Re: [PATCH v3 1/2] clk: Add generic sync_state callback for disabling unused clocks

[Stephen Boyd <sboyd@kernel.org>]

Quoting Abel Vesa (2023-02-20 07:46:36)
> On 23-02-17 21:38:22, Stephen Boyd wrote:
> > Quoting Abel Vesa (2022-12-27 12:45:27)
> > > There are unused clocks that need to remain untouched by clk_disable_unused,
> > > and most likely could be disabled later on sync_state. So provide a generic
> > > sync_state callback for the clock providers that register such clocks.
> > > Then, use the same mechanism as clk_disable_unused from that generic
> > > callback, but pass the device to make sure only the clocks belonging to
> > > the current clock provider get disabled, if unused. Also, during the
> > > default clk_disable_unused, if the driver that registered the clock has
> > > the generic clk_sync_state_disable_unused callback set for sync_state,
> > > skip disabling its clocks.
> > 
> > How does that avoid disabling clks randomly in the clk tree? I'm
> > concerned about disabling an unused clk in the middle of the tree
> > because it doesn't have a driver using sync state, while the clk is the
> > parent of an unused clk that is backed by sync state.
> > 
> >    clk A -->  clk B 
> > 
> > clk A: No sync state
> > clk B: sync state
> > 
> > clk B is left on by the bootloader. __clk_disable_unused(NULL) is called
> > from late init. Imagine clk A is the root of the tree.
> > 
> >       clk_disable_unused_subtree(clk_core A)
> >         clk_disable_unused_subtree(clk_core B)
> >           if (from_sync_state && core->dev != dev)
> >             return;
> >         ...
> >         clk core A->ops->disable()
> > 
> > clk core B is off now?
> 
> Yes, that is correct. But the same thing is happening currently if the
> clk_ignore_unused in not specified.

The existing code traverses the clk tree in depth-first order, disabling
clks from the leaves up to the root. This breaks that tree walk. It is
not the same thing.

> At least with this new approach, we
> get to leave unused clocks enabled either until sync_state is called or forever.
> All the provider has to do is to implement a sync_state callback (or use
> the generic one provided). So the provider of clk A would obviously need
> a sync state callback registered.

Sure.

> 
> > 
> > Also sync_state seems broken right now. I saw mka mentioned that if you
> > have a device node enabled in your DT but never enable a driver for it
> > in the kernel we'll never get sync_state called. This is another
> > problem, but it concerns me that sync_state would make the unused clk
> > disabling happen at some random time or not at all.
> 
> Well, the fact that the sync state not being called because a driver for
> a consumer device doesn't probe does not really mean it is broken. Just
> because the consumer driver hasn't probed yet, doesn't mean it will
> not probe later on.
> 
> That aside, rather than going with clk_ignore_unused all the time on
> qcom platforms, at least in a perfect scenario (where sync state is
> reached for all providers) the clocks get disabled.

The clks will get disabled in some random order though even if every clk
provider has sync_state.

> 
> > 
> > Can the problem be approached more directly? If this is about fixing
> > continuous splash screen, then I wonder why we can't list out the clks
> > that we know are enabled by the bootloader in some new DT binding, e.g.:
> > 
> >       clock-controller {
> >               #clock-cells = <1>;
> >               boot-handoff-clocks = <&consumer_device "clock cells for this clk provider">;
> >       };
> > 
> > Then mark those as "critical/don't turn off" all the way up the clk tree
> > when the clk driver probes by essentially incrementing the
> > prepare/enable count but not actually touching the hardware, and when
> > the clks are acquired by clk_get() for that device that's using them
> > from boot we make the first clk_prepare_enable() do nothing and not
> > increment the count at all. We can probably stick some flag into the
> > 'struct clk' for this when we create the handle in clk_get() so that the
> > prepare and enable functions can special case and skip over.
> 
> Well, that means we need to play whack-a-mole by alsways adding such clocks to
> devicetree.

I don't think the bootloader is constantly changing. Either we want to
hand off the enable state to devices that are using them from boot, or
we don't. I doubt that is changing outside of bootloader development
time.

> 
> > 
> > The sync_state hook operates on a driver level, which is too large when
> > you consider that a single clk driver may register hundreds of clks that
> > are not related. We want to target a solution at the clk level so that
> > any damage from keeping on all the clks provided by the controller is
> > limited to just the drivers that aren't probed and ready to handle their
> > clks. If sync_state could be called whenever a clk consumer consumes a
> > clk it may work? Technically we already have that by the clk_hw_provider
> > function but there isn't enough information being passed there, like the
> > getting device.
> 
> Actually, from the multitude of clocks registered by one provider, the
> ones already explicitely enabled (and obvisously their parents) by thier
> consumer are safe. The only ones we need to worry about are the ones that
> might be enabled by bootloader and need to remain on. With the sync state
> approach, the latter mentioned clocks will either remain on indefinitely
> or will be disabled on sync state. The provider driver is the only level
> that has a registered sync state callback.
> 

The driver has sync_state callback, yes. I'm saying that it is too wide
of a scope to implement disabling unused clks via the sync_state
callback.