Re: snd-usb-audio: quirk for Edirol UA-5 in Advanced Mode

Re: snd-usb-audio: quirk for Edirol UA-5 in Advanced Mode

[Stephane Alnet]

The error is as follows; I've tried to move a couple locks around but I
can't figure it out.

------------------------------------------>8
bad: scheduling while atomic!
Call Trace:
 [<c0116c26>] schedule+0x3a6/0x3b0
 [<c0121eda>] schedule_timeout+0x5a/0xb0
 [<c0121e70>] process_timeout+0x0/0x10
 [<d08afaff>] usb_start_wait_urb+0xaf/0x190 [usbcore]
 [<c0116c80>] default_wake_function+0x0/0x30
 [<d08af72f>] usb_alloc_urb+0x2f/0x50 [usbcore]
 [<d08afc47>] usb_internal_control_msg+0x67/0x80 [usbcore]
 [<d08afcee>] usb_control_msg+0x8e/0xb0 [usbcore]
 [<d08b088d>] usb_set_interface+0xbd/0x220 [usbcore]
 [<d08af2a0>] hcd_endpoint_disable+0x0/0x1c0 [usbcore]
 [<d0960591>] set_format+0xd1/0x3a0 [snd_usb_audio]
 [<d09608c4>] snd_usb_pcm_prepare+0x34/0x50 [snd_usb_audio]
 [<d0948f25>] snd_pcm_do_prepare+0x15/0x40 [snd_pcm]
 [<c013ecc2>] handle_mm_fault+0xe2/0x180
 [<d09483a4>] snd_pcm_action_single+0x34/0x60 [snd_pcm]
 [<d095cbe8>] snd_pcm_action_prepare+0x0/0x18 [snd_pcm]
 [<d09484e6>] snd_pcm_action_lock_irq+0x96/0xa0 [snd_pcm]
 [<d095cbe8>] snd_pcm_action_prepare+0x0/0x18 [snd_pcm]
 [<d0948ff4>] snd_pcm_prepare+0x74/0x90 [snd_pcm]
 [<d095cbe8>] snd_pcm_action_prepare+0x0/0x18 [snd_pcm]
 [<d094bd00>] snd_pcm_playback_ioctl1+0x70/0x430 [snd_pcm]
 [<c0121bc0>] run_timer_softirq+0x100/0x1a0
 [<c015e280>] sys_ioctl+0x100/0x290
 [<c010930b>] syscall_call+0x7/0xb
------------------------------------------>8

Stephane


On Tue, 10 Jun 2003, Stephane Alnet wrote:
> Hi,
>
> (Hopefully this is the right place to post this!)
>
> Attached is a diff file for the Edirol UA-5 Advanced Mode against kernel
> 2.5.70-bk14 sound/usb.
>
> I added a new QUIRK mode (QUIRK_AUDIOSTREAM_INTERFACE) which basically
> forces an interface (and its altsettings) into Audio/AudioStream. When in
> Advanced Mode, the UA-5 exposes all its interfaces in class 255/255
> (Vendor/Vendor), but actually besides this all the descriptors are valid
> USB Audio control/interfaces/endpoints/... So this quirk simply overwrites
> the streaming interfaces class/subclass with the proper values. (My
> understanding is that we don't need to care about the audiocontrol
> interface.)
>
> Recording is working properly (tested with "arecord -c 2 -f S24_3LE -r
> 96000" on x86, for example); playback is crashing due to a "bad:
> scheduling while atomic!" problem I haven't looked at yet.
>
> (Note: this change only applies to Advanced Mode, which provides access to
> 24bits samples and the 96kHz sampling rates. The non-Advanced Mode of the
> UA-5 works just fine with the default driver at 44.1 and 48kHz, giving
> 16bits samples, as far as I can tell.)
>
> Feel free to integrate the diff into the current distribution/kernel as
> needed.
>
> I'd be interested in feedback on whether this is working for other people,
> and also help in troubleshooting the playback crash.
>
> Thank you,
> Stephane




-------------------------------------------------------
This SF.NET email is sponsored by: eBay
Great deals on office technology -- on eBay now! Click here:
http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5

Re: Re: snd-usb-audio: quirk for Edirol UA-5 in Advanced Mode

[Clemens Ladisch]

Stephane Alnet wrote:
> On Tue, 10 Jun 2003, Stephane Alnet wrote:
> > (Hopefully this is the right place to post this!)
> >
> > Attached is a diff file for the Edirol UA-5 Advanced Mode against kernel
> > 2.5.70-bk14 sound/usb.

