Hello
I've found some more info relating to the DVEVM's memory map in
sprue67a.pdf (Codec engine).
It reserves a few MB here and there for DSP code and Shared memory
then a massive 122MB (0x88000000-0x8FA00000) for the "algorithm heap"?.

Is it possible to reconfigure this?
or if I'm not using the DSP (at the moment) is it OK to set mem=240M?

Cheers
Rudy

 
> -----Original Message-----
> From: 
> [EMAIL PROTECTED]
> .davincidsp.com 
> [mailto:davinci-linux-open-source-bounces+rudy.reinsch=aupix.c
> [EMAIL PROTECTED] On Behalf Of Kevin Hilman
> Sent: 20 October 2006 17:45
> To: Ring, Chris
> Cc: [email protected]
> Subject: Re: Free 128MB upgrade for DVEVM
> 
> Ring, Chris wrote:
> 
> > Somebody correct me if I'm wrong, but my understanding is:
> >    * The video drivers utilize the OSD window.
> >    * The video drivers get their buffers from the Linux kernel.
> >    * Therefore the Linux kernel shouldn't be granted memory larger 
> > than
> > 128 MB.
> > 
> > Setting it larger than 128 MB _could_ result in the video drivers 
> > getting buffers in the memory space above 128 MB... which 
> would result 
> > in really strange - and seemingly random(!) OSD related problems.
> 
> Actually, this restriction doesn't affect the video drivers.  
> Kernels can be booted with full memory and the video drivers 
> will only get memory < 128M.
> 
> I patched the DaVinci kernels a while back such that DMA 
> allocations are restricted to be < 128M.  Since the video 
> driver uses the dma_alloc* functions for its memory, it is safe.
> 
> For the gory details, see include/asm-arm/arch-davinci/memory.h.
> Specifically, ISA_DMA_THRESHOLD and arch_adjust_zones()
> 
> Kevin
> _______________________________________________
> Davinci-linux-open-source mailing list
> [email protected]
> http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source
> 
_______________________________________________
Davinci-linux-open-source mailing list
[email protected]
http://linux.davincidsp.com/mailman/listinfo/davinci-linux-open-source

Reply via email to