Thinh Nguyen wrote:
> Hi,
>
> Michael Grzeschik wrote:
>> Hi!
>>
>> On Wed, Jun 24, 2020 at 07:26:42PM +0000, Thinh Nguyen wrote:
>>> Michael Grzeschik wrote:
>>>> From: Michael Olbrich <m.olbrich@xxxxxxxxxxxxxx>
>>>>
>>>> Currently __dwc3_gadget_start_isoc must be called very shortly after
>>>> XferNotReady. Otherwise the frame number is outdated and start transfer
>>>> will fail, even with several retries.
>>> Did you try with the recent update for isoc? (e.i., after 5
>>> START_TRANSFER failures, the driver will issue END_TRANSFER to retry on
>>> the next XferNotReady event)
>>>
>>> Let me know if you still run into issues with that.
>> Yes. Without my patch I still see the issue. Event with the retry
>> machanism. It is even worse. I even missed one additional patch,
>> I had on top this one. See my note below.
> Ok. Can you clarify what issue you're seeing?
>
>>>> DSTS provides the lower 14 bit of the frame number. Use it in
>>>> combination
>>>> with the frame number provided by XferNotReady to guess the current
>>>> frame
>>>> number. This will succeed unless more than one 14 rollover has happened
>>>> since XferNotReady.
>>>>
>>>> Start transfer might still fail if the frame number is increased
>>>> immediately after DSTS is read. So retries are still needed.
>>>> Don't drop the current request if this happens. This way it is not
>>>> lost and
>>>> can be used immediately to try again with the next frame number.
>>>>
>>>> With this change, __dwc3_gadget_start_isoc is still not successfully
>>>> in all
>>>> cases bit it increases the acceptable delay after XferNotReady
>>>> significantly.
>>>>
>>>> Signed-off-by: Michael Olbrich <m.olbrich@xxxxxxxxxxxxxx>
>>>> Signed-off-by: Michael Tretter <m.tretter@xxxxxxxxxxxxxx>
>>>> Signed-off-by: Michael Grzeschik <m.grzeschik@xxxxxxxxxxxxxx>
>>>>
>>>> ---
>>>> v1 -> v2: - removed last_frame_number from struct
>>>> - included rollover variable
>>>>
>>>> drivers/usb/dwc3/gadget.c | 37 +++++++++++++++++++++++++------------
>>>> 1 file changed, 25 insertions(+), 12 deletions(-)
>>>>
>>>> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
>>>> index 421a0f73110a40b..0962ddd7fbf6ae6 100644
>>>> --- a/drivers/usb/dwc3/gadget.c
>>>> +++ b/drivers/usb/dwc3/gadget.c
>>>> @@ -1276,7 +1276,7 @@ static void dwc3_prepare_trbs(struct dwc3_ep
>>>> *dep)
>>>>
>>>> static void dwc3_gadget_ep_cleanup_cancelled_requests(struct
>>>> dwc3_ep *dep);
>>>>
>>>> -static int __dwc3_gadget_kick_transfer(struct dwc3_ep *dep)
>>>> +static int __dwc3_gadget_kick_transfer(struct dwc3_ep *dep, bool
>>>> keep_req)
>>> Any reason to have keep_req? With the recent changes, if
>>> dwc3_send_gadget_ep_cmd() returns -EAGAIN, then the controller driver
>>> won't give back the request.
>> Yes, we don't need the keep_req. I tried this and it worked without.
>> I will remove the parameter in v3.
