Canonicalize all reply structures onto stack allocation and static
initialization, like already done in most other extension. So make
the code easier to understand and allow further simplifications by
subsequent commits. Also gaining a little bit efficiency by skipping
some heap allocations.
Dynamically sized buffers (where the upper bound isn't known), are
still allocated on heap.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The dispatcher functions are much more complex than they're usually are
(just switch/case statement). Bring them in line with the standard scheme
used in the Xserver, so further steps become easier.
It's also much cleaner to use the defines from proto headers instead of
raw numbers.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The current way of switching between Xinerama and single-screen handlers
is quite complicated and needs call vector tables that are changed on
the fly, which in turn makes dispatching more complicated.
Reworking this into a simple and straight code flow, where individual request
procs just look at a flag to decide whether to call the Xinerama or single
screen version.
This isn't just much easier to understand (and debug), but also removes the need
or the call vectors, thus allowing further simplification of the dispatcher.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
In order to allow simplifying the reply send path, collect the reply
fragments into one buffer, instead of arbitrary number of WriteToClient()
calls. This also makes it much easier for potentially new purely packet-based
transports which (eg. binder) that would need their own stream parsing logic.
This xres function is an exceptionally hard case, since payload is constructed
step by step, and it's size only known when finished. The current way of the
fragment handling still has lots of room for improvement (eg. using very small
number of allocations), but leaving this for later exercise.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Collect the few bits in a local array, so one WriteToClient() call is
sufficient. That's also easing further simplifications in upcoming commits.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Simplify allocaton by putting the small temporary int array onto stack.
This also allows further simplifications by upcoming commits.
The upper bound is determined by the number of resource types registered
in the server - this can only be increased by writing new extensions.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Collect the few bits in a local array, so one WriteToClient() call is
sufficient. That's also easing further simplifications in upcoming commits.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Simplify allocaton by putting the small temporary int array onto stack.
This also allows further simplifications by upcoming commits.
Note: there's an upper bound by compile time defines (theoretically, can
be increased by special cmdline args, that virtually nobody ever using).
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
When using static struct initialization, fields not explicitly
stated are automatically zero.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Instead of having lots of #ifdef's, consolidating the conditionally
compiled fsync() call into a tiny inline helper.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Since we're not indirectly writing via FILE anymore, this option has
become meaningless: it meant flushing out our in-process buffer to
the kernel, but we're now doing direct write() calls anyways.
xf86 still accepts the "flush" config file flag for backwards compatibility,
but it hasn't any practical meaning anymore.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Instead of maintaining both the logfile fd, as well as ANSI FILE pointer,
simplify it to just a fd.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Increase driver build test coverage by adding more drivers.
Only left out those which won't compile on Linux (yet).
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new pixmap destroy notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons,
so use the new pixmap destroy notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new pixmap destroy notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new pixmap destroy notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new pixmap destroy notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Right now, extension specific pixmap destruction procedures are implemented
by wrapping the ScreenRec's DestroyPixmap() proc pointer: the extensions are
storing the original pointer in their private data and putting in their own one.
On each call, their proc restores the original one, calls it, and switches back
again. When multiple extensions doing so, they're forming a kind of daisy chain.
(the same is done for lots of other procs)
While that approach is looking nice and elegant on the drawing board, it's
complicated, dangerous like a chainsaw and makes debugging hard, leading to
pretty blurred API borders.
It's even getting worse: the proc also has to do ref counting, and only destroy
the pixmap if refconter reaching zero - that's all done in the individual screen
drivers. Therefore, all extensions must check for refcnt == 1, in order to know
when to really act.
This commit introduces a simple approach for letting extension hook into the
pixmap destruction safely, w/o having to care much about side effects with
the call chain. Extensions now can simply register their destructor proc
(and an opaque pointer) and get called back - w/o ever having to mess with
the ScreenRec's internal structures.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new screen close notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new screen close notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new screen close notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new screen close notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new screen close notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new screen close notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Right now, extension specific actions on screen closing implemented by wrapping
the ScreenRec's PositionWindow() proc pointer: the extensions are storing the
original pointer in their private data and putting in their own one. On each
call, their proc restores the original one, calls it, and switches back again.
When multiple extensions doing so, they're forming a kind of daisy chain.
(the same is done for lots of other procs)
While that approach is looking nice and elegant on the drawing board, it's
complicated, dangerous like a chainsaw and makes debugging hard, leading to
pretty blurred API borders.
This commit introduces a simple approach for letting extension hook into the
screen closing path safely, w/o having to care much about side effects with
the call chain. Extensions now can simply register their hook proc (and an
opaque pointer) and get called back - w/o ever having to mess with the
ScreenRec's internal structures. These hooks are called before the original
vector (usually handled by DDX/screen driver directly) is called.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window position notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window position notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window position notify hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Right now, extension specific actions on window positioning are implemented
by wrapping the ScreenRec's PositionWindow() proc pointer: the extensions are
storing the original pointer in their private data and putting in their own one.
On each call, their proc restores the original one, calls it, and switches back
again. When multiple extensions doing so, they're forming a kind of daisy chain.
(the same is done for lots of other procs)
While that approach is looking nice and elegant on the drawing board, it's
complicated, dangerous like a chainsaw and makes debugging hard, leading to
pretty blurred API borders.
This commit introduces a simple approach for letting extension hook into the
window positioning path safely, w/o having to care much about side effects
with the call chain. Extensions now can simply register their hook proc
(and an opaque pointer) and get called back - w/o ever having to mess with
the ScreenRec's internal structures. These hooks are called before the original
vector (usually handled by DDX/screen driver directly) is called.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
2do: should check whether it's better to directly assign the screen procs,
w/o any wrapping at all.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Wrapping ScreenRec's function pointers is problematic for many reasons, so
use the new window destructor hook instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Enrico Weigelt, metux IT consult