Hi,
While working on peric1, I've noticed a few issues in the peric0 area
and these patches are the result. They should all be pretty
self-explanatory.
Cheers,
Andre'
arch/arm64/boot/dts/exynos/google/gs101.dtsi | 11 ++++++-----
drivers/clk/samsung/clk-gs101.c | 8 +++-----
2 files changed, 9 insertions(+), 10 deletions(-)
Changes in v2:
* new patch #6
- fix a typo in the commit messages of patches #3 and #5 (André)
- add some empty lines to commit messages (Sam)
- collect Reviewed-by: Tested-by: tags
This pclk clock is required any time we access the pinctrl registers of
this block.
Since pinctrl-samsung doesn't support a clock at the moment, we just
keep the kernel from disabling it at boot, until we have an update for
pinctrl-samsung to handle this required clock, at which point we'll be
able to drop the flag again.
Fixes: 893f133a040b ("clk: samsung: gs101: add support for cmu_peric0")
Signed-off-by: André Draszik <[email protected]>
Reviewed-by: Sam Protsenko <[email protected]>
Reviewed-by: Tudor Ambarus <[email protected]>
Reviewed-by: Peter Griffin <[email protected]>
---
v2: collect Reviewed-by: tags
---
drivers/clk/samsung/clk-gs101.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/clk/samsung/clk-gs101.c b/drivers/clk/samsung/clk-gs101.c
index 4a0520e825b6..61bb0dcf84ee 100644
--- a/drivers/clk/samsung/clk-gs101.c
+++ b/drivers/clk/samsung/clk-gs101.c
@@ -2848,7 +2848,7 @@ static const struct samsung_gate_clock peric0_gate_clks[] __initconst = {
GATE(CLK_GOUT_PERIC0_GPIO_PERIC0_PCLK,
"gout_peric0_gpio_peric0_pclk", "mout_peric0_bus_user",
CLK_CON_GAT_GOUT_BLK_PERIC0_UID_GPIO_PERIC0_IPCLKPORT_PCLK,
- 21, 0, 0),
+ 21, CLK_IGNORE_UNUSED, 0),
/* Disabling this clock makes the system hang. Mark the clock as critical. */
GATE(CLK_GOUT_PERIC0_LHM_AXI_P_PERIC0_I_CLK,
"gout_peric0_lhm_axi_p_peric0_i_clk", "mout_peric0_bus_user",
--
2.43.0.429.g432eaa2c6b-goog
Wrong pclk clocks have been used in this usi_uart instance here. For
USI and UART, we need the ipclk and pclk, where pclk is the bus clock.
Without it, nothing can work.
It is unclear what exactly is using USI0_UART_CLK, but it is not
required for the IP to be operational at this stage, while pclk is.
This also brings the DT in line with the clock names expected by the
usi and uart drivers.
Update the DTSI accordingly.
Fixes: d97b6c902a40 ("arm64: dts: exynos: gs101: update USI UART to use peric0 clocks")
Signed-off-by: André Draszik <[email protected]>
Reviewed-by: Sam Protsenko <[email protected]>
Reviewed-by: Tudor Ambarus <[email protected]>
Tested-by: Tudor Ambarus <[email protected]>
---
v2:
* fix a typo in and add an empty line to the commit message
* collect Reviewed-by: Tested-by: tags
---
arch/arm64/boot/dts/exynos/google/gs101.dtsi | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/arm64/boot/dts/exynos/google/gs101.dtsi b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
index e5b665be2d62..f93e937d2726 100644
--- a/arch/arm64/boot/dts/exynos/google/gs101.dtsi
+++ b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
@@ -410,7 +410,7 @@ usi_uart: usi@10a000c0 {
ranges;
#address-cells = <1>;
#size-cells = <1>;
- clocks = <&cmu_peric0 CLK_GOUT_PERIC0_CLK_PERIC0_USI0_UART_CLK>,
+ clocks = <&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP1_PCLK_0>,
<&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP1_IPCLK_0>;
clock-names = "pclk", "ipclk";
samsung,sysreg = <&sysreg_peric0 0x1020>;
@@ -422,7 +422,7 @@ serial_0: serial@10a00000 {
reg = <0x10a00000 0xc0>;
interrupts = <GIC_SPI 634
IRQ_TYPE_LEVEL_HIGH 0>;
- clocks = <&cmu_peric0 CLK_GOUT_PERIC0_CLK_PERIC0_USI0_UART_CLK>,
+ clocks = <&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP1_PCLK_0>,
<&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP1_IPCLK_0>;
clock-names = "uart", "clk_uart_baud0";
samsung,uart-fifosize = <256>;
--
2.43.0.429.g432eaa2c6b-goog
While commit 6d44d1a1fb62 ("arm64: dts: exynos: gs101: define USI8 with
I2C configuration") states that the USI8 CONFIG is 0 at reset, the boot
loader has configured it by the time Linux runs and it has a different
value at this stage.
