diff mbox

[v9,4/5] xhci: mediatek: support MTK xHCI host controller

Message ID 1443495698-32233-5-git-send-email-chunfeng.yun@mediatek.com (mailing list archive)
State New, archived
Headers show

Commit Message

Chunfeng Yun (云春峰) Sept. 29, 2015, 3:01 a.m. UTC
There some vendor quirks for MTK xhci host controller:
1. It defines some extra SW scheduling parameters for HW
  to minimize the scheduling effort for synchronous and
  interrupt endpoints. The parameters are put into reseved
  DWs of slot context and endpoint context.
2. Its IMODI unit for Interrupter Moderation register is
  8 times as much as that defined in xHCI spec.
3. Its TDS in  Normal TRB defines a number of packets that
  remains to be transferred for a TD after processing all
  Max packets in all previous TRBs.

Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
---
 drivers/usb/host/Kconfig        |   9 +
 drivers/usb/host/Makefile       |   4 +
 drivers/usb/host/xhci-mtk-sch.c | 424 ++++++++++++++++++++++
 drivers/usb/host/xhci-mtk.c     | 776 ++++++++++++++++++++++++++++++++++++++++
 drivers/usb/host/xhci-mtk.h     | 156 ++++++++
 drivers/usb/host/xhci-ring.c    |  22 +-
 drivers/usb/host/xhci.c         |  19 +-
 drivers/usb/host/xhci.h         |   1 +
 8 files changed, 1404 insertions(+), 7 deletions(-)
 create mode 100644 drivers/usb/host/xhci-mtk-sch.c
 create mode 100644 drivers/usb/host/xhci-mtk.c
 create mode 100644 drivers/usb/host/xhci-mtk.h

Comments

Daniel Thompson Oct. 1, 2015, 11:44 a.m. UTC | #1
On 29/09/15 04:01, Chunfeng Yun wrote:
> There some vendor quirks for MTK xhci host controller:
> 1. It defines some extra SW scheduling parameters for HW
>    to minimize the scheduling effort for synchronous and
>    interrupt endpoints. The parameters are put into reseved
>    DWs of slot context and endpoint context.
> 2. Its IMODI unit for Interrupter Moderation register is
>    8 times as much as that defined in xHCI spec.
> 3. Its TDS in  Normal TRB defines a number of packets that
>    remains to be transferred for a TD after processing all
>    Max packets in all previous TRBs.
>
> Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>

I've done some basic soak tests, both with a directly attached USB3 HDD 
and, given the extra code to manage isochronous xfer, also with a hub, 
disc and two audio interfaces.

Tested-by: Daniel Thompson <daniel.thompson@linaro.org>


> diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
> index 57f40a1..243f696 100644
> --- a/drivers/usb/host/xhci-ring.c
> +++ b/drivers/usb/host/xhci-ring.c
> @@ -68,6 +68,7 @@
>   #include <linux/slab.h>
>   #include "xhci.h"
>   #include "xhci-trace.h"
> +#include "xhci-mtk.h"
>
>   /*
>    * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
> @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
>   			      struct urb *urb, unsigned int num_trbs_left)
>   {
>   	u32 maxp, total_packet_count;
> +	u32 skip_current_trb = 0;

A bit of a nitpick but why do we need skip_current_trb? Testing 
(xhci->quirks & XHCI_MTK_HOST) twice would make what the code does, and 
why, more obvious.

Anyhow with that looked at, and the caveat that I'm not much of USB 
expert, you're welcome to add my Reviewed-by: to v10.

>
> -	if (xhci->hci_version < 0x100)
> -		return ((td_total_len - transferred) >> 10);
> -
> -	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> -	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> +	if (xhci->hci_version < 0x100) {
> +		/* for MTK xHCI, TD size doesn't include this TRB */
> +		if (xhci->quirks & XHCI_MTK_HOST)
> +			skip_current_trb = 1;
> +		else
> +			return ((td_total_len - transferred) >> 10);
> +	}
>
>   	/* One TRB with a zero-length data packet. */
>   	if (num_trbs_left == 0 || (transferred == 0 && trb_buff_len == 0) ||
>   	    trb_buff_len == td_total_len)
>   		return 0;
>
> +	if (skip_current_trb)
> +		trb_buff_len = 0;
> +
> +	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> +	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> +
>   	/* Queueing functions don't count the current TRB into transferred */
>   	return (total_packet_count - ((transferred + trb_buff_len) / maxp));
>   }


Daniel.
Chunfeng Yun (云春峰) Oct. 8, 2015, 12:05 p.m. UTC | #2
Hi,
On Thu, 2015-10-01 at 12:44 +0100, Daniel Thompson wrote:
> On 29/09/15 04:01, Chunfeng Yun wrote:
> > There some vendor quirks for MTK xhci host controller:
> > 1. It defines some extra SW scheduling parameters for HW
> >    to minimize the scheduling effort for synchronous and
> >    interrupt endpoints. The parameters are put into reseved
> >    DWs of slot context and endpoint context.
> > 2. Its IMODI unit for Interrupter Moderation register is
> >    8 times as much as that defined in xHCI spec.
> > 3. Its TDS in  Normal TRB defines a number of packets that
> >    remains to be transferred for a TD after processing all
> >    Max packets in all previous TRBs.
> >
> > Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
> 
> I've done some basic soak tests, both with a directly attached USB3 HDD 
> and, given the extra code to manage isochronous xfer, also with a hub, 
> disc and two audio interfaces.
> 
> Tested-by: Daniel Thompson <daniel.thompson@linaro.org>
> 
> 
> > diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
> > index 57f40a1..243f696 100644
> > --- a/drivers/usb/host/xhci-ring.c
> > +++ b/drivers/usb/host/xhci-ring.c
> > @@ -68,6 +68,7 @@
> >   #include <linux/slab.h>
> >   #include "xhci.h"
> >   #include "xhci-trace.h"
> > +#include "xhci-mtk.h"
> >
> >   /*
> >    * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
> > @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
> >   			      struct urb *urb, unsigned int num_trbs_left)
> >   {
> >   	u32 maxp, total_packet_count;
> > +	u32 skip_current_trb = 0;
> 
> A bit of a nitpick but why do we need skip_current_trb? Testing 
> (xhci->quirks & XHCI_MTK_HOST) twice would make what the code does, and 
> why, more obvious.
> 
I am not sure which way it is better, so add variable of
skip_current_trb to reduce test times.

> Anyhow with that looked at, and the caveat that I'm not much of USB 
> expert, you're welcome to add my Reviewed-by: to v10.
> 

Thanks a lot
> >
> > -	if (xhci->hci_version < 0x100)
> > -		return ((td_total_len - transferred) >> 10);
> > -
> > -	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> > -	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> > +	if (xhci->hci_version < 0x100) {
> > +		/* for MTK xHCI, TD size doesn't include this TRB */
> > +		if (xhci->quirks & XHCI_MTK_HOST)
> > +			skip_current_trb = 1;
> > +		else
> > +			return ((td_total_len - transferred) >> 10);
> > +	}
> >
> >   	/* One TRB with a zero-length data packet. */
> >   	if (num_trbs_left == 0 || (transferred == 0 && trb_buff_len == 0) ||
> >   	    trb_buff_len == td_total_len)
> >   		return 0;
> >
> > +	if (skip_current_trb)
> > +		trb_buff_len = 0;
> > +
> > +	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> > +	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> > +
> >   	/* Queueing functions don't count the current TRB into transferred */
> >   	return (total_packet_count - ((transferred + trb_buff_len) / maxp));
> >   }
> 
> 
> Daniel.
>
Daniel Thompson Oct. 8, 2015, 12:28 p.m. UTC | #3
On 08/10/15 13:05, chunfeng yun wrote:
> Hi,
> On Thu, 2015-10-01 at 12:44 +0100, Daniel Thompson wrote:
>> On 29/09/15 04:01, Chunfeng Yun wrote:
>>> There some vendor quirks for MTK xhci host controller:
>>> 1. It defines some extra SW scheduling parameters for HW
>>>     to minimize the scheduling effort for synchronous and
>>>     interrupt endpoints. The parameters are put into reseved
>>>     DWs of slot context and endpoint context.
>>> 2. Its IMODI unit for Interrupter Moderation register is
>>>     8 times as much as that defined in xHCI spec.
>>> 3. Its TDS in  Normal TRB defines a number of packets that
>>>     remains to be transferred for a TD after processing all
>>>     Max packets in all previous TRBs.
>>>
>>> Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
>>
>> I've done some basic soak tests, both with a directly attached USB3 HDD
>> and, given the extra code to manage isochronous xfer, also with a hub,
>> disc and two audio interfaces.
>>
>> Tested-by: Daniel Thompson <daniel.thompson@linaro.org>
>>
>>
>>> diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
>>> index 57f40a1..243f696 100644
>>> --- a/drivers/usb/host/xhci-ring.c
>>> +++ b/drivers/usb/host/xhci-ring.c
>>> @@ -68,6 +68,7 @@
>>>    #include <linux/slab.h>
>>>    #include "xhci.h"
>>>    #include "xhci-trace.h"
>>> +#include "xhci-mtk.h"
>>>
>>>    /*
>>>     * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
>>> @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
>>>    			      struct urb *urb, unsigned int num_trbs_left)
>>>    {
>>>    	u32 maxp, total_packet_count;
>>> +	u32 skip_current_trb = 0;
>>
>> A bit of a nitpick but why do we need skip_current_trb? Testing
>> (xhci->quirks & XHCI_MTK_HOST) twice would make what the code does, and
>> why, more obvious.
>>
> I am not sure which way it is better, so add variable of
> skip_current_trb to reduce test times.

I can't imagine either approach impacts performance. However introducing 
skip_current_trb makes the flow of the branches harder to follow.

The first branch was a version check and can remain so; we only need to 
do something special with the quirks because the MTK controller is 0.97 
but with a few bits added:

     /* MTK xHCI is mostly 0.97 but contains some features from 1.0 */
     if (xhci->hci_version < 0x100 && !(xhci->quirks & XHCI_MTK_HOST))
	    return ...

Setting trb_buff_len to zero is accommodating a quirk and this is 
clearer when we have:

     /* for MTK xHCI, TD size doesn't include this TRB */
     if (xhci->quirks & XHCI_MTK_HOST)
	    trb_buff_len = 0;


>> Anyhow with that looked at, and the caveat that I'm not much of USB
>> expert, you're welcome to add my Reviewed-by: to v10.
>>
>
> Thanks a lot

No worries.
Mathias Nyman Oct. 8, 2015, 1:05 p.m. UTC | #4
On 08.10.2015 15:28, Daniel Thompson wrote:
> On 08/10/15 13:05, chunfeng yun wrote:
>> Hi,
>> On Thu, 2015-10-01 at 12:44 +0100, Daniel Thompson wrote:
>>> On 29/09/15 04:01, Chunfeng Yun wrote:
>>>> There some vendor quirks for MTK xhci host controller:
>>>> 1. It defines some extra SW scheduling parameters for HW
>>>>     to minimize the scheduling effort for synchronous and
>>>>     interrupt endpoints. The parameters are put into reseved
>>>>     DWs of slot context and endpoint context.
>>>> 2. Its IMODI unit for Interrupter Moderation register is
>>>>     8 times as much as that defined in xHCI spec.
>>>> 3. Its TDS in  Normal TRB defines a number of packets that
>>>>     remains to be transferred for a TD after processing all
>>>>     Max packets in all previous TRBs.
>>>>
>>>> Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
>>>
>>> I've done some basic soak tests, both with a directly attached USB3 HDD
>>> and, given the extra code to manage isochronous xfer, also with a hub,
>>> disc and two audio interfaces.
>>>
>>> Tested-by: Daniel Thompson <daniel.thompson@linaro.org>
>>>
>>>
>>>> diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
>>>> index 57f40a1..243f696 100644
>>>> --- a/drivers/usb/host/xhci-ring.c
>>>> +++ b/drivers/usb/host/xhci-ring.c
>>>> @@ -68,6 +68,7 @@
>>>>    #include <linux/slab.h>
>>>>    #include "xhci.h"
>>>>    #include "xhci-trace.h"
>>>> +#include "xhci-mtk.h"
>>>>
>>>>    /*
>>>>     * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
>>>> @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
>>>>                      struct urb *urb, unsigned int num_trbs_left)
>>>>    {
>>>>        u32 maxp, total_packet_count;
>>>> +    u32 skip_current_trb = 0;
>>>
>>> A bit of a nitpick but why do we need skip_current_trb? Testing
>>> (xhci->quirks & XHCI_MTK_HOST) twice would make what the code does, and
>>> why, more obvious.
>>>
>> I am not sure which way it is better, so add variable of
>> skip_current_trb to reduce test times.
>
> I can't imagine either approach impacts performance. However introducing skip_current_trb makes the flow of the branches harder to follow.
>
> The first branch was a version check and can remain so; we only need to do something special with the quirks because the MTK controller is 0.97 but with a few bits added:
>
>      /* MTK xHCI is mostly 0.97 but contains some features from 1.0 */
>      if (xhci->hci_version < 0x100 && !(xhci->quirks & XHCI_MTK_HOST))
>          return ...
>
> Setting trb_buff_len to zero is accommodating a quirk and this is clearer when we have:
>
>      /* for MTK xHCI, TD size doesn't include this TRB */
>      if (xhci->quirks & XHCI_MTK_HOST)
>          trb_buff_len = 0;

Both will work, but I'd prefer this solution as well.

-Mathias
Chunfeng Yun (云春峰) Oct. 8, 2015, 2:38 p.m. UTC | #5
On Thu, 2015-10-08 at 13:28 +0100, Daniel Thompson wrote:
> On 08/10/15 13:05, chunfeng yun wrote:
> > Hi,
> > On Thu, 2015-10-01 at 12:44 +0100, Daniel Thompson wrote:
> >> On 29/09/15 04:01, Chunfeng Yun wrote:
> >>> There some vendor quirks for MTK xhci host controller:
> >>> 1. It defines some extra SW scheduling parameters for HW
> >>>     to minimize the scheduling effort for synchronous and
> >>>     interrupt endpoints. The parameters are put into reseved
> >>>     DWs of slot context and endpoint context.
> >>> 2. Its IMODI unit for Interrupter Moderation register is
> >>>     8 times as much as that defined in xHCI spec.
> >>> 3. Its TDS in  Normal TRB defines a number of packets that
> >>>     remains to be transferred for a TD after processing all
> >>>     Max packets in all previous TRBs.
> >>>
> >>> Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
> >>
> >> I've done some basic soak tests, both with a directly attached USB3 HDD
> >> and, given the extra code to manage isochronous xfer, also with a hub,
> >> disc and two audio interfaces.
> >>
> >> Tested-by: Daniel Thompson <daniel.thompson@linaro.org>
> >>
> >>
> >>> diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
> >>> index 57f40a1..243f696 100644
> >>> --- a/drivers/usb/host/xhci-ring.c
> >>> +++ b/drivers/usb/host/xhci-ring.c
> >>> @@ -68,6 +68,7 @@
> >>>    #include <linux/slab.h>
> >>>    #include "xhci.h"
> >>>    #include "xhci-trace.h"
> >>> +#include "xhci-mtk.h"
> >>>
> >>>    /*
> >>>     * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
> >>> @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
> >>>    			      struct urb *urb, unsigned int num_trbs_left)
> >>>    {
> >>>    	u32 maxp, total_packet_count;
> >>> +	u32 skip_current_trb = 0;
> >>
> >> A bit of a nitpick but why do we need skip_current_trb? Testing
> >> (xhci->quirks & XHCI_MTK_HOST) twice would make what the code does, and
> >> why, more obvious.
> >>
> > I am not sure which way it is better, so add variable of
> > skip_current_trb to reduce test times.
> 
> I can't imagine either approach impacts performance. However introducing 
> skip_current_trb makes the flow of the branches harder to follow.
> 
> The first branch was a version check and can remain so; we only need to 
> do something special with the quirks because the MTK controller is 0.97 
> but with a few bits added:
> 
>      /* MTK xHCI is mostly 0.97 but contains some features from 1.0 */
>      if (xhci->hci_version < 0x100 && !(xhci->quirks & XHCI_MTK_HOST))
> 	    return ...
> 
> Setting trb_buff_len to zero is accommodating a quirk and this is 
> clearer when we have:
> 
>      /* for MTK xHCI, TD size doesn't include this TRB */
>      if (xhci->quirks & XHCI_MTK_HOST)
> 	    trb_buff_len = 0;
> 
I will revise the code according to your suggestion, thanks

> 
> >> Anyhow with that looked at, and the caveat that I'm not much of USB
> >> expert, you're welcome to add my Reviewed-by: to v10.
> >>
> >
> > Thanks a lot
> 
> No worries.
>
Chunfeng Yun (云春峰) Oct. 8, 2015, 2:42 p.m. UTC | #6
Hi,
On Thu, 2015-10-08 at 16:05 +0300, Mathias Nyman wrote:
> On 08.10.2015 15:28, Daniel Thompson wrote:
> > On 08/10/15 13:05, chunfeng yun wrote:
> >> Hi,
> >> On Thu, 2015-10-01 at 12:44 +0100, Daniel Thompson wrote:
> >>> On 29/09/15 04:01, Chunfeng Yun wrote:
> >>>> There some vendor quirks for MTK xhci host controller:
> >>>> 1. It defines some extra SW scheduling parameters for HW
> >>>>     to minimize the scheduling effort for synchronous and
> >>>>     interrupt endpoints. The parameters are put into reseved
> >>>>     DWs of slot context and endpoint context.
> >>>> 2. Its IMODI unit for Interrupter Moderation register is
> >>>>     8 times as much as that defined in xHCI spec.
> >>>> 3. Its TDS in  Normal TRB defines a number of packets that
> >>>>     remains to be transferred for a TD after processing all
> >>>>     Max packets in all previous TRBs.
> >>>>
> >>>> Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
> >>>
> >>> I've done some basic soak tests, both with a directly attached USB3 HDD
> >>> and, given the extra code to manage isochronous xfer, also with a hub,
> >>> disc and two audio interfaces.
> >>>
> >>> Tested-by: Daniel Thompson <daniel.thompson@linaro.org>
> >>>
> >>>
> >>>> diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
> >>>> index 57f40a1..243f696 100644
> >>>> --- a/drivers/usb/host/xhci-ring.c
> >>>> +++ b/drivers/usb/host/xhci-ring.c
> >>>> @@ -68,6 +68,7 @@
> >>>>    #include <linux/slab.h>
> >>>>    #include "xhci.h"
> >>>>    #include "xhci-trace.h"
> >>>> +#include "xhci-mtk.h"
> >>>>
> >>>>    /*
> >>>>     * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
> >>>> @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
> >>>>                      struct urb *urb, unsigned int num_trbs_left)
> >>>>    {
> >>>>        u32 maxp, total_packet_count;
> >>>> +    u32 skip_current_trb = 0;
> >>>
> >>> A bit of a nitpick but why do we need skip_current_trb? Testing
> >>> (xhci->quirks & XHCI_MTK_HOST) twice would make what the code does, and
> >>> why, more obvious.
> >>>
> >> I am not sure which way it is better, so add variable of
> >> skip_current_trb to reduce test times.
> >
> > I can't imagine either approach impacts performance. However introducing skip_current_trb makes the flow of the branches harder to follow.
> >
> > The first branch was a version check and can remain so; we only need to do something special with the quirks because the MTK controller is 0.97 but with a few bits added:
> >
> >      /* MTK xHCI is mostly 0.97 but contains some features from 1.0 */
> >      if (xhci->hci_version < 0x100 && !(xhci->quirks & XHCI_MTK_HOST))
> >          return ...
> >
> > Setting trb_buff_len to zero is accommodating a quirk and this is clearer when we have:
> >
> >      /* for MTK xHCI, TD size doesn't include this TRB */
> >      if (xhci->quirks & XHCI_MTK_HOST)
> >          trb_buff_len = 0;
> 
> Both will work, but I'd prefer this solution as well.
> 
Thanks for your suggestion as well

> -Mathias
>
Mathias Nyman Oct. 15, 2015, 2:46 p.m. UTC | #7
On 29.09.2015 06:01, Chunfeng Yun wrote:
> There some vendor quirks for MTK xhci host controller:
> 1. It defines some extra SW scheduling parameters for HW
>    to minimize the scheduling effort for synchronous and
>    interrupt endpoints. The parameters are put into reseved
>    DWs of slot context and endpoint context.
> 2. Its IMODI unit for Interrupter Moderation register is
>    8 times as much as that defined in xHCI spec.
> 3. Its TDS in  Normal TRB defines a number of packets that
>    remains to be transferred for a TD after processing all
>    Max packets in all previous TRBs.
>
> Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>

Looks good in my opinion, There's one part about how we split the ISOC transafers
across multiple microframes that I don't fully understand:

> +	} else if (udev->speed == USB_SPEED_SUPER) {
> +		/* usb3_r1 spec section4.4.7 & 4.4.8 */
> +		sch_ep->cs_count = 0;
> +		esit_pkts = (mult + 1) * (max_burst + 1);

So the device would like the transfers to be esit_pkts packets every esit microframe apart.

> +		if (ep_type == INT_IN_EP || ep_type == INT_OUT_EP) {
> +			sch_ep->pkts = esit_pkts;
> +			sch_ep->num_budget_microframes = 1;
> +			sch_ep->repeat = 0;
> +		}
> +
> +		if (ep_type == ISOC_IN_EP || ep_type == ISOC_OUT_EP) {
> +			if (esit_pkts <= sch_ep->esit)
> +				sch_ep->pkts = 1;
> +			else
> +				sch_ep->pkts = roundup_pow_of_two(esit_pkts)
> +					/ sch_ep->esit;

sch_ep->pkts now contain a rounded up average value of how many packets per microframe
would be transferred in case we'd transfer packets every microframe instead of every esit frame apart.

> +
> +			sch_ep->num_budget_microframes =
> +				DIV_ROUND_UP(esit_pkts, sch_ep->pkts);

sch_ep->num_budget_microframes now contains the number of microframes during an esit needed
to transer all data if each microframe contains the average amount of data.

So basically we are trying to use as many microframes as possible with as few packets
per microframe as possible.

Did I understand this correctly?
How will devices react if they expect to get 16 packets every 16th microframe,
but they get one packet every microframe instead?

Or is this just theoretical bw calculations for the host and it will in the end do
whatever it wants with the information.

Otherwise, by making the changes Daniel Thompson suggested I think the xhci parts looks ready.

