On Mon, 2022-05-09 at 13:09 -0700, Josh Poimboeuf wrote:


> To prevent patching, it would need to be some kind of sustained CPU
> activity, rather than a burst.  I guess we haven't seen that show up
> as
> a real-world problem until now.
> 
It's amazing what you see when you have a few million very
busy servers. The problems you think of as "one in a million"
happen all the time :)

> If you're able to identify which kthreads would be problematic, then
> yeah, defining a "transition point" in their outer loops could be an
> option.
> 
I'm in the middle of creating some scripts to gather kpatch
output from all the systems, and then sort and count the
results so we can easily see what the main pain points are.

We'll be back with patches once we have the data in hand :)


-- 
All Rights Reversed.