MS Lifecam HD-5000: Don't attempt to get the sample rate

Commit Message

Joe Turner Feb. 13, 2015, 7:35 p.m. UTC

Hi Folks,

This is my first time doing anything kernel-related, so apologies in
advance for any faux-pas which are to follow.  I hope this is at least
going to the correct place!

Currently after setting the sample rate of USB devices the driver tries to
get the sample rate back from the device, to check that the two match up.
The Lifecam doesn't support getting the sample rate, and so this operation
times-out, and we continue regardless.

Because the USB timeout is pretty long this causes issues with some apps,
which begin to doubt the ability of the device to produce audio before the
timeout is up and misbehave in various ways.  See
https://www.google.co.uk/search?q=%22cannot+get+freq+at+ep+0x82%22 for a
range of reports relating to this issue.

This patch adds a quirk so that rather than check the sample rate and
timeout, we don't bother checking at all.

Hope this all looks okay.

Cheers,
Joe

Comments

Takashi Iwai Feb. 14, 2015, 5:36 p.m. UTC | #1

At Fri, 13 Feb 2015 19:35:19 +0000,
Joe Turner wrote:
> 
> Hi Folks,
> 
> This is my first time doing anything kernel-related, so apologies in
> advance for any faux-pas which are to follow.  I hope this is at least
> going to the correct place!
> 
> Currently after setting the sample rate of USB devices the driver tries to
> get the sample rate back from the device, to check that the two match up.
> The Lifecam doesn't support getting the sample rate, and so this operation
> times-out, and we continue regardless.
> 
> Because the USB timeout is pretty long this causes issues with some apps,
> which begin to doubt the ability of the device to produce audio before the
> timeout is up and misbehave in various ways.  See
> https://www.google.co.uk/search?q=%22cannot+get+freq+at+ep+0x82%22 for a
> range of reports relating to this issue.
> 
> This patch adds a quirk so that rather than check the sample rate and
> timeout, we don't bother checking at all.
> 
> Hope this all looks okay.

The workaround would be OK as long as it really works.
But the patch needs to follow the standard coding style and resent in
a formal way.  Please read Documentation/SubmittingPatches.  Try to
run scripts/checkpatch.pl at least.


thanks,

Takashi

Patch

diff -uprN -X linux-3.18/Documentation/dontdiff linux-3.18/sound/usb/clock.c linux-3.18-devel/sound/usb/clock.c
--- linux-3.18/sound/usb/clock.c	2014-12-07 22:21:05.000000000 +0000
+++ linux-3.18-devel/sound/usb/clock.c	2015-02-13 18:47:48.179363942 +0000
@@ -303,6 +303,12 @@  static int set_sample_rate_v1(struct snd
 		return err;
 	}
 
+    if (snd_usb_get_sample_rate_quirk(chip)) {
+        /* Don't check the sample rate for devices which we know don't
+         * support reading */
+        return 0;
+    }
+
 	if ((err = snd_usb_ctl_msg(dev, usb_rcvctrlpipe(dev, 0), UAC_GET_CUR,
 				   USB_TYPE_CLASS | USB_RECIP_ENDPOINT | USB_DIR_IN,
 				   UAC_EP_CS_ATTR_SAMPLE_RATE << 8, ep,
diff -uprN -X linux-3.18/Documentation/dontdiff linux-3.18/sound/usb/quirks.c linux-3.18-devel/sound/usb/quirks.c
--- linux-3.18/sound/usb/quirks.c	2014-12-07 22:21:05.000000000 +0000
+++ linux-3.18-devel/sound/usb/quirks.c	2015-02-13 18:54:52.262687929 +0000
@@ -1102,6 +1102,14 @@  void snd_usb_set_format_quirk(struct snd
 	}
 }
 
+int snd_usb_get_sample_rate_quirk(struct snd_usb_audio *chip) {
+	/* MS Lifecam HD-5000 doesn't support reading the sample rate. */
+	if (chip->usb_id == USB_ID(0x045E, 0x076D)) {
+		return 1;
+	}
+    return 0;
+}
+
 void snd_usb_endpoint_start_quirk(struct snd_usb_endpoint *ep)
 {
 	/*
diff -uprN -X linux-3.18/Documentation/dontdiff linux-3.18/sound/usb/quirks.h linux-3.18-devel/sound/usb/quirks.h
--- linux-3.18/sound/usb/quirks.h	2014-12-07 22:21:05.000000000 +0000
+++ linux-3.18-devel/sound/usb/quirks.h	2015-02-13 18:49:24.029361830 +0000
@@ -21,6 +21,9 @@  int snd_usb_apply_boot_quirk(struct usb_
 void snd_usb_set_format_quirk(struct snd_usb_substream *subs,
 			      struct audioformat *fmt);
 
+
+int snd_usb_get_sample_rate_quirk(struct snd_usb_audio *chip);
+
 int snd_usb_is_big_endian_format(struct snd_usb_audio *chip,
 				 struct audioformat *fp);