#tyndur · 2010-01-03

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

  1. quokheya pcmattman :)
  2. pcmattmanHi quok
  3. quokseems we've had a nice conversation via the mailing list about hotplugging :)
  4. pcmattmanYup, fairly successful one too
  5. pcmattmanSoon we'll be able to get the word out about CDI with a couple of reference implementations for perusal - that'll be a good day
  6. pcmattmanImagine being able to just write some glue code and instantly have access to any CDI driver in your own hobby OS
  7. pcmattmanIt's like the ultimate "get out of jail free" card for kids like TM :P
  8. quokyup, I'm quite excited about it :)
  9. quokand it's still much easier to follow than UDI
  10. pcmattmanI still don't get why I would need to write a metalanguage whatsit-called and a full IPC system just to use a UDI driver
  11. SHyx0rmZAnd I still have to do my implementation in ASM (just for the lulz) ;-)
  12. SHyx0rmZDo you mind giving me a link to the mailing list, I think I missed it
  13. pcmattmanWell, that's just the thing: you *can* implement CDI glue in assembly... I highly doubt you can do that with UDI
  14. quokSHyx0rmZ: http://list.tyndur.org/cgi-bin/mailman/listinfo/cdi-devel
  15. SHyx0rmZThank you
  16. pcmattmanDamn, that was fast
  17. pcmattmanquok's still the fastest gun in the west
  18. quok:)
  19. SHyx0rmZYou can do everything in assembly if you have enough time ;)
  20. pcmattmanMy current wireless hotspot gives ~4s of lag ;)
  21. pcmattmanWell, yes... Assembly can be used for anything, but you'd surely go insane before you completed a UDI assembly implementation
  22. pcmattmanA CDI assembly implementation would be quite enjoyable
  23. quokI especially like the way we're going about supporting hotplug
  24. quokit's just informing the OS of a new device, and the rest of it is rather OS defined, so we really don't have to worr about it
  25. quokfor OSes that don't support hotplug, it becomes a no-op
  26. pcmattmanRight
  27. pcmattmanHotplug is a *massive* bonus for OS's which decide to support it
  28. SHyx0rmZSounds easy enough
  29. pcmattmanPedigree already, for instance, has all the hooks in place for new disk devices to pop up and be linked straight into the VFS
  30. SHyx0rmZMy OS is far from handling disks at all :D
  31. pcmattmanWell, CDI can help with that :P
  32. SHyx0rmZI hope so
  33. SHyx0rmZAt least CDI will help rolling out my own filesystem
  34. SHyx0rmZAlthough I doubt anyone will have a use for it
  35. pcmattmanDepends on if you make the tools for said filesystem portable
  36. pcmattman;)
  37. SHyx0rmZIt's rather because of the way it works
  38. SHyx0rmZIt doesn't support traditional folders & files
  39. SHyx0rmZOr, rather, it will
  40. SHyx0rmZI'm not quite finished yet :D
  41. pcmattmanHeh, I see
  42. pcmattmanOne disadvantage to CDI is no binary compatibility
  43. pcmattmanOtherwise you could write a CDI driver for your FS and try it on a variety of CDI-supporting operating systems without a recompile
  44. SHyx0rmZI don't think that's a huge tradeoff, I don't care if I have to recompile a driver :D
  45. SHyx0rmZBut...
  46. pcmattmanHehe
  47. SHyx0rmZWould there be a possibly to still make if binary compatible at some point?
  48. pcmattmanWell, binary compatibility is not an easy thing to implement
  49. SHyx0rmZOr does the way CDI works already permit that
  50. pcmattmanI personally would like to see it at some point, but it doesn't seeem to look like it will happen very soon
  51. SHyx0rmZMake a CDI-VM :D
  52. pcmattmanIt would require some work to be portable across various types of OS
  53. quokwe'd kind of need a stable API at least to begin thinking about ABI issues :)
  54. pcmattmanFor example, tyndur is a microkernel, so its binaries are different to a monolithic one
  55. pcmattmanAnd yeah, a stable API is important
  56. SHyx0rmZHm, would it be possible to implement a CDI-VM as a CDI driver? oO
  57. quokwe need to ensure all the hotplug stuff works quite nicely in userspace :)
  58. pcmattmanIf it were to be implemented it would be after CDI was not changing so much (only recently did a fundamental part of CDI get rewritten)
  59. pcmattmanWe probably also need to finish some of the in-design and proposed subsystems, such as graphics and sound
  60. SHyx0rmZWould be great if there were a working VESA driver
  61. SHyx0rmZFor Long Mode that is
  62. pcmattmanRight, and that can come after the graphics subsystem is designed
  63. quokI thougth someone was working on that, actually?
