Skip to contentSkip to conversation

USB: host: Add the USB3 host driver

PATCH Superseded Sherry Sun 4 participants 15 messages Aug 20, 2019 – Sep 9, 2019

v4 of 5 Superseded Latest is v5 →

lore
Share & tools
lore

Sherry Sun

author
#0 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Marek,

Quoted 29 lines · Sherry Sun
Quoted 20 lines

Hi Marek,

Quoted 18 lines · Sherry Sun
Quoted 5 lines

Add the USB3 host driver for NXP imx8 platform, and the cadence IP is in it. The USB3 host driver support DM mode, it will probe USB3 host node in dts.

[...]

Review

Quoted 5 lines
(quoted diff) +5 -0
+static void xhci_imx8_get_reg_addr(struct udevice *dev) {
+	imx8_data.usb3_ctrl_base =
+			(void __iomem *)devfdt_get_addr_index(dev, 0);
+	imx8_data.usb3_core_base =
+			(void __iomem *)devfdt_get_addr_index(dev, 4); }

Inline this.

Quoted 7 lines
Quoted 3 lines

Also, look at drivers/mtd/renesas_rpc_hf.c to handle both 32bit and 64bit DTs with multiple "reg" entries.

Actually, I don't think it's needed to change the way getting reg in dts. For two reasons:
1. This xhci-imx8 driver is only target on imx8 platform which all use 64bit

Quoted 6 lines

dts.

Quoted 1 line

2. devfdt_get_addr_index() can handle both 32/64 bits dts too.

By hard-coding this "4" offset, you assume the DT is using 64bit addressing, that's wrong, so please fix it. The iMX8 does support aarch32, right ? So someone might pass in 32bit DT and this would break.

Sorry, maybe there are some misunderstandings in the code.
There five reg elements in dts node, which are none_core_regs/ xhci_regs/ dev_regs/ phy_regs/ otg_regs. devfdt_get_addr_index(dev, 4) only want to get the fifth reg value, instead of meanning 64bit address.

586 usbotg3: usb3 at 5b110000 {
587 compatible = "cdns,usb3";
588 reg = <0x0 0x5B110000 0x0 0x10000>,

589             <0x0 0x5B130000 0x0 0x10000>,
590             <0x0 0x5B140000 0x0 0x10000>,
591             <0x0 0x5B160000 0x0 0x40000>,
592             <0x0 0x5B120000 0x0 0x10000>;
Quoted 18 lines
Quoted 8 lines
Quoted 8 lines

[...]

Review

Quoted 2 lines
(quoted diff) +2 -0
+static const struct udevice_id xhci_usb_ids[] = {
+	{ .compatible = "cdns,usb3-host", },

https://lore.ker

nel.org%2Fpatchwork%2Fpatch%2F1059917%2F&amp;data=02%7C01%7Cshe

Quoted 1 line
Quoted 1 line

rry.sun%40nxp.com%7C70e116ce152a4060dbd508d724987b54%7C686ea1d

Quoted 1 line
Quoted 1 line

3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C637018109631327748&amp;sd

Quoted 1 line
Quoted 1 line

ata=GzX5KNQ8m0CCeBa9tPSV%2BOEwwuSsxDdAiJd1T6or8P0%3D&amp;reser

Quoted 3 lines
Quoted 2 lines

ved=0 would suggest that
cdns,usb3-1.0.0 is the compatible. But I might be wrong.

Quoted 6 lines
Quoted 3 lines

This patch has not been accepted by the kernel, right? So I think maybe we can do this change to synchronize with kernel after the cdns3 kernel patch been accepted.

Or, we can start with at least reasonably OK DT compatible right away to reasonably avoid ABI breakage.

Sure, will change the compatible from "cdns,usb3" to "cdns,usb3-1.0.0".

Quoted 9 lines
Quoted 5 lines

And note that in uboot, one dts node can only bind with one driver, so we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two dt nodes to bind with.
And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host driver, the compatible will use "cdns,usb3-1.0.0-host". What do you think about it?

Quoted 18 lines
Quoted 18 lines
Quoted 17 lines
Quoted 4 lines
(quoted diff) +3 -0
+	{ }
+};
+

Review

Quoted 10 lines
(quoted diff) +10 -0
+U_BOOT_DRIVER(xhci_imx8) = {
+	.name	= "xhci_imx8",
+	.id	= UCLASS_USB,
+	.of_match = xhci_usb_ids,
+	.probe = xhci_imx8_probe,
+	.remove = xhci_imx8_remove,
+	.ops	= &xhci_usb_ops,
+	.platdata_auto_alloc_size = sizeof(struct usb_platdata),
+	.priv_auto_alloc_size = sizeof(struct xhci_ctrl),
+	.flags	= DM_FLAG_ALLOC_PRIV_DMA,

I think you also need DM_FLAG_OS_PREPARE to trigger .remove before booting Linux, but I might be wrong. Please check that.

Quoted 7 lines
Quoted 4 lines

When use usb stop command to stop the usb host driver, device_remove(bus, DM_REMOVE_NORMAL) is called by usb_stop(), so normally we don’t need the DM_FLAG_OS_PREPARE flag.
But considering some special circumstances, maybe add this flag is helpful.

I'm not talking about "usb stop", I am talking about booting Linux. That's when then .remove should be called to tear down the driver. Is it?

I know what you mean. The only function of DM_FLAG_OS_PREPARE
is to call .remove before booting kernel, to make sure the device is removed.
But normally, the host device is removed every time after we use it. So even the flag is set, the .remove won't be called again before booting kernel because the device is checked to be removed already.
Add the flag just make sure the device will be removed before booting kernel. Do you think so?

Best regard
Sherry sun

#1 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Quoted 1 line · Sherry Sun

Hi Marek,

Hi,

Quoted 39 lines
Quoted 28 lines · Sherry Sun
Quoted 20 lines

Hi Marek,

Quoted 18 lines · Sherry Sun
Quoted 5 lines

Add the USB3 host driver for NXP imx8 platform, and the cadence IP is in it. The USB3 host driver support DM mode, it will probe USB3 host node in dts.

[...]

