On 04.06.25 14:00, Johannes Weiner wrote:

On Fri, May 30, 2025 at 07:52:28PM +1200, Barry Song wrote:

quoted

On Fri, May 30, 2025 at 9:14 AM Johannes Weiner [off-list ref] wrote:

quoted

On Thu, May 29, 2025 at 04:28:46PM +0100, Matthew Wilcox wrote:

quoted

Barry's problem is that we're all nervous about possibly regressing
performance on some unknown workloads.  Just try Barry's proposal, see
if anyone actually compains or if we're just afraid of our own shadows.

I actually explained why I think this is a terrible idea. But okay, I
tried the patch anyway.

This is 'git log' on a hot kernel repo after a large IO stream:

                                      VANILLA                      BARRY
Real time                 49.93 (    +0.00%)         60.36 (   +20.48%)
User time                 32.10 (    +0.00%)         32.09 (    -0.04%)
System time               14.41 (    +0.00%)         14.64 (    +1.50%)
pgmajfault              9227.00 (    +0.00%)      18390.00 (   +99.30%)
workingset_refault_file  184.00 (    +0.00%)    236899.00 (+127954.05%)

Clearly we can't generally ignore page cache hits just because the
mmaps() are intermittent.

Hi Johannes,
Thanks!

Are you on v1, which lacks folio demotion[1], or v2, which includes it [2]?

[1] https://lore.kernel.org/linux-mm/20250412085852.48524-1-21cnbao@gmail.com/ (local)
[2] https://lore.kernel.org/linux-mm/20250514070820.51793-1-21cnbao@gmail.com/ (local)

The subthread is about whether the reference dismissal / demotion
should be unconditional (v1) or opt-in (v2).

I'm arguing for v2.

+1

-- 
Cheers,

David / dhildenb