70 replies to this topic

[joe75]

Any good news to report, joblo

[BaselineAce]

Did this project die?

[joe75]

BaselineAce, on Dec 5 2006, 03:41 PM, said:

Did this project die?

No
http://x86dev.org/forums/

[Korrupted]

The project was put on halt, so in a sense, yes, it is dead.

[chinak0]

x850 pro (r481) qe/ci can work ?

[ole2]

DaxTsurugi, on Dec 6 2006, 12:22 AM, said:

The project was put on halt, so in a sense, yes, it is dead.

not exactly dead,
ole2 and slice are working on it now

[Korrupted]

Keep in mind I made that post a year ago

[ole2]

based on IOGraphics-191.5.tar.gz archive from Darwin

interesting fact was found from IONDRVFramebuffer.cpp:

IONDRVFramebuffer is no longer manage graphic card framebuffer directly,
instead it forwarding all requests/messages to appropriate display driver

for example IONDRVFramebuffer::setGammaTable
IOReturn IONDRVFramebuffer::setGammaTable( UInt32 channelCount, UInt32 dataCount, UInt32 dataWidth, void * data )

is sending cursor information to packager  _doControl
IOReturn IONDRVFramebuffer::_doControl( IONDRVFramebuffer * self, UInt32 code, void * params )

which is sending command kIONDRVControlCommand type of kIONDRVImmediateIOCommandKind
to dispatcher doDriverIO
IOReturn IONDRVFramebuffer::doDriverIO( UInt32 commandID, void * contents, UInt32 commandCode, UInt32 commandKind )

which is retrieving display driver from NUB
            ndrv = IOBootNDRV::fromRegistryEntry( nub );
and forwarding command to it's own dispatcher with same name doDriverIO
        err = ndrv->doDriverIO( commandID, contents, commandCode, commandKind );

ndrv itself is a member of class IONDRVFramebuffer
    class IONDRV * ndrv;

and, the only method ndrv->doDriverIO (IONDRV::doDriverIO) was found, is in
class IOBootNDRV : public IONDRV (inherited from IONDRV), but virtual one:

IOReturn IOBootNDRV::doDriverIO( UInt32 commandID, void * contents,
                                 UInt32 commandCode, UInt32 commandKind )

which is doing actual command dispatching (depends on control or status)

    switch (commandCode)
    {
        case kIONDRVInitializeCommand:
        case kIONDRVOpenCommand:
            ret = kIOReturnSuccess;
            break;

        case kIONDRVControlCommand:
            ret = doControl( pb->code, pb->params );
            break;
        case kIONDRVStatusCommand:
            ret = doStatus( pb->code, pb->params );
            break;

        default:
            ret = kIOReturnUnsupported;
            break;
    }

which is sent then forward to doControl
IOReturn IOBootNDRV::doControl( UInt32 code, void * params )

which is handing orders (commands), example setting Gamma:

    switch (code)
    {
        case cscSetEntries:
        case cscSetGamma:
            ret = kIOReturnSuccess;
            break;

        default:
            ret = kIOReturnUnsupported;
            break;
    }

    return (ret);

another dispatcher was found in same archive in another source IONDRV.cpp:

inside class IOPEFNDRV : public IONDRV

IOReturn IOPEFNDRV::doDriverIO( UInt32 commandID, void * contents,
                                UInt32 commandCode, UInt32 commandKind )

but following one does almost nothing with commands:

    switch (commandCode)
    {
        case kIONDRVReplaceCommand:
        case kIONDRVInitializeCommand:
            err = _IONDRVLibrariesInitialize( (IOService *) contents );
            if (kIOReturnSuccess != err)
                break;
            /* fall thru */

        case kIONDRVFinalizeCommand:
        case kIONDRVSupersededCommand:
            initInfo.refNum = 0xffff & ((UInt32) this);
            MAKE_REG_ENTRY((&initInfo.deviceEntry), contents)
            contents = &initInfo;
            break;

        case kIONDRVControlCommand:
        case kIONDRVStatusCommand:
        case kIONDRVOpenCommand:
        case kIONDRVCloseCommand:
        case kIONDRVReadCommand:
        case kIONDRVWriteCommand:
        case kIONDRVKillIOCommand:

            pb = (CntrlParam *) contents;
            pb->qLink = 0;
            pb->ioCRefNum = 0xffff & ((UInt32) this);
            break;
    }

most of the commands seems to be just ignored

my question is:

if class IOATYNDRV(or some other name) : public IONDRV inheritance with proper methods implementation, is all what we need for framebuffer functionality on osx86 with ATI chipset?

[papillon68]

There is a test package to try ? I have an X600 Mobility that is running in software mode now ...

[dong]

ole2, on Jan 16 2008, 06:03 AM, said:

not exactly dead,
ole2 and slice are working on it now

How is the IOATINDRV class going on now? I'm also looking into this and try to find a way to implement something like CallistoFB for my ATI X1400 mobility, but still struggling in understanding the source code yet.

[Slice]

Joblo abandon the ship so I think it is better to close the thread and I begin new one.