Re: [PATCH v2 4/6] debugobjects: Don't start fill if there are remaining nodes locally
From: Thomas Gleixner
Date: Mon Sep 09 2024 - 10:35:53 EST
On Mon, Sep 09 2024 at 21:51, Leizhen wrote:
> On 2024/9/9 20:10, Thomas Gleixner wrote:
>> So something like the uncompiled below should reduce lock contention
>> significantly with a reasonable safety net.
>
> It looks very good. Especially flexible use of 'state' and 'cpus_allocating'.
> In this way, there is almost no conflict of lock 'pool_lock', and the more
> cores, the less possible conflict.
>
> Hi Thomas Gleixner:
> Do you plan to post this patch? But this patch will conflict with my patch 5/6.
> If you're going to merge my patch 5/6, hopefully yours will be applied
> after mine.
I'm short of cycles, so the best way is when you pick it up and
integrate it instead of 4/6 and post a v3.
> By the way, Do you have time to review the patches in the link below?
> https://lkml.org/lkml/2024/9/4/1094
Please use lore.kernel.org links. I've seen it but did not come around
to it yet because I was staring at this one. I'm single threaded :)
Thanks,
tglx