 c9 c3 0f 1f 80 00 00 00 00 55 48 89 e5 48 83 ec 10 48 89 1c 24 4c 89 64 24 08 0f 1f 44 00 00 <48> 8b 07 49 89 fc 48 8b 80 00 02 00 00 48 8b 18 48 85 db 74 0c
RIP  [<ffffffff812ac7d6>] scsi_device_put+0x16/0x60
RSP <ffff880002605e28>
---[ end trace 670271d3dcede3bd ]---
mhvtl: vtl_release: lu for minor 12 Release

[Mark Harvey]

Hi Bart,

Many thanks for looking into this error..

I can't say I've seen the gpf you presented.
Although it might be because of non-unique SCSI cordinates in the /etc/mhvtl/device.conf (Chan 0, ID 8, LUN 0) ?

I've not tested / built a recent kernel - but normally run with default RedHat / SLES builds.. (Yeah, I know, they are a long way behind the kernel devel, but these are the platforms I work with in my paid job).

I'll attempt to build a 2.6.38 kernel today and test..

Cheers
mark

[crippa.andrea]

Bart Van Assche wrote

This is what I ran into after having built and loaded the mhvtl kernel module via "/etc/init.d/mhvtl start" (mhvtl version 0.18-17). Has anyone seen this before ?

Hi,
I've took a running mhvtl machine with 2.6.33.7 and updated with vanilla 2.6.38.7 (based on Centos 5.6)

During mhvtl kernel compile I've got this error :

/usr/src/redhat/BUILD/mhvtl-0.18/kernel/fetch27.c: In function âvtl_sg_copy_userâ:
/usr/src/redhat/BUILD/mhvtl-0.18/kernel/fetch27.c:51: warning: cast from pointer to integer of different size
/usr/src/redhat/BUILD/mhvtl-0.18/kernel/fetch27.c:52: warning: cast from pointer to integer of different size

Service start normally ....
Tried tgtd and with windows I can see the devices (tape drives and libraries)

Bart Van Assche wrote

# cat /proc/cmdline
BOOT_IMAGE=/boot/vmlinuz-2.6.38.7-scst-debug root=UUID=c2d91556-8ed3-4a2a-95d9-50d0203bcfcc ro quiet splash slub_debug=FZPU vt.handoff=7

This is mine :
[root@localhost ~]# cat /proc/cmdline
ro root=/dev/VolGroup00/LogVol00 quiet splash slub_debug=FZPU vt.handoff=7

Ant this is the message log : messages.txt at service start.

Bye