Hi Dan,
Sorry to reply that late, frankly I just managed to find some time to look into this. I've just sent relevant patch: [1], hopefully it will be merged during merge window (which just opened). Please let me know if such patch helps with your issue, and what is current status of it.
Thanks!
[1] https://patchwork.ozlabs.org/patch/1129872/
On Fri, Apr 12, 2019 at 9:33 PM Dan Rue dan.rue@linaro.org wrote:
Hi Sam!
I am trying to add the x15 to kernelci and ran into something I'm not sure how to deal with.
In https://projects.linaro.org/browse/LAVA-1048, there is a lot of discussion around console implementation, and the end result is that we turned on CONFIG_SERIAL_8250_OMAP in our LKFT builds and then set the console to ttyS2.
I also noticed that CONFIG_SERIAL_8250_OMAP was enabled in multi_v7_defconfig as of 5.0, via f98d45145e6a ("ARM: multi_v7_defconfig: Enable 8250-omap serial driver and use it by default").
Now, when it comes to kernelci, the first thing that we tried was booting multi_v7_defconfig without any additional config fragments, and due to at least our LAVA implementation, it does not work on kernels older than 5.0.
So, I'm trying to find the best way forward. Here are the options I see:
- Enable x15 in kernelci on 5.0+
- [drue] my view is this would be the easiest, but disappointing outcome
- Backport f98d45145e6a
- [drue] i assume there are implications of this change that prevent
it from being backported
- Update lava implementation to use ttyO2 on older kernels
- [drue] not sure if it will work; implementation complicated; would
require jobs to specify when needing ttyO2
- In kernelci, enable a kernel build which is
defconfig+CONFIG_SERIAL_8250_OMAP, just for x15 on kernels older than 5.0.
- [drue] this would get the x15 booting on older branches in
kernelci, but it would still not be used for testing all of the other various config combinations.
So, I'm leaning toward the first or the last option, and wondering if the second option has any potential, or if you have any other suggestions.
Thanks! Dan