I have a reproducible complete train-wreck of a kernel panic that happens so quickly it's hard to debug and it's related to GenericUSBXHCI and the ebook reader/manager application Calibre. This kernel panic happens with calibre-debug and calibre itself, and happens even when there is no device on the bus that Calibre would want to talk to. It's really something spectacular.
I am thinking that this is probably a bug in implementation of usbmuxd/libimobiledevice in Calibre but the developer of Calibre is not known for, shall we say, enthusiastic responses to criticism in his software . I may have another kernel panic as a result of this driver in the software installer and subsequent use of the driver to control a LeapMotion device. I haven't attempted to reproduce in a while, as I bought a CalDigit combo card but after Yosemite the Caldigit kernel extensions for USB have had issues of their own so I've yanked them, yanked the card, and I'm trying to make do with the Etron devices
So I'm back and considering either getting the better CalDigit card (the one with two internal ESATA connectoids and two external in addition to 2 USB 3.0 that doesn't lose it's effing mind when a USB 3.0 hub is plugged in) or going with another option entirely like upgrading my board to something with on-board Intel USB 3.0.
tl;dr is my reproducible panic something that can be addressed in GenericUSBXHCI or do I need to suck it up and go to battle with Kovid Goyal after swallowing enough anti-anxiety medication to not take his tantrums too seriously? With the original author/maintainer stepping out and two other folks starting to dig in I don't want to get in the way of that but if this is an interesting bug that you think will help get your bearings, I'd be happy to guineapig any suggestions and possible fixes.
My hardware is z68x-ud3h-b3 on UEFI h1l, I boot with Clover and don't have Windows (but if required I can pull down that preview release of Windows 10 and put it on a spare drive or VM.)
FWIW I'm not opposed to learning more about USB myself so I'm not merely volunteering to run builds and whine about them. I'm GMT-5/US-en and familiar with git/svn/rcs/mercurial and arcane knowledge like unified diffs and dtrace.
No hard feelings if nobody wants to run with this with me, I'm fine just throwing hardware and money at it if that's all it takes.
 For example, there is what has to be by now a 200 page thread on launchpad of him arguing why a security vulnerability in his software isn't important and him actually telling well-reguarded researchers and developers that know how to address the issues to go screw.