Hi All
I assume we'll be moving to Linux 3.8 for the January release cycle?
For my part I have prepared a 3.8 branch for vexpress [1] which doesn't yet contain Android patches or bit.LITTLE MP as their respective branches aren't on 3.8 yet.
[1] http://git.linaro.org/gitweb?p=landing-teams/working/arm/kernel.git%3Ba=shor...
On 4 January 2013 15:09, Jon Medhurst (Tixy) tixy@linaro.org wrote:
I assume we'll be moving to Linux 3.8 for the January release cycle?
For my part I have prepared a 3.8 branch for vexpress [1] which doesn't yet contain Android patches or bit.LITTLE MP as their respective branches aren't on 3.8 yet.
big.LITTLE MP is 3.8 based now :)
btw, you wrote bit.LITTLE instead of big.LITTLE :)
On Fri, 2013-01-04 at 15:14 +0530, Viresh Kumar wrote:
On 4 January 2013 15:09, Jon Medhurst (Tixy) tixy@linaro.org wrote:
I assume we'll be moving to Linux 3.8 for the January release cycle?
For my part I have prepared a 3.8 branch for vexpress [1] which doesn't yet contain Android patches or bit.LITTLE MP as their respective branches aren't on 3.8 yet.
big.LITTLE MP is 3.8 based now :)
:-) I've now included that in my integration-next branch and Android boots on this, even though we don't have any of the Android patches?! (I'm sure that never used to be possible.) Of course, we're missing important things like the interactive governor...
If we're definitely moving to 3.8 then I may soon drop my temporary 'next' branches and move 'tracking' branches to 3.8, so if anyone can't see integration-next in my tree, use integration-{linux,android}-vexpress
BTW, when booting I see the lockdep error also reported upstream: https://lkml.org/lkml/2013/1/3/192
On 01/04/2013 02:28 PM, Jon Medhurst (Tixy) wrote:
On Fri, 2013-01-04 at 15:14 +0530, Viresh Kumar wrote:
On 4 January 2013 15:09, Jon Medhurst (Tixy) tixy@linaro.org wrote:
I assume we'll be moving to Linux 3.8 for the January release cycle?
For the 13.01 my plan was to stay at 3.7 plus the stable 3.7.y updates (as by the time of code freeze the mainline would be at v3.8-rc4 or so - not sure if -rc4 would be stable enough).
But if the topic owners fill comfortable with moving to 3.8, we can go that way. Just would like to get confirmations from the owners of the heaviest topics.
For my part I have prepared a 3.8 branch for vexpress [1] which doesn't yet contain Android patches or bit.LITTLE MP as their respective branches aren't on 3.8 yet.
big.LITTLE MP is 3.8 based now :)
I take it as the vexpress topic (for the ll tree) and the big.LITTLE topic (for the llct tree) are ready to switch to 3.8, correct?
This leaves the Android topic for llct (John?) and Samsung LT topic for ll (Tushar?).
If anyone has another topic branch to include into 13.01 release, and moving to 3.8 or staying at 3.7.y is critical for that topic, please chime in!
:-) I've now included that in my integration-next branch and Android boots on this, even though we don't have any of the Android patches?! (I'm sure that never used to be possible.) Of course, we're missing important things like the interactive governor...
If we're definitely moving to 3.8 then I may soon drop my temporary 'next' branches and move 'tracking' branches to 3.8, so if anyone can't see integration-next in my tree, use integration-{linux,android}-vexpress
Yes, "definitely moving to 3.8" would make my life a bit easier too: llct could be moved to 3.8 w/o causing any troubles for topic owners (otherwise they would have to switch to linux-linaro-core-3.7 branch), and there would be no urgency in creating CI jobs for linux-linaro-core-3.7.
BTW, when booting I see the lockdep error also reported upstream: https://lkml.org/lkml/2013/1/3/192
It looks like there is some progress in getting the fix, so it could probably be resolved by 13.01 release. No 100% guarantee of course..
