Re: [PATCH] target/ppc: Stop vCPU thread before calling parent_unrealize

[Shivang Upadhyay <shivangu@linux.ibm.com>]

On Thu, 2026-09-24 at 20:12 +0530, Amit Machhiwal wrote:
> The fix is correct and the commit message is well written.  The ASan
> trace makes
> the race unambiguous: tlb_destroy() frees fast->table and desc-
> >fulltlb while
> the MTTCG thread is still running tlb_mmu_resize_locked() on the same
> structures
> via the queued tcg_commit_cpu work.
> 
> Reviewed-by: Amit Machhiwal <amachhiw@linux.ibm.com>
> 
Thanks for review.

> Minor: I think a short comment above cpu_remove_sync() would make the
> ordering
> constraint self-documenting, e.g.:
> 
>     /*
>      * Stop and join the vCPU thread before tearing down accelerator
>      * state (TLB tables etc.) to prevent the running thread from
>      * accessing freed resources.
>      */
>     cpu_remove_sync(CPU(cpu));

Seems this is already pulled via misc tree.

I plan to work on Philippe's suggestion to pull all cpu_remove_sync()
calls into cpu_common_unrealize(), that should take care of all such
caveats.

~Shivang.