Hi Andrew and Ying，
On 2024/12/1 13:37, Huang, Ying wrote:
Hi, Andrew,
Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> writes:
On Thu, 31 Oct 2024 16:39:53 +0800 "Huang, Ying" <ying.huang@xxxxxxxxx> wrote:
However, on ARM64, it does introduce some regression for clearing pages
from end to start. That needs to be addressed. I guess that the
regression can be resolved via using more clearing from start to end
(but not all). For example, can you take a look at the patch below?
Which uses the similar framework as before, but clear each small trunk
(mpage) from start to end. You can adjust MPAGE_NRPAGES to check when
the regression can be restored.
WARNING: the patch is only build tested.
I'm holding this patch in mm-unstable because it's unclear (to me) that
this regression has been adequately addressed?
Yes. This patch isn't ready to be merged yet. More works are needed to
root cause the problem and implement the proper fix. I guess that
Kefeng will continue to work on this.
Sorry for the late, I've been buried in other tasks.
The two bugfix patches in the next should be merged firstly, which is
not related about the performance issue.
84ddd1450383 mm: use aligned address in copy_user_gigantic_page()
9fbd869b19b8 mm: use aligned address in clear_gigantic_page()
and according to the suggestion of Ying, I am trying to fix the
performance issue on arm64(will in new series), but I need to check no
regression on other arm64's machine, will send the new patches once
the test is complete.
So Andrew, please help to pick up the above two bugfix patches to
v6.13-rcX if no objection.
---
Best Regards,
Huang, Ying
