On 16 February 2012 11:39, Deepti Kalakeri deepti.kalakeri@linaro.org wrote:
Hello Sangwook,
On Thu, Feb 16, 2012 at 4:37 PM, Sangwook Lee sangwook.lee@linaro.org wrote:
The initial problem, you faced is due to the wrong kernel selected by someone. If you tried kernel from CI (arnd) , it should be based on 3.2 kernel, but it was not true. It was based on 2.6.35-rxxx. Where does this kernel come from ? At first, please select the proper kernel and then try LAVA again.
I have rectified this. Actually we were using a repository setup by Arnd on git.linaro.org to build the kernel. We did not notice the move to git.kernel.org and we built the old things. But, otherwise also there are hwpacks built for linus(torvalds), linux-next tree with exynos4 defconfig using the latest upstream changes. I have not seen the hwpacks coming out for these trees either being successful to boot on the LAVA origen boards. Hence this is a problem that the hwpacks built for the origen boards do not work at all and not specific to just arm-soc tree. Can you check as an example if you can verify if the hwpacks built today works fine on your Origen board. http://ci.linaro.org/kernel_hwpack/linux-arm-soc-for-next/linux-arm-soc-for-....
With this image, I couldn't see the booting message.
Do we need to apply the patch you gave yesterday on top of the Arnd latest changes to make it work ?
Yes,
I downloaded git://git.kernel.org/pub/scm/linux/kernel/git/arm/arm-soc.git and then could see the booting message with the patch.
On 15 February 2012 21:35, Paul Larson paul.larson@linaro.org wrote:
On Wed, Feb 15, 2012 at 3:46 AM, Deepti Kalakeri deepti.kalakeri@linaro.org wrote:
On Wed, Feb 15, 2012 at 4:32 PM, Sangwook Lee sangwook.lee@linaro.org wrote:
On 14 February 2012 16:40, Paul Larson paul.larson@linaro.org wrote:
On Tue, Feb 14, 2012 at 2:21 AM, Sangwook Lee sangwook.lee@linaro.org wrote: > > Hi Paul > > On 14 February 2012 01:03, Paul Larson paul.larson@linaro.org > wrote: >> >> Looking at the latest results on.the kernel CI view I pointed out >> earlier, here is a serial.log from a recent one >> http://validation.linaro.org/lava-server/dashboard/streams/anonymous/ci-linu... >> >> You can see the test kernel starting to boot at around line 19560 >> and at 19660 it gets only as far as the starting kernel message, hangs, and >> eventually times out. > > I think there is no option to download this log, maybe only for LAVA > team. > If I download serial.log file as html, I must follow the link to > serial log but again I have a broken link.
Yes, we need a download link there - known feature request that we'll hopefully get to soon. >> >> but replaces the kernel with the output from jenkins building the >> selected kernel. I'm guessing that perhaps we're missing something critical >> from the kernel configuration? > > Yes, That's right. I am attaching the patch to fix this problem. > Unfortunately, currently this is not up-streamable. Hope that we > apply this as doing debian packaing. > please let me if you have further problem.
I'm not sure what you mean here. If this patch fixes the problem, what is preventing it from going upstream?
arnd kernel (mainline kernel) we have to do the following for Origen board 1) Select exynos4_defconfig All boards for exynos CPU are sharing one configuration. So It select all boards configs and then set UART port to 1 2) We have to set UART port to 2 manually only for Origen
The patch I sent to you is to select debug UART port to 2 by default, only for Origen If we release this patch into mainline, all other boards using Exynos will manually select UART port again.
I guess we have this fix somewhere in the kernel produced by the landing team? If so, it seems maybe we should also add the landing team tree here for testing so that we can see it pass.
Here is our branch: Git: git://git.linaro.org/landing-teams/working/samsung/kernel.git branch: tracking
Initially I thought LAVA team is using hwpack built by John
https://code.launchpad.net/~jcrigby/+recipe/linux-linaro-3.2-lt-origen-daily
Does Lave team build their own hwpack from kernel ?
Not the LAVA team. We(Infra) team build the latest upstream changes on ci.linaro.org and then send them to LAVA for testing.
Right, so I guess the better question would be, is there a bug for this already, and is there a better solution? Can we pass something on the boot cmdline that selects the proper uart? Why don't we have this problem with the linaro kernel in our leb images, do they include this workaround there?
Thanks, Paul Larson
-- Thanks and Regards, Deepti Infrastructure Team Member, Linaro Platform Teams Linaro.org | Open source software for ARM SoCs Follow Linaro: http://www.facebook.com/pages/Linaro http://twitter.com/#%21/linaroorg - http://www.linaro.org/linaro-blog