.Alex Kenis wrote:
>> it was just a brainstorm, but I was thinking that it would provide a
>> more workable socket for hardware support which would be presented to
>> linux as PCI to fake a supported archetecture but would be translated to
>> the hardware as nubus to allow the use of some of the pre-existing
>> drivers ...>>
>Ah OK I get your point now. You were originally referring to a NuBus to
>_IDE_ emulation layer, which sounds a little bit strange to me :)
>On the other hand I feel that NuBus itself is supported well enough in
>Linux 2.4 already.
>> as for DMA on a 1400... it SEEMS that each individual controller
>> supports DMA, even though I can't seem to get linux to realize that, but
>> it may be because of the bus design??? (I have noticed that the
>> "idebus=25" Mk boot argument seems to give me fewer drive glitches
>> FWIW). I know that nubus is not cache-coherent, and does not support
>> direct transfer to I/O (but it does to rom and ram) but I can't tell if
>> that trickles down to the 1400/5300/2400 or not...
>Thanks for sharing this piece of information.
>Can you provide a list of links or pointers to relevant specs, please?
>This SF.net email is sponsored by: Splunk Inc.
>Still grepping through log files to find problems? Stop.
>Now Search log events and configuration files using AJAX and a browser.
You are probably right about the nubus support. i was just trying to avoid a bunch of recoding with every successive kernel change.
As far as the DMA stuff... I have a bunch of data sheets, tech notes and the like... I'll se about posting them for download.
Ii think that Nubus' inability in regard to DMA is not the issue since Apple has been using a parallel device for DMA since the early 90's. it really comes down to whether or not the Whitney controller allows it and whether it is true DMA (not likely) or some 'apple-ised' equivalent of it... and if the latter is the case... how and where the calls/channels take place. That might be just as useless as no DMA-proper in regard to using existing drivers, etc... but it me provide a window for a workaround to get sound functioning if it can be figured out... i am a newcomer here, so I'll have to search the archives until I hit on all the work that was done before. i do know (since it is in the developer's notes) that the AMIC chip in the x100 series nubus provides the DMA support for their AWAC sound chip, which is based on the same ASCO 2300 codec as the powerbooks. So it looks like it comes down to how the PBX and Whitney chip handle it, and how that relates to the translation between the 68k style bus to powerpc commands... one of my suspicions is that many of these problems have arisen from Apple's heavy leaning on the 68k emulator in their microkernel--I don't know how many of the calls actually depended explicitly on that code. Apple's developer notes were pretty detailed up through the 601 cpu era... then they get more general after that, so I can only really speculate until I find anything solid.