diff mbox

[RFC,06/10] usb: xhci: Add Tegra XHCI host-controller driver

Message ID 1400113986-339-7-git-send-email-abrestic@chromium.org
State Superseded, archived
Headers show

Commit Message

Andrew Bresticker May 15, 2014, 12:33 a.m. UTC
Add support for the on-chip XHCI host controller present on NVIDIA
Tegra114 and later SoCs.

The driver is currently very basic: it loads the controller with its
firmware, starts the controller, and is able to service messages sent
by the controller's firmware.  The hardware supports device mode as
well as runtime power-gating, but support for these is not yet
implemented here.

Based on work by:
  Ajay Gupta <ajayg@nvidia.com>
  Bharath Yadav <byadav@nvidia.com>

Signed-off-by: Andrew Bresticker <abrestic@chromium.org>
---
 .../devicetree/bindings/usb/nvidia,tegra-xhci.txt  |  42 ++
 drivers/usb/host/Kconfig                           |  12 +
 drivers/usb/host/Makefile                          |   2 +
 drivers/usb/host/xhci-tegra.c                      | 796 +++++++++++++++++++++
 drivers/usb/host/xhci-tegra.h                      | 131 ++++
 include/linux/usb/tegra_xusb.h                     |  66 ++
 6 files changed, 1049 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/usb/nvidia,tegra-xhci.txt
 create mode 100644 drivers/usb/host/xhci-tegra.c
 create mode 100644 drivers/usb/host/xhci-tegra.h
 create mode 100644 include/linux/usb/tegra_xusb.h

Comments

Arnd Bergmann May 15, 2014, 8:17 a.m. UTC | #1
On Wednesday 14 May 2014 17:33:02 Andrew Bresticker wrote:
> +
> +int tegra_xhci_register_mbox_notifier(struct notifier_block *nb)
> +{
> +	int ret;
> +
> +	mutex_lock(&tegra_xhci_mbox_lock);
> +	ret = raw_notifier_chain_register(&tegra_xhci_mbox_notifiers, nb);
> +	mutex_unlock(&tegra_xhci_mbox_lock);
> +
> +	return ret;
> +}
> +EXPORT_SYMBOL(tegra_xhci_register_mbox_notifier);
> +
> +void tegra_xhci_unregister_mbox_notifier(struct notifier_block *nb)
> +{
> +	mutex_lock(&tegra_xhci_mbox_lock);
> +	raw_notifier_chain_unregister(&tegra_xhci_mbox_notifiers, nb);
> +	mutex_unlock(&tegra_xhci_mbox_lock);
> +}
> +EXPORT_SYMBOL(tegra_xhci_unregister_mbox_notifier);

What driver would use these?

My feeling is that if you have a mailbox that is used by multiple
drivers, you should use a proper mailbox driver to operate them,
and have the drivers register with that API instead of a custom one.

> +	/* Create child xhci-plat device */
> +	memset(xhci_resources, 0, sizeof(xhci_resources));
> +	res = platform_get_resource(to_platform_device(dev), IORESOURCE_IRQ, 0);
> +	if (!res) {
> +		dev_err(dev, "Missing XHCI IRQ\n");
> +		ret = -ENODEV;
> +		goto out;
> +	}
> +	xhci_resources[0].start = res->start;
> +	xhci_resources[0].end = res->end;
> +	xhci_resources[0].flags = res->flags;
> +	xhci_resources[0].name = res->name;
> +	res = platform_get_resource(to_platform_device(dev), IORESOURCE_MEM, 0);
> +	if (!res) {
> +		dev_err(dev, "Missing XHCI registers\n");
> +		ret = -ENODEV;
> +		goto out;
> +	}
> +	xhci_resources[1].start = res->start;
> +	xhci_resources[1].end = res->end;
> +	xhci_resources[1].flags = res->flags;
> +	xhci_resources[1].name = res->name;
> +
> +	xhci = platform_device_alloc("xhci-hcd", PLATFORM_DEVID_AUTO);
> +	if (!xhci) {
> +		dev_err(dev, "Failed to allocate XHCI host\n");
> +		ret = -ENOMEM;
> +		goto out;
> +	}

This does not feel appropriate at all: Rather than creating a child device,
you should have a specific driver that hooks into functions exported
by the xhci core. See Documentation/driver-model/design-patterns.txt

	Arnd

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Thierry Reding May 15, 2014, 9:19 a.m. UTC | #2
On Thu, May 15, 2014 at 10:17:10AM +0200, Arnd Bergmann wrote:
> On Wednesday 14 May 2014 17:33:02 Andrew Bresticker wrote:
[...]
> > +	/* Create child xhci-plat device */
> > +	memset(xhci_resources, 0, sizeof(xhci_resources));
> > +	res = platform_get_resource(to_platform_device(dev), IORESOURCE_IRQ, 0);
> > +	if (!res) {
> > +		dev_err(dev, "Missing XHCI IRQ\n");
> > +		ret = -ENODEV;
> > +		goto out;
> > +	}
> > +	xhci_resources[0].start = res->start;
> > +	xhci_resources[0].end = res->end;
> > +	xhci_resources[0].flags = res->flags;
> > +	xhci_resources[0].name = res->name;
> > +	res = platform_get_resource(to_platform_device(dev), IORESOURCE_MEM, 0);
> > +	if (!res) {
> > +		dev_err(dev, "Missing XHCI registers\n");
> > +		ret = -ENODEV;
> > +		goto out;
> > +	}
> > +	xhci_resources[1].start = res->start;
> > +	xhci_resources[1].end = res->end;
> > +	xhci_resources[1].flags = res->flags;
> > +	xhci_resources[1].name = res->name;
> > +
> > +	xhci = platform_device_alloc("xhci-hcd", PLATFORM_DEVID_AUTO);
> > +	if (!xhci) {
> > +		dev_err(dev, "Failed to allocate XHCI host\n");
> > +		ret = -ENOMEM;
> > +		goto out;
> > +	}
> 
> This does not feel appropriate at all: Rather than creating a child device,
> you should have a specific driver that hooks into functions exported
> by the xhci core. See Documentation/driver-model/design-patterns.txt

I don't think Documentation/driver-model/design-patterns.txt documents
this. Perhaps this is what you had in mind?

	http://lwn.net/Articles/336262/

Thierry
Arnd Bergmann May 15, 2014, 1:30 p.m. UTC | #3
On Thursday 15 May 2014 11:19:40 Thierry Reding wrote:
> > > +
> > > +   xhci = platform_device_alloc("xhci-hcd", PLATFORM_DEVID_AUTO);
> > > +   if (!xhci) {
> > > +           dev_err(dev, "Failed to allocate XHCI host\n");
> > > +           ret = -ENOMEM;
> > > +           goto out;
> > > +   }
> > 
> > This does not feel appropriate at all: Rather than creating a child device,
> > you should have a specific driver that hooks into functions exported
> > by the xhci core. See Documentation/driver-model/design-patterns.txt
> 
> I don't think Documentation/driver-model/design-patterns.txt documents
> this. Perhaps this is what you had in mind?
> 
>         http://lwn.net/Articles/336262/

No, I did mean Documentation/driver-model/design-patterns.txt.

The pattern used in the proposed driver is to create the generic platform_device
as the child of the more specific platform_device.

The normal pattern is to have only one device and embed the generic structure
inside of the more specific structure and use container_of to cast between
the two as needed.

	Arnd
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Andrew Bresticker May 15, 2014, 8:18 p.m. UTC | #4
Arnd,

On Thu, May 15, 2014 at 1:17 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> On Wednesday 14 May 2014 17:33:02 Andrew Bresticker wrote:
>> +
>> +int tegra_xhci_register_mbox_notifier(struct notifier_block *nb)
>> +{
>> +     int ret;
>> +
>> +     mutex_lock(&tegra_xhci_mbox_lock);
>> +     ret = raw_notifier_chain_register(&tegra_xhci_mbox_notifiers, nb);
>> +     mutex_unlock(&tegra_xhci_mbox_lock);
>> +
>> +     return ret;
>> +}
>> +EXPORT_SYMBOL(tegra_xhci_register_mbox_notifier);
>> +
>> +void tegra_xhci_unregister_mbox_notifier(struct notifier_block *nb)
>> +{
>> +     mutex_lock(&tegra_xhci_mbox_lock);
>> +     raw_notifier_chain_unregister(&tegra_xhci_mbox_notifiers, nb);
>> +     mutex_unlock(&tegra_xhci_mbox_lock);
>> +}
>> +EXPORT_SYMBOL(tegra_xhci_unregister_mbox_notifier);
>
> What driver would use these?

It's used by just this driver (the host) and the PHY driver (next
patch in series).

> My feeling is that if you have a mailbox that is used by multiple
> drivers, you should use a proper mailbox driver to operate them,
> and have the drivers register with that API instead of a custom one.

Ok, will do.

