usb: ehci-mx6: Enable OTG detection on imx8mm and imx8mn
Subsystems (2)
PATCH v3 1/1 usb: ehci-mx6: Enable OTG detection on imx8mm and imx8mn
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the dr_mode is defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.
The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the USBNC_PHY_STATUS_OFFSET register or it will hang.
Marek requested that I not enable the clocks in ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.
When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
---
V3: Keep ehci_usb_of_to_plat but add the ability to return
and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.
V2: Rename ehci_usb_of_to_plat to ehci_usb_dr_mode and call it
from the probe after the clocks are enabled, but before the data is needed.
drivers/usb/host/ehci-mx6.c +17 -4
@@ -543,7 +543,7 @@ static int ehci_usb_phy_mode(struct udevice *dev) plat->init_type =USB_INIT_DEVICE ; else plat->init_type = USB_INIT_HOST; - }else if (is_mx7 ()) { + }else if (is_mx7 () ||is_imx8mm () ||is_imx8mn ()) { phy_status = (void __iomem *)(addr +USBNC_PHY_STATUS_OFFSET ); val =readl (phy_status);
@@ -573,9 +573,8 @@ static int ehci_usb_of_to_plat(struct udevice *dev)case USB_DR_MODE_PERIPHERAL : plat->init_type =USB_INIT_DEVICE ;break ; -case USB_DR_MODE_OTG : -case USB_DR_MODE_UNKNOWN : -return ehci_usb_phy_mode (dev); +default : + plat->init_type =USB_INIT_UNKNOWN ; };return 0 ;
@@ -677,6 +676,20 @@ static int ehci_usb_probe(struct udevice *dev)mdelay (1 );#endif +/* +* If the device tree didn't specify host or device, +* the default is USB_INIT_UNKNOWN, so we need to check +* the register. For imx8mm and imx8mn, the clocks need to be +* running first, so we defer the check until they are. +*/ +if (priv->init_type ==USB_INIT_UNKNOWN ) { + ret =ehci_usb_phy_mode (dev); +if (ret) + goto err_clk; +else + priv->init_type = plat->init_type; + } +#if CONFIG_IS_ENABLED (DM_REGULATOR ) ret =device_get_supply_regulator (dev,"vbus-supply" , &priv->vbus_supply);
include/usb.h +2 -1
@@ -163,7 +163,8 @@ struct int_queue;*/ enum usb_init_type {USB_INIT_HOST , -USB_INIT_DEVICE +USB_INIT_DEVICE , +USB_INIT_UNKNOWN , };/**********************************************************************
Signature · 1 line
2.32.0
Quoted 92 lines · Adam Ford
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the
dr_modeis defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the
USBNC_PHY_STATUS_OFFSETregister or it will hang.Marek requested that I not enable the clocks in
ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
Signed-off-by Adam Ford <[email protected]>---
V3: Keepehci_usb_of_to_platbut add the ability to return
and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.V2: Rename
ehci_usb_of_to_plattoehci_usb_dr_modeand call it
from the probe after the clocks are enabled, but before the data is needed.drivers/usb/host/ehci-mx6.c +17
-4@@ -543,7 +543,7 @@ static int ehci_usb_phy_mode(struct udevice *dev) plat->init_type =USB_INIT_DEVICE ; else plat->init_type = USB_INIT_HOST; - }else if (is_mx7 ()) { + }else if (is_mx7 () ||is_imx8mm () ||is_imx8mn ()) { phy_status = (void __iomem *)(addr +USBNC_PHY_STATUS_OFFSET ); val =readl (phy_status);@@ -573,9 +573,8 @@ static int ehci_usb_of_to_plat(struct udevice *dev)case USB_DR_MODE_PERIPHERAL : plat->init_type =USB_INIT_DEVICE ;break ; -case USB_DR_MODE_OTG : -case USB_DR_MODE_UNKNOWN : -return ehci_usb_phy_mode (dev); +default : + plat->init_type =USB_INIT_UNKNOWN ; };return 0 ;@@ -677,6 +676,20 @@ static int ehci_usb_probe(struct udevice *dev)mdelay (1 );#endif +/* +* If the device tree didn't specify host or device, +* the default is USB_INIT_UNKNOWN, so we need to check +* the register. For imx8mm and imx8mn, the clocks need to be +* running first, so we defer the check until they are. +*/ +if (priv->init_type ==USB_INIT_UNKNOWN ) { + ret =ehci_usb_phy_mode (dev); +if (ret) + goto err_clk; +else + priv->init_type = plat->init_type; + } +#if CONFIG_IS_ENABLED (DM_REGULATOR ) ret =device_get_supply_regulator (dev,"vbus-supply" , &priv->vbus_supply);include/usb.h +2
-2@@ -163,7 +163,8 @@ struct int_queue;*/ enum usb_init_type {USB_INIT_HOST , -USB_INIT_DEVICE +USB_INIT_DEVICE , +USB_INIT_UNKNOWN , };/********************************************************************** --
Quoted 2 lines
2.32.0
Adam,
For v3:
tested on imx8mm_venice_defconfig as USB host with USB_ETHER_ASIX and as gadget with CMD_USB_MASS_STORAGE (ums 0 mmc 2).
