Re: [Qemu-devel] [PATCH 1 of 6] [UPDATE] vnc dynamic resolution

[Stefano Stabellini <stefano.stabellini@eu.citrix.com>]

Anthony Liguori wrote:

> 
> When testing this patch, gvncviewer hangs when attempting to connect. 
> It's waiting to receive the ServerPixelFormat and it doesn't appear to
> receive it.


I admit I didn't test the patch individually but only the full series at
once.
The reality is that the first patch doesn't make any sense without the
fourth.
Actually now that I split the shared buffer patch into two patches: the
shared buffer in vga.c and the vnc implementation, it would even make
sense to merge the first patch and the fourth.
I'll repost the full series that way, if you agree:

1) shared buffer (vga only and interfaces)
2) shared buffer vnc implementation
3) WMVi implementation in vnc
4) shared buffer sdl implementation
5) opengl rendering implementation

I could even merge 2) and 3) if you prefer.

>> @@ -1318,7 +1440,9 @@ static int protocol_client_init(VncState *vs,
>> uint8_t *data, size_t len)
>>      vnc_write_u16(vs, vs->ds->height);
>>  
>>      vnc_write_u8(vs, vs->depth * 8); /* bits-per-pixel */
>> -    vnc_write_u8(vs, vs->depth * 8); /* depth */
>> +    if (vs->depth == 4) vnc_write_u8(vs, 24); /* depth */
>> +    else vnc_write_u8(vs, vs->depth * 8); /* depth */
>>   
> 
> Why unconditionally change to a depth of 24 instead of 32?  There's no
> savings from a protocol perspective.
> 

I guess this change is not really needed, it is only to better specify
that the format is RGB, 1 byte per channel plus 1 byte for the unused
alpha channel.