ATI HD framebuffer driver (10.5/10.6) with source code

[kiepas]

I've taken a look into the 10/28 source code and think I've found a solution for the skewed mouse problem. The patch and a fixed build for 10.6 i386 are attached.

 

I tested on my 10.6.2 with Mobility Radeon X1300 with success.

 

The cause is explained in the source. Basically it's a wrong assumption made about the format of the cursor data returned by IOFrameBuffer::convertCursorImage (wrapped by VSLPrepareCursorForHardwareCursor).

 

Dong, if you find this useful, feel free to include in your source tree.

 

can you do that fix for 10.5 please? (use the 10/28 kext)

[dong]

FYI, the topic is updated with the hardware cursor fix.

[kiepas]

Big thanks Dong, cursor works great, i'm testing new driver now. Qe/Ci on radeon mobility HD2300 still working.

Im using iAtkos 7 with 1058 update

[nubz69]

ok I installed this driver on my HP 6280s and it works except that now I have no mouse cursor. I can click on things but there is no mouse. I removed the kext and my laptop went back to a lower resolution but I still have no cursor. I can interact with items on the screen using the mouse but you just cannot see the cursor. Anyone able to tell me how to fix this?

 

 

I have now found that if I increase the size of the cursor above a certain size that it reappears. Someone in another post had this problem as well and posted about that is how he was dealing with the problem. I also found a post with a kext for switching from hardware to software for the cursor but I don't know how to do that on my own. Does anyone have any idea for a fix for this?

[kiepas]

Big thanks Dong, cursor works great, i'm testing new driver now. Qe/Ci on radeon mobility HD2300 still working.

Im using iAtkos 7 with 1058 update

 

still some random kernel panics but its better thanks dong

[JoocKer]

i do not know what i`m doing wrong but my cursor is still skewed

 

i have installet the kext on ideneb 10.5.8

[0xdeadbeef]

FYI, the topic is updated with the hardware cursor fix.

 

 

Dong, h/w cursor setting in the latest driver has no effect for me - I get a skewed cursor now even with it disabled.

 

2400 Mobility.

[prowlingfox]

Hey...Dong....i've got a toshiba notebook with ATI Mobility Radeon HD 45xx series....i recently saw a post on netkas's blog about your drivers for 45xx/46xx/48xx for non-mobility. i was wondering if there is any hope for mobility support for that driver in the near future?

[super_engine]

Tested 12/18 version (upgrade for mouse tearing)

 

Perfectly working on my HP6820s with X1350

 

 

 

 

Thanks a lot my big hero DONG

[carlone99]

Strange thing here, while the pre hw coursor is working correctly, the 18-12-2009 version with skewed cursor fixed,

does not load at boot. Kextstat report no radeonhd.kext was loaded. If I try to load manually it load and kextstat detects it. I deleted the Extension.mkext and booted with -f without success.

 

Anyone have ideas ?

 

My configuration:

 

Dell 9400, T7200, ati x1400, 3 gb.

Snow Leopard 10.6.2.

 

Thanks

Giuseppe

[blombergkjell]

Strange thing here, while the pre hw coursor is working correctly, the 18-12-2009 version with skewed cursor fixed,

does not load at boot. Kextstat report no radeonhd.kext was loaded. If I try to load manually it load and kextstat detects it. I deleted the Extension.mkext and booted with -f without success.

 

Anyone have ideas ?

 

My configuration:

 

Dell 9400, T7200, ati x1400, 3 gb.

Snow Leopard 10.6.2.

 

Thanks

Giuseppe

 

Try to Touch extentions folder using OSX86 Tools ...... Might do the trick

[carlone99]

Try to Touch extentions folder using OSX86 Tools ...... Might do the trick

 

Thank you for the advice, using osx86tools fixed the strange behaviour and now the kext is loading and working as expected.

[factro]

 Hi every1, i want to thank all, specially  to dong &  semantics 

 

..i used [12/18/2009 update]:RadeonHD10.5.kext.zip and it works.

 

my graphics card is ati radeon Hd 4530 and now that i used this kext i have my full resolution.Btw before i install RadeonHD10.5.kext

 

i delete ATIRadeonX2000 kexts. also At 1st boot i used "Graphics Mode"="1366X768x32" and there was mouse problem(graphics problem)

 

And when i boot to leopard, later on change info.plist of RadeonHD10.5.kext and i change here!

 

<key>enableHWCursor</key>

<false/>

 

again

 

reinstall RadeonHD10.5.kext

 

now i can see my mouse pointer as well as i can change my resolution to any other resolution too..Bt i think opengl is nt working because

 

icant see coverflow in itune.sorry for my english.

