| Age | Commit message (Collapse) | Author |
|
`slice_length` is defined as an int, but was being zero-extended to 64
bits on x64. We should match what we do for ARM64 and sign-extend, as
discussed in PR #14944.
|
|
|
|
CMake: Bundle dolphin-tool on macOS
|
|
Skylanders: only play audio received on endpoint 2
|
|
GameSettings: Remove SyncOnSkipIdle for Battalion Wars 1&2
|
|
This avoids problems in the unrelated game Band Hero.
|
|
Users on Discord have been reporting issues with this setting enabled. The reason given for disabling `SyncOnSkipIdle` 13 years ago was "Idle skipping slows down the game." Re-enabling it gets rid of the issue entirely.
|
|
|
|
DolphinQt: Use CMAKE_CURRENT_BINARY_DIR to temporarily store translation files during build on macOS
|
|
files during build on macOS
|
|
I do not own an Apple device and as such this is completely untested by myself. A user requesting this feature on Discord has confirmed this makes dolphin-tool show up. Most of these changes come from looking at what already exists for DolphinQt and the Mac Updater.
Related issue: https://bugs.dolphin-emu.org/issues/13259
|
|
Core: Bound legacy savestate LZO decompression
|
|
Core: Bound FIFO memory updates
|
|
VideoCommon: error if the same asset id is used with a different type
|
|
|
|
Legacy savestate headers controlled both the compressed input length
and the decompression buffer size. The input was read into a fixed-size
buffer without validation, and the unsafe LZO API could write past its
output buffer.
Use the legacy block size as the output capacity, reject oversized
compressed blocks, and use the bounds-checking LZO decompressor.
|
|
FIFO logs control each memory update address and data size. Playback
masked only the start address into RAM or EXRAM, then copied the
complete update contiguously. An update reaching the end of a mapped
region could therefore write beyond host-mapped emulated memory.
Resolve the target through MemoryManager and reject updates larger than
the remaining valid span before copying them.
|
|
Replace std::chrono::high_resolution_clock usages with a guaranteed monotonic clock.
|
|
high_resolution_clock.
|
|
high_resolution_clock.
|
|
Revert "GameINI: Add patches to limit internal framerate in Driver: San Francisco."
|
|
Francisco."
This reverts commit b7e447e21258f103cbaa1bda1ec6b7a4ce3d4e97. The patch
breaks the game according to https://bugs.dolphin-emu.org/issues/14156.
|
|
Android: Use TitleDatabase for game list
|
|
AMMediaboard: Update closesocket error code approriately
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Minor cleanups with "parenthesized initialization of aggregates".
|
|
ReceiveL2capData forwarded HID output reports to m_hid_source without
checking it. SetSource accepts nullptr (e.g. Wii Remote set to "Real"
with no real remote connected), and the game's L2CAP channel can still
be open when that happens, so the next output report crashed the CPU
thread with a null dereference.
Seen on Android with Skylanders: Spyro's Adventure and Wii Remote 1 set
to Real Wii Remote.
|
|
EXI: Show on-screen error when USB Gecko fails to bind
|
|
Flatpak: Do not set the flathub::manifest tag
|
|
This resolves the recent build failures introduced by https://github.com/flathub-infra/flatpak-builder-lint/commit/df1c8f992aa5620fc473efe6c6e96b9f3b11e1ba. Nothing seems to break with it being removed.
|
|
Previously, if the USB Gecko server failed to bind to any port in the
55020-55030 range, it silently returned with no indication to the user.
Now displays an on-screen message and logs an error.
|
|
VideoBackends / VideoCommon: split draw calls into separate commit and render functions
|
|
DolphinQt: Fix crash from stale timer event in ToolTipWidget
|
|
workaround.
|
|
initialization of aggregates".
|
|
MathUtil: C++20/23 modernisations
|
|
HW/Wiimote: Speaker logic accuracy fixes, comments, and code cleanups.
|
|
feat(retroachievements): implement achievement notification sounds
|
|
|
|
standard deviation
UpdateStats() was calculating the standard deviation from the entirety of m_dt_queue, which has a noticeable performance impact when the frame rates are very high.
On a scene so light it runs at 3000 FPS, UpdateStats() was taking around 5% of the profiling (at default 1s sample window). Now it doesn't.
|
|
For EFB, optimize the case where bytes_per_row == stride.
For XFB, do that, and make sure that autovectorization is used without intrinsics. They were only used on x86, so that should improve performance on ARM.
(MSVC ARM specifically will load to register on every iteration, and I give up on this stupid compiler, but it should still be better than before.)
For reference, on x86 with SIMD, on a scene that runs at 3000 FPS, this function takes 1.5% of time according to the profiler.
I spent way too much time making all the compilers cooperate, and I hate it.
|
|
render steps
|
|
|
|
Vulkan: Allow custom drivers on Mali devices
|
|
Android: Add global hotkeys
|