[Intel-gfx] [PATCH i-g-t] RFC: split PM workarounds into separate lib

[David Weinehall]

On Tue, Dec 08, 2015 at 01:22:14PM +0000, Zanoni, Paulo R wrote:
> Em Ter, 2015-12-08 às 10:50 +0200, David Weinehall escreveu:
> > Since the defaults for some external power management related
> > settings
> > prevents us from testing our power management functionality properly,
> > we have to work around it. Currently this is done from the individual
> > test cases, but this is sub-optimal.  This patch moves the PM-related
> > workarounds into a separate library, and adds some code to restore
> > the
> > previous settings for the SATA link power management while at it.
> > 
> > This patch should be seen as a RFC; there might be other workarounds
> > for external issues that should be moved into the library, and if
> > those
> > workarounds aren't related to power management it might be better to
> > choose a different name for the library.
> 
> I didn't deeply look the implementation, but you have my Acked-by on
> the idea.
> 
> You may also consider adding a function to just run "sudo powertop --
> auto-tune" in addition to the other things, but you can't undo this
> later.
> 
> But in the end, it all depends on your machine. A bad machine will
> never reach the deepest expected PC states. That's the problem when
> automating things...
> 
> Since you're interested in PM, you may also want to look at:
> http://patchwork.freedesktop.org/patch/66392/
> maybe there's some code there that you may want to take.

Thanks! At the very least it seems like a very useful tool that I can
use as an independent test to verify my own tests against. :)


Kind regards, David