On Sat, Jul 05, 2025 at 05:41:17PM +0300, Laurentiu Mihalcea wrote:

On 7/3/2025 6:25 PM, Mark Brown wrote:

quoted

This also appears to be the source of the boot issues I mentioned on the
EVK, affecting ramdisk only boots:

quoted

   https://lava.sirena.org.uk/scheduler/job/1533032

quoted

as well as NFS.  The board seems to get to userspace but then not
respond to serial input, it looks like it's hit something while loading
modules and locked up but ICBW.

OK, this is very odd. I've tried the ramdisk boot using the files from the CI job: Image,
modules, CPIO and my own flash.bin (u-boot and ATF hashes match those shown in
the CI job) on an EVK board and so far no hang.

It may not actually be hung, it may just be hitting long timeouts in
userspace (I notice that all the boots with a NFS root are doing module
autoloading, and the ramdisk ones have tried to bring up the network).

What I've noticed though is that the bridge does indeed not get probed. So, for now,
I think I'm going to submit a patch to enable the driver config in the defconfig so
we can get the Verdin failures out of the way.

If the driver is now required then it does need to be enabled.