Hi Andrzej,
On 12/4/2018 9:54 PM, Andrzej Pietrasiewicz wrote:
> Hi All,
>
> I have recently noticed a problem with dwc2 and audio gadget.
>
> The expected behavior:
>
> The uac2 function acts as a kind of a pass-through.
> For example to play audio from USB host on an Odroid U3 the following command
> is issued at the host:
>
> aplay -D plughw:CARD=Gadget,DEV=0 /usr/share/sounds/alsa/Front_Center.wav
>
> while the following command is used at the device:
>
> arecord -f dat -t wav -D plughw:CARD=UAC2Gadget,DEV=0 \|
> aplay -D default:CARD=OdroidU3
>
> At the device we "record" from the audio device provided by the usb gadget
> and pipe the resulting stream to the actual hardware device present in
> the chip on board.
>
> What actually happens:
>
> On recent kernels this does not work in a stable manner and sometimes
> the kernel panics. I traced the problem back and it seems that the last
> kernel without the below problems is 4.7.
>
> I am using a compiled-in, legacy audio gadget with only uac2 function
> selected in Kconfig.
>
> 4.7 works
> 4.8 kernel crashes at boot time
> (bisect identifies 381fc8f8228923026b3d75c8230fa2ee4d688f32
> "usb: dwc2: gadget: Add Incomplete ISO IN/OUT Interrupt handlers"),
> compiling as a module does not help (crash at module load time),
> composing the gagdet with configfs does not help either.
>
> 4.9 kernel crashes at boot time
> 4.10 kernel crashes at boot time
> 4.11 kernel crashes at boot time
> 4.12 kernel crashes at boot time
> 4.13 kernel crashes at boot time
> 4.14 kernel usually crashes at boot time, but sometimes it boots,
> if it boots, it crashes the same way as at boot time when
> the gadget is actually used (aplay at host and arecord | aplay
> at the device)
> 4.15 kernel sometimes crashes at boot time, but otherwise it boots,
> if it boots, it crashes the same way as at boot time when
> the gadget is actually used (aplay at host and arecord | aplay
> at the device)
> 4.16 kernel crashes at boot time
> 4.17 kernel sometimes crashes at boot time, but otherwise it boots,
> if it boots, it crashes the same way as at boot time when
> the gadget is actually used (aplay at host and arecord | aplay
> at the device)
> 4.18 kernel sometimes crashes at boot time or even silently freezes,
> but otherwise it boots, if it boots, it crashes the same way as
> at boot time when the gadget is actually used (aplay at host
> and arecord | aplay at the device)
> 4.19 and 4.20-rc4
> kernel boots, but after some time the gadget stops working:
>
> at the device with 4.19:
>
> # arecord -f dat -t wav -D plughw:CARD=UAC2Gadget,DEV=0 | aplay -D default:CARD=OdroidU3
> Recording WAVE 'stdin' : Signed 16 bit Little Endian, Rate 48000 Hz, Stereo
> Playing WAVE 'stdin' : Signed 16 bit Little Endian, Rate 48000 Hz, Stereo
> [ 26.245056] max98090 1-0010: PLL unlocked
> underrun!!! (at least 1256.128 ms long)
> [ 29.363400] max98090 1-0010: PLL unlocked
> underrun!!! (at least 930.612 ms long)
> underrun!!! (at least 447.733 ms long)
> [ 34.475390] max98090 1-0010: PLL unlocked
> underrun!!! (at least 319.622 ms long)
> [ 36.678554] max98090 1-0010: PLL unlocked
> underrun!!! (at least 611.690 ms long)
> [ 39.153660] max98090 1-0010: PLL unlocked
> underrun!!! (at least 724.196 ms long)
> [ 41.741488] max98090 1-0010: PLL unlocked
> underrun!!! (at least 695.816 ms long)
> [ 44.428490] max98090 1-0010: PLL unlocked
> underrun!!! (at least 795.061 ms long)
> [ 47.106276] max98090 1-0010: PLL unlocked
> underrun!!! (at least 760.757 ms long)
> underrun!!! (at least 2685.446 ms long)
> [ 54.288479] max98090 1-0010: PLL unlocked
> underrun!!! (at least 1591.909 ms long)
> [ 57.764445] max98090 1-0010: PLL unlocked
> underrun!!! (at least 223.942 ms long)
> underrun!!! (at least 237.751 ms long)
> [ 61.974963] max98090 1-0010: PLL unlocked
> underrun!!! (at least 677.925 ms long)
> [ 64.645406] max98090 1-0010: PLL unlocked
> underrun!!! (at least 550.651 ms long)
> [ 67.079379] max98090 1-0010: PLL unlocked
> underrun!!! (at least 1575.582 ms long)
> [ 70.521155] max98090 1-0010: PLL unlocked
> underrun!!! (at least 832.604 ms long)
> [ 73.219814] max98090 1-0010: PLL unlocked
> underrun!!! (at least 721.516 ms long)
>
Sometime I also see "underrun!!!" but I think it's not related to dwc2.
