sof-logger: fail when an invalid address is found in the DMA trace

[marc-hb]

Let's test DMA corruption bug #5120 for real.

Signed-off-by: Marc Herbert marc.herbert@intel.com

[marc-hb]

As just explained in #5329 (comment), the corruption left _ZEPHYR tests. Now there's only something "milder" affecting CML... EDIT: filed as #5345

https://sof-ci.01.org/sofpr/PR5340/build11976/devicetest

[marc-hb]

SOFCI TEST

[lgirdwood]

@marc-hb still draft ?

[marc-hb]

Yes still a draft, let's focus first on issues that cause the logs to be entirely empty (#5410 etc.)

PS: please don't look at drafts! They have a grey icon and can be ignored with a simple draft:falsehttps://github.com/thesofproject/sof/pulls?q=is%3Apr+is%3Aopen+draft%3Afalse

[sys-pt1s]

Can one of the admins verify this patch?

[marc-hb]

@dbaluta , @paulstelian97 would like to take over this? Intel doesn't use the sof-logger anymore in the main branch.

[paulstelian97]

I'll consider taking over this patch, but I also want to get some understanding on what logging tools Intel will use from now on, as well as in how said tooling can be ported so we can use it. We'll only use sof-logger until the better tools are confirmed working for us (right now the state is that we haven't gotten them to work just yet, but sof-logger also has reliability issues when we test our Zephyr builds).

Once the weekend passes I'll bring it up in an internal meeting with Daniel and see what we should do.

[lgirdwood]

@paulstelian97 NXP should be migrating to Zephyr tooling for logs

@thesofproject/amd @thesofproject/mediatek fyi this is up to you guys on how/if you want to handle ?

[paulstelian97]

We should migrate to the Zephyr tools but that's something to be done further on in the future. For the moment we still have Xtos builds and sof-logger. I'll take over helping out with the logger until we stop using it.