>> +     /* Create child xhci-plat device */
>> +     memset(xhci_resources, 0, sizeof(xhci_resources));
>> +     res = platform_get_resource(to_platform_device(dev), IORESOURCE_IRQ, 0);
>> +     if (!res) {
>> +             dev_err(dev, "Missing XHCI IRQ\n");
>> +             ret = -ENODEV;
>> +             goto out;
>> +     }
>> +     xhci_resources[0].start = res->start;
>> +     xhci_resources[0].end = res->end;
>> +     xhci_resources[0].flags = res->flags;
>> +     xhci_resources[0].name = res->name;
>> +     res = platform_get_resource(to_platform_device(dev), IORESOURCE_MEM, 0);
>> +     if (!res) {
>> +             dev_err(dev, "Missing XHCI registers\n");
>> +             ret = -ENODEV;
>> +             goto out;
>> +     }
>> +     xhci_resources[1].start = res->start;
>> +     xhci_resources[1].end = res->end;
>> +     xhci_resources[1].flags = res->flags;
>> +     xhci_resources[1].name = res->name;
>> +
>> +     xhci = platform_device_alloc("xhci-hcd", PLATFORM_DEVID_AUTO);
>> +     if (!xhci) {
>> +             dev_err(dev, "Failed to allocate XHCI host\n");
>> +             ret = -ENOMEM;
>> +             goto out;
>> +     }
>
> This does not feel appropriate at all: Rather than creating a child device,
> you should have a specific driver that hooks into functions exported
> by the xhci core. See Documentation/driver-model/design-patterns.txt

This is how DWC3, currently the only in-tree non-PCI XHCI host driver,
is structured - see drivers/usb/dwc3/host.c.  The recently proposed
Armada XHCI driver [1] just adds clock support and a hook in
xhci-plat's probe() to do the platform-specific initialization.
Tegra's XHCI driver initialization is quite a bit more complicated,
mainly due to the need for external firmware and specific ordering
(e.g. firmware messages should only be enabled after the HCD is
created).  I could do away with the xhci-plat sub-device and just
create a Tegra hc_driver, but it seems silly to have three XHCI
platform drivers structured in three different ways.  USB folks, do
you have an opinion on how this should be done?

[1] https://lkml.org/lkml/2014/5/15/208

Thanks,
Andrew
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Alan Stern May 15, 2014, 9:16 p.m. UTC | #5
On Thu, 15 May 2014, Andrew Bresticker wrote:

> > This does not feel appropriate at all: Rather than creating a child device,
> > you should have a specific driver that hooks into functions exported
> > by the xhci core. See Documentation/driver-model/design-patterns.txt
> 
> This is how DWC3, currently the only in-tree non-PCI XHCI host driver,
> is structured - see drivers/usb/dwc3/host.c.  The recently proposed
> Armada XHCI driver [1] just adds clock support and a hook in
> xhci-plat's probe() to do the platform-specific initialization.
> Tegra's XHCI driver initialization is quite a bit more complicated,
> mainly due to the need for external firmware and specific ordering
> (e.g. firmware messages should only be enabled after the HCD is
> created).  I could do away with the xhci-plat sub-device and just
> create a Tegra hc_driver, but it seems silly to have three XHCI
> platform drivers structured in three different ways.  USB folks, do
> you have an opinion on how this should be done?

Felipe and I have differing opinions.  I prefer not to create an extra 
device level, and he does (that's why the non-PCI EHCI drivers don't do 
it and the dwc3 driver does).  I consider Felipe's approach to be an 
example of the "Midlayer Mistake" anti-pattern from the LWN 
article mentioned earlier in this thread -- I don't know what he thinks 
about my approach.

Anyway, unless I'm totally mistaken, you wouldn't end up with three
platform drivers structured in three different ways.  Either way, your
new driver would be structured like one of the existing two.

Alan Stern

--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Thierry Reding May 15, 2014, 9:18 p.m. UTC | #6
On Thu, May 15, 2014 at 01:18:22PM -0700, Andrew Bresticker wrote:
> Arnd,
> 
> On Thu, May 15, 2014 at 1:17 AM, Arnd Bergmann <arnd@arndb.de> wrote:
> > On Wednesday 14 May 2014 17:33:02 Andrew Bresticker wrote:
> >> +
> >> +int tegra_xhci_register_mbox_notifier(struct notifier_block *nb)
> >> +{
> >> +     int ret;
> >> +
> >> +     mutex_lock(&tegra_xhci_mbox_lock);
> >> +     ret = raw_notifier_chain_register(&tegra_xhci_mbox_notifiers, nb);
> >> +     mutex_unlock(&tegra_xhci_mbox_lock);
> >> +
> >> +     return ret;
> >> +}
> >> +EXPORT_SYMBOL(tegra_xhci_register_mbox_notifier);
> >> +
> >> +void tegra_xhci_unregister_mbox_notifier(struct notifier_block *nb)
> >> +{
> >> +     mutex_lock(&tegra_xhci_mbox_lock);
> >> +     raw_notifier_chain_unregister(&tegra_xhci_mbox_notifiers, nb);
> >> +     mutex_unlock(&tegra_xhci_mbox_lock);
> >> +}
> >> +EXPORT_SYMBOL(tegra_xhci_unregister_mbox_notifier);
> >
> > What driver would use these?
> 
> It's used by just this driver (the host) and the PHY driver (next
> patch in series).
> 
> > My feeling is that if you have a mailbox that is used by multiple
> > drivers, you should use a proper mailbox driver to operate them,
> > and have the drivers register with that API instead of a custom one.
> 
> Ok, will do.
> 
> >> +     /* Create child xhci-plat device */
> >> +     memset(xhci_resources, 0, sizeof(xhci_resources));
> >> +     res = platform_get_resource(to_platform_device(dev), IORESOURCE_IRQ, 0);
> >> +     if (!res) {
> >> +             dev_err(dev, "Missing XHCI IRQ\n");
> >> +             ret = -ENODEV;
> >> +             goto out;
> >> +     }
> >> +     xhci_resources[0].start = res->start;
> >> +     xhci_resources[0].end = res->end;
> >> +     xhci_resources[0].flags = res->flags;
> >> +     xhci_resources[0].name = res->name;
> >> +     res = platform_get_resource(to_platform_device(dev), IORESOURCE_MEM, 0);
> >> +     if (!res) {
> >> +             dev_err(dev, "Missing XHCI registers\n");
> >> +             ret = -ENODEV;
> >> +             goto out;
> >> +     }
> >> +     xhci_resources[1].start = res->start;
> >> +     xhci_resources[1].end = res->end;
> >> +     xhci_resources[1].flags = res->flags;
> >> +     xhci_resources[1].name = res->name;
> >> +
> >> +     xhci = platform_device_alloc("xhci-hcd", PLATFORM_DEVID_AUTO);
> >> +     if (!xhci) {
> >> +             dev_err(dev, "Failed to allocate XHCI host\n");
> >> +             ret = -ENOMEM;
> >> +             goto out;
> >> +     }
> >
> > This does not feel appropriate at all: Rather than creating a child device,
> > you should have a specific driver that hooks into functions exported
> > by the xhci core. See Documentation/driver-model/design-patterns.txt
> 
> This is how DWC3, currently the only in-tree non-PCI XHCI host driver,
> is structured - see drivers/usb/dwc3/host.c.  The recently proposed
> Armada XHCI driver [1] just adds clock support and a hook in
> xhci-plat's probe() to do the platform-specific initialization.

Ugh... that very much sounds like the midlayer mistake. Doing that is
not going to scale in the long run. Everybody will just keep adding
quirks to the initialization until it's become a huge mess.

> Tegra's XHCI driver initialization is quite a bit more complicated,
> mainly due to the need for external firmware and specific ordering
> (e.g. firmware messages should only be enabled after the HCD is
> created).  I could do away with the xhci-plat sub-device and just
> create a Tegra hc_driver, but it seems silly to have three XHCI
> platform drivers structured in three different ways.  USB folks, do
> you have an opinion on how this should be done?

The tendency in other subsystems (and I think this is also true to some
degree for USB, though I'm less familiar with it) is to make common
functions available as a library of helpers so that other drivers can
use them (either directly or by wrapping them in platform-specific
implementations). That allows you to keep the platform-specific code
where it belongs: in the platform-specific driver.