-Mathias

  

   
  

















> +	} else if (is_fs_or_ls(udev->speed)) {
> +
> +		/*
> +		 * usb_20 spec section11.18.4
> +		 * assume worst cases
> +		 */
> +		sch_ep->repeat = 0;
> +		sch_ep->pkts = 1; /* at most one packet for each microframe */
> +		if (ep_type == INT_IN_EP || ep_type == INT_OUT_EP) {
> +			sch_ep->cs_count = 3; /* at most need 3 CS*/
> +			/* one for SS and one for budgeted transaction */
> +			sch_ep->num_budget_microframes = sch_ep->cs_count + 2;
> +			sch_ep->bw_cost_per_microframe = max_packet_size;
> +		}
> +		if (ep_type == ISOC_OUT_EP) {
> +
> +			/*
> +			 * the best case FS budget assumes that 188 FS bytes
> +			 * occur in each microframe
> +			 */
> +			sch_ep->num_budget_microframes = DIV_ROUND_UP(
> +				max_packet_size, FS_PAYLOAD_MAX);
> +			sch_ep->bw_cost_per_microframe = FS_PAYLOAD_MAX;
> +			sch_ep->cs_count = sch_ep->num_budget_microframes;
> +		}
> +		if (ep_type == ISOC_IN_EP) {
> +			/* at most need additional two CS. */
> +			sch_ep->cs_count = DIV_ROUND_UP(
> +				max_packet_size, FS_PAYLOAD_MAX) + 2;
> +			sch_ep->num_budget_microframes = sch_ep->cs_count + 2;
> +			sch_ep->bw_cost_per_microframe = FS_PAYLOAD_MAX;
> +		}
> +	}
> +}
> +
> +/* Get maximum bandwidth when we schedule at offset slot. */
> +static u32 get_max_bw(struct mu3h_sch_bw_info *sch_bw,
> +	struct mu3h_sch_ep_info *sch_ep, u32 offset)
> +{
> +	u32 num_esit;
> +	u32 max_bw = 0;
> +	int i;
> +	int j;
> +
> +	num_esit = XHCI_MTK_MAX_ESIT / sch_ep->esit;
> +	for (i = 0; i < num_esit; i++) {
> +		u32 base = offset + i * sch_ep->esit;
> +
> +		for (j = 0; j < sch_ep->num_budget_microframes; j++) {
> +			if (sch_bw->bus_bw[base + j] > max_bw)
> +				max_bw = sch_bw->bus_bw[base + j];
> +		}
> +	}
> +	return max_bw;
> +}
> +
> +static void update_bus_bw(struct mu3h_sch_bw_info *sch_bw,
> +	struct mu3h_sch_ep_info *sch_ep, int bw_cost)
> +{
> +	u32 num_esit;
> +	u32 base;
> +	int i;
> +	int j;
> +
> +	num_esit = XHCI_MTK_MAX_ESIT / sch_ep->esit;
> +	for (i = 0; i < num_esit; i++) {
> +		base = sch_ep->offset + i * sch_ep->esit;
> +		for (j = 0; j < sch_ep->num_budget_microframes; j++)
> +			sch_bw->bus_bw[base + j] += bw_cost;
> +	}
> +}
> +
> +static int check_sch_bw(struct usb_device *udev,
> +	struct mu3h_sch_bw_info *sch_bw, struct mu3h_sch_ep_info *sch_ep)
> +{
> +	u32 offset;
> +	u32 esit;
> +	u32 num_budget_microframes;
> +	u32 min_bw;
> +	u32 min_index;
> +	u32 worst_bw;
> +	u32 bw_boundary;
> +
> +	if (sch_ep->esit > XHCI_MTK_MAX_ESIT)
> +		sch_ep->esit = XHCI_MTK_MAX_ESIT;
> +
> +	esit = sch_ep->esit;
> +	num_budget_microframes = sch_ep->num_budget_microframes;
> +
> +	/*
> +	 * Search through all possible schedule microframes.
> +	 * and find a microframe where its worst bandwidth is minimum.
> +	 */
> +	min_bw = ~0;
> +	min_index = 0;
> +	for (offset = 0; offset < esit; offset++) {
> +		if ((offset + num_budget_microframes) > sch_ep->esit)
> +			break;
> +
> +		/*
> +		 * usb_20 spec section11.18:
> +		 * must never schedule Start-Split in Y6
> +		 */
> +		if (is_fs_or_ls(udev->speed) && (offset % 8 == 6))
> +			continue;
> +
> +		worst_bw = get_max_bw(sch_bw, sch_ep, offset);
> +		if (min_bw > worst_bw) {
> +			min_bw = worst_bw;
> +			min_index = offset;
> +		}
> +		if (min_bw == 0)
> +			break;
> +	}
> +	sch_ep->offset = min_index;
> +
> +	bw_boundary = (udev->speed == USB_SPEED_SUPER)
> +				? SS_BW_BOUNDARY : HS_BW_BOUNDARY;
> +
> +	/* check bandwidth */
> +	if (min_bw + sch_ep->bw_cost_per_microframe > bw_boundary)
> +		return -ERANGE;
> +
> +	/* update bus bandwidth info */
> +	update_bus_bw(sch_bw, sch_ep, sch_ep->bw_cost_per_microframe);
> +
> +	return 0;
> +}
> +
> +static bool need_bw_sch(struct usb_host_endpoint *ep,
> +	enum usb_device_speed speed, int has_tt)
> +{
> +	/* only for periodic endpoints */
> +	if (usb_endpoint_xfer_control(&ep->desc)
> +		|| usb_endpoint_xfer_bulk(&ep->desc))
> +		return false;
> +
> +	/*
> +	 * for LS & FS periodic endpoints which its device don't attach
> +	 * to TT are also ignored, root-hub will schedule them directly
> +	 */
> +	if (is_fs_or_ls(speed) && !has_tt)
> +		return false;
> +
> +	return true;
> +}
> +
> +int xhci_mtk_sch_init(struct xhci_hcd *xhci)
> +{
> +	struct usb_hcd *hcd = xhci_to_hcd(xhci);
> +	struct device *dev = hcd->self.controller;
> +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> +	struct mu3h_sch_bw_info *sch_array;
> +	int num_usb_bus;
> +	int i;
> +
> +	/* ss IN and OUT are separated */
> +	num_usb_bus = xhci->num_usb3_ports * 2 + xhci->num_usb2_ports;
> +
> +	sch_array = kcalloc(num_usb_bus, sizeof(*sch_array), GFP_KERNEL);
> +	if (sch_array == NULL)
> +		return -ENOMEM;
> +
> +	for (i = 0; i < num_usb_bus; i++)
> +		INIT_LIST_HEAD(&sch_array[i].bw_ep_list);
> +
> +	mtk->sch_array = sch_array;
> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_GPL(xhci_mtk_sch_init);
> +
> +void xhci_mtk_sch_exit(struct xhci_hcd *xhci)
> +{
> +	struct usb_hcd *hcd = xhci_to_hcd(xhci);
> +	struct device *dev = hcd->self.controller;
> +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> +
> +	kfree(mtk->sch_array);
> +}
> +EXPORT_SYMBOL_GPL(xhci_mtk_sch_exit);
> +
> +int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> +		struct usb_host_endpoint *ep)
> +{
> +	struct device *dev = hcd->self.controller;
> +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> +	struct xhci_hcd *xhci;
> +	struct xhci_ep_ctx *ep_ctx;
> +	struct xhci_slot_ctx *slot_ctx;
> +	struct xhci_virt_device *virt_dev;
> +	struct mu3h_sch_bw_info *sch_bw;
> +	struct mu3h_sch_ep_info *sch_ep;
> +	struct mu3h_sch_bw_info *sch_array;
> +	unsigned int ep_index;
> +	int bw_index;
> +	int ret = 0;
> +
> +	xhci = hcd_to_xhci(hcd);
> +	virt_dev = xhci->devs[udev->slot_id];
> +	ep_index = xhci_get_endpoint_index(&ep->desc);
> +	slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx);
> +	ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, ep_index);
> +	sch_array = mtk->sch_array;
> +
> +	xhci_dbg(xhci, "%s() type:%d, speed:%d, mpkt:%d, dir:%d, ep:%p\n",
> +		__func__, usb_endpoint_type(&ep->desc), udev->speed,
> +		GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc)),
> +		usb_endpoint_dir_in(&ep->desc), ep);
> +
> +	if (!need_bw_sch(ep, udev->speed, slot_ctx->tt_info & TT_SLOT))
> +		return 0;
> +
> +	bw_index = get_bw_index(xhci, udev, ep);
> +	sch_bw = &sch_array[bw_index];
> +
> +	sch_ep = kzalloc(sizeof(struct mu3h_sch_ep_info), GFP_NOIO);
> +	if (!sch_ep)
> +		return -ENOMEM;
> +
> +	setup_sch_info(udev, ep_ctx, sch_ep);
> +
> +	ret = check_sch_bw(udev, sch_bw, sch_ep);
> +	if (ret) {
> +		xhci_err(xhci, "Not enough bandwidth!\n");
> +		kfree(sch_ep);
> +		return -ENOSPC;
> +	}
> +
> +	list_add_tail(&sch_ep->endpoint, &sch_bw->bw_ep_list);
> +	sch_ep->ep = ep;
> +
> +	ep_ctx->reserved[0] |= cpu_to_le32(EP_BPKTS(sch_ep->pkts)
> +		| EP_BCSCOUNT(sch_ep->cs_count) | EP_BBM(sch_ep->burst_mode));
> +	ep_ctx->reserved[1] |= cpu_to_le32(EP_BOFFSET(sch_ep->offset)
> +		| EP_BREPEAT(sch_ep->repeat));
> +
> +	xhci_dbg(xhci, " PKTS:%x, CSCOUNT:%x, BM:%x, OFFSET:%x, REPEAT:%x\n",
> +			sch_ep->pkts, sch_ep->cs_count, sch_ep->burst_mode,
> +			sch_ep->offset, sch_ep->repeat);
> +
> +	return 0;
> +}
> +EXPORT_SYMBOL_GPL(xhci_mtk_add_ep_quirk);
> +
> +void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> +		struct usb_host_endpoint *ep)
> +{
> +	struct device *dev = hcd->self.controller;
> +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> +	struct xhci_hcd *xhci;
> +	struct xhci_slot_ctx *slot_ctx;
> +	struct xhci_virt_device *virt_dev;
> +	struct mu3h_sch_bw_info *sch_array;
> +	struct mu3h_sch_bw_info *sch_bw;
> +	struct mu3h_sch_ep_info *sch_ep;
> +	int bw_index;
> +
> +	xhci = hcd_to_xhci(hcd);
> +	virt_dev = xhci->devs[udev->slot_id];
> +	slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx);
> +	sch_array = mtk->sch_array;
> +
> +	xhci_dbg(xhci, "%s() type:%d, speed:%d, mpks:%d, dir:%d, ep:%p\n",
> +		__func__, usb_endpoint_type(&ep->desc), udev->speed,
> +		GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc)),
> +		usb_endpoint_dir_in(&ep->desc), ep);
> +
> +	if (!need_bw_sch(ep, udev->speed, slot_ctx->tt_info & TT_SLOT))
> +		return;
> +
> +	bw_index = get_bw_index(xhci, udev, ep);
> +	sch_bw = &sch_array[bw_index];
> +
> +	list_for_each_entry(sch_ep, &sch_bw->bw_ep_list, endpoint) {
> +		if (sch_ep->ep == ep) {
> +			update_bus_bw(sch_bw, sch_ep,
> +				-sch_ep->bw_cost_per_microframe);
> +			list_del(&sch_ep->endpoint);
> +			kfree(sch_ep);
> +			break;
> +		}
> +	}
> +}
> +EXPORT_SYMBOL_GPL(xhci_mtk_drop_ep_quirk);
> diff --git a/drivers/usb/host/xhci-mtk.c b/drivers/usb/host/xhci-mtk.c
> new file mode 100644
> index 0000000..afb52bf
> --- /dev/null
> +++ b/drivers/usb/host/xhci-mtk.c
> @@ -0,0 +1,776 @@
> +/*
> + * Copyright (c) 2015 MediaTek Inc.
> + * Author:
> + *  Chunfeng Yun <chunfeng.yun@mediatek.com>
> + *
> + * This software is licensed under the terms of the GNU General Public
> + * License version 2, as published by the Free Software Foundation, and
> + * may be copied, distributed, and modified under those terms.
> + *
> + * This program is distributed in the hope that 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.
> + *
> + */
> +
> +#include <linux/clk.h>
> +#include <linux/dma-mapping.h>
> +#include <linux/iopoll.h>
> +#include <linux/kernel.h>
> +#include <linux/mfd/syscon.h>
> +#include <linux/module.h>
> +#include <linux/of.h>
> +#include <linux/phy/phy.h>
> +#include <linux/platform_device.h>
> +#include <linux/pm_runtime.h>
> +#include <linux/regmap.h>
> +#include <linux/regulator/consumer.h>
> +#include <linux/slab.h>
> +#include <linux/usb/phy.h>
> +#include <linux/usb/xhci_pdriver.h>
> +
> +#include "xhci.h"
> +#include "xhci-mtk.h"
> +
> +/* ip_pw_ctrl0 register */
> +#define CTRL0_IP_SW_RST	BIT(0)
> +
> +/* ip_pw_ctrl1 register */
> +#define CTRL1_IP_HOST_PDN	BIT(0)
> +
> +/* ip_pw_ctrl2 register */
> +#define CTRL2_IP_DEV_PDN	BIT(0)
> +
> +/* ip_pw_sts1 register */
> +#define STS1_IP_SLEEP_STS	BIT(30)
> +#define STS1_U3_MAC_RST	BIT(16)
> +#define STS1_SYS125_RST	BIT(10)
> +#define STS1_REF_RST		BIT(8)
> +#define STS1_SYSPLL_STABLE	BIT(0)
> +
> +/* ip_pw_sts2 register */
> +#define STS2_U2_MAC_RST	BIT(0)
> +
> +/* ip_xhci_cap register */
> +#define CAP_U3_PORT_NUM(p)	((p) & 0xff)
> +#define CAP_U2_PORT_NUM(p)	(((p) >> 8) & 0xff)
> +
> +/* u3_ctrl_p register */
> +#define CTRL_U3_PORT_HOST_SEL	BIT(2)
> +#define CTRL_U3_PORT_PDN	BIT(1)
> +#define CTRL_U3_PORT_DIS	BIT(0)
> +
> +/* u2_ctrl_p register */
> +#define CTRL_U2_PORT_HOST_SEL	BIT(2)
> +#define CTRL_U2_PORT_PDN	BIT(1)
> +#define CTRL_U2_PORT_DIS	BIT(0)
> +
> +/* u2_phy_pll register */
> +#define CTRL_U2_FORCE_PLL_STB	BIT(28)
> +
> +#define PERI_WK_CTRL0		0x400
> +#define UWK_CTR0_0P_LS_PE	BIT(8)  /* posedge */
> +#define UWK_CTR0_0P_LS_NE	BIT(7)  /* negedge for 0p linestate*/
> +#define UWK_CTL1_1P_LS_C(x)	(((x) & 0xf) << 1)
> +#define UWK_CTL1_1P_LS_E	BIT(0)
> +
> +#define PERI_WK_CTRL1		0x404
> +#define UWK_CTL1_IS_C(x)	(((x) & 0xf) << 26)
> +#define UWK_CTL1_IS_E		BIT(25)
> +#define UWK_CTL1_0P_LS_C(x)	(((x) & 0xf) << 21)
> +#define UWK_CTL1_0P_LS_E	BIT(20)
> +#define UWK_CTL1_IDDIG_C(x)	(((x) & 0xf) << 11)  /* cycle debounce */
> +#define UWK_CTL1_IDDIG_E	BIT(10) /* enable debounce */
> +#define UWK_CTL1_IDDIG_P	BIT(9)  /* polarity */
> +#define UWK_CTL1_0P_LS_P	BIT(7)
> +#define UWK_CTL1_IS_P		BIT(6)  /* polarity for ip sleep */
> +
> +enum ssusb_wakeup_src {
> +	SSUSB_WK_IP_SLEEP = 1,
> +	SSUSB_WK_LINE_STATE = 2,
> +};
> +
> +static int check_ip_clk_status(struct xhci_hcd_mtk *mtk)
> +{
> +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> +	u32 val, check_val;
> +	int ret;
> +
> +	check_val = STS1_SYSPLL_STABLE | STS1_REF_RST | STS1_SYS125_RST;
> +	check_val |= (mtk->num_u3_ports ? STS1_U3_MAC_RST : 0);
> +
> +	ret = readl_poll_timeout(&ippc->ip_pw_sts1, val,
> +			  (check_val == (val & check_val)), 100, 20000);
> +	if (ret) {
> +		dev_err(mtk->dev, "mac3 clocks are not stable (0x%x)\n", val);
> +		return ret;
> +	}
> +
> +	ret = readl_poll_timeout(&ippc->ip_pw_sts2, val,
> +			   (val & STS2_U2_MAC_RST), 100, 10000);
> +	if (ret) {
> +		dev_err(mtk->dev, "mac2 reset is still active!!!\n");
> +		return ret;
> +	}
> +
> +	return 0;
> +}
> +
> +static int xhci_mtk_ports_enable(struct xhci_hcd_mtk *mtk)
> +{
> +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> +	u32 temp;
> +	int i;
> +
> +	/* power on host ip */
> +	temp = readl(&ippc->ip_pw_ctr1);
> +	temp &= ~CTRL1_IP_HOST_PDN;
> +	writel(temp, &ippc->ip_pw_ctr1);
> +
> +	/* power on and enable all u3 ports */
> +	for (i = 0; i < mtk->num_u3_ports; i++) {
> +		temp = readl(&ippc->u3_ctrl_p[i]);
> +		temp &= ~(CTRL_U3_PORT_PDN | CTRL_U3_PORT_DIS);
> +		temp |= CTRL_U3_PORT_HOST_SEL;
> +		writel(temp, &ippc->u3_ctrl_p[i]);
> +	}
> +
> +	/* power on and enable all u2 ports */
> +	for (i = 0; i < mtk->num_u2_ports; i++) {
> +		temp = readl(&ippc->u2_ctrl_p[i]);
> +		temp &= ~(CTRL_U2_PORT_PDN | CTRL_U2_PORT_DIS);
> +		temp |= CTRL_U2_PORT_HOST_SEL;
> +		writel(temp, &ippc->u2_ctrl_p[i]);
> +	}
> +
> +	/*
> +	 * wait for clocks to be stable, and clock domains reset to
> +	 * be inactive after power on and enable ports
> +	 */
> +	return check_ip_clk_status(mtk);
> +}
> +
> +static int xhci_mtk_ports_disable(struct xhci_hcd_mtk *mtk)
> +{
> +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> +	u32 temp;
> +	int ret;
> +	int i;
> +
> +	/* power down all u3 ports */
> +	for (i = 0; i < mtk->num_u3_ports; i++) {
> +		temp = readl(&ippc->u3_ctrl_p[i]);
> +		temp |= CTRL_U3_PORT_PDN;
> +		writel(temp, &ippc->u3_ctrl_p[i]);
> +	}
> +
> +	/* power down all u2 ports */
> +	for (i = 0; i < mtk->num_u2_ports; i++) {
> +		temp = readl(&ippc->u2_ctrl_p[i]);
> +		temp |= CTRL_U2_PORT_PDN;
> +		writel(temp, &ippc->u2_ctrl_p[i]);
> +	}
> +
> +	/* power down host ip */
> +	temp = readl(&ippc->ip_pw_ctr1);
> +	temp |= CTRL1_IP_HOST_PDN;
> +	writel(temp, &ippc->ip_pw_ctr1);
> +
> +	ret = readl_poll_timeout(&ippc->ip_pw_sts1, temp,
> +			  (temp & STS1_IP_SLEEP_STS), 100, 100000);
> +	if (ret) {
> +		dev_err(mtk->dev, "ip sleep failed!!!\n");
> +		return ret;
> +	}
> +	return 0;
> +}
> +
> +static void xhci_mtk_ports_config(struct xhci_hcd_mtk *mtk)
> +{
> +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> +	u32 temp;
> +
> +	/* reset whole ip */
> +	temp = readl(&ippc->ip_pw_ctr0);
> +	temp |= CTRL0_IP_SW_RST;
> +	writel(temp, &ippc->ip_pw_ctr0);
> +	udelay(1);
> +	temp = readl(&ippc->ip_pw_ctr0);
> +	temp &= ~CTRL0_IP_SW_RST;
> +	writel(temp, &ippc->ip_pw_ctr0);
> +
> +	/*
> +	 * device ip is default power-on in fact
> +	 * power down device ip, otherwise ip-sleep will fail
> +	 */
> +	temp = readl(&ippc->ip_pw_ctr2);
> +	temp |= CTRL2_IP_DEV_PDN;
> +	writel(temp, &ippc->ip_pw_ctr2);
> +
> +	temp = readl(&ippc->ip_xhci_cap);
> +	mtk->num_u3_ports = CAP_U3_PORT_NUM(temp);
> +	mtk->num_u2_ports = CAP_U2_PORT_NUM(temp);
> +	dev_dbg(mtk->dev, "%s u2p:%d, u3p:%d\n", __func__,
> +			mtk->num_u2_ports, mtk->num_u3_ports);
> +
> +	xhci_mtk_ports_enable(mtk);
> +}
> +
> +static int xhci_mtk_clks_enable(struct xhci_hcd_mtk *mtk)
> +{
> +	int ret;
> +
> +	ret = clk_prepare_enable(mtk->sys_clk);
> +	if (ret) {
> +		dev_err(mtk->dev, "failed to enable u3phya_ref\n");
> +		goto u3phya_ref_err;
> +	}
> +	ret = clk_prepare_enable(mtk->wk_deb_p0);
> +	if (ret) {
> +		dev_err(mtk->dev, "failed to enable wk_deb_p0\n");
> +		goto usb_p0_err;
> +	}
> +	if (mtk->num_phys > 1) {
> +		ret = clk_prepare_enable(mtk->wk_deb_p1);
> +		if (ret) {
> +			dev_err(mtk->dev, "failed to enable wk_deb_p1\n");
> +			goto usb_p1_err;
> +		}
> +	}
> +	return 0;
> +
> +usb_p1_err:
> +	clk_disable_unprepare(mtk->wk_deb_p0);
> +usb_p0_err:
> +	clk_disable_unprepare(mtk->sys_clk);
> +u3phya_ref_err:
> +	return -EINVAL;
> +}
> +
> +static void xhci_mtk_clks_disable(struct xhci_hcd_mtk *mtk)
> +{
> +	if (mtk->num_phys > 1)
> +		clk_disable_unprepare(mtk->wk_deb_p1);
> +	clk_disable_unprepare(mtk->wk_deb_p0);
> +	clk_disable_unprepare(mtk->sys_clk);
> +}
> +
> +/* only clocks can be turn off for ip-sleep wakeup mode */
> +static void usb_wakeup_ip_sleep_en(struct xhci_hcd_mtk *mtk)
> +{
> +	u32 tmp;
> +	struct regmap *pericfg = mtk->pericfg;
> +
> +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> +	tmp &= ~UWK_CTL1_IS_P;
> +	tmp &= ~(UWK_CTL1_IS_C(0xf));
> +	tmp |= UWK_CTL1_IS_C(0x8);
> +	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
> +	regmap_write(pericfg, PERI_WK_CTRL1, tmp | UWK_CTL1_IS_E);
> +
> +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> +	dev_dbg(mtk->dev, "%s(): WK_CTRL1[P6,E25,C26:29]=%#x\n",
> +		__func__, tmp);
> +}
> +
> +static void usb_wakeup_ip_sleep_dis(struct xhci_hcd_mtk *mtk)
> +{
> +	u32 tmp;
> +
> +	regmap_read(mtk->pericfg, PERI_WK_CTRL1, &tmp);
> +	tmp &= ~UWK_CTL1_IS_E;
> +	regmap_write(mtk->pericfg, PERI_WK_CTRL1, tmp);
> +}
> +
> +/*
> +* for line-state wakeup mode, phy's power should not power-down
> +* and only support cable plug in/out
> +*/
> +static void usb_wakeup_line_state_en(struct xhci_hcd_mtk *mtk)
> +{
> +	u32 tmp;
> +	struct regmap *pericfg = mtk->pericfg;
> +
> +	/* line-state of u2-port0 */
> +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> +	tmp &= ~UWK_CTL1_0P_LS_P;
> +	tmp &= ~(UWK_CTL1_0P_LS_C(0xf));
> +	tmp |= UWK_CTL1_0P_LS_C(0x8);
> +	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
> +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> +	regmap_write(pericfg, PERI_WK_CTRL1, tmp | UWK_CTL1_0P_LS_E);
> +
> +	/* line-state of u2-port1 if support */
> +	if (mtk->num_phys > 1) {
> +		regmap_read(pericfg, PERI_WK_CTRL0, &tmp);
> +		tmp &= ~(UWK_CTL1_1P_LS_C(0xf));
> +		tmp |= UWK_CTL1_1P_LS_C(0x8);
> +		regmap_write(pericfg, PERI_WK_CTRL0, tmp);
> +		regmap_write(pericfg, PERI_WK_CTRL0, tmp | UWK_CTL1_1P_LS_E);
> +	}
> +}
> +
> +static void usb_wakeup_line_state_dis(struct xhci_hcd_mtk *mtk)
> +{
> +	u32 tmp;
> +	struct regmap *pericfg = mtk->pericfg;
> +
> +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> +	tmp &= ~UWK_CTL1_0P_LS_E;
> +	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
> +
> +	if (mtk->num_phys > 1) {
> +		regmap_read(pericfg, PERI_WK_CTRL0, &tmp);
> +		tmp &= ~UWK_CTL1_1P_LS_E;
> +		regmap_write(pericfg, PERI_WK_CTRL0, tmp);
> +	}
> +}
> +
> +static void usb_wakeup_enable(struct xhci_hcd_mtk *mtk)
> +{
> +	if (mtk->wakeup_src == SSUSB_WK_IP_SLEEP)
> +		usb_wakeup_ip_sleep_en(mtk);
> +	else if (mtk->wakeup_src == SSUSB_WK_LINE_STATE)
> +		usb_wakeup_line_state_en(mtk);
> +}
> +
> +static void usb_wakeup_disable(struct xhci_hcd_mtk *mtk)
> +{
> +	if (mtk->wakeup_src == SSUSB_WK_IP_SLEEP)
> +		usb_wakeup_ip_sleep_dis(mtk);
> +	else if (mtk->wakeup_src == SSUSB_WK_LINE_STATE)
> +		usb_wakeup_line_state_dis(mtk);
> +}
> +
> +static int usb_wakeup_of_property_parse(struct xhci_hcd_mtk *mtk,
> +				struct device_node *dn)
> +{
> +	struct device *dev = mtk->dev;
> +
> +	/*
> +	* wakeup function is optional, so it is not an error if this property
> +	* does not exist, and in such case, no need to get relative
> +	* properties anymore.
> +	*/
> +	of_property_read_u32(dn, "mediatek,wakeup-src", &mtk->wakeup_src);
> +	if (!mtk->wakeup_src)
> +		return 0;
> +
> +	mtk->wk_deb_p0 = devm_clk_get(dev, "wakeup_deb_p0");
> +	if (IS_ERR(mtk->wk_deb_p0)) {
> +		dev_err(dev, "fail to get wakeup_deb_p0\n");
> +		return PTR_ERR(mtk->wk_deb_p0);
> +	}
> +
> +	mtk->wk_deb_p1 = devm_clk_get(dev, "wakeup_deb_p1");
> +	if (IS_ERR(mtk->wk_deb_p1)) {
> +		dev_err(dev, "fail to get wakeup_deb_p1\n");
> +		return PTR_ERR(mtk->wk_deb_p1);
> +	}
> +
> +	mtk->pericfg = syscon_regmap_lookup_by_phandle(dn,
> +						"mediatek,syscon-wakeup");
> +	if (IS_ERR(mtk->pericfg)) {
> +		dev_err(dev, "fail to get pericfg regs\n");
> +		return PTR_ERR(mtk->pericfg);
> +	}
> +
> +	return 0;
> +}
> +
> +static int xhci_mtk_setup(struct usb_hcd *hcd);
> +static const struct xhci_driver_overrides xhci_mtk_overrides __initconst = {
> +	.extra_priv_size = sizeof(struct xhci_hcd),
> +	.reset = xhci_mtk_setup,
> +};
> +
> +static struct hc_driver __read_mostly xhci_mtk_hc_driver;
> +
> +static int xhci_mtk_phy_init(struct xhci_hcd_mtk *mtk)
> +{
> +	int i;
> +	int ret;
> +
> +	for (i = 0; i < mtk->num_phys; i++) {
> +		ret = phy_init(mtk->phys[i]);
> +		if (ret)
> +			goto exit_phy;
> +	}
> +	return 0;
> +
> +exit_phy:
> +	for (; i > 0; i--)
> +		phy_exit(mtk->phys[i - 1]);
> +
> +	return ret;
> +}
> +
> +static int xhci_mtk_phy_exit(struct xhci_hcd_mtk *mtk)
> +{
> +	int i;
> +
> +	for (i = 0; i < mtk->num_phys; i++)
