Hi Heinrich,

On Sat, 30 May 2020 at 09:51, Heinrich Schuchardt xypron.glpk@gmx.de wrote:

On 5/30/20 5:24 PM, Heinrich Schuchardt wrote:

...

jAm May 30, 2020 2:44:04 PM UTC schrieb Simon Glass sjg@chromium.org:

...

Hi Heinrich,

I got a failure on this commit:

https://gitlab.denx.de/u-boot/custodians/u-boot-dm/-/jobs/102827

then a retry succeeded:

https://gitlab.denx.de/u-boot/custodians/u-boot-dm/-/jobs/102828

Any ideas on this?

Some time ago I had a problem with buildman where the 4k pipe buffer for console output would happen within a unicode character. I'm not sure if this has anything to do with the problem, or if pytest uses 'bytes' under the hood or not.

Regards, Simon

In the log I see:

[1;32;40mExecuting 'extended text input' succeeded

It is this call that fails:

u_boot_console.restart_uboot()

And QEMU writes:

DRAM: alloc space exhausted

Doesn't this mean that the server does not have enough RAM for all processes running on it? This should be fixed in the Gitlab setup of konsulko-bootbake G37FKVsV.

The message is not from QEMU but from U-Boot:

common/malloc_simple.c:28: log_err("alloc space exhausted\n");

Funny how it only happens sometimes though. I wonder what could be different between runs?

Regards, Simon