Re: [PATCH] mm/swapfile.c: select the swap device with default priority round robin

[Baoquan He <bhe@redhat.com>]

On 09/26/25 at 09:46pm, Chris Li wrote:
> On Fri, Sep 26, 2025 at 8:31 AM Baoquan He <bhe@redhat.com> wrote:
> > On 09/25/25 at 11:25am, Chris Li wrote:
> > > Just curious, is setting to "-1" matches to kernel behavior before
> > > a2468cc9bfdf, if not what is the behavior before a2468cc9bfdf.
> >
> > It should be like below. It's not a real output, I made the data to show
> > what it looks like.
> >
> > # swapon
> > NAME       TYPE      SIZE  USED PRIO
> > /dev/zram0 partition  16G 15.8G   -1
> > /dev/zram1 partition  16G    0B   -2
> > /dev/zram2 partition  16G    0B   -3
> > /dev/zram3 partition  16G    0B   -4
> >
> > I just apply this patch and set the priority to emulate the kerel
> > behavirour before a2468cc9bfdf. In kernel before a2468cc9bfdf, it sets
> > priority to swap device from -1 downwards. There's only one
> > swap_avail_head plist for all CPUs. The behaviour is very much like below:
> >
> > [root@hp-dl385g10-03 ~]# swapon
> > NAME       TYPE      SIZE  USED PRIO
> > /dev/zram0 partition  16G    0B    0
> > /dev/zram1 partition  16G    0B    1
> > /dev/zram2 partition  16G    0B    2
> > /dev/zram3 partition  16G 14.3G    3
> 
> I see, in that case it is not a simple revert. It is revert plus a
> change to make the default to use round robin. I suggest you split the
> patch into two parts, one is the conceptual clean revert, back to the
> pre a2468cc9bfdf without behavior change. Then the second one is
> changed to the round robin. That will make your behavior change more
> obvious.

Make senses, I will make v3 to split it.

> 
> I suspect that even pre a2468cc9bfdf, the round robin might outperform
> per node priority already. Which means there exists a much simpler
> solution all alone. Even though I am curious, I am not demanding the
> answer from you. Your test data against the latest kernel, which shows
> great performance improvement is good enough. We don't have to test
> that really old kernel.

Yeah, I agree. Just nobody realize it then.