> +		phy_exit(mtk->phys[i]);
> +
> +	return 0;
> +}
> +
> +static int xhci_mtk_phy_power_on(struct xhci_hcd_mtk *mtk)
> +{
> +	int i;
> +	int ret;
> +
> +	for (i = 0; i < mtk->num_phys; i++) {
> +		ret = phy_power_on(mtk->phys[i]);
> +		if (ret)
> +			goto power_off_phy;
> +	}
> +	return 0;
> +
> +power_off_phy:
> +	for (; i > 0; i--)
> +		phy_power_off(mtk->phys[i - 1]);
> +
> +	return ret;
> +}
> +
> +static void xhci_mtk_phy_power_off(struct xhci_hcd_mtk *mtk)
> +{
> +	unsigned int i;
> +
> +	for (i = 0; i < mtk->num_phys; i++)
> +		phy_power_off(mtk->phys[i]);
> +}
> +
> +static int xhci_mtk_ldos_enable(struct xhci_hcd_mtk *mtk)
> +{
> +	int ret;
> +
> +	ret = regulator_enable(mtk->vbus);
> +	if (ret) {
> +		dev_err(mtk->dev, "failed to enable vbus\n");
> +		return ret;
> +	}
> +
> +	ret = regulator_enable(mtk->vusb33);
> +	if (ret) {
> +		dev_err(mtk->dev, "failed to enable vusb33\n");
> +		regulator_disable(mtk->vbus);
> +		return ret;
> +	}
> +	return 0;
> +}
> +
> +static void xhci_mtk_ldos_disable(struct xhci_hcd_mtk *mtk)
> +{
> +	regulator_disable(mtk->vbus);
> +	regulator_disable(mtk->vusb33);
> +}
> +
> +static void xhci_mtk_quirks(struct device *dev, struct xhci_hcd *xhci)
> +{
> +	/*
> +	 * As of now platform drivers don't provide MSI support so we ensure
> +	 * here that the generic code does not try to make a pci_dev from our
> +	 * dev struct in order to setup MSI
> +	 */
> +	xhci->quirks |= XHCI_PLAT;
> +	xhci->quirks |= XHCI_MTK_HOST;
> +	/*
> +	 * MTK host controller gives a spurious successful event after a
> +	 * short transfer. Ignore it.
> +	 */
> +	xhci->quirks |= XHCI_SPURIOUS_SUCCESS;
> +}
> +
> +/* called during probe() after chip reset completes */
> +static int xhci_mtk_setup(struct usb_hcd *hcd)
> +{
> +	struct xhci_hcd *xhci;
> +	int ret;
> +
> +	ret = xhci_gen_setup(hcd, xhci_mtk_quirks);
> +	if (ret)
> +		return ret;
> +
> +	if (!usb_hcd_is_primary_hcd(hcd))
> +		return 0;
> +
> +	xhci = hcd_to_xhci(hcd);
> +	return xhci_mtk_sch_init(xhci);
> +}
> +
> +
> +static int xhci_mtk_probe(struct platform_device *pdev)
> +{
> +	struct device *dev = &pdev->dev;
> +	struct device_node *node = dev->of_node;
> +	struct usb_xhci_pdata *pdata = dev_get_platdata(dev);
> +	struct xhci_hcd_mtk *mtk;
> +	const struct hc_driver *driver;
> +	struct xhci_hcd *xhci;
> +	struct resource *res;
> +	struct usb_hcd *hcd;
> +	struct phy *phy;
> +	int phy_num;
> +	int ret = -ENODEV;
> +	int irq;
> +
> +	if (usb_disabled())
> +		return -ENODEV;
> +
> +	driver = &xhci_mtk_hc_driver;
> +	mtk = devm_kzalloc(dev, sizeof(*mtk), GFP_KERNEL);
> +	if (!mtk)
> +		return -ENOMEM;
> +
> +	mtk->dev = dev;
> +	mtk->vbus = devm_regulator_get(dev, "vbus");
> +	if (IS_ERR(mtk->vbus)) {
> +		dev_err(dev, "fail to get vbus\n");
> +		return PTR_ERR(mtk->vbus);
> +	}
> +
> +	mtk->vusb33 = devm_regulator_get(dev, "vusb33");
> +	if (IS_ERR(mtk->vusb33)) {
> +		dev_err(dev, "fail to get vusb33\n");
> +		return PTR_ERR(mtk->vusb33);
> +	}
> +
> +	mtk->sys_clk = devm_clk_get(dev, "sys_ck");
> +	if (IS_ERR(mtk->sys_clk)) {
> +		dev_err(dev, "fail to get sys_ck\n");
> +		return PTR_ERR(mtk->sys_clk);
> +	}
> +
> +	ret = usb_wakeup_of_property_parse(mtk, node);
> +	if (ret)
> +		return ret;
> +
> +	mtk->num_phys = of_count_phandle_with_args(node,
> +			"phys", "#phy-cells");
> +	if (mtk->num_phys > 0) {
> +		mtk->phys = devm_kcalloc(dev, mtk->num_phys,
> +					sizeof(*mtk->phys), GFP_KERNEL);
> +		if (!mtk->phys)
> +			return -ENOMEM;
> +	} else {
> +		mtk->num_phys = 0;
> +	}
> +	pm_runtime_enable(dev);
> +	pm_runtime_get_sync(dev);
> +
> +	ret = xhci_mtk_ldos_enable(mtk);
> +	if (ret)
> +		goto disable_pm;
> +
> +	ret = xhci_mtk_clks_enable(mtk);
> +	if (ret)
> +		goto disable_ldos;
> +
> +	irq = platform_get_irq(pdev, 0);
> +	if (irq < 0)
> +		goto disable_clk;
> +
> +	/* Initialize dma_mask and coherent_dma_mask to 32-bits */
> +	ret = dma_set_coherent_mask(dev, DMA_BIT_MASK(32));
> +	if (ret)
> +		goto disable_clk;
> +
> +	if (!dev->dma_mask)
> +		dev->dma_mask = &dev->coherent_dma_mask;
> +	else
> +		dma_set_mask(dev, DMA_BIT_MASK(32));
> +
> +	hcd = usb_create_hcd(driver, dev, dev_name(dev));
> +	if (!hcd) {
> +		ret = -ENOMEM;
> +		goto disable_clk;
> +	}
> +
> +	/*
> +	 * USB 2.0 roothub is stored in the platform_device.
> +	 * Swap it with mtk HCD.
> +	 */
> +	mtk->hcd = platform_get_drvdata(pdev);
> +	platform_set_drvdata(pdev, mtk);
> +
> +	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> +	hcd->regs = devm_ioremap_resource(dev, res);
> +	if (IS_ERR(hcd->regs)) {
> +		ret = PTR_ERR(hcd->regs);
> +		goto put_usb2_hcd;
> +	}
> +	hcd->rsrc_start = res->start;
> +	hcd->rsrc_len = resource_size(res);
> +
> +	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> +	mtk->ippc_regs = devm_ioremap_resource(dev, res);
> +	if (IS_ERR(mtk->ippc_regs)) {
> +		ret = PTR_ERR(mtk->ippc_regs);
> +		goto put_usb2_hcd;
> +	}
> +
> +	for (phy_num = 0; phy_num < mtk->num_phys; phy_num++) {
> +		phy = devm_of_phy_get_by_index(dev, node, phy_num);
> +		if (IS_ERR(phy)) {
> +			ret = PTR_ERR(phy);
> +			goto put_usb2_hcd;
> +		}
> +		mtk->phys[phy_num] = phy;
> +	}
> +
> +	xhci_mtk_ports_config(mtk);
> +	ret = xhci_mtk_phy_init(mtk);
> +	if (ret)
> +		goto put_usb2_hcd;
> +
> +	ret = xhci_mtk_phy_power_on(mtk);
> +	if (ret)
> +		goto exit_phys;
> +
> +	device_init_wakeup(dev, 1);
> +
> +	xhci = hcd_to_xhci(hcd);
> +	xhci->main_hcd = hcd;
> +	xhci->shared_hcd = usb_create_shared_hcd(driver, dev,
> +			dev_name(dev), hcd);
> +	if (!xhci->shared_hcd) {
> +		ret = -ENOMEM;
> +		goto power_off_phys;
> +	}
> +
> +	if ((node && of_property_read_bool(node, "usb3-lpm-capable")) ||
> +			(pdata && pdata->usb3_lpm_capable))
> +		xhci->quirks |= XHCI_LPM_SUPPORT;
> +
> +	if (HCC_MAX_PSA(xhci->hcc_params) >= 4)
> +		xhci->shared_hcd->can_do_streams = 1;
> +
> +	ret = usb_add_hcd(hcd, irq, IRQF_SHARED);
> +	if (ret)
> +		goto put_usb3_hcd;
> +
> +	ret = usb_add_hcd(xhci->shared_hcd, irq, IRQF_SHARED);
> +	if (ret)
> +		goto dealloc_usb2_hcd;
> +
> +	return 0;
> +
> +dealloc_usb2_hcd:
> +	xhci_mtk_sch_exit(xhci);
> +	usb_remove_hcd(hcd);
> +
> +put_usb3_hcd:
> +	usb_put_hcd(xhci->shared_hcd);
> +
> +power_off_phys:
> +	xhci_mtk_phy_power_off(mtk);
> +	device_init_wakeup(dev, 0);
> +
> +exit_phys:
> +	xhci_mtk_phy_exit(mtk);
> +
> +put_usb2_hcd:
> +	usb_put_hcd(hcd);
> +
> +disable_clk:
> +	xhci_mtk_clks_disable(mtk);
> +
> +disable_ldos:
> +	xhci_mtk_ldos_disable(mtk);
> +
> +disable_pm:
> +	pm_runtime_put_sync(dev);
> +	pm_runtime_disable(dev);
> +	return ret;
> +}
> +
> +static int xhci_mtk_remove(struct platform_device *dev)
> +{
> +	struct xhci_hcd_mtk *mtk = platform_get_drvdata(dev);
> +	struct usb_hcd	*hcd = mtk->hcd;
> +	struct xhci_hcd	*xhci = hcd_to_xhci(hcd);
> +
> +	usb_remove_hcd(xhci->shared_hcd);
> +	xhci_mtk_phy_power_off(mtk);
> +	xhci_mtk_phy_exit(mtk);
> +	device_init_wakeup(&dev->dev, 0);
> +
> +	usb_remove_hcd(hcd);
> +	usb_put_hcd(xhci->shared_hcd);
> +	usb_put_hcd(hcd);
> +	xhci_mtk_sch_exit(xhci);
> +	xhci_mtk_clks_disable(mtk);
> +	xhci_mtk_ldos_disable(mtk);
> +	pm_runtime_put_sync(&dev->dev);
> +	pm_runtime_disable(&dev->dev);
> +
> +	return 0;
> +}
> +
> +#ifdef CONFIG_PM_SLEEP
> +static int xhci_mtk_suspend(struct device *dev)
> +{
> +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> +
> +	xhci_mtk_ports_disable(mtk);
> +	xhci_mtk_phy_power_off(mtk);
> +	xhci_mtk_clks_disable(mtk);
> +	usb_wakeup_enable(mtk);
> +	return 0;
> +}
> +
> +static int xhci_mtk_resume(struct device *dev)
> +{
> +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> +
> +	usb_wakeup_disable(mtk);
> +	xhci_mtk_clks_enable(mtk);
> +	xhci_mtk_phy_power_on(mtk);
> +	xhci_mtk_ports_enable(mtk);
> +	return 0;
> +}
> +
> +static const struct dev_pm_ops xhci_mtk_pm_ops = {
> +	SET_SYSTEM_SLEEP_PM_OPS(xhci_mtk_suspend, xhci_mtk_resume)
> +};
> +#define DEV_PM_OPS	(&xhci_mtk_pm_ops)
> +#else
> +#define DEV_PM_OPS	NULL
> +#endif /* CONFIG_PM */
> +
> +#ifdef CONFIG_OF
> +static const struct of_device_id mtk_xhci_of_match[] = {
> +	{ .compatible = "mediatek,mt8173-xhci"},
> +	{ },
> +};
> +MODULE_DEVICE_TABLE(of, mtk_xhci_of_match);
> +#endif
> +
> +static struct platform_driver mtk_xhci_driver = {
> +	.probe	= xhci_mtk_probe,
> +	.remove	= xhci_mtk_remove,
> +	.driver	= {
> +		.name = "xhci-mtk",
> +		.pm = DEV_PM_OPS,
> +		.of_match_table = of_match_ptr(mtk_xhci_of_match),
> +	},
> +};
> +
> +static int __init xhci_mtk_init(void)
> +{
> +	xhci_init_driver(&xhci_mtk_hc_driver, &xhci_mtk_overrides);
> +	return platform_driver_register(&mtk_xhci_driver);
> +}
> +module_init(xhci_mtk_init);
> +
> +static void __exit xhci_mtk_exit(void)
> +{
> +	platform_driver_unregister(&mtk_xhci_driver);
> +}
> +module_exit(xhci_mtk_exit);
> +
> +MODULE_DESCRIPTION("MediaTek xHCI Host Controller Driver");
> +MODULE_LICENSE("GPL v2");
> diff --git a/drivers/usb/host/xhci-mtk.h b/drivers/usb/host/xhci-mtk.h
> new file mode 100644
> index 0000000..8913c2f
> --- /dev/null
> +++ b/drivers/usb/host/xhci-mtk.h
> @@ -0,0 +1,156 @@
> +/*
> + * Copyright (c) 2015 MediaTek Inc.
> + * Author:
> + *  Zhigang.Wei <zhigang.wei@mediatek.com>
> + *  Chunfeng.Yun <chunfeng.yun@mediatek.com>
> + *
> + * This software is licensed under the terms of the GNU General Public
> + * License version 2, as published by the Free Software Foundation, and
> + * may be copied, distributed, and modified under those terms.
> + *
> + * This program is distributed in the hope that 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.
> + *
> + */
> +
> +#ifndef _XHCI_MTK_H_
> +#define _XHCI_MTK_H_
> +
> +#include "xhci.h"
> +
> +/**
> + * To simplify scheduler algorithm, set a upper limit for ESIT,
> + * if a synchromous ep's ESIT is larger than @XHCI_MTK_MAX_ESIT,
> + * round down to the limit value, that means allocating more
> + * bandwidth to it.
> + */
> +#define XHCI_MTK_MAX_ESIT	64
> +
> +/**
> + * struct mu3h_sch_bw_info: schedule information for bandwidth domain
> + *
> + * @bus_bw: array to keep track of bandwidth already used at each uframes
> + * @bw_ep_list: eps in the bandwidth domain
> + *
> + * treat a HS root port as a bandwidth domain, but treat a SS root port as
> + * two bandwidth domains, one for IN eps and another for OUT eps.
> + */
> +struct mu3h_sch_bw_info {
> +	u32 bus_bw[XHCI_MTK_MAX_ESIT];
> +	struct list_head bw_ep_list;
> +};
> +
> +/**
> + * struct mu3h_sch_ep_info: schedule information for endpoint
> + *
> + * @esit: unit is 125us, equal to 2 << Interval field in ep-context
> + * @num_budget_microframes: number of continuous uframes
> + *		(@repeat==1) scheduled within the interval
> + * @bw_cost_per_microframe: bandwidth cost per microframe
> + * @endpoint: linked into bandwidth domain which it belongs to
> + * @ep: address of usb_host_endpoint struct
> + * @offset: which uframe of the interval that transfer should be
> + *		scheduled first time within the interval
> + * @repeat: the time gap between two uframes that transfers are
> + *		scheduled within a interval. in the simple algorithm, only
> + *		assign 0 or 1 to it; 0 means using only one uframe in a
> + *		interval, and 1 means using @num_budget_microframes
> + *		continuous uframes
> + * @pkts: number of packets to be transferred in the scheduled uframes
> + * @cs_count: number of CS that host will trigger
> + * @burst_mode: burst mode for scheduling. 0: normal burst mode,
> + *		distribute the bMaxBurst+1 packets for a single burst
> + *		according to @pkts and @repeat, repeate the burst multiple
> + *		times; 1: distribute the (bMaxBurst+1)*(Mult+1) packets
> + *		according to @pkts and @repeat. normal mode is used by
> + *		default
> + */
> +struct mu3h_sch_ep_info {
> +	u32 esit;
> +	u32 num_budget_microframes;
> +	u32 bw_cost_per_microframe;
> +	struct list_head endpoint;
> +	void *ep;
> +	/*
> +	 * mtk xHCI scheduling information put into reserved DWs
> +	 * in ep context
> +	 */
> +	u32 offset;
> +	u32 repeat;
> +	u32 pkts;
> +	u32 cs_count;
> +	u32 burst_mode;
> +};
> +
> +#define MU3C_U3_PORT_MAX 4
> +#define MU3C_U2_PORT_MAX 5
> +
> +/**
> + * struct mu3c_ippc_regs: MTK ssusb ip port control registers
> + * @ip_pw_ctr0~3: ip power and clock control registers
> + * @ip_pw_sts1~2: ip power and clock status registers
> + * @ip_xhci_cap: ip xHCI capability register
> + * @u3_ctrl_p[x]: ip usb3 port x control register, only low 4bytes are used
> + * @u2_ctrl_p[x]: ip usb2 port x control register, only low 4bytes are used
> + * @u2_phy_pll: usb2 phy pll control register
> + */
> +struct mu3c_ippc_regs {
> +	__le32 ip_pw_ctr0;
> +	__le32 ip_pw_ctr1;
> +	__le32 ip_pw_ctr2;
> +	__le32 ip_pw_ctr3;
> +	__le32 ip_pw_sts1;
> +	__le32 ip_pw_sts2;
> +	__le32 reserved0[3];
> +	__le32 ip_xhci_cap;
> +	__le32 reserved1[2];
> +	__le64 u3_ctrl_p[MU3C_U3_PORT_MAX];
> +	__le64 u2_ctrl_p[MU3C_U2_PORT_MAX];
> +	__le32 reserved2;
> +	__le32 u2_phy_pll;
> +	__le32 reserved3[33]; /* 0x80 ~ 0xff */
> +};
> +
> +struct xhci_hcd_mtk {
> +	struct device *dev;
> +	struct usb_hcd *hcd;
> +	struct mu3h_sch_bw_info *sch_array;
> +	struct mu3c_ippc_regs __iomem *ippc_regs;
> +	int num_u2_ports;
> +	int num_u3_ports;
> +	struct regulator *vusb33;
> +	struct regulator *vbus;
> +	struct clk *sys_clk;	/* sys and mac clock */
> +	struct clk *wk_deb_p0;	/* port0's wakeup debounce clock */
> +	struct clk *wk_deb_p1;
> +	struct regmap *pericfg;
> +	int wakeup_src;
> +	struct phy **phys;
> +	int num_phys;
> +};
> +
> +#if IS_ENABLED(CONFIG_USB_XHCI_MTK)
> +int xhci_mtk_sch_init(struct xhci_hcd *xhci);
> +void xhci_mtk_sch_exit(struct xhci_hcd *xhci);
> +int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> +		struct usb_host_endpoint *ep);
> +void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> +		struct usb_host_endpoint *ep);
> +
> +#else
> +static inline int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd,
> +	struct usb_device *udev, struct usb_host_endpoint *ep)
> +{
> +	return 0;
> +}
> +
> +static inline void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd,
> +	struct usb_device *udev, struct usb_host_endpoint *ep)
> +{
> +}
> +
> +#endif
> +
> +#endif		/* _XHCI_MTK_H_ */
> diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
> index 57f40a1..243f696 100644
> --- a/drivers/usb/host/xhci-ring.c
> +++ b/drivers/usb/host/xhci-ring.c
> @@ -68,6 +68,7 @@
>   #include <linux/slab.h>
>   #include "xhci.h"
>   #include "xhci-trace.h"
> +#include "xhci-mtk.h"
>
>   /*
>    * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
> @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
>   			      struct urb *urb, unsigned int num_trbs_left)
>   {
>   	u32 maxp, total_packet_count;
> +	u32 skip_current_trb = 0;
>
> -	if (xhci->hci_version < 0x100)
> -		return ((td_total_len - transferred) >> 10);
> -
> -	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> -	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> +	if (xhci->hci_version < 0x100) {
> +		/* for MTK xHCI, TD size doesn't include this TRB */
> +		if (xhci->quirks & XHCI_MTK_HOST)
> +			skip_current_trb = 1;
> +		else
> +			return ((td_total_len - transferred) >> 10);
> +	}
>
>   	/* One TRB with a zero-length data packet. */
>   	if (num_trbs_left == 0 || (transferred == 0 && trb_buff_len == 0) ||
>   	    trb_buff_len == td_total_len)
>   		return 0;
>
> +	if (skip_current_trb)
> +		trb_buff_len = 0;
> +
> +	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> +	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> +
>   	/* Queueing functions don't count the current TRB into transferred */
>   	return (total_packet_count - ((transferred + trb_buff_len) / maxp));
>   }
> @@ -3443,7 +3453,7 @@ int xhci_queue_ctrl_tx(struct xhci_hcd *xhci, gfp_t mem_flags,
>   		field |= 0x1;
>
>   	/* xHCI 1.0 6.4.1.2.1: Transfer Type field */
> -	if (xhci->hci_version == 0x100) {
> +	if ((xhci->hci_version == 0x100) || (xhci->quirks & XHCI_MTK_HOST)) {
>   		if (urb->transfer_buffer_length > 0) {
>   			if (setup->bRequestType & USB_DIR_IN)
>   				field |= TRB_TX_TYPE(TRB_DATA_IN);
> diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
> index 6b0f4a4..4741188 100644
> --- a/drivers/usb/host/xhci.c
> +++ b/drivers/usb/host/xhci.c
> @@ -31,6 +31,7 @@
>
>   #include "xhci.h"
>   #include "xhci-trace.h"
> +#include "xhci-mtk.h"
>
>   #define DRIVER_AUTHOR "Sarah Sharp"
>   #define DRIVER_DESC "'eXtensible' Host Controller (xHC) Driver"
> @@ -623,7 +624,11 @@ int xhci_run(struct usb_hcd *hcd)
>   			"// Set the interrupt modulation register");
>   	temp = readl(&xhci->ir_set->irq_control);
>   	temp &= ~ER_IRQ_INTERVAL_MASK;
> -	temp |= (u32) 160;
> +	/*
> +	 * the increment interval is 8 times as much as that defined
> +	 * in xHCI spec on MTK's controller
> +	 */
> +	temp |= (u32) ((xhci->quirks & XHCI_MTK_HOST) ? 20 : 160);
>   	writel(temp, &xhci->ir_set->irq_control);
>
>   	/* Set the HCD state before we enable the irqs */
> @@ -1693,6 +1698,9 @@ int xhci_drop_endpoint(struct usb_hcd *hcd, struct usb_device *udev,
>
>   	xhci_endpoint_zero(xhci, xhci->devs[udev->slot_id], ep);
>
> +	if (xhci->quirks & XHCI_MTK_HOST)
> +		xhci_mtk_drop_ep_quirk(hcd, udev, ep);
> +
>   	xhci_dbg(xhci, "drop ep 0x%x, slot id %d, new drop flags = %#x, new add flags = %#x\n",
>   			(unsigned int) ep->desc.bEndpointAddress,
>   			udev->slot_id,
> @@ -1788,6 +1796,15 @@ int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev,
>   		return -ENOMEM;
>   	}
>
> +	if (xhci->quirks & XHCI_MTK_HOST) {
> +		ret = xhci_mtk_add_ep_quirk(hcd, udev, ep);
> +		if (ret < 0) {
> +			xhci_free_or_cache_endpoint_ring(xhci,
> +				virt_dev, ep_index);
> +			return ret;
> +		}
> +	}
> +
>   	ctrl_ctx->add_flags |= cpu_to_le32(added_ctxs);
>   	new_add_flags = le32_to_cpu(ctrl_ctx->add_flags);
>
> diff --git a/drivers/usb/host/xhci.h b/drivers/usb/host/xhci.h
> index ea8b904..1536da0 100644
> --- a/drivers/usb/host/xhci.h
> +++ b/drivers/usb/host/xhci.h
> @@ -1577,6 +1577,7 @@ struct xhci_hcd {
>   /* For controllers with a broken beyond repair streams implementation */
>   #define XHCI_BROKEN_STREAMS	(1 << 19)
>   #define XHCI_PME_STUCK_QUIRK	(1 << 20)
> +#define XHCI_MTK_HOST		(1 << 21)
>   	unsigned int		num_active_eps;
>   	unsigned int		limit_active_eps;
>   	/* There are two roothubs to keep track of bus suspend info for */
>
Chunfeng Yun (云春峰) Oct. 18, 2015, 1:25 a.m. UTC | #8
Hi,
On Thu, 2015-10-15 at 17:46 +0300, Mathias Nyman wrote:
> On 29.09.2015 06:01, Chunfeng Yun wrote:
> > There some vendor quirks for MTK xhci host controller:
> > 1. It defines some extra SW scheduling parameters for HW
> >    to minimize the scheduling effort for synchronous and
> >    interrupt endpoints. The parameters are put into reseved
> >    DWs of slot context and endpoint context.
> > 2. Its IMODI unit for Interrupter Moderation register is
> >    8 times as much as that defined in xHCI spec.
> > 3. Its TDS in  Normal TRB defines a number of packets that
> >    remains to be transferred for a TD after processing all
> >    Max packets in all previous TRBs.
> >
> > Signed-off-by: Chunfeng Yun <chunfeng.yun@mediatek.com>
> 
> Looks good in my opinion, There's one part about how we split the ISOC transafers
> across multiple microframes that I don't fully understand:
> 
> > +	} else if (udev->speed == USB_SPEED_SUPER) {
> > +		/* usb3_r1 spec section4.4.7 & 4.4.8 */
> > +		sch_ep->cs_count = 0;
> > +		esit_pkts = (mult + 1) * (max_burst + 1);
> 
> So the device would like the transfers to be esit_pkts packets every esit microframe apart.
> 
> > +		if (ep_type == INT_IN_EP || ep_type == INT_OUT_EP) {
> > +			sch_ep->pkts = esit_pkts;
> > +			sch_ep->num_budget_microframes = 1;
> > +			sch_ep->repeat = 0;
> > +		}
> > +
> > +		if (ep_type == ISOC_IN_EP || ep_type == ISOC_OUT_EP) {
> > +			if (esit_pkts <= sch_ep->esit)
> > +				sch_ep->pkts = 1;
> > +			else
> > +				sch_ep->pkts = roundup_pow_of_two(esit_pkts)
> > +					/ sch_ep->esit;
> 
> sch_ep->pkts now contain a rounded up average value of how many packets per microframe
> would be transferred in case we'd transfer packets every microframe instead of every esit frame apart.
> 
> > +
> > +			sch_ep->num_budget_microframes =
> > +				DIV_ROUND_UP(esit_pkts, sch_ep->pkts);
> 
> sch_ep->num_budget_microframes now contains the number of microframes during an esit needed
> to transer all data if each microframe contains the average amount of data.
> 
> So basically we are trying to use as many microframes as possible with as few packets
> per microframe as possible.
> 
> Did I understand this correctly?
Yes, you are right.

