headers
gmcnutt | 25 Feb 2011 18:31
Favicon

panic in ehci_abort_xfer

By unplugging a certain usb device I can reproducibly hit the panic for 
intr_context in this function. Is there supposed to be something that ensures we 
can never call ehci_abort_xfer on a thread that preempts the soft interrupt? Is 
the soft interrupt not supposed to be pre-emptible by the usb_task_thread? I'm 
running on an embedded MIPS platform.
Permalink | Reply |
headers
gmcnutt | 25 Feb 2011 20:54
Favicon

Re: panic in ehci_abort_xfer

gmcnutt <gmcnutt <at> cradlepoint.com> writes:

I worked around my problem by identifying a xfer callback in a driver that was
doing synch'd transfers to a hub and deferring these to a timer.

But my original questions remain unanswered: are softintr's preemptible? If so,
isn't that panic bound to happen in some cases?
Permalink | Reply |

Gmane