clk: mvebu: cp110 add CLK_IGNORE_UNUSED to pcie_x10, pcie_x11 & pcie_x4

[PATCH 2/2] clk: mvebu: cp110 add CLK_IGNORE_UNUSED to pcie_x10, pcie_x11 & pcie_x4

Posted by Josua Mayer 3 months, 1 week ago

CP110 based platforms rely on the bootloader for pci port
initialization.
TF-A actively prevents non-uboot re-configuration of pci lanes, and many
boards do not have software control over the pci card reset.

If a pci port had link at boot-time and the clock is stopped at a later
point, the link fails and can not be recovered.

PCI controller driver probe - and by extension ownership of a driver for
the pci clocks - may be delayed especially on large modular kernels,
causing the clock core to start disabling unused clocks.

Add the CLK_IGNORE_UNUSED flag to the three pci port's clocks to ensure
they are not stopped before the pci controller driver has taken
ownership and tested for an existing link.

This fixes failed pci link detection when controller driver probes late,
e.g. with arm64 defconfig and CONFIG_PHY_MVEBU_CP110_COMPHY=m.

Closes: https://lore.kernel.org/r/b71596c7-461b-44b6-89ab-3cfbd492639f@solid-run.com
Signed-off-by: Josua Mayer <josua@solid-run.com>
---
 drivers/clk/mvebu/cp110-system-controller.c | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/drivers/clk/mvebu/cp110-system-controller.c b/drivers/clk/mvebu/cp110-system-controller.c
index 03c59bf221060..b47c869060466 100644
--- a/drivers/clk/mvebu/cp110-system-controller.c
+++ b/drivers/clk/mvebu/cp110-system-controller.c
@@ -110,6 +110,25 @@ static const char * const gate_base_names[] = {
 	[CP110_GATE_EIP197]	= "eip197"
 };
 
+static unsigned long gate_flags(const u8 bit_idx)
+{
+	switch (bit_idx) {
+	case CP110_GATE_PCIE_X1_0:
+	case CP110_GATE_PCIE_X1_1:
+	case CP110_GATE_PCIE_X4:
+		/*
+		 * If a port had an active link at boot time, stopping
+		 * the clock creates a failed state from which controller
+		 * driver can not recover.
+		 * Prevent stopping this clock till after a driver has taken
+		 * ownership.
+		 */
+		return CLK_IGNORE_UNUSED;
+	default:
+		return 0;
+	}
+};
+
 struct cp110_gate_clk {
 	struct clk_hw hw;
 	struct regmap *regmap;
@@ -171,6 +190,7 @@ static struct clk_hw *cp110_register_gate(const char *name,
 	init.ops = &cp110_gate_ops;
 	init.parent_names = &parent_name;
 	init.num_parents = 1;
+	init.flags = gate_flags(bit_idx);
 
 	gate->regmap = regmap;
 	gate->bit_idx = bit_idx;

-- 
2.51.0

Re: [PATCH 2/2] clk: mvebu: cp110 add CLK_IGNORE_UNUSED to pcie_x10, pcie_x11 & pcie_x4

Posted by Andrew Lunn 3 months, 1 week ago

On Thu, Oct 30, 2025 at 04:16:26PM +0100, Josua Mayer wrote:
> CP110 based platforms rely on the bootloader for pci port
> initialization.
> TF-A actively prevents non-uboot re-configuration of pci lanes, and many
> boards do not have software control over the pci card reset.
> 
> If a pci port had link at boot-time and the clock is stopped at a later
> point, the link fails and can not be recovered.
> 
> PCI controller driver probe - and by extension ownership of a driver for
> the pci clocks - may be delayed especially on large modular kernels,
> causing the clock core to start disabling unused clocks.
> 
> Add the CLK_IGNORE_UNUSED flag to the three pci port's clocks to ensure
> they are not stopped before the pci controller driver has taken
> ownership and tested for an existing link.
> 
> This fixes failed pci link detection when controller driver probes late,
> e.g. with arm64 defconfig and CONFIG_PHY_MVEBU_CP110_COMPHY=m.