I didn't see your first post (maybe because the current list moderator
is Dave Null), so I'm only guessing what your diff did.

> > I added a new QUIRK mode (QUIRK_AUDIOSTREAM_INTERFACE) which basically
> > forces an interface (and its altsettings) into Audio/AudioStream. When in
> > Advanced Mode, the UA-5 exposes all its interfaces in class 255/255
> > (Vendor/Vendor), but actually besides this all the descriptors are valid
> > USB Audio control/interfaces/endpoints/... So this quirk simply overwrites
> > the streaming interfaces class/subclass with the proper values. (My
> > understanding is that we don't need to care about the audiocontrol
> > interface.)

There already is a quirk type for this, QUIRK_STANDARD_INTERFACE. See
the quirk entry for the UA-20, which behaves in the same way (the
UA-20's interface #3 is for the MIDI ports).


Regards,
Clemens




-------------------------------------------------------
This SF.NET email is sponsored by: eBay
Great deals on office technology -- on eBay now! Click here:
http://adfarm.mediaplex.com/ad/ck/711-11697-6916-5

Re: Re: snd-usb-audio: quirk for Edirol UA-5 in Advanced Mode

[Stephane Alnet]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1526 bytes --]

On Thu, 12 Jun 2003, Clemens Ladisch wrote:
> I didn't see your first post (maybe because the current list moderator
> is Dave Null),

OK, so I won't feel bad sending the diff a second time, then. :)

> so I'm only guessing what your diff did.
>
[...]
> > > When in Advanced Mode, the UA-5 exposes all its interfaces in class
> > > 255/255 (Vendor/Vendor), but actually besides this all the
> > > descriptors are valid USB Audio control/interfaces/endpoints/...
[...]
> There already is a quirk type for this, QUIRK_STANDARD_INTERFACE. See
> the quirk entry for the UA-20, which behaves in the same way (the
> UA-20's interface #3 is for the MIDI ports).

That's what I hoped too (didn't want to have to change usbaudio.c if at
all possible) but it didn't seem to work. (My understanding is that
STANDARD_INTERFACE needs at least the bInterfaceSubClass info to be valid.
This isn't the case on the UA-5.)

Here's the interesting part of the quirk I proposed (mostly a rewrite of
STANDARD_INTERFACE that goes blindly trusting what usbquirks.h says):

  for( i = 0; i < iface->num_altsetting; i++ )
  {
    alts = &iface->altsetting[i];
    altsd = get_iface_desc(alts);
    altsd->bInterfaceClass = USB_CLASS_AUDIO;
    altsd->bInterfaceSubClass = USB_SUBCLASS_AUDIO_STREAMING;
  }

[I'm not sure the overwrite of Vendor with USB_CLASS_AUDIO is strictly
required, but it didn't hurt and is consistent with changing the
subclass.]

Of course since it's my first time looking at alsa/kernel, I may just be
plain off-track. :)
S.

[-- Attachment #2: Type: APPLICATION/octet-stream, Size: 1360 bytes --]

[PATCH] snd-usb-audio: quirk for Edirol UA-5 in Advanced Mode

[Clemens Ladisch]

(Jaroslav/Takashi: please apply the patch below)

Stephane Alnet wrote:
> On Thu, 12 Jun 2003, Clemens Ladisch wrote:
> > There already is a quirk type for this, QUIRK_STANDARD_INTERFACE. See
> > the quirk entry for the UA-20, which behaves in the same way (the
> > UA-20's interface #3 is for the MIDI ports).
>
> That's what I hoped too (didn't want to have to change usbaudio.c if at
> all possible) but it didn't seem to work. (My understanding is that
> STANDARD_INTERFACE needs at least the bInterfaceSubClass info to be valid.
> This isn't the case on the UA-5.)

Yes, the UA-20 does has valid subclass info, so the driver uses it.

