On Thu, Nov 23, 2017 at 11:40 AM, Olof Johansson olof@lixom.net wrote:
On Thu, Nov 23, 2017 at 1:49 AM, Arnd Bergmann arnd@arndb.de wrote:
On Thu, Nov 23, 2017 at 1:20 AM, Olof's autobooter build@lixom.net wrote:
Full logs are available at http://arm-soc.lixom.net/bootlogs/mainline/v4.14/
Failed boards:
arndale arm-multi_v7_nosmp_defconfig : FAILED 1:15.09Successful boards: cubie2 arm-multi_v7_nosmp_defconfig : warnings 0:46.50 ...
Last entries of failed logs below:It looks like Olof's boot bot came back after a time without running any tests, and found a regression on Arndale when running with CONFIG_SMP disabled. cubie2 did not run into the same problem.
The backtrace looks a bit suspicous. I can see __switch_to() calling the notifiers, which would be vfp_notifier_block and contextidr_notifier_block on both those machines.
I was doing a trial run with just a few boards, it's a bit premature to claim it's back. I didn't mean for the email to go out.
Looks like kernelci doesn't cover CONFIG_SMP=n boots, so I guess this might be a new issue.
I haven't automated the new setup yet, so this was a one-off. I'll bisect it over the weekend.
Actually, !SMP on Arndale has been broken for as long as I've been attempting to boot it, but it used to just lock up silently instead. So while this is a bug to sort out at some point, it's nowhere new.
-Olof