All of lore.kernel.org
 help / color / mirror / Atom feed
From: Felipe Balbi <felipe.balbi@linux.intel.com>
To: Bin Liu <b-liu@ti.com>
Cc: linux-usb@vger.kernel.org, linux-media@vger.kernel.org
Subject: Re: g_webcam Isoch high bandwidth transfer
Date: Thu, 22 Sep 2016 11:08:52 +0300	[thread overview]
Message-ID: <87lgyknyp7.fsf@linux.intel.com> (raw)
In-Reply-To: <87oa3go065.fsf@linux.intel.com>

[-- Attachment #1: Type: text/plain, Size: 5383 bytes --]


Hi,

Felipe Balbi <felipe.balbi@linux.intel.com> writes:
> Bin Liu <b-liu@ti.com> writes:
>> On Wed, Sep 21, 2016 at 11:01:21AM +0300, Felipe Balbi wrote:
>>> 
>>> Hi,
>>> 
>>> Bin Liu <b-liu@ti.com> writes:
>>> > Hi,
>>> >
>>> > I am trying to check Isoch high bandwidth transfer with g_webcam.ko in
>>> >  high-speed connection.
>>> >
>>> > First I hacked webcam.c as follows to enable 640x480@30fps mode.
>>> >
>>> > diff --git a/drivers/usb/gadget/legacy/webcam.c b/drivers/usb/gadget/legacy/webcam.c
>>> > index 72c976b..9eb315f 100644
>>> > --- a/drivers/usb/gadget/legacy/webcam.c
>>> > +++ b/drivers/usb/gadget/legacy/webcam.c
>>> > @@ -191,15 +191,15 @@ static const struct UVC_FRAME_UNCOMPRESSED(3) uvc_frame_yuv_360p = {
>>> >         .bFrameIndex            = 1,
>>> >         .bmCapabilities         = 0,
>>> >         .wWidth                 = cpu_to_le16(640),
>>> > -       .wHeight                = cpu_to_le16(360),
>>> > +       .wHeight                = cpu_to_le16(480),
>>> >         .dwMinBitRate           = cpu_to_le32(18432000),
>>> >         .dwMaxBitRate           = cpu_to_le32(55296000),
>>> > -       .dwMaxVideoFrameBufferSize      = cpu_to_le32(460800),
>>> > -       .dwDefaultFrameInterval = cpu_to_le32(666666),
>>> > +       .dwMaxVideoFrameBufferSize      = cpu_to_le32(614400),
>>> > +       .dwDefaultFrameInterval = cpu_to_le32(333333),
>>> >         .bFrameIntervalType     = 3,
>>> > -       .dwFrameInterval[0]     = cpu_to_le32(666666),
>>> > -       .dwFrameInterval[1]     = cpu_to_le32(1000000),
>>> > -       .dwFrameInterval[2]     = cpu_to_le32(5000000),
>>> > +       .dwFrameInterval[0]     = cpu_to_le32(333333),
>>> > +       .dwFrameInterval[1]     = cpu_to_le32(666666),
>>> > +       .dwFrameInterval[2]     = cpu_to_le32(1000000),
>>> >  };
>>> >
>>> > then loaded g_webcam.ko as
>>> >
>>> > # modprobe g_webcam streaming_maxpacket=3072
>>> >
>>> > The endpoint descriptor showing on the host is
>>> >
>>> >       Endpoint Descriptor:
>>> >         bLength                 7
>>> >         bDescriptorType         5
>>> >         bEndpointAddress     0x8d  EP 13 IN
>>> >         bmAttributes            5
>>> >           Transfer Type            Isochronous
>>> >           Synch Type               Asynchronous
>>> >           Usage Type               Data
>>> >         wMaxPacketSize     0x1400  3x 1024 bytes
>>> >         bInterval               1
>>> >
>>> > However the usb bus trace shows only one transaction with 1024-bytes packet in
>>> > every SOF. The host only sends one IN packet in every SOF, I am expecting 2~3
>>> > 1024-bytes transactions, since this would be required to transfer 640x480@30fps
>>> > YUV frames in high-speed.
>>> >
>>> > DId I miss anything in the setup?
>>> 
>>> MUSB or DWC3? This looks like a UDC bug to me. Can you show a screenshot
>>
>> Happened on both MUSB and DWC3.
>>
>>> of your bus analyzer? When host sends IN token, are you replying with
>>
>> The trace screenshot on DWC3 is attached.
>>
>>> DATA0, DATA1 or DATA2?
>>
>> Good hint! It is DATA0!
>
> yeah, should've been DATA2. I'll check if we're missing anything for
> High Bandwidth Iso on DWC3. Can you confirm if it works of tails on
> DWC3? On your follow-up mail you mentioned it's a bug in MUSB. What
> about DWC3?

