From: Peng Fan <[email protected]>
V3:
Add cover-letter
V2:
Added patch [2,3,4]/4 and avoid glitch when prepare
There is two bypass bit in the pll, BYPASS and EXT_BYPASS.
There is also a restriction that to avoid glitch, need set BYPASS
bit when RESETB changed from 0 to 1, otherwise there will be glitch.
However the BYPASS bit is also used as mux bit in imx8mm/imx8mn clk driver.
This means two paths touch the same bit which is wrong. So switch to use
EXT_BYPASS bit as the mux.
Peng Fan (4):
clk: imx: pll14xx: avoid glitch when set rate
clk: imx: clk-pll14xx: unbypass PLL by default
clk: imx: imx8mm: fix pll mux bit
clk: imx: imx8mn: fix pll mux bit
drivers/clk/imx/clk-imx8mm.c | 32 ++++++++++----------------------
drivers/clk/imx/clk-imx8mn.c | 32 ++++++++++----------------------
drivers/clk/imx/clk-pll14xx.c | 27 ++++++++++++++++++++++++++-
3 files changed, 46 insertions(+), 45 deletions(-)
--
2.16.4
From: Peng Fan <[email protected]>
When registering the PLL, unbypass the PLL.
The PLL has two bypass control bit, BYPASS and EXT_BYPASS.
we will expose EXT_BYPASS to clk driver for mux usage, and keep
BYPASS inside pll14xx usage. The PLL has a restriction that
when M/P change, need to RESET/BYPASS pll to avoid glitch, so
we could not expose BYPASS.
To make it easy for clk driver usage, unbypass PLL which does
not hurt current function.
Fixes: 8646d4dcc7fb ("clk: imx: Add PLLs driver for imx8mm soc")
Reviewed-by: Leonard Crestez <[email protected]>
Signed-off-by: Peng Fan <[email protected]>
---
V3:
None
V2:
New patch
drivers/clk/imx/clk-pll14xx.c | 5 +++++
1 file changed, 5 insertions(+)
diff --git a/drivers/clk/imx/clk-pll14xx.c b/drivers/clk/imx/clk-pll14xx.c
index 656f48b002dd..7a815ec76aa5 100644
--- a/drivers/clk/imx/clk-pll14xx.c
+++ b/drivers/clk/imx/clk-pll14xx.c
@@ -368,6 +368,7 @@ struct clk *imx_clk_pll14xx(const char *name, const char *parent_name,
struct clk_pll14xx *pll;
struct clk *clk;
struct clk_init_data init;
+ u32 val;
pll = kzalloc(sizeof(*pll), GFP_KERNEL);
if (!pll)
@@ -399,6 +400,10 @@ struct clk *imx_clk_pll14xx(const char *name, const char *parent_name,
pll->rate_table = pll_clk->rate_table;
pll->rate_count = pll_clk->rate_count;
+ val = readl_relaxed(pll->base + GNRL_CTL);
+ val &= ~BYPASS_MASK;
+ writel_relaxed(val, pll->base + GNRL_CTL);
+
clk = clk_register(NULL, &pll->hw);
if (IS_ERR(clk)) {
pr_err("%s: failed to register pll %s %lu\n",
--
2.16.4
From: Peng Fan <[email protected]>
pll BYPASS bit should be kept inside pll driver for glitchless freq
setting following spec. If exposing the bit, that means pll driver and
clk driver has two paths to touch this bit, which is wrong.
So use EXT_BYPASS bit here.
And drop uneeded set parent, because EXT_BYPASS default is 0.
