Commit Graph

36048 Commits

Author SHA1 Message Date
JosJuice
e316d0e94f JitArm64: Don't update dest reg when load triggers exception, part 2
If a host register has been newly allocated for the destination
guest register, and the load triggers an exception, we must make
sure to not write the old value in the host register into ppcState.
This commit achieves this by not marking the register as dirty
until after the load is done.
2021-11-20 23:39:26 +01:00
JosJuice
96190887ce JitArm64: Don't update dest reg when load triggers exception
Fixes a problem introduced in the previous commit.
2021-11-20 23:39:26 +01:00
JosJuice
ab1ceee16f JitArm64: Implement memcheck for lXX/stX without update 2021-11-20 23:39:26 +01:00
JosJuice
8c905e152a JitArm64: Make WriteConditionalExceptionExit more flexible
You can now specify an already allocated register for it to use as a
temporary register, and it also supports being called while in farcode.
2021-11-20 23:37:02 +01:00
JMC47
e5a4a86672
Merge pull request #10055 from JosJuice/jitarm64-reuse-memory
JitArm64: Codegen space reuse
2021-11-20 17:35:24 -05:00
Filoppi
1badceb455 ControllerInterface: fix UpdateReferences() deadlock
Removed useless locks to DeviceContainer::m_devices_mutex, as they were all already protected by m_devices_population_mutex.
We have no interest in blocking other threads that were potentially reading devices at the same time so this seems fine.
This simplifies the code, and I've adjusted a few comments which mentioned possible deadlock that should now be totally gone.

The deadlock could have happen if a thread directly called EmulatedController::UpdateReferences(), while another another thread also reached EmulatedController::UpdateReferences() within a call to ControllerInterface::UpdateDevices(), as the mentioned function locked both the DeviceContainer::m_devices_mutex and s_get_state_mutex at the same time.

The deadlock was frequent on game emulation startup on Android, due to the UpdateReferences() call in InputConfig::LoadConfig() and the UI thread triggering calls to ControllerInterface::UpdateDevices().
It could also have happened on Desktop if a user pressed "Refresh Devices" manually in the UI while the input config was loading.

Also brought some UpdateReferences() comments and thread safety fixes from https://github.com/dolphin-emu/dolphin/pull/9489
2021-11-20 16:54:36 +02:00
JMC47
6ca2897d20
Merge pull request #10233 from AdmiralCurtiss/riivolution-case-sensitive
RiivolutionPatcher: Use case-insensitive filename comparison when searching for files in a folder patch.
2021-11-19 21:42:27 -05:00
Admiral H. Curtiss
0fc563ee2e
RiivolutionPatcher: Use case-insensitive filename comparison when searching for files in a folder patch. 2021-11-20 01:15:18 +01:00
JosJuice
c2aa2818be Android: Make rescan on app start work again
I haven't fully confirmed why the previous commit broke this,
but I imagine it's due to AfterDirectoryInitializationRunner
executing in a different order than before, resulting in
startRescan running before startLoad.
2021-11-19 22:58:05 +01:00
JosJuice
bb475391d2 Android: Make DirectoryInitialization use LiveData
Gets rid of all remaining uses of the deprecated LocalBroadcastManager.
2021-11-19 22:58:05 +01:00
JosJuice
d3a23fe5c2 Android: Enable R8 code shrinking
This decreases our APK size by a few megabytes. Most of the reduction
is from Java libraries that we only use small parts of. Code shrinking
gets rid of all the unused code from these libraries from the APK.

Because I highly value the ability to get stack traces that make
sense, I have specifically disabled obfuscation (automatic renaming
of symbols to short incomprehensible names).

I've only enabled code shrinking for release builds, purely because
I feel like the extra build time (30 seconds on my machine)
would be annoying when you want to make debug builds rapidly.
2021-11-19 22:56:34 +01:00
JMC47
d73f3183ac Compatibility Settings for Red Steel 2
Red Steel 2 has FMVs that get corrupted with Anisotropic filtering.  It
also hangs on dualcore.  While SyncGPU works, both another user and
myself saw much smoother gameplay when using Single Core for unknown
reasons.
2021-11-18 18:57:45 -05:00
JMC47
dbaebdc585
Merge pull request #10222 from phire/fix-copy-filter-clamping
Fix copy filter clamping
2021-11-18 17:48:33 -05:00
Pokechu22
94ccf765af Add option for setting the PNG zlib compression level 2021-11-18 13:10:22 -08:00
JMC47
6f4bbac528
Merge pull request #9956 from Pokechu22/non-power-of-2-wrap-2
VideoCommon: Manually handle texture wrapping and sampling
2021-11-18 13:08:23 -05:00
Pokechu22
95b9941044 Use Fast Texture Sampling by default
This commit changes the default value of Fast Texture Sampling to true, and also moves the setting that controls it to the experimental section of the advanced tab.  This is its own commit so that it can be easily reverted when we want to default to Manual Texture Sampling.