Thierry
Andrew Bresticker May 16, 2014, 4:52 p.m. UTC | #7
On Thu, May 15, 2014 at 2:18 PM, Thierry Reding
<thierry.reding@gmail.com> wrote:
> On Thu, May 15, 2014 at 01:18:22PM -0700, Andrew Bresticker wrote:
>> Arnd,
>>
>> On Thu, May 15, 2014 at 1:17 AM, Arnd Bergmann <arnd@arndb.de> wrote:
>> > On Wednesday 14 May 2014 17:33:02 Andrew Bresticker wrote:
>> >> +
>> >> +int tegra_xhci_register_mbox_notifier(struct notifier_block *nb)
>> >> +{
>> >> +     int ret;
>> >> +
>> >> +     mutex_lock(&tegra_xhci_mbox_lock);
>> >> +     ret = raw_notifier_chain_register(&tegra_xhci_mbox_notifiers, nb);
>> >> +     mutex_unlock(&tegra_xhci_mbox_lock);
>> >> +
>> >> +     return ret;
>> >> +}
>> >> +EXPORT_SYMBOL(tegra_xhci_register_mbox_notifier);
>> >> +
>> >> +void tegra_xhci_unregister_mbox_notifier(struct notifier_block *nb)
>> >> +{
>> >> +     mutex_lock(&tegra_xhci_mbox_lock);
>> >> +     raw_notifier_chain_unregister(&tegra_xhci_mbox_notifiers, nb);
>> >> +     mutex_unlock(&tegra_xhci_mbox_lock);
>> >> +}
>> >> +EXPORT_SYMBOL(tegra_xhci_unregister_mbox_notifier);
>> >
>> > What driver would use these?
>>
>> It's used by just this driver (the host) and the PHY driver (next
>> patch in series).
>>
>> > My feeling is that if you have a mailbox that is used by multiple
>> > drivers, you should use a proper mailbox driver to operate them,
>> > and have the drivers register with that API instead of a custom one.
>>
>> Ok, will do.
>>
>> >> +     /* Create child xhci-plat device */
>> >> +     memset(xhci_resources, 0, sizeof(xhci_resources));
>> >> +     res = platform_get_resource(to_platform_device(dev), IORESOURCE_IRQ, 0);
>> >> +     if (!res) {
>> >> +             dev_err(dev, "Missing XHCI IRQ\n");
>> >> +             ret = -ENODEV;
>> >> +             goto out;
>> >> +     }
>> >> +     xhci_resources[0].start = res->start;
>> >> +     xhci_resources[0].end = res->end;
>> >> +     xhci_resources[0].flags = res->flags;
>> >> +     xhci_resources[0].name = res->name;
>> >> +     res = platform_get_resource(to_platform_device(dev), IORESOURCE_MEM, 0);
>> >> +     if (!res) {
>> >> +             dev_err(dev, "Missing XHCI registers\n");
>> >> +             ret = -ENODEV;
>> >> +             goto out;
>> >> +     }
>> >> +     xhci_resources[1].start = res->start;
>> >> +     xhci_resources[1].end = res->end;
>> >> +     xhci_resources[1].flags = res->flags;
>> >> +     xhci_resources[1].name = res->name;
>> >> +
>> >> +     xhci = platform_device_alloc("xhci-hcd", PLATFORM_DEVID_AUTO);
>> >> +     if (!xhci) {
>> >> +             dev_err(dev, "Failed to allocate XHCI host\n");
>> >> +             ret = -ENOMEM;
>> >> +             goto out;
>> >> +     }
>> >
>> > This does not feel appropriate at all: Rather than creating a child device,
>> > you should have a specific driver that hooks into functions exported
>> > by the xhci core. See Documentation/driver-model/design-patterns.txt
>>
>> This is how DWC3, currently the only in-tree non-PCI XHCI host driver,
>> is structured - see drivers/usb/dwc3/host.c.  The recently proposed
>> Armada XHCI driver [1] just adds clock support and a hook in
>> xhci-plat's probe() to do the platform-specific initialization.
>
> Ugh... that very much sounds like the midlayer mistake. Doing that is
> not going to scale in the long run. Everybody will just keep adding
> quirks to the initialization until it's become a huge mess.

Agreed - I didn't even bother trying to write the Tegra driver like that.

>> Tegra's XHCI driver initialization is quite a bit more complicated,
>> mainly due to the need for external firmware and specific ordering
>> (e.g. firmware messages should only be enabled after the HCD is
>> created).  I could do away with the xhci-plat sub-device and just
>> create a Tegra hc_driver, but it seems silly to have three XHCI
>> platform drivers structured in three different ways.  USB folks, do
>> you have an opinion on how this should be done?
>
> The tendency in other subsystems (and I think this is also true to some
> degree for USB, though I'm less familiar with it) is to make common
> functions available as a library of helpers so that other drivers can
> use them (either directly or by wrapping them in platform-specific
> implementations). That allows you to keep the platform-specific code
> where it belongs: in the platform-specific driver.

Yes, this is how the non-PCI EHCI host drivers are written.  I'll
structure it more like that for the next spin.

-Andrew
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
diff mbox

Patch

diff --git a/Documentation/devicetree/bindings/usb/nvidia,tegra-xhci.txt b/Documentation/devicetree/bindings/usb/nvidia,tegra-xhci.txt
new file mode 100644
index 0000000..6d94bd6
--- /dev/null
+++ b/Documentation/devicetree/bindings/usb/nvidia,tegra-xhci.txt
@@ -0,0 +1,42 @@ 
+NVIDIA Tegra XHCI controller
+
+The device node for the Tegra XHCI host controller present on Tegra114 and
+later SoCs.
+
+Required properties:
+ - compatible: Should be "nvidia,tegra114-xhci" or "nvidia,tegra124-xhci".
+ - reg: Address and length of the register sets.  There should be three
+   entries in the following order: XHCI host registers, FPCI registers, and
+   IPFS registers.
+ - interrupts: Interrupts used by the controller.  There should be two
+   entries in the following order: XHCI host interrupt and firmware mailbox
+   interrupt.
+ - clocks: Handles to XUSB host and falcon clocks.
+ - clock-names: Should be "xusb_host" and "xusb_falcon_src", respecitively.
+ - resets: Handle to XUSB host reset.
+ - reset-names: Should be "xusb_host".
+ - phys: Handle to the XUSB PHY.
+ - phy-names: Should be "xusb".
+ - s1p05v-supply: 1.05V supply regulator.
+ - s1p8v-supply: 1.8V supply regulator.
+ - s3p3v-supply: 3.3V supply regulator.
+
+Example:
+	usb@0,70090000 {
+		compatible = "nvidia,tegra114-xhci", "nvidia,tegra124-xhci";
+		reg = <0x0 0x70090000 0x0 0x8000>,
+		      <0x0 0x70098000 0x0 0x1000>,
+		      <0x0 0x70099000 0x0 0x1000>;
+		interrupts = <GIC_SPI 39 IRQ_TYPE_LEVEL_HIGH>,
+			     <GIC_SPI 40 IRQ_TYPE_LEVEL_HIGH>;
+		clocks = <&tegra_car TEGRA124_CLK_XUSB_HOST>,
+			 <&tegra_car TEGRA124_CLK_XUSB_FALCON_SRC>;
+		clock-names = "xusb_host", "xusb_falcon_src";
+		resets = <&tegra_car 89>;
+		reset-names = "xusb_host";
+		phy = <&xusb_phy>;
+		phy-names = "xusb";
+		s1p05v-supply = <&vdd_1v05_run>;
+		s3p3v-supply = <&vdd_3v3_lp0>;
+		s1p8v-supply = <&vddio_1v8>;
+	};
diff --git a/drivers/usb/host/Kconfig b/drivers/usb/host/Kconfig
index 3d9e540..43f7977 100644
--- a/drivers/usb/host/Kconfig
+++ b/drivers/usb/host/Kconfig
@@ -29,6 +29,18 @@  if USB_XHCI_HCD
 config USB_XHCI_PLATFORM
 	tristate
 
+config USB_XHCI_TEGRA
+	tristate "Support for NVIDIA Tegra XHCI host controller"
+	depends on ARCH_TEGRA
+	select USB_XHCI_PLATFORM
+	select PHY_TEGRA_XUSB
+	select FW_LOADER
+	---help---
+	  Enables support for the on-chip XHCI controller present on NVIDIA
+	  Tegra114 and later SoCs.
+
+	  If unsure, say N.
+
 endif # USB_XHCI_HCD
 
 config USB_EHCI_HCD
diff --git a/drivers/usb/host/Makefile b/drivers/usb/host/Makefile
index 7530468..5941d36 100644
--- a/drivers/usb/host/Makefile
+++ b/drivers/usb/host/Makefile
@@ -38,6 +38,8 @@  obj-$(CONFIG_USB_EHCI_MSM)	+= ehci-msm.o
 obj-$(CONFIG_USB_EHCI_TEGRA)	+= ehci-tegra.o
 obj-$(CONFIG_USB_W90X900_EHCI)	+= ehci-w90x900.o
 
