Re: [PATCH v4 1/1] mm: vmscan: Reduce throttling due to a failure to make progress

[Thorsten Leemhuis]

On 31.12.21 20:22, Linus Torvalds wrote:
> On Fri, Dec 31, 2021 at 11:21 AM Linus Torvalds
> <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>>
>> Pushed out as 1b4e3f26f9f7 ("mm: vmscan: Reduce throttling due to a
>> failure to make progress")

Thx.

> .. and I _think_ this empties the regzbot queue for this release, Thorsten. No?

Well, it was the regression that bothered me most. But there are still a
few out there that got introduced this cycle.

There is a regression in RDMA/mlx5 introduced in v5.16-rc5:
https://lore.kernel.org/lkml/f298db4ec5fdf7a2d1d166ca2f66020fd9397e5c.1640079962.git.leonro@xxxxxxxxxx/
https://lore.kernel.org/all/EEBA2D1C-F29C-4237-901C-587B60CEE113@xxxxxxxxxx/
A fix is available, but got stuck afaics:
https://lore.kernel.org/lkml/f298db4ec5fdf7a2d1d166ca2f66020fd9397e5c.1640079962.git.leonro@xxxxxxxxxx/
And I only noticed just now: a revert was also discussed, but not performed:
https://lore.kernel.org/all/20211222101312.1358616-1-maorg@xxxxxxxxxx/
Will let Greg know, seems the commit got backported to 5.15.

s0ix suspend broke for some AMD machines, Alex and Mario are busy
looking into it:
https://gitlab.freedesktop.org/drm/amd/-/issues/1821
https://bugzilla.kernel.org/show_bug.cgi?id=215436

Alex is also dealing with another issue where the screen contents now
get restored after some input events:
https://bugzilla.kernel.org/show_bug.cgi?id=215203

There still seems to be a performance regression that Josef and Valentin
try hard to pin down without much success for weeks now:
https://lore.kernel.org/lkml/87tuf07hdk.mognet@xxxxxxx/

And there one more report, but it might be a follow-up error due to
another regression:
https://lore.kernel.org/linux-pm/52933493.dBzk7ret6Y@geek500/