>
> while at the host:
>
> $ aplay -D plughw:CARD=Gadget,DEV=0 /usr/share/sounds/alsa/Front_Center.wav
> Playing WAVE '/usr/share/sounds/alsa/Front_Center.wav' : Signed 16 bit Little Endian, Rate 48000 Hz, Mono
>
> .... several times ok ....
>
> $ aplay -D plughw:CARD=Gadget,DEV=0 /usr/share/sounds/alsa/Front_Center.wav
> Playing WAVE '/usr/share/sounds/alsa/Front_Center.wav' : Signed 16 bit Little Endian, Rate 48000 Hz, Mono
> aplay: set_params:1297: Unable to install hw params:
> ACCESS: RW_INTERLEAVED
> FORMAT: S16_LE
> SUBFORMAT: STD
> SAMPLE_BITS: 16
> FRAME_BITS: 16
> CHANNELS: 1
> RATE: 48000
> PERIOD_TIME: 125000
> PERIOD_SIZE: 6000
> PERIOD_BYTES: 12000
> PERIODS: 4
> BUFFER_TIME: 500000
> BUFFER_SIZE: 24000
> BUFFER_BYTES: 48000
> TICK_TIME: 0
>
> Any ideas on how to test it further to identify the problem?
>
> Definitely the commit identified with bisect was a problem at least
> until 4.13. Perhaps it still is the root cause, but I can't tell.
> It is difficult to revert it, because there were later changes which
> depend on it.
>
> Andrzej
>
Which DMA (BDMA or DDMA) or Slave mode in dwc2 you use for testing?
I run tests on 4.20.0-rc4 in DDMA. By default IN ISOC traffic failed
because of BNA interrupts. It's happen because UAC2 requests count set
by default to 2. Our core and driver can't work in DDMA with descriptor
list length equal to 2. It's not possible on time prepare next
descriptors to avoid BNA interrupt.
By changing UAC2_DEF_REQ_NUM to 4 all audio gadget tests passed smoothly.
Could you please apply this patch and run tests in DDMA mode:
diff --git a/drivers/usb/gadget/function/u_uac2.h
b/drivers/usb/gadget/function/u_uac2.h
index 8362ee572e1e..5e649259ab76 100644
--- a/drivers/usb/gadget/function/u_uac2.h
+++ b/drivers/usb/gadget/function/u_uac2.h
@@ -21,7 +21,7 @@
#define UAC2_DEF_CCHMASK 0x3
#define UAC2_DEF_CSRATE 64000
#define UAC2_DEF_CSSIZE 2
-#define UAC2_DEF_REQ_NUM 2
+#define UAC2_DEF_REQ_NUM 4
struct f_uac2_opts {
struct usb_function_instance func_inst;
If it will OK on your side also then will switch to BDMA mode and debug.
Thanks,
Minas
