[PATCH v3 5/6] ARM: davinci: Add support for PRUSS on DA850

On 10/4/2012 10:05 PM, Matt Porter wrote:

On Thu, Oct 04, 2012 at 05:52:45PM +0530, Sekhar Nori wrote:

quoted

On 10/3/2012 8:25 PM, Matt Porter wrote:

quoted

+static struct clk pruss_clk = {
+	.name		= "pruss",
+	.parent		= &pll0_sysclk2,
+	.lpsc		= DA8XX_LPSC0_PRUSS,
+};
+
 static struct clk uart0_clk = {
 	.name		= "uart0",
 	.parent		= &pll0_sysclk2,

@@ -378,6 +384,7 @@ static struct clk_lookup da850_clks[] = {
 	CLK(NULL,		"tptc1",	&tptc1_clk),
 	CLK(NULL,		"tpcc1",	&tpcc1_clk),
 	CLK(NULL,		"tptc2",	&tptc2_clk),
+	CLK(NULL,		"pruss",	&pruss_clk),

This is actually incorrect since we should use device name rather than
con_id for matching the clock. If there is just one clock that the
driver needs, connection id should be NULL. Looking at the driver now,
the clk_get() call seems to pass a valid device pointer. So, I wonder
how you are able to look up the clock even with a NULL device name.

I doublechecked the clk_get() find implentation to confirm that I indeed
did get lucky here. Since pruss has only one instance and the clk_find()
implementation looks for the best found match, it's able to find the
clock just by the con_id, though it's not the "best possible" match in
the find implementation...it's the "best found" match.. As you noted,
this is incorrect though for a clock expected to be used from a driver
context so I will address this in the update.

Thanks for debugging and clarifying.

Regards,
Sekhar