search

seL4 / util_libs

Other
53 stars 83 forks source link

libplatsupport: use KernelPlatformZynqmp #146

Closed axel-h closed 1 year ago

axel-h commented 1 year ago

Follow-up from https://github.com/seL4/util_libs/pull/142#discussion_r1013337980

chrisguikema commented 1 year ago

Looks good to me!

lsf37 commented 1 year ago

Looks good, just checking that the tests work before I merge.

axel-h commented 1 year ago

Can we merge this now, the ZynqMP failure seem to be cause by something else?

lsf37 commented 1 year ago

Restarted the test to check.

axel-h commented 1 year ago

It seems there is a general problem with starting the board now? Building works, but the image can't be loaded:

 '/tmp/tmp.PTTuIEl9bv' -> '/tftpboot/zcu102/sel4-image'
  Couldn't change mode of /tftpboot/zcu102/sel4-image: [Errno 1] Operation not permitted: '/tftpboot/zcu102/sel4-image'
lsf37 commented 1 year ago

It seems there is a general problem with starting the board now? Building works, but the image can't be loaded:

 '/tmp/tmp.PTTuIEl9bv' -> '/tftpboot/zcu102/sel4-image'
  Couldn't change mode of /tftpboot/zcu102/sel4-image: [Errno 1] Operation not permitted: '/tftpboot/zcu102/sel4-image'

The mode change error is normal (all logs have it, we should really remove that command), but the subsequent boot failure looks like the board is not reachable, yes.

axel-h commented 1 year ago

Not sure why IMX8MM_EVK_debug_gcc_32 failed in the middle, is it a power spike that cause a reset?

Thu, 27 Apr 2023 15:05:35 GMT  Running test SERSERV_PARENT_010 (Test a series of unexpected input values to write())
Thu, 27 Apr 2023 15:05:35 GMT  serial_server_write@clientapi.c:221 Serserv Client: printf: NULL passed for required arguments.
Thu, 27 Apr 2023 15:05:35 GMT   Is connection handle valid?
Thu, 27 Apr 2023 15:05:35 GMT  serial_server_write@clientapi.c:221 Serserv Client: printf: NULL passed for required arguments.
Thu, 27 Apr 2023 15:05:35 GMT   Is connection handle valid?
Thu, 27 Apr 2023 15:05:35 GMT  Test SERSERV_PARENT_010 passed
Thu, 27 Apr 2023 15:05:36 GMT  
Thu, 27 Apr 2023 15:05:36 GMT  U-Boot SPL 2018.03 (Dec 19 2018 - 17:39:10 +0800)
Thu, 27 Apr 2023 15:05:36 GMT  power_bd71837_init

Next test should have been SYNC001

lsf37 commented 1 year ago

Not sure why IMX8MM_EVK_debug_gcc_32 failed in the middle, is it a power spike that cause a reset?

@wom-bat has been trying to figure this out for months now, it's just frequent enough to fail in every other larger test, but too infrequent to reproduce manually. We thought it was temperature, but that does seem to be controlled now. It only seems to be happening when load is high on multiple boards, so maybe it is something with power (when you re-run the one board in isolation it almost always passes everything, i.e. it's not completely gone, but frequency of failure drops strongly). Extremely annoying to debug.

Might just be that some component on the board is old and failing. We have new imx8mm boards on order (also just to have more of them in CI), but delivery time is still a few months, I think.

In any case, it's not related to this PR and Ok to ignore the failure for this.