+obj-$(CONFIG_USB_XHCI_TEGRA)	+= xhci-tegra.o
+
 obj-$(CONFIG_USB_OXU210HP_HCD)	+= oxu210hp-hcd.o
 obj-$(CONFIG_USB_ISP116X_HCD)	+= isp116x-hcd.o
 obj-$(CONFIG_USB_ISP1362_HCD)	+= isp1362-hcd.o
diff --git a/drivers/usb/host/xhci-tegra.c b/drivers/usb/host/xhci-tegra.c
new file mode 100644
index 0000000..5c705b7
--- /dev/null
+++ b/drivers/usb/host/xhci-tegra.c
@@ -0,0 +1,796 @@ 
+/*
+ * NVIDIA Tegra XHCI host controller driver
+ *
+ * Copyright (C) 2014 NVIDIA Corporation
+ * Copyright (C) 2014 Google, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+#include <linux/clk.h>
+#include <linux/delay.h>
+#include <linux/dma-mapping.h>
+#include <linux/firmware.h>
+#include <linux/interrupt.h>
+#include <linux/of_device.h>
+#include <linux/phy/phy.h>
+#include <linux/platform_device.h>
+#include <linux/regulator/consumer.h>
+#include <linux/reset.h>
+#include <linux/usb/tegra_xusb.h>
+
+#include "xhci-tegra.h"
+
+struct tegra_xhci_fw_cfgtbl {
+	u32 boot_loadaddr_in_imem;
+	u32 boot_codedfi_offset;
+	u32 boot_codetag;
+	u32 boot_codesize;
+	u32 phys_memaddr;
+	u16 reqphys_memsize;
+	u16 alloc_phys_memsize;
+	u32 rodata_img_offset;
+	u32 rodata_section_start;
+	u32 rodata_section_end;
+	u32 main_fnaddr;
+	u32 fwimg_cksum;
+	u32 fwimg_created_time;
+	u32 imem_resident_start;
+	u32 imem_resident_end;
+	u32 idirect_start;
+	u32 idirect_end;
+	u32 l2_imem_start;
+	u32 l2_imem_end;
+	u32 version_id;
+	u8 init_ddirect;
+	u8 reserved[3];
+	u32 phys_addr_log_buffer;
+	u32 total_log_entries;
+	u32 dequeue_ptr;
+	u32 dummy_var[2];
+	u32 fwimg_len;
+	u8 magic[8];
+	u32 ss_low_power_entry_timeout;
+	u8 num_hsic_port;
+	u8 padding[139]; /* Padding bytes to make 256-bytes cfgtbl */
+};
+
+struct tegra_xhci_soc_config {
+	const char *firmware_file;
+};
+
+struct tegra_xhci_hcd {
+	struct device *dev;
+	struct platform_device *xhci_pdev;
+
+	int mbox_irq;
+	struct notifier_block mbox_nb;
+
+	phys_addr_t host_phys_base;
+	void __iomem *fpci_base;
+	void __iomem *ipfs_base;
+
+	const struct tegra_xhci_soc_config *soc_config;
+
+	struct regulator *xusb_s1p05v_reg;
+	struct regulator *xusb_s3p3v_reg;
+	struct regulator *xusb_s1p8v_reg;
+
+	/* XUSB host clock */
+	struct clk *host_clk;
+	/* XUSB Falcon controller clock */
+	struct clk *falc_clk;
+	/* XUSB host reset */
+	struct reset_control *host_rst;
+
+	/* XUSB PHY */
+	struct phy *phy;
+
+	/* Firmware loading related */
+	void *fw_data;
+	size_t fw_size;
+	dma_addr_t fw_dma_addr;
+};
+
+static RAW_NOTIFIER_HEAD(tegra_xhci_mbox_notifiers);
+static DEFINE_MUTEX(tegra_xhci_mbox_lock);
+
+int tegra_xhci_register_mbox_notifier(struct notifier_block *nb)
+{
+	int ret;
+
+	mutex_lock(&tegra_xhci_mbox_lock);
+	ret = raw_notifier_chain_register(&tegra_xhci_mbox_notifiers, nb);
+	mutex_unlock(&tegra_xhci_mbox_lock);
+
+	return ret;
+}
+EXPORT_SYMBOL(tegra_xhci_register_mbox_notifier);
+
+void tegra_xhci_unregister_mbox_notifier(struct notifier_block *nb)
+{
+	mutex_lock(&tegra_xhci_mbox_lock);
+	raw_notifier_chain_unregister(&tegra_xhci_mbox_notifiers, nb);
+	mutex_unlock(&tegra_xhci_mbox_lock);
+}
+EXPORT_SYMBOL(tegra_xhci_unregister_mbox_notifier);
+
+static inline u32 fpci_readl(struct tegra_xhci_hcd *tegra, u32 addr)
+{
+	return readl(tegra->fpci_base + addr);
+}
+
+static inline void fpci_writel(struct tegra_xhci_hcd *tegra, u32 val, u32 addr)
+{
+	writel(val, tegra->fpci_base + addr);
+}
+
+static inline u32 ipfs_readl(struct tegra_xhci_hcd *tegra, u32 addr)
+{
+	return readl(tegra->ipfs_base + addr);
+}
+
+static inline void ipfs_writel(struct tegra_xhci_hcd *tegra, u32 val, u32 addr)
+{
+	writel(val, tegra->ipfs_base + addr);
+}
+
+static u32 csb_readl(struct tegra_xhci_hcd *tegra, u32 addr)
+{
+	u32 page, offset;
+
+	page = CSB_PAGE_SELECT(addr);
+	offset = CSB_PAGE_OFFSET(addr);
+	fpci_writel(tegra, page, XUSB_CFG_ARU_C11_CSBRANGE);
+	return fpci_readl(tegra, XUSB_CFG_CSB_BASE_ADDR + offset);
+}
+
+static void csb_writel(struct tegra_xhci_hcd *tegra, u32 val, u32 addr)
+{
+	u32 page, offset;
+
+	page = CSB_PAGE_SELECT(addr);
+	offset = CSB_PAGE_OFFSET(addr);
+	fpci_writel(tegra, page, XUSB_CFG_ARU_C11_CSBRANGE);
+	fpci_writel(tegra, val, XUSB_CFG_CSB_BASE_ADDR + offset);
+}
+
+static void tegra_xhci_cfg(struct tegra_xhci_hcd *tegra)
+{
+	u32 reg;
+
+	reg = ipfs_readl(tegra, IPFS_XUSB_HOST_CONFIGURATION_0);
+	reg |= IPFS_EN_FPCI;
+	ipfs_writel(tegra, reg, IPFS_XUSB_HOST_CONFIGURATION_0);
+	udelay(10);
+
+	/* Program Bar0 Space */
+	reg = fpci_readl(tegra, XUSB_CFG_4);
+	reg |= tegra->host_phys_base;
+	fpci_writel(tegra, reg, XUSB_CFG_4);
+	usleep_range(100, 200);
+
+	/* Enable Bus Master */
+	reg = fpci_readl(tegra, XUSB_CFG_1);
+	reg |= XUSB_IO_SPACE_EN | XUSB_MEM_SPACE_EN | XUSB_BUS_MASTER_EN;
+	fpci_writel(tegra, reg, XUSB_CFG_1);
+
+	/* Set intr mask to enable intr assertion */
+	reg = ipfs_readl(tegra, IPFS_XUSB_HOST_INTR_MASK_0);
+	reg |= IPFS_IP_INT_MASK;
+	ipfs_writel(tegra, reg, IPFS_XUSB_HOST_INTR_MASK_0);
+
+	/* Set hysteris to 0x80 */
+	ipfs_writel(tegra, 0x80, IPFS_XUSB_HOST_CLKGATE_HYSTERESIS_0);
+}
+
+static int tegra_xhci_mbox_send(struct tegra_xhci_hcd *tegra,
+				enum tegra_xusb_mbox_cmd type, u32 data)
+{
+	struct device *dev = tegra->dev;
+	unsigned long timeout;
+	u32 reg;
+
+	dev_dbg(dev, "MBOX send message 0x%x:0x%x\n", type, data);
+	mutex_lock(&tegra_xhci_mbox_lock);
+
+	timeout = jiffies + msecs_to_jiffies(20);
+	/* Wait for mailbox to become idle */
+	while ((fpci_readl(tegra, XUSB_CFG_ARU_MBOX_OWNER) != MBOX_OWNER_NONE)
+	       && time_is_after_jiffies(timeout)) {
+		mutex_unlock(&tegra_xhci_mbox_lock);
+		usleep_range(100, 200);
+		mutex_lock(&tegra_xhci_mbox_lock);
+	}
+	if (fpci_readl(tegra, XUSB_CFG_ARU_MBOX_OWNER) != MBOX_OWNER_NONE) {
+		dev_err(dev, "Mailbox failed to go idle\n");
+		goto timeout;
+	}
+
+	timeout = jiffies + msecs_to_jiffies(10);
+	/* Acquire mailbox */
+	fpci_writel(tegra, MBOX_OWNER_SW, XUSB_CFG_ARU_MBOX_OWNER);