> How will devices react if they expect to get 16 packets every 16th microframe,
> but they get one packet every microframe instead?
I think that the synchronous endpoint must specify its period by
bInterval, but can't specify how data should be transfered during the
period by the host, and it just only receives data passively. So the
device can receive data correctly in the case(bInterval is 5).

quote from usb3_r1.0 section4.4.8 Isochronous Transfers:
"The host can request data from the device or send data to the device at
any time during the service interval for a particular endpoint on that
device"

> 
> Or is this just theoretical bw calculations for the host and it will in the end do
> whatever it wants with the information.
The host will schedule synchronous endpoints in a specific
micro-frame/frame according to this parameters provided by software.

Thanks a lot.
> 
> Otherwise, by making the changes Daniel Thompson suggested I think the xhci parts looks ready.
> 
> -Mathias
> 
>   
> 
>    
>   
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> > +	} else if (is_fs_or_ls(udev->speed)) {
> > +
> > +		/*
> > +		 * usb_20 spec section11.18.4
> > +		 * assume worst cases
> > +		 */
> > +		sch_ep->repeat = 0;
> > +		sch_ep->pkts = 1; /* at most one packet for each microframe */
> > +		if (ep_type == INT_IN_EP || ep_type == INT_OUT_EP) {
> > +			sch_ep->cs_count = 3; /* at most need 3 CS*/
> > +			/* one for SS and one for budgeted transaction */
> > +			sch_ep->num_budget_microframes = sch_ep->cs_count + 2;
> > +			sch_ep->bw_cost_per_microframe = max_packet_size;
> > +		}
> > +		if (ep_type == ISOC_OUT_EP) {
> > +
> > +			/*
> > +			 * the best case FS budget assumes that 188 FS bytes
> > +			 * occur in each microframe
> > +			 */
> > +			sch_ep->num_budget_microframes = DIV_ROUND_UP(
> > +				max_packet_size, FS_PAYLOAD_MAX);
> > +			sch_ep->bw_cost_per_microframe = FS_PAYLOAD_MAX;
> > +			sch_ep->cs_count = sch_ep->num_budget_microframes;
> > +		}
> > +		if (ep_type == ISOC_IN_EP) {
> > +			/* at most need additional two CS. */
> > +			sch_ep->cs_count = DIV_ROUND_UP(
> > +				max_packet_size, FS_PAYLOAD_MAX) + 2;
> > +			sch_ep->num_budget_microframes = sch_ep->cs_count + 2;
> > +			sch_ep->bw_cost_per_microframe = FS_PAYLOAD_MAX;
> > +		}
> > +	}
> > +}
> > +
> > +/* Get maximum bandwidth when we schedule at offset slot. */
> > +static u32 get_max_bw(struct mu3h_sch_bw_info *sch_bw,
> > +	struct mu3h_sch_ep_info *sch_ep, u32 offset)
> > +{
> > +	u32 num_esit;
> > +	u32 max_bw = 0;
> > +	int i;
> > +	int j;
> > +
> > +	num_esit = XHCI_MTK_MAX_ESIT / sch_ep->esit;
> > +	for (i = 0; i < num_esit; i++) {
> > +		u32 base = offset + i * sch_ep->esit;
> > +
> > +		for (j = 0; j < sch_ep->num_budget_microframes; j++) {
> > +			if (sch_bw->bus_bw[base + j] > max_bw)
> > +				max_bw = sch_bw->bus_bw[base + j];
> > +		}
> > +	}
> > +	return max_bw;
> > +}
> > +
> > +static void update_bus_bw(struct mu3h_sch_bw_info *sch_bw,
> > +	struct mu3h_sch_ep_info *sch_ep, int bw_cost)
> > +{
> > +	u32 num_esit;
> > +	u32 base;
> > +	int i;
> > +	int j;
> > +
> > +	num_esit = XHCI_MTK_MAX_ESIT / sch_ep->esit;
> > +	for (i = 0; i < num_esit; i++) {
> > +		base = sch_ep->offset + i * sch_ep->esit;
> > +		for (j = 0; j < sch_ep->num_budget_microframes; j++)
> > +			sch_bw->bus_bw[base + j] += bw_cost;
> > +	}
> > +}
> > +
> > +static int check_sch_bw(struct usb_device *udev,
> > +	struct mu3h_sch_bw_info *sch_bw, struct mu3h_sch_ep_info *sch_ep)
> > +{
> > +	u32 offset;
> > +	u32 esit;
> > +	u32 num_budget_microframes;
> > +	u32 min_bw;
> > +	u32 min_index;
> > +	u32 worst_bw;
> > +	u32 bw_boundary;
> > +
> > +	if (sch_ep->esit > XHCI_MTK_MAX_ESIT)
> > +		sch_ep->esit = XHCI_MTK_MAX_ESIT;
> > +
> > +	esit = sch_ep->esit;
> > +	num_budget_microframes = sch_ep->num_budget_microframes;
> > +
> > +	/*
> > +	 * Search through all possible schedule microframes.
> > +	 * and find a microframe where its worst bandwidth is minimum.
> > +	 */
> > +	min_bw = ~0;
> > +	min_index = 0;
> > +	for (offset = 0; offset < esit; offset++) {
> > +		if ((offset + num_budget_microframes) > sch_ep->esit)
> > +			break;
> > +
> > +		/*
> > +		 * usb_20 spec section11.18:
> > +		 * must never schedule Start-Split in Y6
> > +		 */
> > +		if (is_fs_or_ls(udev->speed) && (offset % 8 == 6))
> > +			continue;
> > +
> > +		worst_bw = get_max_bw(sch_bw, sch_ep, offset);
> > +		if (min_bw > worst_bw) {
> > +			min_bw = worst_bw;
> > +			min_index = offset;
> > +		}
> > +		if (min_bw == 0)
> > +			break;
> > +	}
> > +	sch_ep->offset = min_index;
> > +
> > +	bw_boundary = (udev->speed == USB_SPEED_SUPER)
> > +				? SS_BW_BOUNDARY : HS_BW_BOUNDARY;
> > +
> > +	/* check bandwidth */
> > +	if (min_bw + sch_ep->bw_cost_per_microframe > bw_boundary)
> > +		return -ERANGE;
> > +
> > +	/* update bus bandwidth info */
> > +	update_bus_bw(sch_bw, sch_ep, sch_ep->bw_cost_per_microframe);
> > +
> > +	return 0;
> > +}
> > +
> > +static bool need_bw_sch(struct usb_host_endpoint *ep,
> > +	enum usb_device_speed speed, int has_tt)
> > +{
> > +	/* only for periodic endpoints */
> > +	if (usb_endpoint_xfer_control(&ep->desc)
> > +		|| usb_endpoint_xfer_bulk(&ep->desc))
> > +		return false;
> > +
> > +	/*
> > +	 * for LS & FS periodic endpoints which its device don't attach
> > +	 * to TT are also ignored, root-hub will schedule them directly
> > +	 */
> > +	if (is_fs_or_ls(speed) && !has_tt)
> > +		return false;
> > +
> > +	return true;
> > +}
> > +
> > +int xhci_mtk_sch_init(struct xhci_hcd *xhci)
> > +{
> > +	struct usb_hcd *hcd = xhci_to_hcd(xhci);
> > +	struct device *dev = hcd->self.controller;
> > +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> > +	struct mu3h_sch_bw_info *sch_array;
> > +	int num_usb_bus;
> > +	int i;
> > +
> > +	/* ss IN and OUT are separated */
> > +	num_usb_bus = xhci->num_usb3_ports * 2 + xhci->num_usb2_ports;
> > +
> > +	sch_array = kcalloc(num_usb_bus, sizeof(*sch_array), GFP_KERNEL);
> > +	if (sch_array == NULL)
> > +		return -ENOMEM;
> > +
> > +	for (i = 0; i < num_usb_bus; i++)
> > +		INIT_LIST_HEAD(&sch_array[i].bw_ep_list);
> > +
> > +	mtk->sch_array = sch_array;
> > +
> > +	return 0;
> > +}
> > +EXPORT_SYMBOL_GPL(xhci_mtk_sch_init);
> > +
> > +void xhci_mtk_sch_exit(struct xhci_hcd *xhci)
> > +{
> > +	struct usb_hcd *hcd = xhci_to_hcd(xhci);
> > +	struct device *dev = hcd->self.controller;
> > +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> > +
> > +	kfree(mtk->sch_array);
> > +}
> > +EXPORT_SYMBOL_GPL(xhci_mtk_sch_exit);
> > +
> > +int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> > +		struct usb_host_endpoint *ep)
> > +{
> > +	struct device *dev = hcd->self.controller;
> > +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> > +	struct xhci_hcd *xhci;
> > +	struct xhci_ep_ctx *ep_ctx;
> > +	struct xhci_slot_ctx *slot_ctx;
> > +	struct xhci_virt_device *virt_dev;
> > +	struct mu3h_sch_bw_info *sch_bw;
> > +	struct mu3h_sch_ep_info *sch_ep;
> > +	struct mu3h_sch_bw_info *sch_array;
> > +	unsigned int ep_index;
> > +	int bw_index;
> > +	int ret = 0;
> > +
> > +	xhci = hcd_to_xhci(hcd);
> > +	virt_dev = xhci->devs[udev->slot_id];
> > +	ep_index = xhci_get_endpoint_index(&ep->desc);
> > +	slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx);
> > +	ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, ep_index);
> > +	sch_array = mtk->sch_array;
> > +
> > +	xhci_dbg(xhci, "%s() type:%d, speed:%d, mpkt:%d, dir:%d, ep:%p\n",
> > +		__func__, usb_endpoint_type(&ep->desc), udev->speed,
> > +		GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc)),
> > +		usb_endpoint_dir_in(&ep->desc), ep);
> > +
> > +	if (!need_bw_sch(ep, udev->speed, slot_ctx->tt_info & TT_SLOT))
> > +		return 0;
> > +
> > +	bw_index = get_bw_index(xhci, udev, ep);
> > +	sch_bw = &sch_array[bw_index];
> > +
> > +	sch_ep = kzalloc(sizeof(struct mu3h_sch_ep_info), GFP_NOIO);
> > +	if (!sch_ep)
> > +		return -ENOMEM;
> > +
> > +	setup_sch_info(udev, ep_ctx, sch_ep);
> > +
> > +	ret = check_sch_bw(udev, sch_bw, sch_ep);
> > +	if (ret) {
> > +		xhci_err(xhci, "Not enough bandwidth!\n");
> > +		kfree(sch_ep);
> > +		return -ENOSPC;
> > +	}
> > +
> > +	list_add_tail(&sch_ep->endpoint, &sch_bw->bw_ep_list);
> > +	sch_ep->ep = ep;
> > +
> > +	ep_ctx->reserved[0] |= cpu_to_le32(EP_BPKTS(sch_ep->pkts)
> > +		| EP_BCSCOUNT(sch_ep->cs_count) | EP_BBM(sch_ep->burst_mode));
> > +	ep_ctx->reserved[1] |= cpu_to_le32(EP_BOFFSET(sch_ep->offset)
> > +		| EP_BREPEAT(sch_ep->repeat));
> > +
> > +	xhci_dbg(xhci, " PKTS:%x, CSCOUNT:%x, BM:%x, OFFSET:%x, REPEAT:%x\n",
> > +			sch_ep->pkts, sch_ep->cs_count, sch_ep->burst_mode,
> > +			sch_ep->offset, sch_ep->repeat);
> > +
> > +	return 0;
> > +}
> > +EXPORT_SYMBOL_GPL(xhci_mtk_add_ep_quirk);
> > +
> > +void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> > +		struct usb_host_endpoint *ep)
> > +{
> > +	struct device *dev = hcd->self.controller;
> > +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> > +	struct xhci_hcd *xhci;
> > +	struct xhci_slot_ctx *slot_ctx;
> > +	struct xhci_virt_device *virt_dev;
> > +	struct mu3h_sch_bw_info *sch_array;
> > +	struct mu3h_sch_bw_info *sch_bw;
> > +	struct mu3h_sch_ep_info *sch_ep;
> > +	int bw_index;
> > +
> > +	xhci = hcd_to_xhci(hcd);
> > +	virt_dev = xhci->devs[udev->slot_id];
> > +	slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx);
> > +	sch_array = mtk->sch_array;
> > +
> > +	xhci_dbg(xhci, "%s() type:%d, speed:%d, mpks:%d, dir:%d, ep:%p\n",
> > +		__func__, usb_endpoint_type(&ep->desc), udev->speed,
> > +		GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc)),
> > +		usb_endpoint_dir_in(&ep->desc), ep);
> > +
> > +	if (!need_bw_sch(ep, udev->speed, slot_ctx->tt_info & TT_SLOT))
> > +		return;
> > +
> > +	bw_index = get_bw_index(xhci, udev, ep);
> > +	sch_bw = &sch_array[bw_index];
> > +
> > +	list_for_each_entry(sch_ep, &sch_bw->bw_ep_list, endpoint) {
> > +		if (sch_ep->ep == ep) {
> > +			update_bus_bw(sch_bw, sch_ep,
> > +				-sch_ep->bw_cost_per_microframe);
> > +			list_del(&sch_ep->endpoint);
> > +			kfree(sch_ep);
> > +			break;
> > +		}
> > +	}
> > +}
> > +EXPORT_SYMBOL_GPL(xhci_mtk_drop_ep_quirk);
> > diff --git a/drivers/usb/host/xhci-mtk.c b/drivers/usb/host/xhci-mtk.c
> > new file mode 100644
> > index 0000000..afb52bf
> > --- /dev/null
> > +++ b/drivers/usb/host/xhci-mtk.c
> > @@ -0,0 +1,776 @@
> > +/*
> > + * Copyright (c) 2015 MediaTek Inc.
> > + * Author:
> > + *  Chunfeng Yun <chunfeng.yun@mediatek.com>
> > + *
> > + * This software is licensed under the terms of the GNU General Public
> > + * License version 2, as published by the Free Software Foundation, and
> > + * may be copied, distributed, and modified under those terms.
> > + *
> > + * This program is distributed in the hope that 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.
> > + *
> > + */
> > +
> > +#include <linux/clk.h>
> > +#include <linux/dma-mapping.h>
> > +#include <linux/iopoll.h>
> > +#include <linux/kernel.h>
> > +#include <linux/mfd/syscon.h>
> > +#include <linux/module.h>
> > +#include <linux/of.h>
> > +#include <linux/phy/phy.h>
> > +#include <linux/platform_device.h>
> > +#include <linux/pm_runtime.h>
> > +#include <linux/regmap.h>
> > +#include <linux/regulator/consumer.h>
> > +#include <linux/slab.h>
> > +#include <linux/usb/phy.h>
> > +#include <linux/usb/xhci_pdriver.h>
> > +
> > +#include "xhci.h"
> > +#include "xhci-mtk.h"
> > +
> > +/* ip_pw_ctrl0 register */
> > +#define CTRL0_IP_SW_RST	BIT(0)
> > +
> > +/* ip_pw_ctrl1 register */
> > +#define CTRL1_IP_HOST_PDN	BIT(0)
> > +
> > +/* ip_pw_ctrl2 register */
> > +#define CTRL2_IP_DEV_PDN	BIT(0)
> > +
> > +/* ip_pw_sts1 register */
> > +#define STS1_IP_SLEEP_STS	BIT(30)
> > +#define STS1_U3_MAC_RST	BIT(16)
> > +#define STS1_SYS125_RST	BIT(10)
> > +#define STS1_REF_RST		BIT(8)
> > +#define STS1_SYSPLL_STABLE	BIT(0)
> > +
> > +/* ip_pw_sts2 register */
> > +#define STS2_U2_MAC_RST	BIT(0)
> > +
> > +/* ip_xhci_cap register */
> > +#define CAP_U3_PORT_NUM(p)	((p) & 0xff)
> > +#define CAP_U2_PORT_NUM(p)	(((p) >> 8) & 0xff)
> > +
> > +/* u3_ctrl_p register */
> > +#define CTRL_U3_PORT_HOST_SEL	BIT(2)
> > +#define CTRL_U3_PORT_PDN	BIT(1)
> > +#define CTRL_U3_PORT_DIS	BIT(0)
> > +
> > +/* u2_ctrl_p register */
> > +#define CTRL_U2_PORT_HOST_SEL	BIT(2)
> > +#define CTRL_U2_PORT_PDN	BIT(1)
> > +#define CTRL_U2_PORT_DIS	BIT(0)
> > +
> > +/* u2_phy_pll register */
> > +#define CTRL_U2_FORCE_PLL_STB	BIT(28)
> > +
> > +#define PERI_WK_CTRL0		0x400
> > +#define UWK_CTR0_0P_LS_PE	BIT(8)  /* posedge */
> > +#define UWK_CTR0_0P_LS_NE	BIT(7)  /* negedge for 0p linestate*/
> > +#define UWK_CTL1_1P_LS_C(x)	(((x) & 0xf) << 1)
> > +#define UWK_CTL1_1P_LS_E	BIT(0)
> > +
> > +#define PERI_WK_CTRL1		0x404
> > +#define UWK_CTL1_IS_C(x)	(((x) & 0xf) << 26)
> > +#define UWK_CTL1_IS_E		BIT(25)
> > +#define UWK_CTL1_0P_LS_C(x)	(((x) & 0xf) << 21)
> > +#define UWK_CTL1_0P_LS_E	BIT(20)
> > +#define UWK_CTL1_IDDIG_C(x)	(((x) & 0xf) << 11)  /* cycle debounce */
> > +#define UWK_CTL1_IDDIG_E	BIT(10) /* enable debounce */
> > +#define UWK_CTL1_IDDIG_P	BIT(9)  /* polarity */
> > +#define UWK_CTL1_0P_LS_P	BIT(7)
> > +#define UWK_CTL1_IS_P		BIT(6)  /* polarity for ip sleep */
> > +
> > +enum ssusb_wakeup_src {
> > +	SSUSB_WK_IP_SLEEP = 1,
> > +	SSUSB_WK_LINE_STATE = 2,
> > +};
> > +
> > +static int check_ip_clk_status(struct xhci_hcd_mtk *mtk)
> > +{
> > +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> > +	u32 val, check_val;
> > +	int ret;
> > +
> > +	check_val = STS1_SYSPLL_STABLE | STS1_REF_RST | STS1_SYS125_RST;
> > +	check_val |= (mtk->num_u3_ports ? STS1_U3_MAC_RST : 0);
> > +
> > +	ret = readl_poll_timeout(&ippc->ip_pw_sts1, val,
> > +			  (check_val == (val & check_val)), 100, 20000);
> > +	if (ret) {
> > +		dev_err(mtk->dev, "mac3 clocks are not stable (0x%x)\n", val);
> > +		return ret;
> > +	}
> > +
> > +	ret = readl_poll_timeout(&ippc->ip_pw_sts2, val,
> > +			   (val & STS2_U2_MAC_RST), 100, 10000);
> > +	if (ret) {
> > +		dev_err(mtk->dev, "mac2 reset is still active!!!\n");
> > +		return ret;
> > +	}
> > +
> > +	return 0;
> > +}
> > +
> > +static int xhci_mtk_ports_enable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> > +	u32 temp;
> > +	int i;
> > +
> > +	/* power on host ip */
> > +	temp = readl(&ippc->ip_pw_ctr1);
> > +	temp &= ~CTRL1_IP_HOST_PDN;
> > +	writel(temp, &ippc->ip_pw_ctr1);
> > +
> > +	/* power on and enable all u3 ports */
> > +	for (i = 0; i < mtk->num_u3_ports; i++) {
> > +		temp = readl(&ippc->u3_ctrl_p[i]);
> > +		temp &= ~(CTRL_U3_PORT_PDN | CTRL_U3_PORT_DIS);
> > +		temp |= CTRL_U3_PORT_HOST_SEL;
> > +		writel(temp, &ippc->u3_ctrl_p[i]);
> > +	}
> > +
> > +	/* power on and enable all u2 ports */
> > +	for (i = 0; i < mtk->num_u2_ports; i++) {
> > +		temp = readl(&ippc->u2_ctrl_p[i]);
> > +		temp &= ~(CTRL_U2_PORT_PDN | CTRL_U2_PORT_DIS);
> > +		temp |= CTRL_U2_PORT_HOST_SEL;
> > +		writel(temp, &ippc->u2_ctrl_p[i]);
> > +	}
> > +
> > +	/*
> > +	 * wait for clocks to be stable, and clock domains reset to
> > +	 * be inactive after power on and enable ports
> > +	 */
> > +	return check_ip_clk_status(mtk);
> > +}
> > +
> > +static int xhci_mtk_ports_disable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> > +	u32 temp;
> > +	int ret;
> > +	int i;
> > +
> > +	/* power down all u3 ports */
> > +	for (i = 0; i < mtk->num_u3_ports; i++) {
> > +		temp = readl(&ippc->u3_ctrl_p[i]);
> > +		temp |= CTRL_U3_PORT_PDN;
> > +		writel(temp, &ippc->u3_ctrl_p[i]);
> > +	}
> > +
> > +	/* power down all u2 ports */
> > +	for (i = 0; i < mtk->num_u2_ports; i++) {
> > +		temp = readl(&ippc->u2_ctrl_p[i]);
> > +		temp |= CTRL_U2_PORT_PDN;
> > +		writel(temp, &ippc->u2_ctrl_p[i]);
> > +	}
> > +
> > +	/* power down host ip */
> > +	temp = readl(&ippc->ip_pw_ctr1);
> > +	temp |= CTRL1_IP_HOST_PDN;
> > +	writel(temp, &ippc->ip_pw_ctr1);
> > +
> > +	ret = readl_poll_timeout(&ippc->ip_pw_sts1, temp,
> > +			  (temp & STS1_IP_SLEEP_STS), 100, 100000);
> > +	if (ret) {
> > +		dev_err(mtk->dev, "ip sleep failed!!!\n");
> > +		return ret;
> > +	}
> > +	return 0;
> > +}
> > +
> > +static void xhci_mtk_ports_config(struct xhci_hcd_mtk *mtk)
> > +{
> > +	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
> > +	u32 temp;
> > +
> > +	/* reset whole ip */
> > +	temp = readl(&ippc->ip_pw_ctr0);
> > +	temp |= CTRL0_IP_SW_RST;
> > +	writel(temp, &ippc->ip_pw_ctr0);
> > +	udelay(1);
> > +	temp = readl(&ippc->ip_pw_ctr0);
> > +	temp &= ~CTRL0_IP_SW_RST;
> > +	writel(temp, &ippc->ip_pw_ctr0);
> > +
> > +	/*
> > +	 * device ip is default power-on in fact
> > +	 * power down device ip, otherwise ip-sleep will fail
> > +	 */
> > +	temp = readl(&ippc->ip_pw_ctr2);
> > +	temp |= CTRL2_IP_DEV_PDN;
> > +	writel(temp, &ippc->ip_pw_ctr2);
> > +
> > +	temp = readl(&ippc->ip_xhci_cap);
> > +	mtk->num_u3_ports = CAP_U3_PORT_NUM(temp);
> > +	mtk->num_u2_ports = CAP_U2_PORT_NUM(temp);
> > +	dev_dbg(mtk->dev, "%s u2p:%d, u3p:%d\n", __func__,
> > +			mtk->num_u2_ports, mtk->num_u3_ports);
> > +
> > +	xhci_mtk_ports_enable(mtk);
> > +}
> > +
> > +static int xhci_mtk_clks_enable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	int ret;
> > +
> > +	ret = clk_prepare_enable(mtk->sys_clk);
> > +	if (ret) {
> > +		dev_err(mtk->dev, "failed to enable u3phya_ref\n");
> > +		goto u3phya_ref_err;
> > +	}
> > +	ret = clk_prepare_enable(mtk->wk_deb_p0);
> > +	if (ret) {
> > +		dev_err(mtk->dev, "failed to enable wk_deb_p0\n");
> > +		goto usb_p0_err;
> > +	}
> > +	if (mtk->num_phys > 1) {
> > +		ret = clk_prepare_enable(mtk->wk_deb_p1);
> > +		if (ret) {
> > +			dev_err(mtk->dev, "failed to enable wk_deb_p1\n");
> > +			goto usb_p1_err;
> > +		}
> > +	}
> > +	return 0;
> > +
> > +usb_p1_err:
> > +	clk_disable_unprepare(mtk->wk_deb_p0);
> > +usb_p0_err:
> > +	clk_disable_unprepare(mtk->sys_clk);
> > +u3phya_ref_err:
> > +	return -EINVAL;
> > +}
> > +
> > +static void xhci_mtk_clks_disable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	if (mtk->num_phys > 1)
> > +		clk_disable_unprepare(mtk->wk_deb_p1);
> > +	clk_disable_unprepare(mtk->wk_deb_p0);
> > +	clk_disable_unprepare(mtk->sys_clk);
> > +}
> > +
> > +/* only clocks can be turn off for ip-sleep wakeup mode */
> > +static void usb_wakeup_ip_sleep_en(struct xhci_hcd_mtk *mtk)
> > +{
> > +	u32 tmp;
> > +	struct regmap *pericfg = mtk->pericfg;
> > +
> > +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> > +	tmp &= ~UWK_CTL1_IS_P;
> > +	tmp &= ~(UWK_CTL1_IS_C(0xf));
> > +	tmp |= UWK_CTL1_IS_C(0x8);
> > +	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
> > +	regmap_write(pericfg, PERI_WK_CTRL1, tmp | UWK_CTL1_IS_E);
> > +
> > +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> > +	dev_dbg(mtk->dev, "%s(): WK_CTRL1[P6,E25,C26:29]=%#x\n",
> > +		__func__, tmp);
> > +}
> > +
> > +static void usb_wakeup_ip_sleep_dis(struct xhci_hcd_mtk *mtk)
> > +{
> > +	u32 tmp;
> > +
> > +	regmap_read(mtk->pericfg, PERI_WK_CTRL1, &tmp);
> > +	tmp &= ~UWK_CTL1_IS_E;
> > +	regmap_write(mtk->pericfg, PERI_WK_CTRL1, tmp);
> > +}
> > +
> > +/*
> > +* for line-state wakeup mode, phy's power should not power-down
> > +* and only support cable plug in/out
> > +*/
> > +static void usb_wakeup_line_state_en(struct xhci_hcd_mtk *mtk)
> > +{
> > +	u32 tmp;
> > +	struct regmap *pericfg = mtk->pericfg;
> > +
> > +	/* line-state of u2-port0 */
> > +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> > +	tmp &= ~UWK_CTL1_0P_LS_P;
> > +	tmp &= ~(UWK_CTL1_0P_LS_C(0xf));
> > +	tmp |= UWK_CTL1_0P_LS_C(0x8);
> > +	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
> > +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> > +	regmap_write(pericfg, PERI_WK_CTRL1, tmp | UWK_CTL1_0P_LS_E);
> > +
> > +	/* line-state of u2-port1 if support */
> > +	if (mtk->num_phys > 1) {
> > +		regmap_read(pericfg, PERI_WK_CTRL0, &tmp);
> > +		tmp &= ~(UWK_CTL1_1P_LS_C(0xf));
> > +		tmp |= UWK_CTL1_1P_LS_C(0x8);
> > +		regmap_write(pericfg, PERI_WK_CTRL0, tmp);
> > +		regmap_write(pericfg, PERI_WK_CTRL0, tmp | UWK_CTL1_1P_LS_E);
> > +	}
> > +}
> > +
> > +static void usb_wakeup_line_state_dis(struct xhci_hcd_mtk *mtk)
> > +{
> > +	u32 tmp;
> > +	struct regmap *pericfg = mtk->pericfg;
> > +
> > +	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
> > +	tmp &= ~UWK_CTL1_0P_LS_E;
> > +	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
> > +
> > +	if (mtk->num_phys > 1) {
> > +		regmap_read(pericfg, PERI_WK_CTRL0, &tmp);
> > +		tmp &= ~UWK_CTL1_1P_LS_E;
> > +		regmap_write(pericfg, PERI_WK_CTRL0, tmp);
> > +	}
> > +}
> > +
> > +static void usb_wakeup_enable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	if (mtk->wakeup_src == SSUSB_WK_IP_SLEEP)
> > +		usb_wakeup_ip_sleep_en(mtk);
> > +	else if (mtk->wakeup_src == SSUSB_WK_LINE_STATE)
> > +		usb_wakeup_line_state_en(mtk);
> > +}
> > +
> > +static void usb_wakeup_disable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	if (mtk->wakeup_src == SSUSB_WK_IP_SLEEP)
> > +		usb_wakeup_ip_sleep_dis(mtk);
> > +	else if (mtk->wakeup_src == SSUSB_WK_LINE_STATE)
> > +		usb_wakeup_line_state_dis(mtk);
> > +}
> > +
> > +static int usb_wakeup_of_property_parse(struct xhci_hcd_mtk *mtk,
> > +				struct device_node *dn)
> > +{
> > +	struct device *dev = mtk->dev;
> > +
> > +	/*
> > +	* wakeup function is optional, so it is not an error if this property
> > +	* does not exist, and in such case, no need to get relative
> > +	* properties anymore.
> > +	*/
> > +	of_property_read_u32(dn, "mediatek,wakeup-src", &mtk->wakeup_src);
> > +	if (!mtk->wakeup_src)
> > +		return 0;
> > +
> > +	mtk->wk_deb_p0 = devm_clk_get(dev, "wakeup_deb_p0");
> > +	if (IS_ERR(mtk->wk_deb_p0)) {
> > +		dev_err(dev, "fail to get wakeup_deb_p0\n");
> > +		return PTR_ERR(mtk->wk_deb_p0);
> > +	}
> > +
