Re: [Qemu-devel] [PATCH 5/5] vga optimization.

[Glauber Costa]

On Tue, Nov 11, 2008 at 1:33 PM, Stefano Stabellini
<address@hidden> wrote:
> Anthony Liguori wrote:
>>Glauber Costa wrote:
>>> +static void map_linear_vram(CirrusVGAState *s)
>>> +{
>>> +    int phys_offset = s->vga_io_memory;
>>> +
>>> +    if (!s->map_addr && s->lfb_addr && s->lfb_end) {
>>> +        s->map_addr = s->lfb_addr;
>>> +        s->map_end = s->lfb_end;
>>> +        cpu_register_physical_memory(s->map_addr, s->map_end - 
>>> s->map_addr, s->vram_offset);
>>> +        vga_dirty_log_start((VGAState *)s);
>>>
>>
>>So you register this region and enable dirty tracking.
>
> I don't think it should be enabled here.
>
>>> +    }
>>> +
>>> +    if(!(s->cirrus_srcptr != s->cirrus_srcptr_end)
>>> +        && !((s->sr[0x07] & 0x01) == 0)
>>> +        && !((s->gr[0x0B] & 0x14) == 0x14)
>>> +        && !(s->gr[0x0B] & 0x02)) {
>>> +        phys_offset = s->vram_offset | IO_MEM_RAM;
>>> +    }
>>> +    cpu_register_physical_memory(isa_mem_base + 0xa0000, 0x20000, 
>>> phys_offset);
>>>
>>
>>But also potentially change 0xa0000..0xc0000 to RAM, but you don't
>>enable dirty tracking on this region.  I think you either have to leave
>>this region as MMIO or enable dirty tracking on it too.
>>
>
> To be honest I don't understand why this is part is needed.
>
> @@ -2102,6 +2122,11 @@ static void vga_map(PCIDevice *pci_dev, int region_num,
>     } else {
>         cpu_register_physical_memory(addr, s->vram_size, s->vram_offset);
>     }
> +
> +    s->map_addr = addr;
> +    s->map_end = addr + VGA_RAM_SIZE;
> +
> +    vga_dirty_log_start(s);
>  }
>
> I think you should set lfb_addr and lfb_end here instead of map_addr and
> map_end.
>
It is possible for the lfb to be set, but not mapped.
So map_addr is what really indicates we're interested in tracking this region.
Or I am totally wrong?



-- 
Glauber  Costa.
"Free as in Freedom"
http://glommer.net

"The less confident you are, the more serious you have to act."