| Message ID | 20220117150039.44058-1-jonathanh@nvidia.com |
|---|---|
| State | Handled Elsewhere |
| Headers | show |
| Series | [V2] usb: common: ulpi: Fix crash in ulpi_match() | expand |
On Mon, Jan 17, 2022 at 03:00:39PM +0000, Jon Hunter wrote: > Commit 7495af930835 ("ARM: multi_v7_defconfig: Enable drivers for > DragonBoard 410c") enables the CONFIG_PHY_QCOM_USB_HS for the ARM > multi_v7_defconfig. Enabling this Kconfig is causing the kernel to crash > on the Tegra20 Ventana platform in the ulpi_match() function. > > The Qualcomm USB HS PHY driver that is enabled by CONFIG_PHY_QCOM_USB_HS, > registers a ulpi_driver but this driver does not provide an 'id_table', > so when ulpi_match() is called on the Tegra20 Ventana platform, it > crashes when attempting to deference the id_table pointer which is not > valid. The Qualcomm USB HS PHY driver uses device-tree for matching the > ULPI driver with the device and so fix this crash by using device-tree > for matching if the id_table is not valid. > > Fixes: ef6a7bcfb01c ("usb: ulpi: Support device discovery via DT") > Signed-off-by: Jon Hunter <jonathanh@nvidia.com> No cc: of stable since this is a bug going back to 4.11? thanks, greg k-h
On 17/01/2022 15:40, Greg KH wrote: > On Mon, Jan 17, 2022 at 03:00:39PM +0000, Jon Hunter wrote: >> Commit 7495af930835 ("ARM: multi_v7_defconfig: Enable drivers for >> DragonBoard 410c") enables the CONFIG_PHY_QCOM_USB_HS for the ARM >> multi_v7_defconfig. Enabling this Kconfig is causing the kernel to crash >> on the Tegra20 Ventana platform in the ulpi_match() function. >> >> The Qualcomm USB HS PHY driver that is enabled by CONFIG_PHY_QCOM_USB_HS, >> registers a ulpi_driver but this driver does not provide an 'id_table', >> so when ulpi_match() is called on the Tegra20 Ventana platform, it >> crashes when attempting to deference the id_table pointer which is not >> valid. The Qualcomm USB HS PHY driver uses device-tree for matching the >> ULPI driver with the device and so fix this crash by using device-tree >> for matching if the id_table is not valid. >> >> Fixes: ef6a7bcfb01c ("usb: ulpi: Support device discovery via DT") >> Signed-off-by: Jon Hunter <jonathanh@nvidia.com> > > No cc: of stable since this is a bug going back to 4.11? Yes good point. Heikki, let me know if you want me to resend or if you can add the stable tag? Cheers Jon
On Tue, Jan 18, 2022 at 09:02:36AM +0000, Jon Hunter wrote: > > On 17/01/2022 15:40, Greg KH wrote: > > On Mon, Jan 17, 2022 at 03:00:39PM +0000, Jon Hunter wrote: > > > Commit 7495af930835 ("ARM: multi_v7_defconfig: Enable drivers for > > > DragonBoard 410c") enables the CONFIG_PHY_QCOM_USB_HS for the ARM > > > multi_v7_defconfig. Enabling this Kconfig is causing the kernel to crash > > > on the Tegra20 Ventana platform in the ulpi_match() function. > > > > > > The Qualcomm USB HS PHY driver that is enabled by CONFIG_PHY_QCOM_USB_HS, > > > registers a ulpi_driver but this driver does not provide an 'id_table', > > > so when ulpi_match() is called on the Tegra20 Ventana platform, it > > > crashes when attempting to deference the id_table pointer which is not > > > valid. The Qualcomm USB HS PHY driver uses device-tree for matching the > > > ULPI driver with the device and so fix this crash by using device-tree > > > for matching if the id_table is not valid. > > > > > > Fixes: ef6a7bcfb01c ("usb: ulpi: Support device discovery via DT") > > > Signed-off-by: Jon Hunter <jonathanh@nvidia.com> > > > > No cc: of stable since this is a bug going back to 4.11? > > > Yes good point. > > Heikki, let me know if you want me to resend or if you can add the stable > tag? I can add it myself. I'll do so after 5.17-rc1 is out, when I can apply things to my tree. thanks, greg k-h
diff --git a/drivers/usb/common/ulpi.c b/drivers/usb/common/ulpi.c index 4169cf40a03b..8f8405b0d608 100644 --- a/drivers/usb/common/ulpi.c +++ b/drivers/usb/common/ulpi.c @@ -39,8 +39,11 @@ static int ulpi_match(struct device *dev, struct device_driver *driver) struct ulpi *ulpi = to_ulpi_dev(dev); const struct ulpi_device_id *id; - /* Some ULPI devices don't have a vendor id so rely on OF match */ - if (ulpi->id.vendor == 0) + /* + * Some ULPI devices don't have a vendor id + * or provide an id_table so rely on OF match. + */ + if (ulpi->id.vendor == 0 || !drv->id_table) return of_driver_match_device(dev, driver); for (id = drv->id_table; id->vendor; id++)
Commit 7495af930835 ("ARM: multi_v7_defconfig: Enable drivers for DragonBoard 410c") enables the CONFIG_PHY_QCOM_USB_HS for the ARM multi_v7_defconfig. Enabling this Kconfig is causing the kernel to crash on the Tegra20 Ventana platform in the ulpi_match() function. The Qualcomm USB HS PHY driver that is enabled by CONFIG_PHY_QCOM_USB_HS, registers a ulpi_driver but this driver does not provide an 'id_table', so when ulpi_match() is called on the Tegra20 Ventana platform, it crashes when attempting to deference the id_table pointer which is not valid. The Qualcomm USB HS PHY driver uses device-tree for matching the ULPI driver with the device and so fix this crash by using device-tree for matching if the id_table is not valid. Fixes: ef6a7bcfb01c ("usb: ulpi: Support device discovery via DT") Signed-off-by: Jon Hunter <jonathanh@nvidia.com> --- Changes since V1: - Added fixes tag drivers/usb/common/ulpi.c | 7 +++++-- 1 file changed, 5 insertions(+), 2 deletions(-)