On 03/25/14 22:56, Fred wrote:
On 03/25/14 20:42, Martin Pieuchot wrote:
This diff should help to avoid some races occurring when resuming
a machine with USB devices, please test!
Since the Host Controllers are halted during suspend and reset on
wakeup, all the connected USB devices are disconnected during a
suspend/resume cycle.
When a device is disconnected its pipes are aborted and closed,
right now this is done *after* resuming. But since the HC have
already been reset and some timeout might also fire, that generates
a lot of problems. So the diff below detaches the root hub of every
controller before suspending, so that no device stay "logically"
attached when suspended.
Make sure you have my last commit to all *hci.c before trying this,
and report any regression/success story.
</snipped>
I've just noticed that my ehci.c is at v 1.145 and not 1.146 - I'll
bring my source up to date and try again.
Sorry for the noise.
Fred
