This function is internal to DIX (used for assembling XQueryFont replies
and not used by any drivers (not even proprietary NVidia), and really
shouldn't ever called by them at all. It's been one the many victims of a
really unthoughtful mass export back 17 years ago.
That'S a necessary step for refactoring the RPC reply mechanics.
Since it's never used by any driver at all, we can easily to this change
within the current ABI level.
Fixes: d6cbd4511e
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Fix pedantic warnings:
> ../include/os.h:144:71: warning: ISO C does not allow extra ‘;’ outside of a function [-Wpedantic]
> 144 | XNFalloc(unsigned long /*amount */ ) __attribute__((returns_nonnull));;
> ../include/os.h:158:76: warning: ISO C does not allow extra ‘;’ outside of a function [-Wpedantic]
> 158 | XNFcallocarray(size_t nmemb, size_t size) __attribute__((returns_nonnull));;
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
modesetting has a sort of broken concept of getting the cursor dimensions since 1f41320e1c, which causes issues on virtual machines and older AMD hardware. Our solution is to use SIZE_HINTS which is available in i915 since 2023 otherwise fallback to getting the cursor size provided by the driver, if a driver wants to have a more optimal cursor size, it is more than welcome to implement SIZE_HINTS in it's DRM interface.
Taken from my own fork of Xorg.
v2: Fix some bugs in the implemenation.
It is patch 1/7 of a series that provides a convenient way to specify
IgnoreABI and module search paths on a per-driver basis.
It adds #ifdef CONFIG_LEGACY_NVIDIA_PADDING wherever it should be but
is missed.
Signed-off-by: Oleh Nykyforchyn <oleh.nyk@gmail.com>
Commit 3cb3024fea removed pScratchPixmap field
in ScreenRec, which broke legacy Nvidia (proprietary) drivers (470.x), thus
we should add an empty/dummy field here, to ensure correct padding. But this
would break ABI again - can't do that within minor release line.
As compromise, adding a *build time* option CONFIG_LEGACY_NVIDIA_PADDING for
this, so operators/packagers can opt-in to this change.
As it breaks ABI, the option is disabled by default until the next major release
and intended for EXPERTS ONLY who need nvidia390 or nvidia470 drivers.
Note that ALL DRIVERS should be rebuild if it is applied!
This compile-time option, along with the hacks needed to support it
in a non-abi-breaking way, should be droppen in the next major release.
Co-authored-by: Oleh Nykyforchyn <oleh.nyk@gmail.com>
Signed-off-by: Oleh Nykyforchyn <oleh.nyk@gmail.com>
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The global (exported) serverGeneration field is `unsigned long`, while
many other places copy it and compare it two other integer types, eg.
plain `int` (which is signed). Even if it's unlikely ever reaching such
high number of generations that it will ever make trouble, it's still
a good idea to clean this up and use the same type everywhere.
For clearity, introducing a typedef `x_server_generation_t` which is
used everywhere, instead of raw `unsigned long`.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Fix compiler warnings by explicit type casts.
In this case, we know that those values can't be bigger than int can take,
neither become negative.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The client index (in the client table) can never be negative, thus
make it an unsigned short.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The epoll-shim dependency did not get detected on FreeBSD,
because of missing prefix and dependency arguments in has_function()
call in include/meson.build.
Also added the path of epoll-shim headers include directory
(/usr/local/include/libepoll-shim/) to the compiler include search
paths for proper detection and build.
Signed-off-by: b-aaz <b-aazbsd@proton.me>
some bit better naming for config symbols.
Yet leaving the old one defined, until all drivers have kept up.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
../glx/indirect_table.c
In file included from ../glx/indirect_table.c:28:
In file included from ../glx/glxserver.h:66:
../include/glx_extinit.h:33:28: warning: redefinition of typedef '__GLXscreen' is a C11 feature [-Wtypedef-redefinition]
33 | typedef struct __GLXscreen __GLXscreen;
| ^
../glx/glxscreens.h:113:28: note: previous definition is here
113 | typedef struct __GLXscreen __GLXscreen;
| ^
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Reduce complexity for things that really don't matter much:
The ddxBeforeReset() function is called when the Xserver going to reset
(new server generation). Right now, the only DDX really needing that is
Xwin, on all the others it's just no-op.
We've got an extra complicated build logic, which ifdef's out this all when
Xwin isn't built at all. The saving is extremely minimal - just skipping
few stub functions, which in most sessions aren't even called.
Therefore, get rid of this extra complexity that isn't giving us any
notable gain.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Both are potentially tunable variables, and MAXCLIENTS is (still) used by
intel-driver, but also by os specific parts that must not include misc.h
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
These defines are already public, used by consumers of SetNotifyFd(),
but also needed in places where os.h cannot be included.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
The swapping is so trivial, we don't need several extra functions
for that - just do it right where the header fields are swapped.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Xace callbacks are needed in many places, and the their overhead (when not
actually used) is really minimal. So it doesn't make much sense having
extra complexity for disabling it at build time.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
In contrast to the already existing ScreenClose hook, this one is
called *after* the driver's CloseScreen() proc. That's required for
some extensions (eg. damage) where drivers still need to call in
inside of their CloseScreen procs.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Composite extension is always enabled for pretty long time now, but some
drivers (eg. xf86-video-intel) still relying on this symbol being set,
otherwise assuming to run w/o composite and doing crazy things.
Fixes: d708b28adc
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
It always had it's own lifecycle (not been part of Xorg releases),
doesn't make sense to maintain a competing implementation that we
won't use anyways.
Once that's gone, we can also drop few things in core/dix that had
been added just for xwayland only.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
in struct _Screen (include/scrnintstr.h) there is a PRIVATE_LAST-sized array,
having another private increases PRIVATE_LAST, the array increases in size,
moving everything below it down. this breaks ABI for nvidia driver
Signed-off-by: dasha_uwu <dasha@linuxping.win>
Adding data structure and initial data for namespace configuration.
Built-in namespaces are ROOT and ANONYMOUS.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Add tiny skeleton for the namespace extension. Disabled by default,
can be enabled via +extension arg, but doesn't actually do something yet.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Simplify cases where callers need to check whether an entry already is
in a list before adding / appending.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
This function walks through the list and checks whether an entry is already
present. This check sometimes is neccessary, since trying to add an entry
twice has catastrophic consequences: iteration will become endless loop.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Up until now, xorg_list's always need to be explicitly initialized by
calling xorg_list_init(), otherwise next access attempt will segfault.
This is adding extra complexity and risk of subtle bugs.
Adding some extra NULL check are trivial and their cost is so tiny
that it's even hard to practically measure.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Not used by any drivers, and shouldn't be used by them.
Needs to _X_EXPORT'ed, as long as glx is a separate module.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
This allows us to do further probing in the included meson files:
Individual subdirectories (eg. DDXes, extensions, OS layer, ...)
can now probe things that are only relevant to them - no need to fill
the already too fat includes/meson.build with even more things.
Preparation for upcoming commits that'll make us of that.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Lots of logging functions, especially init and teardown aren't called
by any drivers/modules, so no need to keep them exported.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
systemd is linux specific and the actual implementation is under the
os-support layer of xfree86 ddx. Thus no need to keep it in global
include directory, putting it onto the linux specific os-support instead.
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
