[PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[Stefan Wahren]

The commit 4adc20ba95d4 ("ARM: dts: broadcom: rpi: Switch to V3D firmware
clock") causes a regression in arm64 developer setups, which stores the
kernel modules via NFS. Before this change the involved V3D clock provider
was builtin, but after this DT change the clk-raspberrypi is responsible
for V3D and for arm64/defconfig this driver is build as a kernel module.
In case these kernel modules are provided via NFS this takes too long and
the PM domain core give up before the clock driver could be loaded:

  v3d fec00000.gpu: deferred probe timeout, ignoring dependency

So resolve this issue by making this critical driver builtin.

Reported-by: Mark Brown <broonie@kernel.org>
Closes: https://lore.kernel.org/linux-arm-kernel/9ebda74e-e700-4fbe-bca5-382f92417a9c@sirena.org.uk/
Fixes: 4adc20ba95d4 ("ARM: dts: broadcom: rpi: Switch to V3D firmware clock")
Signed-off-by: Stefan Wahren <wahrenst@gmx.net>
---
 arch/arm64/configs/defconfig | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
index e3a2d37bd104..1a48faad2473 100644
--- a/arch/arm64/configs/defconfig
+++ b/arch/arm64/configs/defconfig
@@ -1341,7 +1341,7 @@ CONFIG_COMMON_CLK_RS9_PCIE=y
 CONFIG_COMMON_CLK_VC3=y
 CONFIG_COMMON_CLK_VC5=y
 CONFIG_COMMON_CLK_BD718XX=m
-CONFIG_CLK_RASPBERRYPI=m
+CONFIG_CLK_RASPBERRYPI=y
 CONFIG_CLK_IMX8MM=y
 CONFIG_CLK_IMX8MN=y
 CONFIG_CLK_IMX8MP=y
-- 
2.43.0

Re: [PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[Florian Fainelli]

On 11/4/25 09:45, Stefan Wahren wrote:
> The commit 4adc20ba95d4 ("ARM: dts: broadcom: rpi: Switch to V3D firmware
> clock") causes a regression in arm64 developer setups, which stores the
> kernel modules via NFS. Before this change the involved V3D clock provider
> was builtin, but after this DT change the clk-raspberrypi is responsible
> for V3D and for arm64/defconfig this driver is build as a kernel module.
> In case these kernel modules are provided via NFS this takes too long and
> the PM domain core give up before the clock driver could be loaded:
> 
>    v3d fec00000.gpu: deferred probe timeout, ignoring dependency
> 
> So resolve this issue by making this critical driver builtin.
> 
> Reported-by: Mark Brown <broonie@kernel.org>
> Closes: https://lore.kernel.org/linux-arm-kernel/9ebda74e-e700-4fbe-bca5-382f92417a9c@sirena.org.uk/
> Fixes: 4adc20ba95d4 ("ARM: dts: broadcom: rpi: Switch to V3D firmware clock")
> Signed-off-by: Stefan Wahren <wahrenst@gmx.net>

Applied, thanks Stefan!
-- 
Florian

Re: [PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[Stefan Wahren]

Hi Mark,

Am 04.11.25 um 18:45 schrieb Stefan Wahren:
> The commit 4adc20ba95d4 ("ARM: dts: broadcom: rpi: Switch to V3D firmware
> clock") causes a regression in arm64 developer setups, which stores the
> kernel modules via NFS. Before this change the involved V3D clock provider
> was builtin, but after this DT change the clk-raspberrypi is responsible
> for V3D and for arm64/defconfig this driver is build as a kernel module.
> In case these kernel modules are provided via NFS this takes too long and
> the PM domain core give up before the clock driver could be loaded:
>
>    v3d fec00000.gpu: deferred probe timeout, ignoring dependency
>
> So resolve this issue by making this critical driver builtin.
>
> Reported-by: Mark Brown <broonie@kernel.org>
> Closes: https://lore.kernel.org/linux-arm-kernel/9ebda74e-e700-4fbe-bca5-382f92417a9c@sirena.org.uk/
> Fixes: 4adc20ba95d4 ("ARM: dts: broadcom: rpi: Switch to V3D firmware clock")
> Signed-off-by: Stefan Wahren <wahrenst@gmx.net>
can you confirm, that this fixes the problem in your case?