> Here's the interesting part of the quirk I proposed (mostly a rewrite of
> STANDARD_INTERFACE that goes blindly trusting what usbquirks.h says):
> ...
>     altsd->bInterfaceClass = USB_CLASS_AUDIO;
>     altsd->bInterfaceSubClass = USB_SUBCLASS_AUDIO_STREAMING;

The _STANDARD_INTERFACE quirk type did almost the same as your
_AUDIOSTREAM_INTERFACE, but was just a special case for the UA-20, so
I've split it into two types to to allow specifying the interface type
(audio or MIDI) without reading the subclass from the descriptors. So
now we have a more generic solution for both the UA-5 and the UA-20.


Regards,
Clemens
-- 


- split QUIRK_STANDARD_INTERFACE into QUIRK_AUDIO_STANDARD_INTERFACE
  and QUIRK_MIDI_STANDARD_INTERFACE
- add quirk for Edirol UA-20


Index: alsa-kernel/usb/usbaudio.c
===================================================================
RCS file: /cvsroot/alsa/alsa-kernel/usb/usbaudio.c,v
retrieving revision 1.56
diff -u -r1.56 usbaudio.c
--- alsa-kernel/usb/usbaudio.c	4 Jun 2003 12:43:55 -0000	1.56
+++ alsa-kernel/usb/usbaudio.c	17 Jun 2003 06:56:16 -0000
@@ -2381,7 +2381,8 @@
  * create a stream for an interface with proper descriptors
  */
 static int create_standard_interface_quirk(snd_usb_audio_t *chip,
-					   struct usb_interface *iface)
+					   struct usb_interface *iface,
+					   const snd_usb_audio_quirk_t *quirk)
 {
 	struct usb_host_interface *alts;
 	struct usb_interface_descriptor *altsd;
@@ -2389,19 +2390,18 @@

 	alts = &iface->altsetting[0];
 	altsd = get_iface_desc(alts);
-	switch (altsd->bInterfaceSubClass) {
-	case USB_SUBCLASS_AUDIO_STREAMING:
+	switch (quirk->type) {
+	case QUIRK_AUDIO_STANDARD_INTERFACE:
 		err = parse_audio_endpoints(chip, altsd->bInterfaceNumber);
 		if (!err)
 			usb_set_interface(chip->dev, altsd->bInterfaceNumber, 0); /* reset the current interface */
 		break;
-	case USB_SUBCLASS_MIDI_STREAMING:
+	case QUIRK_MIDI_STANDARD_INTERFACE:
 		err = snd_usb_create_midi_interface(chip, iface, NULL);
 		break;
 	default:
-		snd_printk(KERN_ERR "if %d: non-supported subclass %d\n",
-			   altsd->bInterfaceNumber, altsd->bInterfaceSubClass);
-		return -ENODEV;
+		snd_printd(KERN_ERR "invalid quirk type %d\n", quirk->type);
+		return -ENXIO;
 	}
 	if (err < 0) {
 		snd_printk(KERN_ERR "cannot setup if %d: error %d\n",
@@ -2495,8 +2495,9 @@
 		return create_composite_quirk(chip, iface, quirk);
 	case QUIRK_AUDIO_FIXED_ENDPOINT:
 		return create_fixed_stream_quirk(chip, iface, quirk);
-	case QUIRK_STANDARD_INTERFACE:
-		return create_standard_interface_quirk(chip, iface);
+	case QUIRK_AUDIO_STANDARD_INTERFACE:
+	case QUIRK_MIDI_STANDARD_INTERFACE:
+		return create_standard_interface_quirk(chip, iface, quirk);
 	default:
 		snd_printd(KERN_ERR "invalid quirk type %d\n", quirk->type);
 		return -ENXIO;
Index: alsa-kernel/usb/usbaudio.h
===================================================================
RCS file: /cvsroot/alsa/alsa-kernel/usb/usbaudio.h,v
retrieving revision 1.17
diff -u -r1.17 usbaudio.h
--- alsa-kernel/usb/usbaudio.h	13 May 2003 10:44:09 -0000	1.17
+++ alsa-kernel/usb/usbaudio.h	17 Jun 2003 06:56:16 -0000
@@ -154,7 +154,8 @@
 #define QUIRK_MIDI_MIDIMAN		2
 #define QUIRK_COMPOSITE			3
 #define QUIRK_AUDIO_FIXED_ENDPOINT	4
-#define QUIRK_STANDARD_INTERFACE	5
+#define QUIRK_AUDIO_STANDARD_INTERFACE	5
+#define QUIRK_MIDI_STANDARD_INTERFACE	6

 typedef struct snd_usb_audio_quirk snd_usb_audio_quirk_t;
 typedef struct snd_usb_midi_endpoint_info snd_usb_midi_endpoint_info_t;
@@ -184,7 +185,7 @@

