Manuel Teira wrote:
> It's nice to hear about you again. I'll test the changes you've made on my
> machine ASAP. BTW, I've take a fast look about the way you are allocating the
> memory for the test:
>
> struct pci_pool *pool;
> void *cpu_addr_table, *cpu_addr_data;
> void *cpu_addr_table, *cpu_addr_data;
> pool = pci_pool_create( "mach64", NULL, 0x4000, 0x4000, 0x4000, SLAB_ATOMIC );
> cpu_addr_table = pci_pool_alloc( pool, SLAB_ATOMIC, &table_addr );
> cpu_addr_data = pci_pool_alloc( pool, SLAB_ATOMIC, &data_addr );
>
> What alignment are you getting on this way? Could we select a diferent
> alignment using this kind of memory alloc?
have a look at <kernel-source-dir/Documentation/DMA-mapping.txt
there's the explaination about this API. I think we should use this if
anybody hasn't any objections.
pci_pool_create is specified:
pool = pci_pool_create( name, dev, size, align, alloc, flags );
in this case a 16kB buffer aligned to 16kB boundry (16384 = 0x4000)
please tell me status of your Xserver after starting. does it start up?
does the hardware lock? are you able to run gears without locking? what
does glxinfo tell you?
I would like to compare to my status in order to have the same working
ground.
Furthermore I suggest to drop the GUI master issue at the moment (it
isn't used in the driver right now or am I wrong? and therefore it is
just a performance issue)
If your gears locks too, then I would first track this down. Debuging
this is very hard because I have to boot after every try.
--Andreas Karrenbauer
_______________________________________________
Dri-devel mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/dri-devel