Review

Quoted 5 lines
(quoted diff) +5 -0
+static void xhci_imx8_get_reg_addr(struct udevice *dev) {
+	imx8_data.usb3_ctrl_base =
+			(void __iomem *)devfdt_get_addr_index(dev, 0);
+	imx8_data.usb3_core_base =
+			(void __iomem *)devfdt_get_addr_index(dev, 4); }

Inline this.

Quoted 7 lines
Quoted 3 lines

Also, look at drivers/mtd/renesas_rpc_hf.c to handle both 32bit and 64bit DTs with multiple "reg" entries.

Actually, I don't think it's needed to change the way getting reg in dts. For two reasons:
1. This xhci-imx8 driver is only target on imx8 platform which all use 64bit

Quoted 6 lines

dts.

Quoted 1 line

2. devfdt_get_addr_index() can handle both 32/64 bits dts too.

By hard-coding this "4" offset, you assume the DT is using 64bit addressing, that's wrong, so please fix it. The iMX8 does support aarch32, right ? So someone might pass in 32bit DT and this would break.

Sorry, maybe there are some misunderstandings in the code.
There five reg elements in dts node, which are none_core_regs/ xhci_regs/ dev_regs/ phy_regs/ otg_regs. devfdt_get_addr_index(dev, 4) only want to get the fifth reg value, instead of meanning 64bit address.

Quoted 7 lines

586 usbotg3: usb3 at 5b110000 {
587 compatible = "cdns,usb3";
588 reg = <0x0 0x5B110000 0x0 0x10000>,

589             <0x0 0x5B130000 0x0 0x10000>,
590             <0x0 0x5B140000 0x0 0x10000>,
591             <0x0 0x5B160000 0x0 0x40000>,
592             <0x0 0x5B120000 0x0 0x10000>;

Look the registers up by name then, each entry should have it's own name in the DT. git grep for "reg-names" .

[...]

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines

And note that in uboot, one dts node can only bind with one driver, so we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 8 lines
Quoted 3 lines
Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two dt nodes to bind with.
And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host driver, the compatible will use "cdns,usb3-1.0.0-host". What do you think about it?

CCing Jean, since I think he did solve similar topic for his platform.

Quoted 22 lines
Quoted 22 lines
Quoted 21 lines
Quoted 8 lines
Quoted 8 lines
(quoted diff) +8 -0
+	{ }
+};
+
+U_BOOT_DRIVER(xhci_imx8) = {
+	.name	= "xhci_imx8",
+	.id	= UCLASS_USB,
+	.of_match = xhci_usb_ids,
+	.probe = xhci_imx8_probe,
Quoted 8 lines

Review

Quoted 5 lines
(quoted diff) +5 -0
+	.remove = xhci_imx8_remove,
+	.ops	= &xhci_usb_ops,
+	.platdata_auto_alloc_size = sizeof(struct usb_platdata),
+	.priv_auto_alloc_size = sizeof(struct xhci_ctrl),
+	.flags	= DM_FLAG_ALLOC_PRIV_DMA,

I think you also need DM_FLAG_OS_PREPARE to trigger .remove before booting Linux, but I might be wrong. Please check that.

When use usb stop command to stop the usb host driver, device_remove(bus, DM_REMOVE_NORMAL) is called by usb_stop(), so normally we don’t need the DM_FLAG_OS_PREPARE flag.
But considering some special circumstances, maybe add this flag is helpful.

Quoted 7 lines
Quoted 2 lines

I'm not talking about "usb stop", I am talking about booting Linux. That's when then .remove should be called to tear down the driver. Is it?

I know what you mean. The only function of DM_FLAG_OS_PREPARE is to call .remove before booting kernel, to make sure the device is removed.
But normally, the host device is removed every time after we use it.

How so ? The controller is left running until you call .remove()

Quoted 4 lines

So even the flag is set, the .remove won't be called again before booting kernel because the device is checked to be removed already.
Add the flag just make sure the device will be removed before booting kernel. Do you think so?

You want to tear the controller down before booting kernel, e.g. to prevent it from doing some rogue DMA into memory, while the kernel starts up. This might mess the system up.

#2 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Marek, Sherry,

Quoted 11 lines
Quoted 10 lines
Quoted 6 lines
Quoted 3 lines

we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two dt nodes to bind with.
And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host driver, the compatible will use "cdns,usb3-1.0.0-host". What do you think about it?

CCing Jean, since I think he did solve similar topic for his platform.

I've been OOO for a few weeks and didn't look at the whole series.

For this particular issue, the solution I used is to let the wrapper do the binding job. The name of the driver to use is hard-coded in the wrapper diver.

This is done in dwc3_glue_bind().

JJ

Quoted 4 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
(quoted diff) +2 -0
+	{ }
+};
author
#3 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Jean,

Quoted 17 lines

Hi Marek, Sherry,

Quoted 8 lines
Quoted 8 lines
Quoted 7 lines
Quoted 4 lines

we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two

dt nodes to bind with.

Quoted 2 lines
Quoted 2 lines

And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host

driver, the compatible will use "cdns,usb3-1.0.0-host".

Quoted 1 line
Quoted 1 line

What do you think about it?

Quoted 8 lines
Quoted 1 line

CCing Jean, since I think he did solve similar topic for his platform.

I've been OOO for a few weeks and didn't look at the whole series.

For this particular issue, the solution I used is to let the wrapper do the binding job. The name of the driver to use is hard-coded in the wrapper diver.

This is done in dwc3_glue_bind().

Thanks for your suggestions.

So if I want to use the cdns3 usb node as both usb gadget device and usb host device,
do you mean that I should make the cdns3 usb node as a usb wrapper device, and create two subnodes in it. Then when binding the wrapper node, it will hard-coded the two subnodes to different driver(gadge/host driver) according to the dr_mode property in nodes.

Best regards
Sherry sun

Quoted 9 lines

JJ

Quoted 4 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
(quoted diff) +2 -0
+	{ }
+};
author
#4 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Jean,

Quoted 32 lines

Hi Jean,

Quoted 17 lines

Hi Marek, Sherry,

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines

we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two

dt nodes to bind with.

Quoted 3 lines
Quoted 3 lines

And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host

Quoted 8 lines

driver, the compatible will use "cdns,usb3-1.0.0-host".

Quoted 2 lines
Quoted 1 line

What do you think about it?

