Re: Possible regression with the FW643 chipset

[Jonathan W. <jw...@ju...>]

On Fri, Jun 20, 2025 at 09:32:57AM +0000, AreYouLoco? wrote:
> Good you reported it even before I did.

Thanks for adding your input.

> I am willing to build kernels and test patches.

Oh ok, that's awesome!  Would you be willing to do what's called a "git
bisect" to discover the kernel that faulted?  In simplified terms the
process is this:

 1. Identify a kernel that works (6.6 in our case) and one that doesn't
    (6.12 for the present situation).  Call one that works version W and the
    one that doesn't F.

 2. Choose a kernel half way between W and F.  Build and test.

 3. If the new kernel works, set W to that kernel version.
    If the new kernel fails, set F to that kernel version.

 4. Go to step 2 until the earliest failing version is found.

In practice the "versions" used here are git commits but the principle
remains the same.  The git revision control system also greatly simplifies
this by automating the management of the versions.  Steps 2-4 come down
to answering a "did it work" question and recompiling/testing a new kernel.

If you do a web search on "linux kernel git bisect" you'll find many
tutorials which walk you through the process.

It would be *really* appreciated if you could give this a go as it should
identify the precise kernel commit that caused the problem.  I'm happy to
assist off-list if required.

Regards
  jonathan