On 17:47-20121106, Joshua Emele wrote:
> +static int __cpuinit omap_iva_init(struct cpufreq_policy *policy)
> +{
> + int result;
> + if (!iva_clk_name) {
> + pr_info("%s: iva unavailable\n", __func__);
> + return 0;
> + }
> + iva_dev = omap_device_get_by_hwmod_name("iva");
NAK. Reasons as follows:
a) cpufreq is purely meant for cpu, not IVA. we should instead be using
devfreq which is designed around the usage for co-processor
b) clubbing ARM's frequency decision is definitely NOT equal to IVA
frequency decision, not only is it wrong in terms of modularization, it
is wrong in terms of power benefits as well (not to mention weirdness
needed when you look at all OMAP SoC variants)
c) DVFS is not trivial around multiple co-processor transitions -> core
OPPs (as dependent OPP) needs to be addressed as well. ideally common
clock framework could take care of clock dependencies.
--
Regards,
Nishanth Menon
--
To unsubscribe from this list: send the line "unsubscribe cpufreq" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