CCing Jean, since I think he did solve similar topic for his platform.

I've been OOO for a few weeks and didn't look at the whole series.

For this particular issue, the solution I used is to let the wrapper do the binding job. The name of the driver to use is hard-coded in the

wrapper diver.

Quoted 2 lines

This is done in dwc3_glue_bind().

Quoted 9 lines

Thanks for your suggestions.

So if I want to use the cdns3 usb node as both usb gadget device and usb host device, do you mean that I should make the cdns3 usb node as a usb wrapper device, and create two subnodes in it.
Then when binding the wrapper node, it will hard-coded the two subnodes to different driver(gadge/host driver) according to the dr_mode property in nodes.

Do you think I understand it right?

Best regards
Sherry sun

Quoted 21 lines

Best regards
Sherry sun

Quoted 9 lines

JJ

Quoted 4 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
(quoted diff) +2 -0
+	{ }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.d
enx.de%2Flistinfo%2Fu- boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7 670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7 C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI GrNOxDH6HC0%3D&amp;reserved=0

#5 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Quoted 34 lines · Sherry Sun

Hi Jean,

Quoted 32 lines

Hi Jean,

Quoted 16 lines

Hi Marek, Sherry,

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines

we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two

dt nodes to bind with.

Quoted 3 lines
Quoted 3 lines

And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host

Quoted 7 lines

driver, the compatible will use "cdns,usb3-1.0.0-host".

Quoted 2 lines
Quoted 1 line

What do you think about it?

CCing Jean, since I think he did solve similar topic for his platform.

I've been OOO for a few weeks and didn't look at the whole series.

For this particular issue, the solution I used is to let the wrapper do the binding job. The name of the driver to use is hard-coded in the

wrapper diver.

Quoted 1 line

This is done in dwc3_glue_bind().

Thanks for your suggestions.

So if I want to use the cdns3 usb node as both usb gadget device and usb host device, do you mean that I should make the cdns3 usb node as a usb wrapper device, and create two subnodes in it.

I think we should not change the binding to adapt to out driver but keep the bindings that exist in linux and adapt the u-boot driver

In the version used by our platform, there is a wrapper around the USB:

    usbss0: cdns_usb at 4104000 {
        compatible = "ti,j721e-usb";
        [....]
        usb0: usb at 6000000 {
            compatible = "cdns,usb3-1.0.1";

The driver selection (host or device) could be done by the wrapper when it binds its children (same as the dwc3).

OR

The "cdns,usb3-1.0.1" could be a dumb driver the role of which would be only to bind a new driver (host or device) based on "dr_mode". The binding could be done using the same node, it doesn't have to be a subnode.

Maybe the second solution will be better, as it would work for platforms that do not use a wrapper.

JJ

Quoted 28 lines
Quoted 4 lines

Then when binding the wrapper node, it will hard-coded the two subnodes to different driver(gadge/host driver) according to the dr_mode property in nodes.

Do you think I understand it right?

Best regards
Sherry sun

Quoted 18 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+	{ }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flists.d
enx.de%2Flistinfo%2Fu- boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7 670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7 C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI GrNOxDH6HC0%3D&amp;reserved=0

author
#6 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Jean,

Quoted 55 lines · Sherry Sun
Quoted 35 lines

Hi Jean,

Quoted 33 lines

Hi Jean,

Quoted 16 lines

Hi Marek, Sherry,

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines

we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two

dt nodes to bind with.

Quoted 3 lines
Quoted 3 lines

And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host

Quoted 8 lines

driver, the compatible will use "cdns,usb3-1.0.0-host".

Quoted 2 lines
Quoted 1 line

What do you think about it?

CCing Jean, since I think he did solve similar topic for his platform.

I've been OOO for a few weeks and didn't look at the whole series.

For this particular issue, the solution I used is to let the wrapper do the binding job. The name of the driver to use is hard-coded in the

wrapper diver.

Quoted 1 line

This is done in dwc3_glue_bind().

Thanks for your suggestions.

So if I want to use the cdns3 usb node as both usb gadget device and usb host device, do you mean that I should make the cdns3 usb node as a usb wrapper device, and create two subnodes in it.

I think we should not change the binding to adapt to out driver but keep the bindings that exist in linux and adapt the u-boot driver

In the version used by our platform, there is a wrapper around the USB:

    usbss0: cdns_usb at 4104000 {
        compatible = "ti,j721e-usb";
        [....]
        usb0: usb at 6000000 {
            compatible = "cdns,usb3-1.0.1";

The driver selection (host or device) could be done by the wrapper when it binds its children (same as the dwc3).

OR

Quoted 8 lines

The "cdns,usb3-1.0.1" could be a dumb driver the role of which would be only to bind a new driver (host or device) based on "dr_mode". The binding could be done using the same node, it doesn't have to be a subnode.

Maybe the second solution will be better, as it would work for platforms that do not use a wrapper.

I just communicated with Vignesh Raghavendra <[email protected]>, and he suggest that I should keep this cdns3 driver under Linux kernel and U-Boot in sync. And show me your downstream code with v10 of Cadence USB3 kernel driver ported to U-Boot. So I decide to follow your way to deal with this issue.

But I want to ask another question:
The two solutions you gave before both make the usb node with compatible "cdns,usb3-1.0.1" as a definite device (host or gadget) by its dr_mode property. If I want use this usb device works as both host and gadget driver, which means I want to change its status runtime, such as I want to use this usb device to run both fastboot or usb start command, how can we deal with this? .

Best regards
Sherry sun

Quoted 36 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 1 line
Quoted 1 line

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 23 lines
Quoted 2 lines

nodes.

Do you think I understand it right?

Best regards
Sherry sun

Quoted 15 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+	{ }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.d
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7

Quoted 1 line
Quoted 1 line

670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7

Quoted 1 line
Quoted 1 line

C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

#7 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Quoted 62 lines · Sherry Sun

Hi Jean,

Quoted 60 lines · Sherry Sun
Quoted 35 lines

Hi Jean,

Quoted 33 lines

Hi Jean,

Quoted 16 lines

Hi Marek, Sherry,

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines

we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two

dt nodes to bind with.

Quoted 3 lines
Quoted 3 lines

And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host

Quoted 8 lines

driver, the compatible will use "cdns,usb3-1.0.0-host".

