Xiangfu Liu wrote: > we do have a bug in usb-mouse. > but I can't reproduce the bug. maybe Werner can give more information.
One bug ? I wish :-) So far, I see at least three different issues. But they may all have thing(s) in common. 1) The large number of RX timeout errors. There's definitely something wrong with the low-level timing there. 2) When I connect my Rii wireless keyboard and the LV3 and then reset the M1, I have about a 30-70% chance that the M1 will come out of reset with USB working, 30-70% of all the USB being completely dead, and a small percentage (maybe < 10%) of the keyboard working but the LV3 being dead. 3) Some mice that worked with the first release (from July) don't work with the current USB firmware. For 2), I looked for shared state allowing troubles at one port to foul up the other, but we don't seem to have anything that's intentionally shared. We should also add payload CRC checking. It may well be that 2) is caused by some lower-level corruption that then yields an a value we don't sanitize and that upsets Navre. (Of course, we should also sanitize all values we receive.) - Werner _______________________________________________ http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org IRC: #milkymist@Freenode
