Hi Felipe, Well, I've only tried on the one device that has a dwc3 controller. Given that the bug was "scheduling while atomic", the stack trace always had dwc3_gadget_ep_dequeue doing the sleeping, and it only happened when I had a gadget plugged in, I felt pretty solid that this was the same bug. Also, Matthias pointed out I was looking at an older series when I was figuring out which patches went together. Check my work, but the series seems to be here: https://patchwork.kernel.org/project/linux-usb/list/?series=42875 The first 3 patches are already backported. Then there were a couple extra I had missed before. So the complete set would look like: d53701067f04 usb: dwc3: gadget: check if dep->frame_number is still valid 3451f6affaef usb: dwc3: gadget: remove unnecessary dev_info() 1517265228b4 usb: dwc3: trace: log ep commands in hex 25abad6a0584 usb: dwc3: gadget: return errors from __dwc3_gadget_start_isoc() fec9095bdef4e usb: dwc3: gadget: remove wait_end_transfer d4f1afe5e896c usb: dwc3: gadget: move requests to cancelled_list d5443bbf5fc8f usb: dwc3: gadget: introduce cancelled_list 7746a8dfb3f9c usb: dwc3: gadget: extract dwc3_gadget_ep_skip_trbs() c3acd59014148 usb: dwc3: gadget: use num_trbs when skipping TRBs on ->dequeue() 09fe1f8d7e2f4 usb: dwc3: gadget: track number of TRBs per request 1a22ec6435806 usb: dwc3: gadget: combine unaligned and zero flags Though perhaps only the stuff up through "move requests to cancelled_list" is the important stuff... Felipe might be able to say better. -Evan
