https://bugs.kde.org/show_bug.cgi?id=522307

--- Comment #8 from [email protected] ---
Thanks for the detailed guidance on points 1-4 above — I really
appreciate the depth of the suggestions. I have to be upfront though:
the gdb breakpoint script, debug symbol setup, and correlating
serial/reply_serial pairs across a multi-million-line dbus-monitor
capture is beyond what I'm able to reliably do myself at this point.
I gave it a try (installed gdb, attempted debuginfod/debug repos) but
didn't get clean, symbol-resolved results, and I don't want to send
you noisy or misleading data.

What I *can* offer, and which I think is solid evidence on its own:

I rolled the system back via a Btrfs/snapper snapshot taken
immediately before pam_kwallet6 was updated from 6.7.0 to 6.7.1, and
have now locked the package at that version:

  pam_kwallet6-6.7.0-1.1.x86_64
  pam_kwallet6-common-6.7.0-1.1.noarch
  kwalletd6-6.27.0-1.1.x86_64 (unchanged)
  kf6-kwallet-6.27.0-1.1.x86_64 (unchanged)

With pam_kwallet6 back at 6.7.0 (and kwalletd6 untouched at 6.27.0),
suspend/resume + screen unlock now works correctly with no CPU loop,
across repeated real-world use (not just a one-off test boot). This
points fairly specifically at something introduced between
pam_kwallet6 6.7.0 and 6.7.1 as the trigger for this behavior in
kwalletd6.

I'm not able to commit to further interactive debugging sessions
(gdb, symbol resolution, raw D-Bus trace correlation) at this point —
it's a bit beyond what I can reliably do on my own. Thanks again for
taking the time to look into this; hopefully the version regression
info is useful in narrowing things down.

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to