mmap()

[Kamal R. Prasad]

John-Mark Gurney wrote:

>Kamal R. Prasad wrote this message on Wed, Dec 08, 2004 at 10:45 +0530:
>  
>
>>Wilko Bulte wrote:
>>
>>    
>>
>>>On Tue, Dec 07, 2004 at 02:19:19PM -0800, John-Mark Gurney wrote..
>>>
>>>      
>>>
>>>>Kamal R. Prasad wrote this message on Tue, Dec 07, 2004 at 22:21 +0530:
>>>>
>>>>        
>>>>
>>>>>There was a bug in my  interface routine which probably resulted in too 
>>>>>many calls.
>>>>>Something like *paddr=vtophys(base) instead of 
>>>>>*paddr=vtophys(base+offset) should result in lots of calls to the 
>>>>>interface.
>>>>>
>>>>>          
>>>>>
>>>>Just remeber that vtophys is not a supported interface in HEAD, (if it
>>>>still works, it will probably be removed shortly).. Also, vtophys only
>>>>works on limited archs, like i386 and alpha...
>>>>        
>>>>
>>>s/limited/some/
>>>      
>>>
>>So what would be the equivalent supported interface in HEAD?
>>    
>>
>
>bus_dma(9)  It's a bit more complicated though...  but with out it,
>you won't be able to get your driver working on sparc64...
>
>  
>
woudn't pmap_extract() provide me with a physical addr of the 
corresponding kernel virtual address?
Also, is there some src code available that  shows bus_dma can be used 
to share a block of memory between kernel and user space?
My usb device probably doesn't have a dma controller, so I don't think 
the bulk pipe can use any memory allocated by bus_dma. [Pl. correct me 
if Im wrong this].
thanks
-kamal

>>BTW -man (4) usb probably needs more info than present.
>>    
>>
>
>I'm sure it does, feel free to send-pr patches.. :)
>
>  
>