#tyndur · 2010-06-25

51 Nachrichten · Seite 1 von 1 · Original-CSV herunterladen

  1. eddybXanClic: found something interesting while a realHW test, so I believe you may want to hear :P...
  2. eddybthe very first TD(SETUP, that is. the request is address change) actually gets no error, but the following IN TD gets stall+crc/timeout. also, all TDs that follows, get the same last error(even if they are all SETUP TDs).
  3. XanClicI would want to hear if I wouldn't have stopped working on USB for a couple of months, as said before ;-)
  4. eddybright
  5. eddyboops :D
  6. kwolfXanClic: You should just start working on it again ;-)
  7. XanClicI knew this would happen xD
  8. eddyb"made my uhci initialization a bit more fancy (read: a bit more like it should be), but still same problems in qemu" << I was preparing to say that, but it's a bit changed, actually
  9. eddybepic fail again
  10. eddybhowcome this can happen
  11. eddybat some point in the whole descriptor reading process
  12. eddybuhci(as in qemu's emulation) starts repeatedly processing my TD
  13. eddybwithout sending an IRQ(AFAIK)
  14. kwolfWell, anything that involves qemu should be easy to debug
  15. eddybthe "uhci: data: xx xx xx..." keeps coming up
  16. kwolfEasier than real hardware anyway
  17. eddybit has something to do with the async crap
  18. eddybfeeling better now :P ?
  19. eddyb(crap as in I really had no idea of what word to use :P)
  20. eddybdammit, I hate it when it happens
  21. eddybBAD TIMING
  22. eddybit could be as simple as uhci_async_find_td not finding the TD XD
  23. eddybI think uhci_async_link gets called after the async callback does...
  24. _eddyb_
  25. eddybeverything looks fine but it isn't is
  26. eddybno wonder I quit trying to find out what it was a few months ago
  27. pc[away]What does it do on real hw?
  28. pcmattmanNot that qemu is wrong, just interested in what physical hardware does
  29. pcmattmaneddyb: ^
  30. eddybback
  31. eddybwell
  32. eddybit seems the request to change address has an error
  33. eddyband the handshake fails
  34. eddybwith stall and crc/timeout bits set
  35. eddyball further TDs on that control pipe, no matter what type they are, get those two bits set
  36. eddybso I supposed the error isn't "cleaned" :)
  37. eddybpcmattman: ^^
  38. pcmattmanSounds fun :)
  39. eddyb*cleared
  40. eddyb*suppose
  41. eddybXD
  42. eddybgreat, now I can't find my CDRW :(
  43. eddybnevermind :)
  44. eddybI believe it happened before, but I failed to see the first SETUP not getting an error
  45. eddybpcmattman: here's something for you: "(NN) KERNELELF: Executing module iso9660" then "(NN) Elf::create: buffer at ........" then #PF :)
  46. eddybdoesn't happen always, but most of the time and it's really annoying
  47. pcmattmanPossibly out of ram
  48. pcmattmanAnyway, I'm off. Night
  49. eddybon 512MB?
  50. eddybI doubt that...
  51. eddybanyway, 'night