[PATCH] clk: Add axi-clkgen driver

[lars@metafoo.de (Lars-Peter Clausen)]

On 01/22/2013 06:55 PM, Mike Turquette wrote:
> Quoting Lars-Peter Clausen (2013-01-09 10:12:00)
> <snip>
>> +static void axi_clkgen_write(struct axi_clkgen *axi_clkgen,
>> +       unsigned int reg, unsigned int val)
>> +{
>> +       iowrite32(val, axi_clkgen->base + reg);
> 
> Silly question: any reason to use this over readl()?  This is more for
> my understanding than a real criticism.

I think I read somewhere at some point that ioread{8,16,32} is preferred
over write{b,h,l} in new code.

> 
>> +}
>> +
>> +static void axi_clkgen_read(struct axi_clkgen *axi_clkgen,
>> +       unsigned int reg, unsigned int *val)
>> +{
>> +       *val = ioread32(axi_clkgen->base + reg);
> 
> Same as above, any reason to use this over writel?

Same answer.

> 
> <snip>
>> +static unsigned long axi_clkgen_recalc_rate(struct clk_hw *clk_hw,
>> +       unsigned long parent_rate)
>> +{
>> +       struct axi_clkgen *axi_clkgen = clk_hw_to_axi_clkgen(clk_hw);
>> +       unsigned int d, m, dout;
>> +       unsigned int reg;
>> +
>> +       axi_clkgen_read(axi_clkgen, AXI_CLKGEN_REG_CLK_OUT1, &reg);
>> +       dout = (reg & 0x3f) + ((reg >> 6) & 0x3f);
>> +       axi_clkgen_read(axi_clkgen, AXI_CLKGEN_REG_CLK_DIV, &reg);
>> +       d = (reg & 0x3f) + ((reg >> 6) & 0x3f);
>> +       axi_clkgen_read(axi_clkgen, AXI_CLKGEN_REG_CLK_FB1, &reg);
>> +       m = (reg & 0x3f) + ((reg >> 6) & 0x3f);
>> +
>> +       if (d == 0 || dout == 0)
>> +               return 0;
>> +
>> +       return parent_rate / d * m / dout;
> 
> Any chance of overflow here?  Maybe do_div should be used?

Not if all the parameters are within spec. But since this is on a slowpath I
guess it does not hurt to use do_div.

Will send a v2.

Thanks for the review,
- Lars