Re: [PATCH 1/2] mm: swap_cgroup: allocate swap_cgroup map using vcalloc()

From: Roman Gushchin
Date: Fri Nov 15 2024 - 18:00:47 EST


On Fri, Nov 15, 2024 at 12:17:43PM -0800, Shakeel Butt wrote:
> On Fri, Nov 15, 2024 at 07:02:28PM +0000, Roman Gushchin wrote:
> > Currently swap_cgroup's map is constructed as a vmalloc()'s-based
> > array of pointers to individual struct pages. This brings an
> > unnecessary complexity into the code.
> >
> > This patch turns the swap_cgroup's map into a single space
> > allocated by vcalloc().
> >
> > Signed-off-by: Roman Gushchin <roman.gushchin@xxxxxxxxx>
>
> Acked-by: Shakeel Butt <shakeel.butt@xxxxxxxxx>

Thank you for reviewing the patchset!

>
> [...]
> > @@ -215,19 +169,8 @@ void swap_cgroup_swapoff(int type)
> > mutex_lock(&swap_cgroup_mutex);
> > ctrl = &swap_cgroup_ctrl[type];
> > map = ctrl->map;
> > - length = ctrl->length;
> > ctrl->map = NULL;
> > - ctrl->length = 0;
> > mutex_unlock(&swap_cgroup_mutex);
> >
> > - if (map) {
> > - for (i = 0; i < length; i++) {
> > - struct page *page = map[i];
> > - if (page)
> > - __free_page(page);
> > - if (!(i % SWAP_CLUSTER_MAX))
> > - cond_resched();
> > - }
> > - vfree(map);
> > - }
> > + kvfree(map);
>
> Any reason to use kvfree() instead of just vfree()?

No, you're right, vfree() is a better choice here.

Andrew, do you prefer a v2 or it's easier to master a fix-up?

Thank you!