| Message ID | 1415408818-28028-1-git-send-email-dianders@chromium.org (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
Quoting Doug Anderson (2014-11-07 17:06:58) > With the existing code, if you find a parent for an orhpan that has > already been prepared / enabled, you won't enable the parent. That > can cause later problems since the clock tree isn't in a consistent > state. Fix by propagating the prepare and enable. > > NOTE: this does bring up the question about whether the enable of the > orphan actually made sense. If the orphan's parent wasn't enabled by > default (by the bootloader or the default state of the hardware) then > the original enable of the orphan probably didn't do what the caller > though it would. Some users of the orphan might have preferred an > EPROBE_DEFER be returned until we had a full path to a root clock. > This patch doesn't address those concerns and really just syncs up the > state. -ECANOFWORMS I'm thinking about this patch but I haven't quite made up my mind. It is reasonable, but also some nice kind of error might be preferable when preparing/enabling an orphaned clock. Under what conditions might a clock be orphaned? An obvious example is just bad luck during the thundering herd of clock registrations from a driver. In this case deferring the probe might be a good idea. However what about the case where a loadable module provides the parent clock? It might be a long time before the orphan clocks gets picked up from the orphanage. Is deferring probe the right thing here as well? Regards, Mike > > Tested on rk3288-evb-rk808 by temporarily considering "sclk_tsadc" as > a critical clock (to simulate a driver enabling it at bootup). > > Before: > > clock enable_cnt prepare_cnt rate accuracy phase > ---------------------------------------------------------------------------------------- > xin32k 0 0 32768 0 0 > sclk_hdmi_cec 0 0 32768 0 0 > sclk_otg_adp 0 0 32768 0 0 > sclk_tsadc 1 1 993 0 0 > > After: > > clock enable_cnt prepare_cnt rate accuracy phase > ---------------------------------------------------------------------------------------- > xin32k 1 1 32768 0 0 > sclk_hdmi_cec 0 0 32768 0 0 > sclk_otg_adp 0 0 32768 0 0 > sclk_tsadc 1 1 993 0 0 > > Note that xin32k on rk808 is a clock that cannot be disabled in > hardware (it's an always on clock), so really all we needed to do was > to sync up the state. > > Signed-off-by: Doug Anderson <dianders@chromium.org> > --- > Changes in v2: > - Only do the work for orphans, not for other users of __clk_reparent(). > > drivers/clk/clk.c | 20 ++++++++++++++++++-- > 1 file changed, 18 insertions(+), 2 deletions(-) > > diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c > index 4896ae9..0f04b7c 100644 > --- a/drivers/clk/clk.c > +++ b/drivers/clk/clk.c > @@ -1652,6 +1652,22 @@ void __clk_reparent(struct clk *clk, struct clk *new_parent) > __clk_recalc_rates(clk, POST_RATE_CHANGE); > } > > +static void __clk_reparent_orphan(struct clk *clk, struct clk *new_parent) > +{ > + __clk_reparent(clk, new_parent); > + > + if (clk->prepare_count) { > + unsigned long flags; > + > + __clk_prepare(new_parent); > + > + flags = clk_enable_lock(); > + if (clk->enable_count) > + __clk_enable(new_parent); > + clk_enable_unlock(flags); > + } > +} > + > /** > * clk_set_parent - switch the parent of a mux clk > * @clk: the mux clk whose input we are switching > @@ -1953,13 +1969,13 @@ int __clk_init(struct device *dev, struct clk *clk) > if (orphan->num_parents && orphan->ops->get_parent) { > i = orphan->ops->get_parent(orphan->hw); > if (!strcmp(clk->name, orphan->parent_names[i])) > - __clk_reparent(orphan, clk); > + __clk_reparent_orphan(orphan, clk); > continue; > } > > for (i = 0; i < orphan->num_parents; i++) > if (!strcmp(clk->name, orphan->parent_names[i])) { > - __clk_reparent(orphan, clk); > + __clk_reparent_orphan(orphan, clk); > break; > } > } > -- > 2.1.0.rc2.206.gedb03e5 >
