On Fri, 11 Jan 2013 22:40:50 +0100
Stephen Warren [off-list ref] wrote:

On 01/11/2013 05:17 AM, Hiroshi Doyu wrote:

quoted

On Fri, 11 Jan 2013 08:46:25 +0100
Prashant Gaikwad [off-list ref] wrote:

quoted

Add tegra30 clock support based on common clock framework.

Signed-off-by: Prashant Gaikwad <pgaikwad@nvidia.com>

.......

quoted

+static void __init tegra30_pll_init(void)
+{
+       struct clk *clk;
+
+       /* PLLC */
+       clk = tegra_clk_pll("pll_c", "pll_ref", clk_base, pmc_base, 0,
+                           0, &pll_c_params,
+                           TEGRA_PLL_HAS_CPCON | TEGRA_PLL_USE_LOCK,
+                           pll_c_freq_table, NULL);
+       clk_register_clkdev(clk, "pll_c", NULL);
+       clks[pll_c] = clk;

Just I noticed that there are quite many same itegration of:

	clk_register_clkdev(clk, <ID name>, ?);
	clks[<ID>] = clk;

ID == <ID name>

Can any macro/func do the above at once?

To my mind, a macro would obfuscate this fairly simple code, unless
there is a table somewhere the maps <ID> to <ID name> which would allow
saving some code space (I just looked; I don't think there is).

Agree that those 2 lines may be too simple. I thought that ##string##
could do the above. One of the benefit to make them at once is that
we'll get a single entry point to register clkdev in tegra. For
example, if you insert some debug print in it, it would shows all
clkdev registration.

Eventually (later cleanup), I wouldn't be surprised if both parameters
to clk_register_clkdev() became NULL in most cases, since most lookups
are through DT by the end of this series.

Ok.