UAC2 Gadget: feedback endpoint support
Subsystems (2)
cover UAC2 Gadget: feedback endpoint support
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).
While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget: f_uac2: add adaptive sync support for capture
usb: gadget: u_audio: add real feedback implementation
Documentation/ABI/testing/configfs-usb-gadget-uac2 | 1 + Documentation/usb/gadget-testing.rst | 1 + drivers/usb/gadget/function/f_uac2.c | 118 ++++++++++- drivers/usb/gadget/function/u_audio.c | 217 ++++++++++++++++++++- drivers/usb/gadget/function/u_audio.h | 10 + drivers/usb/gadget/function/u_uac2.h | 2 + 6 files changed, 345 insertions(+), 4 deletions(-)
Signature · 1 line
1.9.1
PATCH 1/3 usb: gadget: f_uac2/u_audio: add feedback endpoint support
As per USB and UAC2 specs, asynchronous audio sink endpoint requires explicit synchronization mechanism (Isochronous Feedback Endpoint)
Implement feedback companion endpoint for ISO OUT endpoint
This patch adds all required infrastructure and USB requests handling for feedback endpoint. Syncrhonization itself is still dummy (feedback ep always reports 'nomimal frequency'
e.g. no adjustement is needed). This satisfies hosts that
require feedback endpoint (like Win10) and poll it periodically
Actual synchronization mechanism should be implemented separately
--- drivers/usb/gadget/function/f_uac2.c | 34 +++++++++- drivers/usb/gadget/function/u_audio.c | 124 +++++++++++++++++++++++++++++++++- drivers/usb/gadget/function/u_audio.h | 3 + 3 files changed, 158 insertions(+), 3 deletions(-)
drivers/usb/gadget/function/f_uac2.c +33 -1
@@ -237,7 +237,7 @@ enum { .bDescriptorType = USB_DT_INTERFACE, .bAlternateSetting = 1, - .bNumEndpoints = 1, + .bNumEndpoints = 2, .bInterfaceClass = USB_CLASS_AUDIO, .bInterfaceSubClass = USB_SUBCLASS_AUDIOSTREAMING, .bInterfaceProtocol = UAC_VERSION_2,
@@ -296,6 +296,27 @@ enum { .wLockDelay = 0, }; +/* STD AS ISO IN Feedback Endpoint */ +static struct usb_endpoint_descriptor fs_epin_fback_desc = { + .bLength =USB_DT_ENDPOINT_SIZE , + .bDescriptorType =USB_DT_ENDPOINT , + + .bEndpointAddress =USB_DIR_IN , + .bmAttributes =USB_ENDPOINT_XFER_ISOC |USB_ENDPOINT_USAGE_FEEDBACK , + .wMaxPacketSize =cpu_to_le16 (3 ), + .bInterval =1 , +}; + +static struct usb_endpoint_descriptor hs_epin_fback_desc = { + .bLength =USB_DT_ENDPOINT_SIZE , + .bDescriptorType =USB_DT_ENDPOINT , + + .bmAttributes =USB_ENDPOINT_XFER_ISOC |USB_ENDPOINT_USAGE_FEEDBACK , + .wMaxPacketSize =cpu_to_le16 (4 ), + .bInterval =4 , +}; + +/* Audio Streaming IN Interface - Alt0 */ static struct usb_interface_descriptor std_as_in_if0_desc = { .bLength =sizeof std_as_in_if0_desc,
@@ -392,6 +413,7 @@ enum { (struct usb_descriptor_header *)&as_out_fmt1_desc, (struct usb_descriptor_header *)&fs_epout_desc, (struct usb_descriptor_header *)&as_iso_out_desc, + (struct usb_descriptor_header *)&fs_epin_fback_desc, (struct usb_descriptor_header *)&std_as_in_if0_desc, (struct usb_descriptor_header *)&std_as_in_if1_desc,
@@ -422,6 +444,7 @@ enum { (struct usb_descriptor_header *)&as_out_fmt1_desc, (struct usb_descriptor_header *)&hs_epout_desc, (struct usb_descriptor_header *)&as_iso_out_desc, + (struct usb_descriptor_header *)&hs_epin_fback_desc, (struct usb_descriptor_header *)&std_as_in_if0_desc, (struct usb_descriptor_header *)&std_as_in_if1_desc,
@@ -541,6 +564,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) fs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); fs_audio_desc[i++] =USBDHDR (&fs_epout_desc); fs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); + fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); }if (EPIN_EN (opts)) { fs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);
@@ -574,6 +598,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) hs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); hs_audio_desc[i++] =USBDHDR (&hs_epout_desc); hs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); + hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); }if (EPIN_EN (opts)) { hs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);
@@ -681,6 +706,12 @@ static void setup_descriptor(struct f_uac2_opts *opts)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__);return -ENODEV ; } + agdev->in_ep_fback =usb_ep_autoconfig (gadget, + &fs_epin_fback_desc); +if (!agdev->in_ep_fback) { +dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); +return -ENODEV ; + } }if (EPIN_EN (uac2_opts)) {
@@ -699,6 +730,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) le16_to_cpu(hs_epout_desc.wMaxPacketSize)); hs_epout_desc.bEndpointAddress = fs_epout_desc.bEndpointAddress; + hs_epin_fback_desc.bEndpointAddress = fs_epin_fback_desc.bEndpointAddress; hs_epin_desc.bEndpointAddress = fs_epin_desc.bEndpointAddress;setup_descriptor (uac2_opts);
drivers/usb/gadget/function/u_audio.c +122 -2
166 lines not on this page · raw message
drivers/usb/gadget/function/u_audio.h +3 -0
@@ -30,7 +30,10 @@ struct g_audio {struct usb_gadget *gadget;struct usb_ep *in_ep; +struct usb_ep *out_ep; +/* feedback IN endpoint corresponding to out_ep */ +struct usb_ep *in_ep_fback;/* Max packet size for all in_ep possible speeds */ unsigned int in_ep_maxpsize;
Signature · 1 line
1.9.1
Re: [PATCH 1/3] usb: gadget: f_uac2/u_audio: add feedback endpoint support
Quoted 210 lines · Ruslan Bilovol
As per USB and UAC2 specs, asynchronous audio sink endpoint requires explicit synchronization mechanism (Isochronous Feedback Endpoint)
Implement feedback companion endpoint for ISO OUT endpoint
This patch adds all required infrastructure and USB requests handling for feedback endpoint. Syncrhonization itself is still dummy (feedback ep always reports 'nomimal frequency'
e.g. no adjustement is needed). This satisfies hosts that
require feedback endpoint (like Win10) and poll it periodicallyActual synchronization mechanism should be implemented separately
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 34 +++++++++- drivers/usb/gadget/function/u_audio.c | 124 +++++++++++++++++++++++++++++++++- drivers/usb/gadget/function/u_audio.h | 3 + 3 files changed, 158 insertions(+), 3 deletions(-)drivers/usb/gadget/function/f_uac2.c +33
-1@@ -237,7 +237,7 @@ enum { .bDescriptorType = USB_DT_INTERFACE, .bAlternateSetting = 1, - .bNumEndpoints = 1, + .bNumEndpoints = 2, .bInterfaceClass = USB_CLASS_AUDIO, .bInterfaceSubClass = USB_SUBCLASS_AUDIOSTREAMING, .bInterfaceProtocol = UAC_VERSION_2,@@ -296,6 +296,27 @@ enum { .wLockDelay = 0, }; +/* STD AS ISO IN Feedback Endpoint */ +static struct usb_endpoint_descriptor fs_epin_fback_desc = { + .bLength =USB_DT_ENDPOINT_SIZE , + .bDescriptorType =USB_DT_ENDPOINT , + + .bEndpointAddress =USB_DIR_IN , + .bmAttributes =USB_ENDPOINT_XFER_ISOC |USB_ENDPOINT_USAGE_FEEDBACK , + .wMaxPacketSize =cpu_to_le16 (3 ), + .bInterval =1 , +}; + +static struct usb_endpoint_descriptor hs_epin_fback_desc = { + .bLength =USB_DT_ENDPOINT_SIZE , + .bDescriptorType =USB_DT_ENDPOINT , + + .bmAttributes =USB_ENDPOINT_XFER_ISOC |USB_ENDPOINT_USAGE_FEEDBACK , + .wMaxPacketSize =cpu_to_le16 (4 ), + .bInterval =4 , +}; + +/* Audio Streaming IN Interface - Alt0 */ static struct usb_interface_descriptor std_as_in_if0_desc = { .bLength =sizeof std_as_in_if0_desc,@@ -392,6 +413,7 @@ enum { (struct usb_descriptor_header *)&as_out_fmt1_desc, (struct usb_descriptor_header *)&fs_epout_desc, (struct usb_descriptor_header *)&as_iso_out_desc, + (struct usb_descriptor_header *)&fs_epin_fback_desc, (struct usb_descriptor_header *)&std_as_in_if0_desc, (struct usb_descriptor_header *)&std_as_in_if1_desc,@@ -422,6 +444,7 @@ enum { (struct usb_descriptor_header *)&as_out_fmt1_desc, (struct usb_descriptor_header *)&hs_epout_desc, (struct usb_descriptor_header *)&as_iso_out_desc, + (struct usb_descriptor_header *)&hs_epin_fback_desc, (struct usb_descriptor_header *)&std_as_in_if0_desc, (struct usb_descriptor_header *)&std_as_in_if1_desc,@@ -541,6 +564,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) fs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); fs_audio_desc[i++] =USBDHDR (&fs_epout_desc); fs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); + fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); }if (EPIN_EN (opts)) { fs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -574,6 +598,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) hs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); hs_audio_desc[i++] =USBDHDR (&hs_epout_desc); hs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); + hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); }if (EPIN_EN (opts)) { hs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -681,6 +706,12 @@ static void setup_descriptor(struct f_uac2_opts *opts)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__);return -ENODEV ; } + agdev->in_ep_fback =usb_ep_autoconfig (gadget, + &fs_epin_fback_desc); +if (!agdev->in_ep_fback) { +dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); +return -ENODEV ; + } }if (EPIN_EN (uac2_opts)) {@@ -699,6 +730,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) le16_to_cpu(hs_epout_desc.wMaxPacketSize)); hs_epout_desc.bEndpointAddress = fs_epout_desc.bEndpointAddress; + hs_epin_fback_desc.bEndpointAddress = fs_epin_fback_desc.bEndpointAddress; hs_epin_desc.bEndpointAddress = fs_epin_desc.bEndpointAddress;setup_descriptor (uac2_opts);drivers/usb/gadget/function/u_audio.c +65
-0@@ -43,6 +43,10 @@ struct uac_rtd_params {unsigned int max_psize;/* MaxPacketSize of endpoint */ struct uac_req *ureq; +struct usb_request *req_fback;/* Feedback endpoint request */ +unsigned int ffback;/* Real frequency reported by feedback endpoint */ +bool fb_ep_enabled;/* if the ep is enabled */ +spinlock_t lock; };@@ -76,6 +80,35 @@ struct snd_uac_chip { .periods_min = MIN_PERIODS, }; +static void u_audio_set_fback_frequency (enum usb_device_speed speed, +unsigned int freq,void *buf) +{ +u32 ff =0 ; + +if (speed ==USB_SPEED_FULL ) { +/* +* Full-speed feedback endpoints report frequency +* in samples/microframe +* Format is encoded in Q10.10 left-justified in the 24 bits, +* so that it has a Q10.14 format. +*/ + ff =DIV_ROUND_UP ((freq <<14 ),1000 ); + }else { +/* +* High-speed feedback endpoints report frequency +* in samples/microframe. +* Format is encoded in Q12.13 fitted into four bytes so that +* the binary point is located between the second and the third +* byte fromat (that is Q16.16) +* +* Prevent integer overflow by calculating in Q12.13 fromat and +* then shifting to Q16.16 +*/ + ff =DIV_ROUND_UP ((freq <<13 ), (8 *1000 )) <<3 ; + } + *(__le32 *)buf =cpu_to_le32 (ff); +} +static void u_audio_iso_complete (struct usb_ep *ep,struct usb_request *req) {unsigned int pending;@@ -182,6 +215,36 @@ static void u_audio_iso_complete(struct usb_ep *ep, struct usb_request *req)dev_err (uac->card->dev,"%d Error!\n" , __LINE__); } +static void u_audio_iso_fback_complete (struct usb_ep *ep, +struct usb_request *req) +{ +struct uac_rtd_params *prm = req->context; +struct snd_uac_chip *uac = prm->uac; +struct g_audio *audio_dev = uac->audio_dev; +int status = req->status; +unsigned long flags; + +/* i/f shutting down */ +if (!prm->fb_ep_enabled || req->status == -ESHUTDOWN ) +return ; + +/* +* We can't really do much about bad xfers. +* Afterall, the ISOCH xfers could fail legitimately. +*/ +if (status) +pr_debug ("%s: iso_complete status(%d) %d/%d\n" , + __func__, status, req->actual, req->length); + +spin_lock_irqsave (&prm->lock, flags); +u_audio_set_fback_frequency (audio_dev->gadget->speed, + prm->ffback, req->buf); +spin_unlock_irqrestore (&prm->lock, flags); + +if (usb_ep_queue (ep, req,GFP_ATOMIC )) +dev_err (uac->card->dev,"%d Error!\n" , __LINE__); +} +static int uac_pcm_trigger (struct snd_pcm_substream *substream,int cmd) {struct snd_uac_chip *uac =snd_pcm_substream_chip (substream);@@ -346,14 +409,33 @@ static inline void free_ep(struct uac_rtd_params *prm, struct usb_ep *ep)dev_err (uac->card->dev,"%s:%d Error!\n" , __func__, __LINE__); } +static inline void free_ep_fback (struct uac_rtd_params *prm,struct usb_ep *ep) +{
drivers/usb/gadget/function/u_audio.c static inline void free_ep(struct uac_rtd_params *prm, struct usb_ep *ep) · line 423
Quoted 10 lines
drivers/usb/gadget/function/u_audio.c +10
-0+struct snd_uac_chip *uac = prm->uac; + +if (!prm->fb_ep_enabled) +return ; + + prm->fb_ep_enabled = false; + +if (prm->req_fback) { +usb_ep_dequeue (ep, prm->req_fback); +kfree (prm->req_fback->buf);
You may need to free buf at completion, Jack Pham found a use-after-free bug at dwc3 controller due to async execution for usb_ep_dequeue.
