On 27-01-21, 22:27, Dmitry Osipenko wrote: > 27.01.2021 13:16, Viresh Kumar пишет: > > There is no point calling dev_pm_opp_set_clkname() with the "name" > > parameter set to NULL, this is already done by the OPP core at setup > > time and should work as it is. > > > > Signed-off-by: Viresh Kumar <viresh.kumar@xxxxxxxxxx> > > > > --- > > V2: Update tegra124 as well. > > > > Krzysztof, please take this through your tree, it doesn't have any > > dependency in the OPP tree. > > --- > > drivers/memory/tegra/tegra124-emc.c | 13 ++----------- > > drivers/memory/tegra/tegra20-emc.c | 13 ++----------- > > drivers/memory/tegra/tegra30-emc.c | 13 ++----------- > > 3 files changed, 6 insertions(+), 33 deletions(-) > > Ideally drivers should be able to ensure that OPP table received the > clk, IMO. I don't really agree with that, this stuff is internal to the OPP core. And it should all work unless there is a bug, whose effects we will see anyway. > But this is also almost fine with me since realistically > clk_get() shouldn't fail if it already succeeded a moment ago. > > Reviewed-by: Dmitry Osipenko <digetx@xxxxxxxxx> > Tested-by: Dmitry Osipenko <digetx@xxxxxxxxx> Thanks. Krzysztof, please apply the patch to your tree. Thanks. > Could we please fix the _allocate_opp_table() to not ignore clk_get() > errors like -ENOMEM and etc? Maybe we should fail if the error is anything other than -ENODEV ? -- viresh
