#tyndur · 2010-03-31

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

  1. kwolfMorning pcmattman
  2. pcmattmanHi kw
  3. pcmattmaner
  4. pcmattmankwolf:
  5. kwolfIs the v2 of the e1000 patch still on your list?
  6. pcmattmanI've completely forgotten what I was going to do in v2 unfortunately
  7. kwolfDon't you have IRC logs?
  8. kwolfOr just take this one: http://irc.tyndur.org/index.php?pageSize=1000&channelId=4
  9. kwolfOf course, if you are busy with other things, there is no need to hurry. I just want to avoid to drop the patch because everyone forgets about it.
  10. pcmattmanI have been using the e1000 quite extensively in VirtualBox at work and haven't run into any issues with it there
  11. kwolfOn real hardware, too?
  12. pcmattmanI haven't yet had a chance to try real hardware
  13. kwolfOk. A working driver in VBox is nice progress for itself anyway.
  14. JarvixHi
  15. kwolfHi Jarvix
  16. Jarvixhi
  17. eddybXanClic: do you happen to know anything about usb master storage?
  18. XanClicUnless you mean mass storage, no
  19. eddyb*mass
  20. XanClicThen I do ;-9
  21. XanClics/9/)/
  22. eddybi don't know why i say master XD
  23. eddybi've got my driver read some sectors
  24. eddybbut
  25. eddybi found out that there are some not-so-predictable failures
  26. eddybas in
  27. XanClicWell, my drivers (CB/CBI and bulk-only) only work on emulators (but there also with real hardware)
  28. eddybi'm talking about qemu(bulk/SCSI)
  29. XanClicOn completely real hardware (also real HC) I'm unable to read sectors
  30. XanClicok
  31. XanClicHm, I guess I never had any problems with qemu's mass storage devices
  32. eddybi do CBW out packet(endpoint 2), then i do an in to read the data and another for CSW(endpoint 1)
  33. XanClicShould be correct
  34. eddybfrom time to time, i got the IN packet deferred
  35. XanClicStalled?
  36. eddybit seems to be internal
  37. XanClicWell, an NAK wouldn't stop the transfer...
  38. eddybi mean i don't get an irq
  39. eddybit simply retries (i think) the packet later
  40. XanClicSounds like an NAK
  41. XanClicAnd that would be completely fine
  42. XanClicDid you check the status field?
  43. eddybyou don't get me
  44. eddybinside qemu
  45. XanClicYes
  46. eddybusb_msd_handle_data is called twice
  47. eddybbefore i get an irq
  48. eddybwell, i think it's something normal
  49. XanClicWell, all I can say is that I have no problems with that *g*
  50. eddybexcept for teh fact that it sets mode to CSW
  51. XanClicThe only thing was that qemu's bulk-only devices didn't handle the class-specific control requests correctly
  52. eddyband that makes no sense
  53. eddybif (s->data_len == 0) s->mode = USB_MSDM_CSW;
  54. eddybs->usb_len = 0;
  55. eddybso usb_len gets set to 0, while mode may be not set to CSW
  56. eddyb(inside an if statement)
  57. eddybthen if (s->usb_len) deffer packet
  58. eddybit simply shouldn't get there :|
  59. XanClicWell, have you tried the qemu mailing list? *g*
  60. XanClicI think I could help you only if I experienced the same problem, but I didn't
  61. eddybhmm
  62. eddybokey, i'll try with stable qemu
  63. eddybmaybe it's the same problem with the HID not NAKing when not having changed data(aka new kepress) in qemu git
  64. eddybthis is strange
  65. eddybreinstalling qemu did absolutely nothing :|
  66. eddybfunny, it still happens with stable qemu...
  67. eddybalso, it happens mostly on sectors 63 and 64
  68. eddybXanClic: any code I can check out?
  69. XanClicMy USB code?
  70. XanClicWell, if you're brave enough *g*
  71. eddybmass storage
  72. XanClichttp://gitorious.org/paloxena/kernel3/blobs/master/src/kernel/drivers/usb-device/msd-cbi.c <— CBI
  73. XanClichttp://gitorious.org/paloxena/kernel3/blobs/master/src/kernel/drivers/usb-device/msd-bo.c <— Bulk-only
  74. eddybwhat's CBI?
  75. XanClicControl/Bulk/Interrupt
  76. eddybah
  77. XanClicThough I don't use the interrupt part *g*
  78. XanClichttp://gitorious.org/paloxena/kernel3/blobs/master/src/kernel/drivers/bus/scsi.c <— and my SCSI code
  79. eddybusb_bulk_transfer is sync?
  80. eddybah, and what's tag for?
  81. eddybi set it to 0
  82. XanClicYou may use it on your own
  83. XanClicI set it to 42
  84. XanClicI guess it's used to see if the CSW goes with the CBW or anything like that
  85. eddybi think i see
  86. eddybthe data arrives just after the packet get deferred, but before the handler function returns...
  87. eddybqemu code it's a big load of mess
  88. eddybis there any way i can get to talk so some qemu developer(s) ?
  89. XanClicYes
  90. XanClickwolf is one ;-)
  91. eddybbecause i know way too less about the code to start digging myself
  92. eddybsilly me XD
  93. eddybsadly, he's not around :(