Quoted 2 lines
Quoted 1 line

What do you think about it?

CCing Jean, since I think he did solve similar topic for his platform.

I've been OOO for a few weeks and didn't look at the whole series.

For this particular issue, the solution I used is to let the wrapper do the binding job. The name of the driver to use is hard-coded in the

wrapper diver.

Quoted 1 line

This is done in dwc3_glue_bind().

Thanks for your suggestions.

So if I want to use the cdns3 usb node as both usb gadget device and usb host device, do you mean that I should make the cdns3 usb node as a usb wrapper device, and create two subnodes in it.

I think we should not change the binding to adapt to out driver but keep the bindings that exist in linux and adapt the u-boot driver

In the version used by our platform, there is a wrapper around the USB:

    usbss0: cdns_usb at 4104000 {
        compatible = "ti,j721e-usb";
        [....]
        usb0: usb at 6000000 {
            compatible = "cdns,usb3-1.0.1";

The driver selection (host or device) could be done by the wrapper when it binds its children (same as the dwc3).

OR

The "cdns,usb3-1.0.1" could be a dumb driver the role of which would be only to bind a new driver (host or device) based on "dr_mode". The binding could be done using the same node, it doesn't have to be a subnode.

Quoted 9 lines
Quoted 3 lines

Maybe the second solution will be better, as it would work for platforms that do not use a wrapper.

I just communicated with Vignesh Raghavendra <[email protected]>, and he suggest that I should keep this cdns3 driver under Linux kernel and U-Boot in sync. And show me your downstream code with v10 of Cadence USB3 kernel driver ported to U-Boot. So I decide to follow your way to deal with this issue.

But I want to ask another question:
The two solutions you gave before both make the usb node with compatible "cdns,usb3-1.0.1" as a definite device (host or gadget) by its dr_mode property. If I want use this usb device works as both host and gadget driver, which means I want to change its status runtime, such as I want to use this usb device to run both fastboot or usb start command, how can we deal with this? .

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like DWC3/MUSB etc.

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

Regards
Vignesh

Quoted 39 lines

Best regards
Sherry sun

Quoted 36 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 1 line
Quoted 1 line

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 23 lines
Quoted 2 lines

nodes.

Do you think I understand it right?

Best regards
Sherry sun

Quoted 15 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+	{ }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://lists.d
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7

Quoted 1 line
Quoted 1 line

670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7

Quoted 1 line
Quoted 1 line

C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

Signature · 2 lines
Regards
Vignesh
author
#8 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Vignesh,

Quoted 85 lines · Sherry Sun
Quoted 60 lines

Hi Jean,

Quoted 58 lines · Sherry Sun
Quoted 35 lines

Hi Jean,

Quoted 33 lines

Hi Jean,

Quoted 16 lines

Hi Marek, Sherry,

Quoted 9 lines
Quoted 9 lines
Quoted 7 lines
Quoted 4 lines

we keep the cdns3 node for usb gadget driver, then add a usb host node for
xhci-imx8 driver in *-uboot.dtsi. so here is no need to change the host driver

compatible.

Quoted 1 line

But the compatible in gadget driver should be changed later.

We should try avoiding ABI breaks in DT.

But the cdns3 usb gaget driver and host driver in different uclass need two

dt nodes to bind with.

Quoted 3 lines
Quoted 3 lines

And the compatible of the two node cannot be same.
So for gadget driver, the compatible may use "cdns,usb3-1.0.0", for host

Quoted 8 lines

driver, the compatible will use "cdns,usb3-1.0.0-host".

Quoted 2 lines
Quoted 1 line

What do you think about it?

CCing Jean, since I think he did solve similar topic for his platform.

I've been OOO for a few weeks and didn't look at the whole series.

For this particular issue, the solution I used is to let the wrapper do the binding job. The name of the driver to use is hard-coded in the

wrapper diver.

Quoted 1 line

This is done in dwc3_glue_bind().

Thanks for your suggestions.

So if I want to use the cdns3 usb node as both usb gadget device and usb host device, do you mean that I should make the cdns3 usb node as a usb wrapper device, and create two subnodes in it.

I think we should not change the binding to adapt to out driver but keep the bindings that exist in linux and adapt the u-boot driver

In the version used by our platform, there is a wrapper around the USB:

    usbss0: cdns_usb at 4104000 {
        compatible = "ti,j721e-usb";
        [....]
        usb0: usb at 6000000 {
            compatible = "cdns,usb3-1.0.1";

The driver selection (host or device) could be done by the wrapper when it binds its children (same as the dwc3).

OR

The "cdns,usb3-1.0.1" could be a dumb driver the role of which would be only to bind a new driver (host or device) based on "dr_mode". The binding could be done using the same node, it doesn't have to be a

subnode.

Quoted 7 lines
Quoted 5 lines

Maybe the second solution will be better, as it would work for platforms that do not use a wrapper.

I just communicated with Vignesh Raghavendra <[email protected]>, and he

suggest that I should keep this cdns3 driver under Linux kernel and U-Boot in sync. And show me your downstream code with v10 of Cadence USB3 kernel driver ported to U-Boot. So I decide to follow your way to deal with this issue.

Quoted 3 lines

But I want to ask another question:
The two solutions you gave before both make the usb node with

compatible "cdns,usb3-1.0.1" as a definite device (host or gadget) by its dr_mode property. If I want use this usb device works as both host and gadget driver, which means I want to change its status runtime, such as I want to use this usb device to run both fastboot or usb start command, how can we deal with this? .

Quoted 1 line
Quoted 6 lines

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like DWC3/MUSB etc.

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

I wonder if we can add two subnodes under the wrapper node as below, one bind to usb gadget driver and another bind to usb host driver. So if we want to use usb device, just call the gadget driver, and if want to use usb host, just call the host driver. The driver will probe correspond node. I'm not sure if it is feasible, what do you think about it?

      usbss0: cdns_usb at 4104000 {
          compatible = "ti,j721e-usb";
          [....]
          usb0: usb at 6010000 {		/* xhci reg address*/
              compatible = "cdns,usb3-1.0.1";
	dr_mode = "host";
	[....]
	}
          usb1: usb at 6020000 {		/* dev reg address*/
              compatible = "cdns,usb3-1.0.1";
	dr_mode = "peripheral";
	[....]
	}
    }

Best regards
Sherry sun

Quoted 61 lines

Regards
Vignesh

Quoted 33 lines

Best regards
Sherry sun

Quoted 30 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 2 lines
Quoted 2 lines

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 22 lines
Quoted 2 lines

nodes.