Fixes: ba5625c3e272 ("clk: imx: Add clock driver support for imx8mm")
Suggested-by: Jacky Bai <[email protected]>
Reviewed-by: Leonard Crestez <[email protected]>
Signed-off-by: Peng Fan <[email protected]>
---
V3:
None
V2:
New patch
drivers/clk/imx/clk-imx8mm.c | 32 ++++++++++----------------------
1 file changed, 10 insertions(+), 22 deletions(-)
diff --git a/drivers/clk/imx/clk-imx8mm.c b/drivers/clk/imx/clk-imx8mm.c
index 2758e3f0d15d..067ab876911d 100644
--- a/drivers/clk/imx/clk-imx8mm.c
+++ b/drivers/clk/imx/clk-imx8mm.c
@@ -408,28 +408,16 @@ static int imx8mm_clocks_probe(struct platform_device *pdev)
clks[IMX8MM_SYS_PLL3] = imx_clk_pll14xx("sys_pll3", "sys_pll3_ref_sel", base + 0x114, &imx8mm_sys_pll);
/* PLL bypass out */
- clks[IMX8MM_AUDIO_PLL1_BYPASS] = imx_clk_mux_flags("audio_pll1_bypass", base, 4, 1, audio_pll1_bypass_sels, ARRAY_SIZE(audio_pll1_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_AUDIO_PLL2_BYPASS] = imx_clk_mux_flags("audio_pll2_bypass", base + 0x14, 4, 1, audio_pll2_bypass_sels, ARRAY_SIZE(audio_pll2_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_VIDEO_PLL1_BYPASS] = imx_clk_mux_flags("video_pll1_bypass", base + 0x28, 4, 1, video_pll1_bypass_sels, ARRAY_SIZE(video_pll1_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_DRAM_PLL_BYPASS] = imx_clk_mux_flags("dram_pll_bypass", base + 0x50, 4, 1, dram_pll_bypass_sels, ARRAY_SIZE(dram_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_GPU_PLL_BYPASS] = imx_clk_mux_flags("gpu_pll_bypass", base + 0x64, 4, 1, gpu_pll_bypass_sels, ARRAY_SIZE(gpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_VPU_PLL_BYPASS] = imx_clk_mux_flags("vpu_pll_bypass", base + 0x74, 4, 1, vpu_pll_bypass_sels, ARRAY_SIZE(vpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_ARM_PLL_BYPASS] = imx_clk_mux_flags("arm_pll_bypass", base + 0x84, 4, 1, arm_pll_bypass_sels, ARRAY_SIZE(arm_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_SYS_PLL1_BYPASS] = imx_clk_mux_flags("sys_pll1_bypass", base + 0x94, 4, 1, sys_pll1_bypass_sels, ARRAY_SIZE(sys_pll1_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_SYS_PLL2_BYPASS] = imx_clk_mux_flags("sys_pll2_bypass", base + 0x104, 4, 1, sys_pll2_bypass_sels, ARRAY_SIZE(sys_pll2_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MM_SYS_PLL3_BYPASS] = imx_clk_mux_flags("sys_pll3_bypass", base + 0x114, 4, 1, sys_pll3_bypass_sels, ARRAY_SIZE(sys_pll3_bypass_sels), CLK_SET_RATE_PARENT);
-
- /* unbypass all the plls */
- clk_set_parent(clks[IMX8MM_AUDIO_PLL1_BYPASS], clks[IMX8MM_AUDIO_PLL1]);
- clk_set_parent(clks[IMX8MM_AUDIO_PLL2_BYPASS], clks[IMX8MM_AUDIO_PLL2]);
- clk_set_parent(clks[IMX8MM_VIDEO_PLL1_BYPASS], clks[IMX8MM_VIDEO_PLL1]);
- clk_set_parent(clks[IMX8MM_DRAM_PLL_BYPASS], clks[IMX8MM_DRAM_PLL]);