Quoted 42 lines
drivers/usb/gadget/function/u_audio.c +27
-2+usb_ep_free_request (ep, prm->req_fback); + prm->req_fback =NULL ; + } + +if (usb_ep_disable (ep)) +dev_err (uac->card->dev,"%s:%d Error!\n" , __func__, __LINE__); +}int u_audio_start_capture (struct g_audio *audio_dev) {struct snd_uac_chip *uac = audio_dev->uac;struct usb_gadget *gadget = audio_dev->gadget;struct device *dev = &gadget->dev; -struct usb_request *req; -struct usb_ep *ep; +struct usb_request *req, *req_fback; +struct usb_ep *ep, *ep_fback;struct uac_rtd_params *prm;struct uac_params *params = &audio_dev->params;int req_len, i;@@ -386,6 +468,42 @@ int u_audio_start_capture(struct g_audio *audio_dev)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); } + ep_fback = audio_dev->in_ep_fback; +if (!ep_fback) +return 0 ; + +/* Setup feedback endpoint */ +config_ep_by_speed (gadget, &audio_dev->func, ep_fback); + prm->fb_ep_enabled = true; +usb_ep_enable (ep_fback); + req_len = ep_fback->maxpacket; + + req_fback =usb_ep_alloc_request (ep_fback,GFP_ATOMIC ); +if (req_fback ==NULL ) +return -ENOMEM ; + + prm->req_fback = req_fback; + req_fback->zero =0 ; + req_fback->context = prm; + req_fback->length = req_len;
drivers/usb/gadget/function/u_audio.c int u_audio_start_capture(struct g_audio *audio_dev) · line 498
Quoted 10 lines
drivers/usb/gadget/function/u_audio.c +10
-0+ req_fback->complete = u_audio_iso_fback_complete; + + req_fback->buf =kzalloc (req_len,GFP_ATOMIC ); +if (!req_fback->buf) +return -ENOMEM ; + +/* +* Configure the feedback endpoint's reported frequency. +* Always start with original frequency since its deviation can't +* be meauserd at start of playback
%s/meauserd/measured
Quoted 38 lines
drivers/usb/gadget/function/u_audio.c +10
-0+*/ + prm->ffback = params->c_srate; +u_audio_set_fback_frequency (audio_dev->gadget->speed, + prm->ffback, req_fback->buf); + +if (usb_ep_queue (ep_fback, req_fback,GFP_ATOMIC )) +dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); +return 0 ; }EXPORT_SYMBOL_GPL (u_audio_start_capture);@@ -394,6 +512,8 @@ void u_audio_stop_capture(struct g_audio *audio_dev) {struct snd_uac_chip *uac = audio_dev->uac; +if (audio_dev->in_ep_fback) +free_ep_fback (&uac->c_prm, audio_dev->in_ep_fback);free_ep (&uac->c_prm, audio_dev->out_ep); }EXPORT_SYMBOL_GPL (u_audio_stop_capture);drivers/usb/gadget/function/u_audio.h +3
-0@@ -30,7 +30,10 @@ struct g_audio {struct usb_gadget *gadget;struct usb_ep *in_ep; +struct usb_ep *out_ep; +/* feedback IN endpoint corresponding to out_ep */ +struct usb_ep *in_ep_fback;/* Max packet size for all in_ep possible speeds */ unsigned int in_ep_maxpsize;Signature · 1 line
1.9.1
Signature · 2 lines
Thanks, Peter Chen
Re: [PATCH 1/3] usb: gadget: f_uac2/u_audio: add feedback endpoint support
Quoted 226 lines · Peter Chen
Quoted 210 lines · Ruslan Bilovol
As per USB and UAC2 specs, asynchronous audio sink endpoint requires explicit synchronization mechanism (Isochronous Feedback Endpoint)
Implement feedback companion endpoint for ISO OUT endpoint
This patch adds all required infrastructure and USB requests handling for feedback endpoint. Syncrhonization itself is still dummy (feedback ep always reports 'nomimal frequency'
e.g. no adjustement is needed). This satisfies hosts that
require feedback endpoint (like Win10) and poll it periodicallyActual synchronization mechanism should be implemented separately
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 34 +++++++++- drivers/usb/gadget/function/u_audio.c | 124 +++++++++++++++++++++++++++++++++- drivers/usb/gadget/function/u_audio.h | 3 + 3 files changed, 158 insertions(+), 3 deletions(-)drivers/usb/gadget/function/f_uac2.c +33
-1@@ -237,7 +237,7 @@ enum { .bDescriptorType = USB_DT_INTERFACE, .bAlternateSetting = 1, - .bNumEndpoints = 1, + .bNumEndpoints = 2, .bInterfaceClass = USB_CLASS_AUDIO, .bInterfaceSubClass = USB_SUBCLASS_AUDIOSTREAMING, .bInterfaceProtocol = UAC_VERSION_2,@@ -296,6 +296,27 @@ enum { .wLockDelay = 0, }; +/* STD AS ISO IN Feedback Endpoint */ +static struct usb_endpoint_descriptor fs_epin_fback_desc = { + .bLength =USB_DT_ENDPOINT_SIZE , + .bDescriptorType =USB_DT_ENDPOINT , + + .bEndpointAddress =USB_DIR_IN , + .bmAttributes =USB_ENDPOINT_XFER_ISOC |USB_ENDPOINT_USAGE_FEEDBACK , + .wMaxPacketSize =cpu_to_le16 (3 ), + .bInterval =1 , +}; + +static struct usb_endpoint_descriptor hs_epin_fback_desc = { + .bLength =USB_DT_ENDPOINT_SIZE , + .bDescriptorType =USB_DT_ENDPOINT , + + .bmAttributes =USB_ENDPOINT_XFER_ISOC |USB_ENDPOINT_USAGE_FEEDBACK , + .wMaxPacketSize =cpu_to_le16 (4 ), + .bInterval =4 , +}; + +/* Audio Streaming IN Interface - Alt0 */ static struct usb_interface_descriptor std_as_in_if0_desc = { .bLength =sizeof std_as_in_if0_desc,@@ -392,6 +413,7 @@ enum { (struct usb_descriptor_header *)&as_out_fmt1_desc, (struct usb_descriptor_header *)&fs_epout_desc, (struct usb_descriptor_header *)&as_iso_out_desc, + (struct usb_descriptor_header *)&fs_epin_fback_desc, (struct usb_descriptor_header *)&std_as_in_if0_desc, (struct usb_descriptor_header *)&std_as_in_if1_desc,@@ -422,6 +444,7 @@ enum { (struct usb_descriptor_header *)&as_out_fmt1_desc, (struct usb_descriptor_header *)&hs_epout_desc, (struct usb_descriptor_header *)&as_iso_out_desc, + (struct usb_descriptor_header *)&hs_epin_fback_desc, (struct usb_descriptor_header *)&std_as_in_if0_desc, (struct usb_descriptor_header *)&std_as_in_if1_desc,@@ -541,6 +564,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) fs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); fs_audio_desc[i++] =USBDHDR (&fs_epout_desc); fs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); + fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); }if (EPIN_EN (opts)) { fs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -574,6 +598,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) hs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); hs_audio_desc[i++] =USBDHDR (&hs_epout_desc); hs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); + hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); }if (EPIN_EN (opts)) { hs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -681,6 +706,12 @@ static void setup_descriptor(struct f_uac2_opts *opts)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__);return -ENODEV ; } + agdev->in_ep_fback =usb_ep_autoconfig (gadget, + &fs_epin_fback_desc); +if (!agdev->in_ep_fback) { +dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); +return -ENODEV ; + } }if (EPIN_EN (uac2_opts)) {@@ -699,6 +730,7 @@ static void setup_descriptor(struct f_uac2_opts *opts) le16_to_cpu(hs_epout_desc.wMaxPacketSize)); hs_epout_desc.bEndpointAddress = fs_epout_desc.bEndpointAddress; + hs_epin_fback_desc.bEndpointAddress = fs_epin_fback_desc.bEndpointAddress; hs_epin_desc.bEndpointAddress = fs_epin_desc.bEndpointAddress;setup_descriptor (uac2_opts);drivers/usb/gadget/function/u_audio.c +65
-0@@ -43,6 +43,10 @@ struct uac_rtd_params {unsigned int max_psize;/* MaxPacketSize of endpoint */ struct uac_req *ureq; +struct usb_request *req_fback;/* Feedback endpoint request */ +unsigned int ffback;/* Real frequency reported by feedback endpoint */ +bool fb_ep_enabled;/* if the ep is enabled */ +spinlock_t lock; };@@ -76,6 +80,35 @@ struct snd_uac_chip { .periods_min = MIN_PERIODS, }; +static void u_audio_set_fback_frequency (enum usb_device_speed speed, +unsigned int freq,void *buf) +{ +u32 ff =0 ; + +if (speed ==USB_SPEED_FULL ) { +/* +* Full-speed feedback endpoints report frequency +* in samples/microframe +* Format is encoded in Q10.10 left-justified in the 24 bits, +* so that it has a Q10.14 format. +*/ + ff =DIV_ROUND_UP ((freq <<14 ),1000 ); + }else { +/* +* High-speed feedback endpoints report frequency +* in samples/microframe. +* Format is encoded in Q12.13 fitted into four bytes so that +* the binary point is located between the second and the third +* byte fromat (that is Q16.16) +* +* Prevent integer overflow by calculating in Q12.13 fromat and +* then shifting to Q16.16 +*/ + ff =DIV_ROUND_UP ((freq <<13 ), (8 *1000 )) <<3 ; + } + *(__le32 *)buf =cpu_to_le32 (ff); +} +static void u_audio_iso_complete (struct usb_ep *ep,struct usb_request *req) {unsigned int pending;@@ -182,6 +215,36 @@ static void u_audio_iso_complete(struct usb_ep *ep, struct usb_request *req)dev_err (uac->card->dev,"%d Error!\n" , __LINE__); } +static void u_audio_iso_fback_complete (struct usb_ep *ep, +struct usb_request *req) +{ +struct uac_rtd_params *prm = req->context; +struct snd_uac_chip *uac = prm->uac; +struct g_audio *audio_dev = uac->audio_dev; +int status = req->status; +unsigned long flags; + +/* i/f shutting down */ +if (!prm->fb_ep_enabled || req->status == -ESHUTDOWN ) +return ; + +/* +* We can't really do much about bad xfers. +* Afterall, the ISOCH xfers could fail legitimately. +*/ +if (status) +pr_debug ("%s: iso_complete status(%d) %d/%d\n" , + __func__, status, req->actual, req->length); + +spin_lock_irqsave (&prm->lock, flags); +u_audio_set_fback_frequency (audio_dev->gadget->speed, + prm->ffback, req->buf); +spin_unlock_irqrestore (&prm->lock, flags); + +if (usb_ep_queue (ep, req,GFP_ATOMIC )) +dev_err (uac->card->dev,"%d Error!\n" , __LINE__); +} +static int uac_pcm_trigger (struct snd_pcm_substream *substream,int cmd) {struct snd_uac_chip *uac =snd_pcm_substream_chip (substream);@@ -346,14 +409,33 @@ static inline void free_ep(struct uac_rtd_params *prm, struct usb_ep *ep)dev_err (uac->card->dev,"%s:%d Error!\n" , __func__, __LINE__); } +static inline void free_ep_fback (struct uac_rtd_params *prm,struct usb_ep *ep) +{drivers/usb/gadget/function/u_audio.c static inline void free_ep(struct uac_rtd_params *prm, struct usb_ep *ep) · line 423
Quoted 10 lines
drivers/usb/gadget/function/u_audio.c +10
-0+struct snd_uac_chip *uac = prm->uac; + +if (!prm->fb_ep_enabled) +return ; + + prm->fb_ep_enabled = false; + +if (prm->req_fback) { +usb_ep_dequeue (ep, prm->req_fback); +kfree (prm->req_fback->buf);You may need to free buf at completion, Jack Pham found a use-after-free bug at dwc3 controller due to async execution for
usb_ep_dequeue.
Will review and retest with that change
Quoted 47 lines
Quoted 47 lines
drivers/usb/gadget/function/u_audio.c +31
-2+usb_ep_free_request (ep, prm->req_fback); + prm->req_fback =NULL ; + } + +if (usb_ep_disable (ep)) +dev_err (uac->card->dev,"%s:%d Error!\n" , __func__, __LINE__); +}int u_audio_start_capture (struct g_audio *audio_dev) {struct snd_uac_chip *uac = audio_dev->uac;struct usb_gadget *gadget = audio_dev->gadget;struct device *dev = &gadget->dev; -struct usb_request *req; -struct usb_ep *ep; +struct usb_request *req, *req_fback; +struct usb_ep *ep, *ep_fback;struct uac_rtd_params *prm;struct uac_params *params = &audio_dev->params;int req_len, i;@@ -386,6 +468,42 @@ int u_audio_start_capture(struct g_audio *audio_dev)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); } + ep_fback = audio_dev->in_ep_fback; +if (!ep_fback) +return 0 ; + +/* Setup feedback endpoint */ +config_ep_by_speed (gadget, &audio_dev->func, ep_fback); + prm->fb_ep_enabled = true; +usb_ep_enable (ep_fback); + req_len = ep_fback->maxpacket; + + req_fback =usb_ep_alloc_request (ep_fback,GFP_ATOMIC ); +if (req_fback ==NULL ) +return -ENOMEM ; + + prm->req_fback = req_fback; + req_fback->zero =0 ; + req_fback->context = prm; + req_fback->length = req_len; + req_fback->complete = u_audio_iso_fback_complete; + + req_fback->buf =kzalloc (req_len,GFP_ATOMIC ); +if (!req_fback->buf)
Quoted 8 lines
drivers/usb/gadget/function/u_audio.c int u_audio_start_capture(struct g_audio *audio_dev) · line 498
Quoted 6 lines
drivers/usb/gadget/function/u_audio.c +6
-0+return -ENOMEM ; + +/* +* Configure the feedback endpoint's reported frequency. +* Always start with original frequency since its deviation can't +* be meauserd at start of playback %s/meauserd/measured
Good catch!
Thanks,
Ruslan
Quoted 44 lines
Quoted 37 lines
drivers/usb/gadget/function/u_audio.c +10
-0+*/ + prm->ffback = params->c_srate; +u_audio_set_fback_frequency (audio_dev->gadget->speed, + prm->ffback, req_fback->buf); + +if (usb_ep_queue (ep_fback, req_fback,GFP_ATOMIC )) +dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); +return 0 ; }EXPORT_SYMBOL_GPL (u_audio_start_capture);@@ -394,6 +512,8 @@ void u_audio_stop_capture(struct g_audio *audio_dev) {struct snd_uac_chip *uac = audio_dev->uac; +if (audio_dev->in_ep_fback) +free_ep_fback (&uac->c_prm, audio_dev->in_ep_fback);free_ep (&uac->c_prm, audio_dev->out_ep); }EXPORT_SYMBOL_GPL (u_audio_stop_capture);drivers/usb/gadget/function/u_audio.h +3
-1@@ -30,7 +30,10 @@ struct g_audio {struct usb_gadget *gadget;struct usb_ep *in_ep; +struct usb_ep *out_ep; +/* feedback IN endpoint corresponding to out_ep */ +struct usb_ep *in_ep_fback;/* Max packet size for all in_ep possible speeds */ unsigned int in_ep_maxpsize; --Quoted 2 lines
1.9.1
--
Thanks,
Peter Chen
PATCH 2/3 usb: gadget: f_uac2: add adaptive sync support for capture
In reply to Ruslan BilovolCurrent f_uac2 USB OUT (aka 'capture') synchronization implements 'ASYNC' scenario which means USB Gadget has it's own freerunning clock and can update Host about real clock frequency through feedback endpoint so Host can align number of samples sent to the USB gadget to prevent overruns/underruns
In case if Gadget can has no it's internal clock and can consume audio samples at any rate (for example, on the Gadget side someone records audio directly to a file, or audio samples are played through an external DAC as soon as they arrive), UAC2 spec suggests 'ADAPTIVE' synchronization type.
Change UAC2 driver to make it configurable through additional 'c_sync' configfs file.
Default remains 'asynchronous' with possibility to switch it to 'adaptive'
--- Documentation/ABI/testing/configfs-usb-gadget-uac2 | 1 + Documentation/usb/gadget-testing.rst | 1 + drivers/usb/gadget/function/f_uac2.c | 96 ++++++++++++++++++++-- drivers/usb/gadget/function/u_uac2.h | 2 + 4 files changed, 91 insertions(+), 9 deletions(-)
Documentation/ABI/testing/configfs-usb-gadget-uac2 +1 -0
@@ -7,6 +7,7 @@ Description: c_chmask - capture channel mask c_srate - capture sampling rate c_ssize - capture sample size (bytes) + c_sync - capture synchronization type (async/adaptive) p_chmask - playback channel mask p_srate - playback sampling rate p_ssize - playback sample size (bytes)
Documentation/usb/gadget-testing.rst +1 -0
@@ -728,6 +728,7 @@ The uac2 function provides these attributes in its function directory: c_chmask capture channel mask c_srate capture sampling rate c_ssize capture sample size (bytes) + c_sync capture synchronization type (async/adaptive) p_chmask playback channel mask p_srate playback sampling rate p_ssize playback sample size (bytes)
drivers/usb/gadget/function/f_uac2.c +87 -9
168 lines not on this page · raw message
drivers/usb/gadget/function/u_uac2.h +2 -0
@@ -21,6 +21,7 @@#define UAC2_DEF_CCHMASK 0x3#define UAC2_DEF_CSRATE 64000#define UAC2_DEF_CSSIZE 2 +#define UAC2_DEF_CSYNC USB_ENDPOINT_SYNC_ASYNC#define UAC2_DEF_REQ_NUM 2struct f_uac2_opts {
@@ -31,6 +32,7 @@ struct f_uac2_opts {int c_chmask;int c_srate;int c_ssize; +int c_sync;int req_number;bool bound;
Signature · 1 line
1.9.1
Re: [PATCH 2/3] usb: gadget: f_uac2: add adaptive sync support for capture
Quoted 33 lines · Ruslan Bilovol
Current
f_uac2USB OUT (aka 'capture') synchronization implements 'ASYNC' scenario which means USB Gadget has it's own freerunning clock and can update Host about real clock frequency through feedback endpoint so Host can align number of samples sent to the USB gadget to prevent overruns/underrunsIn case if Gadget can has no it's internal clock and can consume audio samples at any rate (for example, on the Gadget side someone records audio directly to a file, or audio samples are played through an external DAC as soon as they arrive), UAC2 spec suggests 'ADAPTIVE' synchronization type.
Change UAC2 driver to make it configurable through additional '
c_sync' configfs file.Default remains 'asynchronous' with possibility to switch it to 'adaptive'
Signed-off-by Ruslan Bilovol <[email protected]>--- Documentation/ABI/testing/configfs-usb-gadget-uac2 | 1 + Documentation/usb/gadget-testing.rst | 1 + drivers/usb/gadget/function/f_uac2.c | 96 ++++++++++++++++++++-- drivers/usb/gadget/function/u_uac2.h | 2 + 4 files changed, 91 insertions(+), 9 deletions(-)Documentation/ABI/testing/configfs-usb-gadget-uac2 +0
-0
Documentation/ABI/testing/configfs-usb-gadget-uac2 Description: · line 10
Quoted 4 lines
Documentation/ABI/testing/configfs-usb-gadget-uac2 +1
-0@@ -7,6 +7,7 @@ Description: c_chmask - capture channel mask c_srate - capture sampling rate c_ssize - capture sample size (bytes) + c_sync - capture synchronization type (async/adaptive)
Can't apply it based on the latest code, this file has changed.
