a guest Dec 11th, 2011 677 Never
Not a member of Pastebin yet? Sign Up, it unlocks many cool features!
  1. [22:40:48] <resetnow_> hi all
  2. [22:41:09] <resetnow_> has anybody tried latest l4t release? 12 beta
  3. [22:41:31] <gordan> resetnow: I tried the Xorg driver from it.
  4. [22:41:40] <resetnow_> yes, i too
  5. [22:41:43] <resetnow_> it works
  6. [22:41:48] <marvin24_DT_> here too
  7. [22:41:51] alex_ac100 [~alex@] зашел на #ac100
  8. [22:41:55] <marvin24_DT_> but not the video decoding ...
  9. [22:41:59] <gordan> Suffers from the same bitmap font corruption bug that manifests on 12 alpha driver.
  10. [22:42:10] <resetnow_> but i'm trying to get hw acceleration of the video
  11. [22:42:15] <resetnow_> with gstreamer
  12. [22:42:32] <resetnow_> so, i copied libopenmax-gst
  13. [22:42:44] <marvin24_DT_> gordan: I checked your url in firefox and I saw no corruption
  14. [22:43:08] leonro [] вышел из IRC: Ping timeout: 240 seconds
  15. [22:43:15] <resetnow_> and when i try to play video file, i get error /dev/tegra_avp: no such file or directory
  16. [22:43:31] <marvin24_DT_> resetnow_: same here
  17. [22:43:40] <marvin24_DT_> maybe some codec files are missing
  18. [22:43:48] <resetnow_> i get the same error in flash
  19. [22:44:01] leonro [] зашел на #ac100
  20. [22:44:12] <resetnow_> does this mean that some flash video acceleration may be available?
  21. [22:44:15] <gordan> It's possible that F13/FF3.6 specifically triggers it.
  22. [22:44:15] <marvin24_DT_> the doc mentions something about restrictedmodules.tbz2
  23. [22:44:19] <marvin24_DT_> which is not included
  24. [22:44:33] <gordan> And I suspect you're not running F13/FF3.6. :)
  25. [22:44:44] <marvin24_DT_> gordan: yes
  26. [22:44:46] <marvin24_DT_> right
  27. [22:44:57] <marvin24_DT_> that makes it hard to reproduce
  28. [22:45:02] <gordan> Not at all.
  29. [22:45:10] <marvin24_DT_> did fedora reached F16 now?
  30. [22:45:26] <gordan> I have 3 different F13 rootfs-es and 3 different AC100s and I can trigger the bug on all of them every time.
  31. [22:45:49] <gordan> It may be hard to reproduce using a different distro/rootfs/driver/FF, sure.
  32. [22:46:17] <gordan> But that doesn't invalidate the bug report validity.
  33. [22:47:02] <gordan> Where does cpufreq_policy get defined?
  34. [22:47:05] <resetnow_> so, may be some request for restrictedmodules.tbz2 to nvidia developers will help?
  35. [22:47:18] <marvin24_DT_> resetnow_: you can ask in the forum
  36. [22:47:23] <gordan> cpu-tegra.c uses it to check which table it should use.
  37. [22:47:44] <gordan> But I can't see where cpufreq_policy is getting defined/populated.
  38. [22:48:52] <marvin24_DT_> juliank: no corruptions for the last three times and the same kernel ...
  39. [22:49:46] vanger [] вышел из IRC: Ping timeout: 252 seconds
  40. [22:51:49] <juliank> resetnow_: Video acceleration worked in the alpha release, but not in the beta.
  41. [22:52:24] <resetnow_> oh, so buggy
  42. [22:52:47] <juliank> The problem with the beta release could be that it is targeted at the nv-tegra 2.6.36 tree and thus does not work with our chromeos trees due to differences
  43. [22:53:02] <resetnow_> restrictedmodules.tbz2 will not help?
  44. [22:53:49] <resetnow_> thank you for answer
  45. [22:54:49] <juliank> NVIDIA is currently not even complying with the gst-openmax license as far as I can tell, because they only provide a binary build and no object files or any reference to the source code.
  46. [22:55:42] <juliank> resetnow_: It does not work with the codec files from the Alpha release, I can't tell you whether the new ones are different.
  47. [22:56:54] <resetnow_> "codec files" means nvmm_h264dec.axf, nvmm_mp4dec.axf, etc.?
  48. [22:57:20] <marvin24_DT_> yes
  49. [22:57:34] <marvin24_DT_> but there are some smaller .bins instead ;-)
  50. [22:58:24] <resetnow_> hm, so estrictedmodules.tbz2 can contain some different codecs
  51. [22:58:30] <juliank> which might contain VP8, Ogg Vorbis, etc. support.