Best Regards,
Tim
Hi Adam,
Quoted 18 lines · Adam Ford
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the
dr_modeis defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the
USBNC_PHY_STATUS_OFFSETregister or it will hang.Marek requested that I not enable the clocks in
ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
Quoted 8 lines
Signed-off-by Adam Ford <[email protected]>---
V3: Keepehci_usb_of_to_platbut add the ability to return
and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.
Unfortunately, v3 breaks 'ums 0 mmc 0' on the imx7s-warp board.
The eMMC is no longer mounted and the board hangs.
Quoted 31 lines · Fabio Estevam
Hi Adam,
Quoted 18 lines · Adam Ford
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the
dr_modeis defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the
USBNC_PHY_STATUS_OFFSETregister or it will hang.Marek requested that I not enable the clocks in
ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
Quoted 8 lines
Signed-off-by Adam Ford <[email protected]>---
V3: Keepehci_usb_of_to_platbut add the ability to return
and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.
Quoted 1 line
Unfortunately, v3 breaks 'ums 0 mmc 0' on the imx7s-warp board.
Thanks for testing it.
I am not really sure what's significantly different between them. Do you get any errors when you run UMS?
Quoted 2 lines
The eMMC is no longer mounted and the board hangs.
Hi Adam,
Quoted 4 lines · Adam Ford
Thanks for testing it.
I am not really sure what's significantly different between them. Do you get any errors when you run UMS?
Just realized the problem I am seeing is unrelated to your patch. I will need to debug it.
Quoted 61 lines · Adam Ford
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the
dr_modeis defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the
USBNC_PHY_STATUS_OFFSETregister or it will hang.Marek requested that I not enable the clocks in
ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
Signed-off-by Adam Ford <[email protected]>---
V3: Keepehci_usb_of_to_platbut add the ability to return and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.V2: Rename
ehci_usb_of_to_plattoehci_usb_dr_modeand call it from the probe after the clocks are enabled, but before the data is needed.drivers/usb/host/ehci-mx6.c +0
-0@@ -543,7 +543,7 @@ static int ehci_usb_phy_mode(struct udevice *dev) plat->init_type = USB_INIT_DEVICE;else
plat->init_type =USB_INIT_HOST;
- } else if (is_mx7()) {
+ } else if (is_mx7()||is_imx8mm()||is_imx8mn()) {phy_status= (void__iomem*)(addr +USBNC_PHY_STATUS_OFFSET); val =readl(phy_status);drivers/usb/host/ehci-mx6.c +0
-0@@ -573,9 +573,8 @@ static int ehci_usb_of_to_plat(struct udevice *dev)case
USB_DR_MODE_PERIPHERAL:
plat->init_type =USB_INIT_DEVICE; break;
- caseUSB_DR_MODE_OTG:
- caseUSB_DR_MODE_UNKNOWN:
- returnehci_usb_phy_mode(dev); + default:
+ plat->init_type =USB_INIT_UNKNOWN; };
return 0;drivers/usb/host/ehci-mx6.c +0
-0@@ -677,6 +676,20 @@ static int ehci_usb_probe(struct udevice *dev) mdelay(1); #endif
+ /*
+ * If the device tree didn't specify host or device,
Quoted 8 lines
+ * the default is
USB_INIT_UNKNOWN, so we need to check + * the register. For imx8mm and imx8mn, the clocks need to be + * running first, so we defer the check until they are. + */
+ if (priv->init_type==USB_INIT_UNKNOWN) {
+ ret =ehci_usb_phy_mode(dev);
+ if (ret)
+ gotoerr_clk;
Hm, unfortunately, that label is gated by an ifdef leading to the following on colibri_imx7:
drivers/usb/host/ehci-mx6.c: In function ‘ehci_usb_probe’:drivers/usb/host/ehci-mx6.c:688:4: error: label ‘err_clk’ used but not defined
688 | goto err_clk;
| ^~~~make[1]: *** [scripts/Makefile.build:254: drivers/usb/host/ehci-mx6.o] Error 1
I have to admit that maybe we should indeed run it with DM_REGULATOR but that won't change anything on this issue. I guess one should really not have any gotos to a label err_clk being ifdefed from somewhere not ifdefed.
