Skip to contentSkip to conversation

phy: spacemit: fix and enable the K1 USB 2.0 PHY

PATCH Applied Bruno Banelli 5 participants 7 messages Aug 25 – Nov 1

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v2 cover phy: spacemit: fix and enable the K1 USB 2.0 PHY

v1 was a single patch adding "default ARCH_SPACEMIT" to PHY_SPACEMIT_K1_USB2. Automated review pointed out two pre-existing Kconfig problems in the same entry, and that change makes them easier to hit: enabling the symbol by default means configurations that would previously have left it off will now build the driver.

So fix the dependencies first, then enable it by default.

Both were verified against the driver source: phy-k1-usb2.c calls devm_platform_ioremap_resource() and devm_regmap_init_mmio(), while the Kconfig entry has neither "depends on HAS_IOMEM" nor
"select REGMAP_MMIO". PHY_SPACEMIT_K1_PCIE, directly above it in the same file, already has the HAS_IOMEM dependency.

Tested on a Milk-V Jupiter (SpacemiT M1) running mainline 7.2: without patch 2/2 a "make defconfig" kernel has no USB at all, because dwc3-generic-plat cannot get its PHY.

v1: https://lore.kernel.org/linux-riscv/[email protected]/

Changes in v2:
- new patch 1/2 adding "depends on HAS_IOMEM" and "select REGMAP_MMIO", with a Fixes: tag
- no change to the patch now numbered 2/2

Bruno Banelli (2):
phy: spacemit: fix K1 USB 2.0 PHY Kconfig dependencies
phy: spacemit: enable K1 USB 2.0 PHY by default on ARCH_SPACEMIT

 drivers/phy/spacemit/Kconfig | 3 +++
 1 file changed, 3 insertions(+)

base-commit: 66498c75b4f8017f62d720d9b59675bdf3abce91

Signature · 1 line
2.43.0
#1 lore raw

PATCH v2 1/2 phy: spacemit: fix K1 USB 2.0 PHY Kconfig dependencies

phy-k1-usb2.c maps its registers with devm_platform_ioremap_resource() and wraps them with devm_regmap_init_mmio(), but PHY_SPACEMIT_K1_USB2 neither depends on HAS_IOMEM nor selects REGMAP_MMIO.

REGMAP_MMIO has no prompt and can only be enabled by select, so a configuration that leaves it off fails to link. In practice the symbol is usually pulled in indirectly, which is why this has gone unnoticed.

PHY_SPACEMIT_K1_PCIE, directly above it in the same file, already has the HAS_IOMEM dependency; without it PHY_SPACEMIT_K1_USB2 can be built under COMPILE_TEST on architectures that provide no I/O memory.

Fixes fe4bc1a08638 ("phy: spacemit: support K1 USB2.0 PHY controller")
---
 drivers/phy/spacemit/Kconfig | 2 ++
 1 file changed, 2 insertions(+)
drivers/phy/spacemit/Kconfig +2 -0
index 5fdf18fce..cd7ae48d8 100644
@@ -18,8 +18,10 @@ config PHY_SPACEMIT_K1_USB2
 	tristate "SpacemiT K1 USB 2.0 PHY support"
 	depends on (ARCH_SPACEMIT || COMPILE_TEST) && OF
 	depends on COMMON_CLK
+	depends on HAS_IOMEM
 	depends on USB_COMMON
 	select GENERIC_PHY
+	select REGMAP_MMIO
 	help
 	  Enable this to support K1 USB 2.0 PHY driver. This driver takes care of
 	  enabling and clock setup and will be used by K1 udc/ehci/otg/xhci driver.
Signature · 1 line
2.43.0
#2 lore raw

PATCH v2 2/2 phy: spacemit: enable K1 USB 2.0 PHY by default on ARCH_SPACEMIT

In reply to Bruno Banelli

PHY_SPACEMIT_K1_PCIE has "default ARCH_SPACEMIT", but PHY_SPACEMIT_K1_USB2, directly below it in the same file, does not. Since the symbol is also absent from arch/riscv/configs/defconfig, a plain "make defconfig" build for a SpacemiT K1 board ends up with the USB 2.0 PHY driver disabled.

dwc3-generic-plat needs that PHY, so the result is a kernel with no working USB at all on boards such as the Milk-V Jupiter, where all four USB 3.0 type-A ports hang off the single dwc3 controller.

