Re: [patch 056/146] mm: rearrange madvise code to allow for reuse

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On Sat, Jan 15, 2022 at 6:16 AM Linus Torvalds
<torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Sat, Jan 15, 2022 at 12:06 AM Andrew Morton
> <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> >
> > Speed up fork() by up to 40% by refcounting the anon vma name field.
>
> What?
>
> No.
>
> This doesn't speed up anything at all.
>
> The refcounting of the anon-vma name field avoids a 40% regression
> that comes from adding the field in the first place, but this commit
> message makes it sound like this series is speeding up fork() by 40%.
>
> I don't mind the series, but I absolutely mind these kinds of horribly
> misleading commit messages.
>
> This is literally the first commit in the series - and the series in
> no way improves performance by 40% in the end, it just first makes it
> worse, and then fixes the regression.
>
> I can speed up any function by a thousand percent - if I'm just
> allowed to make it horribly slow first, and only count the final
> speedup win when I remove the overhead of the garbage I added.

Thanks for fixing it! I see in your tree it is changed to:
"Avoid performance regression of the new anon vma name field refcounting it."
Please let me know if anything else is left to be done here.

>
>                Linus



[Index of Archives]     [Kernel Archive]     [IETF Annouce]     [DCCP]     [Netdev]     [Networking]     [Security]     [Bugtraq]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Security]     [Linux RAID]     [Linux SCSI]

  Powered by Linux