Do you think I understand it right?

Best regards
Sherry sun

Quoted 14 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

Review

Quoted 2 lines
(quoted diff) +2 -0
+	{ }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fl

ists.d&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C7d1d76a7124f4c3f e9

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

9d08d72d3ddf82%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63 70276

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

16080089878&amp;sdata=70BPVQkomokcNpxsHRD3njfZQvuG51VSD1QKBjQ o1kA%3

Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

D&amp;reserved=0
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

--
Regards
Vignesh

#9 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Sherry,

[...]

Quoted 11 lines
Quoted 6 lines

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like DWC3/MUSB etc.

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

I wonder if we can add two subnodes under the wrapper node as below, one bind to usb gadget driver and another bind to usb host driver. So if we want to use usb device, just call the gadget driver, and if want to use usb host, just call the host driver. The driver will probe correspond node. I'm not sure if it is feasible, what do you think about it?

Quoted 15 lines
      usbss0: cdns_usb at 4104000 {
          compatible = "ti,j721e-usb";
          [....]
          usb0: usb at 6010000 {		/* xhci reg address*/
              compatible = "cdns,usb3-1.0.1";
	dr_mode = "host";
	[....]
	}
          usb1: usb at 6020000 {		/* dev reg address*/
              compatible = "cdns,usb3-1.0.1";
	dr_mode = "peripheral";
	[....]
	}
    }

But this is not how DT would look in kernel. There will be single DT node representing both Host and Device node. DT repesentation should not be changed based on OS/bootloader, its HW description and must remain same. Unbinding host/gadget driver and rebinding with a different role should not be hard as the U-Boot core infrastructure exists.

Moreover if xhci reg and dev reg are separated into different nodes dont we still need access to OTG register block to switch b/w host and device mode?

Regards
Vignesh

Quoted 64 lines

Best regards
Sherry sun

Quoted 61 lines

Regards
Vignesh

Quoted 33 lines

Best regards
Sherry sun

Quoted 30 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 2 lines
Quoted 2 lines

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 22 lines
Quoted 2 lines

nodes.

Do you think I understand it right?

Best regards
Sherry sun

Quoted 14 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

+ { }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fl

ists.d&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C7d1d76a7124f4c3f e9

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

9d08d72d3ddf82%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63 70276

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

16080089878&amp;sdata=70BPVQkomokcNpxsHRD3njfZQvuG51VSD1QKBjQ o1kA%3

Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

D&amp;reserved=0
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

--
Regards
Vignesh

author
#10 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Vignesh,

Quoted 34 lines

Hi Sherry,

[...]

Quoted 2 lines
Quoted 2 lines

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like

DWC3/MUSB etc.

Quoted 6 lines
Quoted 4 lines

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

I wonder if we can add two subnodes under the wrapper node as below,

one bind to usb gadget driver and another bind to usb host driver.

Quoted 1 line

So if we want to use usb device, just call the gadget driver, and if want to

use usb host, just call the host driver. The driver will probe correspond node.

Quoted 17 lines

I'm not sure if it is feasible, what do you think about it?

      usbss0: cdns_usb at 4104000 {
          compatible = "ti,j721e-usb";
          [....]
          usb0: usb at 6010000 {		/* xhci reg address*/
              compatible = "cdns,usb3-1.0.1";
	dr_mode = "host";
	[....]
	}
          usb1: usb at 6020000 {		/* dev reg address*/
              compatible = "cdns,usb3-1.0.1";
	dr_mode = "peripheral";
	[....]
	}
    }
Quoted 8 lines

But this is not how DT would look in kernel. There will be single DT node representing both Host and Device node. DT repesentation should not be changed based on OS/bootloader, its HW description and must remain same. Unbinding host/gadget driver and rebinding with a different role should not be hard as the U-Boot core infrastructure exists.

Moreover if xhci reg and dev reg are separated into different nodes dont we still need access to OTG register block to switch b/w host and device mode?

I think we may not need to access OTG register if we add two subnodes for gadget and host.

But I see a for loop in dwc3_glue_bind() as follows, if there only one single node representing both Host and Device node under usb wrapper node, why we need this for loop?

212 for (node = fdt_first_subnode(fdt, dev_of_offset(parent)); node > 0;
213 node = fdt_next_subnode(fdt, node)) {

Best regards
Sherry sun

Quoted 74 lines

Regards
Vignesh

Quoted 41 lines

Best regards
Sherry sun

Quoted 38 lines

Regards
Vignesh

Quoted 33 lines

Best regards
Sherry sun

Quoted 30 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 2 lines
Quoted 2 lines

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 22 lines
Quoted 2 lines

nodes.

Do you think I understand it right?

Best regards
Sherry sun

Quoted 14 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

+ { }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://l

ists.d&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C7d1d76a7124f4c3f

Quoted 3 lines
Quoted 3 lines

e9

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

9d08d72d3ddf82%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63

Quoted 3 lines
Quoted 3 lines

70276

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

16080089878&amp;sdata=70BPVQkomokcNpxsHRD3njfZQvuG51VSD1QKBjQ

Quoted 6 lines
Quoted 6 lines

o1kA%3

Quoted 4 lines
Quoted 4 lines
Quoted 3 lines
Quoted 3 lines

D&amp;reserved=0
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7

Quoted 3 lines
Quoted 3 lines
Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 5 lines
Quoted 5 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

--
Regards
Vignesh

#11 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Quoted 38 lines · Sherry Sun

Hi Vignesh,

Quoted 36 lines

Hi Sherry,

[...]

Quoted 2 lines
Quoted 2 lines

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like

DWC3/MUSB etc.

Quoted 4 lines
Quoted 3 lines

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

I wonder if we can add two subnodes under the wrapper node as below,

one bind to usb gadget driver and another bind to usb host driver.

Quoted 1 line

So if we want to use usb device, just call the gadget driver, and if want to

use usb host, just call the host driver. The driver will probe correspond node.