Quoted 221 lines
Documentation/ABI/testing/configfs-usb-gadget-uac2 +0
-0p_chmask - playback channel mask p_srate - playback sampling rate p_ssize - playback sample size (bytes)Documentation/usb/gadget-testing.rst +1
-0@@ -728,6 +728,7 @@ The uac2 function provides these attributes in its function directory: c_chmask capture channel mask c_srate capture sampling rate c_ssize capture sample size (bytes) + c_sync capture synchronization type (async/adaptive) p_chmask playback channel mask p_srate playback sampling rate p_ssize playback sample size (bytes)drivers/usb/gadget/function/f_uac2.c +87
-9@@ -41,6 +41,7 @@#define EPIN_EN (_opts) ((_opts)->p_chmask != 0)#define EPOUT_EN (_opts) ((_opts)->c_chmask != 0) +#define EPOUT_FBACK_IN_EN (_opts) ((_opts)->c_sync == USB_ENDPOINT_SYNC_ASYNC)struct f_uac2 {struct g_audio g_audio;@@ -237,7 +238,7 @@ enum { .bDescriptorType = USB_DT_INTERFACE, .bAlternateSetting = 1, - .bNumEndpoints = 2, + .bNumEndpoints = 1, .bInterfaceClass = USB_CLASS_AUDIO, .bInterfaceSubClass = USB_SUBCLASS_AUDIOSTREAMING, .bInterfaceProtocol = UAC_VERSION_2,@@ -270,7 +271,7 @@ enum { .bDescriptorType = USB_DT_ENDPOINT, .bEndpointAddress = USB_DIR_OUT, - .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, + .bmAttributes = USB_ENDPOINT_XFER_ISOC, .wMaxPacketSize = cpu_to_le16(1023), .bInterval = 1, };@@ -279,7 +280,7 @@ enum { .bLength = USB_DT_ENDPOINT_SIZE, .bDescriptorType = USB_DT_ENDPOINT, - .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, + .bmAttributes = USB_ENDPOINT_XFER_ISOC, .wMaxPacketSize = cpu_to_le16(1024), .bInterval = 1, };@@ -540,6 +541,19 @@ static void setup_descriptor(struct f_uac2_opts *opts) len +=sizeof (io_out_ot_desc); ac_hdr_desc.wTotalLength =cpu_to_le16 (len); iad_desc.bInterfaceCount++; + +if (EPOUT_FBACK_IN_EN (opts)) { + fs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ASYNC ; + hs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ASYNC ; + std_as_out_if1_desc.bNumEndpoints++; + }else { + fs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ADAPTIVE ; + hs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ADAPTIVE ; + } } i =0 ;@@ -564,7 +578,8 @@ static void setup_descriptor(struct f_uac2_opts *opts) fs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); fs_audio_desc[i++] =USBDHDR (&fs_epout_desc); fs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); - fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); +if (EPOUT_FBACK_IN_EN (opts)) + fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); }if (EPIN_EN (opts)) { fs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -598,7 +613,8 @@ static void setup_descriptor(struct f_uac2_opts *opts) hs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); hs_audio_desc[i++] =USBDHDR (&hs_epout_desc); hs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); - hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); +if (EPOUT_FBACK_IN_EN (opts)) + hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); }if (EPIN_EN (opts)) { hs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -706,11 +722,14 @@ static void setup_descriptor(struct f_uac2_opts *opts)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__);return -ENODEV ; } - agdev->in_ep_fback =usb_ep_autoconfig (gadget, +if (EPOUT_FBACK_IN_EN (uac2_opts)) { + agdev->in_ep_fback =usb_ep_autoconfig (gadget, &fs_epin_fback_desc); -if (!agdev->in_ep_fback) { -dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); -return -ENODEV ; +if (!agdev->in_ep_fback) { +dev_err (dev,"%s:%d Error!\n" , + __func__, __LINE__); +return -ENODEV ; + } } }@@ -1057,11 +1076,68 @@ static void f_uac2_attr_release(struct config_item *item) \ CONFIGFS_ATTR(f_uac2_opts_, name) +#define UAC2_ATTRIBUTE_SYNC (name) \ +static ssize_t f_uac2_opts_##name##_show(struct config_item *item, \ + char *page) \ +{ \ + struct f_uac2_opts *opts = to_f_uac2_opts(item); \ + int result; \ + char *str; \ + \ + mutex_lock(&opts->lock); \ + switch (opts->name) { \ + case USB_ENDPOINT_SYNC_ASYNC: \ + str = "async"; \ + break; \ + case USB_ENDPOINT_SYNC_ADAPTIVE: \ + str = "adaptive"; \ + break; \ + default: \ + str = "unknown"; \ + break; \ + } \ + result = sprintf(page, "%s\n", str); \ + mutex_unlock(&opts->lock); \ + \ + return result; \ +} \ + \ +static ssize_t f_uac2_opts_##name##_store(struct config_item *item, \ + const char *page, size_t len) \ +{ \ + struct f_uac2_opts *opts = to_f_uac2_opts(item); \ + int ret = 0; \ + \ + mutex_lock(&opts->lock); \ + if (opts->refcnt) { \ + ret = -EBUSY; \ + goto end; \ + } \ + \ + if (!strncmp(page, "async", 5)) \ + opts->name = USB_ENDPOINT_SYNC_ASYNC; \ + else if (!strncmp(page, "adaptive", 8)) \ + opts->name = USB_ENDPOINT_SYNC_ADAPTIVE; \ + else { \ + ret = -EINVAL; \ + goto end; \ + } \ + \ + ret = len; \ + \ +end: \ + mutex_unlock(&opts->lock); \ + return ret; \ +} \ + \ +CONFIGFS_ATTR(f_uac2_opts_, name) +UAC2_ATTRIBUTE (p_chmask);UAC2_ATTRIBUTE (p_srate);UAC2_ATTRIBUTE (p_ssize);UAC2_ATTRIBUTE (c_chmask);UAC2_ATTRIBUTE (c_srate); +UAC2_ATTRIBUTE_SYNC (c_sync);UAC2_ATTRIBUTE (c_ssize);UAC2_ATTRIBUTE (req_number);@@ -1072,6 +1148,7 @@ static void f_uac2_attr_release(struct config_item *item) &f_uac2_opts_attr_c_chmask, &f_uac2_opts_attr_c_srate, &f_uac2_opts_attr_c_ssize, + &f_uac2_opts_attr_c_sync, &f_uac2_opts_attr_req_number, NULL, };@@ -1110,6 +1187,7 @@ static struct usb_function_instance *afunc_alloc_inst(void) opts->c_chmask =UAC2_DEF_CCHMASK ; opts->c_srate =UAC2_DEF_CSRATE ; opts->c_ssize =UAC2_DEF_CSSIZE ; + opts->c_sync =UAC2_DEF_CSYNC ; opts->req_number =UAC2_DEF_REQ_NUM ;return &opts->func_inst; }drivers/usb/gadget/function/u_uac2.h +2
-0@@ -21,6 +21,7 @@#define UAC2_DEF_CCHMASK 0x3#define UAC2_DEF_CSRATE 64000#define UAC2_DEF_CSSIZE 2 +#define UAC2_DEF_CSYNC USB_ENDPOINT_SYNC_ASYNC#define UAC2_DEF_REQ_NUM 2struct f_uac2_opts {@@ -31,6 +32,7 @@ struct f_uac2_opts {int c_chmask;int c_srate;int c_ssize; +int c_sync;int req_number;bool bound;Signature · 1 line
1.9.1
Signature · 2 lines
Thanks, Peter Chen
Re: [PATCH 2/3] usb: gadget: f_uac2: add adaptive sync support for capture
Quoted 30 lines · Peter Chen
Quoted 28 lines · Ruslan Bilovol
Current
f_uac2USB OUT (aka 'capture') synchronization implements 'ASYNC' scenario which means USB Gadget has it's own freerunning clock and can update Host about real clock frequency through feedback endpoint so Host can align number of samples sent to the USB gadget to prevent overruns/underrunsIn case if Gadget can has no it's internal clock and can consume audio samples at any rate (for example, on the Gadget side someone records audio directly to a file, or audio samples are played through an external DAC as soon as they arrive), UAC2 spec suggests 'ADAPTIVE' synchronization type.
Change UAC2 driver to make it configurable through additional '
c_sync' configfs file.Default remains 'asynchronous' with possibility to switch it to 'adaptive'
Signed-off-by Ruslan Bilovol <[email protected]>--- Documentation/ABI/testing/configfs-usb-gadget-uac2 | 1 + Documentation/usb/gadget-testing.rst | 1 + drivers/usb/gadget/function/f_uac2.c | 96 ++++++++++++++++++++-- drivers/usb/gadget/function/u_uac2.h | 2 + 4 files changed, 91 insertions(+), 9 deletions(-)
Quoted 11 lines
Documentation/ABI/testing/configfs-usb-gadget-uac2 Description: · line 10
Quoted 9 lines
Documentation/ABI/testing/configfs-usb-gadget-uac2 +1
-0@@ -7,6 +7,7 @@ Description: c_chmask - capture channel mask c_srate - capture sampling rate c_ssize - capture sample size (bytes) + c_sync - capture synchronization type (async/adaptive)Can't apply it based on the latest code, this file has changed.
This patch set is based on Felipe's -next branch which seems to be a little bit outdated.
Thanks,
Ruslan
Re: [PATCH 2/3] usb: gadget: f_uac2: add adaptive sync support for capture
In reply to Ruslan BilovolQuoted 98 lines · Ruslan Bilovol
Current
f_uac2USB OUT (aka 'capture') synchronization implements 'ASYNC' scenario which means USB Gadget has it's own freerunning clock and can update Host about real clock frequency through feedback endpoint so Host can align number of samples sent to the USB gadget to prevent overruns/underrunsIn case if Gadget can has no it's internal clock and can consume audio samples at any rate (for example, on the Gadget side someone records audio directly to a file, or audio samples are played through an external DAC as soon as they arrive), UAC2 spec suggests 'ADAPTIVE' synchronization type.
Change UAC2 driver to make it configurable through additional '
c_sync' configfs file.Default remains 'asynchronous' with possibility to switch it to 'adaptive'
Signed-off-by Ruslan Bilovol <[email protected]>--- Documentation/ABI/testing/configfs-usb-gadget-uac2 | 1 + Documentation/usb/gadget-testing.rst | 1 + drivers/usb/gadget/function/f_uac2.c | 96 ++++++++++++++++++++-- drivers/usb/gadget/function/u_uac2.h | 2 + 4 files changed, 91 insertions(+), 9 deletions(-)Documentation/ABI/testing/configfs-usb-gadget-uac2 +1
-0@@ -7,6 +7,7 @@ Description: c_chmask - capture channel mask c_srate - capture sampling rate c_ssize - capture sample size (bytes) + c_sync - capture synchronization type (async/adaptive) p_chmask - playback channel mask p_srate - playback sampling rate p_ssize - playback sample size (bytes)Documentation/usb/gadget-testing.rst +1
-0@@ -728,6 +728,7 @@ The uac2 function provides these attributes in its function directory: c_chmask capture channel mask c_srate capture sampling rate c_ssize capture sample size (bytes) + c_sync capture synchronization type (async/adaptive) p_chmask playback channel mask p_srate playback sampling rate p_ssize playback sample size (bytes)drivers/usb/gadget/function/f_uac2.c +7
-3@@ -41,6 +41,7 @@#define EPIN_EN (_opts) ((_opts)->p_chmask != 0)#define EPOUT_EN (_opts) ((_opts)->c_chmask != 0) +#define EPOUT_FBACK_IN_EN (_opts) ((_opts)->c_sync == USB_ENDPOINT_SYNC_ASYNC)struct f_uac2 {struct g_audio g_audio;@@ -237,7 +238,7 @@ enum { .bDescriptorType = USB_DT_INTERFACE, .bAlternateSetting = 1, - .bNumEndpoints = 2, + .bNumEndpoints = 1, .bInterfaceClass = USB_CLASS_AUDIO, .bInterfaceSubClass = USB_SUBCLASS_AUDIOSTREAMING, .bInterfaceProtocol = UAC_VERSION_2,@@ -270,7 +271,7 @@ enum { .bDescriptorType = USB_DT_ENDPOINT, .bEndpointAddress = USB_DIR_OUT, - .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, + .bmAttributes = USB_ENDPOINT_XFER_ISOC, .wMaxPacketSize = cpu_to_le16(1023), .bInterval = 1, };@@ -279,7 +280,7 @@ enum { .bLength = USB_DT_ENDPOINT_SIZE, .bDescriptorType = USB_DT_ENDPOINT, - .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, + .bmAttributes = USB_ENDPOINT_XFER_ISOC, .wMaxPacketSize = cpu_to_le16(1024), .bInterval = 1, };@@ -540,6 +541,19 @@ static void setup_descriptor(struct f_uac2_opts *opts) len +=sizeof (io_out_ot_desc); ac_hdr_desc.wTotalLength =cpu_to_le16 (len); iad_desc.bInterfaceCount++; + +if (EPOUT_FBACK_IN_EN (opts)) { + fs_epout_desc.bmAttributes |=
drivers/usb/gadget/function/f_uac2.c static void setup_descriptor(struct f_uac2_opts *opts) · line 556
Quoted 10 lines
drivers/usb/gadget/function/f_uac2.c +10
-0+USB_ENDPOINT_SYNC_ASYNC ; + hs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ASYNC ; + std_as_out_if1_desc.bNumEndpoints++; + }else { + fs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ADAPTIVE ; + hs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ADAPTIVE ; + }
Hi Ruslan,
First, thanks a lot for this series, it is very helpful
Instead of amending the descriptors, could you consider using comments like
/* .bNumEndpoints = DYNAMIC */
or
/* .bmAttributes = DYNAMIC */
It would help understand that the actual value of these parameters will be determine at runtime, making the code easier to follow and maintain.
Also, I wonder if it would be difficult to add implicit feedback support while at it ?
I'm asking this now because it could (possibly) change the driver params (implicit is async as well)
Quoted 146 lines
drivers/usb/gadget/function/f_uac2.c +70
-6} i =0 ;@@ -564,7 +578,8 @@ static void setup_descriptor(struct f_uac2_opts *opts) fs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); fs_audio_desc[i++] =USBDHDR (&fs_epout_desc); fs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); - fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); +if (EPOUT_FBACK_IN_EN (opts)) + fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); }if (EPIN_EN (opts)) { fs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -598,7 +613,8 @@ static void setup_descriptor(struct f_uac2_opts *opts) hs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); hs_audio_desc[i++] =USBDHDR (&hs_epout_desc); hs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); - hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); +if (EPOUT_FBACK_IN_EN (opts)) + hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); }if (EPIN_EN (opts)) { hs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -706,11 +722,14 @@ static void setup_descriptor(struct f_uac2_opts *opts)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__);return -ENODEV ; } - agdev->in_ep_fback =usb_ep_autoconfig (gadget, +if (EPOUT_FBACK_IN_EN (uac2_opts)) { + agdev->in_ep_fback =usb_ep_autoconfig (gadget, &fs_epin_fback_desc); -if (!agdev->in_ep_fback) { -dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); -return -ENODEV ; +if (!agdev->in_ep_fback) { +dev_err (dev,"%s:%d Error!\n" , + __func__, __LINE__); +return -ENODEV ; + } } }@@ -1057,11 +1076,68 @@ static void f_uac2_attr_release(struct config_item *item) \ CONFIGFS_ATTR(f_uac2_opts_, name) +#define UAC2_ATTRIBUTE_SYNC (name) \ +static ssize_t f_uac2_opts_##name##_show(struct config_item *item, \ + char *page) \ +{ \ + struct f_uac2_opts *opts = to_f_uac2_opts(item); \ + int result; \ + char *str; \ + \ + mutex_lock(&opts->lock); \ + switch (opts->name) { \ + case USB_ENDPOINT_SYNC_ASYNC: \ + str = "async"; \ + break; \ + case USB_ENDPOINT_SYNC_ADAPTIVE: \ + str = "adaptive"; \ + break; \ + default: \ + str = "unknown"; \ + break; \ + } \ + result = sprintf(page, "%s\n", str); \ + mutex_unlock(&opts->lock); \ + \ + return result; \ +} \ + \ +static ssize_t f_uac2_opts_##name##_store(struct config_item *item, \ + const char *page, size_t len) \ +{ \ + struct f_uac2_opts *opts = to_f_uac2_opts(item); \ + int ret = 0; \ + \ + mutex_lock(&opts->lock); \ + if (opts->refcnt) { \ + ret = -EBUSY; \ + goto end; \ + } \ + \ + if (!strncmp(page, "async", 5)) \ + opts->name = USB_ENDPOINT_SYNC_ASYNC; \ + else if (!strncmp(page, "adaptive", 8)) \ + opts->name = USB_ENDPOINT_SYNC_ADAPTIVE; \ + else { \ + ret = -EINVAL; \ + goto end; \ + } \ + \ + ret = len; \ + \ +end: \ + mutex_unlock(&opts->lock); \ + return ret; \ +} \ + \ +CONFIGFS_ATTR(f_uac2_opts_, name) +UAC2_ATTRIBUTE (p_chmask);UAC2_ATTRIBUTE (p_srate);UAC2_ATTRIBUTE (p_ssize);UAC2_ATTRIBUTE (c_chmask);UAC2_ATTRIBUTE (c_srate); +UAC2_ATTRIBUTE_SYNC (c_sync);UAC2_ATTRIBUTE (c_ssize);UAC2_ATTRIBUTE (req_number);@@ -1072,6 +1148,7 @@ static void f_uac2_attr_release(struct config_item *item) &f_uac2_opts_attr_c_chmask, &f_uac2_opts_attr_c_srate, &f_uac2_opts_attr_c_ssize, + &f_uac2_opts_attr_c_sync, &f_uac2_opts_attr_req_number, NULL, };@@ -1110,6 +1187,7 @@ static struct usb_function_instance *afunc_alloc_inst(void) opts->c_chmask =UAC2_DEF_CCHMASK ; opts->c_srate =UAC2_DEF_CSRATE ; opts->c_ssize =UAC2_DEF_CSSIZE ; + opts->c_sync =UAC2_DEF_CSYNC ; opts->req_number =UAC2_DEF_REQ_NUM ;return &opts->func_inst; }drivers/usb/gadget/function/u_uac2.h +2
-0@@ -21,6 +21,7 @@#define UAC2_DEF_CCHMASK 0x3#define UAC2_DEF_CSRATE 64000#define UAC2_DEF_CSSIZE 2 +#define UAC2_DEF_CSYNC USB_ENDPOINT_SYNC_ASYNC#define UAC2_DEF_REQ_NUM 2struct f_uac2_opts {@@ -31,6 +32,7 @@ struct f_uac2_opts {int c_chmask;int c_srate;int c_ssize; +int c_sync;int req_number;bool bound;
Re: [PATCH 2/3] usb: gadget: f_uac2: add adaptive sync support for capture
Quoted 117 lines · Jerome Brunet
Quoted 98 lines · Ruslan Bilovol
Current
f_uac2USB OUT (aka 'capture') synchronization implements 'ASYNC' scenario which means USB Gadget has it's own freerunning clock and can update Host about real clock frequency through feedback endpoint so Host can align number of samples sent to the USB gadget to prevent overruns/underrunsIn case if Gadget can has no it's internal clock and can consume audio samples at any rate (for example, on the Gadget side someone records audio directly to a file, or audio samples are played through an external DAC as soon as they arrive), UAC2 spec suggests 'ADAPTIVE' synchronization type.