+	while ((fpci_readl(tegra, XUSB_CFG_ARU_MBOX_OWNER) != MBOX_OWNER_SW) &&
+	       time_is_after_jiffies(timeout)) {
+		mutex_unlock(&tegra_xhci_mbox_lock);
+		usleep_range(100, 200);
+		mutex_lock(&tegra_xhci_mbox_lock);
+		fpci_writel(tegra, MBOX_OWNER_SW, XUSB_CFG_ARU_MBOX_OWNER);
+	}
+	if (fpci_readl(tegra, XUSB_CFG_ARU_MBOX_OWNER) != MBOX_OWNER_SW) {
+		dev_err(dev, "Acquire mailbox timeout\n");
+		goto timeout;
+	}
+
+	reg = (type & CMD_TYPE_MASK) << CMD_TYPE_SHIFT;
+	reg |= (data & CMD_DATA_MASK) << CMD_DATA_SHIFT;
+	fpci_writel(tegra, reg, XUSB_CFG_ARU_MBOX_DATA_IN);
+
+	reg = fpci_readl(tegra, XUSB_CFG_ARU_MBOX_CMD);
+	reg |= MBOX_INT_EN | MBOX_FALC_INT_EN;
+	fpci_writel(tegra, reg, XUSB_CFG_ARU_MBOX_CMD);
+
+	mutex_unlock(&tegra_xhci_mbox_lock);
+
+	return 0;
+timeout:
+	mutex_unlock(&tegra_xhci_mbox_lock);
+	return -ETIMEDOUT;
+}
+
+static irqreturn_t tegra_xhci_mbox_irq(int irq, void *p)
+{
+	struct tegra_xhci_hcd *tegra = (struct tegra_xhci_hcd *)p;
+	struct device *dev = tegra->dev;
+	u32 resp = 0, cmd_type, cmd_data, reg;
+
+	mutex_lock(&tegra_xhci_mbox_lock);
+
+	/* Clear mbox interrupts */
+	reg = fpci_readl(tegra, XUSB_CFG_ARU_SMI_INTR);
+	if (reg & MBOX_SMI_INTR_FW_HANG)
+		dev_err(dev, "Hang up inside firmware\n");
+	fpci_writel(tegra, reg, XUSB_CFG_ARU_SMI_INTR);
+
+	/* Get the mbox message from firmware */
+	reg = fpci_readl(tegra, XUSB_CFG_ARU_MBOX_DATA_OUT);
+	cmd_type = (reg >> CMD_TYPE_SHIFT) & CMD_TYPE_MASK;
+	cmd_data = (reg >> CMD_DATA_SHIFT) & CMD_DATA_MASK;
+
+	/* Decode the message and call the notifiers */
+	dev_dbg(dev, "MBOX receive message 0x%x:0x%x\n", cmd_type, cmd_data);
+	if (cmd_type < MBOX_CMD_MAX) {
+		struct mbox_notifier_data mbox_info;
+
+		mbox_info.msg_data = cmd_data;
+		mbox_info.resp_cmd = 0;
+		mbox_info.resp_data = 0;
+		raw_notifier_call_chain(&tegra_xhci_mbox_notifiers, cmd_type,
+					&mbox_info);
+		if (mbox_info.resp_cmd) {
+			resp = (mbox_info.resp_cmd & CMD_TYPE_MASK) <<
+				CMD_TYPE_SHIFT;
+			resp |= (mbox_info.resp_data & CMD_DATA_MASK) <<
+				CMD_DATA_SHIFT;
+		}
+	} else if (cmd_type == MBOX_CMD_ACK) {
+		dev_dbg(dev, "Firmware responds with ACK\n");
+	} else if (cmd_type == MBOX_CMD_NAK) {
+		dev_err(dev, "Firmware responds with NAK\n");
+	} else {
+		dev_err(dev, "Invalid command %x\n", cmd_type);
+	}
+
+	if (resp) {
+		/* Send ACK/NAK to firmware */
+		fpci_writel(tegra, resp, XUSB_CFG_ARU_MBOX_DATA_IN);
+		reg = fpci_readl(tegra, XUSB_CFG_ARU_MBOX_CMD);
+		reg |= MBOX_INT_EN | MBOX_FALC_INT_EN;
+		fpci_writel(tegra, reg, XUSB_CFG_ARU_MBOX_CMD);
+	} else {
+		/* Clear MBOX_SMI_INT_EN bit */
+		reg = fpci_readl(tegra, XUSB_CFG_ARU_MBOX_CMD);
+		reg &= ~MBOX_SMI_INT_EN;
+		fpci_writel(tegra, reg, XUSB_CFG_ARU_MBOX_CMD);
+		/* Clear mailbox ownership */
+		fpci_writel(tegra, MBOX_OWNER_NONE, XUSB_CFG_ARU_MBOX_OWNER);
+	}
+
+	mutex_unlock(&tegra_xhci_mbox_lock);
+
+	return IRQ_HANDLED;
+}
+
+static int tegra_xhci_default_mbox_notifier(struct notifier_block *nb,
+					    unsigned long event, void *p)
+{
+	struct tegra_xhci_hcd *tegra = container_of(nb, struct tegra_xhci_hcd,
+						    mbox_nb);
+	struct mbox_notifier_data *data = (struct mbox_notifier_data *)p;
+
+	switch (event) {
+	case MBOX_CMD_INC_FALC_CLOCK:
+	case MBOX_CMD_DEC_FALC_CLOCK:
+		data->resp_data = clk_get_rate(tegra->falc_clk) / 1000;
+		if (data->resp_data != data->msg_data)
+			data->resp_cmd = MBOX_CMD_NAK;
+		else
+			data->resp_cmd = MBOX_CMD_ACK;
+		return NOTIFY_STOP;
+	case MBOX_CMD_SET_BW:
+		/* No support for EMC scaling yet */
+		return NOTIFY_STOP;
+	default:
+		return NOTIFY_DONE;
+	}
+}
+
+static int tegra_xhci_load_firmware(struct tegra_xhci_hcd *tegra)
+{
+	struct device *dev = tegra->dev;
+	struct tegra_xhci_fw_cfgtbl *cfg_tbl;
+	u64 fw_base;
+	u32 val;
+	time_t fw_time;
+	struct tm fw_tm;
+
+	if (csb_readl(tegra, XUSB_CSB_MP_ILOAD_BASE_LO) != 0) {
+		dev_info(dev, "Firmware already loaded, Falcon state 0x%x\n",
+			 csb_readl(tegra, XUSB_FALC_CPUCTL));
+		return 0;
+	}
+
+	cfg_tbl = (struct tegra_xhci_fw_cfgtbl *)tegra->fw_data;
+
+	/* Program the size of DFI into ILOAD_ATTR */
+	csb_writel(tegra, tegra->fw_size, XUSB_CSB_MP_ILOAD_ATTR);
+
+	/*
+	 * Boot code of the firmware reads the ILOAD_BASE registers
+	 * to get to the start of the DFI in system memory.
+	 */
+	fw_base = tegra->fw_dma_addr + sizeof(*cfg_tbl);
+	csb_writel(tegra, fw_base, XUSB_CSB_MP_ILOAD_BASE_LO);
+	csb_writel(tegra, fw_base >> 32, XUSB_CSB_MP_ILOAD_BASE_HI);
+
+	/* Set BOOTPATH to 1 in APMAP. */
+	csb_writel(tegra, APMAP_BOOTPATH, XUSB_CSB_MP_APMAP);
+
+	/* Invalidate L2IMEM. */
+	csb_writel(tegra, L2IMEMOP_INVALIDATE_ALL, XUSB_CSB_MP_L2IMEMOP_TRIG);
+
+	/*
+	 * Initiate fetch of bootcode from system memory into L2IMEM.
+	 * Program bootcode location and size in system memory.
+	 */
+	val = (DIV_ROUND_UP(cfg_tbl->boot_codetag, IMEM_BLOCK_SIZE) &
+	       L2IMEMOP_SIZE_SRC_OFFSET_MASK) << L2IMEMOP_SIZE_SRC_OFFSET_SHIFT;
+	val |= (DIV_ROUND_UP(cfg_tbl->boot_codesize, IMEM_BLOCK_SIZE) &
+		L2IMEMOP_SIZE_SRC_COUNT_MASK) << L2IMEMOP_SIZE_SRC_COUNT_SHIFT;
+	csb_writel(tegra, val, XUSB_CSB_MP_L2IMEMOP_SIZE);
+
+	/* Trigger L2IMEM Load operation. */
+	csb_writel(tegra, L2IMEMOP_LOAD_LOCKED_RESULT,
+		   XUSB_CSB_MP_L2IMEMOP_TRIG);
+
+	/* Setup Falcon Auto-fill */
+	val = DIV_ROUND_UP(cfg_tbl->boot_codesize, IMEM_BLOCK_SIZE);
+	csb_writel(tegra, val, XUSB_FALC_IMFILLCTL);
+
+	val = DIV_ROUND_UP(cfg_tbl->boot_codetag, IMEM_BLOCK_SIZE) &
+		IMFILLRNG1_TAG_MASK;
+	val |= DIV_ROUND_UP(cfg_tbl->boot_codetag + cfg_tbl->boot_codesize,
+			    IMEM_BLOCK_SIZE) << IMFILLRNG1_TAG_HI_SHIFT;
+	csb_writel(tegra, val, XUSB_FALC_IMFILLRNG1);
+
+	csb_writel(tegra, 0, XUSB_FALC_DMACTL);
+	msleep(50);
+
+	csb_writel(tegra, cfg_tbl->boot_codetag, XUSB_FALC_BOOTVEC);
+
+	/* Start Falcon CPU */
+	csb_writel(tegra, CPUCTL_STARTCPU, XUSB_FALC_CPUCTL);
