arm64: dts: qcom: glymur: Mark USB controllers dma-coherent
PATCH 1/1 arm64: dts: qcom: glymur: Mark USB controllers dma-coherent
All five USB controller nodes on Glymur (usb_0, usb_1, usb_2, usb_hs, usb_mp) are missing the dma-coherent property, which Hamoa carries on all of its USB controllers. Krishna Kurapati confirmed the Glymur USB controllers are dma-coherent [1].
Without the property the kernel treats USB DMA buffers as non-coherent and performs cache maintenance on every transfer, which coherent hardware makes redundant.
Follow-up to the clock/power/bandwidth votes patch [2], kept separate so the already-reviewed patch stands as is.
Tested on an Asus Zenbook A16 (UX3607OA): all controllers enumerate their devices as before and a 64 MiB write plus two O_DIRECT read-backs over USB mass storage produce identical checksums.
[1] https://lore.kernel.org/linux-arm-msm/[email protected]/
[2] https://lore.kernel.org/linux-arm-msm/[email protected]/
--- arch/arm64/boot/dts/qcom/glymur.dtsi | 10 ++++++++++ 1 file changed, 10 insertions(+)
arch/arm64/boot/dts/qcom/glymur.dtsi +10 -0
@@ -4605,6 +4605,8 @@ usb-role-switch; wakeup-source; + dma-coherent; + status = "disabled"; ports {
@@ -4681,6 +4683,8 @@ usb-role-switch; wakeup-source; + dma-coherent; + status = "disabled"; ports {
@@ -4757,6 +4761,8 @@ usb-role-switch; wakeup-source; + dma-coherent; + status = "disabled"; ports {
@@ -4847,6 +4853,8 @@ usb-role-switch; wakeup-source; + dma-coherent; + status = "disabled"; };
@@ -4919,6 +4927,8 @@ dr_mode = "host"; wakeup-source; + dma-coherent; + status = "disabled"; };
Signature · 1 line
2.50.1
Quoted 19 lines · Greg Ociepka
All five USB controller nodes on Glymur (
usb_0,usb_1,usb_2,usb_hs,usb_mp) are missing the dma-coherent property, which Hamoa carries on all of its USB controllers. Krishna Kurapati confirmed the Glymur USB controllers are dma-coherent [1].Without the property the kernel treats USB DMA buffers as non-coherent and performs cache maintenance on every transfer, which coherent hardware makes redundant.
Follow-up to the clock/power/bandwidth votes patch [2], kept separate so the already-reviewed patch stands as is.
Tested on an Asus Zenbook A16 (UX3607OA): all controllers enumerate their devices as before and a 64 MiB write plus two
O_DIRECTread-backs over USB mass storage produce identical checksums.[1] https://lore.kernel.org/linux-arm-msm/[email protected]/
[2] https://lore.kernel.org/linux-arm-msm/[email protected]/
Quoted 7 lines
Fixes 4eee57dd4df9 ("arm64: dts: qcom: glymur: Add USB related nodes")Signed-off-by Greg Ociepka <[email protected]>Assisted-by Claude:fable-5--- arch/arm64/boot/dts/qcom/glymur.dtsi | 10 ++++++++++ 1 file changed, 10 insertions(+)
Signature · 2 lines
With best wishes Dmitry
Quoted 16 lines · Greg Ociepka
All five USB controller nodes on Glymur (
usb_0,usb_1,usb_2,usb_hs,usb_mp) are missing the dma-coherent property, which Hamoa carries on all of its USB controllers. Krishna Kurapati confirmed the Glymur USB controllers are dma-coherent [1].Without the property the kernel treats USB DMA buffers as non-coherent and performs cache maintenance on every transfer, which coherent hardware makes redundant.
Follow-up to the clock/power/bandwidth votes patch [2], kept separate so the already-reviewed patch stands as is.
Tested on an Asus Zenbook A16 (UX3607OA): all controllers enumerate their devices as before and a 64 MiB write plus two
O_DIRECTread-backs over USB mass storage produce identical checksums.
Quoted 6 lines
[1] https://lore.kernel.org/linux-arm-msm/[email protected]/
[2] https://lore.kernel.org/linux-arm-msm/[email protected]/Fixes 4eee57dd4df9 ("arm64: dts: qcom: glymur: Add USB related nodes")Signed-off-by Greg Ociepka <[email protected]>Assisted-by Claude:fable-5
Quoted 16 lines · Greg Ociepka
All five USB controller nodes on Glymur (
usb_0,usb_1,usb_2,usb_hs,usb_mp) are missing the dma-coherent property, which Hamoa carries on all of its USB controllers. Krishna Kurapati confirmed the Glymur USB controllers are dma-coherent [1].Without the property the kernel treats USB DMA buffers as non-coherent and performs cache maintenance on every transfer, which coherent hardware makes redundant.
Follow-up to the clock/power/bandwidth votes patch [2], kept separate so the already-reviewed patch stands as is.
Tested on an Asus Zenbook A16 (UX3607OA): all controllers enumerate their devices as before and a 64 MiB write plus two
O_DIRECTread-backs over USB mass storage produce identical checksums.
Quoted 6 lines
[1] https://lore.kernel.org/linux-arm-msm/[email protected]/
[2] https://lore.kernel.org/linux-arm-msm/[email protected]/Fixes 4eee57dd4df9 ("arm64: dts: qcom: glymur: Add USB related nodes")Signed-off-by Greg Ociepka <[email protected]>Assisted-by Claude:fable-5
Quoted 10 lines · Greg Ociepka
All five USB controller nodes on Glymur (
usb_0,usb_1,usb_2,usb_hs,usb_mp) are missing the dma-coherent property, which Hamoa carries on all of its USB controllers. Krishna Kurapati confirmed the Glymur USB controllers are dma-coherent [1].Without the property the kernel treats USB DMA buffers as non-coherent and performs cache maintenance on every transfer, which coherent hardware makes redundant.
[...]
Applied, thanks!
[1/1] arm64: dts: qcom: glymur: Mark USB controllers dma-coherent
commit: e62df36f34949aebe016c5c81e34baa7a9eca427
Best regards,
Signature · 1 line
Bjorn Andersson <[email protected]>