| Message ID | 1597043179-17903-1-git-send-email-rnayak@codeaurora.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | opp: Fix dev_pm_opp_set_rate() to not return early | expand |
On 2020-08-10 12:36, Rajendra Nayak wrote: > dev_pm_opp_set_rate() can now be called with freq = 0 inorder > to either drop performance or bandwidth votes or to disable > regulators on platforms which support them. > In such cases, a subsequent call to dev_pm_opp_set_rate() with > the same frequency ends up returning early because 'old_freq == freq' > Instead make it fall through and put back the dropped performance > and bandwidth votes and/or enable back the regulators. > > Fixes: cd7ea582 ("opp: Make dev_pm_opp_set_rate() handle freq = 0 to > drop performance votes") > Reported-by: Sajida Bhanu <sbhanu@codeaurora.org> > Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org> Reviewed-by: Sibi Sankar <sibis@codeaurora.org> > --- > drivers/opp/core.c | 8 +++++++- > 1 file changed, 7 insertions(+), 1 deletion(-) > > diff --git a/drivers/opp/core.c b/drivers/opp/core.c > index 0c8c74a..a994f30 100644 > --- a/drivers/opp/core.c > +++ b/drivers/opp/core.c > @@ -901,6 +901,9 @@ int dev_pm_opp_set_rate(struct device *dev, > unsigned long target_freq) > > /* Return early if nothing to do */ > if (old_freq == freq) { > + if (opp_table->required_opp_tables || opp_table->regulators || > + opp_table->paths) > + goto skip_clk_only; > dev_dbg(dev, "%s: old/new frequencies (%lu Hz) are same, nothing to > do\n", > __func__, freq); > ret = 0; > @@ -919,6 +922,7 @@ int dev_pm_opp_set_rate(struct device *dev, > unsigned long target_freq) > goto put_opp_table; > } > > +skip_clk_only: > temp_freq = old_freq; > old_opp = _find_freq_ceil(opp_table, &temp_freq); > if (IS_ERR(old_opp)) { > @@ -954,8 +958,10 @@ int dev_pm_opp_set_rate(struct device *dev, > unsigned long target_freq) > IS_ERR(old_opp) ? NULL : old_opp->supplies, > opp->supplies); > } else { > + ret = 0; > /* Only frequency scaling */ > - ret = _generic_set_opp_clk_only(dev, clk, freq); > + if (freq != old_freq) > + ret = _generic_set_opp_clk_only(dev, clk, freq); > } > > /* Scaling down? Configure required OPPs after frequency */
On Mon, Aug 10, 2020 at 12:36:19PM +0530, Rajendra Nayak wrote: > dev_pm_opp_set_rate() can now be called with freq = 0 inorder > to either drop performance or bandwidth votes or to disable > regulators on platforms which support them. > In such cases, a subsequent call to dev_pm_opp_set_rate() with > the same frequency ends up returning early because 'old_freq == freq' > Instead make it fall through and put back the dropped performance > and bandwidth votes and/or enable back the regulators. > > Fixes: cd7ea582 ("opp: Make dev_pm_opp_set_rate() handle freq = 0 to drop performance votes") > Reported-by: Sajida Bhanu <sbhanu@codeaurora.org> > Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org> Tested-by: Matthias Kaehlcke <mka@chromium.org> Originally-reported-by: Matthias Kaehlcke <mka@chromium.org> https://patchwork.kernel.org/patch/11675369/#23514895 :P
Quoting Rajendra Nayak (2020-08-10 00:06:19) > dev_pm_opp_set_rate() can now be called with freq = 0 inorder > to either drop performance or bandwidth votes or to disable > regulators on platforms which support them. > In such cases, a subsequent call to dev_pm_opp_set_rate() with > the same frequency ends up returning early because 'old_freq == freq' > Instead make it fall through and put back the dropped performance > and bandwidth votes and/or enable back the regulators. > > Fixes: cd7ea582 ("opp: Make dev_pm_opp_set_rate() handle freq = 0 to drop performance votes") > Reported-by: Sajida Bhanu <sbhanu@codeaurora.org> > Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org> > --- > drivers/opp/core.c | 8 +++++++- > 1 file changed, 7 insertions(+), 1 deletion(-) > > diff --git a/drivers/opp/core.c b/drivers/opp/core.c > index 0c8c74a..a994f30 100644 > --- a/drivers/opp/core.c > +++ b/drivers/opp/core.c > @@ -901,6 +901,9 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq) > > /* Return early if nothing to do */ > if (old_freq == freq) { > + if (opp_table->required_opp_tables || opp_table->regulators || > + opp_table->paths) > + goto skip_clk_only; This is a goto maze! Any chance we can clean this up? if (!opp_table->required_opp_tables && !opp_table->regulators && !opp_table->paths) if (old_freq == freq) { ret = 0 dev_dbg(..) } else if (!