New firewire stack unreliable with Texas Instruments TSB82AA2 IEEE-1394b
| Affects | Status | Importance | Assigned to | Milestone | |
|---|---|---|---|---|---|
| | linux (Ubuntu) |
Medium
|
Unassigned | ||
| | Maverick |
Medium
|
Andy Whitcroft | ||
Bug Description
Both controllers are tested okay working on an Ubuntu LiveCD environment (kernel 2.6.28-4-generic) and the old firewire stack.
New firewire stack works reliably only for the Ricoh (internal) controller. The TI (PCI Express) adapter appears to work but then I see repeated messages filling up dmesg.
I'm able to reproduce the issue exactly.
1. Boot
2. Connect storage device to internal (Ricoh) firewire
3. verify that it is working by running command # dd if=/dev/zero bs=4k of=/dev/sdb count=16k
NOTE there are no error messages yet
4. Unplug storage device from internal (Ricoh) firewire, wait for it to settle
5. Insert PCIE StarTech EC1394B2 (TI) controller, wait for it to settle
6. Connect storage device to PCIE StarTech EC1394B2 (TI) firewire, wait for it to settle
7. verify that it is working by running command # dd if=/dev/zero bs=4k of=/dev/sdb count=16k
NOTE that there are repeated errors filling up dmesg
...
[ 319.997795] sd 6:0:0:0: [sdb] CDB: Write(10): 2a 00 00 00 48 00 00 04 00 00
[ 319.997812] end_request: I/O error, dev sdb, sector 18432
[ 320.511177] sd 6:0:0:0: [sdb] Unhandled error code
[ 320.511183] sd 6:0:0:0: [sdb] Result: hostbyte=
...
This is the most often case. Infrequently, when I boot there is no problem using the TI firewire. After disconnecting and using the Ricoh and then back to the TI again, there are problems.
ProblemType: Bug
DistroRelease: Ubuntu 10.10
Package: linux-image-
Regression: Yes
Reproducible: Yes
ProcVersionSign
Uname: Linux 2.6.35-22-generic i686
AlsaVersion: Advanced Linux Sound Architecture Driver Version 1.0.23.
Architecture: i386
AudioDevicesInUse:
USER PID ACCESS COMMAND
/dev/snd/
Card0.Amixer.info:
Card hw:0 'Intel'/'HDA Intel at 0xfe9fc000 irq 46'
Mixer name : 'SigmaTel STAC9228'
Components : 'HDA:83847616,
Controls : 27
Simple ctrls : 18
Date: Fri Oct 8 15:03:29 2010
HibernationDevice: RESUME=
MachineType: Dell Inc. Inspiron 1420
ProcCmdLine: BOOT_IMAGE=
ProcEnviron:
PATH=(custom, user)
LANG=en_US.UTF-8
SHELL=/bin/bash
RelatedPackageV
SourcePackage: linux
WpaSupplicantLog:
dmi.bios.date: 12/19/2008
dmi.bios.vendor: Dell Inc.
dmi.bios.version: A10
dmi.board.name: 0DT492
dmi.board.vendor: Dell Inc.
dmi.chassis.type: 8
dmi.chassis.vendor: Dell Inc.
dmi.modalias: dmi:bvnDellInc.
dmi.product.name: Inspiron 1420
dmi.sys.vendor: Dell Inc.
| Eric Shattow (eshattow) wrote : | #1 |
| Eric Shattow (eshattow) wrote : | #2 |
| Stefan Richter (stefan-r-ubz) wrote : | #3 |
Discussion on linux1394-user:
http://
| Eric Shattow (eshattow) wrote : | #4 |
Narrowed it down to improvements in 2.6.35 causing the regression.
Using kernels available at http://
Good: 2.6.34-
Bad: 2.6.35-
| Eric Shattow (eshattow) wrote : | #5 |
Fix is in upstream Linux 2.6.36 stable release. Can we see this in Ubuntu Maverick?
| Eric Shattow (eshattow) wrote : | #6 |
It's this one...
commit 7f81c56cf29c0af
Merge: 63f1474 aa0170f
Author: Linus Torvalds
Date: Mon Oct 18 13:09:26 2010 -0700
Merge branch 'for-linus' of git://git.
