kernel (2.6.37-8-generic_amd64) panic on boot (with message "map_single: bounce buffer is not DMA'ble) - possible regression !!!

Bug #688383 reported by carlos palma
16
This bug affects 3 people
Affects Status Importance Assigned to Milestone
linux (Ubuntu)
Fix Released
Undecided
Unassigned

Bug Description

Binary package hint: linux-image-2.6.37-8-generic

I'm currently using the kernel 2.6.35-23 generic (the last known good kernel before upgrade from Maverick to Natty).
It seams a possible regression (this Message is associated to source file "swiotlb.c" which have changed often since 2.6.35).

But the "possible regression" problem appears to be between 2.6.35-23 and 2.6.37-6 (Alfa1 upgrade).
I couldn't follow all the changes since 2.6.35 (at: http://www.linuxhq.com/kernel/file/lib/swiotlb.c).
The laptop is a DELL Vostro 1700 (7700 dual core; 4GB ram).

Also tested:
- Fresh install of Maveric and then an clean upgrade to Natty A1
- Daily builds (since Dec/02) on USB shows the same kernel panic / boot stop

The "/var/log/messages" file doesn't catch all the kernel boot log (see extraction below):

Dec 9 00:17:29 localhost kernel: [ 0.000000] Initializing cgroup subsys cpuset
Dec 9 00:17:29 localhost kernel: [ 0.000000] Initializing cgroup subsys cpu
Dec 9 00:17:29 localhost kernel: [ 0.000000] Linux version 2.6.37-8-generic (buildd@allspice) (gcc version 4.5.2 (Ubuntu/Linaro 4.5.1-10ubuntu3) ) #21-Ubuntu SMP Sun Dec 5 17:56:47 UTC 2010 (Ubuntu 2.6.37-8.21-generic 2.6.37-rc4)
Dec 9 00:17:29 localhost kernel: [ 0.000000] Command line: BOOT_IMAGE=/boot/vmlinuz-2.6.37-8-generic root=UUID=b165a793-8340-48f8-a2b4-091501e00083 ro debug vga=773
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-provided physical RAM map:
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 0000000000000000 - 000000000009f000 (usable)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 000000000009f000 - 00000000000a0000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 0000000000100000 - 00000000dfe6d800 (usable)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000dfe6d800 - 00000000e0000000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000f4000000 - 00000000f8000000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000fec00000 - 00000000fec10000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000fed18000 - 00000000fed1c000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000fed20000 - 00000000fed90000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000feda0000 - 00000000feda6000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000fee00000 - 00000000fee10000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 00000000fff00000 - 0000000100000000 (reserved)
Dec 9 00:17:29 localhost kernel: [ 0.000000] BIOS-e820: 0000000100002000 - 0000000120000000 (usable)
Dec 9 00:17:29 localhost kernel: [ 0.000000] NX (Execute Disable) protection: active
Dec 9 00:17:29 localhost kernel: [ 0.000000] DMI 2.4 present.
Dec 9 00:17:29 localhost kernel: [ 0.000000] No AGP bridge found
Dec 9 00:17:29 localhost kernel: [ 0.000000] last_pfn = 0x120000 max_arch_pfn = 0x400000000
Dec 9 00:17:29 localhost kernel: [ 0.000000] x86 PAT enabled: cpu 0, old 0x7040600070406, new 0x7010600070106
Dec 9 00:17:29 localhost kernel: [ 0.000000] last_pfn = 0xdfe6d max_arch_pfn = 0x400000000
Dec 9 00:17:29 localhost kernel: [ 0.000000] Scanning 0 areas for low memory corruption
Dec 9 00:17:29 localhost kernel: [ 0.000000] init_memory_mapping: 0000000000000000-00000000dfe6d000
Dec 9 00:17:29 localhost kernel: [ 0.000000] init_memory_mapping: 0000000100000000-0000000120000000
Dec 9 00:17:29 localhost kernel: [ 0.000000] RAMDISK: 36752000 - 373a1000
Dec 9 00:17:29 localhost kernel: [ 0.000000] ACPI: RSDP 00000000000fbbf0 00024 (v02 DELL )
Dec 9 00:17:29 localhost kerneDec 9 00:18:49 localhost kernel: imklog 4.6.4, log source = /proc/kmsg started.

Tags: kj-triage
Revision history for this message
carlos palma (carlos-palma) wrote :
Revision history for this message
carlos palma (carlos-palma) wrote :

Additional INFO after a few more tests:
- mainline kernels (http://kernel.ubuntu.com/~kernel-ppa/mainline/)
  - v2.6.36.2-natty (10-Dec-2010 13:42) - WORKS perfectly
  - v2.6.37-rc5-natty (07-Dec-2010 11:31) - NO (same kernel panic)
  - v2.6.37-rc4-natty (30-Nov-2010 10:54) - NO (same kernel panic)

Last screen message shown after the kernel panic is always related to function "swiotlb_map_page()" from swiotlb.c
Then ... complete freeze.
thanks

description: updated
Revision history for this message
carlos palma (carlos-palma) wrote :

Also registered in Bugzilla.kernel.org.
Follow discussion at: https://bugzilla.kernel.org/show_bug.cgi?id=25402

Revision history for this message
carlos palma (carlos-palma) wrote :

Additional INFO on new released kernels
- Kernel 2.6.36.3 (http://kernel.ubuntu.com/~kernel-ppa/mailine/v2.6.36.3-natty/): works fine
- kernel 2.6.37 stable (natty 2.6.37-12 generic amd64). still have the same boot panic (-> hang with all kb lights blinking).

Revision history for this message
carlos palma (carlos-palma) wrote :

I've found a similar problem in redhat.
See: https://bugzilla.redhat.com/show_bug.cgi?id=668088

tags: added: kj-triage
Revision history for this message
carlos palma (carlos-palma) wrote :

There is already a patch to this b44 regression
Proposed in: http://groups.google.com/group/linux.kernel/browse_thread/thread/184c7cd190ae7b5a?pli=1
I hope this one goes to the next rc ...
Thanks.

Revision history for this message
carlos palma (carlos-palma) wrote :

The Fujita patch was already merged to the Linus mainline kernel
See swiotlb.c diff:
 (http://www.kernel.org/diff/diffview.cgi?file=%2Fpub%2Flinux%2Fkernel%2Fv2.6%2Fsnapshots%2Fpatch-2.6.38-rc6-git6.bz2;z=123)

It will come out in the 2.6.38-rc7 for sure.

Changed in linux (Ubuntu):
status: New → Fix Released
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.