Thanks, Andrey
On 01/04/2013 06:33 AM, Andrey Konovalov wrote:
On 01/04/2013 02:28 PM, Jon Medhurst (Tixy) wrote:
On Fri, 2013-01-04 at 15:14 +0530, Viresh Kumar wrote:
On 4 January 2013 15:09, Jon Medhurst (Tixy) tixy@linaro.org wrote:
For my part I have prepared a 3.8 branch for vexpress [1] which doesn't yet contain Android patches or bit.LITTLE MP as their respective branches aren't on 3.8 yet.
big.LITTLE MP is 3.8 based now :)
I take it as the vexpress topic (for the ll tree) and the big.LITTLE topic (for the llct tree) are ready to switch to 3.8, correct?
This leaves the Android topic for llct (John?) and Samsung LT topic for ll (Tushar?).
I'll take a swing at rebasing/merging the Android tree. Although my build box is ordered, but not shipped, so I'm trying to hobble along doing git work from my very slow work VM, so I can't promise it will happen quickly.
That said, as Tixy noted, most of the critical Android support is already upstream (and the 4.2 userland is using it), so some Android testing could likely be done without the linaro.android tree merged in, but it will be missing items like the interactive cpufreq gov, paranoid networking, ion, sync, etc.
thanks -john
On 01/04/2013 12:19 PM, John Stultz wrote:
On 01/04/2013 06:33 AM, Andrey Konovalov wrote:
On 01/04/2013 02:28 PM, Jon Medhurst (Tixy) wrote:
On Fri, 2013-01-04 at 15:14 +0530, Viresh Kumar wrote:
On 4 January 2013 15:09, Jon Medhurst (Tixy) tixy@linaro.org wrote:
For my part I have prepared a 3.8 branch for vexpress [1] which doesn't yet contain Android patches or bit.LITTLE MP as their respective branches aren't on 3.8 yet.
big.LITTLE MP is 3.8 based now :)
I take it as the vexpress topic (for the ll tree) and the big.LITTLE topic (for the llct tree) are ready to switch to 3.8, correct?
This leaves the Android topic for llct (John?) and Samsung LT topic for ll (Tushar?).
I'll take a swing at rebasing/merging the Android tree. Although my build box is ordered, but not shipped, so I'm trying to hobble along doing git work from my very slow work VM, so I can't promise it will happen quickly.
Ok, I've got a first-pass branch here: git://git.linaro.org/people/jstultz/android.git linaro-android-3.8-jstultz-test
Instead of doing the full re-base, I just merged the 3.8-rc2+ tree into the linaro-android-3.7-anton-rebase branch, and the collisions were seemingly manageable.
I've only compile tested on x86_64, since I don't have my cross-tools setup yet.
If someone has the time to test merge that branch and make sure it boots, I'd really appreciate it.
thanks -john
On Fri, 2013-01-04 at 14:06 -0800, John Stultz wrote:
Ok, I've got a first-pass branch here: git://git.linaro.org/people/jstultz/android.git linaro-android-3.8-jstultz-test
Thanks for that.
Instead of doing the full re-base, I just merged the 3.8-rc2+ tree into the linaro-android-3.7-anton-rebase branch, and the collisions were seemingly manageable.
I've only compile tested on x86_64, since I don't have my cross-tools setup yet.
If someone has the time to test merge that branch and make sure it boots, I'd really appreciate it.
I get this compilation error:
arch/arm/mm/cache-l2x0.c: In function 'l2x0_init': arch/arm/mm/cache-l2x0.c:376:3: error: 'cache_id' undeclared (first use in this function) arch/arm/mm/cache-l2x0.c:376:3: note: each undeclared identifier is reported only once for each function it appears in arch/arm/mm/cache-l2x0.c: At top level: arch/arm/mm/cache-l2x0.c:38:21: warning: 'l2x0_sets' defined but not used [-Wunused-variable]
If I revert Android's cache-l2x0.c change then it builds and boots fine on vexpress. I'm not sure how to test the Android patches specifically but I can confirm the interactive governor was running after boot and the system generally seemed to work as expected.