+	usleep_range(1000, 2000);
+
+	fw_time = cfg_tbl->fwimg_created_time;
+	time_to_tm(fw_time, 0, &fw_tm);
+	dev_info(dev,
+		 "Firmware timestamp: %ld-%02d-%02d %02d:%02d:%02d UTC, "
+		 "Falcon state 0x%x\n", fw_tm.tm_year + 1900,
+		 fw_tm.tm_mon + 1, fw_tm.tm_mday, fw_tm.tm_hour,
+		 fw_tm.tm_min, fw_tm.tm_sec,
+		 csb_readl(tegra, XUSB_FALC_CPUCTL));
+
+	/* Bail out if Falcon CPU is not in a good state */
+	if (csb_readl(tegra, XUSB_FALC_CPUCTL) == CPUCTL_STATE_HALTED)
+		return -EIO;
+
+	return 0;
+}
+
+static int tegra_xhci_regulator_init(struct tegra_xhci_hcd *tegra)
+{
+	struct device *dev = tegra->dev;
+	int err = 0;
+
+	tegra->xusb_s3p3v_reg = devm_regulator_get(dev, "s3p3v");
+	if (IS_ERR(tegra->xusb_s3p3v_reg)) {
+		dev_err(dev, "s3p3v regulator not found: %ld.",
+			PTR_ERR(tegra->xusb_s3p3v_reg));
+		return PTR_ERR(tegra->xusb_s3p3v_reg);
+	}
+	err = regulator_enable(tegra->xusb_s3p3v_reg);
+	if (err < 0) {
+		dev_err(dev, "s3p3v regulator enable failed:%d\n", err);
+		return err;
+	}
+
+	tegra->xusb_s1p8v_reg = devm_regulator_get(dev, "s1p8v");
+	if (IS_ERR(tegra->xusb_s1p8v_reg)) {
+		dev_err(dev, "s1p8v regulator not found: %ld.",
+			PTR_ERR(tegra->xusb_s1p8v_reg));
+		err = PTR_ERR(tegra->xusb_s1p8v_reg);
+		goto err_put_s3p3v_reg;
+	} else {
+		err = regulator_enable(tegra->xusb_s1p8v_reg);
+		if (err < 0) {
+			dev_err(dev, "s1p8v regulator enable failed:%d\n", err);
+			goto err_put_s3p3v_reg;
+		}
+	}
+
+	tegra->xusb_s1p05v_reg = devm_regulator_get(dev, "s1p05v");
+	if (IS_ERR(tegra->xusb_s1p05v_reg)) {
+		dev_err(dev, "s1p05v regulator not found: %ld.",
+			PTR_ERR(tegra->xusb_s1p05v_reg));
+		err = PTR_ERR(tegra->xusb_s1p05v_reg);
+		goto err_put_s1p8v_reg;
+	} else {
+		err = regulator_enable(tegra->xusb_s1p05v_reg);
+		if (err < 0) {
+			dev_err(dev,
+				"s1p05v regulator enable failed:%d\n", err);
+			goto err_put_s1p8v_reg;
+		}
+	}
+
+	return 0;
+
+err_put_s1p8v_reg:
+	regulator_disable(tegra->xusb_s1p8v_reg);
+err_put_s3p3v_reg:
+	regulator_disable(tegra->xusb_s3p3v_reg);
+	return err;
+}
+
+static void tegra_xhci_regulator_deinit(struct tegra_xhci_hcd *tegra)
+{
+	regulator_disable(tegra->xusb_s1p05v_reg);
+	regulator_disable(tegra->xusb_s1p8v_reg);
+	regulator_disable(tegra->xusb_s3p3v_reg);
+}
+
+static int tegra_xhci_clk_init(struct tegra_xhci_hcd *tegra)
+{
+	struct device *dev = tegra->dev;
+	int err = 0;
+
+	tegra->host_clk = devm_clk_get(dev, "xusb_host");
+	if (IS_ERR(tegra->host_clk)) {
+		dev_err(dev, "Failed to get xusb_host clk\n");
+		err = PTR_ERR(tegra->host_clk);
+		return err;
+	}
+
+	tegra->falc_clk = devm_clk_get(dev, "xusb_falcon_src");
+	if (IS_ERR(tegra->falc_clk)) {
+		dev_err(dev, "Failed to get xusb_falcon_src clk\n");
+		err = PTR_ERR(tegra->falc_clk);
+		return err;
+	}
+
+	clk_prepare_enable(tegra->host_clk);
+	clk_prepare_enable(tegra->falc_clk);
+
+	return 0;
+}
+
+static void tegra_xhci_clk_deinit(struct tegra_xhci_hcd *tegra)
+{
+	clk_disable_unprepare(tegra->host_clk);
+	clk_disable_unprepare(tegra->falc_clk);
+}
+
+static const struct tegra_xhci_soc_config tegra114_soc_config = {
+	.firmware_file = "tegra11x/tegra_xusb_firmware",
+};
+
+static const struct tegra_xhci_soc_config tegra124_soc_config = {
+	.firmware_file = "tegra12x/tegra_xusb_firmware",
+};
+
+static struct of_device_id tegra_xhci_of_match[] = {
+	{ .compatible = "nvidia,tegra114-xhci", .data = &tegra114_soc_config },
+	{ .compatible = "nvidia,tegra124-xhci", .data = &tegra124_soc_config },
+	{ },
+};
+MODULE_DEVICE_TABLE(of, tegra_xhci_of_match);
+
+static void tegra_xhci_probe_finish(const struct firmware *fw, void *context)
+{
+	struct tegra_xhci_hcd *tegra = context;
+	struct device *dev = tegra->dev;
+	struct platform_device *xhci = NULL;
+	struct resource xhci_resources[2];
+	struct resource	*res;
+	struct tegra_xhci_fw_cfgtbl *cfg_tbl;
+	int ret;
+
+	if (!fw) {
+		dev_err(dev, "Failed to load firmware file\n");
+		ret = -ENOENT;
+		goto out;
+	}
+
+	/* Load Falcon controller with its firmware */
+	cfg_tbl = (struct tegra_xhci_fw_cfgtbl *)fw->data;
+	tegra->fw_size = cfg_tbl->fwimg_len;
+	tegra->fw_data = dma_alloc_coherent(dev, tegra->fw_size,
+					    &tegra->fw_dma_addr,
+					    GFP_KERNEL);
+	if (!tegra->fw_data) {
+		dev_err(dev, "Failed to allocate DMA buffer for firmware\n");
+		ret = -ENOMEM;
+		goto out;
+	}
+	memcpy(tegra->fw_data, fw->data, tegra->fw_size);
+
+	ret = tegra_xhci_load_firmware(tegra);
+	if (ret < 0) {
+		dev_err(dev, "Failed to load controller firmware\n");
+		goto out;
+	}
+
+	/* Set up mailbox notifier and interrupt handler */
+	tegra->mbox_nb.notifier_call = tegra_xhci_default_mbox_notifier;
+	ret = tegra_xhci_register_mbox_notifier(&tegra->mbox_nb);
+	if (ret < 0) {
+		dev_err(dev, "Failed to registe mbox handler\n");
+		goto out;
+	}
+	tegra->mbox_irq = platform_get_irq(to_platform_device(dev), 1);
+	if (!tegra->mbox_irq) {
+		dev_err(dev, "Missing MBOX IRQ\n");
+		ret = -ENODEV;
+		goto out;
+	}
+	ret = devm_request_threaded_irq(dev, tegra->mbox_irq, NULL,
+					tegra_xhci_mbox_irq, IRQF_ONESHOT,
+					"tegra_xhci_mbox_irq", tegra);
+	if (ret != 0) {
+		dev_err(dev, "Failed to request MBOX IRQ\n");
+		goto out;
+	}
+
+	/* Create child xhci-plat device */
+	memset(xhci_resources, 0, sizeof(xhci_resources));
+	res = platform_get_resource(to_platform_device(dev), IORESOURCE_IRQ, 0);
+	if (!res) {
+		dev_err(dev, "Missing XHCI IRQ\n");
+		ret = -ENODEV;
+		goto out;
+	}
+	xhci_resources[0].start = res->start;
+	xhci_resources[0].end = res->end;
+	xhci_resources[0].flags = res->flags;
+	xhci_resources[0].name = res->name;
+	res = platform_get_resource(to_platform_device(dev), IORESOURCE_MEM, 0);
+	if (!res) {
+		dev_err(dev, "Missing XHCI registers\n");
+		ret = -ENODEV;
+		goto out;
+	}
+	xhci_resources[1].start = res->start;
+	xhci_resources[1].end = res->end;
+	xhci_resources[1].flags = res->flags;