Quoted 21 lines
+ else
+ priv->init_type =plat->init_type; + }
+
#ifCONFIG_IS_ENABLED(DM_REGULATOR)
ret =device_get_supply_regulator(dev, "vbus-supply", &priv->vbus_supply);include/usb.h +2
-1@@ -163,7 +163,8 @@ struct int_queue;*/ enum usb_init_type { USB_INIT_HOST, - USB_INIT_DEVICE + USB_INIT_DEVICE, + USB_INIT_UNKNOWN, };/**********************************************************************
Quoted 61 lines · Adam Ford
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the
dr_modeis defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the
USBNC_PHY_STATUS_OFFSETregister or it will hang.Marek requested that I not enable the clocks in
ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
Signed-off-by Adam Ford <[email protected]>---
V3: Keepehci_usb_of_to_platbut add the ability to return and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.V2: Rename
ehci_usb_of_to_plattoehci_usb_dr_modeand call it from the probe after the clocks are enabled, but before the data is needed.drivers/usb/host/ehci-mx6.c +0
-0@@ -543,7 +543,7 @@ static int ehci_usb_phy_mode(struct udevice *dev) plat->init_type = USB_INIT_DEVICE;else
plat->init_type =USB_INIT_HOST;
- } else if (is_mx7()) {
+ } else if (is_mx7()||is_imx8mm()||is_imx8mn()) {phy_status= (void__iomem*)(addr +USBNC_PHY_STATUS_OFFSET); val =readl(phy_status);drivers/usb/host/ehci-mx6.c +0
-0@@ -573,9 +573,8 @@ static int ehci_usb_of_to_plat(struct udevice *dev)case
USB_DR_MODE_PERIPHERAL:
plat->init_type =USB_INIT_DEVICE; break;
- caseUSB_DR_MODE_OTG:
- caseUSB_DR_MODE_UNKNOWN:
- returnehci_usb_phy_mode(dev); + default:
+ plat->init_type =USB_INIT_UNKNOWN; };
return 0;drivers/usb/host/ehci-mx6.c +0
-0@@ -677,6 +676,20 @@ static int ehci_usb_probe(struct udevice *dev) mdelay(1); #endif
+ /*
+ * If the device tree didn't specify host or device,
Quoted 8 lines
+ * the default is
USB_INIT_UNKNOWN, so we need to check + * the register. For imx8mm and imx8mn, the clocks need to be + * running first, so we defer the check until they are. + */
+ if (priv->init_type==USB_INIT_UNKNOWN) {
+ ret =ehci_usb_phy_mode(dev);
+ if (ret)
+ gotoerr_clk;
Hm, unfortunately, that label is gated by an ifdef leading to the following on colibri_imx7:
drivers/usb/host/ehci-mx6.c: In function ‘ehci_usb_probe’:drivers/usb/host/ehci-mx6.c:688:4: error: label ‘err_clk’ used but not defined
688 | goto err_clk;
| ^~~~make[1]: *** [scripts/Makefile.build:254: drivers/usb/host/ehci-mx6.o] Error 1
I have to admit that maybe we should indeed run it with DM_REGULATOR but that won't change anything on this issue. I guess one should really not have any gotos to a label err_clk being ifdefed from somewhere not ifdefed.
