Re: 32 bit qemu regression from v6.5 tip pull [6c480f222128 x86/alternative: Rewrite optimize_nops() some]

[Paul Gortmaker]

[Re: 32 bit qemu regression from v6.5 tip pull [6c480f222128 x86/alternative: Rewrite optimize_nops() some]] On 29/11/2023 (Wed 09:57) Thomas Gleixner wrote:

[...]

> > Paul, was this regression ever solved? I wonder as I could not find
> > anything with a "Fixes: 6c480f222128 [...]" tag on lore and this thread
> > looks stalled.
> 
> Hmm. I was waiting for the .config file to materialize and for an
> eventual analysis of the alternatives debug output....

Sorry - was off for a week vacation, and was playing catch-up on my
return and 32 bit non-KVM qemu wasn't top on my priority list.

Then I did a batch of runs to get a fail, but without also booting with
"ignore-loglevel" - since the lines are debug and don't show on the
crashed console by default - start over!  :(

The config file and both "good" and "bad" files are there now:

https://bugzilla.yoctoproject.org/show_bug.cgi?id=15230

There are a lot more debug lines in the good/working output!

$ grep alternatives: dmesg-debug-alt-good.txt | wc -l
4868
$ grep alternatives: dmesg-debug-alt.txt | wc -l
19

Paul
--