- clk_set_parent(clks[IMX8MM_GPU_PLL_BYPASS], clks[IMX8MM_GPU_PLL]);
- clk_set_parent(clks[IMX8MM_VPU_PLL_BYPASS], clks[IMX8MM_VPU_PLL]);
- clk_set_parent(clks[IMX8MM_ARM_PLL_BYPASS], clks[IMX8MM_ARM_PLL]);
- clk_set_parent(clks[IMX8MM_SYS_PLL1_BYPASS], clks[IMX8MM_SYS_PLL1]);
- clk_set_parent(clks[IMX8MM_SYS_PLL2_BYPASS], clks[IMX8MM_SYS_PLL2]);
- clk_set_parent(clks[IMX8MM_SYS_PLL3_BYPASS], clks[IMX8MM_SYS_PLL3]);
+ clks[IMX8MM_AUDIO_PLL1_BYPASS] = imx_clk_mux_flags("audio_pll1_bypass", base, 16, 1, audio_pll1_bypass_sels, ARRAY_SIZE(audio_pll1_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_AUDIO_PLL2_BYPASS] = imx_clk_mux_flags("audio_pll2_bypass", base + 0x14, 16, 1, audio_pll2_bypass_sels, ARRAY_SIZE(audio_pll2_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_VIDEO_PLL1_BYPASS] = imx_clk_mux_flags("video_pll1_bypass", base + 0x28, 16, 1, video_pll1_bypass_sels, ARRAY_SIZE(video_pll1_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_DRAM_PLL_BYPASS] = imx_clk_mux_flags("dram_pll_bypass", base + 0x50, 16, 1, dram_pll_bypass_sels, ARRAY_SIZE(dram_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_GPU_PLL_BYPASS] = imx_clk_mux_flags("gpu_pll_bypass", base + 0x64, 28, 1, gpu_pll_bypass_sels, ARRAY_SIZE(gpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_VPU_PLL_BYPASS] = imx_clk_mux_flags("vpu_pll_bypass", base + 0x74, 28, 1, vpu_pll_bypass_sels, ARRAY_SIZE(vpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_ARM_PLL_BYPASS] = imx_clk_mux_flags("arm_pll_bypass", base + 0x84, 28, 1, arm_pll_bypass_sels, ARRAY_SIZE(arm_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_SYS_PLL1_BYPASS] = imx_clk_mux_flags("sys_pll1_bypass", base + 0x94, 28, 1, sys_pll1_bypass_sels, ARRAY_SIZE(sys_pll1_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_SYS_PLL2_BYPASS] = imx_clk_mux_flags("sys_pll2_bypass", base + 0x104, 28, 1, sys_pll2_bypass_sels, ARRAY_SIZE(sys_pll2_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MM_SYS_PLL3_BYPASS] = imx_clk_mux_flags("sys_pll3_bypass", base + 0x114, 28, 1, sys_pll3_bypass_sels, ARRAY_SIZE(sys_pll3_bypass_sels), CLK_SET_RATE_PARENT);
/* PLL out gate */
clks[IMX8MM_AUDIO_PLL1_OUT] = imx_clk_gate("audio_pll1_out", "audio_pll1_bypass", base, 13);
--
2.16.4
From: Peng Fan <[email protected]>
According to PLL1443XA and PLL1416X spec,
"When BYPASS is 0 and RESETB is changed from 0 to 1, FOUT starts to
output unstable clock until lock time passes. PLL1416X/PLL1443XA may
generate a glitch at FOUT."
So set BYPASS when RESETB is changed from 0 to 1 to avoid glitch.
In the end of set rate, BYPASS will be cleared.
When prepare clock, also need to take care to avoid glitch. So
we also follow Spec to set BYPASS before RESETB changed from 0 to 1.