Since we want board DTS files to explicitly select the mode, we should
set it to none here so as to ensure things don't work by accident and
to make it clear that board DTS actually need to set the mode based on
the configuration.
Fixes: 6d44d1a1fb62 ("arm64: dts: exynos: gs101: define USI8 with I2C configuration")
Signed-off-by: André Draszik <[email protected]>
Reviewed-by: Sam Protsenko <[email protected]>
Reviewed-by: Tudor Ambarus <[email protected]>
Reviewed-by: Peter Griffin <[email protected]>
---
v2: collect Reviewed-by: tags
---
arch/arm64/boot/dts/exynos/google/gs101.dtsi | 1 +
1 file changed, 1 insertion(+)
diff --git a/arch/arm64/boot/dts/exynos/google/gs101.dtsi b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
index aaac04df5e65..bc251e565be6 100644
--- a/arch/arm64/boot/dts/exynos/google/gs101.dtsi
+++ b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
@@ -384,6 +384,7 @@ usi8: usi@109700c0 {
<&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP0_IPCLK_7>;
clock-names = "pclk", "ipclk";
samsung,sysreg = <&sysreg_peric0 0x101c>;
+ samsung,mode = <USI_V2_NONE>;
status = "disabled";
hsi2c_8: i2c@10970000 {
--
2.43.0.429.g432eaa2c6b-goog
The preferred order for these is pinctrl-0 pinctrl-names.
Update the DTSI accordingly.
Signed-off-by: André Draszik <[email protected]>
Suggested-by: Sam Protsenko <[email protected]>
---
v2: new patch in this series
---
arch/arm64/boot/dts/exynos/google/gs101.dtsi | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/arch/arm64/boot/dts/exynos/google/gs101.dtsi b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
index f93e937d2726..195533fe04c6 100644
--- a/arch/arm64/boot/dts/exynos/google/gs101.dtsi
+++ b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
@@ -394,8 +394,8 @@ hsi2c_8: i2c@10970000 {
interrupts = <GIC_SPI 642 IRQ_TYPE_LEVEL_HIGH 0>;
#address-cells = <1>;
#size-cells = <0>;
- pinctrl-names = "default";
pinctrl-0 = <&hsi2c8_bus>;
+ pinctrl-names = "default";
clocks = <&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP0_IPCLK_7>,
<&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP0_PCLK_7>;
clock-names = "hsi2c", "hsi2c_pclk";
--
2.43.0.429.g432eaa2c6b-goog
The peric0_top1_ipclk_0 and peric0_top1_pclk_0 are the clocks going to
peric0/uart_usi, with pclk being the bus clock. Without pclk running,
any bus access will hang.
Unfortunately, in commit d97b6c902a40 ("arm64: dts: exynos: gs101:
update USI UART to use peric0 clocks") the gs101 DT ended up specifying
an incorrect pclk in the respective node and instead the two clocks
here were marked as critical.
We have fixed the gs101 DT and can therefore drop this incorrect
work-around here, the uart driver will claim these clocks as needed.