Change UAC2 driver to make it configurable through additional '
c_sync' configfs file.Default remains 'asynchronous' with possibility to switch it to 'adaptive'
Signed-off-by Ruslan Bilovol <[email protected]>--- Documentation/ABI/testing/configfs-usb-gadget-uac2 | 1 + Documentation/usb/gadget-testing.rst | 1 + drivers/usb/gadget/function/f_uac2.c | 96 ++++++++++++++++++++-- drivers/usb/gadget/function/u_uac2.h | 2 + 4 files changed, 91 insertions(+), 9 deletions(-)Documentation/ABI/testing/configfs-usb-gadget-uac2 +1
-0@@ -7,6 +7,7 @@ Description: c_chmask - capture channel mask c_srate - capture sampling rate c_ssize - capture sample size (bytes) + c_sync - capture synchronization type (async/adaptive) p_chmask - playback channel mask p_srate - playback sampling rate p_ssize - playback sample size (bytes)Documentation/usb/gadget-testing.rst +1
-0@@ -728,6 +728,7 @@ The uac2 function provides these attributes in its function directory: c_chmask capture channel mask c_srate capture sampling rate c_ssize capture sample size (bytes) + c_sync capture synchronization type (async/adaptive) p_chmask playback channel mask p_srate playback sampling rate p_ssize playback sample size (bytes)drivers/usb/gadget/function/f_uac2.c +7
-3@@ -41,6 +41,7 @@#define EPIN_EN (_opts) ((_opts)->p_chmask != 0)#define EPOUT_EN (_opts) ((_opts)->c_chmask != 0) +#define EPOUT_FBACK_IN_EN (_opts) ((_opts)->c_sync == USB_ENDPOINT_SYNC_ASYNC)struct f_uac2 {struct g_audio g_audio;@@ -237,7 +238,7 @@ enum { .bDescriptorType = USB_DT_INTERFACE, .bAlternateSetting = 1, - .bNumEndpoints = 2, + .bNumEndpoints = 1, .bInterfaceClass = USB_CLASS_AUDIO, .bInterfaceSubClass = USB_SUBCLASS_AUDIOSTREAMING, .bInterfaceProtocol = UAC_VERSION_2,@@ -270,7 +271,7 @@ enum { .bDescriptorType = USB_DT_ENDPOINT, .bEndpointAddress = USB_DIR_OUT, - .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, + .bmAttributes = USB_ENDPOINT_XFER_ISOC, .wMaxPacketSize = cpu_to_le16(1023), .bInterval = 1, };@@ -279,7 +280,7 @@ enum { .bLength = USB_DT_ENDPOINT_SIZE, .bDescriptorType = USB_DT_ENDPOINT, - .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, + .bmAttributes = USB_ENDPOINT_XFER_ISOC, .wMaxPacketSize = cpu_to_le16(1024), .bInterval = 1, };@@ -540,6 +541,19 @@ static void setup_descriptor(struct f_uac2_opts *opts) len +=sizeof (io_out_ot_desc); ac_hdr_desc.wTotalLength =cpu_to_le16 (len); iad_desc.bInterfaceCount++; + +if (EPOUT_FBACK_IN_EN (opts)) { + fs_epout_desc.bmAttributes |=drivers/usb/gadget/function/f_uac2.c static void setup_descriptor(struct f_uac2_opts *opts) · line 556
Quoted 10 lines
drivers/usb/gadget/function/f_uac2.c +10
-0+USB_ENDPOINT_SYNC_ASYNC ; + hs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ASYNC ; + std_as_out_if1_desc.bNumEndpoints++; + }else { + fs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ADAPTIVE ; + hs_epout_desc.bmAttributes |= +USB_ENDPOINT_SYNC_ADAPTIVE ; + }Hi Ruslan,
First, thanks a lot for this series, it is very helpful
Quoted 7 lines
Instead of amending the descriptors, could you consider using comments like
/* .bNumEndpoints = DYNAMIC */
or
/* .bmAttributes = DYNAMIC */
Agree, I just forgot to do that
Quoted 9 lines
It would help understand that the actual value of these parameters will be determine at runtime, making the code easier to follow and maintain.
Also, I wonder if it would be difficult to add implicit feedback support while at it ?
I'm asking this now because it could (possibly) change the driver params (implicit is async as well)
It seems it should be quite easy to add it, so at least worth a try. Will experiment this in next patch set
Thanks,
Ruslan
Quoted 148 lines
Quoted 147 lines
drivers/usb/gadget/function/f_uac2.c +70
-6} i =0 ;@@ -564,7 +578,8 @@ static void setup_descriptor(struct f_uac2_opts *opts) fs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); fs_audio_desc[i++] =USBDHDR (&fs_epout_desc); fs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); - fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); +if (EPOUT_FBACK_IN_EN (opts)) + fs_audio_desc[i++] =USBDHDR (&fs_epin_fback_desc); }if (EPIN_EN (opts)) { fs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -598,7 +613,8 @@ static void setup_descriptor(struct f_uac2_opts *opts) hs_audio_desc[i++] =USBDHDR (&as_out_fmt1_desc); hs_audio_desc[i++] =USBDHDR (&hs_epout_desc); hs_audio_desc[i++] =USBDHDR (&as_iso_out_desc); - hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); +if (EPOUT_FBACK_IN_EN (opts)) + hs_audio_desc[i++] =USBDHDR (&hs_epin_fback_desc); }if (EPIN_EN (opts)) { hs_audio_desc[i++] =USBDHDR (&std_as_in_if0_desc);@@ -706,11 +722,14 @@ static void setup_descriptor(struct f_uac2_opts *opts)dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__);return -ENODEV ; } - agdev->in_ep_fback =usb_ep_autoconfig (gadget, +if (EPOUT_FBACK_IN_EN (uac2_opts)) { + agdev->in_ep_fback =usb_ep_autoconfig (gadget, &fs_epin_fback_desc); -if (!agdev->in_ep_fback) { -dev_err (dev,"%s:%d Error!\n" , __func__, __LINE__); -return -ENODEV ; +if (!agdev->in_ep_fback) { +dev_err (dev,"%s:%d Error!\n" , + __func__, __LINE__); +return -ENODEV ; + } } }@@ -1057,11 +1076,68 @@ static void f_uac2_attr_release(struct config_item *item) \ CONFIGFS_ATTR(f_uac2_opts_, name) +#define UAC2_ATTRIBUTE_SYNC (name) \ +static ssize_t f_uac2_opts_##name##_show(struct config_item *item, \ + char *page) \ +{ \ + struct f_uac2_opts *opts = to_f_uac2_opts(item); \ + int result; \ + char *str; \ + \ + mutex_lock(&opts->lock); \ + switch (opts->name) { \ + case USB_ENDPOINT_SYNC_ASYNC: \ + str = "async"; \ + break; \ + case USB_ENDPOINT_SYNC_ADAPTIVE: \ + str = "adaptive"; \ + break; \ + default: \ + str = "unknown"; \ + break; \ + } \ + result = sprintf(page, "%s\n", str); \ + mutex_unlock(&opts->lock); \ + \ + return result; \ +} \ + \ +static ssize_t f_uac2_opts_##name##_store(struct config_item *item, \ + const char *page, size_t len) \ +{ \ + struct f_uac2_opts *opts = to_f_uac2_opts(item); \ + int ret = 0; \ + \ + mutex_lock(&opts->lock); \ + if (opts->refcnt) { \ + ret = -EBUSY; \ + goto end; \ + } \ + \ + if (!strncmp(page, "async", 5)) \ + opts->name = USB_ENDPOINT_SYNC_ASYNC; \ + else if (!strncmp(page, "adaptive", 8)) \ + opts->name = USB_ENDPOINT_SYNC_ADAPTIVE; \ + else { \ + ret = -EINVAL; \ + goto end; \ + } \ + \ + ret = len; \ + \ +end: \ + mutex_unlock(&opts->lock); \ + return ret; \ +} \ + \ +CONFIGFS_ATTR(f_uac2_opts_, name) +UAC2_ATTRIBUTE (p_chmask);UAC2_ATTRIBUTE (p_srate);UAC2_ATTRIBUTE (p_ssize);UAC2_ATTRIBUTE (c_chmask);UAC2_ATTRIBUTE (c_srate); +UAC2_ATTRIBUTE_SYNC (c_sync);UAC2_ATTRIBUTE (c_ssize);UAC2_ATTRIBUTE (req_number);@@ -1072,6 +1148,7 @@ static void f_uac2_attr_release(struct config_item *item) &f_uac2_opts_attr_c_chmask, &f_uac2_opts_attr_c_srate, &f_uac2_opts_attr_c_ssize, + &f_uac2_opts_attr_c_sync, &f_uac2_opts_attr_req_number, NULL, };@@ -1110,6 +1187,7 @@ static struct usb_function_instance *afunc_alloc_inst(void) opts->c_chmask =UAC2_DEF_CCHMASK ; opts->c_srate =UAC2_DEF_CSRATE ; opts->c_ssize =UAC2_DEF_CSSIZE ; + opts->c_sync =UAC2_DEF_CSYNC ; opts->req_number =UAC2_DEF_REQ_NUM ;return &opts->func_inst; }drivers/usb/gadget/function/u_uac2.h +2
-0@@ -21,6 +21,7 @@#define UAC2_DEF_CCHMASK 0x3#define UAC2_DEF_CSRATE 64000#define UAC2_DEF_CSSIZE 2 +#define UAC2_DEF_CSYNC USB_ENDPOINT_SYNC_ASYNC#define UAC2_DEF_REQ_NUM 2struct f_uac2_opts {@@ -31,6 +32,7 @@ struct f_uac2_opts {int c_chmask;int c_srate;int c_ssize; +int c_sync;int req_number;bool bound;
PATCH 3/3 usb: gadget: u_audio: add real feedback implementation
In reply to Ruslan BilovolThis adds interface between userspace and feedback endpoint to report real feedback frequency to the Host.
Current implementation adds new userspace interface ALSA mixer control "PCM Feedback Frequency Hz" (similar to aloop driver's "PCM Rate Shift 100000" mixer control)
We allow +/-20% deviation of nominal sampling frequency, that usually is more than enough in real-world usecases
Usage of this new control is easy to implement in existing userspace tools like alsaloop from alsa-utils.
--- drivers/usb/gadget/function/f_uac2.c | 4 ++ drivers/usb/gadget/function/u_audio.c | 93 +++++++++++++++++++++++++++++++++++ drivers/usb/gadget/function/u_audio.h | 7 +++ 3 files changed, 104 insertions(+)
drivers/usb/gadget/function/f_uac2.c +4 -0
@@ -487,6 +487,10 @@ static void set_ep_max_packet_size(const struct f_uac2_opts *uac2_opts, max_packet_size =num_channels (chmask) * ssize *DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + +if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC )) + max_packet_size = max_packet_size *FBACK_FREQ_MAX /100 ; + ep_desc->wMaxPacketSize =cpu_to_le16 (min_t (u16, max_packet_size,le16_to_cpu (ep_desc->wMaxPacketSize))); }
drivers/usb/gadget/function/u_audio.c +93 -0
117 lines not on this page · raw message
drivers/usb/gadget/function/u_audio.h +7 -0
@@ -11,6 +11,13 @@#include <linux/usb/composite.h> +/* +* Min/max percentage of nominal sampling frequency deviation +* reported through feedback endpoint to the host +*/ +#define FBACK_FREQ_MIN 80 +#define FBACK_FREQ_MAX 120 +struct uac_params {/* playback */ int p_chmask;/* channel mask */
Signature · 1 line
1.9.1
Re: [PATCH 3/3] usb: gadget: u_audio: add real feedback implementation
Hi Bilovol,
Dne 08. 11. 20 v 1:18 Ruslan Bilovol napsal(a):
Quoted 13 lines
This adds interface between userspace and feedback endpoint to report real feedback frequency to the Host.
Current implementation adds new userspace interface ALSA mixer control "PCM Feedback Frequency Hz" (similar to aloop driver's "PCM Rate Shift 100000" mixer control)
We allow +/-20% deviation of nominal sampling frequency, that usually is more than enough in real-world usecases
Usage of this new control is easy to implement in existing userspace tools like alsaloop from alsa-utils.
Quoted 6 lines
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++ drivers/usb/gadget/function/u_audio.c | 93 +++++++++++++++++++++++++++++++++++ drivers/usb/gadget/function/u_audio.h | 7 +++ 3 files changed, 104 insertions(+)
Thanks a lot for the great implementation. IIUC the control element sets integer frequency in Hz. Often the clocks deviate by small fractions of Hz. Please have you considered the value to be e.g. in 100th of Hz for finer control of the samplerate? Similar to the PCM Rate Shift which has a step 100000th of the samplerate.
Thanks a lot.
Best regards,
Pavel.
Re: [PATCH 3/3] usb: gadget: u_audio: add real feedback implementation
Hi Ruslan, please excuse my stupid error, I really apologize...
Dne 09. 11. 20 v 9:24 Pavel Hofman napsal(a):
Quoted 35 lines
Hi Bilovol,
Dne 08. 11. 20 v 1:18 Ruslan Bilovol napsal(a):
Quoted 13 lines
This adds interface between userspace and feedback endpoint to report real feedback frequency to the Host.
Current implementation adds new userspace interface ALSA mixer control "PCM Feedback Frequency Hz" (similar to aloop driver's "PCM Rate Shift 100000" mixer control)
We allow +/-20% deviation of nominal sampling frequency, that usually is more than enough in real-world usecases
Usage of this new control is easy to implement in existing userspace tools like alsaloop from alsa-utils.
Quoted 6 lines
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++ drivers/usb/gadget/function/u_audio.c | 93 +++++++++++++++++++++++++++++++++++ drivers/usb/gadget/function/u_audio.h | 7 +++ 3 files changed, 104 insertions(+)Thanks a lot for the great implementation. IIUC the control element sets integer frequency in Hz. Often the clocks deviate by small fractions of Hz. Please have you considered the value to be e.g. in 100th of Hz for finer control of the samplerate? Similar to the PCM Rate Shift which has a step 100000th of the samplerate.
Thanks a lot.
Best regards,
Pavel.
Re: [PATCH 3/3] usb: gadget: u_audio: add real feedback implementation
In reply to Pavel HofmanDne 09. 11. 20 v 9:24 Pavel Hofman napsal(a):
Quoted 23 lines
Hi Ruslan,
Dne 08. 11. 20 v 1:18 Ruslan Bilovol napsal(a):
Quoted 13 lines
This adds interface between userspace and feedback endpoint to report real feedback frequency to the Host.
