* Tero Kristo [off-list ref] [160628 23:19]:

On 28/06/16 09:57, Tony Lindgren wrote:

quoted

Hi,

* Tero Kristo [off-list ref] [160613 12:07]:

quoted

The new API avoids the need to add clock aliases for most of the clocks,
should use of it is preferred. Many of the existing clock aliases are
only created because of hwmod data.

Signed-off-by: Tero Kristo <redacted>
---
  arch/arm/mach-omap2/omap_hwmod.c | 6 +++---
  1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/arch/arm/mach-omap2/omap_hwmod.c b/arch/arm/mach-omap2/omap_hwmod.c
index 83cb527..0ea869c 100644
--- a/arch/arm/mach-omap2/omap_hwmod.c
+++ b/arch/arm/mach-omap2/omap_hwmod.c

@@ -786,7 +786,7 @@ static int _init_main_clk(struct omap_hwmod *oh)
  	if (!oh->main_clk)
  		return 0;

-	oh->_clk = clk_get(NULL, oh->main_clk);
+	oh->_clk = ti_clk_get(oh->main_clk);
  	if (IS_ERR(oh->_clk)) {
  		pr_warn("omap_hwmod: %s: cannot clk_get main_clk %s\n",
  			oh->name, oh->main_clk);

After thinking about this for a while I think code outside TI specific
clock implementation should use just clk_get(). Otherwise we create more
dependencies to move code to live under drivers subdirectory.

Can't we have clk_get() call a SoC specific helper function if clock
is not found?

That is for Mike / Stephen to answer. Implementing something like this
sounds trivial, I could just hook up the ti_clk_get being called here.
However, any attempted changes to the clkdev implementation have been shot
down so far, thus I did this.

Probably best to send a minimal patch for that for review. We want to
have the clock framework consumers using the standard interface.

The problem is still the clocks which are just referred via a name only, not
via a DT node.

OK

Regards,

Tony