Android Tablets Forum banner
1 - 20 of 24 Posts

·
Registered
Joined
·
576 Posts
well... while wandering through some PDN threads(I like to see what they're up to, along w/A81 Es...) I came across something that made me want to check through some logs that I saved a while back(dmesg, logcat, etc.) and stumbled across:
Code:
D/libEGL  (  914): loaded /system/lib/egl/libGLES_android.soD/libEGL  (  914): loaded /system/lib/egl/libEGL_mali.soD/libEGL  (  914): loaded /system/lib/egl/libGLESv1_CM_mali.soD/libEGL  (  914): loaded /system/lib/egl/libGLESv2_mali.soI/SurfaceFlinger(  914): EGL [URL=informations:I/SurfaceFlinger(]informations:I/SurfaceFlinger([/URL]  914): # of configs : 20I/SurfaceFlinger(  914): vendor    : AndroidI/SurfaceFlinger(  914): version   : 1.4 Android META-EGLI/SurfaceFlinger(  914): extensions: EGL_KHR_image EGL_KHR_image_base EGL_KHR_image_pixmap EGL_ANDROID_image_native_buffer EGL_ANDROID_swap_rectangle EGL_ANDROID_get_render_buffer I/SurfaceFlinger(  914): Client API: OpenGL ESI/SurfaceFlinger(  914): EGLSurface: 8-8-8-8, config=0x7E/libEGL  (  914): [URL=eglSetSwapRectangleANDROID:1743]eglSetSwapRectangleANDROID:1743[/URL] error 3008 (EGL_BAD_DISPLAY)I/SurfaceFlinger(  914): flags     : 00100000I/SurfaceFlinger(  914): OpenGL [URL=informations:I/SurfaceFlinger(]informations:I/SurfaceFlinger([/URL]  914): vendor    : ARMI/SurfaceFlinger(  914): renderer  : Mali-200I/SurfaceFlinger(  914): version   : OpenGL ES-CM 1.1I/SurfaceFlinger(  914): extensions: GL_OES_byte_coordinates GL_OES_fixed_point GL_OES_single_precision GL_OES_matrix_get GL_OES_read_format GL_OES_compressed_paletted_texture GL_OES_point_size_array GL_OES_point_sprite GL_OES_query_matrix GL_OES_matrix_palette GL_OES_compressed_ETC1_RGB8_texture GL_OES_EGL_image GL_OES_draw_texture D/SurfaceFlinger(  914): Assuming uncached graphics buffers.I/sysproc (  914): System server: entering thread pool.I/SystemServer(  914): Entropy ServiceI/SystemServer(  914): Power ManagerI/SystemServer(  914): Activity ManagerI/ActivityManager(  914): Memory class: 16D/libEGL  (  926): loaded /system/lib/egl/libGLES_android.soE/SurfaceFlinger(  914): eglSwapBuffers: EGL error 0x3008 (EGL_BAD_DISPLAY)D/libEGL  (  926): loaded /system/lib/egl/libEGL_mali.soD/libEGL  (  926): loaded /system/lib/egl/libGLESv1_CM_mali.soD/libEGL  (  926): loaded /system/lib/egl/libGLESv2_mali.so
in logcat.i.e. If I'm not mistaken it DOES look like it's trying to load the full opengl es stack, but for some reason doesn't seem to show and/or maybe behave properly...anyone have any ideas? insights?[EDIT]I notice two error messages in there... maybe they don't have the LCD driver properly configured and it's causing OGL to fail to work properly?
Code:
V/RenderScript( 1032): initEGL startD/libEGL  ( 1032): loaded /system/lib/egl/libGLES_android.soD/libEGL  ( 1032): loaded /system/lib/egl/libEGL_mali.soD/libEGL  ( 1032): loaded /system/lib/egl/libGLESv1_CM_mali.soD/libEGL  ( 1032): loaded /system/lib/egl/libGLESv2_mali.soE/RenderScript( 1032): EGL/Surface mismatch  EGL (1 x 1)  SF (800 x 480)V/RenderScript( 1032): GL Version OpenGL ES 2.0V/RenderScript( 1032): GL Renderer Mali-200
later in the startup process...[/EDIT]
 

·
Registered
Joined
·
53 Posts
It's well known that our driver stack isn't really utilizing the Mali200 properly. The driver we have doesn't seem to carry the full HW support in the userspace HW specific driver for the Mali200. I'm curious what you were trying to run when this logcat dump was generated, and what thread in the PDN group caught your eye that might've referred to this.
 