And add a check if the RESETB is already 0, directly return 0;
Fixes: 8646d4dcc7fb ("clk: imx: Add PLLs driver for imx8mm soc")
Reviewed-by: Leonard Crestez <[email protected]>
Signed-off-by: Peng Fan <[email protected]>
---
V3:
None
V2:
Avoid glitch when prepare
update commit log
drivers/clk/imx/clk-pll14xx.c | 22 +++++++++++++++++++++-
1 file changed, 21 insertions(+), 1 deletion(-)
diff --git a/drivers/clk/imx/clk-pll14xx.c b/drivers/clk/imx/clk-pll14xx.c
index b7213023b238..656f48b002dd 100644
--- a/drivers/clk/imx/clk-pll14xx.c
+++ b/drivers/clk/imx/clk-pll14xx.c
@@ -191,6 +191,10 @@ static int clk_pll1416x_set_rate(struct clk_hw *hw, unsigned long drate,
tmp &= ~RST_MASK;
writel_relaxed(tmp, pll->base);
+ /* Enable BYPASS */
+ tmp |= BYPASS_MASK;
+ writel(tmp, pll->base);
+
div_val = (rate->mdiv << MDIV_SHIFT) | (rate->pdiv << PDIV_SHIFT) |
(rate->sdiv << SDIV_SHIFT);
writel_relaxed(div_val, pll->base + 0x4);
@@ -250,6 +254,10 @@ static int clk_pll1443x_set_rate(struct clk_hw *hw, unsigned long drate,
tmp &= ~RST_MASK;
writel_relaxed(tmp, pll->base);
+ /* Enable BYPASS */
+ tmp |= BYPASS_MASK;
+ writel_relaxed(tmp, pll->base);
+
div_val = (rate->mdiv << MDIV_SHIFT) | (rate->pdiv << PDIV_SHIFT) |
(rate->sdiv << SDIV_SHIFT);
writel_relaxed(div_val, pll->base + 0x4);
@@ -283,16 +291,28 @@ static int clk_pll14xx_prepare(struct clk_hw *hw)
{
struct clk_pll14xx *pll = to_clk_pll14xx(hw);
u32 val;
+ int ret;
/*
* RESETB = 1 from 0, PLL starts its normal
* operation after lock time
*/
val = readl_relaxed(pll->base + GNRL_CTL);
+ if (val & RST_MASK)
+ return 0;
+ val |= BYPASS_MASK;
+ writel_relaxed(val, pll->base + GNRL_CTL);
val |= RST_MASK;
writel_relaxed(val, pll->base + GNRL_CTL);
- return clk_pll14xx_wait_lock(pll);
+ ret = clk_pll14xx_wait_lock(pll);
+ if (ret)
+ return ret;
+
+ val &= ~BYPASS_MASK;
+ writel_relaxed(val, pll->base + GNRL_CTL);
+
+ return 0;
}
static int clk_pll14xx_is_prepared(struct clk_hw *hw)
--
2.16.4
From: Peng Fan <[email protected]>
pll BYPASS bit should be kept inside pll driver for glitchless freq
setting following spec. If exposing the bit, that means pll driver and
clk driver has two paths to touch this bit, which is wrong.
So use EXT_BYPASS bit here.
And drop uneeded set parent, because EXT_BYPASS default is 0.
Suggested-by: Jacky Bai <[email protected]>
Reviewed-by: Leonard Crestez <[email protected]>
Signed-off-by: Peng Fan <[email protected]>
---
V3:
None
V2:
New patch
drivers/clk/imx/clk-imx8mn.c | 32 ++++++++++----------------------
1 file changed, 10 insertions(+), 22 deletions(-)
diff --git a/drivers/clk/imx/clk-imx8mn.c b/drivers/clk/imx/clk-imx8mn.c
index f41116d59749..f767d18679ea 100644
--- a/drivers/clk/imx/clk-imx8mn.c
+++ b/drivers/clk/imx/clk-imx8mn.c
@@ -421,28 +421,16 @@ static int imx8mn_clocks_probe(struct platform_device *pdev)
clks[IMX8MN_SYS_PLL3] = imx_clk_pll14xx("sys_pll3", "sys_pll3_ref_sel", base + 0x114, &imx8mn_sys_pll);
/* PLL bypass out */