Note that this commit has the side-effect of causing earlycon to stop
to work sometime into the boot for two reasons:
* peric0_top1_ipclk_0 requires its parent gout_cmu_peric0_ip to be
running, but because earlycon doesn't deal with clocks that
parent will be disabled when none of the other drivers that
actually deal with clocks correctly require it to be running and
the real serial driver (which does deal with clocks) hasn't taken
over yet
* hand-over between earlycon and serial driver appears to be
fragile and clocks get enabled and disabled a few times, which
also causes register access to hang while earlycon is still
active
Nonetheless we shouldn't keep these clocks running unconditionally just
for earlycon. Clocks should be disabled where possible. If earlycon is
required in the future, e.g. for debug, this commit can simply be
reverted (locally!).
Fixes: 893f133a040b ("clk: samsung: gs101: add support for cmu_peric0")
Signed-off-by: André Draszik <[email protected]>
Reviewed-by: Tudor Ambarus <[email protected]>
Reviewed-by: Sam Protsenko <[email protected]>
---
v2:
* collect Reviewed-by: tags
---
drivers/clk/samsung/clk-gs101.c | 6 ++----
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/drivers/clk/samsung/clk-gs101.c b/drivers/clk/samsung/clk-gs101.c
index 61bb0dcf84ee..5c338ac9231c 100644
--- a/drivers/clk/samsung/clk-gs101.c
+++ b/drivers/clk/samsung/clk-gs101.c
@@ -2982,20 +2982,18 @@ static const struct samsung_gate_clock peric0_gate_clks[] __initconst = {
"gout_peric0_peric0_top0_pclk_9", "mout_peric0_bus_user",
CLK_CON_GAT_GOUT_BLK_PERIC0_UID_PERIC0_TOP0_IPCLKPORT_PCLK_9,
21, 0, 0),
- /* Disabling this clock makes the system hang. Mark the clock as critical. */
GATE(CLK_GOUT_PERIC0_PERIC0_TOP1_IPCLK_0,
"gout_peric0_peric0_top1_ipclk_0", "dout_peric0_usi0_uart",
CLK_CON_GAT_GOUT_BLK_PERIC0_UID_PERIC0_TOP1_IPCLKPORT_IPCLK_0,
- 21, CLK_IS_CRITICAL, 0),
+ 21, 0, 0),
GATE(CLK_GOUT_PERIC0_PERIC0_TOP1_IPCLK_2,
"gout_peric0_peric0_top1_ipclk_2", "dout_peric0_usi14_usi",
CLK_CON_GAT_GOUT_BLK_PERIC0_UID_PERIC0_TOP1_IPCLKPORT_IPCLK_2,
21, 0, 0),
- /* Disabling this clock makes the system hang. Mark the clock as critical. */
GATE(CLK_GOUT_PERIC0_PERIC0_TOP1_PCLK_0,
"gout_peric0_peric0_top1_pclk_0", "mout_peric0_bus_user",
CLK_CON_GAT_GOUT_BLK_PERIC0_UID_PERIC0_TOP1_IPCLKPORT_PCLK_0,
- 21, CLK_IS_CRITICAL, 0),
+ 21, 0, 0),
GATE(CLK_GOUT_PERIC0_PERIC0_TOP1_PCLK_2,
"gout_peric0_peric0_top1_pclk_2", "mout_peric0_bus_user",
CLK_CON_GAT_GOUT_BLK_PERIC0_UID_PERIC0_TOP1_IPCLKPORT_PCLK_2,
--
2.43.0.429.g432eaa2c6b-goog
Wrong pclk clocks have been used in this usi8 instance here. For USI
and I2C, we need the ipclk and pclk, where pclk is the bus clock.
Without it, nothing can work.
It is unclear what exactly is using USI8_USI_CLK, but it is not
required for the IP to be operational at this stage, while pclk is.
This also brings the DT in line with the clock names expected by the
usi and i2c drivers.
Update the DTSI accordingly.