Thanks
Stefan
> ---
>   arch/arm64/configs/defconfig | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/arch/arm64/configs/defconfig b/arch/arm64/configs/defconfig
> index e3a2d37bd104..1a48faad2473 100644
> --- a/arch/arm64/configs/defconfig
> +++ b/arch/arm64/configs/defconfig
> @@ -1341,7 +1341,7 @@ CONFIG_COMMON_CLK_RS9_PCIE=y
>   CONFIG_COMMON_CLK_VC3=y
>   CONFIG_COMMON_CLK_VC5=y
>   CONFIG_COMMON_CLK_BD718XX=m
> -CONFIG_CLK_RASPBERRYPI=m
> +CONFIG_CLK_RASPBERRYPI=y
>   CONFIG_CLK_IMX8MM=y
>   CONFIG_CLK_IMX8MN=y
>   CONFIG_CLK_IMX8MP=y

Re: [PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 622 bytes --]

On Wed, Nov 12, 2025 at 06:20:38PM +0100, Stefan Wahren wrote:
> Am 04.11.25 um 18:45 schrieb Stefan Wahren:

> > In case these kernel modules are provided via NFS this takes too long and
> > the PM domain core give up before the clock driver could be loaded:
> > 
> >    v3d fec00000.gpu: deferred probe timeout, ignoring dependency
> > 
> > So resolve this issue by making this critical driver builtin.

> can you confirm, that this fixes the problem in your case?

Yes, it seems to work.  Given that this patch was applied I had been
expecting it to show up in -next but that doesn't seem to be the case?

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[Florian Fainelli]

On 11/16/2025 5:52 AM, Mark Brown wrote:
> On Wed, Nov 12, 2025 at 06:20:38PM +0100, Stefan Wahren wrote:
>> Am 04.11.25 um 18:45 schrieb Stefan Wahren:
> 
>>> In case these kernel modules are provided via NFS this takes too long and
>>> the PM domain core give up before the clock driver could be loaded:
>>>
>>>     v3d fec00000.gpu: deferred probe timeout, ignoring dependency
>>>
>>> So resolve this issue by making this critical driver builtin.
> 
>> can you confirm, that this fixes the problem in your case?
> 
> Yes, it seems to work.  Given that this patch was applied I had been
> expecting it to show up in -next but that doesn't seem to be the case?

It was applied on Nov 14th:

https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc.git/commit/?id=020e792ca39b
-- 
Florian

Re: [PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[Mark Brown]

[-- Attachment #1: Type: text/plain, Size: 616 bytes --]

On Sun, Nov 16, 2025 at 09:14:39AM -0800, Florian Fainelli wrote:
> On 11/16/2025 5:52 AM, Mark Brown wrote:

> > Yes, it seems to work.  Given that this patch was applied I had been
> > expecting it to show up in -next but that doesn't seem to be the case?

> It was applied on Nov 14th:

> https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc.git/commit/?id=020e792ca39b

Ah, that's the pull request you sent with it in getting pulled in to the
main SoC tree.  I was expecting your tree to be in -next directly as
well (a lot of individual SoC trees are) so when you applied it on the
4th it'd start appearing.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 488 bytes --]

Re: [PATCH] arm64: defconfig: Fix V3D deferred probe timeout

[Krzysztof Kozlowski]

On 16/11/2025 18:56, Mark Brown wrote:
> On Sun, Nov 16, 2025 at 09:14:39AM -0800, Florian Fainelli wrote:
>> On 11/16/2025 5:52 AM, Mark Brown wrote:
> 
>>> Yes, it seems to work.  Given that this patch was applied I had been
>>> expecting it to show up in -next but that doesn't seem to be the case?
> 
>> It was applied on Nov 14th:
> 
>> https://git.kernel.org/pub/scm/linux/kernel/git/soc/soc.git/commit/?id=020e792ca39b
> 
> Ah, that's the pull request you sent with it in getting pulled in to the
> main SoC tree.  I was expecting your tree to be in -next directly as
> well (a lot of individual SoC trees are) so when you applied it on the

And every individual maintainer's SoC tree should be there... There is
an exception for media because their upstream does not pull from
sub-maintainers but rebases, which would lead to duplicated commits, but
that not the case for SoC tree.

> 4th it'd start appearing.


Best regards,
Krzysztof