Skip to contentSkip to conversation

Weekly Meeting 2019-03-04 #minutes

DISCUSSION Dan Rue 2 participants 2 messages Mar 4, 2019 – Mar 6, 2019 1 message not in the archive

lore
Share & tools
lore

A message that is not in the archive (sent off-list or not yet received): [email protected]

Dan Rue

author
#1 lore raw
Quoted 12 lines · Dan Rue via Groups.Io

- Attendees: Charles Oliveira, Guenter Roeck, Dan Rue, Mark Brown, Matt Hart, Amro Hassan, Anders Roxell, Michal Galka, Daniel Díaz, Guillaume Tucker, Alice Ferrazzi
- Matt
- I will do the PR to move clang builds to production
- Will also move the google builders
- Using idle cpu time on builders to run randconfig builds
- Ask Arnd
- Broonie
- rmk complaining it’s taking ~4 days to build his tree
- There was a massive backlog last week somehow
- Looks like it might’ve been to do with rolling out gtucker’s

Quoted 8 lines

changes, cleared now anyway.
- Dan
- Adding x15’s and naming convention
- Linaro lab uses name ‘x15’ for board, not the dtb name. Won’t be possible to rename boards in linaro lab.
- Lava supports aliases
- But do they work as expected?
- [drue]....no.

About this, there are a few options and I'm wondering which is preferred by the kernelci community:

Option 1:
- I can set the name of my x15 in lab-drue to am57xx-beagle-x15 and then I can add it to kernelci once my lab is in production (it's in staging as of today)
- Maybe eventually either lava or kernelci will be improved to be able to deal with alternative device names, at which time lkft's devices can be added

Option 2:
- I can add x15 to kernelci using the name 'x15', in which case the 11 x15 devices[1] in the lkft lab would immediately become available

Opinions?

Thanks,
Dan

[1] https://lkft.validation.linaro.org/scheduler/device_type/x15

Quoted 19 lines

- There are two different boards named x15, and x15 can boot with two different DTBs
- Will look at adding a naming convention page to kernelci wiki - i guess i’ll post to mailing list first for review.
- Daniel
- Working on a bisection shell script for lkft
- Interested in helping decouple the kernelci bisection from jenkins and working together
- [guillaume] it’s not really coupled with jenkins right now, though it is hard to use on the command line.
- Intel freedesktop folks have some advanced bisection tools too
https://gitlab.freedesktop.org/gfx-ci
- Amro
- Just touching base on behalf of Charles on his PR for extra callback to ES experimental. Any further holdoffs?
- https://github.com/kernelci/kernelci-core-staging/pull/68

Signature · 1 line
Linaro - Kernel Validation
#2 lore raw
Quoted 7 lines · Dan Rue
Quoted 5 lines · Dan Rue via Groups.Io

- Attendees: Charles Oliveira, Guenter Roeck, Dan Rue, Mark Brown, Matt Hart, Amro Hassan, Anders Roxell, Michal Galka, Daniel Díaz, Guillaume Tucker, Alice Ferrazzi
- Matt
- I will do the PR to move clang builds to production

Is the agreement as discussed on IRC, to enable clang-8 in production and have all the linux-next build results in the same email report?

In this scenario, the clang-9 builds would still be run on staging and sent to a smaller audience.

Quoted 9 lines
Quoted 8 lines

- Will also move the google builders
- Using idle cpu time on builders to run randconfig builds
- Ask Arnd
- Broonie
- rmk complaining it’s taking ~4 days to build his tree
- There was a massive backlog last week somehow
- Looks like it might’ve been to do with rolling out gtucker’s changes, cleared now anyway.

Yes, that was the glitch I mentioned on IRC. It's all fixed now. There wasn't actually a big build backlog, just a problem with the tree monitor job that got jammed.

Guillaume