Current implementation adds new userspace interface ALSA mixer control "PCM Feedback Frequency Hz" (similar to aloop driver's "PCM Rate Shift 100000" mixer control)
We allow +/-20% deviation of nominal sampling frequency, that usually is more than enough in real-world usecases
Usage of this new control is easy to implement in existing userspace tools like alsaloop from alsa-utils.
Quoted 6 lines
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++ drivers/usb/gadget/function/u_audio.c | 93 +++++++++++++++++++++++++++++++++++ drivers/usb/gadget/function/u_audio.h | 7 +++ 3 files changed, 104 insertions(+)
Quoted 6 lines
Thanks a lot for the great implementation. IIUC the control element sets integer frequency in Hz. Often the clocks deviate by small fractions of Hz. Please have you considered the value to be e.g. in 100th of Hz for finer control of the samplerate? Similar to the PCM Rate Shift which has a step 100000th of the samplerate.
My stupid, I did not realize that one Hz in the samplerate value is basically the same order of magnitude as the 100,000ths in the "PCM Rate Shift 100000" mixer control. Sorry for disturbing.
Thanks,
Pavel.
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementation
Hi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmask
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed > /sys/kernel/config/usb_gadget/g1/max_speed fi
Signature · 2 lines
Thanks, Peter Chen
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmask
Quoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfi
Hmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0
# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_srate
mkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo 0x0101 > idProduct
echo 0x1d6b > idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------
Thanks,
Ruslan
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------
Quoted 2 lines
Thanks,
Ruslan
Hi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
Thanks,
Glenn
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settings
Please test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From 51516435bbf2486574ec7bc9fd4726677cd474a4 Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget: f_uac2: always increase endpoint max_packet_size
by one audio slot
As per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint max_packet_size calculation code was adding sometimes +1 audio slot due to DIV_ROUND_UP behaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.
It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
(quoted diff) +0 -0
b/drivers/usb/gadget/function/f_uac2.c
index ebdb2a1..1698587 100644
drivers/usb/gadget/function/f_uac2.c +0 -0
@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const struct
f_uac2_opts *uac2_opts,
}
max_packet_size = num_channels(chmask) * ssize *
- DIV_ROUND_UP(srate, factor / (1 << (ep_desc->bInterval - 1)));
+ ((srate / (factor / (1 << (ep_desc->bInterval - 1)))) + 1);
if (!is_playback && (uac2_opts->c_sync == USB_ENDPOINT_SYNC_ASYNC)) max_packet_size = max_packet_size * FBACK_FREQ_MAX / 100;
Signature · 24 lines
1.9.1 ---------------------------------------------8<---------------------------------------- From c8f2f2b414af672ec40841e75fb1ea761ae29122 Mon Sep 17 00:00:00 2001 From: Ruslan Bilovol <[email protected]> Date: Sun, 16 Feb 2020 22:40:23 +0200 Subject: [PATCH] usb: gadget: f_uac2: change bInterval to 1 for 8ch/24bit/44.1kHz With bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time. USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1. In the future, bInterval should be dynamically calculated depending on bandwith requirementes Signed-off-by: Ruslan Bilovol <[email protected]> --- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)
(quoted diff) +0 -0
b/drivers/usb/gadget/function/f_uac2.c
index 3633df6..fb9b875 100644
drivers/usb/gadget/function/f_uac2.c +2 -2
@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */
@@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */
Signature · 7 lines
1.9.1 ---------------------------------------------8<---------------------------------------- Thanks, Ruslan
Quoted 237 lines · Ruslan Bilovol
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settingsPlease test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From
51516435bbf2486574ec7bc9fd4726677cd474a4Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget:f_uac2: always increase endpointmax_packet_size
by one audio slotAs per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint
max_packet_sizecalculation code was adding sometimes +1 audio slot due toDIV_ROUND_UPbehaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/usb/gadget/function/f_uac2.c +1
-1@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const structf_uac2_opts *uac2_opts, } max_packet_size =num_channels (chmask) * ssize * -DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + ((srate / (factor / (1 << (ep_desc->bInterval -1 )))) +1 );if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC ))
max_packet_size=max_packet_size*FBACK_FREQ_MAX/ 100;
--
1.9.1---------------------------------------------8<---------------------------------------- From
c8f2f2b414af672ec40841e75fb1ea761ae29122Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 16 Feb 2020 22:40:23 +0200
Subject: [PATCH] usb: gadget:f_uac2: change bInterval to 1 for
8ch/24bit/44.1kHzWith bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time.
USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1.In the future, bInterval should be dynamically calculated depending on bandwith requirementes
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/usb/gadget/function/f_uac2.c +2
-3@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */ @@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */ --1.9.1
Quoted 4 lines
---------------------------------------------8<----------------------------------------
Thanks,
Ruslan
Hi Ruslan -
I have had some mixed success with your additional patches. First, the UAC2 gadget does correctly register with Windows 10 but only for 1-31 channels. For a fully specified channel map (32 channels):
c_chmask = 0xFFFFFFFF ( Does not load)p_chmask = 0xFFFFFFFF ( Does not load)
the driver fails to load (code=10). This seems similar to the prior problem. Could there be another off-by-one problem here?
With a 31 channel mask the Windows 10 usbaudio2.sys loads successfully, e.g.
c_chmask = 0x7FFFFFFF (loads)p_chmask = 0x7FFFFFFF (loads)
Once Windows 10 loads the driver successfully, however, I can't seem to select the device through Audacity on Windows. The device appears in the Windows "Device Manager" under "Sound, video and game controllers -> Source/Sink". How are you testing audio playback/capture under Windows? Perhaps my test approach is wrong?
I used the same approach on my Linux host where the USB UAC2 device appears as an ALSA device and I use speaker-test on the target (BeagleBone Black) to play and then capture the audio using Audacity on the Linux host. I was hoping to do the same or something similar under Windows.
Below is the script I've been using on the target to configure my UAC2 device:
mkdir -p "/sys/kernel/config/usb_gadget/g_multi"
echo "${idVendor:-0x11F5}" > "/sys/kernel/config/usb_gadget/g_multi/idVendor" echo "${idProduct:-0x0009}" > "/sys/kernel/config/usb_gadget/g_multi/idProduct" echo "${bcdDevice:-0x0001}" > "/sys/kernel/config/usb_gadget/g_multi/bcdDevice" echo "${bDeviceClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceClass"
echo "${bDeviceSubClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceSubClass"
echo "${bDeviceProtocol:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceProtocol"
echo "${bcdUSB:-0x0200}" > "/sys/kernel/config/usb_gadget/g_multi/bcdUSB" echo "${bMaxPacketSize0:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bMaxPacketSize0"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/strings/0x409" echo "${serialnumber:-000000000000}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/serialnumber" echo "${product:-UAC2}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/product"
echo "${manufacturer:-www.acme.com}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/manufacturer" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" echo "${req_number:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/req_number" echo "${c_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_ssize" echo "${c_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_srate" echo "${c_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_chmask" echo "${c_sync:-asynchronous}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_sync" echo "${p_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_ssize" echo "${p_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_srate" echo "${p_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_chmask" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1"
echo "${bmAttributes:-0x80}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/bmAttributes" echo "${MaxPower:-500}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/MaxPower"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409" echo "${configuration:-BeagleBone Composite}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409/configuration" ln -s "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/uac2.usb0" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/os_desc"
ln -s "//sys/kernel/config/usb_gadget/g_multi/configs/c.1" "/sys/kernel/config/usb_gadget/g_multi/os_desc/c.1"
echo "${qw_sign:-MSFT100}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/qw_sign"
echo "${b_vendor_code:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/b_vendor_code"
echo "${use:-1}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/use"
basename /sys/class/udc/* > /sys/kernel/config/usb_gadget/g_multi/UDC
This is only a two channel playback/capture configuration. I hoped to get this working under Windows before increasing my channel count to a full 32 channel mask with a 48K sample rate.
I look forward to hearing your suggestions and possibly pointing out where I've gone astray.
Thanks,
Glenn
Quoted 239 lines · Glenn Schmottlach
Quoted 237 lines · Ruslan Bilovol
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settingsPlease test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From
51516435bbf2486574ec7bc9fd4726677cd474a4Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget:f_uac2: always increase endpointmax_packet_size
by one audio slotAs per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint
max_packet_sizecalculation code was adding sometimes +1 audio slot due toDIV_ROUND_UPbehaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/usb/gadget/function/f_uac2.c +1
-1@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const structf_uac2_opts *uac2_opts, } max_packet_size =num_channels (chmask) * ssize * -DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + ((srate / (factor / (1 << (ep_desc->bInterval -1 )))) +1 );if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC ))
max_packet_size=max_packet_size*FBACK_FREQ_MAX/ 100;
--
1.9.1---------------------------------------------8<---------------------------------------- From
c8f2f2b414af672ec40841e75fb1ea761ae29122Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 16 Feb 2020 22:40:23 +0200
Subject: [PATCH] usb: gadget:f_uac2: change bInterval to 1 for
8ch/24bit/44.1kHzWith bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time.
USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1.In the future, bInterval should be dynamically calculated depending on bandwith requirementes
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/usb/gadget/function/f_uac2.c +2
-3@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */ @@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */ --1.9.1
Quoted 8 lines
Quoted 4 lines
---------------------------------------------8<----------------------------------------
Thanks,
RuslanHi Ruslan -
I have had some mixed success with your additional patches. First, the
That's good news, thanks for testing!
Quoted 6 lines
UAC2 gadget does correctly register with Windows 10 but only for 1-31 channels. For a fully specified channel map (32 channels):
c_chmask=0xFFFFFFFF( Does not load)p_chmask=0xFFFFFFFF( Does not load)
Quoted 7 lines
the driver fails to load (code=10). This seems similar to the prior problem. Could there be another off-by-one problem here?
With a 31 channel mask the Windows 10 usbaudio2.sys loads successfully, e.g.
c_chmask=0x7FFFFFFF(loads)p_chmask=0x7FFFFFFF(loads)
That's correct behavior. UAC2 spec says that bits D27..D30 are reserved. Also bit D31 is mutually exclusive with other bits and means 'raw data'. It means that with current UAC2 gadget implementation you can enable only 27 channels.
It seems like it's possible to add more channels by adding more audio clusters but that requires UAC2 driver modifications. Also it's not clear whether it's implemented in Host OS drivers
I personally tested as many as 8x8 playback/capture channels simultaneously.
Quoted 6 lines
Once Windows 10 loads the driver successfully, however, I can't seem to select the device through Audacity on Windows. The device appears in the Windows "Device Manager" under "Sound, video and game controllers -> Source/Sink". How are you testing audio playback/capture under Windows? Perhaps my test approach is wrong?
You need to go to sound settings -> manage sound devices and enable UAC2 gadget there, after that it will appear in the system.
I also used Audacity for testing this and it works fine for me
Thanks,
Ruslan
Quoted 77 lines
I used the same approach on my Linux host where the USB UAC2 device appears as an ALSA device and I use speaker-test on the target (BeagleBone Black) to play and then capture the audio using Audacity on the Linux host. I was hoping to do the same or something similar under Windows.
Below is the script I've been using on the target to configure my UAC2 device:
mkdir -p "
/sys/kernel/config/usb_gadget/g_multi"
echo "${idVendor:-0x11F5}" > "/sys/kernel/config/usb_gadget/g_multi/idVendor" echo "${idProduct:-0x0009}" > "/sys/kernel/config/usb_gadget/g_multi/idProduct" echo "${bcdDevice:-0x0001}" > "/sys/kernel/config/usb_gadget/g_multi/bcdDevice" echo "${bDeviceClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceClass"
echo "${bDeviceSubClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceSubClass"
echo "${bDeviceProtocol:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceProtocol"
echo "${bcdUSB:-0x0200}" > "/sys/kernel/config/usb_gadget/g_multi/bcdUSB" echo "${bMaxPacketSize0:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bMaxPacketSize0"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/strings/0x409" echo "${serialnumber:-000000000000}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/serialnumber" echo "${product:-UAC2}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/product"
echo "${manufacturer:-www.acme.com}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/manufacturer" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" echo "${req_number:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/req_number" echo "${c_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_ssize" echo "${c_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_srate" echo "${c_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_chmask" echo "${c_sync:-asynchronous}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_sync" echo "${p_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_ssize" echo "${p_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_srate" echo "${p_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_chmask" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1"
echo "${bmAttributes:-0x80}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/bmAttributes" echo "${MaxPower:-500}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/MaxPower"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409" echo "${configuration:-BeagleBone Composite}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409/configuration" ln -s "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/uac2.usb0" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/os_desc"
ln -s "//sys/kernel/config/usb_gadget/g_multi/configs/c.1" "/sys/kernel/config/usb_gadget/g_multi/os_desc/c.1"
echo "${qw_sign:-MSFT100}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/qw_sign"
echo "${b_vendor_code:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/b_vendor_code"
echo "${use:-1}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/use"basename
/sys/class/udc/*>/sys/kernel/config/usb_gadget/g_multi/UDCThis is only a two channel playback/capture configuration. I hoped to get this working under Windows before increasing my channel count to a full 32 channel mask with a 48K sample rate.
I look forward to hearing your suggestions and possibly pointing out where I've gone astray.
Thanks,
Glenn
Quoted 285 lines · Ruslan Bilovol
Quoted 239 lines · Glenn Schmottlach
Quoted 237 lines · Ruslan Bilovol
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settingsPlease test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From
51516435bbf2486574ec7bc9fd4726677cd474a4Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget:f_uac2: always increase endpointmax_packet_size
by one audio slotAs per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint
max_packet_sizecalculation code was adding sometimes +1 audio slot due toDIV_ROUND_UPbehaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/usb/gadget/function/f_uac2.c +1
-1@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const structf_uac2_opts *uac2_opts, } max_packet_size =num_channels (chmask) * ssize * -DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + ((srate / (factor / (1 << (ep_desc->bInterval -1 )))) +1 );if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC ))
max_packet_size=max_packet_size*FBACK_FREQ_MAX/ 100;
--
1.9.1---------------------------------------------8<---------------------------------------- From
c8f2f2b414af672ec40841e75fb1ea761ae29122Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 16 Feb 2020 22:40:23 +0200
Subject: [PATCH] usb: gadget:f_uac2: change bInterval to 1 for
8ch/24bit/44.1kHzWith bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time.
USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1.In the future, bInterval should be dynamically calculated depending on bandwith requirementes
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/usb/gadget/function/f_uac2.c +2
-3@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */ @@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */ --1.9.1
Quoted 8 lines
Quoted 4 lines
---------------------------------------------8<----------------------------------------
Thanks,
RuslanHi Ruslan -
I have had some mixed success with your additional patches. First, the
That's good news, thanks for testing!
Quoted 6 lines
UAC2 gadget does correctly register with Windows 10 but only for 1-31 channels. For a fully specified channel map (32 channels):
c_chmask=0xFFFFFFFF( Does not load)p_chmask=0xFFFFFFFF( Does not load)Quoted 7 lines
the driver fails to load (code=10). This seems similar to the prior problem. Could there be another off-by-one problem here?
With a 31 channel mask the Windows 10 usbaudio2.sys loads successfully, e.g.
c_chmask=0x7FFFFFFF(loads)p_chmask=0x7FFFFFFF(loads)That's correct behavior. UAC2 spec says that bits D27..D30 are reserved. Also bit D31 is mutually exclusive with other bits and means 'raw data'. It means that with current UAC2 gadget implementation you can enable only 27 channels.
It seems like it's possible to add more channels by adding more audio clusters but that requires UAC2 driver modifications. Also it's not clear whether it's implemented in Host OS drivers
I personally tested as many as 8x8 playback/capture channels simultaneously.
Quoted 6 lines
Once Windows 10 loads the driver successfully, however, I can't seem to select the device through Audacity on Windows. The device appears in the Windows "Device Manager" under "Sound, video and game controllers -> Source/Sink". How are you testing audio playback/capture under Windows? Perhaps my test approach is wrong?
Quoted 85 lines
You need to go to sound settings -> manage sound devices and enable UAC2 gadget there, after that it will appear in the system.
I also used Audacity for testing this and it works fine for me
Thanks,
RuslanQuoted 77 lines
I used the same approach on my Linux host where the USB UAC2 device appears as an ALSA device and I use speaker-test on the target (BeagleBone Black) to play and then capture the audio using Audacity on the Linux host. I was hoping to do the same or something similar under Windows.