Fixes: 6d44d1a1fb62 ("arm64: dts: exynos: gs101: define USI8 with I2C configuration")
Signed-off-by: André Draszik <[email protected]>
Reviewed-by: Sam Protsenko <[email protected]>
Reviewed-by: Tudor Ambarus <[email protected]>
Tested-by: Tudor Ambarus <[email protected]>
---
v2:
* add an empty line to commit message
* collect Reviewed-by: Tested-by: tags
---
arch/arm64/boot/dts/exynos/google/gs101.dtsi | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/arch/arm64/boot/dts/exynos/google/gs101.dtsi b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
index bc251e565be6..e5b665be2d62 100644
--- a/arch/arm64/boot/dts/exynos/google/gs101.dtsi
+++ b/arch/arm64/boot/dts/exynos/google/gs101.dtsi
@@ -380,7 +380,7 @@ usi8: usi@109700c0 {
ranges;
#address-cells = <1>;
#size-cells = <1>;
- clocks = <&cmu_peric0 CLK_GOUT_PERIC0_CLK_PERIC0_USI8_USI_CLK>,
+ clocks = <&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP0_PCLK_7>,
<&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP0_IPCLK_7>;
clock-names = "pclk", "ipclk";
samsung,sysreg = <&sysreg_peric0 0x101c>;
@@ -397,7 +397,7 @@ hsi2c_8: i2c@10970000 {
pinctrl-names = "default";
pinctrl-0 = <&hsi2c8_bus>;
clocks = <&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP0_IPCLK_7>,
- <&cmu_peric0 CLK_GOUT_PERIC0_CLK_PERIC0_USI8_USI_CLK>;
+ <&cmu_peric0 CLK_GOUT_PERIC0_PERIC0_TOP0_PCLK_7>;
clock-names = "hsi2c", "hsi2c_pclk";
status = "disabled";
};
--
2.43.0.429.g432eaa2c6b-goog
On 30/01/2024 10:36, André Draszik wrote:
> This pclk clock is required any time we access the pinctrl registers of
> this block.
>
> Since pinctrl-samsung doesn't support a clock at the moment, we just
> keep the kernel from disabling it at boot, until we have an update for
> pinctrl-samsung to handle this required clock, at which point we'll be
> able to drop the flag again.
>
> Fixes: 893f133a040b ("clk: samsung: gs101: add support for cmu_peric0")
Dropped fixes tag. The driver looks correct, it's pinctrl issue.
I really dislike how these patches are inter-mixed with DTS. Makes
applying difficult and confuses about dependencies.
I assume there are no dependencies here.
I am repeating this and repeating, but in future I will just reject the
patches:
Your DTS and driver changes cannot depend on each other for new feature
submissions.
Best regards,
Krzysztof
On 30/01/2024 10:36, André Draszik wrote:
> The peric0_top1_ipclk_0 and peric0_top1_pclk_0 are the clocks going to
> peric0/uart_usi, with pclk being the bus clock. Without pclk running,
> any bus access will hang.
> Unfortunately, in commit d97b6c902a40 ("arm64: dts: exynos: gs101:
> update USI UART to use peric0 clocks") the gs101 DT ended up specifying
> an incorrect pclk in the respective node and instead the two clocks
> here were marked as critical.
>
> We have fixed the gs101 DT and can therefore drop this incorrect
> work-around here, the uart driver will claim these clocks as needed.
How did you fixed the DTS? Which commit did it? Are we going back to
basics of driver changes depending on DTS?
Best regards,
Krzysztof
On 30/01/2024 10:36, André Draszik wrote:
> The preferred order for these is pinctrl-0 pinctrl-names.
>
> Update the DTSI accordingly.
>
> Signed-off-by: André Draszik <[email protected]>
> Suggested-by: Sam Protsenko <[email protected]>
>
> ---
> v2: new patch in this series
> ---
> arch/arm64/boot/dts/exynos/google/gs101.dtsi | 2 +-
This is trivial cleanup, which is fine but then do it for entire Google
GS DTSI and DTS, not only one case.
Best regards,
Krzysztof
On 30/01/2024 10:36, André Draszik wrote:
> While commit 6d44d1a1fb62 ("arm64: dts: exynos: gs101: define USI8 with
> I2C configuration") states that the USI8 CONFIG is 0 at reset, the boot
> loader has configured it by the time Linux runs and it has a different
> value at this stage.
This issue was pointed during review:
https://lore.kernel.org/all/CAPLW+4=U9DBmwgxyWz3cy=V-Ui7s2Z9um4xbEuyax1o=0zB_NA@mail.gmail.com/
Yet you posted new version of patchset not implementing this, just to do
it week later as new patch.