Quoted 17 lines

I'm not sure if it is feasible, what do you think about it?

       usbss0: cdns_usb at 4104000 {
           compatible = "ti,j721e-usb";
           [....]
           usb0: usb at 6010000 {		/* xhci reg address*/
               compatible = "cdns,usb3-1.0.1";
	dr_mode = "host";
	[....]
	}
           usb1: usb at 6020000 {		/* dev reg address*/
               compatible = "cdns,usb3-1.0.1";
	dr_mode = "peripheral";
	[....]
	}
     }

But this is not how DT would look in kernel. There will be single DT node representing both Host and Device node. DT repesentation should not be changed based on OS/bootloader, its HW description and must remain same. Unbinding host/gadget driver and rebinding with a different role should not be hard as the U-Boot core infrastructure exists.

Quoted 8 lines
Quoted 2 lines

Moreover if xhci reg and dev reg are separated into different nodes dont we still need access to OTG register block to switch b/w host and device mode?

I think we may not need to access OTG register if we add two subnodes for gadget and host.

But I see a for loop in dwc3_glue_bind() as follows, if there only one single node representing both Host and Device node under usb wrapper node, why we need this for loop?

212 for (node = fdt_first_subnode(fdt, dev_of_offset(parent)); node > 0;
213 node = fdt_next_subnode(fdt, node)) {

I believe this is a legacy from the code it was inspired from.

Indeed the loop is not required.

Like Vignesh I think we must stick with the dt-bindings used by the kernel.

BTW we are working on the USB3 support right now that is lacking in our tree. I choose to keep the PHY drivers as close as possible to their linux version. I'll probably start posting preliminary patches this week.

If you have the USB3 working in the kernel, can you point to a tree where I can have a look at the drivers and dt-bindings ?

JJ

Quoted 59 lines

Best regards
Sherry sun

Quoted 55 lines

Regards
Vignesh

Quoted 36 lines

Best regards
Sherry sun

Quoted 33 lines

Regards
Vignesh

Quoted 30 lines

Best regards
Sherry sun

Quoted 27 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 2 lines
Quoted 2 lines

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 20 lines
Quoted 2 lines

nodes.

Do you think I understand it right?
Best regards
Sherry sun

Quoted 14 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

+ { }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://l

ists.d&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C7d1d76a7124f4c3f

Quoted 1 line
Quoted 1 line

e9

9d08d72d3ddf82%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63

Quoted 1 line
Quoted 1 line

70276

16080089878&amp;sdata=70BPVQkomokcNpxsHRD3njfZQvuG51VSD1QKBjQ

Quoted 4 lines
Quoted 4 lines

o1kA%3

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

D&amp;reserved=0
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7 670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7 C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 4 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

--
Regards
Vignesh

author
#12 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Jean,

Quoted 55 lines · Sherry Sun
Quoted 17 lines

Hi Vignesh,

Quoted 15 lines

Hi Sherry,

[...]

Quoted 3 lines
Quoted 3 lines

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like

DWC3/MUSB etc.

Quoted 4 lines
Quoted 3 lines

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

I wonder if we can add two subnodes under the wrapper node as below,

one bind to usb gadget driver and another bind to usb host driver.

Quoted 2 lines

So if we want to use usb device, just call the gadget driver, and if want to

use usb host, just call the host driver. The driver will probe correspond

node.

Quoted 17 lines
Quoted 17 lines
Quoted 17 lines

I'm not sure if it is feasible, what do you think about it?

       usbss0: cdns_usb at 4104000 {
           compatible = "ti,j721e-usb";
           [....]
           usb0: usb at 6010000 {		/* xhci reg address*/
               compatible = "cdns,usb3-1.0.1";
	dr_mode = "host";
	[....]
	}
           usb1: usb at 6020000 {		/* dev reg address*/
               compatible = "cdns,usb3-1.0.1";
	dr_mode = "peripheral";
	[....]
	}
     }
Quoted 3 lines
Quoted 3 lines

But this is not how DT would look in kernel. There will be single DT node representing both Host and Device node. DT repesentation should not be changed based on OS/bootloader, its HW description and must

remain same.

Quoted 5 lines
Quoted 5 lines

Unbinding host/gadget driver and rebinding with a different role should not be hard as the U-Boot core infrastructure exists.

Moreover if xhci reg and dev reg are separated into different nodes dont we still need access to OTG register block to switch b/w host and

device mode?

Quoted 1 line

I think we may not need to access OTG register if we add two subnodes for

gadget and host.

Quoted 2 lines

But I see a for loop in dwc3_glue_bind() as follows, if there only one single

node representing both Host and Device node under usb wrapper node, why we need this for loop?

Quoted 1 line
Quoted 8 lines
Quoted 2 lines

212 for (node = fdt_first_subnode(fdt, dev_of_offset(parent)); node > 0;
213 node = fdt_next_subnode(fdt, node)) {

I believe this is a legacy from the code it was inspired from.

Indeed the loop is not required.

Like Vignesh I think we must stick with the dt-bindings used by the kernel.

Thanks for your reply, I understand now.

Quoted 7 lines

BTW we are working on the USB3 support right now that is lacking in our tree. I choose to keep the PHY drivers as close as possible to their linux version. I'll probably start posting preliminary patches this week.

If you have the USB3 working in the kernel, can you point to a tree where I can have a look at the drivers and dt-bindings ?

Sure, you can see our downstream kernel code at my github, here is the link address: https://github.com/sherrysun1/linux-imx.git.
And look forward to seeing your patches in uboot maillist.

Best regards
Sherry sun

Quoted 68 lines

JJ

Quoted 44 lines

Best regards
Sherry sun

Quoted 40 lines

Regards
Vignesh

Quoted 36 lines

Best regards
Sherry sun

Quoted 33 lines

Regards
Vignesh

Quoted 30 lines

Best regards
Sherry sun

Quoted 27 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 2 lines
Quoted 2 lines

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 20 lines
Quoted 2 lines

nodes.

