Commit Graph

5074 Commits

Author SHA1 Message Date
JosJuice
c204b33314 VideoCommon/BPStructs: Add a missing bounds check
Happened to find this when working on the previous commit.
2024-04-20 18:31:08 +02:00
JosJuice
3cfa233b63 VideoCommon: Use GetSpanForAddress safely in texture decoding
Now only VertexLoader remains... But that one might be tricky.
2024-04-20 18:31:08 +02:00
JosJuice
5c9bb80638 Memmap: Replace GetPointer with GetSpanForAddress
To ensure memory safety, callers of GetPointer have to perform a bounds
check. But how is this bounds check supposed to be performed?
GetPointerForRange contained one implementation of a bounds check, but
it was cumbersome, and it also isn't obvious why it's correct.

To make doing the right thing easier, this commit changes GetPointer to
return a span that tells the caller how many bytes it's allowed to
access.
2024-04-20 18:31:08 +02:00
Jordan Woyak
df7dd5d36a Config: Expose GFX_OVERLAY_PROJ_STATS setting in UI. 2024-04-18 14:45:28 -05:00
Jordan Woyak
1e27183dff VideoCommon: Show the average projection to viewport ratio in statistics
window.
2024-04-18 14:45:28 -05:00
Admiral H. Curtiss
8268ae1cf7
Merge pull request #12698 from iwubcode/xfb_copy_hash_in_present
VideoCommon: add xfbs hashes to present info if available
2024-04-18 17:11:49 +02:00
Admiral H. Curtiss
b3939052b4
Merge pull request #12436 from Filoppi/frame-dump-raw-internal-resolution
Frame dump at raw internal resolution
2024-04-13 03:39:37 +02:00
Admiral H. Curtiss
2f13be5a2d
VideoConfig: Adjust FrameDumpResolutionType enum class to style guidelines 2024-04-13 03:21:39 +02:00
iwubcode
abf380b705 VideoCommon: add xfb hash values (if available) to present information 2024-04-09 18:59:33 -05:00
iwubcode
e5b7b2e9ee VideoCommon: add xfbs presented to video events 2024-04-09 18:57:32 -05:00
JosJuice
54773bc5d2 VideoCommon: Remove calls to GetPointer
This fourth part of my series of patches to get rid of unsafe uses of
GetPointer takes care of the "easy" cases in VideoCommon. Three uses of
GetPointer now remain in Dolphin: VertexLoaderManager, TextureInfo, and
the software renderer's TextureSampler.
2024-04-09 21:08:57 +02:00
mitaclaw
eb92d6f0a8 Core::GetState: Avoid Global System Accessor 2024-04-08 16:23:23 -07:00
Filoppi
66592f79f2 Video: remove enforced resolution least common multiple of 4 when dumping screenshots and not videos (only videos encoders have this limit).
NOTE: this will likely trigger FIFOCI differences.
2024-04-08 22:54:45 +03:00
Filoppi
72db62e178 Video: split frame dumping settings into 3 resolution dumping modes
also polish aspect ratio related code for clarity
2024-04-08 22:54:45 +03:00
Pokechu22
a3951dc2d7 Fix out of bounds accesses for invalid vertex component formats
On all platforms, this would result in out of bounds accesses when getting the component sizes (which uses stuff from VertexLoader_Position.h/VertexLoader_TextCoord.h/VertexLoader_Normal.h). On platforms other than x64 and ARM64, this would also be out of bounds accesses when getting function pointers for the non-JIT vertex loader (in VertexLoader_Position.cpp etc.). Usually both of these would get data from other entries in the same multi-dimensional array, but the last few entries would be truly out of bounds. This does mean that an out of bounds function pointer can be called on platforms that don't have a JIT vertex loader, but it is limited to invalid component formats with values 5/6/7 due to the size of the bitfield the formats come from, so it seems unlikely that this could be exploited in practice.

This issue affects a few games; Def Jam: Fight for New York (https://bugs.dolphin-emu.org/issues/12719) and Fifa Street are known to be affected.

I have not done any hardware testing for this PR specifically, though I *think* I previously determined that at least a value of 5 behaves the same as float (4). That's what I implemented in any case. I did previously determine that both Def Jam: Fight for New York and Fifa Street use an invalid normal format, but don't actually have lighting enabled when that normal vector is used, so it doesn't change rendering in practice.