Co-authored-by: JosJuice <josjuice@gmail.com>
2021-11-17 21:29:57 -08:00
Pokechu22
1adff1c467 VideoCommon: Skip textureQueryLevels if it doesn't exist 2021-11-17 21:28:39 -08:00
Pokechu22
bdcfb31187 VideoCommon: Handle custom texture sizes correctly
Specifically, when using Manual Texture Sampling, if textures sizes don't match the size the game specifies, things previously broke.  That can happen with custom textures, and also with scaled EFB copies at non-native IRs.  It breaks most obviously by not scaling the texture coordinates (so only part of the texture shows up), but the hardware wrapping functionality also assumes texture sizes are a power of 2 (or else it will behave weirdly in a way that matches how hardware behaves weirdly).  The fix is to provide alternative texture wrapping logic when custom texture sizes are possible.
2021-11-17 21:28:36 -08:00
Pokechu22
93eea7cb13 VideoCommon: Add option to use old behavior (Fast Texture Sampling)
Co-authored-by: JosJuice <josjuice@gmail.com>
2021-11-17 21:27:32 -08:00
Pokechu22
ee80298ca4 VideoCommon: Implement diagonal LOD
Note that both GLSL and HLSL provide a fwidth (fragment width) function defined as `fwidth(p) = abs(dFdx(p)) + abs(dFdy(p))`.  However, it's easy enough to implement this ourselves (and it makes the code a bit more obvious).
2021-11-17 20:04:34 -08:00
Pokechu22
b9288212a0 Software: Adjust diagonal LOD implementation
This produces behavior matching the behavior on hardware (see Wario's Gold Mine in Mario Kart Wii).
2021-11-17 20:04:34 -08:00
Pokechu22
51e3334526 VideoCommon: Use coarse derivatives for Manual Texture Sampling if possible 2021-11-17 20:04:34 -08:00
Pokechu22
ddf2691395 VideoCommon: Manually handle texture wrapping and sampling 2021-11-17 20:04:34 -08:00
Pokechu22
4a9b26de86 VideoCommon: Expose SamplerState to shaders
The benefit to exposing this over the raw BP state is that adjustments Dolphin makes, such as LOD biases from arbitrary mipmap detection, will work properly.
2021-11-17 20:04:34 -08:00
Pokechu22
9ef228503a VideoCommon: Provide raw texdims to shaders 2021-11-17 20:04:34 -08:00
Pokechu22
a273b65566 RenderState: Use operator== for operator!= and adjust constructors 2021-11-17 20:04:34 -08:00
Pokechu22
6236a0d494 Eliminate SamplerCommon 2021-11-17 20:04:34 -08:00
Pokechu22
3096f77ba0 Eliminate SamplerCommon::AreBpTexMode0MipmapsEnabled
This was added in 0b9a72a62d but became irrelevant in 70f9fc4e75 as the check is now self-explanatory due to a rejiggering of the bitfields.
2021-11-17 20:04:34 -08:00
Pokechu22
d2041b4c2a VideoCommon: Add signed version of BitfieldExtract 2021-11-17 20:04:33 -08:00
Pokechu22
555a93057c VideoCommon: Allow BitfieldExtract in specialized shaders 2021-11-17 20:04:33 -08:00
Mai M
8b57aad8ed
Merge pull request #10229 from JosJuice/android-game-list-livedata
Android: Make GameFileCacheManager use LiveData
2021-11-17 17:21:39 -05:00
JosJuice
2941cf8d94 Android: Make GameFileCacheManager use LiveData, part 2
Gets rid some uses of the deprecated LocalBroadcastManager.

One note about the changes in GameFileCacheManager itself:
The change from compareAndSet to getValue followed by
setValue is actually safe, because startLoad and startRescan
only run from the main thread, and only the main thread ever
sets the flags to true. So it's impossible for any other thread
to change the flag in between the getValue and the setValue.
2021-11-17 21:49:51 +01:00
JosJuice
857963b336 Android: Make GameFileCacheManager use LiveData, part 1 2021-11-17 21:29:11 +01:00
Mai M
735fd60e52
Merge pull request #10217 from JMC47/bombermanjetters
Force Dual Core on for Bomberman Jetters
2021-11-16 23:29:40 -05:00
Mai M
b14d982d36
Merge pull request #10225 from JosJuice/android-game-settings-title
Android: Properly set game settings title when navigating backwards
2021-11-16 23:26:38 -05:00
Mai M
f4a32834b9
Merge pull request #10223 from JosJuice/enforce-skirts
GameINI: Fix comments causing errors in GALE01r0.ini
2021-11-16 23:24:52 -05:00
Mai M
f10d95e2b2
Merge pull request #10224 from JosJuice/android-not-service
Android: Make GameFileCacheService not be a service
2021-11-16 23:22:22 -05:00
JosJuice
bb2ee279cd Android: Properly set game settings title when navigating backwards 2021-11-16 22:09:40 +01:00
JosJuice
ffd8cd059c Android: Make GameFileCacheService not be a service
The past few Android releases have been adding restrictions
to what services are allowed to do, for the sake of stopping
services from using up too much battery in the background.
The IntentService class, which GameFileCacheService uses,
was even deprecated in Android 11 in light of this.

