Re: squashfs performance regression and readahea

[Xiongwei Song <sxwjean@gmail.com>]

On Sat, May 14, 2022 at 2:12 AM Matthew Wilcox <willy@infradead.org> wrote:
>
> On Sat, May 14, 2022 at 12:43:47AM +0800, Hsin-Yi Wang wrote:
> > > One stupid question, see below code from your patch:
> > >
> > >  +       }
> > >  +
> > >  +       kfree(actor);
> > >  +       return;
> > >  +
> > >  +skip_pages:
> > >
> > > when release page pointers array after pages cached? I don't see
> > > any chance to do that.
> > >
> > actor is not a page pointer. This is allocated from
> > squashfs_page_actor_init() and should be freed after use. Or do you
> > mean skip_pages? There are some situations where we can't decompress
> > the whole block, so we will skip those pages.
>
> I think the concern is that you don't seem to kfree(pages) on this
> exit path.

That's exactly what I wanted to say. Thanks Matthew.