All of lore.kernel.org
 help / color / mirror / Atom feed
From: Stephen Boyd <sboyd@codeaurora.org>
To: Mike Turquette <mturquette@linaro.org>,
	Stephen Boyd <sboyd@codeaurora.org>
Cc: linux-kernel@vger.kernel.org, linux-arm-msm@vger.kernel.org,
	linux-arm-kernel@lists.infradead.org,
	Kenneth Westfield <kwestfie@codeaurora.org>
Subject: [PATCH v2] clk: qcom: Properly change rates for ahbix clock
Date: Fri,  6 Mar 2015 15:41:53 -0800	[thread overview]
Message-ID: <1425685313-6005-1-git-send-email-sboyd@codeaurora.org> (raw)

The ahbix clock can never be turned off in practice. To change the
rates we need to switch the mux off the M/N counter to an always on
source (XO), reprogram the M/N counter to get the rate we want and
finally switch back to the M/N counter. Add a new ops structure
for this type of clock so that we can set the rate properly.

Fixes: c99e515a92e9 "clk: qcom: Add IPQ806X LPASS clock controller (LCC) driver"
Cc: Kenneth Westfield <kwestfie@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
---

Changes from v1:
 * Handle the case where clock feeding M/N counter is off

 drivers/clk/qcom/clk-rcg.c     | 62 ++++++++++++++++++++++++++++++++++++++++++
 drivers/clk/qcom/clk-rcg.h     |  1 +
 drivers/clk/qcom/lcc-ipq806x.c |  5 ++--
 3 files changed, 65 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/qcom/clk-rcg.c b/drivers/clk/qcom/clk-rcg.c
index 0039bd7d3965..466f30ca65c2 100644
--- a/drivers/clk/qcom/clk-rcg.c
+++ b/drivers/clk/qcom/clk-rcg.c
@@ -495,6 +495,57 @@ static int clk_rcg_bypass_set_rate(struct clk_hw *hw, unsigned long rate,
 	return __clk_rcg_set_rate(rcg, rcg->freq_tbl);
 }
 
+/*
+ * This type of clock has a glitch-free mux that switches between the output of
+ * the M/N counter and an always on clock source (XO). When clk_set_rate() is
+ * called we need to make sure that we don't switch to the M/N counter if it
+ * isn't clocking because the mux will get stuck and the clock will stop
+ * outputting a clock. This can happen if the framework isn't aware that this
+ * clock is on and so clk_set_rate() doesn't turn on the new parent. To fix
+ * this we switch the mux in the enable/disable ops and reprogram the M/N
+ * counter in the set_rate op. We also make sure to switch away from the M/N
+ * counter in set_rate if software thinks the clock is off.
+ */
+static int clk_rcg_lcc_set_rate(struct clk_hw *hw, unsigned long rate,
+				unsigned long parent_rate)
+{
+	struct clk_rcg *rcg = to_clk_rcg(hw);
+	const struct freq_tbl *f;
+	int ret;
+	u32 gfm = BIT(10);
+
+	f = qcom_find_freq(rcg->freq_tbl, rate);
+	if (!f)
+		return -EINVAL;
+
+	/* Switch to XO to avoid glitches */
+	regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, 0);
+	ret = __clk_rcg_set_rate(rcg, f);
+	/* Switch back to M/N if it's clocking */
+	if (__clk_is_enabled(hw->clk))
+		regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, gfm);
+
+	return ret;
+}
+
+static int clk_rcg_lcc_enable(struct clk_hw *hw)
+{
+	struct clk_rcg *rcg = to_clk_rcg(hw);
+	u32 gfm = BIT(10);
+
+	/* Use M/N */
+	return regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, gfm);
+}
+
+static void clk_rcg_lcc_disable(struct clk_hw *hw)
+{
+	struct clk_rcg *rcg = to_clk_rcg(hw);
+	u32 gfm = BIT(10);
+
+	/* Use XO */
+	regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, 0);
+}
+
 static int __clk_dyn_rcg_set_rate(struct clk_hw *hw, unsigned long rate)
 {
 	struct clk_dyn_rcg *rcg = to_clk_dyn_rcg(hw);
@@ -543,6 +594,17 @@ const struct clk_ops clk_rcg_bypass_ops = {
 };
 EXPORT_SYMBOL_GPL(clk_rcg_bypass_ops);
 
