2012/10/12 Daniel Vetter <daniel.vetter at ffwll.ch>: > On Fri, Oct 12, 2012 at 7:17 PM, Paulo Zanoni <przanoni at gmail.com> wrote: >> Ok, so please do a final test: try to write something to those >> "must-be-preserved" bits and check if the values stay or not. If after >> writing 1 to bits 17-18, 20-23 you read 0, then you have my >> Reviewed-by: Paulo Zanoni <paulo.r.zanoni at intel.com>. I still do plan >> to test the 6 patches of the series on hsw later btw. > > I've tested a few bits on a few machines, and some do stick and some > cause ... strange things (like a seemingly random set of other > register bits also being set). I guess that's not the answer you've > been looking for. I'd still prefer if we just try to clear things, > worst case we can easily read out the current state of these reserved > bits at boot up and restore them at resume time. But I'd really prefer > if we reduce our reliance on the boot-up state from the bios as much > as possible. And in this case here the only way to figure things out > is to merge it (it's really early for 3.8 anyway) and see what > happens. Well, ok then... Both the old and the new version don't exactly follow the spec, but I guess writing zero makes more sense than enabling random bits, especially because you said you checked and your machines actually have zeros on those bits. Reviewed-by: Paulo Zanoni <paulo.r.zanoni at intel.com> > > Cheers, Daniel > -- > Daniel Vetter > Software Engineer, Intel Corporation > +41 (0) 79 365 57 48 - http://blog.ffwll.ch -- Paulo Zanoni