  52. [22:58:48] <resetnow_> i'll ask in the foeum
  53. [22:59:40] <juliank> We might need to merge nvavp from the nv-tegra tree to get things working  --;a=tree;f=drivers/media/video/tegra/nvavp;h=bba13489d4ee0753d3105a4559a8354d6cdc2f8a;hb=refs/heads/android-tegra-2.6.36
  54. [23:01:19] <marvin24_DT_> juliank: mmh
  55. [23:01:31] <marvin24_DT_> we do not even have this directory
  56. [23:01:37] Vraister [] зашел на #ac100
  57. [23:01:53] <juliank> marvin24_DT_: Yes, it's a new driver.
  58. [23:02:01] <juliank> It exposes /dev/tegra_avpchannel
  59. [23:05:30] <marvin24_DT_> mmh,  sprintf(fw_ucode_file, "nvavp_vid_ucode.bin");
  60. [23:05:34] socketa4techx7 [5dbece2d@gateway/web/freenode/ip.] зашел на #ac100
  61. [23:05:37] <marvin24_DT_> we don't have this file
  62. [23:07:18] <juliank> marvin24_DT_: Then it's likely unrelated to the problem.
  63. [23:07:43] <alex_ac100> not only this link, but also some code in fuse, as it seems that last nvidia modules queries hardware revision via that interface
  64. [23:08:13] <marvin24_DT_> alex_ac100: but the message says it is normal for tegra2 to fail
  65. [23:09:24] <marvin24_DT_> I guess we need this file (and the driver), but it is in the secret blob
  66. [23:09:44] <marvin24_DT_> also nvavp_os.bin
  67. [23:11:10] <marvin24_DT_> mmh, maybe it is called nvrm_avp ...
  68. [23:11:20] <marvin24_DT_> crap!
  69. [23:11:43] <resetnow_> rename will help, lol?)
  70. [23:12:21] <marvin24_DT_>;a=commitdiff;h=5c5bab1dbedcf72f95257f14f6f549357c82962c
  71. [23:13:56] <juliank> marvin24_DT_: What do you want to tell us with this?
  72. [23:14:12] <juliank> That's a Tegra3 change.
  73. [23:14:13] <marvin24_DT_> this driver loads our binary blob
  74. [23:14:27] <marvin24_DT_> I mean drivers/media/video/tegra/avp
  75. [23:15:27] <juliank> marvin24_DT_: Yes, I know. And our kernel loads nvrm_avp.bin as well
  76. [23:15:40] <marvin24_DT_> does it?
  77. [23:15:52] <marvin24_DT_> I mean in reality ...
  78. [23:16:37] epicmonkey [~epicmonke@] зашел на #ac100
  79. [23:17:06] <juliank> marvin24_DT_: If it didn't, it wouldn't work with the alpha1 userspace either
  80. [23:17:12] <resetnow_> nvrm_avp.bin? l4t package has, isn't it?
  81. [23:17:21] <resetnow_> * only
  82. [23:17:28] <juliank> And WebM video acceleration works happily there.
  83. [23:17:45] <juliank> (in chromium)
  84. [23:17:50] <marvin24_DT_> with beta1?
  85. [23:18:11] <juliank> marvin24_DT_: No with alpha1 it works happily, with beta1 everything's broken.
  86. [23:18:19] <marvin24_DT_> ok
  87. [23:19:01] <marvin24_DT_> I guess the chromeos kernel is already too far away from the nv-tegra one
  88. [23:20:51] <juliank> marvin24_DT_: Alpha1 was cool because it was targeted at the chromeos kernel. Beta isn't. And the final release won't support harmony anymore.
  89. [23:21:32] <juliank> The "Additionally, developers should note that support for Harmony devices will be deprecated following this release.  Support for Harmony will transition to community-supported Linux kernels, and Ventana will be the Tegra 2 reference platform." part of the announcement
  90. [23:21:48] <marvin24_DT_> yeah, I read that also
  91. [23:21:59] <marvin24_DT_> but ok, harmony or ventana
  92. [23:22:09] <marvin24_DT_> not so much difference I think
  93. [23:22:51] <marvin24_DT_> ventana is also tegra2
  94. [23:23:20] mase76 [] вышел из IRC: Read error: No route to host
  95. [23:23:29] <juliank> And the xv support is there now, but it's not working correctly from what I can tell. It crashes the system if the video is too large and makes the colors look bad.
