Hello Tom, On 08/09/18(Sat) 12:07, Tom Murphy wrote: > On Thu, Sep 06, 2018 at 01:06:50PM -0300, Martin Pieuchot wrote: > > Tom, as I said previously you've found a race in the ugen(4) driver. > > > > That's the symptom: > > > > > [...] > > > usb_detach_wait: ugen1 didn't detach > > > > To be able to understand which race we are chasing, could you rebuild a > > kernel with UGEN_DEBUG defined and set `ugendebug' to 6? > > OK here's the output per each step. Below that will be the dmesg and the > backtrace.
Thanks a lot, but I need the same outputs with both UGEN_DEBUG and XHCI_DEBUG, and of course `ugendebug' set to 6 :) The interaction between ugen(4) the stack and xhci(4) is what will tell us where is the use-after-free :)