Quoted 21 lines
+ else
+ priv->init_type =plat->init_type; + }
+
#ifCONFIG_IS_ENABLED(DM_REGULATOR)
ret =device_get_supply_regulator(dev, "vbus-supply", &priv->vbus_supply);include/usb.h +2
-1@@ -163,7 +163,8 @@ struct int_queue;*/ enum usb_init_type { USB_INIT_HOST, - USB_INIT_DEVICE + USB_INIT_DEVICE, + USB_INIT_UNKNOWN, };/**********************************************************************
Quoted 76 lines · Marcel Ziswiler
Quoted 59 lines · Adam Ford
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the
dr_modeis defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the
USBNC_PHY_STATUS_OFFSETregister or it will hang.Marek requested that I not enable the clocks in
ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
Signed-off-by Adam Ford <[email protected]>---
V3: Keepehci_usb_of_to_platbut add the ability to return
and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.V2: Rename
ehci_usb_of_to_plattoehci_usb_dr_modeand call it
from the probe after the clocks are enabled, but before the data is needed.drivers/usb/host/ehci-mx6.c +3
-4@@ -543,7 +543,7 @@ static int ehci_usb_phy_mode(struct udevice *dev) plat->init_type =USB_INIT_DEVICE ; else plat->init_type = USB_INIT_HOST; - }else if (is_mx7 ()) { + }else if (is_mx7 () ||is_imx8mm () ||is_imx8mn ()) { phy_status = (void __iomem *)(addr +USBNC_PHY_STATUS_OFFSET ); val =readl (phy_status);@@ -573,9 +573,8 @@ static int ehci_usb_of_to_plat(struct udevice *dev)case USB_DR_MODE_PERIPHERAL : plat->init_type =USB_INIT_DEVICE ;break ; -case USB_DR_MODE_OTG : -case USB_DR_MODE_UNKNOWN : -return ehci_usb_phy_mode (dev); +default : + plat->init_type =USB_INIT_UNKNOWN ; };return 0 ;@@ -677,6 +676,20 @@ static int ehci_usb_probe(struct udevice *dev)mdelay (1 );#endif drivers/usb/host/ehci-mx6.c static int ehci_usb_probe(struct udevice *dev) · line 688
Quoted 10 lines
drivers/usb/host/ehci-mx6.c +10
-0+/* +* If the device tree didn't specify host or device, +* the default is USB_INIT_UNKNOWN, so we need to check +* the register. For imx8mm and imx8mn, the clocks need to be +* running first, so we defer the check until they are. +*/ +if (priv->init_type ==USB_INIT_UNKNOWN ) { + ret =ehci_usb_phy_mode (dev); +if (ret) + goto err_clk;Hm, unfortunately, that label is gated by an ifdef leading to the following on
colibri_imx7:
drivers/usb/host/ehci-mx6.c: In function ‘ehci_usb_probe’:drivers/usb/host/ehci-mx6.c:688:4: error: label ‘err_clk’ used but not defined
Quoted 8 lines
688 | goto err_clk; | ^~~~make[1]: *** [
scripts/Makefile.build:254:drivers/usb/host/ehci-mx6.o] Error 1I have to admit that maybe we should indeed run it with
DM_REGULATORbut that won't change anything on this issue. I guess one should really not have any gotos to a labelerr_clkbeing ifdefed from somewhere not ifdefed.
Adam,
Can you re-submit this with an ifdef to handle the correct goto when DM_REGULATOR is undefined?
It looks like Fabio figured out his issue with his board and it was unrelated to this (fixed by 'mmc: fsl_esdhc_imx: fix watermark level in dma') so I think Marcel's point is the only blocker on this patch unless there is more discussion needed with Michael or some feedback needed from Marek?