Typically, the reason why you would want use a service instead of
using a simple thread or some other concurrency mechanism from the
Java standard library is if you want to be able to run code in the
background while the user isn't using your app. This isn't actually
something we care about for GameFileCacheService -- if Android wants
to kill Dolphin there's no reason to keep GameFileCacheService
running -- so let's make it not be a service.

I'm changing this mainly for the sake of future proofing, but there
is one immediate (minor) benefit: Previously, if you tried to launch
Dolphin from Android Studio while your phone was locked, the whole
app would fail to launch because launching GameFileCacheService
wasn't allowed because Dolphin wasn't considered a foreground app.
2021-11-16 21:45:06 +01:00
Scott Mansell
9bbc843542 VideoSoftware: Fix copy filter clamping 2021-11-17 09:29:16 +13:00
JosJuice
31bfbca923 Android: Split up GameFileCacheService.onHandleIntent 2021-11-16 21:15:49 +01:00
Scott Mansell
7128befb39 Fix copy filter clamping regression in Spyro
This fixes horizontal lines in the bloom effect of Spyro: A Hero's Tail,
which is a regression caused by PR #10204

Screenshot of regression:
https://user-images.githubusercontent.com/138484/142030503-90fcd8d5-63d3-4820-874a-72e9be0c4768.png

Fixed:
https://user-images.githubusercontent.com/138484/142031598-b85ff55c-1302-4e4d-bcb2-57848974056b.png

Spyro uses an 640x80 pixel sub-buffer within the EFB to calculate
it's bloom effects, which it places below the main 640x448 buffer.

EFB layout:
https://user-images.githubusercontent.com/138484/142030573-e933b6ae-c37e-4be6-86d4-0bc779b92535.png
Note: Colors are wrong because the main color buffer uses RGBA6,
      while the bloom is calculated in RGB8

This allows it to do bloom without backing up part of the EFB to
main memory, as most games do.

But, since some of the sub-buffers used in the bloom effect are taller
than 80 pixels, they need to be sliced up into smaller sub, sub buffers
which get combined later when copied to main memory.

At one point, a 320x224 buffer is broken up into 320x80, 320x64 and
320x80 slices. These are copied out with the copy filter set to a
vertical blur.

Because there was an off-by-one errror in the clamping coordinates,
the bottom line of the color buffer would be blurred into
the top of each slice.

Final combined EFB copy:
https://user-images.githubusercontent.com/138484/142031360-2c076839-7c96-4b3b-a093-d899d0a2c7ae.png

Fixed version:
https://user-images.githubusercontent.com/138484/142031370-72e41a35-3b3e-4662-a483-79203e357ecc.png

Before #10204 the copy filter wasn't enabled for efb copies, and most
other games don't do this type of slicing.

FIFO CI shows that a few other games are effected, it's always just a minor difference to the top line where there was previously a slight hint of garbage.
2021-11-17 06:12:46 +13:00
JosJuice
89cb311c86 GameINI: Fix comments causing errors in GALE01r0.ini
Same problem as 658eed4... Except instead of just making the
comments use #, I'm actually removing some of the codes entirely
because they are of questionable value.

Should fix https://bugs.dolphin-emu.org/issues/12737.
2021-11-16 17:58:16 +01:00
Mai M
0b81640dd1
Merge pull request #10221 from shuffle2/msbuild
msbuild: remove workaround for older winsdk
2021-11-15 19:53:35 -05:00
Shawn Hoffman
5e1fee7dbd msbuild: remove workaround for older winsdk 2021-11-15 16:23:33 -08:00
JosJuice
7292ac0ce5 Android: Remove DirectoryInitialization from AndroidManifest.xml
It isn't a service, so it shouldn't be listed as one.
2021-11-15 23:23:17 +01:00
JMC47
26fdc19c8d
Merge pull request #10204 from Pokechu22/efb-copy-filter
VideoCommon: Use the copy filter for EFB copies as well as XFB copies
2021-11-15 17:04:27 -05:00
Mai M
b919557e25
Merge pull request #10219 from shuffle2/extern
msbuild: use /external:anglebrackets
2021-11-15 11:10:38 -05:00
Léo Lam
d17dcf72ff
Merge pull request #10201 from JMC47/mp6-codeadjust
GameINI: Fix Comment Causing Parsing Errors in Mario Party 6/7 INIs.
2021-11-15 14:59:31 +01:00
JMC47
658eed405a GameINI: Fix Comments in Mario Party 6/7 AR Codes
Using the "*" symbol as a comment only works for Gecko Codes for some
reason.  I changed it to "#".
2021-11-15 07:33:49 -05:00