 /* for QUIRK_AUDIO_FIXED_ENDPOINT, data points to an audioformat structure */

-/* for QUIRK_STANDARD_INTERFACE, data is NULL */
+/* for QUIRK_AUDIO/MIDI_STANDARD_INTERFACE, data is NULL */

 /*
  */
Index: alsa-kernel/usb/usbquirks.h
===================================================================
RCS file: /cvsroot/alsa/alsa-kernel/usb/usbquirks.h,v
retrieving revision 1.20
diff -u -r1.20 usbquirks.h
--- alsa-kernel/usb/usbquirks.h	13 May 2003 10:44:09 -0000	1.20
+++ alsa-kernel/usb/usbquirks.h	17 Jun 2003 06:56:16 -0000
@@ -453,6 +453,36 @@
 		}
 	}
 },
+{	/*
+	 * This quirk is for the "Advanced Driver" mode of the Edirol UA-5.
+	 * If the advanced mode switch at the back of the unit is off, the
+	 * UA-5 has ID 0x0582/0x0011 and is standard compliant (no quirks),
+	 * but offers only 16-bit PCM.
+	 * In advanced mode, the UA-5 will output S24_3LE samples (two
+	 * channels) at the rate indicated on the front switch, including
+	 * the 96kHz sample rate.
+	 */
+	USB_DEVICE(0x0582, 0x0010),
+	.driver_info = (unsigned long) & (const snd_usb_audio_quirk_t) {
+		.vendor_name = "EDIROL",
+		.product_name = "UA-5",
+		.ifnum = QUIRK_ANY_INTERFACE,
+		.type = QUIRK_COMPOSITE,
+		.data = & (const snd_usb_audio_quirk_t[]) {
+			{
+				.ifnum = 1,
+				.type = QUIRK_AUDIO_STANDARD_INTERFACE
+			},
+			{
+				.ifnum = 2,
+				.type = QUIRK_AUDIO_STANDARD_INTERFACE
+			},
+			{
+				.ifnum = -1
+			}
+		}
+	}
+},
 {
 	USB_DEVICE(0x0582, 0x0012),
 	.driver_info = (unsigned long) & (const snd_usb_audio_quirk_t) {
@@ -520,15 +550,15 @@
 		.data = & (const snd_usb_audio_quirk_t[]) {
 			{
 				.ifnum = 1,
-				.type = QUIRK_STANDARD_INTERFACE
+				.type = QUIRK_AUDIO_STANDARD_INTERFACE
 			},
 			{
 				.ifnum = 2,
-				.type = QUIRK_STANDARD_INTERFACE
+				.type = QUIRK_AUDIO_STANDARD_INTERFACE
 			},
 			{
 				.ifnum = 3,
-				.type = QUIRK_STANDARD_INTERFACE
+				.type = QUIRK_MIDI_STANDARD_INTERFACE
 			},
 			{
 				.ifnum = -1




-------------------------------------------------------
This SF.Net email is sponsored by: INetU
Attention Web Developers & Consultants: Become An INetU Hosting Partner.
Refer Dedicated Servers. We Manage Them. You Get 10% Monthly Commission!
INetU Dedicated Managed Hosting http://www.inetu.net/partner/index.php

Re: [PATCH] snd-usb-audio: quirk for Edirol UA-5 in Advanced Mode

[Takashi Iwai]

At Tue, 17 Jun 2003 11:28:25 +0200 (METDST),
Clemens Ladisch wrote:
> 
> (Jaroslav/Takashi: please apply the patch below)

applied now.  thanks.


Takashi


-------------------------------------------------------
This SF.Net email is sponsored by: INetU
Attention Web Developers & Consultants: Become An INetU Hosting Partner.
Refer Dedicated Servers. We Manage Them. You Get 10% Monthly Commission!
INetU Dedicated Managed Hosting http://www.inetu.net/partner/index.php