> > +	mtk->wk_deb_p1 = devm_clk_get(dev, "wakeup_deb_p1");
> > +	if (IS_ERR(mtk->wk_deb_p1)) {
> > +		dev_err(dev, "fail to get wakeup_deb_p1\n");
> > +		return PTR_ERR(mtk->wk_deb_p1);
> > +	}
> > +
> > +	mtk->pericfg = syscon_regmap_lookup_by_phandle(dn,
> > +						"mediatek,syscon-wakeup");
> > +	if (IS_ERR(mtk->pericfg)) {
> > +		dev_err(dev, "fail to get pericfg regs\n");
> > +		return PTR_ERR(mtk->pericfg);
> > +	}
> > +
> > +	return 0;
> > +}
> > +
> > +static int xhci_mtk_setup(struct usb_hcd *hcd);
> > +static const struct xhci_driver_overrides xhci_mtk_overrides __initconst = {
> > +	.extra_priv_size = sizeof(struct xhci_hcd),
> > +	.reset = xhci_mtk_setup,
> > +};
> > +
> > +static struct hc_driver __read_mostly xhci_mtk_hc_driver;
> > +
> > +static int xhci_mtk_phy_init(struct xhci_hcd_mtk *mtk)
> > +{
> > +	int i;
> > +	int ret;
> > +
> > +	for (i = 0; i < mtk->num_phys; i++) {
> > +		ret = phy_init(mtk->phys[i]);
> > +		if (ret)
> > +			goto exit_phy;
> > +	}
> > +	return 0;
> > +
> > +exit_phy:
> > +	for (; i > 0; i--)
> > +		phy_exit(mtk->phys[i - 1]);
> > +
> > +	return ret;
> > +}
> > +
> > +static int xhci_mtk_phy_exit(struct xhci_hcd_mtk *mtk)
> > +{
> > +	int i;
> > +
> > +	for (i = 0; i < mtk->num_phys; i++)
> > +		phy_exit(mtk->phys[i]);
> > +
> > +	return 0;
> > +}
> > +
> > +static int xhci_mtk_phy_power_on(struct xhci_hcd_mtk *mtk)
> > +{
> > +	int i;
> > +	int ret;
> > +
> > +	for (i = 0; i < mtk->num_phys; i++) {
> > +		ret = phy_power_on(mtk->phys[i]);
> > +		if (ret)
> > +			goto power_off_phy;
> > +	}
> > +	return 0;
> > +
> > +power_off_phy:
> > +	for (; i > 0; i--)
> > +		phy_power_off(mtk->phys[i - 1]);
> > +
> > +	return ret;
> > +}
> > +
> > +static void xhci_mtk_phy_power_off(struct xhci_hcd_mtk *mtk)
> > +{
> > +	unsigned int i;
> > +
> > +	for (i = 0; i < mtk->num_phys; i++)
> > +		phy_power_off(mtk->phys[i]);
> > +}
> > +
> > +static int xhci_mtk_ldos_enable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	int ret;
> > +
> > +	ret = regulator_enable(mtk->vbus);
> > +	if (ret) {
> > +		dev_err(mtk->dev, "failed to enable vbus\n");
> > +		return ret;
> > +	}
> > +
> > +	ret = regulator_enable(mtk->vusb33);
> > +	if (ret) {
> > +		dev_err(mtk->dev, "failed to enable vusb33\n");
> > +		regulator_disable(mtk->vbus);
> > +		return ret;
> > +	}
> > +	return 0;
> > +}
> > +
> > +static void xhci_mtk_ldos_disable(struct xhci_hcd_mtk *mtk)
> > +{
> > +	regulator_disable(mtk->vbus);
> > +	regulator_disable(mtk->vusb33);
> > +}
> > +
> > +static void xhci_mtk_quirks(struct device *dev, struct xhci_hcd *xhci)
> > +{
> > +	/*
> > +	 * As of now platform drivers don't provide MSI support so we ensure
> > +	 * here that the generic code does not try to make a pci_dev from our
> > +	 * dev struct in order to setup MSI
> > +	 */
> > +	xhci->quirks |= XHCI_PLAT;
> > +	xhci->quirks |= XHCI_MTK_HOST;
> > +	/*
> > +	 * MTK host controller gives a spurious successful event after a
> > +	 * short transfer. Ignore it.
> > +	 */
> > +	xhci->quirks |= XHCI_SPURIOUS_SUCCESS;
> > +}
> > +
> > +/* called during probe() after chip reset completes */
> > +static int xhci_mtk_setup(struct usb_hcd *hcd)
> > +{
> > +	struct xhci_hcd *xhci;
> > +	int ret;
> > +
> > +	ret = xhci_gen_setup(hcd, xhci_mtk_quirks);
> > +	if (ret)
> > +		return ret;
> > +
> > +	if (!usb_hcd_is_primary_hcd(hcd))
> > +		return 0;
> > +
> > +	xhci = hcd_to_xhci(hcd);
> > +	return xhci_mtk_sch_init(xhci);
> > +}
> > +
> > +
> > +static int xhci_mtk_probe(struct platform_device *pdev)
> > +{
> > +	struct device *dev = &pdev->dev;
> > +	struct device_node *node = dev->of_node;
> > +	struct usb_xhci_pdata *pdata = dev_get_platdata(dev);
> > +	struct xhci_hcd_mtk *mtk;
> > +	const struct hc_driver *driver;
> > +	struct xhci_hcd *xhci;
> > +	struct resource *res;
> > +	struct usb_hcd *hcd;
> > +	struct phy *phy;
> > +	int phy_num;
> > +	int ret = -ENODEV;
> > +	int irq;
> > +
> > +	if (usb_disabled())
> > +		return -ENODEV;
> > +
> > +	driver = &xhci_mtk_hc_driver;
> > +	mtk = devm_kzalloc(dev, sizeof(*mtk), GFP_KERNEL);
> > +	if (!mtk)
> > +		return -ENOMEM;
> > +
> > +	mtk->dev = dev;
> > +	mtk->vbus = devm_regulator_get(dev, "vbus");
> > +	if (IS_ERR(mtk->vbus)) {
> > +		dev_err(dev, "fail to get vbus\n");
> > +		return PTR_ERR(mtk->vbus);
> > +	}
> > +
> > +	mtk->vusb33 = devm_regulator_get(dev, "vusb33");
> > +	if (IS_ERR(mtk->vusb33)) {
> > +		dev_err(dev, "fail to get vusb33\n");
> > +		return PTR_ERR(mtk->vusb33);
> > +	}
> > +
> > +	mtk->sys_clk = devm_clk_get(dev, "sys_ck");
> > +	if (IS_ERR(mtk->sys_clk)) {
> > +		dev_err(dev, "fail to get sys_ck\n");
> > +		return PTR_ERR(mtk->sys_clk);
> > +	}
> > +
> > +	ret = usb_wakeup_of_property_parse(mtk, node);
> > +	if (ret)
> > +		return ret;
> > +
> > +	mtk->num_phys = of_count_phandle_with_args(node,
> > +			"phys", "#phy-cells");
> > +	if (mtk->num_phys > 0) {
> > +		mtk->phys = devm_kcalloc(dev, mtk->num_phys,
> > +					sizeof(*mtk->phys), GFP_KERNEL);
> > +		if (!mtk->phys)
> > +			return -ENOMEM;
> > +	} else {
> > +		mtk->num_phys = 0;
> > +	}
> > +	pm_runtime_enable(dev);
> > +	pm_runtime_get_sync(dev);
> > +
> > +	ret = xhci_mtk_ldos_enable(mtk);
> > +	if (ret)
> > +		goto disable_pm;
> > +
> > +	ret = xhci_mtk_clks_enable(mtk);
> > +	if (ret)
> > +		goto disable_ldos;
> > +
> > +	irq = platform_get_irq(pdev, 0);
> > +	if (irq < 0)
> > +		goto disable_clk;
> > +
> > +	/* Initialize dma_mask and coherent_dma_mask to 32-bits */
> > +	ret = dma_set_coherent_mask(dev, DMA_BIT_MASK(32));
> > +	if (ret)
> > +		goto disable_clk;
> > +
> > +	if (!dev->dma_mask)
> > +		dev->dma_mask = &dev->coherent_dma_mask;
> > +	else
> > +		dma_set_mask(dev, DMA_BIT_MASK(32));
> > +
> > +	hcd = usb_create_hcd(driver, dev, dev_name(dev));
> > +	if (!hcd) {
> > +		ret = -ENOMEM;
> > +		goto disable_clk;
> > +	}
> > +
> > +	/*
> > +	 * USB 2.0 roothub is stored in the platform_device.
> > +	 * Swap it with mtk HCD.
> > +	 */
> > +	mtk->hcd = platform_get_drvdata(pdev);
> > +	platform_set_drvdata(pdev, mtk);
> > +
> > +	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
> > +	hcd->regs = devm_ioremap_resource(dev, res);
> > +	if (IS_ERR(hcd->regs)) {
> > +		ret = PTR_ERR(hcd->regs);
> > +		goto put_usb2_hcd;
> > +	}
> > +	hcd->rsrc_start = res->start;
> > +	hcd->rsrc_len = resource_size(res);
> > +
> > +	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
> > +	mtk->ippc_regs = devm_ioremap_resource(dev, res);
> > +	if (IS_ERR(mtk->ippc_regs)) {
> > +		ret = PTR_ERR(mtk->ippc_regs);
> > +		goto put_usb2_hcd;
> > +	}
> > +
> > +	for (phy_num = 0; phy_num < mtk->num_phys; phy_num++) {
> > +		phy = devm_of_phy_get_by_index(dev, node, phy_num);
> > +		if (IS_ERR(phy)) {
> > +			ret = PTR_ERR(phy);
> > +			goto put_usb2_hcd;
> > +		}
> > +		mtk->phys[phy_num] = phy;
> > +	}
> > +
> > +	xhci_mtk_ports_config(mtk);
> > +	ret = xhci_mtk_phy_init(mtk);
> > +	if (ret)
> > +		goto put_usb2_hcd;
> > +
> > +	ret = xhci_mtk_phy_power_on(mtk);
> > +	if (ret)
> > +		goto exit_phys;
> > +
> > +	device_init_wakeup(dev, 1);
> > +
> > +	xhci = hcd_to_xhci(hcd);
> > +	xhci->main_hcd = hcd;
> > +	xhci->shared_hcd = usb_create_shared_hcd(driver, dev,
> > +			dev_name(dev), hcd);
> > +	if (!xhci->shared_hcd) {
> > +		ret = -ENOMEM;
> > +		goto power_off_phys;
> > +	}
> > +
> > +	if ((node && of_property_read_bool(node, "usb3-lpm-capable")) ||
> > +			(pdata && pdata->usb3_lpm_capable))
> > +		xhci->quirks |= XHCI_LPM_SUPPORT;
> > +
> > +	if (HCC_MAX_PSA(xhci->hcc_params) >= 4)
> > +		xhci->shared_hcd->can_do_streams = 1;
> > +
> > +	ret = usb_add_hcd(hcd, irq, IRQF_SHARED);
> > +	if (ret)
> > +		goto put_usb3_hcd;
> > +
> > +	ret = usb_add_hcd(xhci->shared_hcd, irq, IRQF_SHARED);
> > +	if (ret)
> > +		goto dealloc_usb2_hcd;
> > +
> > +	return 0;
> > +
> > +dealloc_usb2_hcd:
> > +	xhci_mtk_sch_exit(xhci);
> > +	usb_remove_hcd(hcd);
> > +
> > +put_usb3_hcd:
> > +	usb_put_hcd(xhci->shared_hcd);
> > +
> > +power_off_phys:
> > +	xhci_mtk_phy_power_off(mtk);
> > +	device_init_wakeup(dev, 0);
> > +
> > +exit_phys:
> > +	xhci_mtk_phy_exit(mtk);
> > +
> > +put_usb2_hcd:
> > +	usb_put_hcd(hcd);
> > +
> > +disable_clk:
> > +	xhci_mtk_clks_disable(mtk);
> > +
> > +disable_ldos:
> > +	xhci_mtk_ldos_disable(mtk);
> > +
> > +disable_pm:
> > +	pm_runtime_put_sync(dev);
> > +	pm_runtime_disable(dev);
> > +	return ret;
> > +}
> > +
> > +static int xhci_mtk_remove(struct platform_device *dev)
> > +{
> > +	struct xhci_hcd_mtk *mtk = platform_get_drvdata(dev);
> > +	struct usb_hcd	*hcd = mtk->hcd;
> > +	struct xhci_hcd	*xhci = hcd_to_xhci(hcd);
> > +
> > +	usb_remove_hcd(xhci->shared_hcd);
> > +	xhci_mtk_phy_power_off(mtk);
> > +	xhci_mtk_phy_exit(mtk);
> > +	device_init_wakeup(&dev->dev, 0);
> > +
> > +	usb_remove_hcd(hcd);
> > +	usb_put_hcd(xhci->shared_hcd);
> > +	usb_put_hcd(hcd);
> > +	xhci_mtk_sch_exit(xhci);
> > +	xhci_mtk_clks_disable(mtk);
> > +	xhci_mtk_ldos_disable(mtk);
> > +	pm_runtime_put_sync(&dev->dev);
> > +	pm_runtime_disable(&dev->dev);
> > +
> > +	return 0;
> > +}
> > +
> > +#ifdef CONFIG_PM_SLEEP
> > +static int xhci_mtk_suspend(struct device *dev)
> > +{
> > +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> > +
> > +	xhci_mtk_ports_disable(mtk);
> > +	xhci_mtk_phy_power_off(mtk);
> > +	xhci_mtk_clks_disable(mtk);
> > +	usb_wakeup_enable(mtk);
> > +	return 0;
> > +}
> > +
> > +static int xhci_mtk_resume(struct device *dev)
> > +{
> > +	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
> > +
> > +	usb_wakeup_disable(mtk);
> > +	xhci_mtk_clks_enable(mtk);
> > +	xhci_mtk_phy_power_on(mtk);
> > +	xhci_mtk_ports_enable(mtk);
> > +	return 0;
> > +}
> > +
> > +static const struct dev_pm_ops xhci_mtk_pm_ops = {
> > +	SET_SYSTEM_SLEEP_PM_OPS(xhci_mtk_suspend, xhci_mtk_resume)
> > +};
> > +#define DEV_PM_OPS	(&xhci_mtk_pm_ops)
> > +#else
> > +#define DEV_PM_OPS	NULL
> > +#endif /* CONFIG_PM */
> > +
> > +#ifdef CONFIG_OF
> > +static const struct of_device_id mtk_xhci_of_match[] = {
> > +	{ .compatible = "mediatek,mt8173-xhci"},
> > +	{ },
> > +};
> > +MODULE_DEVICE_TABLE(of, mtk_xhci_of_match);
> > +#endif
> > +
> > +static struct platform_driver mtk_xhci_driver = {
> > +	.probe	= xhci_mtk_probe,
> > +	.remove	= xhci_mtk_remove,
> > +	.driver	= {
> > +		.name = "xhci-mtk",
> > +		.pm = DEV_PM_OPS,
> > +		.of_match_table = of_match_ptr(mtk_xhci_of_match),
> > +	},
> > +};
> > +
> > +static int __init xhci_mtk_init(void)
> > +{
> > +	xhci_init_driver(&xhci_mtk_hc_driver, &xhci_mtk_overrides);
> > +	return platform_driver_register(&mtk_xhci_driver);
> > +}
> > +module_init(xhci_mtk_init);
> > +
> > +static void __exit xhci_mtk_exit(void)
> > +{
> > +	platform_driver_unregister(&mtk_xhci_driver);
> > +}
> > +module_exit(xhci_mtk_exit);
> > +
> > +MODULE_DESCRIPTION("MediaTek xHCI Host Controller Driver");
> > +MODULE_LICENSE("GPL v2");
> > diff --git a/drivers/usb/host/xhci-mtk.h b/drivers/usb/host/xhci-mtk.h
> > new file mode 100644
> > index 0000000..8913c2f
> > --- /dev/null
> > +++ b/drivers/usb/host/xhci-mtk.h
> > @@ -0,0 +1,156 @@
> > +/*
> > + * Copyright (c) 2015 MediaTek Inc.
> > + * Author:
> > + *  Zhigang.Wei <zhigang.wei@mediatek.com>
> > + *  Chunfeng.Yun <chunfeng.yun@mediatek.com>
> > + *
> > + * This software is licensed under the terms of the GNU General Public
> > + * License version 2, as published by the Free Software Foundation, and
> > + * may be copied, distributed, and modified under those terms.
> > + *
> > + * This program is distributed in the hope that 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.
> > + *
> > + */
> > +
> > +#ifndef _XHCI_MTK_H_
> > +#define _XHCI_MTK_H_
> > +
> > +#include "xhci.h"
> > +
> > +/**
> > + * To simplify scheduler algorithm, set a upper limit for ESIT,
> > + * if a synchromous ep's ESIT is larger than @XHCI_MTK_MAX_ESIT,
> > + * round down to the limit value, that means allocating more
> > + * bandwidth to it.
> > + */
> > +#define XHCI_MTK_MAX_ESIT	64
> > +
> > +/**
> > + * struct mu3h_sch_bw_info: schedule information for bandwidth domain
> > + *
> > + * @bus_bw: array to keep track of bandwidth already used at each uframes
> > + * @bw_ep_list: eps in the bandwidth domain
> > + *
> > + * treat a HS root port as a bandwidth domain, but treat a SS root port as
> > + * two bandwidth domains, one for IN eps and another for OUT eps.
> > + */
> > +struct mu3h_sch_bw_info {
> > +	u32 bus_bw[XHCI_MTK_MAX_ESIT];
> > +	struct list_head bw_ep_list;
> > +};
> > +
> > +/**
> > + * struct mu3h_sch_ep_info: schedule information for endpoint
> > + *
> > + * @esit: unit is 125us, equal to 2 << Interval field in ep-context
> > + * @num_budget_microframes: number of continuous uframes
> > + *		(@repeat==1) scheduled within the interval
> > + * @bw_cost_per_microframe: bandwidth cost per microframe
> > + * @endpoint: linked into bandwidth domain which it belongs to
> > + * @ep: address of usb_host_endpoint struct
> > + * @offset: which uframe of the interval that transfer should be
> > + *		scheduled first time within the interval
> > + * @repeat: the time gap between two uframes that transfers are
> > + *		scheduled within a interval. in the simple algorithm, only
> > + *		assign 0 or 1 to it; 0 means using only one uframe in a
> > + *		interval, and 1 means using @num_budget_microframes
> > + *		continuous uframes
> > + * @pkts: number of packets to be transferred in the scheduled uframes
> > + * @cs_count: number of CS that host will trigger
> > + * @burst_mode: burst mode for scheduling. 0: normal burst mode,
> > + *		distribute the bMaxBurst+1 packets for a single burst
> > + *		according to @pkts and @repeat, repeate the burst multiple
> > + *		times; 1: distribute the (bMaxBurst+1)*(Mult+1) packets
> > + *		according to @pkts and @repeat. normal mode is used by
> > + *		default
> > + */
> > +struct mu3h_sch_ep_info {
> > +	u32 esit;
> > +	u32 num_budget_microframes;
> > +	u32 bw_cost_per_microframe;
> > +	struct list_head endpoint;
> > +	void *ep;
> > +	/*
> > +	 * mtk xHCI scheduling information put into reserved DWs
> > +	 * in ep context
> > +	 */
> > +	u32 offset;
> > +	u32 repeat;
> > +	u32 pkts;
> > +	u32 cs_count;
> > +	u32 burst_mode;
> > +};
> > +
> > +#define MU3C_U3_PORT_MAX 4
> > +#define MU3C_U2_PORT_MAX 5
> > +
> > +/**
> > + * struct mu3c_ippc_regs: MTK ssusb ip port control registers
> > + * @ip_pw_ctr0~3: ip power and clock control registers
> > + * @ip_pw_sts1~2: ip power and clock status registers
> > + * @ip_xhci_cap: ip xHCI capability register
> > + * @u3_ctrl_p[x]: ip usb3 port x control register, only low 4bytes are used
> > + * @u2_ctrl_p[x]: ip usb2 port x control register, only low 4bytes are used
> > + * @u2_phy_pll: usb2 phy pll control register
> > + */
> > +struct mu3c_ippc_regs {
> > +	__le32 ip_pw_ctr0;
> > +	__le32 ip_pw_ctr1;
> > +	__le32 ip_pw_ctr2;
> > +	__le32 ip_pw_ctr3;
> > +	__le32 ip_pw_sts1;
> > +	__le32 ip_pw_sts2;
> > +	__le32 reserved0[3];
> > +	__le32 ip_xhci_cap;
> > +	__le32 reserved1[2];
> > +	__le64 u3_ctrl_p[MU3C_U3_PORT_MAX];
> > +	__le64 u2_ctrl_p[MU3C_U2_PORT_MAX];
> > +	__le32 reserved2;
> > +	__le32 u2_phy_pll;
> > +	__le32 reserved3[33]; /* 0x80 ~ 0xff */
> > +};
> > +
> > +struct xhci_hcd_mtk {
> > +	struct device *dev;
> > +	struct usb_hcd *hcd;
> > +	struct mu3h_sch_bw_info *sch_array;
> > +	struct mu3c_ippc_regs __iomem *ippc_regs;
> > +	int num_u2_ports;
> > +	int num_u3_ports;
> > +	struct regulator *vusb33;
> > +	struct regulator *vbus;
> > +	struct clk *sys_clk;	/* sys and mac clock */
> > +	struct clk *wk_deb_p0;	/* port0's wakeup debounce clock */
> > +	struct clk *wk_deb_p1;
> > +	struct regmap *pericfg;
> > +	int wakeup_src;
> > +	struct phy **phys;
> > +	int num_phys;
> > +};
> > +
> > +#if IS_ENABLED(CONFIG_USB_XHCI_MTK)
> > +int xhci_mtk_sch_init(struct xhci_hcd *xhci);
> > +void xhci_mtk_sch_exit(struct xhci_hcd *xhci);
> > +int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> > +		struct usb_host_endpoint *ep);
> > +void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
> > +		struct usb_host_endpoint *ep);
> > +
> > +#else
> > +static inline int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd,
> > +	struct usb_device *udev, struct usb_host_endpoint *ep)
> > +{
> > +	return 0;
> > +}
> > +
> > +static inline void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd,
> > +	struct usb_device *udev, struct usb_host_endpoint *ep)
> > +{
> > +}
> > +
> > +#endif
> > +
> > +#endif		/* _XHCI_MTK_H_ */
> > diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
> > index 57f40a1..243f696 100644
> > --- a/drivers/usb/host/xhci-ring.c
> > +++ b/drivers/usb/host/xhci-ring.c
> > @@ -68,6 +68,7 @@
> >   #include <linux/slab.h>
> >   #include "xhci.h"
> >   #include "xhci-trace.h"
> > +#include "xhci-mtk.h"
> >
> >   /*
> >    * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
> > @@ -3044,18 +3045,27 @@ static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
> >   			      struct urb *urb, unsigned int num_trbs_left)
> >   {
> >   	u32 maxp, total_packet_count;
> > +	u32 skip_current_trb = 0;
> >
> > -	if (xhci->hci_version < 0x100)
> > -		return ((td_total_len - transferred) >> 10);
> > -
> > -	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> > -	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> > +	if (xhci->hci_version < 0x100) {
> > +		/* for MTK xHCI, TD size doesn't include this TRB */
> > +		if (xhci->quirks & XHCI_MTK_HOST)
> > +			skip_current_trb = 1;
> > +		else
> > +			return ((td_total_len - transferred) >> 10);
> > +	}
> >
> >   	/* One TRB with a zero-length data packet. */
> >   	if (num_trbs_left == 0 || (transferred == 0 && trb_buff_len == 0) ||
> >   	    trb_buff_len == td_total_len)
> >   		return 0;
> >
> > +	if (skip_current_trb)
> > +		trb_buff_len = 0;
> > +
> > +	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
> > +	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
> > +
> >   	/* Queueing functions don't count the current TRB into transferred */
> >   	return (total_packet_count - ((transferred + trb_buff_len) / maxp));
> >   }
> > @@ -3443,7 +3453,7 @@ int xhci_queue_ctrl_tx(struct xhci_hcd *xhci, gfp_t mem_flags,
> >   		field |= 0x1;
> >
> >   	/* xHCI 1.0 6.4.1.2.1: Transfer Type field */
> > -	if (xhci->hci_version == 0x100) {
> > +	if ((xhci->hci_version == 0x100) || (xhci->quirks & XHCI_MTK_HOST)) {
> >   		if (urb->transfer_buffer_length > 0) {
> >   			if (setup->bRequestType & USB_DIR_IN)
> >   				field |= TRB_TX_TYPE(TRB_DATA_IN);
> > diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
> > index 6b0f4a4..4741188 100644
> > --- a/drivers/usb/host/xhci.c
> > +++ b/drivers/usb/host/xhci.c
> > @@ -31,6 +31,7 @@
> >
> >   #include "xhci.h"
> >   #include "xhci-trace.h"
> > +#include "xhci-mtk.h"
> >
> >   #define DRIVER_AUTHOR "Sarah Sharp"
> >   #define DRIVER_DESC "'eXtensible' Host Controller (xHC) Driver"
> > @@ -623,7 +624,11 @@ int xhci_run(struct usb_hcd *hcd)
> >   			"// Set the interrupt modulation register");
> >   	temp = readl(&xhci->ir_set->irq_control);
> >   	temp &= ~ER_IRQ_INTERVAL_MASK;
> > -	temp |= (u32) 160;
> > +	/*
> > +	 * the increment interval is 8 times as much as that defined
> > +	 * in xHCI spec on MTK's controller
> > +	 */
> > +	temp |= (u32) ((xhci->quirks & XHCI_MTK_HOST) ? 20 : 160);
> >   	writel(temp, &xhci->ir_set->irq_control);
> >
> >   	/* Set the HCD state before we enable the irqs */
> > @@ -1693,6 +1698,9 @@ int xhci_drop_endpoint(struct usb_hcd *hcd, struct usb_device *udev,
> >
> >   	xhci_endpoint_zero(xhci, xhci->devs[udev->slot_id], ep);
> >
> > +	if (xhci->quirks & XHCI_MTK_HOST)
> > +		xhci_mtk_drop_ep_quirk(hcd, udev, ep);
> > +
> >   	xhci_dbg(xhci, "drop ep 0x%x, slot id %d, new drop flags = %#x, new add flags = %#x\n",
> >   			(unsigned int) ep->desc.bEndpointAddress,
> >   			udev->slot_id,
> > @@ -1788,6 +1796,15 @@ int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev,
> >   		return -ENOMEM;
> >   	}
> >
> > +	if (xhci->quirks & XHCI_MTK_HOST) {
> > +		ret = xhci_mtk_add_ep_quirk(hcd, udev, ep);
> > +		if (ret < 0) {
> > +			xhci_free_or_cache_endpoint_ring(xhci,
> > +				virt_dev, ep_index);
> > +			return ret;
> > +		}
> > +	}
> > +
> >   	ctrl_ctx->add_flags |= cpu_to_le32(added_ctxs);
> >   	new_add_flags = le32_to_cpu(ctrl_ctx->add_flags);
> >
> > diff --git a/drivers/usb/host/xhci.h b/drivers/usb/host/xhci.h
> > index ea8b904..1536da0 100644
> > --- a/drivers/usb/host/xhci.h
> > +++ b/drivers/usb/host/xhci.h
> > @@ -1577,6 +1577,7 @@ struct xhci_hcd {
> >   /* For controllers with a broken beyond repair streams implementation */
> >   #define XHCI_BROKEN_STREAMS	(1 << 19)
> >   #define XHCI_PME_STUCK_QUIRK	(1 << 20)
> > +#define XHCI_MTK_HOST		(1 << 21)
> >   	unsigned int		num_active_eps;
> >   	unsigned int		limit_active_eps;
> >   	/* There are two roothubs to keep track of bus suspend info for */
> >
>
Mathias Nyman Oct. 19, 2015, 11:25 a.m. UTC | #9
>>
>> So basically we are trying to use as many microframes as possible with as few packets
>> per microframe as possible.
>>
>> Did I understand this correctly?
> Yes, you are right.
>
>> How will devices react if they expect to get 16 packets every 16th microframe,
>> but they get one packet every microframe instead?
> I think that the synchronous endpoint must specify its period by
> bInterval, but can't specify how data should be transfered during the
> period by the host, and it just only receives data passively. So the
> device can receive data correctly in the case(bInterval is 5).
>
> quote from usb3_r1.0 section4.4.8 Isochronous Transfers:
> "The host can request data from the device or send data to the device at
> any time during the service interval for a particular endpoint on that
> device"
>

