Hi Uwe,

On Tue, Jan 04, 2011 at 02:58:18PM +0100, Uwe Kleine-K?nig wrote:

Hello Shawn,

On Tue, Jan 04, 2011 at 05:24:12PM +0800, Shawn Guo wrote:

quoted

Register clocks fec.0 and fec.1 for dual fec support.

Signed-off-by: Shawn Guo <redacted>
---
Changes for v2:
 - Rebase the patch against patch below which gets fundamental
   clocks explicitly called in clk_enable.
   [PATCH v4] ARM: mxs: Change duart device to use amba-pl011
 - Register clocks fec.0 and fec.1 respectively than use wildcard
   NULL for both instances.

 arch/arm/mach-mxs/clock-mx28.c |    1 +
 1 files changed, 1 insertions(+), 0 deletions(-)

diff --git a/arch/arm/mach-mxs/clock-mx28.c b/arch/arm/mach-mxs/clock-mx28.c
index f20b254..f79587b 100644
--- a/arch/arm/mach-mxs/clock-mx28.c
+++ b/arch/arm/mach-mxs/clock-mx28.c

@@ -607,6 +607,7 @@ static struct clk_lookup lookups[] = {
 	/* for amba-pl011 driver */
 	_REGISTER_CLOCK("duart", NULL, uart_clk)
 	_REGISTER_CLOCK("fec.0", NULL, fec_clk)
+	_REGISTER_CLOCK("fec.1", NULL, fec_clk)

Did you verify this works as intended?  I didn't and consider it
possible that after

	clk_enable("fec.0");
	clk_enable("fec.1");
	clk_disable("fec.1");

the clk is off.  Probably it just works, just want to prevent hard to
find bugs ...

The clk should not be off.  Looking at _REGISTER_CLOCK below, we are
registering both fec.0 and fec.1 on the same one fec_clk.  IOW, the
usecount of the clk is working to control the on/off of the clk.

#define _REGISTER_CLOCK(d, n, c) \
	{ \
		.dev_id = d, \
		.con_id = n, \
		.clk = &c, \
	},

-- 
Regards,
Shawn