Below is the script I've been using on the target to configure my UAC2 device:
mkdir -p "
/sys/kernel/config/usb_gadget/g_multi"
echo "${idVendor:-0x11F5}" > "/sys/kernel/config/usb_gadget/g_multi/idVendor" echo "${idProduct:-0x0009}" > "/sys/kernel/config/usb_gadget/g_multi/idProduct" echo "${bcdDevice:-0x0001}" > "/sys/kernel/config/usb_gadget/g_multi/bcdDevice" echo "${bDeviceClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceClass"
echo "${bDeviceSubClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceSubClass"
echo "${bDeviceProtocol:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceProtocol"
echo "${bcdUSB:-0x0200}" > "/sys/kernel/config/usb_gadget/g_multi/bcdUSB" echo "${bMaxPacketSize0:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bMaxPacketSize0"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/strings/0x409" echo "${serialnumber:-000000000000}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/serialnumber" echo "${product:-UAC2}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/product"
echo "${manufacturer:-www.acme.com}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/manufacturer" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" echo "${req_number:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/req_number" echo "${c_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_ssize" echo "${c_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_srate" echo "${c_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_chmask" echo "${c_sync:-asynchronous}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_sync" echo "${p_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_ssize" echo "${p_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_srate" echo "${p_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_chmask" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1"
echo "${bmAttributes:-0x80}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/bmAttributes" echo "${MaxPower:-500}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/MaxPower"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409" echo "${configuration:-BeagleBone Composite}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409/configuration" ln -s "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/uac2.usb0" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/os_desc"
ln -s "//sys/kernel/config/usb_gadget/g_multi/configs/c.1" "/sys/kernel/config/usb_gadget/g_multi/os_desc/c.1"
echo "${qw_sign:-MSFT100}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/qw_sign"
echo "${b_vendor_code:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/b_vendor_code"
echo "${use:-1}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/use"basename
/sys/class/udc/*>/sys/kernel/config/usb_gadget/g_multi/UDCThis is only a two channel playback/capture configuration. I hoped to get this working under Windows before increasing my channel count to a full 32 channel mask with a 48K sample rate.
I look forward to hearing your suggestions and possibly pointing out where I've gone astray.
Thanks,
Glenn
Hi Ruslan -
Thanks for the feedback but unfortunately I've experienced mixed results with the gadget UAC2 driver on both Windows/Linux. Let me describe my environment. My host platform is either a Linux Ubuntu 18.04 or Windows 10 laptop while the target environment is a BeagleBone Black (Linux beaglebone 5.4.74-g9574bba32a #1 PREEMPT). I'm testing two different scenarios:
Scenario #1:
BeagleBone Black (BBB) runs speaker-test generating a single channel (S32_LE) audio stream containing a 1KHz tone with a 48K sample rate, e.g.
Quoted 1 line
speaker-test -D hw:1,0 -r 48000 -c 1 -f 1000 -F
S32_LE-t sine
The host laptop is running Audacity and recording the tone over the UAC2 adapter. On the Linux host the capture is correct and the tone is bit-perfect. On the Windows 10 the capture contains numerous missing samples which translates into a lot of audible pops and clicks.
Scenario #2:
The Linux/Windows host plays a single channel, 48K, S32_LE 1K sine tone to the target using either Audacity (on Windows) or 'aplay' (on Linux), e.g.
Quoted 1 line
aplay -D hw:4,0 -c 1 -r 48000 -t wav tone_1k.wav (Linux)
On the BBB target I use 'arecord' to record the tone to a RAM disk and then copy the recorded file back to the host where I can verify the quality of the recording. In both instances (e.g. using either Windows or Linux for playback) the recording on the target results in a captured file with missing samples and audible pops/clicks. In this scenario the UAC2 gadget is configured with c_sync == asynchronous. I wouldn't expect things to improve with c_sync == adaptive since you mentioned in your patch that it always reports back the nominal frequency to the host from the feedback endpoint.
Do you have any suggestions that might explain (the above) behavior. Can you describe your test environment in more detail so that I can perhaps re-create it? What Linux target are you using with your tests? You mentioned you tested an 8x8 playback/capture scenario. Can you provide any details of how you performed this test and the method you used to confirm the audio quality for the capture/playback?
Thanks for any insights you might be able to offer . . .
Glenn
Quoted 418 lines · Glenn Schmottlach
Quoted 285 lines · Ruslan Bilovol
Quoted 239 lines · Glenn Schmottlach
Quoted 237 lines · Ruslan Bilovol
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settingsPlease test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From
51516435bbf2486574ec7bc9fd4726677cd474a4Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget:f_uac2: always increase endpointmax_packet_size
by one audio slotAs per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint
max_packet_sizecalculation code was adding sometimes +1 audio slot due toDIV_ROUND_UPbehaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/usb/gadget/function/f_uac2.c +1
-1@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const structf_uac2_opts *uac2_opts, } max_packet_size =num_channels (chmask) * ssize * -DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + ((srate / (factor / (1 << (ep_desc->bInterval -1 )))) +1 );if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC ))
max_packet_size=max_packet_size*FBACK_FREQ_MAX/ 100;
--
1.9.1---------------------------------------------8<---------------------------------------- From
c8f2f2b414af672ec40841e75fb1ea761ae29122Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 16 Feb 2020 22:40:23 +0200
Subject: [PATCH] usb: gadget:f_uac2: change bInterval to 1 for
8ch/24bit/44.1kHzWith bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time.
USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1.In the future, bInterval should be dynamically calculated depending on bandwith requirementes
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/usb/gadget/function/f_uac2.c +2
-3@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */ @@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */ --1.9.1
Quoted 8 lines
Quoted 4 lines
---------------------------------------------8<----------------------------------------
Thanks,
RuslanHi Ruslan -
I have had some mixed success with your additional patches. First, the
That's good news, thanks for testing!
Quoted 6 lines
UAC2 gadget does correctly register with Windows 10 but only for 1-31 channels. For a fully specified channel map (32 channels):
c_chmask=0xFFFFFFFF( Does not load)p_chmask=0xFFFFFFFF( Does not load)Quoted 7 lines
the driver fails to load (code=10). This seems similar to the prior problem. Could there be another off-by-one problem here?
With a 31 channel mask the Windows 10 usbaudio2.sys loads successfully, e.g.
c_chmask=0x7FFFFFFF(loads)p_chmask=0x7FFFFFFF(loads)That's correct behavior. UAC2 spec says that bits D27..D30 are reserved. Also bit D31 is mutually exclusive with other bits and means 'raw data'. It means that with current UAC2 gadget implementation you can enable only 27 channels.
It seems like it's possible to add more channels by adding more audio clusters but that requires UAC2 driver modifications. Also it's not clear whether it's implemented in Host OS drivers
I personally tested as many as 8x8 playback/capture channels simultaneously.
Quoted 6 lines
Once Windows 10 loads the driver successfully, however, I can't seem to select the device through Audacity on Windows. The device appears in the Windows "Device Manager" under "Sound, video and game controllers -> Source/Sink". How are you testing audio playback/capture under Windows? Perhaps my test approach is wrong?
Quoted 85 lines
You need to go to sound settings -> manage sound devices and enable UAC2 gadget there, after that it will appear in the system.
I also used Audacity for testing this and it works fine for me
Thanks,
RuslanQuoted 77 lines
I used the same approach on my Linux host where the USB UAC2 device appears as an ALSA device and I use speaker-test on the target (BeagleBone Black) to play and then capture the audio using Audacity on the Linux host. I was hoping to do the same or something similar under Windows.
Below is the script I've been using on the target to configure my UAC2 device:
mkdir -p "
/sys/kernel/config/usb_gadget/g_multi"
echo "${idVendor:-0x11F5}" > "/sys/kernel/config/usb_gadget/g_multi/idVendor" echo "${idProduct:-0x0009}" > "/sys/kernel/config/usb_gadget/g_multi/idProduct" echo "${bcdDevice:-0x0001}" > "/sys/kernel/config/usb_gadget/g_multi/bcdDevice" echo "${bDeviceClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceClass"
echo "${bDeviceSubClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceSubClass"
echo "${bDeviceProtocol:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceProtocol"
echo "${bcdUSB:-0x0200}" > "/sys/kernel/config/usb_gadget/g_multi/bcdUSB" echo "${bMaxPacketSize0:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bMaxPacketSize0"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/strings/0x409" echo "${serialnumber:-000000000000}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/serialnumber" echo "${product:-UAC2}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/product"
echo "${manufacturer:-www.acme.com}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/manufacturer" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" echo "${req_number:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/req_number" echo "${c_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_ssize" echo "${c_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_srate" echo "${c_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_chmask" echo "${c_sync:-asynchronous}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_sync" echo "${p_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_ssize" echo "${p_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_srate" echo "${p_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_chmask" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1"
echo "${bmAttributes:-0x80}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/bmAttributes" echo "${MaxPower:-500}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/MaxPower"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409" echo "${configuration:-BeagleBone Composite}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409/configuration" ln -s "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/uac2.usb0" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/os_desc"
ln -s "//sys/kernel/config/usb_gadget/g_multi/configs/c.1" "/sys/kernel/config/usb_gadget/g_multi/os_desc/c.1"
echo "${qw_sign:-MSFT100}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/qw_sign"
echo "${b_vendor_code:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/b_vendor_code"
echo "${use:-1}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/use"basename
/sys/class/udc/*>/sys/kernel/config/usb_gadget/g_multi/UDCThis is only a two channel playback/capture configuration. I hoped to get this working under Windows before increasing my channel count to a full 32 channel mask with a 48K sample rate.
I look forward to hearing your suggestions and possibly pointing out where I've gone astray.
Thanks,
Glenn
Hi Ruslan -
Thanks for the feedback but unfortunately I've experienced mixed results with the gadget UAC2 driver on both Windows/Linux. Let me describe my environment. My host platform is either a Linux Ubuntu 18.04 or Windows 10 laptop while the target environment is a BeagleBone Black (Linux beaglebone 5.4.74-g9574bba32a #1 PREEMPT). I'm testing two different scenarios:
Scenario #1:
BeagleBone Black (BBB) runs speaker-test generating a single channel (S32_LE) audio stream containing a 1KHz tone with a 48K sample rate, e.g.Quoted 1 line
speaker-test -D hw:1,0 -r 48000 -c 1 -f 1000 -F
S32_LE-t sineThe host laptop is running Audacity and recording the tone over the UAC2 adapter. On the Linux host the capture is correct and the tone is bit-perfect. On the Windows 10 the capture contains numerous missing samples which translates into a lot of audible pops and clicks.
Scenario #2:
The Linux/Windows host plays a single channel, 48K,S32_LE1K sine tone to the target using either Audacity (on Windows) or 'aplay' (on Linux), e.g.Quoted 1 line
aplay -D hw:4,0 -c 1 -r 48000 -t wav tone_1k.wav (Linux)
On the BBB target I use 'arecord' to record the tone to a RAM disk and then copy the recorded file back to the host where I can verify the quality of the recording. In both instances (e.g. using either Windows or Linux for playback) the recording on the target results in a captured file with missing samples and audible pops/clicks. In this scenario the UAC2 gadget is configured with
c_sync== asynchronous. I wouldn't expect things to improve withc_sync== adaptive since you mentioned in your patch that it always reports back the nominal frequency to the host from the feedback endpoint.Do you have any suggestions that might explain (the above) behavior. Can you describe your test environment in more detail so that I can perhaps re-create it? What Linux target are you using with your tests? You mentioned you tested an 8x8 playback/capture scenario. Can you provide any details of how you performed this test and the method you used to confirm the audio quality for the capture/playback?
Quoted 3 lines
Thanks for any insights you might be able to offer . . .
Glenn
Hi Ruslan -
This is a follow-up from my post yesterday. I recompiled my kernel *WITHOUT* your UAC2 patches and repeated Scenario #2 where the Linux PC plays a single channel tone to the BeagleBone Black where it's recorded with 'arecord'. Yesterday, I recorded garbled audio on the target but today, without any UAC2 kernel patches, the recorded audio on the target is glitch-free and appears to be bit-perfect.
This experiment leads me to believe your patches may be inadvertently corrupting the data-path. Have you been able to repeat my experiment and either confirm or refute my findings? I am interested to learn more how you tested your patches and whether it's something I can recreate here.
Assuming we can sort out this data corruption issue, what are your thoughts on how the Linux target device can properly provide the Windows feedback endpoint with real frequency updates rather than the constant nominal frequency. If I understood your patch notes correctly it seems this is an outstanding issue that requires additional attention. I'm a bit of a noob when it comes to how this might be addressed.
Thanks for your continued insights and support . . .
Glenn
Quoted 73 lines · Glenn Schmottlach
Quoted 45 lines · Glenn Schmottlach
Hi Ruslan -
Thanks for the feedback but unfortunately I've experienced mixed results with the gadget UAC2 driver on both Windows/Linux. Let me describe my environment. My host platform is either a Linux Ubuntu 18.04 or Windows 10 laptop while the target environment is a BeagleBone Black (Linux beaglebone 5.4.74-g9574bba32a #1 PREEMPT). I'm testing two different scenarios:
Scenario #1:
BeagleBone Black (BBB) runs speaker-test generating a single channel (S32_LE) audio stream containing a 1KHz tone with a 48K sample rate, e.g.Quoted 1 line
speaker-test -D hw:1,0 -r 48000 -c 1 -f 1000 -F
S32_LE-t sineThe host laptop is running Audacity and recording the tone over the UAC2 adapter. On the Linux host the capture is correct and the tone is bit-perfect. On the Windows 10 the capture contains numerous missing samples which translates into a lot of audible pops and clicks.
Scenario #2:
The Linux/Windows host plays a single channel, 48K,S32_LE1K sine tone to the target using either Audacity (on Windows) or 'aplay' (on Linux), e.g.Quoted 1 line
aplay -D hw:4,0 -c 1 -r 48000 -t wav tone_1k.wav (Linux)
On the BBB target I use 'arecord' to record the tone to a RAM disk and then copy the recorded file back to the host where I can verify the quality of the recording. In both instances (e.g. using either Windows or Linux for playback) the recording on the target results in a captured file with missing samples and audible pops/clicks. In this scenario the UAC2 gadget is configured with
c_sync== asynchronous. I wouldn't expect things to improve withc_sync== adaptive since you mentioned in your patch that it always reports back the nominal frequency to the host from the feedback endpoint.Do you have any suggestions that might explain (the above) behavior. Can you describe your test environment in more detail so that I can perhaps re-create it? What Linux target are you using with your tests? You mentioned you tested an 8x8 playback/capture scenario. Can you provide any details of how you performed this test and the method you used to confirm the audio quality for the capture/playback?
Quoted 3 lines
Thanks for any insights you might be able to offer . . .
Glenn
Hi Ruslan -
This is a follow-up from my post yesterday. I recompiled my kernel *WITHOUT* your UAC2 patches and repeated Scenario #2 where the Linux PC plays a single channel tone to the BeagleBone Black where it's recorded with 'arecord'. Yesterday, I recorded garbled audio on the target but today, without any UAC2 kernel patches, the recorded audio on the target is glitch-free and appears to be bit-perfect.
This experiment leads me to believe your patches may be inadvertently corrupting the data-path. Have you been able to repeat my experiment and either confirm or refute my findings? I am interested to learn more how you tested your patches and whether it's something I can recreate here.
Assuming we can sort out this data corruption issue, what are your thoughts on how the Linux target device can properly provide the Windows feedback endpoint with real frequency updates rather than the constant nominal frequency. If I understood your patch notes correctly it seems this is an outstanding issue that requires additional attention. I'm a bit of a noob when it comes to how this might be addressed.
Quoted 3 lines
Thanks for your continued insights and support . . .
Glenn
Hi Glenn & Ruslan,
Do you know why WIN10 can't recognized UAC2 device if I configure the sample rate as 48000HZ? Configuring sample rate as 44100HZ, the playback function would work well at my platforms (chipidea IP), no glitch is heard. At WIN10, I use Windows Media Player, at board side I use command:
arecord -f cd -t wav -D hw:4,0 | aplay -f cd -Dplughw:3,0 &
From the USB Bus analyzer:
Feedback EP is scheduled every 1ms, there are nine 176-byte packets and one 180-byte packet among 10ms transfers.
Signature · 2 lines
Thanks, Peter Chen
Quoted 90 lines · Peter Chen
Quoted 73 lines · Glenn Schmottlach
Quoted 45 lines · Glenn Schmottlach
Hi Ruslan -
Thanks for the feedback but unfortunately I've experienced mixed results with the gadget UAC2 driver on both Windows/Linux. Let me describe my environment. My host platform is either a Linux Ubuntu 18.04 or Windows 10 laptop while the target environment is a BeagleBone Black (Linux beaglebone 5.4.74-g9574bba32a #1 PREEMPT). I'm testing two different scenarios:
Scenario #1:
BeagleBone Black (BBB) runs speaker-test generating a single channel (S32_LE) audio stream containing a 1KHz tone with a 48K sample rate, e.g.Quoted 1 line
speaker-test -D hw:1,0 -r 48000 -c 1 -f 1000 -F
S32_LE-t sineThe host laptop is running Audacity and recording the tone over the UAC2 adapter. On the Linux host the capture is correct and the tone is bit-perfect. On the Windows 10 the capture contains numerous missing samples which translates into a lot of audible pops and clicks.
