Re: [PATCH v2 1/5] clk: qcom: regmap-mux: add pipe clk implementation

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 13/04/2022 12:15, Johan Hovold wrote:
On Tue, Apr 12, 2022 at 10:38:35PM +0300, Dmitry Baryshkov wrote:
On recent Qualcomm platforms the QMP PIPE clocks feed into a set of
muxes which must be parked to the "safe" source (bi_tcxo) when
corresponding GDSC is turned off and on again. Currently this is
handcoded in the PCIe driver by reparenting the gcc_pipe_N_clk_src
clock. However the same code sequence should be applied in the
pcie-qcom endpoint, USB3 and UFS drivers.

I'm starting to think this really belongs in the PHY driver which is the
provider of the pipe clock. Moving it there would also allow the code to
be shared between PCIe, USB, and UFS.

The PHY driver enables the pipe clock by starting the PHY and before
doing so there's no point in updating the mux. Similarly, the PHY driver
can restore the "safe" source after disabling the pipe clock.


I thought about this at some point. However it would still mean that the driver does the dance manually: disable pipe_clock, switch parent, sleep, switch the parent back, enable pipe clock. Switching parents is tied to disabling pipe_clock, so enforcing this link seems like a better option to me.

No to mention that it would complicate already overcomplicated QMP driver.

That way there's no magic happening behind scenes, the clock framework
always reports the actual state of the tree, and the reason for all of
this can be documented in the QMP PHY driver once and for all.

We already have such 'magic' for the RCG2 (clk_rcg2_shared_ops), with the very practical reason. If the clock is running from the tcxo, it is as good as disabled from the practical purpose.

The only change to the bindings compared to what this series proposes is
that the PHY driver also needs a reference to bi_tcxo.

And this looks as bad, as providing bi_tcxo to the PCI device. From the schematics/silicon point of view neither of them actually uses these parents. Neither of them uses the pipe_clock_src. What do they need is just the pipe_clock. The rest should be in the programming API.


Also note that updating the mux separately from starting the PHY as this
series allows for, doesn't really make the pipe clock any safer to use.

Either way, there are also some problems with this safe-mux
implementation that I point out below.

Rather than copying this sequence over and over again, follow the
example of clk_rcg2_shared_ops and implement this parking in the
enable() and disable() clock operations. As we are changing the parent
behind the back of the clock framework, also implement custom
set_parent() and get_parent() operations behaving accroding to the clock
framework expectations (cache the new parent if the clock is in disabled
state, return cached parent).

Reviewed-by: Bjorn Andersson <bjorn.andersson@xxxxxxxxxx>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@xxxxxxxxxx>
---
  drivers/clk/qcom/clk-regmap-mux.c | 78 +++++++++++++++++++++++++++++++
  drivers/clk/qcom/clk-regmap-mux.h |  3 ++
  2 files changed, 81 insertions(+)