Do you think I understand it right?
Best regards
Sherry sun

Quoted 14 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines
Quoted 2 lines

+ { }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://l

ists.d&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C7d1d76a7124f4c3f

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

e9

9d08d72d3ddf82%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line

70276

16080089878&amp;sdata=70BPVQkomokcNpxsHRD3njfZQvuG51VSD1QKBjQ

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines
Quoted 4 lines

o1kA%3

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

D&amp;reserved=0
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7

Quoted 1 line
Quoted 1 line

670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7

Quoted 1 line
Quoted 1 line

C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

--
Regards
Vignesh

#13 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Sherry,

Quoted 63 lines · Sherry Sun

Hi Jean,

Quoted 52 lines · Sherry Sun
Quoted 17 lines

Hi Vignesh,

Quoted 15 lines

Hi Sherry,

[...]

Quoted 3 lines
Quoted 3 lines

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like

DWC3/MUSB etc.

Quoted 4 lines
Quoted 3 lines

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

I wonder if we can add two subnodes under the wrapper node as below,

one bind to usb gadget driver and another bind to usb host driver.

Quoted 2 lines

So if we want to use usb device, just call the gadget driver, and if want to

use usb host, just call the host driver. The driver will probe correspond

node.

Quoted 17 lines
Quoted 17 lines
Quoted 17 lines

I'm not sure if it is feasible, what do you think about it?

        usbss0: cdns_usb at 4104000 {
            compatible = "ti,j721e-usb";
            [....]
            usb0: usb at 6010000 {		/* xhci reg address*/
                compatible = "cdns,usb3-1.0.1";
	dr_mode = "host";
	[....]
	}
            usb1: usb at 6020000 {		/* dev reg address*/
                compatible = "cdns,usb3-1.0.1";
	dr_mode = "peripheral";
	[....]
	}
      }
Quoted 3 lines
Quoted 3 lines

But this is not how DT would look in kernel. There will be single DT node representing both Host and Device node. DT repesentation should not be changed based on OS/bootloader, its HW description and must

remain same.

Quoted 5 lines
Quoted 5 lines

Unbinding host/gadget driver and rebinding with a different role should not be hard as the U-Boot core infrastructure exists.

Moreover if xhci reg and dev reg are separated into different nodes dont we still need access to OTG register block to switch b/w host and

device mode?

Quoted 1 line

I think we may not need to access OTG register if we add two subnodes for

gadget and host.

Quoted 1 line

But I see a for loop in dwc3_glue_bind() as follows, if there only one single

node representing both Host and Device node under usb wrapper node, why we need this for loop?

Quoted 7 lines
Quoted 2 lines

212 for (node = fdt_first_subnode(fdt, dev_of_offset(parent)); node > 0;
213 node = fdt_next_subnode(fdt, node)) {

I believe this is a legacy from the code it was inspired from.

Indeed the loop is not required.

Like Vignesh I think we must stick with the dt-bindings used by the kernel.

Thanks for your reply, I understand now.

Quoted 8 lines
Quoted 6 lines

BTW we are working on the USB3 support right now that is lacking in our tree. I choose to keep the PHY drivers as close as possible to their linux version. I'll probably start posting preliminary patches this week.

If you have the USB3 working in the kernel, can you point to a tree where I can have a look at the drivers and dt-bindings ?

Sure, you can see our downstream kernel code at my github, here is the link address: https://github.com/sherrysun1/linux-imx.git.
And look forward to seeing your patches in uboot maillist.

It will take some times before I can post on the mailing list because I did the work on top TI's u-boot. But you can find the patches under
https://github.com/jjhiblot/u-boot/tree/usb3_cleanup_v2.
<https://github.com/jjhiblot/u-boot/tree/usb3_cleanup_v2>

The DTS bindings are the same as under Linux. The CDNS3 driver is a port of the iteration v11 of the linux driver that has just been merged in the linux usb tree
(https://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb.git/log/?h=next)

Care has been taken to keep the changes with linux minimal to ease the
maintenance: the drivers are quite new and will probably evolve quite a bit in the next few months.

JJ

Quoted 64 lines

Best regards
Sherry sun

Quoted 60 lines

JJ

Quoted 42 lines

Best regards
Sherry sun

Quoted 39 lines

Regards
Vignesh

Quoted 36 lines

Best regards
Sherry sun

Quoted 33 lines

Regards
Vignesh

Quoted 30 lines

Best regards
Sherry sun

Quoted 27 lines

JJ

Quoted 1 line
Quoted 1 line

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 2 lines
Quoted 2 lines

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 20 lines
Quoted 2 lines

nodes.

Do you think I understand it right?
Best regards
Sherry sun

Quoted 14 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines
Quoted 2 lines

+ { }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://l

ists.d&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C7d1d76a7124f4c3f

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

e9

9d08d72d3ddf82%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63

Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

70276

16080089878&amp;sdata=70BPVQkomokcNpxsHRD3njfZQvuG51VSD1QKBjQ

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

o1kA%3

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

D&amp;reserved=0
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7 670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7 C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

--
Regards
Vignesh

author
#14 lore raw

PATCH v4 2/4 USB: host: Add the USB3 host driver

Hi Jean,

Quoted 118 lines

Hi Sherry,

Quoted 61 lines · Sherry Sun

Hi Jean,

Quoted 59 lines · Sherry Sun
Quoted 19 lines

Hi Vignesh,

Quoted 17 lines

Hi Sherry,

[...]

Quoted 3 lines
Quoted 3 lines

AFAIK, U-Boot does not support runtime switching of USB port to host from device and vice versa. This is the case for existing driver like

DWC3/MUSB etc.

Quoted 5 lines
Quoted 3 lines

Ideally we would need a role switch driver that unbinds and rebinds host vs device driver as when required based on U-Boot cmd or via an API (if required to be done during SPL stage etc).

I wonder if we can add two subnodes under the wrapper node as below,

one bind to usb gadget driver and another bind to usb host driver.

Quoted 2 lines

So if we want to use usb device, just call the gadget driver, and if want to

use usb host, just call the host driver. The driver will probe correspond

node.

