On Wed, 2010-10-13 at 16:29 +0100, Mark Brown wrote:
On Wed, Oct 13, 2010 at 08:27:47AM -0700, Joe Perches wrote:
On Wed, 2010-10-13 at 16:11 +0100, Mark Brown wrote:
If you check the code again you will notice that these functons are all used in ops structures which presents obvious issues when trying to replace with open coded checks. I hope their usefulness is clear given the above.
There is some value in using consistent function styles for table driven and non-table driven instances.
I'm sorry, I don't understand what you are saying here. What do you believe to be inconsistent?
I don't believe anything to be inconsistent about the volatile uses and was agreeing with you.
Some codecs functions for volatile access use a switch/case,
static int wm8994_volatile(unsigned int reg) { if (reg >= WM8994_REG_CACHE_SIZE) return 1;
switch (reg) { case WM8994_SOFTWARE_RESET: case WM8994_CHIP_REVISION: case WM8994_DC_SERVO_1: case WM8994_DC_SERVO_READBACK: case WM8994_RATE_STATUS: case WM8994_LDO_1: case WM8994_LDO_2: return 1; default: return 0; } }
others use the register variable from the struct
static int wm8962_volatile_register(unsigned int reg) { if (wm8962_reg_access[reg].volatile_register) return 1; else return 0; }
so I'm agreeing that it's useful to keep the same access style in multiple codecs instead of using separate styles in each one.
It'd be even better to use a similarly consistent function naming scheme.
cheers, Joe