Wake up daisy chain activation failed on omap3

Bug #594382 reported by Tobin Davis on 2010-06-14
8
This bug affects 1 person
Affects Status Importance Assigned to Milestone
linux (Ubuntu)
Critical
Mathieu Poirier
Maverick
Critical
Mathieu Poirier

Bug Description

On the 2.6.35-RC[1|2} kernels, the system gets stuck in a loop during boot with the message "Wake up daisy chain activation failed." repeating until the system is rebooted. No user interaction is available. See attached kernel logs.

Tobin Davis (gruemaster) wrote :
Tobin Davis (gruemaster) wrote :

Log for 2.6.35-RC2.

Changed in linux-ti-omap (Ubuntu):
assignee: nobody → Mathieu Poirier (mathieu.poirier)
tags: added: armel
Oliver Grawert (ogra) wrote :

the code in question seems to have come in with the patch from: http://<email address hidden>/msg18213.html

and seems to show up on the merge request here: http://www.gossamer-threads.com/lists/linux/kernel/1228828?do=post_view_threaded

Changed in linux-ti-omap (Ubuntu Maverick):
milestone: none → maverick-alpha-2
importance: Undecided → High
status: New → Confirmed
Tobin Davis (gruemaster) wrote :
Oliver Grawert (ogra) wrote :

this bug is a release blocker (renders omap3 unbootable), setting to critical

Changed in linux-ti-omap (Ubuntu Maverick):
importance: High → Critical

This seems to be related to yet another kernel config option:

- The code producing the error message is compiled in both upstream and maverick branches.
- If the upstream branch is compiled with upstreams' .config file, the daisy chain error messages don't show.
- If the upstream branch is compiled with maverick's .config file, the messages show up.
- If the maverick codebase is compiled with upstream's .config file, the messages don't show up.

The problem seams to be related to a timing issue when reading PM_WKST_WKUP, pulling for a daisy chain event. Introducing a simple mdelay(5) between the write and read operation fixes the problem but it is highly inefficient.

A real fix should follow shortly but in the mean time a kernel config with CONFIG_CPU_IDLE turned off has been checked in - this will allow people to keep working.

omap3 in Maverick is maintained in the linux master branch, thus I'm reassigning this to the linux kernel package.

The temporary work around to disable CONFIG_CPU_IDLE has been committed.

http://kernel.ubuntu.com/git?p=ubuntu/ubuntu-maverick.git;a=commit;h=026e03a3447b42645914c4bf81b58c5a359bf7c4

affects: linux-ti-omap (Ubuntu Maverick) → linux (Ubuntu Maverick)
Changed in linux (Ubuntu Maverick):
status: Confirmed → Fix Committed

Bah, I've been reminded that the git url will be invalid the next time we rebase the Maverick git tree as the sha1 id will change. And it's actually already changed as the BugLink needing fixing up in the commit message:

UBUNTU: [Config] ARM: Turning off CONFIG_CPU_IDLE on omap

This is a temperary fix that allow the system (beagleboard) to be
usable. The real fix, the one that really correct what seems to
be an issue with retreiving the value of the PM_WKST_WKUP register
should follow shortly.

BugLink: https://bugs.launchpad.net/bugs/594382
https://lists.ubuntu.com/archives/kernel-team/2010-June/011260.html

Signed-off-by: Mathieu Poirier <email address hidden>
Signed-off-by: Leann Ogasawara <email address hidden>

Launchpad Janitor (janitor) wrote :

This bug was fixed in the package linux - 2.6.35-6.7

---------------
linux (2.6.35-6.7) maverick; urgency=low

  [ Alberto Milone ]

  * [Upstream] Add support for the ATIF ACPI method to the radeon driver

  [ Chase Douglas ]

  * [Upstream] HID: magicmouse: scroll on entire surface, not just middle
    of mouse
  * [Upstream] HID: magicmouse: disable and add module param for scroll
    acceleration
  * [Upstream] HID: magicmouse: properly account for scroll movement in
    state
  * [Upstream] HID: magicmouse: add param for scroll speed
  * [Upstream] HID: magicmouse: enable horizontal scrolling

  [ Henrik Rydberg ]

  * [Upstream] Input: evdev - convert to dynamic event buffer
  * [Upstream] Input: evdev - use driver hint to compute size of event
    buffer
  * [Upstream] Input: bcm5974 - set the average number of events per MT
    event packet
  * [Upstream] Input: hid-input - use a larger event buffer for MT devices
  * [Upstream] Input: evdev - never leave the client buffer empty after
    write

  [ John Johansen ]

  * SAUCE: AppArmor: -- mainline 2010-06-23
  * SAUCE: AppArmor 2.4 compatibility patch
  * SAUCE: fs: block hardlinks to non-accessible sources AppArmor portion

  [ Leann Ogasawara ]

  * [Config] Enable CONFIG_INTR_REMAP=y
    - LP: #597091
  * [Config] Enable CONFIG_X86_X2APIC
    - LP: #597091

  [ Mathieu J. Poirier ]

  * [Config] ARM: Turning off CONFIG_CPU_IDLE on omap
    - LP: #594382
 -- Leann Ogasawara <email address hidden> Thu, 24 Jun 2010 12:19:48 -0700

Changed in linux (Ubuntu Maverick):
status: Fix Committed → Fix Released
Tobin Davis (gruemaster) wrote :

I have verified that this fixes the issue, albeit temporarily. Not sure how to mark it as still needing attention.

Tobin Davis (gruemaster) wrote :

Mathieu Poirier's test kernel seems to work fine. Not sure what all was changed, but it seams to be booting just fine.

To post a comment you must log in.
This report contains Public information  Edit
Everyone can see this information.

Other bug subscribers