lockdep warning during boot with tilt-3.4
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Linaro Ubuntu |
Fix Released
|
Low
|
Ricardo Salveti | ||
linaro-landing-team-ti |
Fix Released
|
Low
|
David Long |
Bug Description
[ 1.726135] ------------[ cut here ]------------
[ 1.731018] WARNING: at /build/
[ 1.742492] Modules linked in:
[ 1.745758] [<c0011011>] (unwind_
[ 1.755554] [<c002b409>] (warn_slowpath_
[ 1.765594] [<c002b435>] (warn_slowpath_
[ 1.775573] [<c005ce89>] (lockdep_
[ 1.785980] [<c00b8231>] (kmem_cache_
[ 1.796691] [<c0023011>] (omap2_
[ 1.807800] [<c0024417>] (omap4460_
[ 1.818664] [<c0022bcd>] (omap2_
[ 1.828277] [<c0025877>] (clk_round_
[ 1.837982] [<c071a0e1>] (omap_init_
[ 1.847686] [<c071a21b>] (omap4_
[ 1.856933] [<c00085f5>] (do_one_
[ 1.866180] [<c0711683>] (do_basic_
[ 1.875061] [<c07116f7>] (kernel_
[ 1.884155] ---[ end trace 0da3e76b87a358c1 ]---
Using kernel:
linux-linaro-
* Packaged version of lt-omap kernel created from:
Kernel:
Repo: git://git.
Branch: tilt-3.4
Head:
commit 61abde21bf62729
Author: Jaswinder Singh <jaswinder.
Date: Fri Jun 22 11:26:22 2012 +0800
omapdss temp hack ignore resume runtime_pm fail
Board config fragment:
Repo: git://git.
Branch: config-boards-3.4
Head:
commit b7274f7e11c7820
Author: Ricardo Salveti de Araujo <ricardo.
Date: Mon Jun 25 15:30:04 2012 -0300
Ubuntu and Linaro Base config fragments:
Repo: git://git.
Branch: config-core-3.4
Head:
commit 072bd6cb29aa25a
Author: Ricardo Salveti de Araujo <ricardo.
Date: Mon Jun 25 18:06:06 2012 -0300
Not yet stable enough to be used as default for all boards we currently
support at Linaro.
Packaging template:
Repo: git://git.
Branch: linaro-
Head:
commit 3da2b9971c9fbd9
Author: John Rigby <john.rigby.org>
Date: Mon Jun 25 15:04:23 2012 -0600
LINARO: template debian.linaro based on omap only version
-- John Rigby <email address hidden> Mon, 25 Jun 2012 15:04:22 -0600
visibility: | private → public |
Changed in linaro-ubuntu: | |
status: | New → Confirmed |
Changed in linaro-landing-team-ti: | |
status: | New → Confirmed |
assignee: | nobody → David Long (dave-long) |
Changed in linaro-landing-team-ti: | |
status: | Fix Committed → Fix Released |
Changed in linaro-landing-team-ti: | |
importance: | Undecided → Low |
Changed in linaro-ubuntu: | |
status: | Confirmed → Fix Committed |
importance: | Undecided → Low |
assignee: | nobody → Ricardo Salveti (rsalveti) |
Changed in linaro-ubuntu: | |
status: | Fix Committed → Fix Released |
Dave it's complaining about the kzalloc in omap2_dpll_ round_rate( ) which you added to generate the rate cache.
Actually I don't know what it's complaining about exactly except somehow it's a "bad time to do kzalloc()", and it's the only code that triggers the problem.
Since there are only a handful of discrete frequencies, I guess you should make a patch which adapts this code to use a static array of struct dpll_rate_list and a static "high water mark" replacing kzalloc. If it reaches the size of the array, just don't cache the result any more. Or, if you can see what the actual complaint is really about attacking it that way is also an option.