>>
>>>> {
>>>> struct dwc3_gadget_ep_cmd_params params;
>>>> struct dwc3_request *req;
>>>> @@ -1314,12 +1314,9 @@ static int __dwc3_gadget_kick_transfer(struct
>>>> dwc3_ep *dep)
>>>> }
>>>>
>>>> ret = dwc3_send_gadget_ep_cmd(dep, cmd, ¶ms);
>>>> - if (ret < 0) {
>>>> + if (ret < 0 && (!keep_req || ret != -EAGAIN)) {
>>>> struct dwc3_request *tmp;
>>>>
>>>> - if (ret == -EAGAIN)
>>>> - return ret;
>>>> -
>>>> dwc3_stop_active_transfer(dep, true, true);
>>>>
>>>> list_for_each_entry_safe(req, tmp, &dep->started_list, list)
>>>> @@ -1335,7 +1332,7 @@ static int __dwc3_gadget_kick_transfer(struct
>>>> dwc3_ep *dep)
>>>> if (dep->stream_capable && req->request.is_last)
>>>> dep->flags |= DWC3_EP_WAIT_TRANSFER_COMPLETE;
>>>>
>>>> - return 0;
>>>> + return ret;
>>>> }
>>>>
>>>> static int __dwc3_gadget_get_frame(struct dwc3 *dwc)
>>>> @@ -1458,7 +1455,7 @@ static int dwc3_gadget_start_isoc_quirk(struct
>>>> dwc3_ep *dep)
>>>> dep->start_cmd_status = 0;
>>>> dep->combo_num = 0;
>>>>
>>>> - return __dwc3_gadget_kick_transfer(dep);
>>>> + return __dwc3_gadget_kick_transfer(dep, false);
>>>> }
>>>>
>>>> static int __dwc3_gadget_start_isoc(struct dwc3_ep *dep)
>>>> @@ -1481,9 +1478,25 @@ static int __dwc3_gadget_start_isoc(struct
>>>> dwc3_ep *dep)
>>>> }
>>>>
>>>> for (i = 0; i < DWC3_ISOC_MAX_RETRIES; i++) {
>>>> - dep->frame_number = DWC3_ALIGN_FRAME(dep, i + 1);
>>>> + /*
>>>> + * frame_number is set from XferNotReady and may be already
>>>> + * out of date. DSTS only provides the lower 14 bit of the
>>>> + * current frame number. So add the upper two bits of
>>>> + * frame_number and handle a possible rollover.
>>>> + * This will provide the correct frame_number unless more than
>>>> + * rollover has happened since XferNotReady.
>>>> + */
>>>> + u32 frame = __dwc3_gadget_get_frame(dwc);
>>>> + bool rollover = frame < (dep->frame_number & 0x3fff);
>>>> +
>>>> + dep->frame_number = (dep->frame_number & ~0x3fff) | frame;
>>>> + if (rollover)
>>>> + dep->frame_number += BIT(14);
>>>> +
>>>> + dep->frame_number = DWC3_ALIGN_FRAME(dep, 1);
>> This is not enough, we have to add i into the alignment to ensure
>> that the stream is not failing:
>>
>> dep->frame_number = DWC3_ALIGN_FRAME(dep, i + 2);
>>
>> I am even thiking to move the frame number calculation into the
>> DWC3_DEPCMD_STARTTRANSFER code path of dwc3_send_gadget_ep_cmd. But this
>> will break the dwc3_gadget_start_isoc_quirk function. What do you think?
> You shouldn't be keep calling __dwc3_gadget_get_frame() in a loop. You
> should do all these calculation to get the current frame_number only
> once before entering the retry loop.
>
> The issue here is we don't know when the XferNotReady event will be
> handled, which may be too late and multiple uframe had gone by. But once
> the event is being handled, it shouldn't take much time at all. That
> means __dwc3_gadget_get_frame() will return the same value.
>
> So, we need something like this:
>
> diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
> index f66ff7fd87a9..1cd73c2dbe71 100644
> --- a/drivers/usb/dwc3/gadget.c
> +++ b/drivers/usb/dwc3/gadget.c
> @@ -1709,6 +1709,8 @@ static int dwc3_gadget_start_isoc_quirk(struct
> dwc3_ep *dep)
> static int __dwc3_gadget_start_isoc(struct dwc3_ep *dep)
> {
> struct dwc3 *dwc = dep->dwc;
> + u32 current_frame;
> + bool rollover;
> int ret;
> int i;
>
> @@ -1725,6 +1727,13 @@ static int __dwc3_gadget_start_isoc(struct
> dwc3_ep *dep)
> return dwc3_gadget_start_isoc_quirk(dep);
> }
>
> + current_frame = __dwc3_gadget_get_frame(dwc);
> + rollover = current_frame < (dep->frame_number & 0x3fff);
> +
> + dep->frame_number = (dep->frame_number & ~0x3fff) | current_frame;
> + if (rollover)
> + dep->frame_number += BIT(14);
> +
> for (i = 0; i < DWC3_ISOC_MAX_RETRIES; i++) {
> dep->frame_number = DWC3_ALIGN_FRAME(dep, i + 1);
>
>
> (Please create a macro for 0x3fff mask)
>
Forgot a couple of notes:
1) If bInterval is greater than 14, don't attempt to get current frame
from DSTS and ignore this mechanism.
2) The rollover check needs to account for number of uframes per
interval. If the difference is less than an interval length, then no
need to update dep->frame_number.
BR,
Thinh
