Hi Dan, On Thu, Dec 08, 2022, Dan Scally wrote: > Good morning > > On 08/12/2022 06:11, Felipe Balbi wrote: > > Hi, > > > > Dan Scally <dan.scally@xxxxxxxxxxxxxxxx> writes: > > > I'm having an issue with DWC3 which I'm hoping you might be able to shed > > > light on. I'm using the UVC gadget function to stream video from an > > > imx8mp platform, which works just fine. Once I have stopped streaming > > > however and after some time has passed (the exact duration seems to vary > > > quite a lot from 1-2 minutes to 15+ mins) I get a kernel warning like so: > > As Greg mentioned, please add Thinh, the new maintainer, to the loop. > > > Apologies to both of you; for some reason I thought you had taken over from > Thinh rather than the other way around. Thanks for the heads up Greg. > > > > > > [ 737.996842] configfs-gadget gadget.0: uvc: uvc_function_disable() > > > [ 738.519582] configfs-gadget gadget.0: uvc: uvc_function_set_alt(0, 0) > > > [ 738.526060] configfs-gadget gadget.0: uvc: reset UVC interrupt endpoint > > > [ 738.532713] ------------[ cut here ]------------ > > > [ 738.537683] dwc3 38100000.usb: No resource for ep1in > > > [ 738.542712] WARNING: CPU: 0 PID: 645 at drivers/usb/dwc3/gadget.c:384 > > > dwc3_send_gadget_ep_cmd+0x478/0x580 > > > [ 738.552314] Modules linked in: > > > [ 738.555402] CPU: 0 PID: 645 Comm: irq/208-dwc3 Not tainted > > > 5.19.0-00034-gf017ce943b32 #82 > > > [ 738.563601] Hardware name: Polyhex Debix Model A i.MX8MPlus board (DT) > > > [ 738.570145] pstate: 600000c5 (nZCv daIF -PAN -UAO -TCO -DIT -SSBS > > > BTYPE=--) > > > [ 738.577128] pc : dwc3_send_gadget_ep_cmd+0x478/0x580 > > > [ 738.582116] lr : dwc3_send_gadget_ep_cmd+0x478/0x580 > > > [ 738.587102] sp : ffffffc00c3c39e0 > > > [ 738.590433] x29: ffffffc00c3c39e0 x28: 0000000000031006 x27: > > > 00000000ffffffea > > > [ 738.597618] x26: 0000000000000006 x25: ffffff8004f32880 x24: > > > ffffffc00c3c3abc > > > [ 738.604801] x23: 0000000000000406 x22: ffffffffffff3f00 x21: > > > ffffffffffff3f0c > > > [ 738.611984] x20: ffffff8003872600 x19: 0000000000000001 x18: > > > ffffffffffffffff > > > [ 738.619166] x17: 000000000000001c x16: 00000000c99c42a5 x15: > > > ffffffc08c3c36b7 > > > [ 738.626347] x14: 0000000000000000 x13: 6e6931706520726f x12: > > > 6620656372756f73 > > > [ 738.633528] x11: 00000000000c02a7 x10: ffffffc009af1ac0 x9 : > > > ffffffc0080c42bc > > > [ 738.640710] x8 : 00000000ffffefff x7 : ffffffc009af1ac0 x6 : > > > 0000000000000000 > > > [ 738.647891] x5 : ffffff807fb2eb08 x4 : 0000000000000000 x3 : > > > 0000000000000027 > > > [ 738.655072] x2 : 0000000000000000 x1 : 0000000000000000 x0 : > > > ffffff80158a8000 > > > [ 738.662255] Call trace: > > > [ 738.664721] dwc3_send_gadget_ep_cmd+0x478/0x580 > > > [ 738.669362] __dwc3_gadget_ep_enable+0x4f4/0x714 > > > [ 738.674004] dwc3_gadget_ep_enable+0x6c/0x15c > > > [ 738.678382] usb_ep_enable+0x4c/0x1bc > > > [ 738.682067] uvc_function_set_alt+0xcc/0x264 > > > [ 738.686362] composite_setup+0x7ec/0x195c > > > [ 738.690395] configfs_composite_setup+0x90/0xc0 > > > [ 738.694950] dwc3_ep0_interrupt+0x834/0x9e0 > > > [ 738.699156] dwc3_thread_interrupt+0x994/0xd70 > > > [ 738.703624] irq_thread_fn+0x38/0xa4 > > > [ 738.707221] irq_thread+0x184/0x230 > > > [ 738.710733] kthread+0x118/0x120 > > > [ 738.713981] ret_from_fork+0x10/0x20 > > > [ 738.717582] irq event stamp: 41021 > > > [ 738.720998] hardirqs last enabled at (41019): [<ffffffc008082858>] > > > finish_task_switch.isra.0+0xe8/0x264 > > > [ 738.730501] hardirqs last disabled at (41021): [<ffffffc008f15058>] > > > _raw_spin_lock_irqsave+0xc4/0x170 > > > [ 738.739747] softirqs last enabled at (41014): [<ffffffc008a25b90>] > > > dwc3_thread_interrupt+0x1c0/0xd70 > > > [ 738.748988] softirqs last disabled at (41020): [<ffffffc008a259dc>] > > > dwc3_thread_interrupt+0xc/0xd70 > > > > > > > > > ep1in in this instance refers to a Status Interrupt Endpoint under the > > > UVC specification, which is being "reset" by uvc_function_set_alt() > > > against the VideoControl interface (meaning a call to usb_ep_disable() > > > followed by usb_ep_enable(), see [1]). The isochronous endpoint that > > > data is streamed across is treated similarly in the same function (when > > > called for the VideoStreaming interface) and never seems to show the > > > same warning and as far as I can tell the operation ought to be safe, so > > > I'm struggling to see anything in the f_uvc code that's misbehaving to > > > cause the problem. I wondered if you might be able to take a look at the > > > trace and regdump of the dwc3 (which was taken immediately after the > > > warning is thrown) and see if that sheds any light on what might be > > > going on? > > Best way forward is to collect dwc3 trace points, so we can see what the > > HW is doing. For details, see [1]. > > > > [1] https://urldefense.com/v3/__https://kernel.org/doc/html/latest/driver-api/usb/dwc3.html*reporting-bugs__;Iw!!A4F2R9G_pg!Zb5Cx7r3McJ2FCbH_oK14MqINrzor9SihvyA9qFxrlmzU-d_VcBhZYs2jTpWMNE8oxhIH8PMIdk0Nhqj8xBEmkEwe9H6$ > > > Both the trace and regdump should be in the attached .tar.gz > > > Looks like we're missing the check to ignore delay End Transfer. Can you try this to see if it fixes your issue? diff --git a/drivers/usb/dwc3/gadget.c b/drivers/usb/dwc3/gadget.c index ebc0e147cc71..65500246323b 100644 --- a/drivers/usb/dwc3/gadget.c +++ b/drivers/usb/dwc3/gadget.c @@ -2008,6 +2008,7 @@ static int __dwc3_stop_active_transfer(struct dwc3_ep *dep, bool force, bool int else if (!ret) dep->flags |= DWC3_EP_END_TRANSFER_PENDING; + dep->flags &= ~DWC3_EP_DELAY_STOP; return ret; } @@ -4279,8 +4280,10 @@ void dwc3_stop_active_transfer(struct dwc3_ep *dep, bool force, dwc->hiber_state != DWC3_HIBER_ENTERING_DISCONNECTED) return; + if (interrupt && (dep->flags & DWC3_EP_DELAY_STOP)) + return; + if (!(dep->flags & DWC3_EP_TRANSFER_STARTED) || - (dep->flags & DWC3_EP_DELAY_STOP) || (dep->flags & DWC3_EP_END_TRANSFER_PENDING)) return; Thanks, Thinh
