[Maverick] [PATCH] UBUNTU - ARM: Reverting patch that break mmc init

Amit Kucheria amit.kucheria at canonical.com
Wed Aug 11 07:36:11 UTC 2010


(Argh. Sent it before I was finished)

On 10 Aug 10, Mathieu Poirier wrote:
> On Mon, 2010-08-09 at 16:00 -0700, Tim Gardner wrote:
> > On 08/09/2010 09:58 AM, Mathieu Poirier wrote:
> > > Ensure regulator enable
> > 
> > What does TI think about this? Its not been reverted upstream. Is there 
> > a better way to implement the functionality in the original patch?
> > 
> > rtg
> 
> All,
> 
> The patch hasn't been reverted upstream because no one upstream is
> seeing this issue.  Only Ubuntu compiles with all 4 flags enabled.  I
> tried to discuss my findings on #beagle but only received rude comments
> like "ubuntu and their crazy ways".
 
The #beagle community has a few juveniles in their midst. I strongly suggest
_not_ using them as the first stop for discussions and discussing this on
linux-omap mailing list which is the upstream for the omap kernels.
 
> I thought about leaving the patch intact and reverting only the
> initialization of "host->power_mode" but decided against it for
> consistency reason.  Fixing only the second part of the patch was my
> preferred choice.
> 
> I am well aware that reverting the patch is not the real solution.  On
> the other hand there is no denying the feature is broken and probably
> would not have been accepted upstream had anyone been running their
> system with the flags enabled.

Not completely true.

Nokia ships the N900 kernel with all those feature enabled, but AFAICT they
ship with CONFIG_PREEMPT (low-latency desktop) instead of VOLUNTARY. Have you
tried with that?

> I submitted the patch to allow the mobile team to move forward. The
> system doesn't boot if the mmc card doesn't get initialized properly,
> something that blocks them in all their omap3 endeavours.  

Please mention it in the patch description - something like "HACK: foo bar"
in that case. That way we can review it in the next cycle.

> All that being said, I can submit another patch that will only address
> the "power_mode" initialization and leave the pairing of regulator
> enables/disables intact.  I will also get in touch with the implementer
> of the patch and expose the problem to him.

Yes, please get in touch with Adrian Hunter about the patch. They've
probably never tested with PREEMPT_VOLUNTARY because N900 is not a desktop.

/Amit
-- 
----------------------------------------------------------------------
Amit Kucheria, Kernel Engineer || amit.kucheria at canonical.com
----------------------------------------------------------------------




More information about the kernel-team mailing list