On 2023/4/25 0:17, Luck, Tony wrote:
This change seems to not related to what you try to fix.
Could this break some other workloads like copying from user address?
Yes, this move MCE_IN_KERNEL_COPYIN set into next case, both COPY and
MCE_SAFE type will set MCE_IN_KERNEL_COPYIN, for EX_TYPE_COPY, we don't
break it.
Should Linux even try to take a core dump for a SIGBUS generated because
the application accessed a poisoned page?
It doesn't seem like it would be useful. Core dumps are for debugging s/w
program errors in applications and libraries. That isn't the case when there
is a poison consumption. The application did nothing wrong.
This patch is still useful though. There may be an undiscovered poison
page in the application. Avoiding a kernel crash when dumping core
is still a good thing.
Thanks for your confirm, and what your option about add
MCE_IN_KERNEL_COPYIN to EX_TYPE_DEFAULT_MCE_SAFE/FAULT_MCE_SAFE type
to let do_machine_check call queue_task_work(&m, msg, kill_me_never),
which kill every call memory_failure_queue() after mc safe copy return?
-Tony
