On Tue, Jul 22, 2008 at 11:01:35PM -0400, Jon Smirl wrote:
Convert bitfields in ASOC into full int width. This needs testing, I don't have sample controls for all of the different configurations. This is a simple mechanical conversion, behavior should not be changed. I did find two places in the DAPM code where max was being used as a mask. I fixed them to convert max to a mask, was this right?
Might want to look at your line wrapping here :) .
The patch looks like a good approach and should work - the issues below a look fixable.
When I tried testing this patch I found that SOC_ENUM no longer builds - the first error is 'error: extra brace group at end of initializer'. Looks like that needs adjusting as well. There's also rather a lot of checkpatch warnings, mostly lines over 80 columns and whitespace errors.
+/* mixer control */ +struct mixer_control {
- int min, max;
- uint reg, rreg, shift, rshift, invert;
+};
This should be namespaced - soc_mixer_control, snd_soc_mixer_control or simlar.
@@ -104,10 +104,12 @@ static void dapm_set_path_status(struct snd_soc_dapm_widget *w, case snd_soc_dapm_switch: case snd_soc_dapm_mixer: { int val;
int reg = w->kcontrols[i].private_value & 0xff;int shift = (w->kcontrols[i].private_value >> 8) & 0x0f;int mask = (w->kcontrols[i].private_value >> 16) & 0xff;int invert = (w->kcontrols[i].private_value >> 24) & 0x01;
struct mixer_control *mc = (struct mixer_control *)w->kcontrols[i].private_value;uint reg = mc->reg;uint shift = mc->shift;int max = mc->max;uint mask = (1 << fls(max)) - 1;uint invert = mc->invert;
What the original code is doing here is checking to see if the control is on at all.