Sorry guys, it seems you need much more time to accept and go through
review, I will use two weeks delay time for applying GS patches.
Now, for the patch, we don't do it for any other nodes which have 0 as
reset value and we do not know what bootloader does there. Bootloader
also can change.
This is a required property, therefore please explain me how, really
how, this can happen:
" we should
set it to none here so as to ensure things don't work by accident"
NAK
Best regards,
Krzysztof
On Tue, 30 Jan 2024 09:36:42 +0000, André Draszik wrote:
> Wrong pclk clocks have been used in this usi8 instance here. For USI
> and I2C, we need the ipclk and pclk, where pclk is the bus clock.
> Without it, nothing can work.
>
> It is unclear what exactly is using USI8_USI_CLK, but it is not
> required for the IP to be operational at this stage, while pclk is.
> This also brings the DT in line with the clock names expected by the
> usi and i2c drivers.
>
> [...]
Applied, thanks!
[3/6] arm64: dts: exynos: gs101: use correct clocks for usi8
https://git.kernel.org/krzk/linux/c/512b5a875cd8b8352257fefe71513097ee97be77
Best regards,
--
Krzysztof Kozlowski <[email protected]>
On Tue, 30 Jan 2024 09:36:43 +0000, André Draszik wrote:
> Wrong pclk clocks have been used in this usi_uart instance here. For
> USI and UART, we need the ipclk and pclk, where pclk is the bus clock.
> Without it, nothing can work.
>
> It is unclear what exactly is using USI0_UART_CLK, but it is not
> required for the IP to be operational at this stage, while pclk is.
> This also brings the DT in line with the clock names expected by the
> usi and uart drivers.
>
> [...]
Applied, thanks!
[4/6] arm64: dts: exynos: gs101: use correct clocks for usi_uart
https://git.kernel.org/krzk/linux/c/21e4e8807bfc7acc0fc9fe3ab69e1dc0662e7ce4
Best regards,
--
Krzysztof Kozlowski <[email protected]>
On Tue, 30 Jan 2024 09:36:40 +0000, André Draszik wrote:
> This pclk clock is required any time we access the pinctrl registers of
> this block.
>
> Since pinctrl-samsung doesn't support a clock at the moment, we just
> keep the kernel from disabling it at boot, until we have an update for
> pinctrl-samsung to handle this required clock, at which point we'll be
> able to drop the flag again.
>
> [...]
Applied, thanks!
[1/6] clk: samsung: gs101: gpio_peric0_pclk needs to be kept on
https://git.kernel.org/krzk/linux/c/8a96d2701f7c794e45102a9cc7fc4a5c4951e699
Best regards,
--
Krzysztof Kozlowski <[email protected]>
Hi Krzysztof,
On Thu, 2024-02-01 at 11:02 +0100, Krzysztof Kozlowski wrote:
> On 30/01/2024 10:36, André Draszik wrote:
> > The peric0_top1_ipclk_0 and peric0_top1_pclk_0 are the clocks going to
> > peric0/uart_usi, with pclk being the bus clock. Without pclk running,
> > any bus access will hang.
> > Unfortunately, in commit d97b6c902a40 ("arm64: dts: exynos: gs101:
> > update USI UART to use peric0 clocks") the gs101 DT ended up specifying
> > an incorrect pclk in the respective node and instead the two clocks
> > here were marked as critical.
> >
> > We have fixed the gs101 DT and can therefore drop this incorrect
> > work-around here, the uart driver will claim these clocks as needed.
>
> How did you fixed the DTS? Which commit did it? Are we going back to
> basics of driver changes depending on DTS?
Sorry if the description isn't clear.
a) these clocks are not critical for the system to work, and this patch fixes that.
b) the initial DTSI for gs101 used incorrect clocks for the serial, and it didn't
work. The work-around was to specify these clocks here as critical instead. Patch
#4 in this series has corrected the DTSI.
So there is no dependency between the DTS update and the driver update here as such,
no new properties, or otherwise.
That said, now that b) above has been fixed (in patch #4), it is OK to mark these
clocks as non-critical without any ill effects. That's all that is happening. I was
merely referencing that in the commit message.
I can rephrase things if you wish.
Cheers,
Andre'