- clks[IMX8MN_AUDIO_PLL1_BYPASS] = imx_clk_mux_flags("audio_pll1_bypass", base, 4, 1, audio_pll1_bypass_sels, ARRAY_SIZE(audio_pll1_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_AUDIO_PLL2_BYPASS] = imx_clk_mux_flags("audio_pll2_bypass", base + 0x14, 4, 1, audio_pll2_bypass_sels, ARRAY_SIZE(audio_pll2_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_VIDEO_PLL1_BYPASS] = imx_clk_mux_flags("video_pll1_bypass", base + 0x28, 4, 1, video_pll1_bypass_sels, ARRAY_SIZE(video_pll1_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_DRAM_PLL_BYPASS] = imx_clk_mux_flags("dram_pll_bypass", base + 0x50, 4, 1, dram_pll_bypass_sels, ARRAY_SIZE(dram_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_GPU_PLL_BYPASS] = imx_clk_mux_flags("gpu_pll_bypass", base + 0x64, 4, 1, gpu_pll_bypass_sels, ARRAY_SIZE(gpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_VPU_PLL_BYPASS] = imx_clk_mux_flags("vpu_pll_bypass", base + 0x74, 4, 1, vpu_pll_bypass_sels, ARRAY_SIZE(vpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_ARM_PLL_BYPASS] = imx_clk_mux_flags("arm_pll_bypass", base + 0x84, 4, 1, arm_pll_bypass_sels, ARRAY_SIZE(arm_pll_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_SYS_PLL1_BYPASS] = imx_clk_mux_flags("sys_pll1_bypass", base + 0x94, 4, 1, sys_pll1_bypass_sels, ARRAY_SIZE(sys_pll1_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_SYS_PLL2_BYPASS] = imx_clk_mux_flags("sys_pll2_bypass", base + 0x104, 4, 1, sys_pll2_bypass_sels, ARRAY_SIZE(sys_pll2_bypass_sels), CLK_SET_RATE_PARENT);
- clks[IMX8MN_SYS_PLL3_BYPASS] = imx_clk_mux_flags("sys_pll3_bypass", base + 0x114, 4, 1, sys_pll3_bypass_sels, ARRAY_SIZE(sys_pll3_bypass_sels), CLK_SET_RATE_PARENT);
-
- /* unbypass all the plls */
- clk_set_parent(clks[IMX8MN_AUDIO_PLL1_BYPASS], clks[IMX8MN_AUDIO_PLL1]);
- clk_set_parent(clks[IMX8MN_AUDIO_PLL2_BYPASS], clks[IMX8MN_AUDIO_PLL2]);
- clk_set_parent(clks[IMX8MN_VIDEO_PLL1_BYPASS], clks[IMX8MN_VIDEO_PLL1]);
- clk_set_parent(clks[IMX8MN_DRAM_PLL_BYPASS], clks[IMX8MN_DRAM_PLL]);
- clk_set_parent(clks[IMX8MN_GPU_PLL_BYPASS], clks[IMX8MN_GPU_PLL]);
- clk_set_parent(clks[IMX8MN_VPU_PLL_BYPASS], clks[IMX8MN_VPU_PLL]);
- clk_set_parent(clks[IMX8MN_ARM_PLL_BYPASS], clks[IMX8MN_ARM_PLL]);
- clk_set_parent(clks[IMX8MN_SYS_PLL1_BYPASS], clks[IMX8MN_SYS_PLL1]);
- clk_set_parent(clks[IMX8MN_SYS_PLL2_BYPASS], clks[IMX8MN_SYS_PLL2]);
- clk_set_parent(clks[IMX8MN_SYS_PLL3_BYPASS], clks[IMX8MN_SYS_PLL3]);
+ clks[IMX8MN_AUDIO_PLL1_BYPASS] = imx_clk_mux_flags("audio_pll1_bypass", base, 16, 1, audio_pll1_bypass_sels, ARRAY_SIZE(audio_pll1_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_AUDIO_PLL2_BYPASS] = imx_clk_mux_flags("audio_pll2_bypass", base + 0x14, 16, 1, audio_pll2_bypass_sels, ARRAY_SIZE(audio_pll2_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_VIDEO_PLL1_BYPASS] = imx_clk_mux_flags("video_pll1_bypass", base + 0x28, 16, 1, video_pll1_bypass_sels, ARRAY_SIZE(video_pll1_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_DRAM_PLL_BYPASS] = imx_clk_mux_flags("dram_pll_bypass", base + 0x50, 16, 1, dram_pll_bypass_sels, ARRAY_SIZE(dram_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_GPU_PLL_BYPASS] = imx_clk_mux_flags("gpu_pll_bypass", base + 0x64, 28, 1, gpu_pll_bypass_sels, ARRAY_SIZE(gpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_VPU_PLL_BYPASS] = imx_clk_mux_flags("vpu_pll_bypass", base + 0x74, 28, 1, vpu_pll_bypass_sels, ARRAY_SIZE(vpu_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_ARM_PLL_BYPASS] = imx_clk_mux_flags("arm_pll_bypass", base + 0x84, 28, 1, arm_pll_bypass_sels, ARRAY_SIZE(arm_pll_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_SYS_PLL1_BYPASS] = imx_clk_mux_flags("sys_pll1_bypass", base + 0x94, 28, 1, sys_pll1_bypass_sels, ARRAY_SIZE(sys_pll1_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_SYS_PLL2_BYPASS] = imx_clk_mux_flags("sys_pll2_bypass", base + 0x104, 28, 1, sys_pll2_bypass_sels, ARRAY_SIZE(sys_pll2_bypass_sels), CLK_SET_RATE_PARENT);
+ clks[IMX8MN_SYS_PLL3_BYPASS] = imx_clk_mux_flags("sys_pll3_bypass", base + 0x114, 28, 1, sys_pll3_bypass_sels, ARRAY_SIZE(sys_pll3_bypass_sels), CLK_SET_RATE_PARENT);
/* PLL out gate */
clks[IMX8MN_AUDIO_PLL1_OUT] = imx_clk_gate("audio_pll1_out", "audio_pll1_bypass", base, 13);
--
2.16.4
Hi Stephen, Shawn,
> Subject: [PATCH V3 0/4] clk: imx8m: fix glitch/mux
Sorry to ping early. Is there a chance to land this patchset in 5.3 release?
Thanks,
Peng.
>
> From: Peng Fan <[email protected]>
>
> V3:
> Add cover-letter
>
> V2:
> Added patch [2,3,4]/4 and avoid glitch when prepare
>
> There is two bypass bit in the pll, BYPASS and EXT_BYPASS.
> There is also a restriction that to avoid glitch, need set BYPASS bit when
> RESETB changed from 0 to 1, otherwise there will be glitch.
>
> However the BYPASS bit is also used as mux bit in imx8mm/imx8mn clk driver.
>
> This means two paths touch the same bit which is wrong. So switch to use
> EXT_BYPASS bit as the mux.
>
> Peng Fan (4):
> clk: imx: pll14xx: avoid glitch when set rate
> clk: imx: clk-pll14xx: unbypass PLL by default
> clk: imx: imx8mm: fix pll mux bit
> clk: imx: imx8mn: fix pll mux bit
>
> drivers/clk/imx/clk-imx8mm.c | 32 ++++++++++----------------------
> drivers/clk/imx/clk-imx8mn.c | 32 ++++++++++----------------------
> drivers/clk/imx/clk-pll14xx.c | 27 ++++++++++++++++++++++++++-
> 3 files changed, 46 insertions(+), 45 deletions(-)
>
> --
> 2.16.4
Quoting Peng Fan (2019-09-16 23:20:15)
> Hi Stephen, Shawn,
>
> > Subject: [PATCH V3 0/4] clk: imx8m: fix glitch/mux
>
> Sorry to ping early. Is there a chance to land this patchset in 5.3 release?
>
No, it won't be in 5.3 because that version is released. Shawn already
sent the PR for 5.4 too so this will most likely be in v5.5 at the
earliest.
Hi Stephen,
> Subject: RE: [PATCH V3 0/4] clk: imx8m: fix glitch/mux
>
> Quoting Peng Fan (2019-09-16 23:20:15)
> > Hi Stephen, Shawn,
> >
> > > Subject: [PATCH V3 0/4] clk: imx8m: fix glitch/mux
> >
> > Sorry to ping early. Is there a chance to land this patchset in 5.3 release?
> >
>
> No, it won't be in 5.3 because that version is released. Shawn already sent the
> PR for 5.4 too so this will most likely be in v5.5 at the earliest.
Thanks for the info. But this patchset is bugfix, so hope this could be accepted in 5.4.
Thanks,
Peng.