Quoted 17 lines
Quoted 17 lines
Quoted 17 lines

I'm not sure if it is feasible, what do you think about it?

        usbss0: cdns_usb at 4104000 {
            compatible = "ti,j721e-usb";
            [....]
            usb0: usb at 6010000 {		/* xhci reg address*/
                compatible = "cdns,usb3-1.0.1";
	dr_mode = "host";
	[....]
	}
            usb1: usb at 6020000 {		/* dev reg address*/
                compatible = "cdns,usb3-1.0.1";
	dr_mode = "peripheral";
	[....]
	}
      }
Quoted 4 lines
Quoted 4 lines

But this is not how DT would look in kernel. There will be single DT node representing both Host and Device node. DT repesentation should not be changed based on OS/bootloader, its HW description and must

remain same.

Quoted 6 lines
Quoted 6 lines

Unbinding host/gadget driver and rebinding with a different role should not be hard as the U-Boot core infrastructure exists.

Moreover if xhci reg and dev reg are separated into different nodes dont we still need access to OTG register block to switch b/w host and

device mode?

Quoted 2 lines

I think we may not need to access OTG register if we add two subnodes for

gadget and host.

Quoted 2 lines

But I see a for loop in dwc3_glue_bind() as follows, if there only one single

node representing both Host and Device node under usb wrapper node, why we need this for loop?

Quoted 1 line

212 for (node = fdt_first_subnode(fdt, dev_of_offset(parent)); node >

0;

Quoted 9 lines
Quoted 6 lines
Quoted 1 line

213 node = fdt_next_subnode(fdt, node)) {

I believe this is a legacy from the code it was inspired from.

Indeed the loop is not required.

Like Vignesh I think we must stick with the dt-bindings used by the kernel.

Thanks for your reply, I understand now.

Quoted 1 line

BTW we are working on the USB3 support right now that is lacking in our

tree.

Quoted 6 lines
Quoted 5 lines

I choose to keep the PHY drivers as close as possible to their linux version. I'll probably start posting preliminary patches this week.

If you have the USB3 working in the kernel, can you point to a tree where I can have a look at the drivers and dt-bindings ?

Sure, you can see our downstream kernel code at my github, here is the

link address:
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub
.com%2Fsherrysun1%2Flinux- imx.git&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C368ba21c33c048 fd137208d735199870%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7 C637036256383501199&amp;sdata=y4AB0RpG%2F7Skk34yccpkW9TQsDWpi 6OJhtlRWid1DM0%3D&amp;reserved=0.

Quoted 1 line

And look forward to seeing your patches in uboot maillist.

It will take some times before I can post on the mailing list because I did the work on top TI's u-boot. But you can find the patches under
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub
.com%2Fjjhiblot%2Fu- boot%2Ftree%2Fusb3_cleanup_v2&amp;data=02%7C01%7Csherry.sun%40n xp.com%7C368ba21c33c048fd137208d735199870%7C686ea1d3bc2b4c6fa92c d99c5c301635%7C0%7C0%7C637036256383511193&amp;sdata=ozU9gYDTO %2FkoB4NwU1R2omGw9%2F%2Fyf6MPxBz1%2FtQ8Lgg%3D&amp;reserved=
0.
<https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithu
b.com%2Fjjhiblot%2Fu- boot%2Ftree%2Fusb3_cleanup_v2&amp;data=02%7C01%7Csherry.sun%40n xp.com%7C368ba21c33c048fd137208d735199870%7C686ea1d3bc2b4c6fa92c d99c5c301635%7C0%7C0%7C637036256383511193&amp;sdata=ozU9gYDTO %2FkoB4NwU1R2omGw9%2F%2Fyf6MPxBz1%2FtQ8Lgg%3D&amp;reserved=
0>

The DTS bindings are the same as under Linux. The CDNS3 driver is a port of the iteration v11 of the linux driver that has just been merged in the linux usb tree
(https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgit.ke
rnel.org%2Fpub%2Fscm%2Flinux%2Fkernel%2Fgit%2Fbalbi%2Fusb.git%2Flog %2F%3Fh%3Dnext&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C368b a21c33c048fd137208d735199870%7C686ea1d3bc2b4c6fa92cd99c5c301635% 7C0%7C0%7C637036256383511193&amp;sdata=OxjRaHtDmkJBJ41qwfOtf2Zj WACskvcAuUUPDRuYDcA%3D&amp;reserved=0)

Quoted 3 lines

Care has been taken to keep the changes with linux minimal to ease the
maintenance: the drivers are quite new and will probably evolve quite a bit in the next few months.

Thanks for your information, I will try to test your cdns3 patches on i.MX8 platform recently.

Best regards
Sherry Sun

Quoted 74 lines

JJ

Quoted 50 lines

Best regards
Sherry sun

Quoted 46 lines

JJ

Quoted 43 lines

Best regards
Sherry sun

Quoted 40 lines

Regards
Vignesh

Quoted 37 lines

Best regards
Sherry sun

Quoted 34 lines

Regards
Vignesh

Quoted 31 lines

Best regards
Sherry sun

Quoted 28 lines

JJ

Quoted 2 lines
Quoted 2 lines

Then when binding the wrapper node, it will hard-coded the two

subnodes

Quoted 2 lines
Quoted 2 lines

to different driver(gadge/host driver) according to the dr_mode property

in

Quoted 20 lines
Quoted 2 lines

nodes.

Do you think I understand it right?
Best regards
Sherry sun

Quoted 14 lines

Best regards
Sherry sun

Quoted 6 lines

JJ

Quoted 2 lines

+ { }
+};

_______________________________________________
U-Boot mailing list
U-Boot at lists.denx.de
https://l

ists.d&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C7d1d76a7124f4c3f

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

e9

9d08d72d3ddf82%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C63

Quoted 2 lines
Quoted 2 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

70276

16080089878&amp;sdata=70BPVQkomokcNpxsHRD3njfZQvuG51VSD1QKBjQ

Quoted 5 lines
Quoted 5 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines

o1kA%3

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

D&amp;reserved=0
enx.de%2Flistinfo%2Fu-

boot&amp;data=02%7C01%7Csherry.sun%40nxp.com%7C35f1d34da1ea4b7

Quoted 1 line
Quoted 1 line

670ba08d72b823e9a%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7

Quoted 1 line
Quoted 1 line

C637025710721487079&amp;sdata=Nfk0qWtSViz60wJHAOr2m5tgIwTWjNwI

Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line
Quoted 1 line
Quoted 1 line

GrNOxDH6HC0%3D&amp;reserved=0

--
Regards
Vignesh