Re: [PATCH 1/2] mmc: agressive clocking framework v8

[Chris Ball <cjb@laptop.org>]

On Wed, Nov 24, 2010 at 02:15:57PM +0100, Linus Walleij wrote:
> > [   10.136067] mmc0: frequency set to 0 in disable function, this means the clock is already disabled.
> 
> It didn't use to be like that back when I first wrote the patch, but
> it seems to be like
> that nowadays. So yes.

Hm, okay.  Printing the same message seven times in a second is a little
undesirable, but we could leave it in until we're more confident that
the code's all working well.  It might help if the message made it
clearer that nothing is wrong, and this is an expected event.

> Did the gating work by the way? No strangeness because of it?

Yep, no strangeness so far.

> It's pr_debug() already in host.c:
> pr_debug("%s: frequency set to 0 in disable function, "
> 	 "this means the clock is already disabled.\n",
> 	 mmc_hostname(host));
> 
> pr_debug is only printed when -DDEBUG is set, which is what
> MMC_DEBUG=y does isn't it?

Ah, apologies, I understand now.  My linux-next build was too old for
"mmc: protect against clock-gating races", which does s/pr_err/pr_debug/
on that printk, so it was coming out unconditional on MMC_DEBUG.  I've
moved up to today's linux-next now.

-- 
Chris Ball   <cjb@laptop.org>   <http://printf.net/>
One Laptop Per Child