Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

interesting. I see that when i try to play a video it tries to use 3d acceleration (which it should) but it craps out there. This is more of a userland issue i suppose

Wirelessly posted (Opera/9.80 (Windows Mobile; Opera Mini/5.1.21594/21.572; U; en) Presto/2.5.25 Version/10.54)

Happens on haret as well. I think I remember Alex saying it might be gralloc related. Arrrghhh and I were testing various things for him on this right after he got autobl working.

Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

ic.. Just finished talking to alex and he got the whole .35 kernel to come up for Kovsky. So hopefully jonpry and alex and compare notes so our .35 can fire up. In the mean time i'll concentrate on fixing the panel down issue now that we have a new FB.

did you get a chance to test it yet?

Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

Well hey, glad to see they finally brought the site back up... I did give it a full day of testing, and the new FB is excellent. I updated to the new ext2 at the same time, so I'm having a hard time telling what caused what, but here's some of the things I've noticed.

Startup animation is way faster than it ever has been. I'm actually getting tearing sometimes.

SD video playback is much improved. I used to get a few dropped frames here and there, but this is much smoother. The Youtube app is actually on par with the WM one now for SD playback.

HQ video playback is actually worse, but it doesn't really matter since it was never viewable to begin with. It used to show two or three frames every 10 seconds or so, and now it seems to just lock up and stick on the same frame for the whole video.

Overall, the UI seems a little smoother. Not sure if it's just been a while since I've spent so much time in Android, but the edges and corners of buttons and widgets look cleaner than I remember.

Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

Did you test with 3d enabled ? the tearing and missing frames are probaby due to copybits. This is a userland lib that may need adjusting to play nice with the new framebuffer. I tried cyanogen again and the startupscreen there is actually really smooth but i get the same copybit error there as well. For 3d to work well we need copybit and gralloc to be flawless. right now they are failing.

Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

Yeah, 3D is always enabled for me. I still have the force command in tinboot, but I thought for sure you had hard coded it to always be enabled since it was working properly.

Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

cool

i reading some logs now since i know the hd2 guys had some similar crap when they just got 3d working which was related to copybits. If you see something let me know.. it might help us. copybits works with mdp on the kernel to blit and do all the fancy work.

Here is the exact error i see popping up.

E/copybit ( 1392): copyBits failed (Invalid argument)
D/copybit ( 1392): 0: src={w=480, h=800, f=0, rect={0,0,9,38}}
D/copybit ( 1392): dst={w=480, h=800, f=0, rect={0,0,9,38}}
D/copybit ( 1392): flags=00020000
D/copybit ( 1392): 1: src={w=480, h=800, f=0, rect={406,0,74,38}}
D/copybit ( 1392): dst={w=480, h=800, f=0, rect={406,0,74,38}}
D/copybit ( 1392): flags=00020000
D/copybit ( 1392): 2: src={w=480, h=800, f=0, rect={0,38,480,762}}
D/copybit ( 1392): dst={w=480, h=800, f=0, rect={0,38,480,762}}
D/copybit ( 1392): flags=00020000
E/libagl ( 1392): copybit failed (Unknown error: -22)

Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

Haven't had a chance to test this and won't until tomorrow, but if you guys want the modded rootfs with the new gralloc, you can grab it here:

http://dl.dropbox.com/u/11790134/roo...29-42b44e2.img

I am also working on compiling the new NBH and modules. I'm not sure if autobuild is working peroply, but I'll have a new package up shortly with the 3D driver in it as well.

Also, libGLES will still need to be in the root of the SD card, but it will automount during boot. I left it this way so we can just replace it with a newer version for testing. Eventually I bet we will put it in the rootfs or something.

Re: NAND Boot Testing - Auto-brightness Testing: 10-28. Automatic Keyboard Detection!
 

New package containing the rootfs and new NBH/modules (and libGLES) is in the first post. Let me know how it works. I have a long drive tonight and can't flash it (stupid bluetooth...) until tomorrow.

Re: NAND Boot Testing - 11-29: New Framebuffer, gralloc, and libGLES
 

Boot animation here is flashing badly at first then slower and lots of tearing/artifacts on startup. Near the time it fully loads and boots to launcher it flashes a few times with a few tears but smooth, still slower than when 3d was enabled.At first I don't think it loaded my modules file. I had renamed it to modules-modules-2.6.27.46.tar. Then changed it back I think it loaded this time ok but no change on startup.

3d app drawer just as smooth as before if not a tad better.

Going to wipe my data.img file to see.... Should be almost standard process anymore but I hate waiting on restoring my data.

Note: Now that the screen density is set properly one my apps I had issues with is now loading. Thanks!