Re: [PATCH v14 23/26] mm: zswap: Tie per-CPU acomp_ctx lifetime to the pool.

["Kanchana P. Sridhar" <kanchanapsridhar2026@gmail.com>]

On Wed, 4 Feb 2026 16:29:35 +0000, Yosry Ahmed wrote:
[...]
> LGTM with a small nit below:
> 
> Acked-by: Yosry Ahmed <yosry.ahmed@linux.dev>
>

Thanks!

[...]
> > +static void acomp_ctx_dealloc(struct crypto_acomp_ctx *acomp_ctx)
> > +{
> > +	if (IS_ERR_OR_NULL(acomp_ctx))
> > +		return;
> > +
> > +	if (!IS_ERR_OR_NULL(acomp_ctx->req))
> > +		acomp_request_free(acomp_ctx->req);
> > +
> > +	if (!IS_ERR_OR_NULL(acomp_ctx->acomp))
> > +		crypto_free_acomp(acomp_ctx->acomp);
> 
> Should we set acomp_ctx->req, acomp_ctx->acomp, and acomp_ctx->buffer to
> NULL here?
> 
> zswap_cpu_comp_prepare() uses NULL to detect that we need to initialize
> acomp_ctx.
>

I don't think this is needed, because with the changes made in this
patch, there is no scenario in which zswap_cpu_comp_prepare() will be
called *after* acomp_ctx_dealloc() deletes these resources.

zswap_cpu_comp_prepare() is called only during pool creation or after a
subsequent CPU hotplug operation (for a CPU getting onlined when it was
offline at pool creation time, or, due to an offline-online
transition of a CPU that was online at pool creation time). In both
these cases, the per-CPU acomp_ctx members are either NULL or valid.

Thanks,
Kanchana

> > +
> > +	kfree(acomp_ctx->buffer);
> > +}
> > +