[Spice-devel] [protocol RFC 0/2] RANDR support via QXLHead + SpiceHead

[Alon Levy]

Hi,

 Currently we support multiple monitors by having:
  single pci = single display channel = single client window
 
 The RANDR architecture doesn't lend itself to this, but on the other hand it
 makes it very easy to have an alternative scheme:
  single pci = single display channel = multiple client windows
 
 RANDR introduces a concept of a CRTC and an OUTPUT. The CRTC scansout a
 portion of the framebuffer onto one or more OUTPUTs. I propose having a 1:1 CRTC:OUTPUT correspondence and introducing two new commands, one on PCI and one for the protocol:
 QXLHead = (id, x, y, width, height)
 SpiceHead = (id, x, y, width, height)

 1. Guest enables a new monitor:
 2. Guest pushes QXLHead command to command ring
 3. Server sends SpiceHead message on the ring's display channel.
 4. Client creates a window out of [x, y, width, height] scanning out of the primary surface (there is one associated primary with the display channel)

Some other notes:
 a. To disable a monitor, send (id, 0, 0, 0, 0) (So there remain invalid range of
 values {(id, x, y, width, height) | max(x, y, width, height)!=0 and min(width,
 height) = 0})
 b. Guest keeps track of the head id.
 c. No idea if this is easy / natural for a WDDM driver, if not we will keep having multiple display channels there, or have to extend the concept of primary to have a number of primary surfaces. Another option is to extend the QXLHead/SpiceHead to have a surface id argument. But I suggest to postpone this to when we actually work on the WDDM driver (unless someone wants to look at it now).
 d. Improvements from this method are:
  single vram bar for multiple heads, single ram bar - more efficient memory allocation
  single display channel means single glz dictionary, single image cache, again more efficient.
  limit of 4 monitors is lifted (the limit of four display channels remains)
  implementation for guest is trivial:
   keep track of enclosing size of primary and resize it as needed
   issue a QXLHead for each output enabled / disabled (corresponds with crtc since we keep a 1:1 relation in the driver).

I have a proof of concept driver, but needs some work yet. I want to get input on this scheme, it is also very related to application remoting, but I haven't given it thought so far.

Alon Levy (2):
  qxl_dev: dynamic head support on single display channel
  protocol: add SPICE_DISPLAY_CAP_HEAD

 spice/protocol.h |    4 ++++
 spice/qxl_dev.h  |   12 ++++++++++++
 2 files changed, 16 insertions(+)

-- 
1.7.10.1