On 01/07/2013 07:24 AM, Jon Medhurst (Tixy) wrote:
On Fri, 2013-01-04 at 14:06 -0800, John Stultz wrote:
Instead of doing the full re-base, I just merged the 3.8-rc2+ tree into the linaro-android-3.7-anton-rebase branch, and the collisions were seemingly manageable.
I've only compile tested on x86_64, since I don't have my cross-tools setup yet.
If someone has the time to test merge that branch and make sure it boots, I'd really appreciate it.
I get this compilation error:
arch/arm/mm/cache-l2x0.c: In function 'l2x0_init': arch/arm/mm/cache-l2x0.c:376:3: error: 'cache_id' undeclared (first use in this function) arch/arm/mm/cache-l2x0.c:376:3: note: each undeclared identifier is reported only once for each function it appears in arch/arm/mm/cache-l2x0.c: At top level: arch/arm/mm/cache-l2x0.c:38:21: warning: 'l2x0_sets' defined but not used [-Wunused-variable]
If I revert Android's cache-l2x0.c change then it builds and boots fine on vexpress. I'm not sure how to test the Android patches specifically but I can confirm the interactive governor was running after boot and the system generally seemed to work as expected.
Great! Thanks for the testing! I'll revert the change you pointed out and will push it out for merging.
-john
On 01/07/2013 07:24 AM, Jon Medhurst (Tixy) wrote:
I get this compilation error:
arch/arm/mm/cache-l2x0.c: In function 'l2x0_init': arch/arm/mm/cache-l2x0.c:376:3: error: 'cache_id' undeclared (first use in this function) arch/arm/mm/cache-l2x0.c:376:3: note: each undeclared identifier is reported only once for each function it appears in arch/arm/mm/cache-l2x0.c: At top level: arch/arm/mm/cache-l2x0.c:38:21: warning: 'l2x0_sets' defined but not used [-Wunused-variable]
If I revert Android's cache-l2x0.c change then it builds and boots fine on vexpress. I'm not sure how to test the Android patches specifically but I can confirm the interactive governor was running after boot and the system generally seemed to work as expected.
Yea, *looks* like the PL310 errata 727915 issue fixed in the Android commit id: 74b6cdd9573abba116584c08003ee5e87e96ea14 have been fixed upstream already.
Reverted that change and pushed the whole thing out here: git://git.linaro.org/people/jstultz/android.git linaro-android-3.8-jstultz-merge
thanks -john
On 01/04/2013 08:03 PM, Andrey Konovalov wrote:
On 01/04/2013 02:28 PM, Jon Medhurst (Tixy) wrote:
On Fri, 2013-01-04 at 15:14 +0530, Viresh Kumar wrote:
On 4 January 2013 15:09, Jon Medhurst (Tixy) tixy@linaro.org wrote:
I assume we'll be moving to Linux 3.8 for the January release cycle?
For the 13.01 my plan was to stay at 3.7 plus the stable 3.7.y updates (as by the time of code freeze the mainline would be at v3.8-rc4 or so - not sure if -rc4 would be stable enough).
But if the topic owners fill comfortable with moving to 3.8, we can go that way. Just would like to get confirmations from the owners of the heaviest topics.
For my part I have prepared a 3.8 branch for vexpress [1] which doesn't yet contain Android patches or bit.LITTLE MP as their respective branches aren't on 3.8 yet.
big.LITTLE MP is 3.8 based now :)
I take it as the vexpress topic (for the ll tree) and the big.LITTLE topic (for the llct tree) are ready to switch to 3.8, correct?
This leaves the Android topic for llct (John?) and Samsung LT topic for ll (Tushar?).
I am ok with 3.8 migration. Samsung LT tree would be available around -rc3 time.
Thanks, Andrey
-
Andrey Konovalov -
John Stultz -
Jon Medhurst (Tixy) -
Tushar Behera -
Viresh Kumar