+	xhci_resources[1].name = res->name;
+
+	xhci = platform_device_alloc("xhci-hcd", PLATFORM_DEVID_AUTO);
+	if (!xhci) {
+		dev_err(dev, "Failed to allocate XHCI host\n");
+		ret = -ENOMEM;
+		goto out;
+	}
+	xhci->dev.parent = dev;
+	ret = dma_coerce_mask_and_coherent(&xhci->dev,
+					   dev->coherent_dma_mask);
+	if (ret) {
+		dev_err(dev, "Failed to set XHCI dma mask\n");
+		goto out;
+	}
+	ret = platform_device_add_resources(xhci, xhci_resources,
+					    ARRAY_SIZE(xhci_resources));
+	if (ret) {
+		dev_err(dev, "Failed to add XHCI resources\n");
+		goto out;
+	}
+	ret = platform_device_add(xhci);
+	if (ret) {
+		dev_err(dev, "failed to register XHCI device\n");
+		goto out;
+	}
+	tegra->xhci_pdev = xhci;
+
+	/* Enable firmware messages from controller */
+	tegra_xhci_mbox_send(tegra, MBOX_CMD_MSG_ENABLED, 0);
+
+out:
+	release_firmware(fw);
+	if (ret)
+		platform_device_put(xhci);
+}
+
+static int tegra_xhci_probe(struct platform_device *pdev)
+{
+	struct tegra_xhci_hcd *tegra;
+	struct resource	*res;
+	const struct of_device_id *match;
+	int ret;
+
+	BUILD_BUG_ON(sizeof(struct tegra_xhci_fw_cfgtbl) != 256);
+
+	tegra = devm_kzalloc(&pdev->dev, sizeof(*tegra), GFP_KERNEL);
+	if (!tegra) {
+		dev_err(&pdev->dev, "memory alloc failed\n");
+		return -ENOMEM;
+	}
+	platform_set_drvdata(pdev, tegra);
+	tegra->dev = &pdev->dev;
+
+	match = of_match_device(tegra_xhci_of_match, &pdev->dev);
+	if (!match) {
+		dev_err(&pdev->dev, "No device match found\n");
+		return -ENODEV;
+	}
+	tegra->soc_config = match->data;
+
+	/*
+	 * Right now device-tree probed devices don't get dma_mask set.
+	 * Since shared usb code relies on it, set it here for now.
+	 * Once we have dma capability bindings this can go away.
+	 */
+	ret = dma_coerce_mask_and_coherent(&pdev->dev, DMA_BIT_MASK(64));
+	if (ret)
+		return ret;
+
+	/*
+	 * The firmware needs the XHCI host's base physical address.
+	 * Get it here and save it for later.
+	 */
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	if (!res) {
+		dev_err(&pdev->dev, "Missing XHCI registers\n");
+		return -ENODEV;
+	}
+	tegra->host_phys_base = res->start;
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
+	tegra->fpci_base = devm_ioremap_resource(&pdev->dev, res);
+	if (!tegra->fpci_base) {
+		dev_err(&pdev->dev, "Failed to map FPCI registers\n");
+		return -ENOMEM;
+	}
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 2);
+	tegra->ipfs_base = devm_ioremap_resource(&pdev->dev, res);
+	if (!tegra->ipfs_base) {
+		dev_err(&pdev->dev, "Failed to map IPFS registers\n");
+		return -ENOMEM;
+	}
+
+	tegra->phy = devm_phy_get(&pdev->dev, "xusb");
+	if (IS_ERR(tegra->phy)) {
+		dev_err(&pdev->dev, "Failed to get PHY\n");
+		return PTR_ERR(tegra->phy);
+	}
+	ret = phy_init(tegra->phy);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "PHY initialization failed\n");
+		return ret;
+	}
+
+	tegra->host_rst = devm_reset_control_get(&pdev->dev, "xusb_host");
+	if (IS_ERR(tegra->host_rst)) {
+		dev_err(&pdev->dev, "Failed to get host reset\n");
+		ret = PTR_ERR(tegra->host_rst);
+		goto err_deinit_phy;
+	}
+
+	ret = tegra_xhci_clk_init(tegra);
+	if (ret) {
+		dev_err(&pdev->dev, "Failed to initialize XHCI clocks\n");
+		goto err_deinit_phy;
+	}
+
+	ret = tegra_xhci_regulator_init(tegra);
+	if (ret) {
+		dev_err(&pdev->dev, "Failed to initialize XHCI regulators\n");
+		goto err_deinit_clk;
+	}
+
+	/* Setup IPFS access and BAR0 space */
+	tegra_xhci_cfg(tegra);
+
+	ret = phy_power_on(tegra->phy);
+	if (ret < 0) {
+		dev_err(&pdev->dev, "PHY power up failed\n");
+		goto err_deinit_regulator;
+	}
+
+	ret = request_firmware_nowait(THIS_MODULE, true,
+				      tegra->soc_config->firmware_file,
+				      tegra->dev, GFP_KERNEL, tegra,
+				      tegra_xhci_probe_finish);
+	if (ret < 0) {
+		dev_err(tegra->dev, "Failed to request firmware\n");
+		goto err_shutdown_phy;
+	}
+
+	return 0;
+
+err_shutdown_phy:
+	phy_power_off(tegra->phy);
+err_deinit_regulator:
+	tegra_xhci_regulator_deinit(tegra);
+err_deinit_clk:
+	tegra_xhci_clk_deinit(tegra);
+err_deinit_phy:
+	phy_exit(tegra->phy);
+
+	return ret;
+}
+
+static int tegra_xhci_remove(struct platform_device *pdev)
+{
+	struct tegra_xhci_hcd *tegra = platform_get_drvdata(pdev);
+
+	phy_power_off(tegra->phy);
+	phy_exit(tegra->phy);
+
+	if (tegra->fw_data)
+		dma_free_coherent(tegra->dev, tegra->fw_size, tegra->fw_data,
+				  tegra->fw_dma_addr);
+
+	tegra_xhci_unregister_mbox_notifier(&tegra->mbox_nb);
+	tegra_xhci_regulator_deinit(tegra);
+	tegra_xhci_clk_deinit(tegra);
+
+	return 0;
+}
+
+static struct platform_driver tegra_xhci_driver = {
+	.probe	= tegra_xhci_probe,
+	.remove	= tegra_xhci_remove,
+	.driver	= {
+		.name = "tegra-xhci",
+		.of_match_table = of_match_ptr(tegra_xhci_of_match),
+	},
+};
+module_platform_driver(tegra_xhci_driver);
+
+MODULE_DESCRIPTION("NVIDIA Tegra XHCI host-controller driver");
+MODULE_LICENSE("GPL v2");
+MODULE_ALIAS("platform:tegra-xhci");
diff --git a/drivers/usb/host/xhci-tegra.h b/drivers/usb/host/xhci-tegra.h
new file mode 100644
index 0000000..b65bfdb
--- /dev/null
+++ b/drivers/usb/host/xhci-tegra.h
@@ -0,0 +1,131 @@ 
+/*
+ * Copyright (C) 2014 NVIDIA Corporation
+ * Copyright (C) 2014 Google, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __XHCI_TEGRA_H
+#define __XHCI_TEGRA_H
+
+/* FPCI CFG registers */
+#define XUSB_CFG_0				0x0
+#define XUSB_CFG_1				0x4
+#define  XUSB_IO_SPACE_EN			BIT(0)
+#define  XUSB_MEM_SPACE_EN			BIT(1)
+#define  XUSB_BUS_MASTER_EN			BIT(2)
+#define XUSB_CFG_4				0x10
+#define XUSB_CFG_16				0x40
+#define XUSB_CFG_24				0x60
+#define XUSB_CFG_ARU_MBOX_CMD			0xe4
+#define  MBOX_FALC_INT_EN			BIT(27)
+#define  MBOX_PME_INT_EN			BIT(28)
+#define  MBOX_SMI_INT_EN			BIT(29)
+#define  MBOX_XHCI_INT_EN			BIT(30)
+#define  MBOX_INT_EN				BIT(31)
+#define XUSB_CFG_ARU_MBOX_DATA_IN		0xe8
+#define  CMD_DATA_SHIFT				0
+#define  CMD_DATA_MASK				0xffffff
