Re: [PATCH v4 0/2] alpha: complete the direct-mm shootdown fixes

[Matt Turner]

On Sat, Oct 10, 2026 at 16:33 +0200, Magnus Lindholm wrote:
> This is the second of two series. Apply all eight patches of "alpha: fix
> stale TLB translations breaking copy-on-write and writeback" v4 first,
> then these two patches. The base-commit and prerequisite-patch-id lines
> below describe that ordering from for-next at 33465f6ab697.

The first two prerequisite-patch-id lines (d194ee1170a4, 17df30a2da26) do
not match what is in for-next. for-next carries the v5 versions of "load
the MMU context when switch_mm() switches the current task" and "run
check_mmu_context() from finish_arch_post_lock_switch()". The other six
match, and both patches here apply unchanged on top, so this only matters
to someone following the cover letter literally.

> Fresh fork-throughput measurements and an SMP run under compaction are
> still pending for this revision. Targeted Alpha SMP, UP and SMP with
> compaction builds passed, as did a full ES40 CONFIG_COMPACTION=y build.

Thanks for being explicit about this. The current->mm test in
ipi_flush_mm_and_page() is the one change in v4 that no run has
exercised, and the series is in for-next already. I would like to see an
SMP soak with CONFIG_COMPACTION=y and compaction actually triggered
before this goes further.

I went through v4 again and both tags stand. In particular I checked
that the only caller reaching the ev5_switch_mm() store is
context_switch(); kthread_use_mm(), sched_force_init_mm() and
idle_task_exit() all pass current and take the __load_new_mm_context()
branch, so the arch_spin_unlock() argument covers every publication from
that store. Two small comments on patch 2.