·
Registered
Joined
·
784 Posts
This is from the boot netcat log. You can see it on similar tablets as wel (i.e. the HSG-X5A). If you google for some of the terms in this log, there are similar logs on pastebin.
 

·
Registered
Joined
·
53 Posts
As I've stated before. I suspect that all these TCC8902 equipped Tab ( of which, the HSG-X5A is one ), all suffer from the same malady of a non-fully implemented or improperly implemented OpenGL stack.
 

·
Registered
Joined
·
576 Posts
IIRC it was fairly close to initial booting so Android was still attempting to come up fully... sort of looks like to me that they screwed up on configuring the display and memory buffers for the Mali GPU... (which was one of my thoughts as to why we had no OGL accel...) Thinking misconfigured driver when built now... at least it tells us that they have the DDK... even if they can't figure out completely what to do with it...[EDIT]Hah! Just checked that logfile dump again and it was dated 8/14/10 about 3h into my ownership! :DNot sure any longer exactly what I had done to it at that point... probably just rooting, v2 patch, as IIRC I didn't bother w/a market fix until Sunday morning... probably something buried in the forums here, but it's irrelevant.Probably should try to do another one with current fw + fixes... should probably also do one on one of the failing opengl apps to see what error messages it spits out...Also googling a little turned up this topic, cyanogenmod, and another... looks like not all features are enabled w/the driver either v. those logs... driver stack is probably just misconfigured... [email protected] the x5a and like report an opengl es version in the Android System Info app? (It's under the 2nd tab, "features" heading, bottom entry glesVer or something says 0.0 on GT) And more to the point does it seem that they are using the Mali? (GT "feels" like its doing sw rendering...)[/EDIT]
 

·
Registered
Joined
·
53 Posts
I can actually confirm that we're experiencing software rendering. I've been doing some reading into the Mali Open Source code drop, and it looks like there's a bit of things that have to be configured into one of the source files so that the driver properly initializes and reports it's properties properly. If the the user space driver we have were properly configured, MAYBE, OGL might come up properly. I'm curious to know of the other TCC8902 tablets out there, the version of their user space driver. And perhaps trying out one of theirs on the GT to see if anyone actually has one that supports HW acceleration.
 

·
Registered
Joined
·
53 Posts
I'm exchanging some messages with someone at ARM currently who's familiar with the Mali200, and I'm getting some interesting responses from him.Apparently, if my understanding is correct, any ODM who licenses any Arm based technology that also includes the MaliXXX should receive a license to the Linux DDK, which would include the full Mali Driver stack, and which would be a reference point from which to configure their driver for the desired display technology. Which means........Augen's ODM "should" be capable of this. They should be in position of a full Mali Linux DDK, in which case, should include a Android Specific/User Space driver's code which more than likely by configuration be compilable into a working 3D accelerated driver.Looks like the onus falls onto the ODM to provide a working 3D accelerated driver here folks.
 

·
Registered
Joined
·
246 Posts
[quote name='marcelol;54585]I'm exchanging some messages with someone at ARM currently who's familiar with the Mali200' date=' and I'm getting some interesting responses from him.Apparently, if my understanding is correct, any ODM who licenses any Arm based technology that also includes the MaliXXX should receive a license to the Linux DDK, which would include the full Mali Driver stack, and which would be a reference point from which to configure their driver for the desired display technology. Which means........Augen's ODM "should" be capable of this. They should be in position of a full Mali Linux DDK, in which case, should include a Android Specific/User Space driver's code which more than likely by configuration be compilable into a working 3D accelerated driver.Looks like the onus falls onto the ODM to provide a working 3D accelerated driver here folks.[/QUOTE']I think someone posted about this exactly already.. Cutterjohn even might have.CPG
 