* 'for-linus' of git://git.
firewire: ohci: fix TI TSB82AA2 regression since 2.6.35
| Stefan Richter (stefan-r-ubz) wrote : | #7 |
This patch will be included in upstream 2.6.35.8, expected to be released early next week. From there it will find its way into an Ubuntu 2.6.35 package update, I suppose.
I am experiencing problems with this chipset on a Eurotech COM1461 PC104 pci firewire card (ieee1394 a and b) and a Unibrain Fire-i camera in Ubuntu 10.04 (with linux-image-
I tried to update the kernel to linux-image-
Sorry but I can not attach the kernel log excerpts until Wednesday.
By the time, where can I get the 2.6.35.8 kernel image (or just the modules) from? Are there any precompiled .deb packages or should I compile them by myself?
Sorry, but I forgot to mention that the camera works perfectly under Windows on the same machine. It also works on another Ubuntu 10.04 machine (with linux-image-
Sorry for the recomment.
| Stefan Richter (stefan-r-ubz) wrote : | #10 |
Joan:
1. "I am experiencing problems" is not a bug report.
2. Unibrain Fire-i is neither a storage device nor a 1394b device, like in this bug.
3. This bug here occurs only in firewire-ohci of kernel 2.6.35...2.6.35.7 and nothing else.
4. Ubuntu 10.04 with kernel 2.6.32 uses the older ohci1394 driver per default.
Why are you commenting at this bug at all?
Open a new bug with a detailed problem description. Or better yet, mail a detailed problem description to a relevant upstream mailinglist: linux1394-user if it seems to be a generic FireWire problem, coriander-user if the problems specifically occur in the Coriander capture software, libdc1394-devel if the problems occur in another libdc1394 based capture software. Because it is almost guaranteed that nobody who could help with problems like you might have is going to accidently see a report here on bugs.launchpad.net. (Cf. how Eric got his problem resolved.)
My intention was just to point out that on linux-image-
Since the title of this bug is "New firewire stack unreliable with Texas Instruments TSB82AA2 IEEE-1394b" I was afraid of open a new bug that may be a replicate of this one.
But of course, if this bug is only about 1394b STORAGE devices and TI TSB82AA2, I agree that this is not the correct place. Following your advice, I'll comment the errors on linux1394-user because from the kernel logs it seems clear that the problem are related with the firewire modules.
Thank you!
| tags: | added: acpi |
| Changed in linux (Ubuntu): | |
| assignee: | nobody → Andy Whitcroft (apw) |
| importance: | Undecided → Medium |
| status: | New → Fix Released |
| Changed in linux (Ubuntu Maverick): | |
| assignee: | nobody → Andy Whitcroft (apw) |
| importance: | Undecided → Medium |
| status: | New → Fix Committed |
| Julian Wiedmann (jwiedmann) wrote : | #12 |
Eric,
the patch you mentioned was released for Maverick with 2.6.35-24.42 - is this still an issue?
| Eric Shattow (eshattow) wrote : | #13 |
OK The updated kernel has fewer problems with my TSB82AA2 hardware than before.
| Eric Shattow (eshattow) wrote : | #14 |
Do not have this device now and so marking the bug Incomplete.
| Changed in linux (Ubuntu Maverick): | |
| status: | Fix Committed → Incomplete |
| Changed in linux (Ubuntu): | |
| status: | Fix Released → Incomplete |
| Julian Wiedmann (jwiedmann) wrote : | #15 |
This release has reached end-of-life [0].
| Changed in linux (Ubuntu Maverick): | |
| status: | Incomplete → Invalid |
| Changed in linux (Ubuntu): | |
| assignee: | Andy Whitcroft (apw) → nobody |
| Launchpad Janitor (janitor) wrote : | #16 |
[Expired for linux (Ubuntu) because there has been no activity for 60 days.]
| Changed in linux (Ubuntu): | |
| status: | Incomplete → Expired |


Repeated trials with new and old stack shows a regression from 2.6.32-24-generic to 2.6.35-22-generic. Attached.