  64. SHyx0rmZI'm really looking forward at implementing CDI, but I'm sure it will take me several months to understand it
  65. SHyx0rmZquok, I think so too
  66. pcmattmanquok: hm? Who?
  67. SHyx0rmZI know there is a VESA module in development for Týndur, but I don't know if it's written for CDI
  68. quokhttp://git.tyndur.org/?p=tyndur/alex.git;a=shortlog;h=refs/heads/video-devel
  69. pcmattmanMaybe I can work on documentation while I'm not properly internet-connected
  70. quokOh, I haven't bothered to look and see if it's being written for CDI. I think it is, though
  71. SHyx0rmZhttp://git.tyndur.org/?p=tyndur/alex.git;a=tree;f=src/modules/cdi/vesa;h=80273050d2ab4d60be355d245f6d3e3bba2028b1;hb=e6857c8907d43b8bf8e011afa4cd95a453d88c5f
  72. quokthere you have it, then :)
  73. SHyx0rmZAh, it looks like I'll have to do the implementation for Long Mode though
  74. SHyx0rmZTýndur's current implementation uses vm86 -.-
  75. quokeww :)
  76. pcmattmanHmm, I don't know many implementations which don't
  77. SHyx0rmZI think there aren't many people out there who develop x86-64 OSs
  78. FreakyPenguinMoin
  79. taljethXanClic: Ich habe schon verstanden, dass du im Wesentlich das gleiche sagst wie ich. Nur hast du es halt später gesagt als ich. ;-)
  80. XanClicAber so viel auch nicht ;-)
  81. taljethJedenfalls hast du erstmal geschrieben bevor du meine Mail gelesen hast, oder? ;-)
  82. XanClicNö
  83. XanClicNe, falschrum *g*
  84. XanClicJa ;-)
  85. FreakyPenguintaljeth: Ich habe jetzt nach einem git pull folgendes Problem in ata:
  86. FreakyPenguinmain.c:202: Fehler: Redefinition von »__cdi_driver___COUNTER__«
  87. FreakyPenguinmain.c:201: Fehler: Vorherige Definition von »__cdi_driver___COUNTER__« war hier
  88. XanClicDa hat das glue() wohl nicht so richtig funktioniert
  89. FreakyPenguinTut das bei dir? ;-)
  90. FreakyPenguinSieht so aus ;-)
  91. FreakyPenguinAh gcc-4.3....
  92. FreakyPenguinDas glue hat wohl doch funktioniert ;-)
  93. XanClicAber zu gut *g*
  94. taljethHm, und wer hat jetzt recht? ;-)
  95. XanClicDer Op in #lost hat immer recht ;-)
  96. FreakyPenguinIm Zweifelsfall ich ;-)
  97. quokMorning folks
  98. XanClicGood evening ;-)
  99. quokit's 9:50am here :)
  100. FreakyPenguinMorning ;-)
  101. XanClic4:50 pm here, nearly weekend-end
  102. taljethMoin
  103. taljethThat works always ;-)
  104. XanClic*g*
  105. quokI just checked my email. I kind of expected hotplugging patches from Matt :)
  106. taljethMaybe tomorrow
  107. taljethOr you could just do it yourself ;-)
  108. quokPerhaps I will get to work on it later. Still need to hammer out a few VMM bugs in QuokOS.
  109. quokbut before I can do that, I need to go finish working in my son's bedroom and get it all cleaned up so he has somewhere to sleep tonight. :)
  110. quokI suppose the sooner I get started on that, the sooner I can get finished and then get back to OS dev
  111. taljethI suppose you're right ;-)
  112. taljethAs always
  113. quokdon't let my wife know that. she has this crazy idea that she's the one that is always right!
  114. taljeth*lol*
  115. FreakyPenguin*g*
  116. quokokay, off I go!
  117. FreakyPenguinhf ;-)
  118. taljethAnd of course XanClic is going to tell you that it's the party that is always right ;-)
  119. XanClic:-P
  120. XanClicYou are a party? I didn't got that until now...