[PATCHv2 3/6] ARM: socfpga: Add support to gate peripheral clocks

[Dinh Nguyen]

Hi Pavel,

On 05/17/2013 06:09 AM, Pavel Machek wrote:
> Hi!
> 
>> From: Dinh Nguyen <dinguyen at altera.com>
>>
>> Add support to gate the clocks that directly feed peripherals. For clocks
>> with multiple parents, add the ability to determine the correct parent,
>> and also set parents. Also add support to calculate and set the clocks'
>> rate.
>>
>> Signed-off-by: Dinh Nguyen <dinguyen at altera.com>
>> Reviewed-by: Pavel Machek <pavel at denx.de>
>> CC: Mike Turquette <mturquette at linaro.org>
>> CC: Arnd Bergmann <arnd at arndb.de>
>> CC: Olof Johansson <olof at lixom.net>
>> Cc: Pavel Machek <pavel at denx.de>
>> CC: <linux at arm.linux.org.uk>
>>
>> v2:
>> - Fix space/indent errors
>> - Add streq for strcmp == 0
>> ---
> 
>> +static u8 socfpga_clk_get_parent(struct clk_hw *hwclk)
>> +{
>> +	u32 l4_src;
>> +	u32 perpll_src;
>> +	u8 parent;
>> +
>> +	if (streq(hwclk->init->name, SOCFPGA_L4_MP_CLK)) {
>> +		l4_src = readl(clk_mgr_base_addr + CLKMGR_L4SRC);
>> +		l4_src &= 0x1;
>> +		parent = l4_src;
>> +	} else if (streq(hwclk->init->name, SOCFPGA_L4_SP_CLK)) {
>> +		l4_src = readl(clk_mgr_base_addr + CLKMGR_L4SRC);
>> +		l4_src = ((l4_src & 0x2) >> 1);
>> +		parent = l4_src;
>> +	} else {
>> +		perpll_src = readl(clk_mgr_base_addr + CLKMGR_PERPLL_SRC);
>> +		if (streq(hwclk->init->name, SOCFPGA_MMC_CLK))
>> +			perpll_src &= 0x3;
>> +		else if (streq(hwclk->init->name, SOCFPGA_NAND_CLK) ||
>> +			streq(hwclk->init->name, SOCFPGA_NAND_X_CLK))
>> +			perpll_src = ((perpll_src & 0xC) >> 2);
>> +		else /*QSPI clock */
>> +			perpll_src = ((perpll_src & 0x30) >> 4);
>> +		parent = perpll_src;
> 
> I really don't like the "else" here. If some new clock name appears in
> hwclk->init->name or if there's problem somewhere, it will just
> silently operate wrong clock. WARN_ON(!streq(...))?
> 
> (I tried to append patch to previous email, with suggested
> cleanups....)
> 

Ok, will fix in rev3.

>> +		} else {/*QSPI clock */
>> +			src_reg &= ~0x30;
>> +			src_reg |= (parent << 4);
>> +		}
> 
> Same here. (And there should be space after /* ).
>

OK...

>> +	socfpga_clk = kzalloc(sizeof(*socfpga_clk), GFP_KERNEL);
>> +	if (WARN_ON(!socfpga_clk))
>> +		return;
>> +
>> +	rc = of_property_read_u32_array(node, "clk-gate", clk_gate, 2);
>> +	if (rc)
>> +		clk_gate[0] = 0;
>> +
>> +	rc = of_property_read_u32(node, "fixed-divider", &fixed_div);
>> +	if (rc)
>> +		socfpga_clk->fixed_div = 0;
>> +	else
>> +		socfpga_clk->fixed_div = fixed_div;
>> +
>> +	if (clk_gate[0]) {
>> +		socfpga_clk->hw.reg = clk_mgr_base_addr + clk_gate[0];
>> +		socfpga_clk->hw.bit_idx = clk_gate[1];
>> +
>> +		gateclk_ops.enable = clk_gate_ops.enable;
>> +		gateclk_ops.disable = clk_gate_ops.disable;
>> +	}
>> +
> 
> Could if (clk_gate[]) be moved one block above? It uses
> partially-initialized array, so it would be nice to have code as clear
> as possible.
> 

OK..

> 
>> +	of_property_read_string(node, "clock-output-names", &clk_name);
>> +
>> +	init.name = clk_name;
>> +	init.ops = ops;
>> +	init.flags = 0;
>> +	while (i < SOCFGPA_MAX_PARENTS && (parent_name[i] =
>> +			of_clk_get_parent_name(node, i)) != NULL)
>> +		i++;
> 
> I'd suggest explicit loop, like
> 
> +       while (i < SOCFGPA_MAX_PARENTS) {
> +               char *name = of_clk_get_parent_name(node, i);
> +               parent_name[i] = name;
> +               if (name == NULL)
> +                       break;
>                 i++;
> +       }
> 
> This is a bit too subtle.

I don't know, it seems like the previous simple loop is pretty explicit.

> 
>> +	init.parent_names = parent_name;
>> +	init.num_parents = i;
>> +	socfpga_clk->hw.hw.init = &init;
>> +
>> +	clk = clk_register(NULL, &socfpga_clk->hw.hw);
> 
> init is local variable, yet pointer passed to it is in globally
> alocated structure. What is going on there? Are there some comments
> needed?

Same code is used everywhere in drivers/clk.

Dinh
> 
> Thanks,
> 									Pavel
>