·
Registered
Joined
·
576 Posts
[quote name='marcelol;54585]I'm exchanging some messages with someone at ARM currently who's familiar with the Mali200' date=' and I'm getting some interesting responses from him.Apparently, if my understanding is correct, any ODM who licenses any Arm based technology that also includes the MaliXXX should receive a license to the Linux DDK, which would include the full Mali Driver stack, and which would be a reference point from which to configure their driver for the desired display technology. Which means........Augen's ODM "should" be capable of this. They should be in position of a full Mali Linux DDK, in which case, should include a Android Specific/User Space driver's code which more than likely by configuration be compilable into a working 3D accelerated driver.Looks like the onus falls onto the ODM to provide a working 3D accelerated driver here folks.[/QUOTE']I know. I noticed that when I referenced the mali dev site for augendev in that other thread(the one where they sort of announced new tablets in 2011) where we started asking about some additional fw updates and things... juts wasn't sure how the licensing scheme worked other than Telechips got the Linux Mali DDK as part of their licensing Mali agreement... which makes sense since arm wants to license hw, and their licensee should want to sell hw, which should imply that they'd want to pass on the necessary sw pieces.[EDIT]Actually I just noticed it when I pointed out the mali dev site to augen dev, because I first hit the mali dev forums and saw some ARM employee responses about a commercial Linux DDK license and so didn't bother to look closely at the website when I finally got there......still Telechips could possibly be a hold up if they want additional $$$ for Telechips sw support... hopefully they'll see it's in their best interests not to do that, but you never know...Actually IIRC too, I suggested to Daniel if the fw guys were uncooperative to try contacting Telechips and/or ARM anyways... I figured its his company/product so let them do the legwork...BTW ever hear anything more about the kernel source? Someone posted something on the xda-devel thread about via a filed ticket ON hold priority low...[/EDIT][EDIT2]OT: man vbulletin forum sw sucks... wish that they'd change to phpBB...[/EDIT2]
 

·
Registered
Joined
·
53 Posts
I guess I missed your other thread. This was my understanding all along, but I'm the sort to get specific corroboration from a source before I speak about ( wasn't saying anyone else was unnecessarily, just that that's my S.O.P. ). About Daniel, I'm actually going to call him here in a little bit, because I haven't heard back after the write up I sent ( ask CPG for a copy of it, he's seen it ) them before about this whole thing. The way the person at ARM said it, Telechips could have done what you said...1. Giving them an Android Software stack that ONLY supports software rendering.or2. Should've given then a working Mali200 driver that they would then tweak according to their display technology.If the case is #1, the it gets a little hairier because getting the ODM to go back to Telechips and say "Give me a 3D accelerated driver for Android"...good luck with that.If it's case #2, and they simply chose to not bother with trying to set up things correctly, then MAYBE we can get them to do it. Either way, the other alternative is for Augen ( or one or more of us ) to license the Linux Mali DDK, and build a driver that is properly configured.
 

·
Registered
Joined
·
576 Posts
[quote name='marcelol;54622]I guess I missed your other thread. This was my understanding all along' date=' but I'm the sort to get specific corroboration from a source before I speak about ( wasn't saying anyone else was unnecessarily, just that that's my S.O.P. ). About Daniel, I'm actually going to call him here in a little bit, because I haven't heard back after the write up I sent ( ask CPG for a copy of it, he's seen it ) them before about this whole thing. The way the person at ARM said it, Telechips could have done what you said...1. Giving them an Android Software stack that ONLY supports software rendering.or2. Should've given then a working Mali200 driver that they would then tweak according to their display technology.If the case is #1, the it gets a little hairier because getting the ODM to go back to Telechips and say "Give me a 3D accelerated driver for Android"...good luck with that.If it's case #2, and they simply chose to not bother with trying to set up things correctly, then MAYBE we can get them to do it. Either way, the other alternative is for Augen ( or one or more of us ) to license the Linux Mali DDK, and build a driver that is properly configured.[/QUOTE']I'm guessing #2 given the logs... did you forward those to the ARM guy to see if he could tell just from those which was what? I'm still inclined to believe that the fw guy just hasn't figured out how to configure the driver properly so it falls back to sw rendering...[EDIT]Dug up the link to the post in the thread...http://www.slatedroid.com/augen-gen...multi-touch-screen-coming-soon.html#post47930It's in the 2nd edit part...I still can't believe that they'd give them a sw only driver. It just makes no sense. It's like pre-crippling your product that you're trying to get people to buy, which is why ARM gives the licensees the kit in the first place when they license Mali instead of making them buy it separately...[/EDIT]
 

