Use sched_get_priority_min for priority
Accommodates other unixes with odd minimum priorities.
Manually handle save/restore of scheduling.
Add space. (formatting)
Exit on failure to get scheduler parameters.
Add missing braces.
Copy and edit existing sched_parm rather than new.
May be relevant on some exotic unixes.
Gamescope currently uses 1024 VkFence objects to
synchronize CPU and GPU work. The NVIDIA Vulkan
driver currently creates a file descriptor for
each VKFence object created. To ensure there are
sufficient file descriptors, attempt to raise the
limit to 2048 file descriptors if it is currently
lower than that. Failure to raise the limit for
any reason is non-fatal, as the increased limit
is not required on some drivers.
getopt() will eat the "--" separator if there's one. Rely on the
value of optind after the getopt() loop finishes to figure out
whether a command has been specified.
Calling nice() in the very beginning ensures that Mesa worker threads in
gamescope and Xwayland can properly reduce their priority without us stomping
it later.
Move away from posix_spawn, since there's no easy way (that seems to work)
to keep priority normal for the spawned process and get the right behaviour
for gamescope and Xwayland.
The intermediate fork() lets us simplify the logic for LD_PRELOAD rewriting.
Avoid setting it for our children processes, by spawning them in a new
thread group and setting group-wide priority, which also conveniently
gets Xwayland.
wlroots headers can mostly be included with extern "C" blocks. Two
exceptions are the use of `static` for array args and `class` as a
struct field. These can be #define'ed to avoid C++ build errors.
This allows the whole project to be built as a single C++ codebase,
without having to maintain hybrid header files.
For now just mark one repaint per refresh interval without any particular
alignment, as that can be hard to get in nested. We'll use the display
timing extensions for that later.
We'll have it align properly in embedded mode soon.
- Add a commit wait thread that just waits for the work associated with
an in-bound dma-buff is done. We'd like to be getting a fence directly
from Wayland, but in the absence of that we'll make our own by tapping
into implicit-sync, DX11-style. gpuvis logging of the outcome of the
waits in the thread are consistent with the end of GPU work.
- Rework wayland commit handoff between wlserver and steamcompmgr to let
steamcompmgr have full access to the commit queue, letting it remove
elements in the middle and still keep going, compared to the previous
interface. This lets steamcompmgr get one commit per window per loop,
for now. We should be able to kick off waits and ref/unref resources
for all pending commits soon, with some work on the accounting first.
This should also fix an issue introduced recently where the overlay
could get really starved in embedded mode, as we were just getting
the first commit in the queue for a single frame.
Lower half is getting eaten by something, presumably underflow as it's black.
Make EGL and DRM coexist; make ensure_win_resources safe to call regardless
of what resources we want, so outer code doesn't have to care.
