Re: [PATCH 0/2] pipe: Fixes [ver #2]

[Linus Torvalds]

On Sun, Dec 8, 2019 at 8:46 AM Akemi Yagi <toracat@xxxxxxxxxx> wrote:
>
> I forgot to mention that the aforementioned bug in make was originally
> reported for Fedora. They now have a patched version of make
> (make-4.2.1-15.fc32) in Rawhide.

Yes, as mentioned I do expect it's some jobserver interaction - I've
seen bad jobserver behavior before. But my 'make' hasn't been upgraded
since May, as far as I can tell from my logs, so the huge performance
regression was new.

So the huge slowdown is real, and was triggered by the kernel changes.
But yes, it's entirely possible - even likely - that it's due to a
jobserver bug that ends up being sensitive to scheduling.

We do treat even "user space is buggy and shouldn't have done that" as
a regression.

It's annoying. Timing-related issues are some of the worst to debug.
Particularly when (at least I) primarily debug by looking at the code
and trying to see what's wrong that way.

                Linus