+#define  CMD_TYPE_SHIFT				24
+#define  CMD_TYPE_MASK				0xff
+#define XUSB_CFG_ARU_MBOX_DATA_OUT		0xec
+#define XUSB_CFG_ARU_MBOX_OWNER			0xf0
+#define  MBOX_OWNER_NONE			0
+#define  MBOX_OWNER_FW				1
+#define  MBOX_OWNER_SW				2
+#define XUSB_CFG_FPCICFG			0xf8
+#define XUSB_CFG_ARU_C11PAGESEL0		0x400
+#define XUSB_CFG_ARU_C11PAGESEL1		0x404
+#define XUSB_CFG_ARU_C11_CSBRANGE		0x41c
+#define XUSB_CFG_ARU_SMI_INTR			0x428
+#define  MBOX_SMI_INTR_FW_HANG			BIT(1)
+#define  MBOX_SMI_INTR_EN			BIT(3)
+#define XUSB_CFG_ARU_RST			0x42c
+#define XUSB_CFG_ARU_CONTEXT			0x43c
+#define XUSB_CFG_ARU_FW_SCRATCH			0x440
+#define XUSB_CFG_ARU_CONTEXT_HS_PLS		0x478
+#define XUSB_CFG_ARU_CONTEXT_FS_PLS		0x47c
+#define XUSB_CFG_ARU_CONTEXT_HSFS_SPEED		0x480
+#define XUSB_CFG_ARU_CONTEXT_HSFS_PP		0x484
+#define XUSB_CFG_HSPX_CORE_HSICWRAP		0x658
+#define XUSB_CFG_CSB_BASE_ADDR			0x800
+
+/* IPFS registers */
+#define IPFS_XUSB_HOST_MSI_BAR_SZ_0		0xc0
+#define IPFS_XUSB_HOST_MSI_AXI_BAR_ST_0		0xc4
+#define IPFS_XUSB_HOST_FPCI_BAR_ST_0		0xc8
+#define IPFS_XUSB_HOST_MSI_VEC0_0		0x100
+#define IPFS_XUSB_HOST_MSI_EN_VEC0_0		0x140
+#define IPFS_XUSB_HOST_CONFIGURATION_0		0x180
+#define  IPFS_EN_FPCI				BIT(0)
+#define IPFS_XUSB_HOST_FPCI_ERROR_MASKS_0	0x184
+#define IPFS_XUSB_HOST_INTR_MASK_0		0x188
+#define  IPFS_IP_INT_MASK			BIT(16)
+#define IPFS_XUSB_HOST_IPFS_INTR_ENABLE_0	0x198
+#define IPFS_XUSB_HOST_UFPCI_CONFIG_0		0x19c
+#define IPFS_XUSB_HOST_CLKGATE_HYSTERESIS_0	0x1bc
+#define IPFS_XUSB_HOST_MCCIF_FIFOCTRL_0		0x1dc
+
+#define IMEM_BLOCK_SIZE				256
+
+#define CSB_PAGE_SELECT_MASK			0x7fffff
+#define CSB_PAGE_SELECT_SHIFT			9
+#define CSB_PAGE_OFFSET_MASK			0x1ff
+#define CSB_PAGE_SELECT(addr)	((addr) >> (CSB_PAGE_SELECT_SHIFT) &	\
+				 CSB_PAGE_SELECT_MASK)
+#define CSB_PAGE_OFFSET(addr)	((addr) & CSB_PAGE_OFFSET_MASK)
+
+/* Falcon CSB registers */
+#define XUSB_FALC_CPUCTL			0x100
+#define  CPUCTL_STARTCPU			BIT(1)
+#define  CPUCTL_STATE_HALTED			BIT(4)
+#define XUSB_FALC_BOOTVEC			0x104
+#define XUSB_FALC_DMACTL			0x10c
+#define XUSB_FALC_IMFILLRNG1			0x154
+#define  IMFILLRNG1_TAG_MASK			0xffff
+#define  IMFILLRNG1_TAG_HI_SHIFT		16
+#define XUSB_FALC_IMFILLCTL			0x158
+#define XUSB_FALC_CMEMBASE			0x160
+#define XUSB_FALC_DMEMAPERT			0x164
+#define XUSB_FALC_IMEMC_START			0x180
+#define XUSB_FALC_IMEMD_START			0x184
+#define XUSB_FALC_IMEMT_START			0x188
+#define XUSB_FALC_ICD_CMD			0x200
+#define XUSB_FALC_ICD_RDATA			0x20C
+#define XUSB_FALC_SS_PVTPORTSC1			0x116000
+#define XUSB_FALC_SS_PVTPORTSC2			0x116004
+#define XUSB_FALC_SS_PVTPORTSC3			0x116008
+#define XUSB_FALC_HS_PVTPORTSC1			0x116800
+#define XUSB_FALC_HS_PVTPORTSC2			0x116804
+#define XUSB_FALC_HS_PVTPORTSC3			0x116808
+#define XUSB_FALC_FS_PVTPORTSC1			0x117000
+#define XUSB_FALC_FS_PVTPORTSC2			0x117004
+#define XUSB_FALC_FS_PVTPORTSC3			0x117008
+
+/* MP CSB registers */
+#define XUSB_CSB_MP_ILOAD_ATTR			0x101a00
+#define XUSB_CSB_MP_ILOAD_BASE_LO		0x101a04
+#define XUSB_CSB_MP_ILOAD_BASE_HI		0x101a08
+#define XUSB_CSB_MP_L2IMEMOP_SIZE		0x101a10
+#define  L2IMEMOP_SIZE_SRC_OFFSET_SHIFT		8
+#define  L2IMEMOP_SIZE_SRC_OFFSET_MASK		0x3ff
+#define  L2IMEMOP_SIZE_SRC_COUNT_SHIFT		24
+#define  L2IMEMOP_SIZE_SRC_COUNT_MASK		0xff
+#define XUSB_CSB_MP_L2IMEMOP_TRIG		0x101a14
+#define  L2IMEMOP_ACTION_SHIFT			24
+#define  L2IMEMOP_INVALIDATE_ALL		(0x40 << L2IMEMOP_ACTION_SHIFT)
+#define  L2IMEMOP_LOAD_LOCKED_RESULT		(0x11 << L2IMEMOP_ACTION_SHIFT)
+#define XUSB_CSB_MP_APMAP			0x10181c
+#define  APMAP_BOOTPATH				BIT(31)
+
+#endif /* __XHCI_TEGRA_H */
diff --git a/include/linux/usb/tegra_xusb.h b/include/linux/usb/tegra_xusb.h
new file mode 100644
index 0000000..1a654e7
--- /dev/null
+++ b/include/linux/usb/tegra_xusb.h
@@ -0,0 +1,66 @@ 
+/*
+ * Copyright (C) 2014 NVIDIA Corporation
+ * Copyright (C) 2014 Google, Inc.
+ *
+ * This program is free software; you can redistribute it and/or modify it
+ * under the terms and conditions of the GNU General Public License,
+ * version 2, as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope it will be useful, but WITHOUT
+ * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ * FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ * more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program.  If not, see <http://www.gnu.org/licenses/>.
+ */
+
+#ifndef __TEGRA_XUSB_H
+#define __TEGRA_XUSB_H
+
+/* Command requests from the firmware */
+enum tegra_xusb_mbox_cmd {
+	MBOX_CMD_MSG_ENABLED = 1,
+	MBOX_CMD_INC_FALC_CLOCK,
+	MBOX_CMD_DEC_FALC_CLOCK,
+	MBOX_CMD_INC_SSPI_CLOCK,
+	MBOX_CMD_DEC_SSPI_CLOCK,
+	MBOX_CMD_SET_BW, /* no ACK/NAK required */
+	MBOX_CMD_SET_SS_PWR_GATING,
+	MBOX_CMD_SET_SS_PWR_UNGATING,
+	MBOX_CMD_SAVE_DFE_CTLE_CTX,
+	MBOX_CMD_AIRPLANE_MODE_ENABLED, /* unused */
+	MBOX_CMD_AIRPLANE_MODE_DISABLED, /* unused */
+	MBOX_CMD_START_HSIC_IDLE,
+	MBOX_CMD_STOP_HSIC_IDLE,
+	MBOX_CMD_DBC_WAKE_STACK, /* unused */
+	MBOX_CMD_HSIC_PRETEND_CONNECT,
+
+	MBOX_CMD_MAX,
+
+	/* Response message to above commands */
+	MBOX_CMD_ACK = 128,
+	MBOX_CMD_NAK
+};
+
+struct notifier_block;
+
+/*
+ * Tegra XUSB MBOX handler interface:
+ *   - Drivers which may handle mbox messages should register a notifier.
+ *   - The notifier event will be an mbox command (above) and the data will
+ *     be a pointer to struct mbox_notifier_data.
+ *   - If a notifier has handled the message, it should return NOTIFY_STOP
+ *     and populate resp_{cmd,data} appropriately.
+ *   - A resp_cmd of 0 indicates that no response should be sent.
+ */
+struct mbox_notifier_data {
+	u32 msg_data; /* Inbound message data */
+	u32 resp_cmd; /* Response message command */
+	u32 resp_data; /* Response message data */
+};
+
+extern int tegra_xhci_register_mbox_notifier(struct notifier_block *nb);
+extern void tegra_xhci_unregister_mbox_notifier(struct notifier_block *nb);
+
+#endif /* __TEGRA_XUSB_H */