Scenario #2:
The Linux/Windows host plays a single channel, 48K,S32_LE1K sine tone to the target using either Audacity (on Windows) or 'aplay' (on Linux), e.g.Quoted 1 line
aplay -D hw:4,0 -c 1 -r 48000 -t wav tone_1k.wav (Linux)
On the BBB target I use 'arecord' to record the tone to a RAM disk and then copy the recorded file back to the host where I can verify the quality of the recording. In both instances (e.g. using either Windows or Linux for playback) the recording on the target results in a captured file with missing samples and audible pops/clicks. In this scenario the UAC2 gadget is configured with
c_sync== asynchronous. I wouldn't expect things to improve withc_sync== adaptive since you mentioned in your patch that it always reports back the nominal frequency to the host from the feedback endpoint.Do you have any suggestions that might explain (the above) behavior. Can you describe your test environment in more detail so that I can perhaps re-create it? What Linux target are you using with your tests? You mentioned you tested an 8x8 playback/capture scenario. Can you provide any details of how you performed this test and the method you used to confirm the audio quality for the capture/playback?
Quoted 3 lines
Thanks for any insights you might be able to offer . . .
Glenn
Hi Ruslan -
This is a follow-up from my post yesterday. I recompiled my kernel *WITHOUT* your UAC2 patches and repeated Scenario #2 where the Linux PC plays a single channel tone to the BeagleBone Black where it's recorded with 'arecord'. Yesterday, I recorded garbled audio on the target but today, without any UAC2 kernel patches, the recorded audio on the target is glitch-free and appears to be bit-perfect.
This experiment leads me to believe your patches may be inadvertently corrupting the data-path. Have you been able to repeat my experiment and either confirm or refute my findings? I am interested to learn more how you tested your patches and whether it's something I can recreate here.
Assuming we can sort out this data corruption issue, what are your thoughts on how the Linux target device can properly provide the Windows feedback endpoint with real frequency updates rather than the constant nominal frequency. If I understood your patch notes correctly it seems this is an outstanding issue that requires additional attention. I'm a bit of a noob when it comes to how this might be addressed.
Quoted 3 lines
Thanks for your continued insights and support . . .
Glenn
Hi Glenn & Ruslan,
Do you know why WIN10 can't recognized UAC2 device if I configure the sample rate as 48000HZ? Configuring sample rate as 44100HZ, the playback function would work well at my platforms (chipidea IP), no glitch is heard. At WIN10, I use Windows Media Player, at board side I use command:
arecord -f cd -t wav -D hw:4,0 | aplay -f cd -Dplughw:3,0 &
From the USB Bus analyzer:
Quoted 7 lines
Feedback EP is scheduled every 1ms, there are nine 176-byte packets and one 180-byte packet among 10ms transfers.
--
Thanks,
Peter Chen
Hi Peter -
I have not experienced the issue you are reporting. My configuration uses a 48Khz sample rate exclusively and Windows 10 appears to recognize the UAC2 device properly with all of Ruslan's patches. I have not done any serious testing at the 44.1KHz rate so I don't recall if that is recognized correctly on my Windows 10 box.
It looks like you've established a UAC2 loopback on your target board. How are you verifying the captured audio on the Window 10 PC? I didn't realize Windows Media Player allows you to both playback and capture from the same (UAC2) device simultaneously.
Thanks,
Glenn
Quoted 77 lines · Peter Chen
Quoted 75 lines · Glenn Schmottlach
Quoted 45 lines · Glenn Schmottlach
Hi Ruslan -
Thanks for the feedback but unfortunately I've experienced mixed results with the gadget UAC2 driver on both Windows/Linux. Let me describe my environment. My host platform is either a Linux Ubuntu 18.04 or Windows 10 laptop while the target environment is a BeagleBone Black (Linux beaglebone 5.4.74-g9574bba32a #1 PREEMPT). I'm testing two different scenarios:
Scenario #1:
BeagleBone Black (BBB) runs speaker-test generating a single channel (S32_LE) audio stream containing a 1KHz tone with a 48K sample rate, e.g.Quoted 1 line
speaker-test -D hw:1,0 -r 48000 -c 1 -f 1000 -F
S32_LE-t sineThe host laptop is running Audacity and recording the tone over the UAC2 adapter. On the Linux host the capture is correct and the tone is bit-perfect. On the Windows 10 the capture contains numerous missing samples which translates into a lot of audible pops and clicks.
Scenario #2:
The Linux/Windows host plays a single channel, 48K,S32_LE1K sine tone to the target using either Audacity (on Windows) or 'aplay' (on Linux), e.g.Quoted 1 line
aplay -D hw:4,0 -c 1 -r 48000 -t wav tone_1k.wav (Linux)
On the BBB target I use 'arecord' to record the tone to a RAM disk and then copy the recorded file back to the host where I can verify the quality of the recording. In both instances (e.g. using either Windows or Linux for playback) the recording on the target results in a captured file with missing samples and audible pops/clicks. In this scenario the UAC2 gadget is configured with
c_sync== asynchronous. I wouldn't expect things to improve withc_sync== adaptive since you mentioned in your patch that it always reports back the nominal frequency to the host from the feedback endpoint.Do you have any suggestions that might explain (the above) behavior. Can you describe your test environment in more detail so that I can perhaps re-create it? What Linux target are you using with your tests? You mentioned you tested an 8x8 playback/capture scenario. Can you provide any details of how you performed this test and the method you used to confirm the audio quality for the capture/playback?
Quoted 3 lines
Thanks for any insights you might be able to offer . . .
Glenn
Hi Ruslan -
This is a follow-up from my post yesterday. I recompiled my kernel *WITHOUT* your UAC2 patches and repeated Scenario #2 where the Linux PC plays a single channel tone to the BeagleBone Black where it's recorded with 'arecord'. Yesterday, I recorded garbled audio on the target but today, without any UAC2 kernel patches, the recorded audio on the target is glitch-free and appears to be bit-perfect.
This experiment leads me to believe your patches may be inadvertently corrupting the data-path. Have you been able to repeat my experiment and either confirm or refute my findings? I am interested to learn more how you tested your patches and whether it's something I can recreate here.
Assuming we can sort out this data corruption issue, what are your thoughts on how the Linux target device can properly provide the Windows feedback endpoint with real frequency updates rather than the constant nominal frequency. If I understood your patch notes correctly it seems this is an outstanding issue that requires additional attention. I'm a bit of a noob when it comes to how this might be addressed.
Thanks for your continued insights and support . . .
Quoted 8 lines
Quoted 1 line
Glenn
Hi Glenn & Ruslan,
Do you know why WIN10 can't recognized UAC2 device if I configure the sample rate as 48000HZ? Configuring sample rate as 44100HZ, the playback function would work well at my platforms (chipidea IP), no glitch is heard. At WIN10, I use Windows Media Player, at board side I use command:
That's known issue, Windows is more strict with UAC2 descriptors, try to apply my patch "usb: gadget: f_uac2: always increase endpoint max_packet_size by one audio slot" that I shared in this email:
https://www.spinics.net/lists/linux-usb/msg205077.html
Thanks,
Ruslan
Quoted 87 lines · Ruslan Bilovol
Quoted 77 lines · Peter Chen
Quoted 75 lines · Glenn Schmottlach
Quoted 45 lines · Glenn Schmottlach
Hi Ruslan -
Thanks for the feedback but unfortunately I've experienced mixed results with the gadget UAC2 driver on both Windows/Linux. Let me describe my environment. My host platform is either a Linux Ubuntu 18.04 or Windows 10 laptop while the target environment is a BeagleBone Black (Linux beaglebone 5.4.74-g9574bba32a #1 PREEMPT). I'm testing two different scenarios:
Scenario #1:
BeagleBone Black (BBB) runs speaker-test generating a single channel (S32_LE) audio stream containing a 1KHz tone with a 48K sample rate, e.g.Quoted 1 line
speaker-test -D hw:1,0 -r 48000 -c 1 -f 1000 -F
S32_LE-t sineThe host laptop is running Audacity and recording the tone over the UAC2 adapter. On the Linux host the capture is correct and the tone is bit-perfect. On the Windows 10 the capture contains numerous missing samples which translates into a lot of audible pops and clicks.
Scenario #2:
The Linux/Windows host plays a single channel, 48K,S32_LE1K sine tone to the target using either Audacity (on Windows) or 'aplay' (on Linux), e.g.Quoted 1 line
aplay -D hw:4,0 -c 1 -r 48000 -t wav tone_1k.wav (Linux)
On the BBB target I use 'arecord' to record the tone to a RAM disk and then copy the recorded file back to the host where I can verify the quality of the recording. In both instances (e.g. using either Windows or Linux for playback) the recording on the target results in a captured file with missing samples and audible pops/clicks. In this scenario the UAC2 gadget is configured with
c_sync== asynchronous. I wouldn't expect things to improve withc_sync== adaptive since you mentioned in your patch that it always reports back the nominal frequency to the host from the feedback endpoint.Do you have any suggestions that might explain (the above) behavior. Can you describe your test environment in more detail so that I can perhaps re-create it? What Linux target are you using with your tests? You mentioned you tested an 8x8 playback/capture scenario. Can you provide any details of how you performed this test and the method you used to confirm the audio quality for the capture/playback?
Quoted 3 lines
Thanks for any insights you might be able to offer . . .
Glenn
Hi Ruslan -
This is a follow-up from my post yesterday. I recompiled my kernel *WITHOUT* your UAC2 patches and repeated Scenario #2 where the Linux PC plays a single channel tone to the BeagleBone Black where it's recorded with 'arecord'. Yesterday, I recorded garbled audio on the target but today, without any UAC2 kernel patches, the recorded audio on the target is glitch-free and appears to be bit-perfect.
This experiment leads me to believe your patches may be inadvertently corrupting the data-path. Have you been able to repeat my experiment and either confirm or refute my findings? I am interested to learn more how you tested your patches and whether it's something I can recreate here.
Assuming we can sort out this data corruption issue, what are your thoughts on how the Linux target device can properly provide the Windows feedback endpoint with real frequency updates rather than the constant nominal frequency. If I understood your patch notes correctly it seems this is an outstanding issue that requires additional attention. I'm a bit of a noob when it comes to how this might be addressed.
Thanks for your continued insights and support . . .
Quoted 8 lines
Quoted 1 line
Glenn
Hi Glenn & Ruslan,
Do you know why WIN10 can't recognized UAC2 device if I configure the sample rate as 48000HZ? Configuring sample rate as 44100HZ, the playback function would work well at my platforms (chipidea IP), no glitch is heard. At WIN10, I use Windows Media Player, at board side I use command:
Quoted 5 lines
That's known issue, Windows is more strict with UAC2 descriptors, try to apply my patch "usb: gadget:
f_uac2: always increase endpointmax_packet_sizeby one audio slot" that I shared in this email:
https://eur01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.spinics.net%2Flists%2Flinux-usb%2Fmsg205077.html&data=04%7C01%7Cpeter.chen%40nxp.com%7C1c037c406a47423725e008d89d0b77c1%7C686ea1d3bc2b4c6fa92cd99c5c301635%7C0%7C0%7C637432019865845335%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=7jE6EEi4ZifLMVNgs3%2FygUkBjCrcIrtvIjX2%2FVwbwUM%3D&reserved=0
Better than before, the Device Manager could recognize this sound device, But Windows Media Player can't play it, see attached.
Meanwhile, I find your proposal patch [1] is not based on your UAC2 patch which changed bmAttributes as USB_ENDPOINT_XFER_ISOC only at descriptor.
[1] https://www.spinics.net/lists/linux-usb/msg205077.html
Signature · 2 lines
Thanks, Peter Chen
Quoted 411 lines · Glenn Schmottlach
Quoted 285 lines · Ruslan Bilovol
Quoted 239 lines · Glenn Schmottlach
Quoted 237 lines · Ruslan Bilovol
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settingsPlease test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From
51516435bbf2486574ec7bc9fd4726677cd474a4Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget:f_uac2: always increase endpointmax_packet_size
by one audio slotAs per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint
max_packet_sizecalculation code was adding sometimes +1 audio slot due toDIV_ROUND_UPbehaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/usb/gadget/function/f_uac2.c +1
-1@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const structf_uac2_opts *uac2_opts, } max_packet_size =num_channels (chmask) * ssize * -DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + ((srate / (factor / (1 << (ep_desc->bInterval -1 )))) +1 );if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC ))
max_packet_size=max_packet_size*FBACK_FREQ_MAX/ 100;
--
1.9.1---------------------------------------------8<---------------------------------------- From
c8f2f2b414af672ec40841e75fb1ea761ae29122Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 16 Feb 2020 22:40:23 +0200
Subject: [PATCH] usb: gadget:f_uac2: change bInterval to 1 for
8ch/24bit/44.1kHzWith bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time.
USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1.In the future, bInterval should be dynamically calculated depending on bandwith requirementes
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/usb/gadget/function/f_uac2.c +2
-3@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */ @@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */ --1.9.1
Quoted 8 lines
Quoted 4 lines
---------------------------------------------8<----------------------------------------
Thanks,
RuslanHi Ruslan -
I have had some mixed success with your additional patches. First, the
That's good news, thanks for testing!
Quoted 6 lines
UAC2 gadget does correctly register with Windows 10 but only for 1-31 channels. For a fully specified channel map (32 channels):
c_chmask=0xFFFFFFFF( Does not load)p_chmask=0xFFFFFFFF( Does not load)Quoted 7 lines
the driver fails to load (code=10). This seems similar to the prior problem. Could there be another off-by-one problem here?
With a 31 channel mask the Windows 10 usbaudio2.sys loads successfully, e.g.
c_chmask=0x7FFFFFFF(loads)p_chmask=0x7FFFFFFF(loads)That's correct behavior. UAC2 spec says that bits D27..D30 are reserved. Also bit D31 is mutually exclusive with other bits and means 'raw data'. It means that with current UAC2 gadget implementation you can enable only 27 channels.
It seems like it's possible to add more channels by adding more audio clusters but that requires UAC2 driver modifications. Also it's not clear whether it's implemented in Host OS drivers
I personally tested as many as 8x8 playback/capture channels simultaneously.
Quoted 6 lines
Once Windows 10 loads the driver successfully, however, I can't seem to select the device through Audacity on Windows. The device appears in the Windows "Device Manager" under "Sound, video and game controllers -> Source/Sink". How are you testing audio playback/capture under Windows? Perhaps my test approach is wrong?
Quoted 85 lines
You need to go to sound settings -> manage sound devices and enable UAC2 gadget there, after that it will appear in the system.
I also used Audacity for testing this and it works fine for me
Thanks,
RuslanQuoted 77 lines
I used the same approach on my Linux host where the USB UAC2 device appears as an ALSA device and I use speaker-test on the target (BeagleBone Black) to play and then capture the audio using Audacity on the Linux host. I was hoping to do the same or something similar under Windows.
Below is the script I've been using on the target to configure my UAC2 device:
mkdir -p "
/sys/kernel/config/usb_gadget/g_multi"
echo "${idVendor:-0x11F5}" > "/sys/kernel/config/usb_gadget/g_multi/idVendor" echo "${idProduct:-0x0009}" > "/sys/kernel/config/usb_gadget/g_multi/idProduct" echo "${bcdDevice:-0x0001}" > "/sys/kernel/config/usb_gadget/g_multi/bcdDevice" echo "${bDeviceClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceClass"
echo "${bDeviceSubClass:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceSubClass"
echo "${bDeviceProtocol:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bDeviceProtocol"
echo "${bcdUSB:-0x0200}" > "/sys/kernel/config/usb_gadget/g_multi/bcdUSB" echo "${bMaxPacketSize0:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/bMaxPacketSize0"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/strings/0x409" echo "${serialnumber:-000000000000}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/serialnumber" echo "${product:-UAC2}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/product"
echo "${manufacturer:-www.acme.com}" > "/sys/kernel/config/usb_gadget/g_multi/strings/0x409/manufacturer" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" echo "${req_number:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/req_number" echo "${c_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_ssize" echo "${c_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_srate" echo "${c_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_chmask" echo "${c_sync:-asynchronous}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/c_sync" echo "${p_ssize:-4}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_ssize" echo "${p_srate:-48000}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_srate" echo "${p_chmask:-0x00000003}" > "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0/p_chmask" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1"
echo "${bmAttributes:-0x80}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/bmAttributes" echo "${MaxPower:-500}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/MaxPower"
mkdir -p "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409" echo "${configuration:-BeagleBone Composite}" > "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/strings/0x409/configuration" ln -s "/sys/kernel/config/usb_gadget/g_multi/functions/uac2.usb0" "/sys/kernel/config/usb_gadget/g_multi/configs/c.1/uac2.usb0" mkdir -p "/sys/kernel/config/usb_gadget/g_multi/os_desc"
ln -s "//sys/kernel/config/usb_gadget/g_multi/configs/c.1" "/sys/kernel/config/usb_gadget/g_multi/os_desc/c.1"
echo "${qw_sign:-MSFT100}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/qw_sign"
echo "${b_vendor_code:-0x00}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/b_vendor_code"
echo "${use:-1}" > "/sys/kernel/config/usb_gadget/g_multi/os_desc/use"basename
/sys/class/udc/*>/sys/kernel/config/usb_gadget/g_multi/UDCThis is only a two channel playback/capture configuration. I hoped to get this working under Windows before increasing my channel count to a full 32 channel mask with a 48K sample rate.
I look forward to hearing your suggestions and possibly pointing out where I've gone astray.