I'm assuming DWC3 really breaks. Here's a patch for that:

8<---------------------------------- cut here ----------------------------------
From 62807011c00055785575bb39d92bfe8836817e2f Mon Sep 17 00:00:00 2001
From: Felipe Balbi <felipe.balbi@linux.intel.com>
Date: Thu, 22 Sep 2016 11:01:01 +0300
Subject: [PATCH] usb: dwc3: gadget: set PCM1 field of isochronous-first TRBs

In case of High-Speed, High-Bandwidth endpoints, we
need to tell DWC3 that we have more than one packet
per interval. We do that by setting PCM1 field of
Isochronous-First TRB.

Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
---
 drivers/usb/dwc3/gadget.c | 14 ++++++++++++--
 1 file changed, 12 insertions(+), 2 deletions(-)

diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c
index 602f12254161..106623faf060 100644
--- a/drivers/usb/dwc3/gadget.c
+++ b/drivers/usb/dwc3/gadget.c
@@ -787,6 +787,9 @@ static void dwc3_prepare_one_trb(struct dwc3_ep *dep,
 		unsigned length, unsigned chain, unsigned node)
 {
 	struct dwc3_trb		*trb;
+	struct dwc3		*dwc = dep->dwc;
+	struct usb_gadget	*gadget = &dwc->gadget;
+	enum usb_device_speed	speed = speed;
 
 	dwc3_trace(trace_dwc3_gadget, "%s: req %p dma %08llx length %d%s",
 			dep->name, req, (unsigned long long) dma,
@@ -813,10 +816,17 @@ static void dwc3_prepare_one_trb(struct dwc3_ep *dep,
 		break;
 
 	case USB_ENDPOINT_XFER_ISOC:
-		if (!node)
+		if (!node) {
 			trb->ctrl = DWC3_TRBCTL_ISOCHRONOUS_FIRST;
-		else
+
+			if (speed == USB_SPEED_HIGH) {
+				struct usb_endpoint *ep = &dep->endpoint;
+				u8 pkts = DIV_ROUNDUP(ep->maxpacket, 1024);
+				trb->size |= DWC3_TRB_SIZE_PCM1(pkts - 1);
+			}
+		} else {
 			trb->ctrl = DWC3_TRBCTL_ISOCHRONOUS;
+		}
 
 		/* always enable Interrupt on Missed ISOC */
 		trb->ctrl |= DWC3_TRB_CTRL_ISP_IMI;
-- 
2.10.0

Let me know if this helps DWC3, also, let me know how you're using
g_webcam, I'd like to test that out on my side too ;-)

-- 
balbi

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 800 bytes --]

  reply	other threads:[~2016-09-22  8:09 UTC|newest]

Thread overview: 19+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-09-20 17:04 g_webcam Isoch high bandwidth transfer Bin Liu
2016-09-21  8:01 ` Felipe Balbi
2016-09-21 13:27   ` Bin Liu
2016-09-21 14:58     ` Bin Liu
2016-09-22  7:37     ` Felipe Balbi
2016-09-22  8:08       ` Felipe Balbi [this message]
2016-09-22 10:06         ` Felipe Balbi
2016-09-22 13:33           ` Bin Liu
2016-09-22 19:48             ` Felipe Balbi
2016-09-22 20:11               ` Bin Liu
2016-09-23  0:23                 ` yfw
2016-09-23  7:49                 ` Felipe Balbi
2016-09-23  8:10                   ` yfw
2016-09-23  8:27                     ` Felipe Balbi
2016-09-26 16:06                       ` Laurent Pinchart
2016-09-27  7:10                         ` Felipe Balbi
2016-09-26 14:02                   ` Bin Liu
2016-09-27  7:14                     ` Felipe Balbi
2016-09-26 16:03                   ` Laurent Pinchart

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=87lgyknyp7.fsf@linux.intel.com \
    --to=felipe.balbi@linux.intel.com \
    --cc=b-liu@ti.com \
    --cc=linux-media@vger.kernel.org \
    --cc=linux-usb@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.