+const struct clk_ops clk_rcg_lcc_ops = {
+	.enable = clk_rcg_lcc_enable,
+	.disable = clk_rcg_lcc_disable,
+	.get_parent = clk_rcg_get_parent,
+	.set_parent = clk_rcg_set_parent,
+	.recalc_rate = clk_rcg_recalc_rate,
+	.determine_rate = clk_rcg_determine_rate,
+	.set_rate = clk_rcg_lcc_set_rate,
+};
+EXPORT_SYMBOL_GPL(clk_rcg_lcc_ops);
+
 const struct clk_ops clk_dyn_rcg_ops = {
 	.enable = clk_enable_regmap,
 	.is_enabled = clk_is_enabled_regmap,
diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
index 687e41f91d7c..d09d06ba278e 100644
--- a/drivers/clk/qcom/clk-rcg.h
+++ b/drivers/clk/qcom/clk-rcg.h
@@ -96,6 +96,7 @@ struct clk_rcg {
 
 extern const struct clk_ops clk_rcg_ops;
 extern const struct clk_ops clk_rcg_bypass_ops;
+extern const struct clk_ops clk_rcg_lcc_ops;
 
 #define to_clk_rcg(_hw) container_of(to_clk_regmap(_hw), struct clk_rcg, clkr)
 
diff --git a/drivers/clk/qcom/lcc-ipq806x.c b/drivers/clk/qcom/lcc-ipq806x.c
index 9e27d82c0201..a6d3a6745c4d 100644
--- a/drivers/clk/qcom/lcc-ipq806x.c
+++ b/drivers/clk/qcom/lcc-ipq806x.c
@@ -386,13 +386,12 @@ static struct clk_rcg ahbix_clk = {
 	.freq_tbl = clk_tbl_ahbix,
 	.clkr = {
 		.enable_reg = 0x38,
-		.enable_mask = BIT(10), /* toggle the gfmux to select mn/pxo */
+		.enable_mask = BIT(11),
 		.hw.init = &(struct clk_init_data){
 			.name = "ahbix",
 			.parent_names = lcc_pxo_pll4,
 			.num_parents = 2,
-			.ops = &clk_rcg_ops,
-			.flags = CLK_SET_RATE_GATE,
+			.ops = &clk_rcg_lcc_ops,
 		},
 	},
 };
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

WARNING: multiple messages have this Message-ID (diff)
From: sboyd@codeaurora.org (Stephen Boyd)
To: linux-arm-kernel@lists.infradead.org
Subject: [PATCH v2] clk: qcom: Properly change rates for ahbix clock
Date: Fri,  6 Mar 2015 15:41:53 -0800	[thread overview]
Message-ID: <1425685313-6005-1-git-send-email-sboyd@codeaurora.org> (raw)

The ahbix clock can never be turned off in practice. To change the
rates we need to switch the mux off the M/N counter to an always on
source (XO), reprogram the M/N counter to get the rate we want and
finally switch back to the M/N counter. Add a new ops structure
for this type of clock so that we can set the rate properly.

Fixes: c99e515a92e9 "clk: qcom: Add IPQ806X LPASS clock controller (LCC) driver"
Cc: Kenneth Westfield <kwestfie@codeaurora.org>
Signed-off-by: Stephen Boyd <sboyd@codeaurora.org>
---

Changes from v1:
 * Handle the case where clock feeding M/N counter is off

 drivers/clk/qcom/clk-rcg.c     | 62 ++++++++++++++++++++++++++++++++++++++++++
 drivers/clk/qcom/clk-rcg.h     |  1 +
 drivers/clk/qcom/lcc-ipq806x.c |  5 ++--
 3 files changed, 65 insertions(+), 3 deletions(-)

diff --git a/drivers/clk/qcom/clk-rcg.c b/drivers/clk/qcom/clk-rcg.c
index 0039bd7d3965..466f30ca65c2 100644
--- a/drivers/clk/qcom/clk-rcg.c
+++ b/drivers/clk/qcom/clk-rcg.c
@@ -495,6 +495,57 @@ static int clk_rcg_bypass_set_rate(struct clk_hw *hw, unsigned long rate,
 	return __clk_rcg_set_rate(rcg, rcg->freq_tbl);
 }
 