As I understand the 4.4.8 section it just means the device can't assume a fixed
time interval between transfers, meaning that the host can use the last microframe
in one esit and the first microframe in the next esit, but still only use 1 microframe
per esit.

Section 8.12.6.1 describes how a 11 packet isoc transfer is allowed to be split
to 1 burst of 11 packets, 2 burst (8 + 3),  3 burst (4+4+3) 6 bursts (2+2+2+2+2+1) or
11 bursts of 1. These are however all within the same microframe. Splitting the
transfer into several microframes in a esit kind of makes the whole interval concept pointless.

-Mathias
Chunfeng Yun (云春峰) Oct. 20, 2015, 6:29 a.m. UTC | #10
hi,
On Mon, 2015-10-19 at 14:25 +0300, Mathias Nyman wrote:
> >>
> >> So basically we are trying to use as many microframes as possible with as few packets
> >> per microframe as possible.
> >>
> >> Did I understand this correctly?
> > Yes, you are right.
> >
> >> How will devices react if they expect to get 16 packets every 16th microframe,
> >> but they get one packet every microframe instead?
> > I think that the synchronous endpoint must specify its period by
> > bInterval, but can't specify how data should be transfered during the
> > period by the host, and it just only receives data passively. So the
> > device can receive data correctly in the case(bInterval is 5).
> >
> > quote from usb3_r1.0 section4.4.8 Isochronous Transfers:
> > "The host can request data from the device or send data to the device at
> > any time during the service interval for a particular endpoint on that
> > device"
> >
> 
> As I understand the 4.4.8 section it just means the device can't assume a fixed
> time interval between transfers, meaning that the host can use the last microframe
> in one esit and the first microframe in the next esit, but still only use 1 microframe
> per esit.
> 
> Section 8.12.6.1 describes how a 11 packet isoc transfer is allowed to be split
> to 1 burst of 11 packets, 2 burst (8 + 3),  3 burst (4+4+3) 6 bursts (2+2+2+2+2+1) or
> 11 bursts of 1. These are however all within the same microframe. Splitting the
> transfer into several microframes in a esit kind of makes the whole interval concept pointless.
> 
It doesn't say that the packets should be transfered within the same
microframe (bus interval), as I understand it means service interval;

The direct prove resides in figure 8-56/8-57.

Term: 
1. BI, bus interval, a 125 us period that establishes the internal
boundary of service interval, aka uframe;
2. SSI, Support Smart Isochronous;
3. DBI, Data in this Bus Interval is done;
4. NBI, Numbers of Bus Interval;

As the figure shows, the service interval = 8 BI, that host distribute 2
packets @1st uframe, keep U1/U2 state for the next 3uframe, then
transmit 4 packets @4th uframe, and the remaining 3 packet in the last
frame.

Please notice that this just is an example illustrated by spec, but we
can derive the conclusion that the distribution of packet in a service
interval is completely decided by host, and can split isoc transfers
across multiple uframes.

PS: as you can see, MTK implementation of schedule algorithms is an
implementation of Smart Isochronous of which the smart side resides in
software.


> -Mathias
> 
> 
> 
>   
>
Mathias Nyman Oct. 20, 2015, 8:25 a.m. UTC | #11
On 20.10.2015 09:29, chunfeng yun wrote:
> hi,
> On Mon, 2015-10-19 at 14:25 +0300, Mathias Nyman wrote:
>>>>
>>>> So basically we are trying to use as many microframes as possible with as few packets
>>>> per microframe as possible.
>>>>
>>>> Did I understand this correctly?
>>> Yes, you are right.
>>>
>>>> How will devices react if they expect to get 16 packets every 16th microframe,
>>>> but they get one packet every microframe instead?
>>> I think that the synchronous endpoint must specify its period by
>>> bInterval, but can't specify how data should be transfered during the
>>> period by the host, and it just only receives data passively. So the
>>> device can receive data correctly in the case(bInterval is 5).
>>>
>>> quote from usb3_r1.0 section4.4.8 Isochronous Transfers:
>>> "The host can request data from the device or send data to the device at
>>> any time during the service interval for a particular endpoint on that
>>> device"
>>>
>>
>> As I understand the 4.4.8 section it just means the device can't assume a fixed
>> time interval between transfers, meaning that the host can use the last microframe
>> in one esit and the first microframe in the next esit, but still only use 1 microframe
>> per esit.
>>
>> Section 8.12.6.1 describes how a 11 packet isoc transfer is allowed to be split
>> to 1 burst of 11 packets, 2 burst (8 + 3),  3 burst (4+4+3) 6 bursts (2+2+2+2+2+1) or
>> 11 bursts of 1. These are however all within the same microframe. Splitting the
>> transfer into several microframes in a esit kind of makes the whole interval concept pointless.
>>
> It doesn't say that the packets should be transfered within the same
> microframe (bus interval), as I understand it means service interval;
>
> The direct prove resides in figure 8-56/8-57.
>
> Term:
> 1. BI, bus interval, a 125 us period that establishes the internal
> boundary of service interval, aka uframe;
> 2. SSI, Support Smart Isochronous;
> 3. DBI, Data in this Bus Interval is done;
> 4. NBI, Numbers of Bus Interval;
>
> As the figure shows, the service interval = 8 BI, that host distribute 2
> packets @1st uframe, keep U1/U2 state for the next 3uframe, then
> transmit 4 packets @4th uframe, and the remaining 3 packet in the last
> frame.
>
> Please notice that this just is an example illustrated by spec, but we
> can derive the conclusion that the distribution of packet in a service
> interval is completely decided by host, and can split isoc transfers
> across multiple uframes.

So it seems. You're right

>
> PS: as you can see, MTK implementation of schedule algorithms is an
> implementation of Smart Isochronous of which the smart side resides in
> software.

Thanks for the clarification, I now understand how the implementation works

-Mathias
diff mbox

Patch

diff --git a/drivers/usb/host/Kconfig b/drivers/usb/host/Kconfig
index 079991e..4674118 100644
--- a/drivers/usb/host/Kconfig
+++ b/drivers/usb/host/Kconfig
@@ -41,6 +41,15 @@  config USB_XHCI_PLATFORM
 
 	  If unsure, say N.
 
+config USB_XHCI_MTK
+	tristate "xHCI support for Mediatek MT65xx"
+	select MFD_SYSCON
+	depends on ARCH_MEDIATEK || COMPILE_TEST
+	---help---
+	  Say 'Y' to enable the support for the xHCI host controller
+	  found in Mediatek MT65xx SoCs.
+	  If unsure, say N.
+
 config USB_XHCI_MVEBU
 	tristate "xHCI support for Marvell Armada 375/38x"
 	select USB_XHCI_PLATFORM
diff --git a/drivers/usb/host/Makefile b/drivers/usb/host/Makefile
index 754efaa..00401f9 100644
--- a/drivers/usb/host/Makefile
+++ b/drivers/usb/host/Makefile
@@ -13,6 +13,9 @@  fhci-$(CONFIG_FHCI_DEBUG) += fhci-dbg.o
 xhci-hcd-y := xhci.o xhci-mem.o
 xhci-hcd-y += xhci-ring.o xhci-hub.o xhci-dbg.o
 xhci-hcd-y += xhci-trace.o
+ifneq ($(CONFIG_USB_XHCI_MTK), )
+	xhci-hcd-y += xhci-mtk-sch.o
+endif
 
 xhci-plat-hcd-y := xhci-plat.o
 ifneq ($(CONFIG_USB_XHCI_MVEBU), )
@@ -30,6 +33,7 @@  endif
 
 obj-$(CONFIG_USB_XHCI_PCI)	+= xhci-pci.o
 obj-$(CONFIG_USB_XHCI_PLATFORM) += xhci-plat-hcd.o
+obj-$(CONFIG_USB_XHCI_MTK)	+= xhci-mtk.o
 
