openwrt/target/linux/bcm27xx/patches-5.10/950-0136-clk-bcm2835-Add-support-for-setting-leaf-clock-rates.patch
Rui Salvaterra 1343acc8cd kernel: bump 5.10 to 5.10.94
Deleted (upstreamed):
bcm27xx/patches-5.10/950-0669-drm-vc4-hdmi-Make-sure-the-device-is-powered-with-CE.patch [1]
bcm27xx/patches-5.10/950-0672-drm-vc4-hdmi-Move-initial-register-read-after-pm_run.patch [1]
gemini/patches-5.10/0003-ARM-dts-gemini-NAS4220-B-fis-index-block-with-128-Ki.patch [2]

Manually rebased:
bcm27xx/patches-5.10/950-0675-drm-vc4-hdmi-Drop-devm-interrupt-handler-for-CEC-int.patch

Manually reverted:
generic/pending-5.10/860-Revert-ASoC-mediatek-Check-for-error-clk-pointer.patch [3]

[1] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v5.10.94&id=55b10b88ac8654fc2f31518aa349a2e643b37f18
[2] https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/commit/?h=v5.10.94&id=958a8819d41420d7a74ed922a09cacc0ba3a4218
[3] https://lore.kernel.org/all/trinity-2a727d96-0335-4d03-8f30-e22a0e10112d-1643363480085@3c-app-gmx-bap33/

Signed-off-by: Rui Salvaterra <rsalvaterra@gmail.com>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
2022-02-01 23:34:07 +00:00

54 lines
1.8 KiB
Diff

From a1dcaba80910c0196b6e51111ba3890372067e0e Mon Sep 17 00:00:00 2001
From: Eric Anholt <eric@anholt.net>
Date: Thu, 2 May 2019 15:11:05 -0700
Subject: [PATCH] clk: bcm2835: Add support for setting leaf clock
rates while running.
As long as you wait for !BUSY, you can do glitch-free updates of clock
rate while the clock is running.
Signed-off-by: Eric Anholt <eric@anholt.net>
---
drivers/clk/bcm/clk-bcm2835.c | 22 +++++++++++++---------
1 file changed, 13 insertions(+), 9 deletions(-)
--- a/drivers/clk/bcm/clk-bcm2835.c
+++ b/drivers/clk/bcm/clk-bcm2835.c
@@ -1110,15 +1110,19 @@ static int bcm2835_clock_set_rate(struct
spin_lock(&cprman->regs_lock);
- /*
- * Setting up frac support
- *
- * In principle it is recommended to stop/start the clock first,
- * but as we set CLK_SET_RATE_GATE during registration of the
- * clock this requirement should be take care of by the
- * clk-framework.
+ ctl = cprman_read(cprman, data->ctl_reg);
+
+ /* If the clock is running, we have to pause clock generation while
+ * updating the control and div regs. This is glitchless (no clock
+ * signals generated faster than the rate) but each reg access is two
+ * OSC cycles so the clock will slow down for a moment.
*/
- ctl = cprman_read(cprman, data->ctl_reg) & ~CM_FRAC;
+ if (ctl & CM_ENABLE) {
+ cprman_write(cprman, data->ctl_reg, ctl & ~CM_ENABLE);
+ bcm2835_clock_wait_busy(clock);
+ }
+
+ ctl &= ~CM_FRAC;
ctl |= (div & CM_DIV_FRAC_MASK) ? CM_FRAC : 0;
cprman_write(cprman, data->ctl_reg, ctl);
@@ -1494,7 +1498,7 @@ static struct clk_hw *bcm2835_register_c
init.ops = &bcm2835_vpu_clock_clk_ops;
} else {
init.ops = &bcm2835_clock_clk_ops;
- init.flags |= CLK_SET_RATE_GATE | CLK_SET_PARENT_GATE;
+ init.flags |= CLK_SET_PARENT_GATE;
/* If the clock wasn't actually enabled at boot, it's not
* critical.