Yes, two changes in one commit. Sorry 'bout that.
The first change ensures that when pending property values have been
changed, a mode set to the current mode will actually do something, rather
than being identified as a no-op. In addition, the driver no longer needs to
manage the migration of pending to current values, that is handled both
within the xf86 mode setting code (to deal with non-RandR changes) as well
as within the RandR extension itself.
The second change eliminates the two-call Create/AttachScreen stuff that was
done in a failed attempt to create RandR resources before the screen
structures were allocated. Merging these back into the Create function is
cleaner.
Pending Properties take effect when the driver says they do, so provide an
API to tell DIX when a property effect is made. Also, allow driver
to reject property values in RRChangeOutputProperty.
RRModes are referenced by the resource db, RROutput and RRCrtc structures.
Ensure that the mode reference count is decremented each time a reference is
lost from one of these sources. The missing destroys were in
RRCrtcDestroyResource and RROutputDestroyResource, which only happen at
server reset time, so modes would be unavailable in subsequent server
generations.
The xf86 mode setting code was mis-using this field to try and store a
pointer to a DisplayModeRec, however, each output has its own copy of every
DisplayModeRec leaving the one in in the RRModeRec devPrivate field pointing
at a random DisplayModeRec.
Instead of attempting to rectify this, eliminating the devPrivate entirely
turned out to be very easy; the DDX code now accepts an arbitrary RRModeRec
structure and set that to the hardware, converting it on the fly to a
DisplayModeRec as needed.
The RandR protocol spec has several requests in support of user-defined
modes, but the implementation was stubbed out inside the X server. Fill out
the DIX portion and start on the xf86 DDX portion. It might be necessary to
add more code to the DDX to insert the user-defined modes into the output
mode list.
RandR 1.0 sizeID must be computed the same way every time, so when reporting
it in the ScreenChangeNotify event, just construct the usual 1.0 data block
and use that.
subpixel geometry information can be computed by looking at the connected
outputs and finding any with subpixel geometry and using one of those for
the global screen subpixel geometry. This might be improved by reporting
None if more than one screen has information and they conflict.
It was using REQUEST_SIZE_MATCH (client request length must equal request size)
rather than REQUEST_AT_LEAST_SIZE (client request length must be at least
big enough for request size), and this request has data following the request
structure.
(cherry picked from commit 4f2f3233c8)
The RRScreenSizeSetRange function is used externally for 1.2 API drivers,
but can also be used in the 1.0 compatibility code. This also ensures that
the right changed bits are set so that clients are correctly notified when
the range changes.
RRGetInfo can return an error, use that to return BadAlloc to clients
instead of blindly going on with various requests.
Events and internal data structures need to be updated whenever the physical
or pixel size of the screen changes. The code was ignoring the physical
size, so changing only that would not be registered anywhere.
Mode lines reflect the monitor mode, not the projected size into the frame
buffer. Flip width/height around so that the dimensions are oriented
correctly.
RandR 1.0 refresh rates were scrambled when working with a 1.2 driver that
returned sizes in a mixed order. SetScreenConfig was treating RRCrtcSet as
returning an RandR status instead of a Bool.
The config time in the RandR protocol reflects when the hardware state has
changed. It was getting changed anytime the driver changed the usage
of the hardware as well.
When an output no longer reports the current mode, it must still be included
in the list advertised by the X server. Walk the crtcs to ensure it is
included.
RRCrtcDestroyResource and RROutputDestroyResource had matching
bugs that would overwrite memory past the end of the storage
of the crtc or output arrays. Oops.
Now that resources can be created during server initialization, make sure
the crtc, output and mode resource types are created before attempting to
create associated resources.
This patch tracks the protocol changes which introduce more complex
semantics for RandR output properties including pending and valid value
information.
To allow RandR objects to be created before the screen object exists,
the resource types must be registered with the resource database.
A driver wishing to create RandR objects must call RRInit before doing so.
Also, fix a segfault when setting Output data before it is associated with a
screen.
xf86 drivers need to create RandR object in the PreInit stage,
before the ScreenRec is allocated. Changing the RandR DIX code
to permit this required the addition of functions that later associate the
objects with the related screen.
An additional change is that modes are now global, and no longer associated
with a specific screen. This change actually makes mode management cleaner
as there is no more per-screen list of modes to deal with.
This changes the RandR 1.2 ABI/API for drivers.
Add function to keep pointer within valid crtc areas.
Finish event delivery and swapping code.
Separate configuration from layout changes to send correct events.
Output options and mode origins both affected driver ABI. memmove mistakes
were causing 'Freeing resource which isn't there' messages.
Prune unused non-user defined modes from available list now.
Provide a Xinerama implementation when DIX version isn't enabled. This
version exposes each crtc as a separate 'screen' and reports the size of
that patch. The extension also sends ConfigureNotify events to the root
window whenever crtcs change so that applications will re-fetch xinerama
information. This actually works for metacity.