Re: [LKP] [gpio] f69e00bd21: unixbench.score -24.2% regression

[Feng Tang]

Hi Linus,

On Wed, Jul 17, 2019 at 05:42:15AM +0800, Linus Walleij wrote:
> On Wed, Jul 10, 2019 at 2:15 PM kernel test robot <rong.a.chen@xxxxxxxxx> wrote:
> 
> > FYI, we noticed a -24.2% regression of unixbench.score due to commit:
> > commit: f69e00bd21aa6a1961c521b6eb199137fcb8a76a ("gpio: mmio: Support two direction registers")
> > https://kernel.googlesource.com/pub/scm/linux/kernel/git/torvalds/linux.git master
> 
> That's pretty bogus since the test doesn't even seem to be using GPIO.
> Further AFAIK Intel chips don't even use that driver.
> 
> Can you provide some rootcausing?

We did some further check, and found this is a false alarm. That some recent 
change to LKP itself has enabled the latencytop for newer kernel(post 5.1-rc1),
and latencytop is known to bring extra system load to scheduling related
benchmarks like unixbench, which caused this -24.7 difference of unixbench
score. Sorry for the noise.

Thanks,
Feng


> If you are using GPIOs from userspace in the test somehow I am
> sure both me and Bartosz would be interested to hear how.
> 
> Yours,
> Linus Walleij
> _______________________________________________
> LKP mailing list
> LKP@xxxxxxxxxxxx
> https://lists.01.org/mailman/listinfo/lkp