Seems like a reasonable compromise, given that TF-A could be classed
as broken. This must also prevent suspend/resume powering off PCI
devices, and then reconnecting them on resume.

Reviewed-by: Andrew Lunn <andrew@lunn.ch>

    Andrew

Re: [PATCH 2/2] clk: mvebu: cp110 add CLK_IGNORE_UNUSED to pcie_x10, pcie_x11 & pcie_x4

Posted by Josua Mayer 3 months, 1 week ago

I missed a colon in the subject line "cp110:",
should I roll v2 for this?

Am 30.10.25 um 16:33 schrieb Andrew Lunn:
> On Thu, Oct 30, 2025 at 04:16:26PM +0100, Josua Mayer wrote:
>> CP110 based platforms rely on the bootloader for pci port
>> initialization.
>> TF-A actively prevents non-uboot re-configuration of pci lanes, and many
>> boards do not have software control over the pci card reset.
>>
>> If a pci port had link at boot-time and the clock is stopped at a later
>> point, the link fails and can not be recovered.
>>
>> PCI controller driver probe - and by extension ownership of a driver for
>> the pci clocks - may be delayed especially on large modular kernels,
>> causing the clock core to start disabling unused clocks.
>>
>> Add the CLK_IGNORE_UNUSED flag to the three pci port's clocks to ensure
>> they are not stopped before the pci controller driver has taken
>> ownership and tested for an existing link.
>>
>> This fixes failed pci link detection when controller driver probes late,
>> e.g. with arm64 defconfig and CONFIG_PHY_MVEBU_CP110_COMPHY=m.
> Seems like a reasonable compromise, given that TF-A could be classed
> as broken. This must also prevent suspend/resume powering off PCI
> devices, and then reconnecting them on resume.
Currently pcie-armada8k (unlike e.g. pci-imx6) does not currently define
any dev_pm_ops - so we should be safe from any power-management.
>
> Reviewed-by: Andrew Lunn <andrew@lunn.ch>
>
>     Andrew

Re: [PATCH 2/2] clk: mvebu: cp110 add CLK_IGNORE_UNUSED to pcie_x10, pcie_x11 & pcie_x4

Posted by Gregory CLEMENT 3 months ago

Josua Mayer <josua@solid-run.com> writes:

> I missed a colon in the subject line "cp110:",
> should I roll v2 for this?

it is up to the clock maintainer.

>
> Am 30.10.25 um 16:33 schrieb Andrew Lunn:
>> On Thu, Oct 30, 2025 at 04:16:26PM +0100, Josua Mayer wrote:
>>> CP110 based platforms rely on the bootloader for pci port
>>> initialization.
>>> TF-A actively prevents non-uboot re-configuration of pci lanes, and many
>>> boards do not have software control over the pci card reset.
>>>
>>> If a pci port had link at boot-time and the clock is stopped at a later
>>> point, the link fails and can not be recovered.
>>>
>>> PCI controller driver probe - and by extension ownership of a driver for
>>> the pci clocks - may be delayed especially on large modular kernels,
>>> causing the clock core to start disabling unused clocks.
>>>
>>> Add the CLK_IGNORE_UNUSED flag to the three pci port's clocks to ensure
>>> they are not stopped before the pci controller driver has taken
>>> ownership and tested for an existing link.
>>>
>>> This fixes failed pci link detection when controller driver probes late,
>>> e.g. with arm64 defconfig and CONFIG_PHY_MVEBU_CP110_COMPHY=m.
>> Seems like a reasonable compromise, given that TF-A could be classed
>> as broken. This must also prevent suspend/resume powering off PCI
>> devices, and then reconnecting them on resume.
> Currently pcie-armada8k (unlike e.g. pci-imx6) does not currently define
> any dev_pm_ops - so we should be safe from any power-management.
>>
>> Reviewed-by: Andrew Lunn <andrew@lunn.ch>


Acked-by: Gregory CLEMENT <gregory.clement@bootlin.com>

Thanks,

Gregory


>>
>>     Andrew

-- 
Grégory CLEMENT, Bootlin
Embedded Linux and Kernel engineering
https://bootlin.com