Display transfers with the horizontal downscaling flag were calculating
the wrong output size, causing them to write double the amount of data
intended. It is likely that this was perceived as correct due to a
separate bug in calculating source indices which caused the image to be
padded unless the previous bug was present.
This fixes both issues, correcting flickering issues in 3dscraft,
blargSnes and more (caused by the transfer overwriting the back buffer
which followed) as well as potentially fixing other crashes.
Hardware testing determined that the GSP processes shared memory
framebuffer update info even when no memory transfer or filling GX
commands are used. They are now updated on every interrupt, which isn't
confirmed correct but matches hardware behaviour more closely.
This also reverts the hack introduced in #404. It made a few games
behave better, but I believe it's incorrect and also breaks other games.
PDC0 and PDC1 are both VBlank interrupts. PDC0 was being treated as a
HBlank interrupt and fired many more times than it should. They now both
fire together at 60 Hz. This puzzlingly *improves* apparent framerate on
many applications.
A few other interrupts were being fired inside the GSP command
processing instead of on the actual GPU register writes, so they were
moved there, which should cover direct writes tho those registers not
going through the GX command queue.
This enables the /MP compiler flag, which parallelizes builds of by
distributing compilation of individual object files across workes
processes, instead of being limited to per-project parallelism.
Reduces the time for a full compile from 72 s to 45 s on my machine.
Replace all the C-style complicated buffer management with a std::deque.
In addition to making the code easier to understand it also adds support
for non-POD IdTypes.
Also clean the rest of the code to follow our code style.
Replace the loop-based texture address swizzling code by a bit-twiddling
implementation, providing a very small speed up. Also simplify
addressing code.
Use a new buffer management scheme in the clipper that allows using a
bounded minimal amount of buffer space. Even though it copies more data
it is still slightly faster likely due to using less cache.
The triangle clipper was allocating its temporary input, output and work
buffers using a std::vector. Since this is a hot path, it's desirable to
use stack allocation instead.
Unused OutputVertex attributes were being left un-initialized. The
leftover garbage sometimes decoded as floating-point denormalized
values, causing fallbacks to microcode and massive slowdowns in the rest
of the rasterization pipeline even though the results were unused. By
zeroing the structure we ensure these attributes only contain harmless
zeros.
This handle manager more closely mirrors the behaviour of the CTR-OS
one. In addition object ref-counts and support for DuplicateHandle have
been added.
Note that support for DuplicateHandle is still experimental, since parts
of the kernel still use Handles internally, which will likely cause
troubles if two different handles to the same object are used to e.g.
wait on a synchronization primitive.
- Refactor FS::Archive internals to make Archive creation and lifetime
management clearer.
- Remove the "Archive as a File" hack.
- Implement 64-bit Archive handles.
This is a first step at fixing the conceptual insanity that is our
handling of service and IPC calls. For now, interfaces still directly
derived from Session because we don't have the infrastructure to do it
properly. (That is, Processes and scheduling them.)
All service calls in the CTR OS return result codes indicating the
success or failure of the call. Previous to this commit, Citra's HLE
emulation of services and the kernel universally either ignored errors
or returned dummy -1 error codes.
This commit makes an initial effort to provide an infrastructure for
error reporting and propagation which can be use going forward to make
HLE calls accurately return errors as the original system. A few parts
of the code have been updated to use the new system where applicable.
One part of this effort is the definition of the `ResultCode` type,
which provides facilities for constructing and parsing error codes in
the structured format used by the CTR.
The `ResultVal` type builds on `ResultCode` by providing a container for
values returned by function that can report errors. It enforces that
correct error checking will be done on function returns by preventing
the use of the return value if the function returned an error code.
Currently this change is mostly internal since errors are still
suppressed on the ARM<->HLE border, as a temporary compatibility hack.
As functionality is implemented and tested this hack can be eventually
removed.
This prevents a crash when the buffer size returned by the driver is 0,
in which case no space is allocated to store even the NULL byte and
glGetShaderInfoLog errors out.
Thanks to @Relys for the bug report.
The OpenGL renderer has been revised, with the following changes:
- Initialization and rendering have been refactored to reduce the number of
redundant objects used.
- Framebuffer rotation is now done directly, using texture mapping.
- Vertex coordinates are now given in pixels, and the projection matrix
isn't hardcoded anymore.
This name better represents what the enum does, and is less overloaded
in the context. (The whole register the enum is part of is also called
'format'.)
Several cleanups to the buildsystem:
- Do better factoring of common libs between platforms.
- Add support to building on Windows.
- Remove Qt4 support.
- Re-sort file lists and add missing headers.
This should fix the GL loading errors that occur in some drivers due to
the use of deprecated functions by GLEW. Side benefits are more accurate
auto-completion (deprecated function and symbols don't exist) and faster
pointer loading (less entrypoints to load). In addition it removes an
external library depency, simplifying the build system a bit and
eliminating one set of binary libraries for Windows.
The Git for Windows installer doesn't add the Git binaries to the path
by default. (Due to risk of conflicts with built-in windows commands.)
Unless you have configured your system specially this causes the
scm_rev_gen.js script to fail to find Git. Added more paths to the
script so that it searches in the default msysgit installation
directory, eliminating the need to set the PATH for most environments.
