FS#45548 - [linux] Kernel panic during boot with LVM RAID1 root volume

Attached to Project: Arch Linux
Opened by abc (Xiflite) - Saturday, 04 July 2015, 15:59 GMT
Last edited by Evangelos Foutras (foutrelis) - Wednesday, 15 July 2015, 21:25 GMT
Task Type Bug Report
Category Packages: Testing
Status Closed
Assigned To Tobias Powalowski (tpowa)
Thomas Bächler (brain0)
Evangelos Foutras (foutrelis)
Architecture All
Severity Critical
Priority Normal
Reported Version
Due in Version Undecided
Due Date Undecided
Percent Complete 100%
Votes 2
Private No

Details

Description:
The following error prevents LVM2 volumes to work. If the root fs is on a lvm volume it will probably not boot.
Even if the Kernel boots, "vgchange -aly <VG Name>" will hang an though not activate the volumes.

Seems to be this bug:
https://bugzilla.kernel.org/show_bug.cgi?id=100491

linux 4.1.1-1
linux-headers 4.1.1-1

dmesg:
[Sa Jul 4 17:37:49 2015] md: raid6 personality registered for level 6
[Sa Jul 4 17:37:49 2015] md: raid5 personality registered for level 5
[Sa Jul 4 17:37:49 2015] md: raid4 personality registered for level 4
[Sa Jul 4 17:37:49 2015] device-mapper: raid: Loading target version 1.6.0
[Sa Jul 4 17:37:49 2015] md: raid1 personality registered for level 1
[Sa Jul 4 17:37:49 2015] md/raid1:mdX: active with 2 out of 2 mirrors
[Sa Jul 4 17:37:49 2015] md-cluster module not found.
[Sa Jul 4 17:37:49 2015] mdX: Could not setup cluster service (256)
[Sa Jul 4 17:37:49 2015] mdX: bitmap file superblock:
[Sa Jul 4 17:37:49 2015] magic: 6d746962
[Sa Jul 4 17:37:49 2015] version: 4
[Sa Jul 4 17:37:49 2015] uuid: 00000000.00000000.00000000.00000000
[Sa Jul 4 17:37:49 2015] events: 127
[Sa Jul 4 17:37:49 2015] events cleared: 1
[Sa Jul 4 17:37:49 2015] state: 00000000
[Sa Jul 4 17:37:49 2015] chunksize: 524288 B
[Sa Jul 4 17:37:49 2015] daemon sleep: 5s
[Sa Jul 4 17:37:49 2015] sync size: 62914560 KB
[Sa Jul 4 17:37:49 2015] max write behind: 0
[Sa Jul 4 17:37:49 2015] BUG: unable to handle kernel NULL pointer dereference at 0000000000000100
[Sa Jul 4 17:37:49 2015] IP: [<ffffffff8158acac>] _raw_spin_lock_irq+0x1c/0x50
[Sa Jul 4 17:37:49 2015] PGD 0
[Sa Jul 4 17:37:49 2015] Oops: 0002 [#1] PREEMPT SMP
[Sa Jul 4 17:37:49 2015] Modules linked in: raid1 dm_raid raid456 md_mod async_raid6_recov async_memcpy async_pq async_xor async_tx ipmi_ssif joydev mousedev hid_generic usbhid hid iTCO_wdt iTCO_vendor_support ast ttm drm_kms_helper intel_rapl iosf_mbi x86_pkg_temp_thermal intel_powerclamp coretemp crct10dif_pclmul crc32_pclmul ghash_clmulni_intel igb aesni_intel drm aes_x86_64 lrw gf128mul glue_helper ablk_helper cryptd ptp uas sb_edac pps_core syscopyarea pcspkr mei_me i2c_i801 sysfillrect edac_core usb_storage sysimgblt ioatdma i2c_algo_bit mei lpc_ich evdev i2c_core mac_hid dca tpm_tis shpchp tpm wmi ipmi_si ipmi_msghandler acpi_power_meter acpi_pad processor button sch_fq_codel loop kvm_intel kvm ip_tables x_tables xhci_pci xhci_hcd usbcore ahci usb_common libahci dm_snapshot dm_bufio dm_mod sd_mod pm80xx
This task depends upon

Closed by  Evangelos Foutras (foutrelis)
Wednesday, 15 July 2015, 21:25 GMT
Reason for closing:  Fixed
Additional comments about closing:  linux 4.1.2-2
Comment by Kiril Vladimiroff (Vladimiroff) - Wednesday, 15 July 2015, 08:40 GMT
This patch seems to have fixed the issue: https://bugzilla.kernel.org/attachment.cgi?id=182511

It's now released under 4.1.2 which is already in testing. Could you please test and confirm?
Comment by Evangelos Foutras (foutrelis) - Wednesday, 15 July 2015, 08:52 GMT
That patch isn't included or committed upstream yet. (But we might backport it to our package.)
Comment by abc (Xiflite) - Wednesday, 15 July 2015, 19:51 GMT
Patch works for me. Thanks!
Comment by Evangelos Foutras (foutrelis) - Wednesday, 15 July 2015, 20:47 GMT
I've included the patch in linux 4.1.2-2 so that should fix it.

Weirdly enough, I was unable to reproduce the problem again in a new virtual machine. (I was able to repro it a few days ago with both 4.1.1 and 4.1.2 but I didn't keep that VM.)

Loading...