Mike, On Wed, Nov 19, 2014 at 6:30 PM, Mike Turquette <mturquette@linaro.org> wrote: > Quoting Doug Anderson (2014-11-07 17:06:58) >> With the existing code, if you find a parent for an orhpan that has >> already been prepared / enabled, you won't enable the parent. That >> can cause later problems since the clock tree isn't in a consistent >> state. Fix by propagating the prepare and enable. >> >> NOTE: this does bring up the question about whether the enable of the >> orphan actually made sense. If the orphan's parent wasn't enabled by >> default (by the bootloader or the default state of the hardware) then >> the original enable of the orphan probably didn't do what the caller >> though it would. Some users of the orphan might have preferred an >> EPROBE_DEFER be returned until we had a full path to a root clock. >> This patch doesn't address those concerns and really just syncs up the >> state. > > -ECANOFWORMS > > I'm thinking about this patch but I haven't quite made up my mind. It is > reasonable, but also some nice kind of error might be preferable when > preparing/enabling an orphaned clock. > > Under what conditions might a clock be orphaned? An obvious example is > just bad luck during the thundering herd of clock registrations from a > driver. In this case deferring the probe might be a good idea. > > However what about the case where a loadable module provides the parent > clock? It might be a long time before the orphan clocks gets picked up > from the orphanage. Is deferring probe the right thing here as well? I will defer to your wisdom here. I agree that these are the two primary solutions and I've picked one, but I have no idea which will be more of a PITA in the long run. Note: I'm not sure that anyone expects EPROBE_DEFER to be returned from a clk_enable() (do they?). It almost seems like the right answer is to fail to allow anyone to clk_get() any clock that doesn't have a path to root. I will say that without this patch or the EPROBE_DEFER solution we have a clear bug. You can get into a situation where a clock is enabled/prepared but its parent isn't. -Doug
On Wed, Nov 19, 2014 at 09:15:41PM -0800, Doug Anderson wrote: > Mike, > > On Wed, Nov 19, 2014 at 6:30 PM, Mike Turquette <mturquette@linaro.org> wrote: > > Quoting Doug Anderson (2014-11-07 17:06:58) > >> With the existing code, if you find a parent for an orhpan that has > >> already been prepared / enabled, you won't enable the parent. That > >> can cause later problems since the clock tree isn't in a consistent > >> state. Fix by propagating the prepare and enable. > >> > >> NOTE: this does bring up the question about whether the enable of the > >> orphan actually made sense. If the orphan's parent wasn't enabled by > >> default (by the bootloader or the default state of the hardware) then > >> the original enable of the orphan probably didn't do what the caller > >> though it would. Some users of the orphan might have preferred an > >> EPROBE_DEFER be returned until we had a full path to a root clock. > >> This patch doesn't address those concerns and really just syncs up the > >> state. > > > > -ECANOFWORMS > > > > I'm thinking about this patch but I haven't quite made up my mind. It is > > reasonable, but also some nice kind of error might be preferable when > > preparing/enabling an orphaned clock. > > > > Under what conditions might a clock be orphaned? An obvious example is > > just bad luck during the thundering herd of clock registrations from a > > driver. In this case deferring the probe might be a good idea. > > > > However what about the case where a loadable module provides the parent > > clock? It might be a long time before the orphan clocks gets picked up > > from the orphanage. Is deferring probe the right thing here as well? > > I will defer to your wisdom here. I agree that these are the two > primary solutions and I've picked one, but I have no idea which will > be more of a PITA in the long run. > > Note: I'm not sure that anyone expects EPROBE_DEFER to be returned > from a clk_enable() (do they?). It almost seems like the right answer > is to fail to allow anyone to clk_get() any clock that doesn't have a > path to root. EPROBE_DEFER only makes sense in driver's probe paths and so I would be very against adding it to clk_enable() which is called from many places in the kernel. If we decide to go with EPROBE_DEFER then returning it from clk_get() seems like a much better choice since it is normally called during probing. Thanks.