Best regards,
Tim
Quoted 90 lines · Tim Harvey
Quoted 76 lines · Marcel Ziswiler
Quoted 59 lines · Adam Ford
The imx8mm and imx8mn appear compatible with imx7d-usb flags in the OTG driver. If the
dr_modeis defined as host or peripheral, the device appears to operate correctly, however the auto host/peripheral detection results in an error.The solution isn't just adding checks for imx8mm and imx8mn to the check for imx7, because the USB clock needs to be running to read from the
USBNC_PHY_STATUS_OFFSETregister or it will hang.Marek requested that I not enable the clocks in
ehci_usb_of_to_plat, so I modified that function to return an unknown state if the device tree does not explicitly state whether it is a host or a peripheral.When the driver probes, it looks to see if it's in the unknown state, and only then will it read the register to auto-detect.
Signed-off-by Adam Ford <[email protected]>---
V3: Keepehci_usb_of_to_platbut add the ability to return
and unknown state instead of reading the register. If the probe determines the states is unknown, it will query the register after the clocks have been enabled. Because of the slight behavior change, I removed any review or tested tags.V2: Rename
ehci_usb_of_to_plattoehci_usb_dr_modeand call it
from the probe after the clocks are enabled, but before the data is needed.drivers/usb/host/ehci-mx6.c +3
-4@@ -543,7 +543,7 @@ static int ehci_usb_phy_mode(struct udevice *dev) plat->init_type =USB_INIT_DEVICE ; else plat->init_type = USB_INIT_HOST; - }else if (is_mx7 ()) { + }else if (is_mx7 () ||is_imx8mm () ||is_imx8mn ()) { phy_status = (void __iomem *)(addr +USBNC_PHY_STATUS_OFFSET ); val =readl (phy_status);@@ -573,9 +573,8 @@ static int ehci_usb_of_to_plat(struct udevice *dev)case USB_DR_MODE_PERIPHERAL : plat->init_type =USB_INIT_DEVICE ;break ; -case USB_DR_MODE_OTG : -case USB_DR_MODE_UNKNOWN : -return ehci_usb_phy_mode (dev); +default : + plat->init_type =USB_INIT_UNKNOWN ; };return 0 ;@@ -677,6 +676,20 @@ static int ehci_usb_probe(struct udevice *dev)mdelay (1 );#endif drivers/usb/host/ehci-mx6.c static int ehci_usb_probe(struct udevice *dev) · line 688
Quoted 10 lines
drivers/usb/host/ehci-mx6.c +10
-0+/* +* If the device tree didn't specify host or device, +* the default is USB_INIT_UNKNOWN, so we need to check +* the register. For imx8mm and imx8mn, the clocks need to be +* running first, so we defer the check until they are. +*/ +if (priv->init_type ==USB_INIT_UNKNOWN ) { + ret =ehci_usb_phy_mode (dev); +if (ret) + goto err_clk;Hm, unfortunately, that label is gated by an ifdef leading to the following on
colibri_imx7:
drivers/usb/host/ehci-mx6.c: In function ‘ehci_usb_probe’:drivers/usb/host/ehci-mx6.c:688:4: error: label ‘err_clk’ used but not definedQuoted 8 lines
688 | goto err_clk; | ^~~~make[1]: *** [
scripts/Makefile.build:254:drivers/usb/host/ehci-mx6.o] Error 1I have to admit that maybe we should indeed run it with
DM_REGULATORbut that won't change anything on this issue. I guess one should really not have any gotos to a labelerr_clkbeing ifdefed from somewhere not ifdefed.Adam,
Quoted 8 lines
Can you re-submit this with an ifdef to handle the correct goto when
DM_REGULATORis undefined?It looks like Fabio figured out his issue with his board and it was unrelated to this (fixed by 'mmc:
fsl_esdhc_imx: fix watermark level in dma') so I think Marcel's point is the only blocker on this patch unless there is more discussion needed with Michael or some feedback needed from Marek?
Tim,
Not a problem. I was waiting for feedback from Marek in case there were more issues, but I can fix the goto reference and resubmit. I'm build-testing the colibri_imx7 now.
I'm just moving the 'err_clk' reference out of the #ifdef so the same reference can be used. It doesn't look like it should be inside the #ifdef anyway.
You should see something shortly, and I'll add you to the CC list.
adam
Quoted 4 lines
Best regards,
Tim