Quoting Peng Fan (2019-09-08 20:39:34)
> From: Peng Fan <[email protected]>
>
> According to PLL1443XA and PLL1416X spec,
> "When BYPASS is 0 and RESETB is changed from 0 to 1, FOUT starts to
> output unstable clock until lock time passes. PLL1416X/PLL1443XA may
> generate a glitch at FOUT."
>
> So set BYPASS when RESETB is changed from 0 to 1 to avoid glitch.
> In the end of set rate, BYPASS will be cleared.
>
> When prepare clock, also need to take care to avoid glitch. So
> we also follow Spec to set BYPASS before RESETB changed from 0 to 1.
> And add a check if the RESETB is already 0, directly return 0;
>
> Fixes: 8646d4dcc7fb ("clk: imx: Add PLLs driver for imx8mm soc")
> Reviewed-by: Leonard Crestez <[email protected]>
> Signed-off-by: Peng Fan <[email protected]>
> ---
Applied to clk-next
Quoting Peng Fan (2019-09-08 20:39:39)
> From: Peng Fan <[email protected]>
>
> When registering the PLL, unbypass the PLL.
> The PLL has two bypass control bit, BYPASS and EXT_BYPASS.
> we will expose EXT_BYPASS to clk driver for mux usage, and keep
> BYPASS inside pll14xx usage. The PLL has a restriction that
> when M/P change, need to RESET/BYPASS pll to avoid glitch, so
> we could not expose BYPASS.
>
> To make it easy for clk driver usage, unbypass PLL which does
> not hurt current function.
>
> Fixes: 8646d4dcc7fb ("clk: imx: Add PLLs driver for imx8mm soc")
> Reviewed-by: Leonard Crestez <[email protected]>
> Signed-off-by: Peng Fan <[email protected]>
> ---
Applied to clk-next
Quoting Peng Fan (2019-09-08 20:39:44)
> From: Peng Fan <[email protected]>
>
> pll BYPASS bit should be kept inside pll driver for glitchless freq
> setting following spec. If exposing the bit, that means pll driver and
> clk driver has two paths to touch this bit, which is wrong.
>
> So use EXT_BYPASS bit here.
>
> And drop uneeded set parent, because EXT_BYPASS default is 0.
>
> Fixes: ba5625c3e272 ("clk: imx: Add clock driver support for imx8mm")
> Suggested-by: Jacky Bai <[email protected]>
> Reviewed-by: Leonard Crestez <[email protected]>
> Signed-off-by: Peng Fan <[email protected]>
> ---
Applied to clk-next
Quoting Peng Fan (2019-09-08 20:39:50)
> From: Peng Fan <[email protected]>
>
> pll BYPASS bit should be kept inside pll driver for glitchless freq
> setting following spec. If exposing the bit, that means pll driver and
> clk driver has two paths to touch this bit, which is wrong.
>
> So use EXT_BYPASS bit here.
>
> And drop uneeded set parent, because EXT_BYPASS default is 0.
>
> Suggested-by: Jacky Bai <[email protected]>
> Reviewed-by: Leonard Crestez <[email protected]>
> Signed-off-by: Peng Fan <[email protected]>
> ---
Applied to clk-next
Quoting Peng Fan (2019-09-17 22:45:20)
> Hi Stephen,
>
> > Subject: RE: [PATCH V3 0/4] clk: imx8m: fix glitch/mux
> >
> > Quoting Peng Fan (2019-09-16 23:20:15)
> > > Hi Stephen, Shawn,
> > >
> > > > Subject: [PATCH V3 0/4] clk: imx8m: fix glitch/mux
> > >
> > > Sorry to ping early. Is there a chance to land this patchset in 5.3 release?
> > >
> >
> > No, it won't be in 5.3 because that version is released. Shawn already sent the
> > PR for 5.4 too so this will most likely be in v5.5 at the earliest.
>
> Thanks for the info. But this patchset is bugfix, so hope this could be accepted in 5.4.
>
Ok. Then let's throw it into 5.4 PR and see what goes wrong.