 obj-$(CONFIG_USB_EHCI_HCD)	+= ehci-hcd.o
 obj-$(CONFIG_USB_EHCI_PCI)	+= ehci-pci.o
diff --git a/drivers/usb/host/xhci-mtk-sch.c b/drivers/usb/host/xhci-mtk-sch.c
new file mode 100644
index 0000000..163bc6d
--- /dev/null
+++ b/drivers/usb/host/xhci-mtk-sch.c
@@ -0,0 +1,424 @@ 
+/*
+ * Copyright (c) 2015 MediaTek Inc.
+ * Author:
+ *  Zhigang.Wei <zhigang.wei@mediatek.com>
+ *  Chunfeng.Yun <chunfeng.yun@mediatek.com>
+ *
+ * This software is licensed under the terms of the GNU General Public
+ * License version 2, as published by the Free Software Foundation, and
+ * may be copied, distributed, and modified under those terms.
+ *
+ * This program is distributed in the hope that 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.
+ *
+ */
+
+#include <linux/kernel.h>
+#include <linux/module.h>
+#include <linux/slab.h>
+
+#include "xhci.h"
+#include "xhci-mtk.h"
+
+#define SS_BW_BOUNDARY	51000
+/* table 5-5. High-speed Isoc Transaction Limits in usb_20 spec */
+#define HS_BW_BOUNDARY	6144
+/* usb2 spec section11.18.1: at most 188 FS bytes per microframe */
+#define FS_PAYLOAD_MAX 188
+
+/* mtk scheduler bitmasks */
+#define EP_BPKTS(p)	((p) & 0x3f)
+#define EP_BCSCOUNT(p)	(((p) & 0x7) << 8)
+#define EP_BBM(p)	((p) << 11)
+#define EP_BOFFSET(p)	((p) & 0x3fff)
+#define EP_BREPEAT(p)	(((p) & 0x7fff) << 16)
+
+static int is_fs_or_ls(enum usb_device_speed speed)
+{
+	return speed == USB_SPEED_FULL || speed == USB_SPEED_LOW;
+}
+
+/*
+* get the index of bandwidth domains array which @ep belongs to.
+*
+* the bandwidth domain array is saved to @sch_array of struct xhci_hcd_mtk,
+* each HS root port is treated as a single bandwidth domain,
+* but each SS root port is treated as two bandwidth domains, one for IN eps,
+* one for OUT eps.
+* @real_port value is defined as follow according to xHCI spec:
+* 1 for SSport0, ..., N+1 for SSportN, N+2 for HSport0, N+3 for HSport1, etc
+* so the bandwidth domain array is organized as follow for simplification:
+* SSport0-OUT, SSport0-IN, ..., SSportX-OUT, SSportX-IN, HSport0, ..., HSportY
+*/
+static int get_bw_index(struct xhci_hcd *xhci, struct usb_device *udev,
+	struct usb_host_endpoint *ep)
+{
+	struct xhci_virt_device *virt_dev;
+	int bw_index;
+
+	virt_dev = xhci->devs[udev->slot_id];
+
+	if (udev->speed == USB_SPEED_SUPER) {
+		if (usb_endpoint_dir_out(&ep->desc))
+			bw_index = (virt_dev->real_port - 1) * 2;
+		else
+			bw_index = (virt_dev->real_port - 1) * 2 + 1;
+	} else {
+		/* add one more for each SS port */
+		bw_index = virt_dev->real_port + xhci->num_usb3_ports - 1;
+	}
+
+	return bw_index;
+}
+
+static void setup_sch_info(struct usb_device *udev,
+		struct xhci_ep_ctx *ep_ctx, struct mu3h_sch_ep_info *sch_ep)
+{
+	u32 ep_type;
+	u32 ep_interval;
+	u32 max_packet_size;
+	u32 max_burst;
+	u32 mult;
+	u32 esit_pkts;
+
+	ep_type = CTX_TO_EP_TYPE(le32_to_cpu(ep_ctx->ep_info2));
+	ep_interval = CTX_TO_EP_INTERVAL(le32_to_cpu(ep_ctx->ep_info));
+	max_packet_size = MAX_PACKET_DECODED(le32_to_cpu(ep_ctx->ep_info2));
+	max_burst = CTX_TO_MAX_BURST(le32_to_cpu(ep_ctx->ep_info2));
+	mult = CTX_TO_EP_MULT(le32_to_cpu(ep_ctx->ep_info));
+
+	sch_ep->esit = 1 << ep_interval;
+	sch_ep->offset = 0;
+	sch_ep->burst_mode = 0;
+
+	if (udev->speed == USB_SPEED_HIGH) {
+		sch_ep->cs_count = 0;
+
+		/*
+		 * usb_20 spec section5.9
+		 * a single microframe is enough for HS synchromous endpoints
+		 * in a interval
+		 */
+		sch_ep->num_budget_microframes = 1;
+		sch_ep->repeat = 0;
+
+		/*
+		 * xHCI spec section6.2.3.4
+		 * @max_burst is the number of additional transactions
+		 * opportunities per microframe
+		 */
+		sch_ep->pkts = max_burst + 1;
+		sch_ep->bw_cost_per_microframe = max_packet_size * sch_ep->pkts;
+	} else if (udev->speed == USB_SPEED_SUPER) {
+		/* usb3_r1 spec section4.4.7 & 4.4.8 */
+		sch_ep->cs_count = 0;
+		esit_pkts = (mult + 1) * (max_burst + 1);
+		if (ep_type == INT_IN_EP || ep_type == INT_OUT_EP) {
+			sch_ep->pkts = esit_pkts;
+			sch_ep->num_budget_microframes = 1;
+			sch_ep->repeat = 0;
+		}
+
+		if (ep_type == ISOC_IN_EP || ep_type == ISOC_OUT_EP) {
+			if (esit_pkts <= sch_ep->esit)
+				sch_ep->pkts = 1;
+			else
+				sch_ep->pkts = roundup_pow_of_two(esit_pkts)
+					/ sch_ep->esit;
+
+			sch_ep->num_budget_microframes =
+				DIV_ROUND_UP(esit_pkts, sch_ep->pkts);
+
+			if (sch_ep->num_budget_microframes > 1)
+				sch_ep->repeat = 1;
+			else
+				sch_ep->repeat = 0;
+		}
+		sch_ep->bw_cost_per_microframe = max_packet_size * sch_ep->pkts;
+	} else if (is_fs_or_ls(udev->speed)) {
+
+		/*
+		 * usb_20 spec section11.18.4
+		 * assume worst cases
+		 */
+		sch_ep->repeat = 0;
+		sch_ep->pkts = 1; /* at most one packet for each microframe */
+		if (ep_type == INT_IN_EP || ep_type == INT_OUT_EP) {
+			sch_ep->cs_count = 3; /* at most need 3 CS*/
+			/* one for SS and one for budgeted transaction */
+			sch_ep->num_budget_microframes = sch_ep->cs_count + 2;
+			sch_ep->bw_cost_per_microframe = max_packet_size;
+		}
+		if (ep_type == ISOC_OUT_EP) {
+
+			/*
+			 * the best case FS budget assumes that 188 FS bytes
+			 * occur in each microframe
+			 */
+			sch_ep->num_budget_microframes = DIV_ROUND_UP(
+				max_packet_size, FS_PAYLOAD_MAX);
+			sch_ep->bw_cost_per_microframe = FS_PAYLOAD_MAX;
+			sch_ep->cs_count = sch_ep->num_budget_microframes;
+		}
+		if (ep_type == ISOC_IN_EP) {
+			/* at most need additional two CS. */
+			sch_ep->cs_count = DIV_ROUND_UP(
+				max_packet_size, FS_PAYLOAD_MAX) + 2;
+			sch_ep->num_budget_microframes = sch_ep->cs_count + 2;
+			sch_ep->bw_cost_per_microframe = FS_PAYLOAD_MAX;
+		}
+	}
+}
+
+/* Get maximum bandwidth when we schedule at offset slot. */
+static u32 get_max_bw(struct mu3h_sch_bw_info *sch_bw,
+	struct mu3h_sch_ep_info *sch_ep, u32 offset)
+{
+	u32 num_esit;
+	u32 max_bw = 0;
+	int i;
+	int j;
+
+	num_esit = XHCI_MTK_MAX_ESIT / sch_ep->esit;
+	for (i = 0; i < num_esit; i++) {
+		u32 base = offset + i * sch_ep->esit;
+
+		for (j = 0; j < sch_ep->num_budget_microframes; j++) {
+			if (sch_bw->bus_bw[base + j] > max_bw)
+				max_bw = sch_bw->bus_bw[base + j];
+		}
+	}
+	return max_bw;
+}
+
+static void update_bus_bw(struct mu3h_sch_bw_info *sch_bw,
+	struct mu3h_sch_ep_info *sch_ep, int bw_cost)
+{
+	u32 num_esit;
+	u32 base;
+	int i;
+	int j;
+
+	num_esit = XHCI_MTK_MAX_ESIT / sch_ep->esit;
+	for (i = 0; i < num_esit; i++) {
+		base = sch_ep->offset + i * sch_ep->esit;
+		for (j = 0; j < sch_ep->num_budget_microframes; j++)
+			sch_bw->bus_bw[base + j] += bw_cost;
+	}
+}
+
+static int check_sch_bw(struct usb_device *udev,
+	struct mu3h_sch_bw_info *sch_bw, struct mu3h_sch_ep_info *sch_ep)
+{
+	u32 offset;
+	u32 esit;
+	u32 num_budget_microframes;
+	u32 min_bw;
+	u32 min_index;
+	u32 worst_bw;
+	u32 bw_boundary;
+
+	if (sch_ep->esit > XHCI_MTK_MAX_ESIT)
+		sch_ep->esit = XHCI_MTK_MAX_ESIT;
+
+	esit = sch_ep->esit;
+	num_budget_microframes = sch_ep->num_budget_microframes;
+
+	/*
+	 * Search through all possible schedule microframes.
+	 * and find a microframe where its worst bandwidth is minimum.
+	 */
+	min_bw = ~0;
+	min_index = 0;
+	for (offset = 0; offset < esit; offset++) {
+		if ((offset + num_budget_microframes) > sch_ep->esit)
+			break;
+
+		/*
+		 * usb_20 spec section11.18:
+		 * must never schedule Start-Split in Y6
+		 */
+		if (is_fs_or_ls(udev->speed) && (offset % 8 == 6))
+			continue;
+
+		worst_bw = get_max_bw(sch_bw, sch_ep, offset);
+		if (min_bw > worst_bw) {
+			min_bw = worst_bw;
+			min_index = offset;
+		}
+		if (min_bw == 0)
+			break;
+	}
+	sch_ep->offset = min_index;
+
+	bw_boundary = (udev->speed == USB_SPEED_SUPER)
+				? SS_BW_BOUNDARY : HS_BW_BOUNDARY;
+
+	/* check bandwidth */
+	if (min_bw + sch_ep->bw_cost_per_microframe > bw_boundary)
+		return -ERANGE;
+
+	/* update bus bandwidth info */
+	update_bus_bw(sch_bw, sch_ep, sch_ep->bw_cost_per_microframe);
+
+	return 0;
+}
+
+static bool need_bw_sch(struct usb_host_endpoint *ep,
+	enum usb_device_speed speed, int has_tt)
+{
+	/* only for periodic endpoints */
+	if (usb_endpoint_xfer_control(&ep->desc)
+		|| usb_endpoint_xfer_bulk(&ep->desc))
+		return false;
+
+	/*
+	 * for LS & FS periodic endpoints which its device don't attach
+	 * to TT are also ignored, root-hub will schedule them directly
+	 */
+	if (is_fs_or_ls(speed) && !has_tt)
+		return false;
+
+	return true;
+}
+
+int xhci_mtk_sch_init(struct xhci_hcd *xhci)
+{
+	struct usb_hcd *hcd = xhci_to_hcd(xhci);
+	struct device *dev = hcd->self.controller;
+	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
+	struct mu3h_sch_bw_info *sch_array;
+	int num_usb_bus;
+	int i;
+
+	/* ss IN and OUT are separated */
+	num_usb_bus = xhci->num_usb3_ports * 2 + xhci->num_usb2_ports;
+
+	sch_array = kcalloc(num_usb_bus, sizeof(*sch_array), GFP_KERNEL);
+	if (sch_array == NULL)
+		return -ENOMEM;
+
+	for (i = 0; i < num_usb_bus; i++)
+		INIT_LIST_HEAD(&sch_array[i].bw_ep_list);
+
+	mtk->sch_array = sch_array;
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xhci_mtk_sch_init);
+
+void xhci_mtk_sch_exit(struct xhci_hcd *xhci)
+{
+	struct usb_hcd *hcd = xhci_to_hcd(xhci);
+	struct device *dev = hcd->self.controller;
+	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
+
+	kfree(mtk->sch_array);
+}
+EXPORT_SYMBOL_GPL(xhci_mtk_sch_exit);
+
+int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
+		struct usb_host_endpoint *ep)
+{
+	struct device *dev = hcd->self.controller;
+	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
+	struct xhci_hcd *xhci;
+	struct xhci_ep_ctx *ep_ctx;
+	struct xhci_slot_ctx *slot_ctx;
+	struct xhci_virt_device *virt_dev;
+	struct mu3h_sch_bw_info *sch_bw;
+	struct mu3h_sch_ep_info *sch_ep;
+	struct mu3h_sch_bw_info *sch_array;
+	unsigned int ep_index;
+	int bw_index;
+	int ret = 0;
+
+	xhci = hcd_to_xhci(hcd);
+	virt_dev = xhci->devs[udev->slot_id];
+	ep_index = xhci_get_endpoint_index(&ep->desc);
+	slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx);
+	ep_ctx = xhci_get_ep_ctx(xhci, virt_dev->in_ctx, ep_index);
+	sch_array = mtk->sch_array;
+
+	xhci_dbg(xhci, "%s() type:%d, speed:%d, mpkt:%d, dir:%d, ep:%p\n",
+		__func__, usb_endpoint_type(&ep->desc), udev->speed,
+		GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc)),
+		usb_endpoint_dir_in(&ep->desc), ep);
+
+	if (!need_bw_sch(ep, udev->speed, slot_ctx->tt_info & TT_SLOT))
+		return 0;
+
+	bw_index = get_bw_index(xhci, udev, ep);
+	sch_bw = &sch_array[bw_index];
+
+	sch_ep = kzalloc(sizeof(struct mu3h_sch_ep_info), GFP_NOIO);
+	if (!sch_ep)
+		return -ENOMEM;
+
+	setup_sch_info(udev, ep_ctx, sch_ep);
+
+	ret = check_sch_bw(udev, sch_bw, sch_ep);
+	if (ret) {
+		xhci_err(xhci, "Not enough bandwidth!\n");
+		kfree(sch_ep);
+		return -ENOSPC;
+	}
+
+	list_add_tail(&sch_ep->endpoint, &sch_bw->bw_ep_list);
+	sch_ep->ep = ep;
+
+	ep_ctx->reserved[0] |= cpu_to_le32(EP_BPKTS(sch_ep->pkts)
+		| EP_BCSCOUNT(sch_ep->cs_count) | EP_BBM(sch_ep->burst_mode));
+	ep_ctx->reserved[1] |= cpu_to_le32(EP_BOFFSET(sch_ep->offset)
+		| EP_BREPEAT(sch_ep->repeat));
+
+	xhci_dbg(xhci, " PKTS:%x, CSCOUNT:%x, BM:%x, OFFSET:%x, REPEAT:%x\n",
+			sch_ep->pkts, sch_ep->cs_count, sch_ep->burst_mode,
+			sch_ep->offset, sch_ep->repeat);
+
+	return 0;
+}
+EXPORT_SYMBOL_GPL(xhci_mtk_add_ep_quirk);
+
+void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
+		struct usb_host_endpoint *ep)
+{
+	struct device *dev = hcd->self.controller;
+	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
+	struct xhci_hcd *xhci;
+	struct xhci_slot_ctx *slot_ctx;
+	struct xhci_virt_device *virt_dev;
+	struct mu3h_sch_bw_info *sch_array;
+	struct mu3h_sch_bw_info *sch_bw;
+	struct mu3h_sch_ep_info *sch_ep;
+	int bw_index;
+
+	xhci = hcd_to_xhci(hcd);
+	virt_dev = xhci->devs[udev->slot_id];
+	slot_ctx = xhci_get_slot_ctx(xhci, virt_dev->in_ctx);
+	sch_array = mtk->sch_array;
+
+	xhci_dbg(xhci, "%s() type:%d, speed:%d, mpks:%d, dir:%d, ep:%p\n",
+		__func__, usb_endpoint_type(&ep->desc), udev->speed,
+		GET_MAX_PACKET(usb_endpoint_maxp(&ep->desc)),
+		usb_endpoint_dir_in(&ep->desc), ep);
+
+	if (!need_bw_sch(ep, udev->speed, slot_ctx->tt_info & TT_SLOT))
+		return;
+
+	bw_index = get_bw_index(xhci, udev, ep);
+	sch_bw = &sch_array[bw_index];
+
+	list_for_each_entry(sch_ep, &sch_bw->bw_ep_list, endpoint) {
+		if (sch_ep->ep == ep) {
+			update_bus_bw(sch_bw, sch_ep,
+				-sch_ep->bw_cost_per_microframe);
+			list_del(&sch_ep->endpoint);
+			kfree(sch_ep);
+			break;
+		}
+	}
+}
+EXPORT_SYMBOL_GPL(xhci_mtk_drop_ep_quirk);
diff --git a/drivers/usb/host/xhci-mtk.c b/drivers/usb/host/xhci-mtk.c
new file mode 100644
index 0000000..afb52bf
--- /dev/null
+++ b/drivers/usb/host/xhci-mtk.c
@@ -0,0 +1,776 @@ 
+/*
+ * Copyright (c) 2015 MediaTek Inc.
+ * Author:
+ *  Chunfeng Yun <chunfeng.yun@mediatek.com>
+ *
+ * This software is licensed under the terms of the GNU General Public
+ * License version 2, as published by the Free Software Foundation, and
+ * may be copied, distributed, and modified under those terms.
+ *
+ * This program is distributed in the hope that 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.
+ *
+ */
+
+#include <linux/clk.h>
+#include <linux/dma-mapping.h>
+#include <linux/iopoll.h>
+#include <linux/kernel.h>
+#include <linux/mfd/syscon.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/phy/phy.h>
+#include <linux/platform_device.h>
+#include <linux/pm_runtime.h>
+#include <linux/regmap.h>
+#include <linux/regulator/consumer.h>
+#include <linux/slab.h>
+#include <linux/usb/phy.h>
+#include <linux/usb/xhci_pdriver.h>
+
+#include "xhci.h"
+#include "xhci-mtk.h"
+
+/* ip_pw_ctrl0 register */
+#define CTRL0_IP_SW_RST	BIT(0)
+
+/* ip_pw_ctrl1 register */
+#define CTRL1_IP_HOST_PDN	BIT(0)
+
+/* ip_pw_ctrl2 register */
+#define CTRL2_IP_DEV_PDN	BIT(0)
+
+/* ip_pw_sts1 register */
+#define STS1_IP_SLEEP_STS	BIT(30)
+#define STS1_U3_MAC_RST	BIT(16)
+#define STS1_SYS125_RST	BIT(10)
+#define STS1_REF_RST		BIT(8)
+#define STS1_SYSPLL_STABLE	BIT(0)
+
+/* ip_pw_sts2 register */
+#define STS2_U2_MAC_RST	BIT(0)
+
+/* ip_xhci_cap register */
+#define CAP_U3_PORT_NUM(p)	((p) & 0xff)
+#define CAP_U2_PORT_NUM(p)	(((p) >> 8) & 0xff)
+
+/* u3_ctrl_p register */
+#define CTRL_U3_PORT_HOST_SEL	BIT(2)
+#define CTRL_U3_PORT_PDN	BIT(1)
+#define CTRL_U3_PORT_DIS	BIT(0)
+
+/* u2_ctrl_p register */
+#define CTRL_U2_PORT_HOST_SEL	BIT(2)
+#define CTRL_U2_PORT_PDN	BIT(1)
+#define CTRL_U2_PORT_DIS	BIT(0)
+
+/* u2_phy_pll register */
+#define CTRL_U2_FORCE_PLL_STB	BIT(28)
+
+#define PERI_WK_CTRL0		0x400
+#define UWK_CTR0_0P_LS_PE	BIT(8)  /* posedge */
+#define UWK_CTR0_0P_LS_NE	BIT(7)  /* negedge for 0p linestate*/
+#define UWK_CTL1_1P_LS_C(x)	(((x) & 0xf) << 1)
+#define UWK_CTL1_1P_LS_E	BIT(0)
+
+#define PERI_WK_CTRL1		0x404
+#define UWK_CTL1_IS_C(x)	(((x) & 0xf) << 26)
+#define UWK_CTL1_IS_E		BIT(25)
+#define UWK_CTL1_0P_LS_C(x)	(((x) & 0xf) << 21)
+#define UWK_CTL1_0P_LS_E	BIT(20)
+#define UWK_CTL1_IDDIG_C(x)	(((x) & 0xf) << 11)  /* cycle debounce */
+#define UWK_CTL1_IDDIG_E	BIT(10) /* enable debounce */
+#define UWK_CTL1_IDDIG_P	BIT(9)  /* polarity */
+#define UWK_CTL1_0P_LS_P	BIT(7)
+#define UWK_CTL1_IS_P		BIT(6)  /* polarity for ip sleep */
+
+enum ssusb_wakeup_src {
+	SSUSB_WK_IP_SLEEP = 1,
+	SSUSB_WK_LINE_STATE = 2,
+};
+
+static int check_ip_clk_status(struct xhci_hcd_mtk *mtk)
+{
+	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
+	u32 val, check_val;
+	int ret;
+
+	check_val = STS1_SYSPLL_STABLE | STS1_REF_RST | STS1_SYS125_RST;
+	check_val |= (mtk->num_u3_ports ? STS1_U3_MAC_RST : 0);
+
+	ret = readl_poll_timeout(&ippc->ip_pw_sts1, val,
+			  (check_val == (val & check_val)), 100, 20000);
+	if (ret) {
+		dev_err(mtk->dev, "mac3 clocks are not stable (0x%x)\n", val);
+		return ret;
+	}
+
+	ret = readl_poll_timeout(&ippc->ip_pw_sts2, val,
+			   (val & STS2_U2_MAC_RST), 100, 10000);
+	if (ret) {
+		dev_err(mtk->dev, "mac2 reset is still active!!!\n");
+		return ret;
+	}
+
+	return 0;
+}
+
+static int xhci_mtk_ports_enable(struct xhci_hcd_mtk *mtk)
+{
+	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
+	u32 temp;
+	int i;
+
+	/* power on host ip */
+	temp = readl(&ippc->ip_pw_ctr1);
+	temp &= ~CTRL1_IP_HOST_PDN;
+	writel(temp, &ippc->ip_pw_ctr1);
+
+	/* power on and enable all u3 ports */
+	for (i = 0; i < mtk->num_u3_ports; i++) {
+		temp = readl(&ippc->u3_ctrl_p[i]);
+		temp &= ~(CTRL_U3_PORT_PDN | CTRL_U3_PORT_DIS);
+		temp |= CTRL_U3_PORT_HOST_SEL;
+		writel(temp, &ippc->u3_ctrl_p[i]);
+	}
+
+	/* power on and enable all u2 ports */
+	for (i = 0; i < mtk->num_u2_ports; i++) {
+		temp = readl(&ippc->u2_ctrl_p[i]);
+		temp &= ~(CTRL_U2_PORT_PDN | CTRL_U2_PORT_DIS);
+		temp |= CTRL_U2_PORT_HOST_SEL;
+		writel(temp, &ippc->u2_ctrl_p[i]);
+	}
+
+	/*
+	 * wait for clocks to be stable, and clock domains reset to
+	 * be inactive after power on and enable ports
+	 */
+	return check_ip_clk_status(mtk);
+}
+
+static int xhci_mtk_ports_disable(struct xhci_hcd_mtk *mtk)
+{
+	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
+	u32 temp;
+	int ret;
+	int i;
+
+	/* power down all u3 ports */
+	for (i = 0; i < mtk->num_u3_ports; i++) {
+		temp = readl(&ippc->u3_ctrl_p[i]);
+		temp |= CTRL_U3_PORT_PDN;
+		writel(temp, &ippc->u3_ctrl_p[i]);
+	}
+
+	/* power down all u2 ports */
+	for (i = 0; i < mtk->num_u2_ports; i++) {
+		temp = readl(&ippc->u2_ctrl_p[i]);
+		temp |= CTRL_U2_PORT_PDN;
+		writel(temp, &ippc->u2_ctrl_p[i]);
+	}
+
+	/* power down host ip */
+	temp = readl(&ippc->ip_pw_ctr1);
+	temp |= CTRL1_IP_HOST_PDN;
+	writel(temp, &ippc->ip_pw_ctr1);
+
+	ret = readl_poll_timeout(&ippc->ip_pw_sts1, temp,
+			  (temp & STS1_IP_SLEEP_STS), 100, 100000);
+	if (ret) {
+		dev_err(mtk->dev, "ip sleep failed!!!\n");
+		return ret;
+	}
+	return 0;
+}
+
+static void xhci_mtk_ports_config(struct xhci_hcd_mtk *mtk)
+{
+	struct mu3c_ippc_regs __iomem *ippc = mtk->ippc_regs;
+	u32 temp;
+
+	/* reset whole ip */
+	temp = readl(&ippc->ip_pw_ctr0);
+	temp |= CTRL0_IP_SW_RST;
+	writel(temp, &ippc->ip_pw_ctr0);
+	udelay(1);
+	temp = readl(&ippc->ip_pw_ctr0);
+	temp &= ~CTRL0_IP_SW_RST;
+	writel(temp, &ippc->ip_pw_ctr0);
+
+	/*
+	 * device ip is default power-on in fact
+	 * power down device ip, otherwise ip-sleep will fail
+	 */
+	temp = readl(&ippc->ip_pw_ctr2);
+	temp |= CTRL2_IP_DEV_PDN;
+	writel(temp, &ippc->ip_pw_ctr2);
+
+	temp = readl(&ippc->ip_xhci_cap);
+	mtk->num_u3_ports = CAP_U3_PORT_NUM(temp);
+	mtk->num_u2_ports = CAP_U2_PORT_NUM(temp);
+	dev_dbg(mtk->dev, "%s u2p:%d, u3p:%d\n", __func__,
+			mtk->num_u2_ports, mtk->num_u3_ports);
+
+	xhci_mtk_ports_enable(mtk);
+}
+
+static int xhci_mtk_clks_enable(struct xhci_hcd_mtk *mtk)
+{
+	int ret;
+
+	ret = clk_prepare_enable(mtk->sys_clk);
+	if (ret) {
+		dev_err(mtk->dev, "failed to enable u3phya_ref\n");
+		goto u3phya_ref_err;
+	}
+	ret = clk_prepare_enable(mtk->wk_deb_p0);
+	if (ret) {
+		dev_err(mtk->dev, "failed to enable wk_deb_p0\n");
+		goto usb_p0_err;
+	}
+	if (mtk->num_phys > 1) {
+		ret = clk_prepare_enable(mtk->wk_deb_p1);
+		if (ret) {
+			dev_err(mtk->dev, "failed to enable wk_deb_p1\n");
+			goto usb_p1_err;
+		}
+	}
+	return 0;
+
+usb_p1_err:
+	clk_disable_unprepare(mtk->wk_deb_p0);
+usb_p0_err:
+	clk_disable_unprepare(mtk->sys_clk);
+u3phya_ref_err:
+	return -EINVAL;
+}
+
+static void xhci_mtk_clks_disable(struct xhci_hcd_mtk *mtk)
+{
+	if (mtk->num_phys > 1)
+		clk_disable_unprepare(mtk->wk_deb_p1);
+	clk_disable_unprepare(mtk->wk_deb_p0);
+	clk_disable_unprepare(mtk->sys_clk);
+}
+
+/* only clocks can be turn off for ip-sleep wakeup mode */
+static void usb_wakeup_ip_sleep_en(struct xhci_hcd_mtk *mtk)
+{
+	u32 tmp;
+	struct regmap *pericfg = mtk->pericfg;
+
+	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
+	tmp &= ~UWK_CTL1_IS_P;
+	tmp &= ~(UWK_CTL1_IS_C(0xf));
+	tmp |= UWK_CTL1_IS_C(0x8);
+	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
+	regmap_write(pericfg, PERI_WK_CTRL1, tmp | UWK_CTL1_IS_E);
+
+	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
+	dev_dbg(mtk->dev, "%s(): WK_CTRL1[P6,E25,C26:29]=%#x\n",
+		__func__, tmp);
+}
+
+static void usb_wakeup_ip_sleep_dis(struct xhci_hcd_mtk *mtk)
+{
+	u32 tmp;
+
+	regmap_read(mtk->pericfg, PERI_WK_CTRL1, &tmp);
+	tmp &= ~UWK_CTL1_IS_E;
+	regmap_write(mtk->pericfg, PERI_WK_CTRL1, tmp);
+}
+
+/*
+* for line-state wakeup mode, phy's power should not power-down
+* and only support cable plug in/out
+*/
+static void usb_wakeup_line_state_en(struct xhci_hcd_mtk *mtk)
+{
+	u32 tmp;
+	struct regmap *pericfg = mtk->pericfg;
+
+	/* line-state of u2-port0 */
+	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
+	tmp &= ~UWK_CTL1_0P_LS_P;
+	tmp &= ~(UWK_CTL1_0P_LS_C(0xf));
+	tmp |= UWK_CTL1_0P_LS_C(0x8);
+	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
+	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
+	regmap_write(pericfg, PERI_WK_CTRL1, tmp | UWK_CTL1_0P_LS_E);
+
+	/* line-state of u2-port1 if support */
+	if (mtk->num_phys > 1) {
+		regmap_read(pericfg, PERI_WK_CTRL0, &tmp);
+		tmp &= ~(UWK_CTL1_1P_LS_C(0xf));
+		tmp |= UWK_CTL1_1P_LS_C(0x8);
+		regmap_write(pericfg, PERI_WK_CTRL0, tmp);
+		regmap_write(pericfg, PERI_WK_CTRL0, tmp | UWK_CTL1_1P_LS_E);
+	}
+}
+
+static void usb_wakeup_line_state_dis(struct xhci_hcd_mtk *mtk)
+{
+	u32 tmp;
+	struct regmap *pericfg = mtk->pericfg;
+
+	regmap_read(pericfg, PERI_WK_CTRL1, &tmp);
+	tmp &= ~UWK_CTL1_0P_LS_E;
+	regmap_write(pericfg, PERI_WK_CTRL1, tmp);
+
+	if (mtk->num_phys > 1) {
+		regmap_read(pericfg, PERI_WK_CTRL0, &tmp);
+		tmp &= ~UWK_CTL1_1P_LS_E;
+		regmap_write(pericfg, PERI_WK_CTRL0, tmp);
+	}
+}
+
+static void usb_wakeup_enable(struct xhci_hcd_mtk *mtk)
+{
+	if (mtk->wakeup_src == SSUSB_WK_IP_SLEEP)
+		usb_wakeup_ip_sleep_en(mtk);
+	else if (mtk->wakeup_src == SSUSB_WK_LINE_STATE)
+		usb_wakeup_line_state_en(mtk);
+}
+
+static void usb_wakeup_disable(struct xhci_hcd_mtk *mtk)
+{
+	if (mtk->wakeup_src == SSUSB_WK_IP_SLEEP)
+		usb_wakeup_ip_sleep_dis(mtk);
+	else if (mtk->wakeup_src == SSUSB_WK_LINE_STATE)
+		usb_wakeup_line_state_dis(mtk);
+}
+
+static int usb_wakeup_of_property_parse(struct xhci_hcd_mtk *mtk,
+				struct device_node *dn)
+{
+	struct device *dev = mtk->dev;
+
+	/*
+	* wakeup function is optional, so it is not an error if this property
+	* does not exist, and in such case, no need to get relative
+	* properties anymore.
+	*/
+	of_property_read_u32(dn, "mediatek,wakeup-src", &mtk->wakeup_src);
+	if (!mtk->wakeup_src)
+		return 0;
+
+	mtk->wk_deb_p0 = devm_clk_get(dev, "wakeup_deb_p0");
+	if (IS_ERR(mtk->wk_deb_p0)) {
+		dev_err(dev, "fail to get wakeup_deb_p0\n");
+		return PTR_ERR(mtk->wk_deb_p0);
+	}
+
+	mtk->wk_deb_p1 = devm_clk_get(dev, "wakeup_deb_p1");
+	if (IS_ERR(mtk->wk_deb_p1)) {
+		dev_err(dev, "fail to get wakeup_deb_p1\n");
+		return PTR_ERR(mtk->wk_deb_p1);
+	}
+
+	mtk->pericfg = syscon_regmap_lookup_by_phandle(dn,
+						"mediatek,syscon-wakeup");
+	if (IS_ERR(mtk->pericfg)) {
+		dev_err(dev, "fail to get pericfg regs\n");
+		return PTR_ERR(mtk->pericfg);
+	}
+
+	return 0;
+}
+
+static int xhci_mtk_setup(struct usb_hcd *hcd);
+static const struct xhci_driver_overrides xhci_mtk_overrides __initconst = {
+	.extra_priv_size = sizeof(struct xhci_hcd),
+	.reset = xhci_mtk_setup,
+};
+
+static struct hc_driver __read_mostly xhci_mtk_hc_driver;
+
+static int xhci_mtk_phy_init(struct xhci_hcd_mtk *mtk)
+{
+	int i;
+	int ret;
+
+	for (i = 0; i < mtk->num_phys; i++) {
+		ret = phy_init(mtk->phys[i]);
+		if (ret)
+			goto exit_phy;
+	}
+	return 0;
+
+exit_phy:
+	for (; i > 0; i--)
+		phy_exit(mtk->phys[i - 1]);
+
+	return ret;
+}
+
+static int xhci_mtk_phy_exit(struct xhci_hcd_mtk *mtk)
+{
+	int i;
+
+	for (i = 0; i < mtk->num_phys; i++)
+		phy_exit(mtk->phys[i]);
+
+	return 0;
+}
+
+static int xhci_mtk_phy_power_on(struct xhci_hcd_mtk *mtk)
+{
+	int i;
+	int ret;
+
+	for (i = 0; i < mtk->num_phys; i++) {
+		ret = phy_power_on(mtk->phys[i]);
+		if (ret)
+			goto power_off_phy;
+	}
+	return 0;
+
+power_off_phy:
+	for (; i > 0; i--)
+		phy_power_off(mtk->phys[i - 1]);
+
+	return ret;
+}
+
+static void xhci_mtk_phy_power_off(struct xhci_hcd_mtk *mtk)
+{
+	unsigned int i;
+
+	for (i = 0; i < mtk->num_phys; i++)
+		phy_power_off(mtk->phys[i]);
+}
+
+static int xhci_mtk_ldos_enable(struct xhci_hcd_mtk *mtk)
+{
+	int ret;
+
+	ret = regulator_enable(mtk->vbus);
+	if (ret) {
+		dev_err(mtk->dev, "failed to enable vbus\n");
+		return ret;
+	}
+
+	ret = regulator_enable(mtk->vusb33);
+	if (ret) {
+		dev_err(mtk->dev, "failed to enable vusb33\n");
+		regulator_disable(mtk->vbus);
+		return ret;
+	}
+	return 0;
+}
+
+static void xhci_mtk_ldos_disable(struct xhci_hcd_mtk *mtk)
+{
+	regulator_disable(mtk->vbus);
+	regulator_disable(mtk->vusb33);
+}
+
+static void xhci_mtk_quirks(struct device *dev, struct xhci_hcd *xhci)
+{
+	/*
+	 * As of now platform drivers don't provide MSI support so we ensure
+	 * here that the generic code does not try to make a pci_dev from our
+	 * dev struct in order to setup MSI
+	 */
+	xhci->quirks |= XHCI_PLAT;
+	xhci->quirks |= XHCI_MTK_HOST;
+	/*
+	 * MTK host controller gives a spurious successful event after a
+	 * short transfer. Ignore it.
+	 */
+	xhci->quirks |= XHCI_SPURIOUS_SUCCESS;
+}
+
+/* called during probe() after chip reset completes */
+static int xhci_mtk_setup(struct usb_hcd *hcd)
+{
+	struct xhci_hcd *xhci;
+	int ret;
+
+	ret = xhci_gen_setup(hcd, xhci_mtk_quirks);
+	if (ret)
+		return ret;
+
+	if (!usb_hcd_is_primary_hcd(hcd))
+		return 0;
+
+	xhci = hcd_to_xhci(hcd);
+	return xhci_mtk_sch_init(xhci);
+}
+
+
+static int xhci_mtk_probe(struct platform_device *pdev)
+{
+	struct device *dev = &pdev->dev;
+	struct device_node *node = dev->of_node;
+	struct usb_xhci_pdata *pdata = dev_get_platdata(dev);
+	struct xhci_hcd_mtk *mtk;
+	const struct hc_driver *driver;
+	struct xhci_hcd *xhci;
+	struct resource *res;
+	struct usb_hcd *hcd;
+	struct phy *phy;
+	int phy_num;
+	int ret = -ENODEV;
+	int irq;
+
+	if (usb_disabled())
+		return -ENODEV;
+
+	driver = &xhci_mtk_hc_driver;
+	mtk = devm_kzalloc(dev, sizeof(*mtk), GFP_KERNEL);
+	if (!mtk)
+		return -ENOMEM;
+
+	mtk->dev = dev;
+	mtk->vbus = devm_regulator_get(dev, "vbus");
+	if (IS_ERR(mtk->vbus)) {
+		dev_err(dev, "fail to get vbus\n");
+		return PTR_ERR(mtk->vbus);
+	}
+
+	mtk->vusb33 = devm_regulator_get(dev, "vusb33");
+	if (IS_ERR(mtk->vusb33)) {
+		dev_err(dev, "fail to get vusb33\n");
+		return PTR_ERR(mtk->vusb33);
+	}
+
+	mtk->sys_clk = devm_clk_get(dev, "sys_ck");
+	if (IS_ERR(mtk->sys_clk)) {
+		dev_err(dev, "fail to get sys_ck\n");
+		return PTR_ERR(mtk->sys_clk);
+	}
+
+	ret = usb_wakeup_of_property_parse(mtk, node);
+	if (ret)
+		return ret;
+
+	mtk->num_phys = of_count_phandle_with_args(node,
+			"phys", "#phy-cells");
+	if (mtk->num_phys > 0) {
+		mtk->phys = devm_kcalloc(dev, mtk->num_phys,
+					sizeof(*mtk->phys), GFP_KERNEL);
+		if (!mtk->phys)
+			return -ENOMEM;
+	} else {
+		mtk->num_phys = 0;
+	}
+	pm_runtime_enable(dev);
+	pm_runtime_get_sync(dev);
+
+	ret = xhci_mtk_ldos_enable(mtk);
+	if (ret)
+		goto disable_pm;
+
+	ret = xhci_mtk_clks_enable(mtk);
+	if (ret)
+		goto disable_ldos;
+
+	irq = platform_get_irq(pdev, 0);
+	if (irq < 0)
+		goto disable_clk;
+
+	/* Initialize dma_mask and coherent_dma_mask to 32-bits */
+	ret = dma_set_coherent_mask(dev, DMA_BIT_MASK(32));
+	if (ret)
+		goto disable_clk;
+
+	if (!dev->dma_mask)
+		dev->dma_mask = &dev->coherent_dma_mask;
+	else
+		dma_set_mask(dev, DMA_BIT_MASK(32));
+
+	hcd = usb_create_hcd(driver, dev, dev_name(dev));
+	if (!hcd) {
+		ret = -ENOMEM;
+		goto disable_clk;
+	}
+
+	/*
+	 * USB 2.0 roothub is stored in the platform_device.
+	 * Swap it with mtk HCD.
+	 */
+	mtk->hcd = platform_get_drvdata(pdev);
+	platform_set_drvdata(pdev, mtk);
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
+	hcd->regs = devm_ioremap_resource(dev, res);
+	if (IS_ERR(hcd->regs)) {
+		ret = PTR_ERR(hcd->regs);
+		goto put_usb2_hcd;
+	}
+	hcd->rsrc_start = res->start;
+	hcd->rsrc_len = resource_size(res);
+
+	res = platform_get_resource(pdev, IORESOURCE_MEM, 1);
+	mtk->ippc_regs = devm_ioremap_resource(dev, res);
+	if (IS_ERR(mtk->ippc_regs)) {
+		ret = PTR_ERR(mtk->ippc_regs);
+		goto put_usb2_hcd;
+	}
+
+	for (phy_num = 0; phy_num < mtk->num_phys; phy_num++) {
+		phy = devm_of_phy_get_by_index(dev, node, phy_num);
+		if (IS_ERR(phy)) {
+			ret = PTR_ERR(phy);
+			goto put_usb2_hcd;
+		}
+		mtk->phys[phy_num] = phy;
+	}
+
+	xhci_mtk_ports_config(mtk);
+	ret = xhci_mtk_phy_init(mtk);
+	if (ret)
+		goto put_usb2_hcd;
+
+	ret = xhci_mtk_phy_power_on(mtk);
+	if (ret)
+		goto exit_phys;
+
+	device_init_wakeup(dev, 1);
+
+	xhci = hcd_to_xhci(hcd);
+	xhci->main_hcd = hcd;
+	xhci->shared_hcd = usb_create_shared_hcd(driver, dev,
+			dev_name(dev), hcd);
+	if (!xhci->shared_hcd) {
+		ret = -ENOMEM;
+		goto power_off_phys;
+	}
+
+	if ((node && of_property_read_bool(node, "usb3-lpm-capable")) ||
+			(pdata && pdata->usb3_lpm_capable))
+		xhci->quirks |= XHCI_LPM_SUPPORT;
+
+	if (HCC_MAX_PSA(xhci->hcc_params) >= 4)
+		xhci->shared_hcd->can_do_streams = 1;
+
+	ret = usb_add_hcd(hcd, irq, IRQF_SHARED);
+	if (ret)
+		goto put_usb3_hcd;
+
+	ret = usb_add_hcd(xhci->shared_hcd, irq, IRQF_SHARED);
+	if (ret)
+		goto dealloc_usb2_hcd;
+
+	return 0;
+
+dealloc_usb2_hcd:
+	xhci_mtk_sch_exit(xhci);
+	usb_remove_hcd(hcd);
+
+put_usb3_hcd:
+	usb_put_hcd(xhci->shared_hcd);
+
+power_off_phys:
+	xhci_mtk_phy_power_off(mtk);
+	device_init_wakeup(dev, 0);
+
+exit_phys:
+	xhci_mtk_phy_exit(mtk);
+
+put_usb2_hcd:
+	usb_put_hcd(hcd);
+
+disable_clk:
+	xhci_mtk_clks_disable(mtk);
+
+disable_ldos:
+	xhci_mtk_ldos_disable(mtk);
+
+disable_pm:
+	pm_runtime_put_sync(dev);
+	pm_runtime_disable(dev);
+	return ret;
+}
+
+static int xhci_mtk_remove(struct platform_device *dev)
+{
+	struct xhci_hcd_mtk *mtk = platform_get_drvdata(dev);
+	struct usb_hcd	*hcd = mtk->hcd;
+	struct xhci_hcd	*xhci = hcd_to_xhci(hcd);
+
+	usb_remove_hcd(xhci->shared_hcd);
+	xhci_mtk_phy_power_off(mtk);
+	xhci_mtk_phy_exit(mtk);
+	device_init_wakeup(&dev->dev, 0);
+
+	usb_remove_hcd(hcd);
+	usb_put_hcd(xhci->shared_hcd);
+	usb_put_hcd(hcd);
+	xhci_mtk_sch_exit(xhci);
+	xhci_mtk_clks_disable(mtk);
+	xhci_mtk_ldos_disable(mtk);
+	pm_runtime_put_sync(&dev->dev);
+	pm_runtime_disable(&dev->dev);
+
+	return 0;
+}
+
+#ifdef CONFIG_PM_SLEEP
+static int xhci_mtk_suspend(struct device *dev)
+{
+	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
+
+	xhci_mtk_ports_disable(mtk);
+	xhci_mtk_phy_power_off(mtk);
+	xhci_mtk_clks_disable(mtk);
+	usb_wakeup_enable(mtk);
+	return 0;
+}
+
+static int xhci_mtk_resume(struct device *dev)
+{
+	struct xhci_hcd_mtk *mtk = dev_get_drvdata(dev);
+
+	usb_wakeup_disable(mtk);
+	xhci_mtk_clks_enable(mtk);
+	xhci_mtk_phy_power_on(mtk);
+	xhci_mtk_ports_enable(mtk);
+	return 0;
+}
+
+static const struct dev_pm_ops xhci_mtk_pm_ops = {
+	SET_SYSTEM_SLEEP_PM_OPS(xhci_mtk_suspend, xhci_mtk_resume)
+};
+#define DEV_PM_OPS	(&xhci_mtk_pm_ops)
+#else
+#define DEV_PM_OPS	NULL
+#endif /* CONFIG_PM */
+
+#ifdef CONFIG_OF
+static const struct of_device_id mtk_xhci_of_match[] = {
+	{ .compatible = "mediatek,mt8173-xhci"},
+	{ },
+};
+MODULE_DEVICE_TABLE(of, mtk_xhci_of_match);
+#endif
+
+static struct platform_driver mtk_xhci_driver = {
+	.probe	= xhci_mtk_probe,
+	.remove	= xhci_mtk_remove,
+	.driver	= {
+		.name = "xhci-mtk",
+		.pm = DEV_PM_OPS,
+		.of_match_table = of_match_ptr(mtk_xhci_of_match),
+	},
+};
+
+static int __init xhci_mtk_init(void)
+{
+	xhci_init_driver(&xhci_mtk_hc_driver, &xhci_mtk_overrides);
+	return platform_driver_register(&mtk_xhci_driver);
+}
+module_init(xhci_mtk_init);
+
+static void __exit xhci_mtk_exit(void)
+{
+	platform_driver_unregister(&mtk_xhci_driver);
+}
+module_exit(xhci_mtk_exit);
+
+MODULE_DESCRIPTION("MediaTek xHCI Host Controller Driver");
+MODULE_LICENSE("GPL v2");
diff --git a/drivers/usb/host/xhci-mtk.h b/drivers/usb/host/xhci-mtk.h
new file mode 100644
index 0000000..8913c2f
--- /dev/null
+++ b/drivers/usb/host/xhci-mtk.h
@@ -0,0 +1,156 @@ 
+/*
+ * Copyright (c) 2015 MediaTek Inc.
+ * Author:
+ *  Zhigang.Wei <zhigang.wei@mediatek.com>
+ *  Chunfeng.Yun <chunfeng.yun@mediatek.com>
+ *
+ * This software is licensed under the terms of the GNU General Public
+ * License version 2, as published by the Free Software Foundation, and
+ * may be copied, distributed, and modified under those terms.
+ *
+ * This program is distributed in the hope that 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.
+ *
+ */
+
+#ifndef _XHCI_MTK_H_
+#define _XHCI_MTK_H_
+
+#include "xhci.h"
+
+/**
+ * To simplify scheduler algorithm, set a upper limit for ESIT,
+ * if a synchromous ep's ESIT is larger than @XHCI_MTK_MAX_ESIT,
+ * round down to the limit value, that means allocating more
+ * bandwidth to it.
+ */
+#define XHCI_MTK_MAX_ESIT	64
+
+/**
+ * struct mu3h_sch_bw_info: schedule information for bandwidth domain
+ *
+ * @bus_bw: array to keep track of bandwidth already used at each uframes
+ * @bw_ep_list: eps in the bandwidth domain
+ *
+ * treat a HS root port as a bandwidth domain, but treat a SS root port as
+ * two bandwidth domains, one for IN eps and another for OUT eps.
+ */
+struct mu3h_sch_bw_info {
+	u32 bus_bw[XHCI_MTK_MAX_ESIT];
+	struct list_head bw_ep_list;
+};
+
+/**
+ * struct mu3h_sch_ep_info: schedule information for endpoint
+ *
+ * @esit: unit is 125us, equal to 2 << Interval field in ep-context
+ * @num_budget_microframes: number of continuous uframes
+ *		(@repeat==1) scheduled within the interval
+ * @bw_cost_per_microframe: bandwidth cost per microframe
+ * @endpoint: linked into bandwidth domain which it belongs to
+ * @ep: address of usb_host_endpoint struct
+ * @offset: which uframe of the interval that transfer should be
+ *		scheduled first time within the interval
+ * @repeat: the time gap between two uframes that transfers are
+ *		scheduled within a interval. in the simple algorithm, only
+ *		assign 0 or 1 to it; 0 means using only one uframe in a
+ *		interval, and 1 means using @num_budget_microframes
+ *		continuous uframes
+ * @pkts: number of packets to be transferred in the scheduled uframes
+ * @cs_count: number of CS that host will trigger
+ * @burst_mode: burst mode for scheduling. 0: normal burst mode,
+ *		distribute the bMaxBurst+1 packets for a single burst
+ *		according to @pkts and @repeat, repeate the burst multiple
+ *		times; 1: distribute the (bMaxBurst+1)*(Mult+1) packets
+ *		according to @pkts and @repeat. normal mode is used by
+ *		default
+ */
+struct mu3h_sch_ep_info {
+	u32 esit;
+	u32 num_budget_microframes;
+	u32 bw_cost_per_microframe;
+	struct list_head endpoint;
+	void *ep;
+	/*
+	 * mtk xHCI scheduling information put into reserved DWs
+	 * in ep context
+	 */
+	u32 offset;
+	u32 repeat;
+	u32 pkts;
+	u32 cs_count;
+	u32 burst_mode;
+};
+
+#define MU3C_U3_PORT_MAX 4
+#define MU3C_U2_PORT_MAX 5
+
+/**
+ * struct mu3c_ippc_regs: MTK ssusb ip port control registers
+ * @ip_pw_ctr0~3: ip power and clock control registers
+ * @ip_pw_sts1~2: ip power and clock status registers
+ * @ip_xhci_cap: ip xHCI capability register
+ * @u3_ctrl_p[x]: ip usb3 port x control register, only low 4bytes are used
+ * @u2_ctrl_p[x]: ip usb2 port x control register, only low 4bytes are used
+ * @u2_phy_pll: usb2 phy pll control register
+ */
+struct mu3c_ippc_regs {
+	__le32 ip_pw_ctr0;
+	__le32 ip_pw_ctr1;
+	__le32 ip_pw_ctr2;
+	__le32 ip_pw_ctr3;
+	__le32 ip_pw_sts1;
+	__le32 ip_pw_sts2;
+	__le32 reserved0[3];
+	__le32 ip_xhci_cap;
+	__le32 reserved1[2];
+	__le64 u3_ctrl_p[MU3C_U3_PORT_MAX];
+	__le64 u2_ctrl_p[MU3C_U2_PORT_MAX];
+	__le32 reserved2;
+	__le32 u2_phy_pll;
+	__le32 reserved3[33]; /* 0x80 ~ 0xff */
+};
+
+struct xhci_hcd_mtk {
+	struct device *dev;
+	struct usb_hcd *hcd;
+	struct mu3h_sch_bw_info *sch_array;
+	struct mu3c_ippc_regs __iomem *ippc_regs;
+	int num_u2_ports;
+	int num_u3_ports;
+	struct regulator *vusb33;
+	struct regulator *vbus;
+	struct clk *sys_clk;	/* sys and mac clock */
+	struct clk *wk_deb_p0;	/* port0's wakeup debounce clock */
+	struct clk *wk_deb_p1;
+	struct regmap *pericfg;
+	int wakeup_src;
+	struct phy **phys;
+	int num_phys;
+};
+
+#if IS_ENABLED(CONFIG_USB_XHCI_MTK)
+int xhci_mtk_sch_init(struct xhci_hcd *xhci);
+void xhci_mtk_sch_exit(struct xhci_hcd *xhci);
+int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
+		struct usb_host_endpoint *ep);
+void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd, struct usb_device *udev,
+		struct usb_host_endpoint *ep);
+
+#else
+static inline int xhci_mtk_add_ep_quirk(struct usb_hcd *hcd,
+	struct usb_device *udev, struct usb_host_endpoint *ep)
+{
+	return 0;
+}
+
+static inline void xhci_mtk_drop_ep_quirk(struct usb_hcd *hcd,
+	struct usb_device *udev, struct usb_host_endpoint *ep)
+{
+}
+
+#endif
+
+#endif		/* _XHCI_MTK_H_ */
diff --git a/drivers/usb/host/xhci-ring.c b/drivers/usb/host/xhci-ring.c
index 57f40a1..243f696 100644
--- a/drivers/usb/host/xhci-ring.c
+++ b/drivers/usb/host/xhci-ring.c
@@ -68,6 +68,7 @@ 
 #include <linux/slab.h>
 #include "xhci.h"
 #include "xhci-trace.h"
+#include "xhci-mtk.h"
 