Thanks,
Glenn
Hi Ruslan -
Thanks for the feedback but unfortunately I've experienced mixed results with the gadget UAC2 driver on both Windows/Linux. Let me describe my environment. My host platform is either a Linux Ubuntu 18.04 or Windows 10 laptop while the target environment is a BeagleBone Black (Linux beaglebone 5.4.74-g9574bba32a #1 PREEMPT). I'm testing two different scenarios:
Scenario #1:
BeagleBone Black (BBB) runs speaker-test generating a single channel (S32_LE) audio stream containing a 1KHz tone with a 48K sample rate, e.g.Quoted 1 line
speaker-test -D hw:1,0 -r 48000 -c 1 -f 1000 -F
S32_LE-t sineThe host laptop is running Audacity and recording the tone over the UAC2 adapter. On the Linux host the capture is correct and the tone is bit-perfect. On the Windows 10 the capture contains numerous missing samples which translates into a lot of audible pops and clicks.
Scenario #2:
The Linux/Windows host plays a single channel, 48K,S32_LE1K sine tone to the target using either Audacity (on Windows) or 'aplay' (on Linux), e.g.Quoted 1 line
aplay -D hw:4,0 -c 1 -r 48000 -t wav tone_1k.wav (Linux)
On the BBB target I use 'arecord' to record the tone to a RAM disk and then copy the recorded file back to the host where I can verify the quality of the recording. In both instances (e.g. using either Windows or Linux for playback) the recording on the target results in a captured file with missing samples and audible pops/clicks. In this scenario the UAC2 gadget is configured with
c_sync== asynchronous. I wouldn't expect things to improve withc_sync== adaptive since you mentioned in your patch that it always reports back the nominal frequency to the host from the feedback endpoint.
Quoted 6 lines
Do you have any suggestions that might explain (the above) behavior. Can you describe your test environment in more detail so that I can perhaps re-create it? What Linux target are you using with your tests? You mentioned you tested an 8x8 playback/capture scenario. Can you provide any details of how you performed this test and the method you used to confirm the audio quality for the capture/playback?
I also use BeagleBone black in my development.
I have two testing scenarios:
1) BeagleBone PC loopback (e.g. redirect play->capture in BBB), here I play sine wave on PC to BBB and record back (simultaneously in Audacity), it looks like:
Audiacity (PC playback) -> BBB capture -> alsaloop -> BBB playback -> Audacity (PC capture)
2) I connect another USB audio card to BBB, and redirect audio samples from PC through BBB UAC2 audio gadget to USB audio card, this path looks like:
PC playback) -> BBB capture -> alsaloop -> BBB USB audio card
With #1 I tested simultaneous 8x8ch 24bit 44.1 kHz loopback (also 4x4ch and 2x2 ch)
With #2 it's limited to USB audio card I have (usually I do 2ch 16bit 44.1 or 48 kHz)
I see you use 32bit samples, I tested it a few times before on the previous patch set versions.
Can you try to check 2ch 16bit 44.1 or 48 kHz? I'm thinking if this issue is limited to some combination of sampling rate, channels and sample resolution
Thanks,
Ruslan
Quoted 4 lines
Thanks for any insights you might be able to offer . . .
Glenn
Quoted 182 lines · Ruslan Bilovol
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settingsPlease test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From
51516435bbf2486574ec7bc9fd4726677cd474a4Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget:f_uac2: always increase endpointmax_packet_size
by one audio slotAs per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint
max_packet_sizecalculation code was adding sometimes +1 audio slot due toDIV_ROUND_UPbehaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/usb/gadget/function/f_uac2.c +1
-1@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const structf_uac2_opts *uac2_opts, } max_packet_size =num_channels (chmask) * ssize * -DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + ((srate / (factor / (1 << (ep_desc->bInterval -1 )))) +1 );if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC ))
Quoted 1 line
max_packet_size = max_packet_size * FBACK_FREQ_MAX / 100;
I think "reserving" 20% additional bandwidth is a bit extreme. In the end, the purpose is composate the drift which appears between different XTALs
Allocating bandwidth for 1 more sample than the stream should require (49 instead of 48 for 48kHz with 4 bIntervals) should allow to compensate any realistic drift, don't you think ?
Quoted 20 lines
Signature · 19 lines
1.9.1 ---------------------------------------------8<---------------------------------------- From c8f2f2b414af672ec40841e75fb1ea761ae29122 Mon Sep 17 00:00:00 2001 From: Ruslan Bilovol <[email protected]> Date: Sun, 16 Feb 2020 22:40:23 +0200 Subject: [PATCH] usb: gadget: f_uac2: change bInterval to 1 for 8ch/24bit/44.1kHz With bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time. USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1. In the future, bInterval should be dynamically calculated depending on bandwith requirementes
Probably better to do it from the start to avoid breaking stuff for people who have been using the gadget with the current badnwidth so far
Quoted 29 lines
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/usb/gadget/function/f_uac2.c +2
-2@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */ @@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */
Dne 26. 11. 20 v 14:16 Jerome Brunet napsal(a):
Quoted 6 lines
Quoted 6 lines
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
Should anyone test the patches with RPi4 dwc2 as the gadget, please note the recently accepted patch https://patchwork.kernel.org/project/linux-usb/patch/[email protected]/ which allows using full 1024 bytes maxpacketsize on EP OUT. It is not an enumeration issue, but the old (= existing) RX FIFO size drops data with packet sizes above 960 bytes.
Thanks a lot for the great work.
Best regards,
Pavel.
Quoted 9 lines · Pavel Hofman
Dne 26. 11. 20 v 14:16 Jerome Brunet napsal(a):
Quoted 6 lines
Quoted 6 lines
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
Quoted 6 lines
Should anyone test the patches with RPi4 dwc2 as the gadget, please note the recently accepted patch
https://patchwork.kernel.org/project/linux-usb/patch/[email protected]/
which allows using full 1024 bytes maxpacketsize on EP OUT. It is not an enumeration issue, but the old (= existing) RX FIFO size drops data with packet sizes above 960 bytes.
Thanks for bringing this to our attention. Does it affect RPi 3 as well? I'm going to test feedback feature on this board
Regards,
Ruslan
Dne 04. 12. 20 v 15:39 Ruslan Bilovol napsal(a):
Quoted 17 lines · Pavel Hofman
Quoted 9 lines
Dne 26. 11. 20 v 14:16 Jerome Brunet napsal(a):
Quoted 6 lines
Quoted 6 lines
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
Quoted 6 lines
Should anyone test the patches with RPi4 dwc2 as the gadget, please note the recently accepted patch
https://patchwork.kernel.org/project/linux-usb/patch/[email protected]/
which allows using full 1024 bytes maxpacketsize on EP OUT. It is not an enumeration issue, but the old (= existing) RX FIFO size drops data with packet sizes above 960 bytes.
Quoted 2 lines
Thanks for bringing this to our attention. Does it affect RPi 3 as well? I'm going to test feedback feature on this board
It configures the dwc2 IP of the broadcom which IIRC is in all RPis. The affected files bcm283x-rpi-usb-otg.dtsi and bcm283x-rpi-usb-peripheral.dtsi do not seem RPi4 specific. But I had only RPi4 to analyze the problem and check the fix.
Ruslan, please what do you think about the existing behaviour of gaudio in case of USB stream disconnection? The elapsed-period callback stops being called and the user space will eventually time out. How about closing the PCM stream immediately in case of a problem in the USB layer? E.g. some SPDIF receivers do that in case of change in the incoming samplerate
https://github.com/torvalds/linux/blob/master/sound/i2c/other/ak4117.c#L503
.
Thanks a lot for the great work.
With regards,
Pavel.
Quoted 185 lines · Jerome Brunet
Quoted 181 lines · Ruslan Bilovol
Quoted 111 lines · Glenn Schmottlach
Quoted 95 lines · Ruslan Bilovol
Quoted 55 lines · Peter Chen
Quoted 31 lines · Ruslan Bilovol
Current UAC2 gadget implements capture/sync paths as two USB ISO ASYNC endpoints (IN and OUT).
This violates USB spec which says that ISO ASYNC OUT endpoint should have feedback companion endpoint.
See USB2.0 spec "5.12.4.1 Synchronization Type": asynchronous sink provides explicit feedback (isochronous pipe). Interesting that for ISO ASYNC *IN* endpoint respective feedback isn't required since source provides implicit feedforward (data stream).While it's not an issue if UAC2 Gadget is connected to Linux host (Linux ignores missing feedback endpoint), with other hosts like Windows or MacOS the UAC2 Gadget isn't enumerated due to missing feedback endpoint.
This patch series adds feedback endpoint support to UAC2 function, new control to UAC2 mixer which can be used by userspace tools (like alsaloop from alsa-utils) for updating feedback frequency reported to the host. This is useful for usecases when UAC2 Gadget's audio samples are played to another codec or audio card with its own internal freerunning clock so host can be notified that more/less samples are required.
The alsaloop tool requires some (relatively small) modifications in order to start support driving feedback frequency through UAC2 mixer control. That change will be sent as a separate patch to ALSA community.
Quoted 8 lines
Also added ability to switch ISO ASYNC OUT endpoint into adaptive endpoint which doesn't require feedback endpoint (as per USB spec).
Ruslan Bilovol (3):
usb: gadget: f_uac2/u_audio: add feedback endpoint support
usb: gadget:f_uac2: add adaptive sync support for capture
usb: gadget:u_audio: add real feedback implementationHi Ruslan,
I applied your patches, but WIN10 still can't recognize it well. The UAC1 is OK for WIN10 with the below same configuration. Any debug information you would like to know to check it?
if [ "$FUNC" == "uac2" ]; then
mkdir functions/$FUNC".0"
echo 2 > functions/$FUNC".0"/p_ssize
echo 48000 > functions/$FUNC".0"/p_srate
echo 3 > functions/$FUNC".0"/p_chmaskQuoted 8 lines
echo 2 > functions/$FUNC".0"/c_ssize
echo 48000 > functions/$FUNC".0"/c_srate
echo 3 > functions/$FUNC".0"/c_chmask
#echo 4 > functions/$FUNC".0"/req_number
ln -s functions/$FUNC".0" configs/c.1
echo high-speed >/sys/kernel/config/usb_gadget/g1/max_speedfiHmm... I just tested below config and it works fine with my Win10. The only thing I noticed is Windows doesn't enable UAC2 gadget by default, but this can be done from Win10 sound settings
--------------------------------->8-------------------------------------- mkdir cfg
mount none cfg -t configfs
mkdir cfg/usb_gadget/g1
cd cfg/usb_gadget/g1
mkdir configs/c.1
mkdir functions/uac2.0# 44.1 kHz sample rate
echo 44100 > functions/uac2.0/c_srate
echo 44100 > functions/uac2.0/p_sratemkdir strings/0x409
mkdir configs/c.1/strings/0x409
echo0x0101> idProduct
echo0x1d6b> idVendor
echo my-serial-num > strings/0x409/serialnumber
echo my-manufacturer > strings/0x409/manufacturer
echo "Test gadget" > strings/0x409/product
echo "Conf 1" > configs/c.1/strings/0x409/configuration
echo 120 > configs/c.1/MaxPower
ln -s functions/uac2.0 configs/c.1
echo musb-hdrc.0 > UDC --------------------------------->8--------------------------------------Quoted 2 lines
Thanks,
RuslanHi Ruslan -
With your configuration (above) Win10 was able to recognize and load the driver. What appears to prevent my configuration from loading is the fact that I selected 48K as my sample rate and my capture/playback channel mask includes more than two (2) channels. If I take your configuration and change the sample rate (c_srate/p_srate) to 48000 Windows will fail to load the driver. Likewise, setting the c_chmask/p_chmask to 7 (three channels) will also cause the driver to fail to load.
Quoted 9 lines
You mentioned there is an option in Win10 Sound Settings to "enable" UAC2 by default. I cannot find that option and I wonder if this is what is preventing me from changing the sample rate or channel mask? Could Windows be treating my audio gadget as a UAC1 device rather than a fully multi-channel audio device (even though the usbaudio2.sys driver is loaded)? Have you tried other configurations to verify your Win10 box supports more than two channels and a 44.1K sample rate? I look forward to your feedback and any suggestions you might offer.
I was able to reproduce the issue and prepared a patch below, please try it and see if it fixes the issue.
Maximum data rates that I used were (AFAIR) 8 channel 192kHz/32bit, but there is another issue with high data rate if someone uses so many channels, very high sampling frequency or sample size that data can't fit into allowed (by USB spec) max packet size of endpoint. In this case need to decrease bInterval of endpoint.
I test it in a high-performance audio case so always use a minimal possible bInterval (set to '1').
Of course in the future that need to be calculated dynamically depending on the UAC2 settingsPlease test patches below and see it it helps
---------------------------------------------8<---------------------------------------- From
51516435bbf2486574ec7bc9fd4726677cd474a4Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 22 Nov 2020 21:05:38 +0200
Subject: [PATCH] usb: gadget:f_uac2: always increase endpointmax_packet_size
by one audio slotAs per UAC2 Audio Data Formats spec (2.3.1.1 USB Packets), if the sampling rate is a constant, the allowable variation of number of audio slots per virtual frame is +/- 1 audio slot.
It means that endpoint should be able to accept/send +1 audio slot.
Previous endpoint
max_packet_sizecalculation code was adding sometimes +1 audio slot due toDIV_ROUND_UPbehaviour which was rounding up to closest integer. However this doesn't work if the numbers are divisible.It had no any impact with Linux hosts which ignore this issue, but in case of more strict Windows it caused rejected enumeration
Thus always add +1 audio slot to endpoint's max packet size
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)drivers/usb/gadget/function/f_uac2.c +1
-1@@ -486,7 +486,7 @@ static void set_ep_max_packet_size(const structf_uac2_opts *uac2_opts, } max_packet_size =num_channels (chmask) * ssize * -DIV_ROUND_UP (srate, factor / (1 << (ep_desc->bInterval -1 ))); + ((srate / (factor / (1 << (ep_desc->bInterval -1 )))) +1 );
Quoted 6 lines
Quoted 2 lines
drivers/usb/gadget/function/f_uac2.c +0
-0if (!is_playback && (uac2_opts->c_sync ==USB_ENDPOINT_SYNC_ASYNC ))max_packet_size = max_packet_size * FBACK_FREQ_MAX / 100;I think "reserving" 20% additional bandwidth is a bit extreme. In the end, the purpose is composate the drift which appears between different XTALs
That isn't extreme, I checked snd-usb ALSA driver before selecting number and it assumes the drift is between -25% ... +50% so +/-20% should work fine here (see snd_usb_handle_sync_urb() in sound/usb/endpoint.c).
Since our UAC2 gadget is a kind of "reversed" USB audio card, that also works fine with alsaloop connecting it to other audio cards in the system.
Quoted 4 lines
Allocating bandwidth for 1 more sample than the stream should require (49 instead of 48 for 48kHz with 4 bIntervals) should allow to compensate any realistic drift, don't you think ?
With cheap USB Audio adapters I saw +/-100 Hz drift on 44100 Hz
Anyway I don't think this can cause any problem since it is OK as per UAC2 spec and hosts (at least Linux) can handle such values
Thanks,
Ruslan
Quoted 19 lines
Quoted 19 lines
--
1.9.1---------------------------------------------8<---------------------------------------- From
c8f2f2b414af672ec40841e75fb1ea761ae29122Mon Sep 17 00:00:00 2001
From: Ruslan Bilovol <[email protected]>
Date: Sun, 16 Feb 2020 22:40:23 +0200
Subject: [PATCH] usb: gadget:f_uac2: change bInterval to 1 for
8ch/24bit/44.1kHzWith bInterval=4 one audio slot in case of 8ch/24bit/44.1kHz exeeds MaxPacket size of ISO endpoint (>1024) and can't be transferred in time.
USB spec ("5.6.3 Isochronous Transfer Packet Size Constraints") says if we need to transfer more than 1024 it is a high speed, high bandwidth endpoint and should have bInterval=1.
Quoted 5 lines
Quoted 2 lines
In the future, bInterval should be dynamically calculated depending on bandwith requirementes
Probably better to do it from the start to avoid breaking stuff for people who have been using the gadget with the current badnwidth so far
That is an existing issue with UAC2 driver since its initial implementation, I'm going to fix it in the near future as a separate patch. The issue is not related to the feedback synchronization support patch.
I actually developed a few patches (not published yet) to check user parameters and refuse to start UAC1/2 gadget if they are wrong (currently UAC1/2 gagdet accepts incorrect parameters and then silently fails to work correctly)
Thanks,
Ruslan
Quoted 31 lines
Quoted 30 lines
Signed-off-by Ruslan Bilovol <[email protected]>--- drivers/usb/gadget/function/f_uac2.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-)drivers/usb/gadget/function/f_uac2.c +2
-2@@ -281,7 +281,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO OUT Endpoint */ @@ -358,7 +358,7 @@ enum { .bmAttributes = USB_ENDPOINT_XFER_ISOC | USB_ENDPOINT_SYNC_ASYNC, .wMaxPacketSize = cpu_to_le16(1024), - .bInterval = 4, + .bInterval = 1, };/* CS AS ISO IN Endpoint */