Slightly different chip, but yes it might be of use as the main parts of the interface will doubtless be similar.
Originally Posted by
The Dell Axim x50v and x51v had this chip with an OpenGL ES driver for Windows Mobile. There is a port of Quake III: Arena that was playable at 640x480 on the x50v/x51v. I'm guessing it'd be a good target for reverse engineering drivers, if MBX and MBX lite are fully compatible.
Reverse engineering a Windows CE driver for a slightly different 3D chipset attached in a slightly different way. hmm
You'd probably be better off looking at the existing 2.4.x omap2430 driver imho, unless you already know lots about Windows CE "kernel" drivers that is.
I've been going through the Android source code, and I found some interesting stuff....
In kernel/drivers/video/mbx there is some MBX source code, intended for Intel's 2700G which packs an MBX lite. The code is recent as well, 2007 copyright (GPLed).
I was looking for some closed userspace libs, but couldn't find any. Looks like all the magic is contained within reg_bits.h and regs.h.
Yes, Intel 2700G != Omap 2420, but I do believe that the source code should be fairly compatible with some tweaking. Anyone want to support me/shoot me down?
I attached the source for those that don't want to download the 2 Gigs of Android source (cool stuff though).
I've been going through the Android source code, and I found some interesting stuff....
In kernel/drivers/video/mbx there is some MBX source code, intended for Intel's 2700G which packs an MBX lite. The code is recent as well, 2007 copyright (GPLed).
I was looking for some closed userspace libs, but couldn't find any. Looks like all the magic is contained within reg_bits.h and regs.h.
Yes, Intel 2700G != Omap 2420, but I do believe that the source code should be fairly compatible with some tweaking. Anyone want to support me/shoot me down?
I attached the source for those that don't want to download the 2 Gigs of Android source (cool stuff though).
Jon
Well spotted Jon,
However, there is no interaction with the powervr chip within this driver.
this driver code is for the 2d normal framebuffer.
This is equivalent to the already open omapfb.c
however, the "mbxdebugfs.c" file might have something, but it looks no more than test probing.
Ah ha... makes sense, fb drivers.... ok. But the chipset *is* capable of 3d work... so wouldn't it stand to reason that there are other drivers somewhere else for 3d? Or maybe it's up to the manufacturers to put insert their licensed proprio drivers? (I guess t-mobile inserting the drivers into a production G1....)
Awww, man! I got all flushed and excited there for a second. And here I am, trying to be the guy who is saying, "don't wait for it, it ain't coming..." Don't I look like a fool now.
Oh it'll come. Whether it's "found" or happens to be left "unattended" at the corner of the street....
Or a full out reverse engineering.... I think we have enough pieces to reverse ourselves a good driver. We have bits and pieces of different sources from different devices. I think all we need is someone to step up with a device that has an MBX chip fully functional, and the games can begin.
I'm glad you're so enthusiastic. I'm very skeptical. I think our only shot at this is if Nokia decides to let the N8x0 have a "Hackers Edition" of Fremantle. Otherwise, I fear that we're S.O.L...
I think it's doable, I started writing some code from the ASM, pH5 wrote a kernel patch to switch the chipset on. Doable, but time consuming. I'm happy to come back to it if we don't get a driver when Fremantle is released.
I have gotten software-rendered OpenGL running on the tablet, using the snail-slow Mesa software rasterizer (10fps in the GLGears demo). I, like lcuk, wonder if it would be possible to do some kind of software workaround for the fact we have no drivers; replace some of those slow software routines in that mesa package with some liqbase magic; we could call it liqGL!