On Wed, Nov 19, 2014 at 11:45:14PM -0800, Dmitry Torokhov wrote: > On Wed, Nov 19, 2014 at 09:15:41PM -0800, Doug Anderson wrote: > > I will defer to your wisdom here. I agree that these are the two > > primary solutions and I've picked one, but I have no idea which will > > be more of a PITA in the long run. > > > > Note: I'm not sure that anyone expects EPROBE_DEFER to be returned > > from a clk_enable() (do they?). It almost seems like the right answer > > is to fail to allow anyone to clk_get() any clock that doesn't have a > > path to root. > > EPROBE_DEFER only makes sense in driver's probe paths and so I would be > very against adding it to clk_enable() which is called from many places > in the kernel. If we decide to go with EPROBE_DEFER then returning it > from clk_get() seems like a much better choice since it is normally > called during probing. Absolutely correct. EINVAL would be better for clk_prepare() since it isn't something that can be recovered from by just retrying a bit later. You're absolutely correct that EPROBE_DEFER has no business being returned in any path other than a driver's probe function; it is not a user visible error code, it is a special internal Linux error code which only the driver model understands to mean "add this device to the deferred probe list and try again a while later." Userspace, especially, should never see this error code.
diff --git a/drivers/clk/clk.c b/drivers/clk/clk.c index 4896ae9..0f04b7c 100644 --- a/drivers/clk/clk.c +++ b/drivers/clk/clk.c @@ -1652,6 +1652,22 @@ void __clk_reparent(struct clk *clk, struct clk *new_parent) __clk_recalc_rates(clk, POST_RATE_CHANGE); } +static void __clk_reparent_orphan(struct clk *clk, struct clk *new_parent) +{ + __clk_reparent(clk, new_parent); + + if (clk->prepare_count) { + unsigned long flags; + + __clk_prepare(new_parent); + + flags = clk_enable_lock(); + if (clk->enable_count) + __clk_enable(new_parent); + clk_enable_unlock(flags); + } +} + /** * clk_set_parent - switch the parent of a mux clk * @clk: the mux clk whose input we are switching @@ -1953,13 +1969,13 @@ int __clk_init(struct device *dev, struct clk *clk) if (orphan->num_parents && orphan->ops->get_parent) { i = orphan->ops->get_parent(orphan->hw); if (!strcmp(clk->name, orphan->parent_names[i])) - __clk_reparent(orphan, clk); + __clk_reparent_orphan(orphan, clk); continue; } for (i = 0; i < orphan->num_parents; i++) if (!strcmp(clk->name, orphan->parent_names[i])) { - __clk_reparent(orphan, clk); + __clk_reparent_orphan(orphan, clk); break; } }
With the existing code, if you find a parent for an orhpan that has already been prepared / enabled, you won't enable the parent. That can cause later problems since the clock tree isn't in a consistent state. Fix by propagating the prepare and enable. NOTE: this does bring up the question about whether the enable of the orphan actually made sense. If the orphan's parent wasn't enabled by default (by the bootloader or the default state of the hardware) then the original enable of the orphan probably didn't do what the caller though it would. Some users of the orphan might have preferred an EPROBE_DEFER be returned until we had a full path to a root clock. This patch doesn't address those concerns and really just syncs up the state. Tested on rk3288-evb-rk808 by temporarily considering "sclk_tsadc" as a critical clock (to simulate a driver enabling it at bootup). Before: clock enable_cnt prepare_cnt rate accuracy phase ---------------------------------------------------------------------------------------- xin32k 0 0 32768 0 0 sclk_hdmi_cec 0 0 32768 0 0 sclk_otg_adp 0 0 32768 0 0 sclk_tsadc 1 1 993 0 0 After: clock enable_cnt prepare_cnt rate accuracy phase ---------------------------------------------------------------------------------------- xin32k 1 1 32768 0 0 sclk_hdmi_cec 0 0 32768 0 0 sclk_otg_adp 0 0 32768 0 0 sclk_tsadc 1 1 993 0 0 Note that xin32k on rk808 is a clock that cannot be disabled in hardware (it's an always on clock), so really all we needed to do was to sync up the state. Signed-off-by: Doug Anderson <dianders@chromium.org> --- Changes in v2: - Only do the work for orphans, not for other users of __clk_reparent(). drivers/clk/clk.c | 20 ++++++++++++++++++-- 1 file changed, 18 insertions(+), 2 deletions(-)