Give it the same default as its sibling.

---
 drivers/phy/spacemit/Kconfig | 1 +
 1 file changed, 1 insertion(+)
drivers/phy/spacemit/Kconfig +1 -0
index cd7ae48d8..b5f36621b 100644
@@ -22,6 +22,7 @@ config PHY_SPACEMIT_K1_USB2
 	depends on USB_COMMON
 	select GENERIC_PHY
 	select REGMAP_MMIO
+	default ARCH_SPACEMIT
 	help
 	  Enable this to support K1 USB 2.0 PHY driver. This driver takes care of
 	  enabling and clock setup and will be used by K1 udc/ehci/otg/xhci driver.
Signature · 1 line
2.43.0
#3 lore raw

Re: [PATCH v2 2/2] phy: spacemit: enable K1 USB 2.0 PHY by default on ARCH_SPACEMIT

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

Thank you for your contribution! Sashiko AI review found 3 potential issue(s) to consider:

Pre-existing issues:
- [High] Missing select MFD_SYSCON or depends on MFD_SYSCON for PHY_SPACEMIT_K1_PCIE
- [Medium] Incomplete fix: PHY_SPACEMIT_K3_COMBO_PHY is missing default ARCH_SPACEMIT
- [Medium] PHY_SPACEMIT_K3_COMBO_PHY is missing depends on HAS_IOMEM
--

--- Patch [2]: [PATCH v2 2/2] phy: spacemit: enable K1 USB 2.0 PHY by default on ARCH_SPACEMIT --- commit 24c737090e69ed7db66ac129f68b1a45eb92ca1e
Author: Bruno Banelli <[email protected]>

phy: spacemit: enable K1 USB 2.0 PHY by default on ARCH_SPACEMIT

This commit adds a default ARCH_SPACEMIT to the PHY_SPACEMIT_K1_USB2 Kconfig entry. This resolves an issue where the USB 2.0 PHY driver was disabled by default on SpacemiT K1 boards, preventing USB from working.

drivers/phy/spacemit/Kconfig config PHY_SPACEMIT_K1_USB2 · line 25

Quoted 9 lines
drivers/phy/spacemit/Kconfig +1 -0
index cd7ae48d8ea2f..b5f36621b2075 100644
@@ -22,6 +22,7 @@ config PHY_SPACEMIT_K1_USB2
 	depends on USB_COMMON
 	select GENERIC_PHY
 	select REGMAP_MMIO
+	default ARCH_SPACEMIT

[Severity: High]
This is a pre-existing issue, but I noticed PHY_SPACEMIT_K1_PCIE is missing a dependency on MFD_SYSCON:

drivers/phy/spacemit/Kconfig:PHY_SPACEMIT_K1_PCIE {
    config PHY_SPACEMIT_K1_PCIE
        tristate "PCIe and combo PHY driver for the SpacemiT K1 SoC"
        depends on ARCH_SPACEMIT || COMPILE_TEST
        ...
}

If MFD_SYSCON is not enabled, the kernel will still build successfully, but at runtime the driver probe will silently fail returning -ENOTSUPP:

drivers/phy/spacemit/phy-k1-pcie.c:k1_pcie_combo_phy_probe() {
    ...
    regmap = syscon_regmap_lookup_by_phandle(dev_of_node(dev), SYSCON_APMU);
    ...
}

Does this config need to select or depend on MFD_SYSCON to prevent this?

[Severity: Medium]
This is a pre-existing issue, but PHY_SPACEMIT_K3_COMBO_PHY seems to be missing the same default configuration being added here:

drivers/phy/spacemit/Kconfig:PHY_SPACEMIT_K3_COMBO_PHY {
    config PHY_SPACEMIT_K3_COMBO_PHY
        tristate "SpacemiT K3 USB3/PCIe PHY support"
        depends on (ARCH_SPACEMIT || COMPILE_TEST) && OF
        ...
}

Since the K3 combo PHY driver manages USB3/PCIe on SpacemiT SoCs, could this missing default lead to similar broken USB behavior on K3 boards? Would it make sense to add default ARCH_SPACEMIT there as well?