 /*
  * Returns zero if the TRB isn't in this segment, otherwise it returns the DMA
@@ -3044,18 +3045,27 @@  static u32 xhci_td_remainder(struct xhci_hcd *xhci, int transferred,
 			      struct urb *urb, unsigned int num_trbs_left)
 {
 	u32 maxp, total_packet_count;
+	u32 skip_current_trb = 0;
 
-	if (xhci->hci_version < 0x100)
-		return ((td_total_len - transferred) >> 10);
-
-	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
-	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
+	if (xhci->hci_version < 0x100) {
+		/* for MTK xHCI, TD size doesn't include this TRB */
+		if (xhci->quirks & XHCI_MTK_HOST)
+			skip_current_trb = 1;
+		else
+			return ((td_total_len - transferred) >> 10);
+	}
 
 	/* One TRB with a zero-length data packet. */
 	if (num_trbs_left == 0 || (transferred == 0 && trb_buff_len == 0) ||
 	    trb_buff_len == td_total_len)
 		return 0;
 
+	if (skip_current_trb)
+		trb_buff_len = 0;
+
+	maxp = GET_MAX_PACKET(usb_endpoint_maxp(&urb->ep->desc));
+	total_packet_count = DIV_ROUND_UP(td_total_len, maxp);
+
 	/* Queueing functions don't count the current TRB into transferred */
 	return (total_packet_count - ((transferred + trb_buff_len) / maxp));
 }
@@ -3443,7 +3453,7 @@  int xhci_queue_ctrl_tx(struct xhci_hcd *xhci, gfp_t mem_flags,
 		field |= 0x1;
 
 	/* xHCI 1.0 6.4.1.2.1: Transfer Type field */
-	if (xhci->hci_version == 0x100) {
+	if ((xhci->hci_version == 0x100) || (xhci->quirks & XHCI_MTK_HOST)) {
 		if (urb->transfer_buffer_length > 0) {
 			if (setup->bRequestType & USB_DIR_IN)
 				field |= TRB_TX_TYPE(TRB_DATA_IN);
diff --git a/drivers/usb/host/xhci.c b/drivers/usb/host/xhci.c
index 6b0f4a4..4741188 100644
--- a/drivers/usb/host/xhci.c
+++ b/drivers/usb/host/xhci.c
@@ -31,6 +31,7 @@ 
 
 #include "xhci.h"
 #include "xhci-trace.h"
+#include "xhci-mtk.h"
 
 #define DRIVER_AUTHOR "Sarah Sharp"
 #define DRIVER_DESC "'eXtensible' Host Controller (xHC) Driver"
@@ -623,7 +624,11 @@  int xhci_run(struct usb_hcd *hcd)
 			"// Set the interrupt modulation register");
 	temp = readl(&xhci->ir_set->irq_control);
 	temp &= ~ER_IRQ_INTERVAL_MASK;
-	temp |= (u32) 160;
+	/*
+	 * the increment interval is 8 times as much as that defined
+	 * in xHCI spec on MTK's controller
+	 */
+	temp |= (u32) ((xhci->quirks & XHCI_MTK_HOST) ? 20 : 160);
 	writel(temp, &xhci->ir_set->irq_control);
 
 	/* Set the HCD state before we enable the irqs */
@@ -1693,6 +1698,9 @@  int xhci_drop_endpoint(struct usb_hcd *hcd, struct usb_device *udev,
 
 	xhci_endpoint_zero(xhci, xhci->devs[udev->slot_id], ep);
 
+	if (xhci->quirks & XHCI_MTK_HOST)
+		xhci_mtk_drop_ep_quirk(hcd, udev, ep);
+
 	xhci_dbg(xhci, "drop ep 0x%x, slot id %d, new drop flags = %#x, new add flags = %#x\n",
 			(unsigned int) ep->desc.bEndpointAddress,
 			udev->slot_id,
@@ -1788,6 +1796,15 @@  int xhci_add_endpoint(struct usb_hcd *hcd, struct usb_device *udev,
 		return -ENOMEM;
 	}
 
+	if (xhci->quirks & XHCI_MTK_HOST) {
+		ret = xhci_mtk_add_ep_quirk(hcd, udev, ep);
+		if (ret < 0) {
+			xhci_free_or_cache_endpoint_ring(xhci,
+				virt_dev, ep_index);
+			return ret;
+		}
+	}
+
 	ctrl_ctx->add_flags |= cpu_to_le32(added_ctxs);
 	new_add_flags = le32_to_cpu(ctrl_ctx->add_flags);
 
diff --git a/drivers/usb/host/xhci.h b/drivers/usb/host/xhci.h
index ea8b904..1536da0 100644
--- a/drivers/usb/host/xhci.h
+++ b/drivers/usb/host/xhci.h
@@ -1577,6 +1577,7 @@  struct xhci_hcd {
 /* For controllers with a broken beyond repair streams implementation */
 #define XHCI_BROKEN_STREAMS	(1 << 19)
 #define XHCI_PME_STUCK_QUIRK	(1 << 20)
+#define XHCI_MTK_HOST		(1 << 21)
 	unsigned int		num_active_eps;
 	unsigned int		limit_active_eps;
 	/* There are two roothubs to keep track of bus suspend info for */