diff --git a/drivers/clk/qcom/clk-regmap-mux.c b/drivers/clk/qcom/clk-regmap-mux.c
index 45d9cca28064..c39ee783ee83 100644
--- a/drivers/clk/qcom/clk-regmap-mux.c
+++ b/drivers/clk/qcom/clk-regmap-mux.c
@@ -49,9 +49,87 @@ static int mux_set_parent(struct clk_hw *hw, u8 index)
  	return regmap_update_bits(clkr->regmap, mux->reg, mask, val);
  }
+static u8 mux_safe_get_parent(struct clk_hw *hw)
+{
+	struct clk_regmap_mux *mux = to_clk_regmap_mux(hw);
+	unsigned int val;
+
+	if (clk_hw_is_enabled(hw))
+		return mux_get_parent(hw);
+
+	val = mux->stored_parent_cfg;
+
+	if (mux->parent_map)
+		return qcom_find_cfg_index(hw, mux->parent_map, val);
+
+	return val;
+}
+
+static int mux_safe_set_parent(struct clk_hw *hw, u8 index)
+{
+	struct clk_regmap_mux *mux = to_clk_regmap_mux(hw);
+
+	if (clk_hw_is_enabled(hw))
+		return mux_set_parent(hw, index);
+
+	if (mux->parent_map)
+		index = mux->parent_map[index].cfg;
+
+	mux->stored_parent_cfg = index;
+
+	return 0;
+}
+
+static void mux_safe_disable(struct clk_hw *hw)
+{
+	struct clk_regmap_mux *mux = to_clk_regmap_mux(hw);
+	struct clk_regmap *clkr = to_clk_regmap(hw);
+	unsigned int mask = GENMASK(mux->width + mux->shift - 1, mux->shift);
+	unsigned int val;
+
+	regmap_read(clkr->regmap, mux->reg, &val);
+
+	mux->stored_parent_cfg = (val & mask) >> mux->shift;
+
+	val = mux->safe_src_parent;
+	if (mux->parent_map) {
+		int index = qcom_find_src_index(hw, mux->parent_map, val);
+
+		if (WARN_ON(index < 0))
+			return;
+
+		val = mux->parent_map[index].cfg;
+	}
+	val <<= mux->shift;
+
+	regmap_update_bits(clkr->regmap, mux->reg, mask, val);
+}
+
+static int mux_safe_enable(struct clk_hw *hw)
+{
+	struct clk_regmap_mux *mux = to_clk_regmap_mux(hw);
+	struct clk_regmap *clkr = to_clk_regmap(hw);
+	unsigned int mask = GENMASK(mux->width + mux->shift - 1, mux->shift);
+	unsigned int val;
+
+	val = mux->stored_parent_cfg;
+	val <<= mux->shift;
+
+	return regmap_update_bits(clkr->regmap, mux->reg, mask, val);
+}

The caching of the parent is broken since set_parent() is typically not
called before enabling the clock.

This means that the above code will set the mux to its zero-initialised
value, which currently only works by chance as the pipe clock config
value happens to be zero.

For this to work generally, you'd also need to define also the
(default/initial) non-safe parent for each mux. Handling handover from
the bootloader might also be tricky.

Furthermore, the current implementation appears to ignore locking and
doesn't handle the case where set_parent() races with enable(). The
former is protected by the prepare mutex and the latter by the enable
spinlock and a driver that needs to serialise the two needs to handle
that itself.

+
  const struct clk_ops clk_regmap_mux_closest_ops = {
  	.get_parent = mux_get_parent,
  	.set_parent = mux_set_parent,
  	.determine_rate = __clk_mux_determine_rate_closest,
  };
  EXPORT_SYMBOL_GPL(clk_regmap_mux_closest_ops);
+
+const struct clk_ops clk_regmap_mux_safe_ops = {
+	.enable = mux_safe_enable,
+	.disable = mux_safe_disable,
+	.get_parent = mux_safe_get_parent,
+	.set_parent = mux_safe_set_parent,
+	.determine_rate = __clk_mux_determine_rate_closest,
+};
+EXPORT_SYMBOL_GPL(clk_regmap_mux_safe_ops);
diff --git a/drivers/clk/qcom/clk-regmap-mux.h b/drivers/clk/qcom/clk-regmap-mux.h
index db6f4cdd9586..f86c674ce139 100644
--- a/drivers/clk/qcom/clk-regmap-mux.h
+++ b/drivers/clk/qcom/clk-regmap-mux.h
@@ -14,10 +14,13 @@ struct clk_regmap_mux {
  	u32			reg;
  	u32			shift;
  	u32			width;
+	u8			safe_src_parent;
+	u8			stored_parent_cfg;
  	const struct parent_map	*parent_map;
  	struct clk_regmap	clkr;
  };
extern const struct clk_ops clk_regmap_mux_closest_ops;
+extern const struct clk_ops clk_regmap_mux_safe_ops;
#endif

Johan


--
With best wishes
Dmitry



[Index of Archives]     [Linux ARM Kernel]     [Linux ARM]     [Linux Omap]     [Fedora ARM]     [Linux for Sparc]     [IETF Annouce]     [Security]     [Bugtraq]     [Linux MIPS]     [ECOS]     [Asterisk Internet PBX]     [Linux API]

  Powered by Linux