kgdb: Flush console before entering kgdb on panic
authorDouglas Anderson <dianders@chromium.org>
Tue, 22 Aug 2023 20:19:46 +0000 (13:19 -0700)
committerDaniel Thompson <daniel.thompson@linaro.org>
Sun, 15 Oct 2023 17:07:50 +0000 (18:07 +0100)
commitdd712d3d45807db9fcae28a522deee85c1f2fde6
tree1ad1c32118cb9baa41194cad6760fd7b6faa0878
parent94f6f0550c625fab1f373bb86a6669b45e9748b3
kgdb: Flush console before entering kgdb on panic

When entering kdb/kgdb on a kernel panic, it was be observed that the
console isn't flushed before the `kdb` prompt came up. Specifically,
when using the buddy lockup detector on arm64 and running:
  echo HARDLOCKUP > /sys/kernel/debug/provoke-crash/DIRECT

I could see:
  [   26.161099] lkdtm: Performing direct entry HARDLOCKUP
  [   32.499881] watchdog: Watchdog detected hard LOCKUP on cpu 6
  [   32.552865] Sending NMI from CPU 5 to CPUs 6:
  [   32.557359] NMI backtrace for cpu 6
  ... [backtrace for cpu 6] ...
  [   32.558353] NMI backtrace for cpu 5
  ... [backtrace for cpu 5] ...
  [   32.867471] Sending NMI from CPU 5 to CPUs 0-4,7:
  [   32.872321] NMI backtrace forP cpuANC: Hard LOCKUP

  Entering kdb (current=..., pid 0) on processor 5 due to Keyboard Entry
  [5]kdb>

As you can see, backtraces for the other CPUs start printing and get
interleaved with the kdb PANIC print.

Let's replicate the commands to flush the console in the kdb panic
entry point to avoid this.

Signed-off-by: Douglas Anderson <dianders@chromium.org>
Link: https://lore.kernel.org/r/20230822131945.1.I5b460ae8f954e4c4f628a373d6e74713c06dd26f@changeid
Signed-off-by: Daniel Thompson <daniel.thompson@linaro.org>
kernel/debug/debug_core.c