On Wed, 2008-07-30 at 12:28 +0200, Manuel Lauss wrote:
> Hi Kevin,
> > 1. I like the interface you added in patch #10. Much better than the
> > old /proc one and flexible enough for a lot of different boards. I
> > agree with your own comment that maybe it should be in the
> > board-specific directories so that people can name the nodes better, but
> > for now I think this is great.
> Thinking about my board here in particular: I need to save/restore a
> few bytes in an FPGA (in the ->enter() callback) and call a few other
> pm related callbacks; the gpio nodes are set internally by or'ing together
> other wake sources (think carddetects, Wake on lan, GSM modem irq, ...).
> So if we want to keep it the way it is now, we should give boards a means
> to disable exposure of each of the "standard" wakesources of the Au1000 chip,
> to provide their own nodes and suspend_ops_t callbacks.
I agree - what I should have said is that the current method is good for
this patch set. It should be enhanced but that should not block the
acceptance of these patches.
> > 2. If I use the db1200_defconfig and enable Power Management
> > (CONFIG_PM), the build fails on the Au1xxx IDE and fb drivers. Are you
> > seeing this too? I see no reason to reject this patch if they don't
> > build with CONFIG_PM, I just want to make sure I'm not doing something
> > wrong.
> The au1200fb failure, yes. I also have a patch to fix it, but it needs
> a bit more love: X for example does not always survive the framebuffer
> suspend/resume cycle (it complains about changed parameters after resume).
Sounds good. Looking forward to it.
> IDE I can't use so didn't test.
We have it on our board so I'll take care of it. I have a partial
solution working already.
> > 3. In my preliminary testing, the system was able to suspend and resume
> > correctly on a DB1200 board. I will do some stress testing in the next
> > couple of days to make sure that it is stable in the long term.
> Very much appreciated!
> Thanks Kevin,
> Manuel Lauss