·
Registered
Joined
·
53 Posts
Hey I agree with you #1 seems more far fetched, but....plausible....I wouldn't put anything past penny-pinchers....No I hadn't shared that log snippet, but I will.....do you have a fuller log like let's say, from a clean boot up ? Matter of fact, I may check that for myself on mine tonight. Just for grins and giggle to see what the ARM guy says.
 

·
Registered
Joined
·
53 Posts
Folks......excellent news. GLBenchmark has been updated to support the Gentouch78. I've posted some preliminary results, so we'll be seeing them soon once they update the public DB.Here are the results as follows:CPU Integer Performance - 8019CPU Float Performance - 1791HD ES 1.1 CPU Skinning - 1423 Frames ( 47.4 fps )HD ES 1.1 GPU Skinning - 1399 Frame ( 46.6 fps )Looking at these numbers pretty much cements that the Mali driver is punting to the CPU.More results as I run and post them
 

·
Registered
Joined
·
53 Posts
Please confirm for me folks...some of you..if your egl.cfg looks like this....# This confidential and proprietary software may be used only as# authorised by a licensing agreement from ARM Limited# (C) COPYRIGHT 2010 ARM Limited# ALL RIGHTS RESERVED# The entire notice above must be reproduced on all authorised# copies and copies may only be made to the extent permitted# by a licensing agreement from ARM Limited.#0 0 android0 1 mali
 

·
Registered
Joined
·
58 Posts
Also of interest....strings (Gentouch System image)/lib/modules/mali.ko | grep GPL -b4
Code:
6507-External Memory6523-Mali block allocator page tables6556-version=6565-author=ARM [URL=Ltd.6581:license=GPL6593-parm=mali_max_job_runtime:Maximum]Ltd.6581:license=GPL6593-parm=mali_max_job_runtime:Maximum[/URL] allowed job runtime in msecs.6657-Jobs will be killed after this no matter what6703-parmtype=mali_max_job_runtime:int6737-parm=mali_hang_check_interval:Interval at which to check for progress after the hw watchdog has been triggered
 

·
Registered
Joined
·
576 Posts
Discussion Starter · #18 ·
Dandel;55259 said:
Also of interest....strings (Gentouch System image)/lib/modules/mali.ko | grep GPL -b4
Code:
6507-External Memory6523-Mali block allocator page tables6556-version=6565-author=ARM [URL=Ltd.6581:license=GPL6593-parm=mali_max_job_runtime:Maximum]Ltd.6581:license=GPL6593-parm=mali_max_job_runtime:Maximum[/URL] allowed job runtime in msecs.6657-Jobs will be killed after this no matter what6703-parmtype=mali_max_job_runtime:int6737-parm=mali_hang_check_interval:Interval at which to check for progress after the hw watchdog has been triggered
Parts of the mali driver are GPL, but the complete driver with openvg/gl es is NOT. The GPLed drivers just give basic access to the hw.
 

·
Registered
Joined
·
53 Posts
all they've GPL'd is the actual Kernel driver....nothing moreNone of the Android specific/user space drivers are GPL'd. Which perhaps should be the focus for now, to get ARM to release the user space drivers to the GPL just as Atheros has theirs.
 

·
Registered
Joined
·
784 Posts
marcelol;55255 said:
Please confirm for me folks...some of you..if your egl.cfg looks like this....0 0 android0 1 mali
On HSG-X5A version #369 (sept 1 2010) rom, this is exactly what's in the EGL.CFG.
 
1 - 20 of 24 Posts
Top