[SoOlDiEr]

i know this is my first reply on this website which i think anyway thank you for this great release im using it for my dv7 hp im very greatfall for ur work anyway i just have a question if anyone can handle my situation i have to boot to snow using cpus=1 so its 100% i cant get a qe cause of this stupid 256mb memory of the ati i tried the dsdt fix it worked 50 % cause my osx is better on everything but its still on one cpu and i did flash my bios twice (ok its too risky i know) so i can learn something new anyway still i have to boot using cpus=1 so all fixes failed me to boot without cpus=1 i just wanna know if theres anything i can do to make my ati work as 512mb i have 4530 ati and t6400 cpu 2ghz and ati atom bios thanks once again

 

dont worry i still got my vanilla on my desktop 3000000% working xd

 

 

@ factro so u have that problem too i just wanna give u a little info that new update of dong it should fix ur mouse problem whish it didnt as obvious so what u wrote is an old info

 

man this cpus=1 cant be fixed UNLESS someone can just point into the right direction and ya dont think about efi cause that gave me boot problems.... anyone :S

[asstastic]

Got some strange behavior when trying to use this kext on a dual-monitor laptop setup. I won't even bother testing again until someone starts on the code to get that sort of setup working.

[kiepas]

works great under iDeneb 10.5.8, everythink is OK. Big Thanks Dong

[ole2]

still some random kernel panics but its better thanks dong

just discovered much less panics after temporary disabling voodoohda.kext loading.

could it be they both conflicting for resources? PCI bus DMA or IRQ or something else?

any thoughts, Dong? are you using voodooHDA on your development workstation along with RadeonHD.kext?

thanks for comments, will be waiting.

[dong]

just discovered much less panics after temporary disabling voodoohda.kext loading.

could it be they both conflicting for resources? PCI bus DMA or IRQ or something else?

any thoughts, Dong? are you using voodooHDA on your development workstation along with RadeonHD.kext?

thanks for comments, will be waiting.

Yes, I use VoodooHDA at the same time. It works well in 10.6. I seldom test with 10.5, so they may conflict in 10.5, possibly because of the reason that they are using the same way to manage memory as what I explained below.

 

I now think it could be the kernel_os_malloc/kernel_os_free pair of function that caused the instable behavior for the driver especially in 10.5. I used them to provide memory malloc/free service to AtomBios. It works quite well in 10.6, but may be very unstable in 10.5. ATI driver defined a MemoryAllocator class for this purpose, it wraps IOMalloc/IOFree with size information stored together in a structure. We may need do this too to avoid the problems.

[0xdeadbeef]

I now think it could be the kernel_os_malloc/kernel_os_free pair of function that caused the instable behavior for the driver especially in 10.5. I used them to provide memory malloc/free service to AtomBios. It works quite well in 10.6, but may be very unstable in 10.5. ATI driver defined a MemoryAllocator class for this purpose, it wraps IOMalloc/IOFree with size information stored together in a structure. We may need do this too to avoid the problems.

 

Funnily enough, RadeonHD has never panicked for me with or without VoodooHDA. In general though anything that is IOKit based should in general use the IOkit mem funcs instead of the kernel ones. ( Dunno why, because I thought they were just wrappers, silly me )

 

I made a picture for you :-)

 

[Dime333]

Hi Dong, any chance of getting mobility 4650 to work (maybe you're planning to work on it later, or not?)? Btw you're doing a very good job!

 

P.S. Are there some tricks to change at least resolution with this card (even with QE/CI not working)?

[kiepas]

just discovered much less panics after temporary disabling voodoohda.kext loading.

could it be they both conflicting for resources? PCI bus DMA or IRQ or something else?

any thoughts, Dong? are you using voodooHDA on your development workstation along with RadeonHD.kext?

thanks for comments, will be waiting.

 

im not using voodoo hda

[paddy159]

Does anyone managed to get QE/CI with a x1250 ? RadeonHD works well ...

 

Regards Patrick

[fediaFedia]

I gave up on mobility 4xxx, seems like it's never gonna have full support, and OSX is just no fun without QE/CI

[nomaj]

just discovered much less panics after temporary disabling voodoohda.kext loading.

could it be they both conflicting for resources? PCI bus DMA or IRQ or something else?

any thoughts, Dong? are you using voodooHDA on your development workstation along with RadeonHD.kext?

thanks for comments, will be waiting.

 

I'm using VoodooHDA all the time and i never get panic with RadeonHD ...

 

dell studio 1737 ATI 3650

[HaTaX]

Just wanted to add my experience with Dong's driver, it works well for me!

 

I have a Toshiba U500 with the ATI Radeon Mobility 4570 in it. The native resolution on the panel is 1280x800 and it was previously being run at 1024x768. If I use the ATIRadeonX2000.kext then the screen is unreadable and goes white with some pixelation. If I JUST use dong's RadeonHD.kext file and change the plist so that the hardware cursor is set to false (I get the corruption without setting this to false), then everything works perfectly!

 

I still have some more tweaking to get OSX 10.6 to like this laptop, but it's looking promising so far..

 

Thanks again for your work on this Dong!!! Hopefully now that there's a working frame buffer for the Mobility 4500 series (M92 R710 Family), someone can port a full on driver that we can get QE/CI working with these things.