_get_opp_count(opp_table)) { ret = _generic_set_opp_clk_only(dev, clk, freq); } } else { temp_freq = old_freq; old_opp = _find_freq_ceil(opp_table, &temp_freq); ... dev_pm_opp_put(opp); put_old_opp: if (!IS_ERR(old_opp)) dev_pm_opp_put(old_opp); } put_opp_table: dev_pm_opp_put_opp_table(opp_table); And that stuff in the else should probably go to another function. > dev_dbg(dev, "%s: old/new frequencies (%lu Hz) are same, nothing to do\n", > __func__, freq); > ret = 0; > @@ -919,6 +922,7 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq) > goto put_opp_table; > } > > +skip_clk_only: > temp_freq = old_freq; > old_opp = _find_freq_ceil(opp_table, &temp_freq); > if (IS_ERR(old_opp)) { > @@ -954,8 +958,10 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq) > IS_ERR(old_opp) ? NULL : old_opp->supplies, > opp->supplies); > } else { > + ret = 0; > /* Only frequency scaling */ > - ret = _generic_set_opp_clk_only(dev, clk, freq); > + if (freq != old_freq) > + ret = _generic_set_opp_clk_only(dev, clk, freq); > } And write this as else if (freq != old_freq) { ret = _generic_set_opp_clk_only(..) } else { ret = 0; }
On 8/11/2020 10:42 PM, Matthias Kaehlcke wrote: > On Mon, Aug 10, 2020 at 12:36:19PM +0530, Rajendra Nayak wrote: >> dev_pm_opp_set_rate() can now be called with freq = 0 inorder >> to either drop performance or bandwidth votes or to disable >> regulators on platforms which support them. >> In such cases, a subsequent call to dev_pm_opp_set_rate() with >> the same frequency ends up returning early because 'old_freq == freq' >> Instead make it fall through and put back the dropped performance >> and bandwidth votes and/or enable back the regulators. >> >> Fixes: cd7ea582 ("opp: Make dev_pm_opp_set_rate() handle freq = 0 to drop performance votes") >> Reported-by: Sajida Bhanu <sbhanu@codeaurora.org> >> Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org> > > Tested-by: Matthias Kaehlcke <mka@chromium.org> > > Originally-reported-by: Matthias Kaehlcke <mka@chromium.org> > https://patchwork.kernel.org/patch/11675369/#23514895 :P Sorry to have missed that :/ Thanks for testing.
On 11-08-20, 14:09, Stephen Boyd wrote:
> This is a goto maze! Any chance we can clean this up?
I have sent a short series in reply to this series, please have a
look. It should look better now.
On 8/13/2020 9:59 AM, Viresh Kumar wrote: > On 11-08-20, 14:09, Stephen Boyd wrote: >> This is a goto maze! Any chance we can clean this up? > > I have sent a short series in reply to this series, please have a > look. It should look better now. Thanks, I was out a few days so could not get to the cleanups that Stephen was suggesting. I will give your series a try.
diff --git a/drivers/opp/core.c b/drivers/opp/core.c index 0c8c74a..a994f30 100644 --- a/drivers/opp/core.c +++ b/drivers/opp/core.c @@ -901,6 +901,9 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq) /* Return early if nothing to do */ if (old_freq == freq) { + if (opp_table->required_opp_tables || opp_table->regulators || + opp_table->paths) + goto skip_clk_only; dev_dbg(dev, "%s: old/new frequencies (%lu Hz) are same, nothing to do\n", __func__, freq); ret = 0; @@ -919,6 +922,7 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq) goto put_opp_table; } +skip_clk_only: temp_freq = old_freq; old_opp = _find_freq_ceil(opp_table, &temp_freq); if (IS_ERR(old_opp)) { @@ -954,8 +958,10 @@ int dev_pm_opp_set_rate(struct device *dev, unsigned long target_freq) IS_ERR(old_opp) ? NULL : old_opp->supplies, opp->supplies); } else { + ret = 0; /* Only frequency scaling */ - ret = _generic_set_opp_clk_only(dev, clk, freq); + if (freq != old_freq) + ret = _generic_set_opp_clk_only(dev, clk, freq); } /* Scaling down? Configure required OPPs after frequency */
dev_pm_opp_set_rate() can now be called with freq = 0 inorder to either drop performance or bandwidth votes or to disable regulators on platforms which support them. In such cases, a subsequent call to dev_pm_opp_set_rate() with the same frequency ends up returning early because 'old_freq == freq' Instead make it fall through and put back the dropped performance and bandwidth votes and/or enable back the regulators. Fixes: cd7ea582 ("opp: Make dev_pm_opp_set_rate() handle freq = 0 to drop performance votes") Reported-by: Sajida Bhanu <sbhanu@codeaurora.org> Signed-off-by: Rajendra Nayak <rnayak@codeaurora.org> --- drivers/opp/core.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-)