[Severity: Medium]
This is a pre-existing issue, but PHY_SPACEMIT_K3_COMBO_PHY also appears to be missing a dependency on HAS_IOMEM:

drivers/phy/spacemit/Kconfig:PHY_SPACEMIT_K3_COMBO_PHY {
config PHY_SPACEMIT_K3_COMBO_PHY
tristate "SpacemiT K3 USB3/PCIe PHY support" depends on (ARCH_SPACEMIT || COMPILE_TEST) && OF depends on COMMON_CLK
select PHY_SPACEMIT_K3_COMMON_OPS
}

The PHY_SPACEMIT_K3_COMBO_PHY driver uses devm_platform_ioremap_resource(), which operates on MMIO memory. Could omitting the HAS_IOMEM dependency lead to build or probe failures under COMPILE_TEST?

Quoted 3 lines
drivers/phy/spacemit/Kconfig +0 -0
 	help
 	  Enable this to support K1 USB 2.0 PHY driver. This driver takes care of
 	  enabling and clock setup and will be used by K1 udc/ehci/otg/xhci driver.
Signature · 6 lines
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=2

-- 
linux-phy mailing list
[email protected]
https://lists.infradead.org/mailman/listinfo/linux-phy
maintainer Reviewed-by on the v2 series
#4 lore raw
In reply to Bruno Banelli

Hi Bruno,

Quoted 18 lines · Bruno Banelli

v1 was a single patch adding "default ARCH_SPACEMIT" to PHY_SPACEMIT_K1_USB2. Automated review pointed out two pre-existing Kconfig problems in the same entry, and that change makes them easier to hit: enabling the symbol by default means configurations that would previously have left it off will now build the driver.

So fix the dependencies first, then enable it by default.

Both were verified against the driver source: phy-k1-usb2.c calls devm_platform_ioremap_resource() and devm_regmap_init_mmio(), while the Kconfig entry has neither "depends on HAS_IOMEM" nor
"select REGMAP_MMIO". PHY_SPACEMIT_K1_PCIE, directly above it in the same file, already has the HAS_IOMEM dependency.

Tested on a Milk-V Jupiter (SpacemiT M1) running mainline 7.2: without patch 2/2 a "make defconfig" kernel has no USB at all, because dwc3-generic-plat cannot get its PHY.

Quoted 7 lines

v1: https://lore.kernel.org/linux-riscv/[email protected]/

Changes in v2:
- new patch 1/2 adding "depends on HAS_IOMEM" and "select REGMAP_MMIO", with a Fixes: tag
- no change to the patch now numbered 2/2

Thanks for this v2

Signature · 1 line
Yixun Lan (dlan)
maintainer Applied
#5 lore raw
In reply to Bruno Banelli
Quoted 9 lines · Bruno Banelli

v1 was a single patch adding "default ARCH_SPACEMIT" to PHY_SPACEMIT_K1_USB2. Automated review pointed out two pre-existing Kconfig problems in the same entry, and that change makes them easier to hit: enabling the symbol by default means configurations that would previously have left it off will now build the driver.

So fix the dependencies first, then enable it by default.

[...]

Applied, thanks!

[1/2] phy: spacemit: fix K1 USB 2.0 PHY Kconfig dependencies

      commit: 882af55e42a4cd5bd61db0244c0b78941884518a
[2/2] phy: spacemit: enable K1 USB 2.0 PHY by default on ARCH_SPACEMIT
      commit: 392101240b22256105da3b5af6348d1889c63112

Best regards,

Signature · 1 line
~Vinod
#6 lore raw
In reply to Bruno Banelli
This series was applied to riscv/linux.git (fixes) by Vinod Koul <[email protected]>:

Hello:

This series was applied to riscv/linux.git (fixes) by Vinod Koul <[email protected]>:

Quoted 9 lines · On Tue, 25 Aug 2026 08:50:05 +0200 you wrote:

v1 was a single patch adding "default ARCH_SPACEMIT" to PHY_SPACEMIT_K1_USB2. Automated review pointed out two pre-existing Kconfig problems in the same entry, and that change makes them easier to hit: enabling the symbol by default means configurations that would previously have left it off will now build the driver.

So fix the dependencies first, then enable it by default.

[...]

Here is the summary with links:
- [v2,1/2] phy: spacemit: fix K1 USB 2.0 PHY Kconfig dependencies

    https://git.kernel.org/riscv/c/882af55e42a4
  - [v2,2/2] phy: spacemit: enable K1 USB 2.0 PHY by default on ARCH_SPACEMIT
    https://git.kernel.org/riscv/c/392101240b22

You are awesome, thank you!

Signature · 2 lines
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html