Closed selltc closed 7 years ago
Tim, is this fixed already? I thought I address this.
Yeah, I thought you did too, but I will admit I haven't recently gone thru all of the github issues and updated them to reflect what has been submitted.
Agreed, I'm spending some time trying to clean up the Kanboard, I will look up the patch and close this one as well.
Thanks, David Kershner
Closing fixed with ca1cbf9
KanBoard-25114
After commit 87241ab85930519a962a567de58d959d3029b419 (staging-next), it looks like we are now sending double-responses for DEVICE_CREATE, DEVICE_CHANGESTATE, and DEVICE_DESTROY.
Look at the change to device_epilog():
The key point is that chipset_device_create(), chipset_device_resume(), chipset_device_pause(), and chipset_device_destroy() all have device_responder() (or device_changestate_responder(), which also signalinserts a response) calls embedded in them. Yet when we break out of the above case, we end up at out_respond, which also calls device_responder().