  96. [23:24:49] <marvin24_DT> I wonder if nv is willing to support such "strange" hw/software configs
  97. [23:25:54] <marvin24_DT> I also wonder if google has dropped tegra development on chromeos
  98. [23:26:06] <marvin24_DT> seems they concentrate on atoms now
  99. [23:26:19] <gordan> Atoms?! WTF?
  100. [23:26:42] <juliank> marvin24_DT: Given that the last ARM commit was yesterday, that seems unlikely
  101. [23:27:04] <marvin24_DT> maybe
  102. [23:27:24] <marvin24_DT> maybe they just "drive on two roads"
  103. [23:27:44] <juliank> marvin24_DT: The chromebooks currently all have atom CPUs.
  104. [23:27:50] <marvin24_DT> given that the arm chromebook never materialized
  105. [23:28:08] <juliank> The chrome tablets are planned to have ARM cpu
  106. [23:28:11] <marvin24_DT> on the other hand, the atom ones flopped
  107. [23:28:41] <gordan> There is a certain amount of irony in the fact that Toshiba pretty much washed their hands of the AC100 when it is not only an extremely popular ARM/Linux platform but also the only ChromeOS ARM platform.
  108. [23:29:02] alex_ac100 [~alex@] вышел из IRC: Remote host closed the connection
  109. [23:29:17] marvin24_DT never saw chromoos running on an ac100
  110. [23:29:44] <gordan> Me either, but I remember several people discussing it here a while back, and we are using the ChromeOS ARM kernel.
  111. [23:30:13] <juliank> The most important problem for everyone is Flash support anyway.
  112. [23:30:20] <gordan> LOL!
  113. [23:30:56] <gordan> I really wich Flash would just do the decent thing and die, personally.
  114. [23:31:04] <gordan> s/wich/wish/
  115. [23:31:18] <juliank> No seriously, why produce an ARM device without flash if you can have an x86 one with Flash
  116. [23:31:36] vanger [] зашел на #ac100
  117. [23:32:24] <gordan> Except there is flash for ARM/Linux.
  118. [23:32:34] <gordan> At least two were mentioned here with links.
  119. [23:32:44] <gordan> phh, IIRC has the Tegra one somewhere.
  120. [23:32:51] <juliank> gordan: Yes, but it just does not work as well as the x86 one
  121. [23:33:26] <gordan> Somebody else posted a link to a different one that doesn't have nvidia dependencies, so should work on more generic platforms including on the basic frame buffer X.
  122. [23:33:42] <gordan> I find it works fine on the AC100.
  123. [23:33:59] <gordan> Youtube worked passably well.
  124. [23:34:14] <socketa4techx7> gordan, was it gnash?
  125. [23:34:32] <gordan> socketa4techx7: No, a proper
  126. [23:34:53] <marvin24_DT> adobe announces the dead of flash on ARM a few weeks ago
  127. [23:34:54] <gordan> With nv extensions in there (so it requires the nvidia binary driver).
  128. [23:35:10] <marvin24_DT> so, no flash in the future anymore
  129. [23:35:22] <gordan> GOOD!
  130. [23:35:40] <gordan> The faster it dies the better.
  131. [23:35:41] <juliank> gordan: It works. But only 360p IIRC, whereas I can watch 720p videos using youtube's HTML5 variant.
  132. [23:35:54] <gordan> Indeed, so no point in flash. :)
  133. [23:35:58] <gordan> Why would you want it?
  134. [23:36:18] <resetnow_> juliank: 720p on ac100?
  135. [23:36:26] <juliank> gordan: Not all youtube videos work without flash yet, and there are some more specialised websites requiring flash
  136. [23:36:33] <juliank> resetnow_: Yes.
  137. [23:36:47] <juliank> 720p WebM video on the AC100.
  138. [23:37:06] <gordan> juliank: Well, they are due a kick in the backside to switch to using non-proprietary technologies. :)
  139. [23:37:14] <gordan> IMO 99% of use of flash is for youtube.
  140. [23:37:22] <resetnow_> all the videoos with ads (80%) are still in flash
  141. [23:38:16] <juliank> gordan: Most of my flash use does not come from youtube, I mostly have the HTML5 player there.
  142. [23:38:49] <gordan> Yeah, me too re: html5 on youtube.
  143. [23:39:01] <gordan> There's a reason I use noflash. :)
  144. [23:39:22] <juliank> BTW, gnome-shell now starts on the AC100. The rendering is completely broken, but it runs (for some time)
  145. [23:42:51] alex_ac100 [~alex@] зашел на #ac1
RAW Paste Data
We use cookies for various purposes including analytics. By continuing to use Pastebin, you agree to our use of cookies as described in the Cookies Policy. OK, I Understand