On Thu, Jan 30, 2020 at 12:21 PM Viresh Kumar [off-list ref] wrote:

On 29-01-20, 19:04, Sibi Sankar wrote:

quoted

I don't have a gen-pd use case to test against but with the is_genpd
check removed it works as expected when I used it against this
series: https://patchwork.kernel.org/cover/11353185/

In the lazy_link_required_opps fn shouldn't we skip the dynamic
opps in the the opp list?

Tables with dynamic OPPs should not be there in pending_opp_tables
list and so that function shall never get called for them.

quoted

With ^^ addressed:
Reviewed-by: Sibi Sankar <redacted>
Tested-by: Sibi Sankar <redacted>

Thanks Sibi.

@Saravana: Can you please give your feedback as well? I don't want to
push something that may end up breaking something else :)

Hi Viresh and Saravana,

Do you still have plans to push this? I've tested on mt8183 cci with:

1. [v4,0/5] Add required-opps support to devfreq passive gov
(https://patchwork.kernel.org/project/linux-pm/cover/20190724014222.110767-1-saravanak@google.com/):
patch 2, 4, 5

2. opp: Allow lazy-linking of required-opps
(https://patchwork.kernel.org/project/linux-pm/patch/20190717222340.137578-4-saravanak@google.com/#23020727),
with minor diff to let non genpd use required-opp as well:

@@ -474,13 +474,6 @@ static void lazy_link_required_opp_table(struct

opp_table *required_opp_table)
        struct device_node *required_np, *opp_np, *required_table_np;
        int i, ret;

-       /*
-        * We only support genpd's OPPs in the "required-opps" for now,
-        * as we don't know much about other cases.
-        */
-       if (!required_opp_table->is_genpd)
-               return;
-
        mutex_lock(&opp_table_lock);

        list_for_each_entry_safe(opp_table, temp, &pending_opp_tables,
pending) {

3. PM / devfreq: Add cpu based scaling support to passive_governor and
mt8183 cci, cpufreq series
(https://patchwork.kernel.org/project/linux-mediatek/cover/1594348284-14199-1-git-send-email-andrew-sh.cheng@mediatek.com/)

Thanks

--
viresh