Sriram,
How did you manage to get a hold of the 2.6.18 source?
Contacts inside TI and montavista have eluded to it but they said it's not
ready for distrubution and it doesn't show up on Montavista's list.
I'd like to see what they've done with the FBDev driver... someone from TES
electronic solutions said that the current FBDev driver has a lot of
performance issues and the new one from Montavista is supposed to fix all
of that.
I had a chance to play around with the davncifb patch and I missed a few
more IOCTLs so I'll post when I can run dfbinfo without it complaining...
stay tuned.
I also created a shell script that maps all of the arm_v5t_le-* tools to
the now supported standard of the platform triplet
armv5tl-montavista-linuxeabi so autogen and configure can actually find the
proper tools. I'll send that one also.
If we're able to find the either the source or the binary for the driver, I
can incorporate that into the kernel module patch so that we can build from
within the kernel tree, that is if I knew where it was.
It's probably easier for now to just include the binary image if we can
find or make it because it would require the TI c64x+ code generation tools
and I don't think that the kernel make will call them properly. Besides, it
would be easier to just throw the project in CodeComposer and generate a
binary.
-- David
DAVID A. KONDRAD
Software Design Engineer
On-Q/Legrand
www.onqlegrand.com
"Sriram
Neelakandan"
<sriram.neelakand To
[EMAIL PROTECTED]> [EMAIL PROTECTED]
cc
03/06/2008 10:07 [EMAIL PROTECTED],
AM directfb-dev@directfb.org
Subject
Re: [directfb-dev] How do I get
c64x.ko?
Hi David,
If the latter, what patch level of the MV kernel are you using?
I use a kernel based on MV 2.6.18
I've made a patch that integrates the missing IOCTLs in the supported
kernel as well as the code from the patch located in the directfb source.
I'm in the process of testing this patch and can post if anyone needs it.
The missing IOCTLs is a lot of trouble
I think even if the patch does not apply neatly, u can post it out..
We need it badly.
Currently though there is a davinci driver .. i say gfxdrivers=none.. lest
it wont compile
and using the generic software rastarizer
I can make a patch that will integrate the directfb c64x driver
into the kernel tree.
Will this include the DSP code (c64x_drv.bin) as well ?
Also, be aware that if you decided to play with ++DFB or some of the
other
wrappers for DirectFB, you'll need to patch the latest sources to account
for the __uxx to uxx variable renamings in 1.1.x.
I've also done this, but I'm unsure if there are any other API changes
that
are going to creep up on me.
So the verdict seems to be that even we manage to get the module
compiled,
either it won't load or it's missing the firmware file to do so.
I think if the code is available .. people can make use of it and
contribute back ..
i see a lot of traction especially on Davinci ..
On other note, how did you set up your cross-compile environment to build
DirectFB.
This is how i do it
export LDFLAGS="-L/home/sriram/davinci/freetype_eq/
-L/home/sriram/libpng/lib -L/home/sriram/zlib/lib
-L/home/sriram/davinci/freetype_eq" && export
CPPFLAGS="-I/home/sriram/libpng/include -I/home/sriram/zlib/include
-I/home/sriram/davinci/freetype_eq/include && ./configure --host=arm-linux
--prefix=/home/sriram/directfb/ --enable-fbdev --enable-devmem
--disable-x11 --with-inputdrivers=lirc --disable-video4linux2
--disable-sysfs --disable-multi --disable-sdl --disable-video4linux
--datadir=/home/sriram/directfb/usr/lib --with-png=yes --with-freetype=yes
--disable-debug --disable-debug-support --with-gfxdrivers=none
Hope that helps
--
Sriram Neelakandan
Author - Embedded Linux System Design And Development (
http://tinyurl.com/2doosu)
_______________________________________________
directfb-dev mailing list
directfb-dev@directfb.org
http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