+/*
+ * This type of clock has a glitch-free mux that switches between the output of
+ * the M/N counter and an always on clock source (XO). When clk_set_rate() is
+ * called we need to make sure that we don't switch to the M/N counter if it
+ * isn't clocking because the mux will get stuck and the clock will stop
+ * outputting a clock. This can happen if the framework isn't aware that this
+ * clock is on and so clk_set_rate() doesn't turn on the new parent. To fix
+ * this we switch the mux in the enable/disable ops and reprogram the M/N
+ * counter in the set_rate op. We also make sure to switch away from the M/N
+ * counter in set_rate if software thinks the clock is off.
+ */
+static int clk_rcg_lcc_set_rate(struct clk_hw *hw, unsigned long rate,
+				unsigned long parent_rate)
+{
+	struct clk_rcg *rcg = to_clk_rcg(hw);
+	const struct freq_tbl *f;
+	int ret;
+	u32 gfm = BIT(10);
+
+	f = qcom_find_freq(rcg->freq_tbl, rate);
+	if (!f)
+		return -EINVAL;
+
+	/* Switch to XO to avoid glitches */
+	regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, 0);
+	ret = __clk_rcg_set_rate(rcg, f);
+	/* Switch back to M/N if it's clocking */
+	if (__clk_is_enabled(hw->clk))
+		regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, gfm);
+
+	return ret;
+}
+
+static int clk_rcg_lcc_enable(struct clk_hw *hw)
+{
+	struct clk_rcg *rcg = to_clk_rcg(hw);
+	u32 gfm = BIT(10);
+
+	/* Use M/N */
+	return regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, gfm);
+}
+
+static void clk_rcg_lcc_disable(struct clk_hw *hw)
+{
+	struct clk_rcg *rcg = to_clk_rcg(hw);
+	u32 gfm = BIT(10);
+
+	/* Use XO */
+	regmap_update_bits(rcg->clkr.regmap, rcg->ns_reg, gfm, 0);
+}
+
 static int __clk_dyn_rcg_set_rate(struct clk_hw *hw, unsigned long rate)
 {
 	struct clk_dyn_rcg *rcg = to_clk_dyn_rcg(hw);
@@ -543,6 +594,17 @@ const struct clk_ops clk_rcg_bypass_ops = {
 };
 EXPORT_SYMBOL_GPL(clk_rcg_bypass_ops);
 
+const struct clk_ops clk_rcg_lcc_ops = {
+	.enable = clk_rcg_lcc_enable,
+	.disable = clk_rcg_lcc_disable,
+	.get_parent = clk_rcg_get_parent,
+	.set_parent = clk_rcg_set_parent,
+	.recalc_rate = clk_rcg_recalc_rate,
+	.determine_rate = clk_rcg_determine_rate,
+	.set_rate = clk_rcg_lcc_set_rate,
+};
+EXPORT_SYMBOL_GPL(clk_rcg_lcc_ops);
+
 const struct clk_ops clk_dyn_rcg_ops = {
 	.enable = clk_enable_regmap,
 	.is_enabled = clk_is_enabled_regmap,
diff --git a/drivers/clk/qcom/clk-rcg.h b/drivers/clk/qcom/clk-rcg.h
index 687e41f91d7c..d09d06ba278e 100644
--- a/drivers/clk/qcom/clk-rcg.h
+++ b/drivers/clk/qcom/clk-rcg.h
@@ -96,6 +96,7 @@ struct clk_rcg {
 
 extern const struct clk_ops clk_rcg_ops;
 extern const struct clk_ops clk_rcg_bypass_ops;
+extern const struct clk_ops clk_rcg_lcc_ops;
 
 #define to_clk_rcg(_hw) container_of(to_clk_regmap(_hw), struct clk_rcg, clkr)
 
diff --git a/drivers/clk/qcom/lcc-ipq806x.c b/drivers/clk/qcom/lcc-ipq806x.c
index 9e27d82c0201..a6d3a6745c4d 100644
--- a/drivers/clk/qcom/lcc-ipq806x.c
+++ b/drivers/clk/qcom/lcc-ipq806x.c
@@ -386,13 +386,12 @@ static struct clk_rcg ahbix_clk = {
 	.freq_tbl = clk_tbl_ahbix,
 	.clkr = {
 		.enable_reg = 0x38,
-		.enable_mask = BIT(10), /* toggle the gfmux to select mn/pxo */
+		.enable_mask = BIT(11),
 		.hw.init = &(struct clk_init_data){
 			.name = "ahbix",
 			.parent_names = lcc_pxo_pll4,
 			.num_parents = 2,
-			.ops = &clk_rcg_ops,
-			.flags = CLK_SET_RATE_GATE,
+			.ops = &clk_rcg_lcc_ops,
 		},
 	},
 };
-- 
The Qualcomm Innovation Center, Inc. is a member of the Code Aurora Forum,
a Linux Foundation Collaborative Project

             reply	other threads:[~2015-03-06 23:41 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-03-06 23:41 Stephen Boyd [this message]
2015-03-06 23:41 ` [PATCH v2] clk: qcom: Properly change rates for ahbix clock Stephen Boyd
2015-03-12  0:43 ` Stephen Boyd
2015-03-12  0:43   ` Stephen Boyd
2015-03-12  3:42 ` Kenneth Westfield
2015-03-12  3:42   ` Kenneth Westfield

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=1425685313-6005-1-git-send-email-sboyd@codeaurora.org \
    --to=sboyd@codeaurora.org \
    --cc=kwestfie@codeaurora.org \
    --cc=linux-arm-kernel@lists.infradead.org \
    --cc=linux-arm-msm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=mturquette@linaro.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.