The color component format also has two invalid values, but VertexLoader_Color.h/.cpp do check for those invalid ones and return a default value instead of doing an out of bounds access.
2024-04-04 12:50:34 -07:00
Filoppi
1f34adf216 Video: move all padding added for frame dumping to a single function,
which also avoids the output window from being resized randomly to be a multiple of 4
2024-04-03 13:32:01 +03:00
Filoppi
d6230bbad8 Video: Change the frame dumper to actually use the raw emulation output resolution, avoiding any scaling if possible.
This should make comparisons much more reliable as pixels wouldn't be smushed together or stretched.
2024-04-03 13:32:01 +03:00
Jordan Woyak
550e008774 VideoCommon: Scan texture dumping directory + subdirectories to not re-dump existing files. 2024-03-21 20:27:34 -05:00
Jordan Woyak
ac5c2d9bf2 VideoCommon: Prefer D3D11/12 over OpenGL on Windows. 2024-03-15 19:51:04 -05:00
Admiral H. Curtiss
58616a6e4c
Merge pull request #12438 from Filoppi/custom_relative_aspect_ratio
Add Custom Relative and Raw (Squared Pixels) aspect ratios
2024-03-02 14:14:19 +01:00
Filoppi
41b19e262f Add custom relative and raw (squared pixels) aspect ratio modes 2024-02-29 21:11:19 +02:00
Lioncash
7bf77a56f4 VertexManagerBase: Initialize m_ticks_elapsed on construction
Ensures that this always has a deterministic value on construction like
everything else in the class.
2024-02-28 10:21:08 -05:00
Admiral H. Curtiss
5090a028e6
Merge pull request #12435 from Filoppi/fix-aspect-ratio-stuck
Fix aspect ratio heuristics getting stuck to a state
2024-02-20 21:36:43 +01:00
Filoppi
48fbbdba7c Video: update widescreen heuristic code to never get stuck to specific old values when changing settings 2024-02-20 22:26:19 +02:00
Filoppi
3f102ea8c2 Video: Make the game resolution (within the window) snap to the XFB size if they are within a ~1 pixel treshold on one axis only.
This takes care of making the image clearer in some edge cases where the game was already running at near perfect
 4:3 with no stretching, and the VI aspect ratio didn't match the XFB by one pixel, making the image stretched and blurry.
-Video: Fix `FindClosestIntegerResolution() using the window aspect ratio and not the draw aspect ratio, causing it to prefer
 stretching over black bars in cases when it wasn't desirable.
2024-02-20 03:09:11 +02:00
Filoppi
95ee0ac781 Video: Fix aspect ratio heuristics getting stuck to widescreen (or to non widescreen) (m_is_game_widescreen variable) if the user first forced the aspect ratio to 16:9/4:3 and then set it back to Auto. 2024-02-20 02:42:52 +02:00
iwubcode
a1147dae6e VideoCommon: move factory names to be a static inside each action class, so that they can be reused in the future for serialization 2024-02-18 15:45:10 -06:00
Admiral H. Curtiss
9b5fd5d34e
Merge pull request #12281 from TellowKrinkle/AsahiGL33
VideoCommon: Don't use indexed output for fbfetch
2024-02-18 02:33:50 +01:00
Mai
21300bb21b
Merge pull request #12457 from iwubcode/asset_memory_limit
VideoCommon: handle asset memory going over reserved limit correctly
2024-02-16 15:46:52 -05:00
iwubcode
2ab877586d VideoCommon: make mesh asset data loadable by asset loader 2024-02-12 21:45:32 -06:00
iwubcode
60772ed9d2 VideoCommon: add functionality to prepare for a mesh asset that is loaded from a GLTF file 2024-02-11 13:28:00 -06:00
TellowKrinkle
5949911a5a VideoCommon: Don't use indexed output for fbfetch
A nonzero index makes no sense, and Mesa doesn't like it when you supply an index
2024-02-07 03:52:31 +01:00
Admiral H. Curtiss
df5baab873
Merge pull request #12550 from lioncash/dead
VertexLoaderManager: Remove unused entry struct
2024-02-03 02:38:02 +01:00
Lioncash
4f40bdf501 VertexLoaderManager: Use fill() in Init()
Same behavior, less code.
2024-02-01 23:04:40 -05:00
Lioncash
ea95c82a01 VertexLoaderManager: Remove unused entry struct
This isn't used anywhere, so it can be removed.
2024-02-01 23:04:39 -05:00
Lioncash
7096f99f79 CustomPipeline: Mark arrays as constexpr
Ensures that these go into the ro section.
2024-02-01 23:02:45 -05:00
Lioncash
59211589b9 CustomPipeline: Make use of emplace_back() in GlobalConflicts()
We can use the string_view arguments to directly construct strings
inside of the global_result vector.
2024-02-01 23:02:45 -05:00
Lioncash
353ceedb50 CustomPipeline: Resolve unused variable warning
We can just use holds_alternative here instead.
2024-02-01 23:02:45 -05:00
Admiral H. Curtiss
da6b5dd38a
Merge pull request #12546 from lioncash/event
VideoCommon/Statistics: Remove global system accessor from s_after_frame_event
2024-01-31 21:16:21 +01:00
Admiral H. Curtiss
18abf7c768
Merge pull request #12544 from lioncash/getmod
GraphicsModGroup: Allow heterogenous lookup for GetMod()
2024-01-31 20:17:26 +01:00
Lioncash
cac66317aa VideoCommon/Statistics: Remove global system accessor from s_after_frame_event
Instead, we make the event take a reference to the system and then pass
it in when the event is triggered.

This does introduce two other accessors, but these are much easier to
refactor out over time, and without modification to the existing event
interface.
2024-01-31 13:12:09 -05:00
Lioncash
0dfefacdf4 VertexLoaderBase: Collapse std namespace for hash and mark noexcept
Makes the hash specialization a little less noisy. Also we mark it
noexcept, since hashes shouldn't be throwing exceptions (and this can be
optimized on).
2024-01-31 12:43:00 -05:00
Lioncash
8e4b2565cd TextureConfig: Collapse std namespace for hash
Lets us collapse the namespacing and make the specialization a little
less noisy.
2024-01-31 12:41:33 -05:00
Lioncash
b63dcd504d RenderState: Collapse std namespace for hash
We can specify the namespace on the hash to make the specialization a
little less noisy.
2024-01-31 12:40:10 -05:00
Lioncash
5bfaa3a966 NativeVertexFormat: Collapse std namespace and mark hash noexcept
We can just tag the std:: onto the end of the specialization to make it
less noisy.

Also mark it as noexcept, since hashes shouldn't throw exceptions.
2024-01-31 12:37:44 -05:00
Lioncash
40b050fe37 GraphicsModGroup: std::move graphics_mod in Load()
The config object is quite heavyweight, so we should move this instead
of copying.
2024-01-31 12:27:43 -05:00
Lioncash
ccacda5e2c GraphicsModGroup: Simplify try_add_mod()
We can use contains() here, and also move the mod config if it's valid
instead of copying it, since it contains quite a bit of allocated data.
2024-01-31 12:23:21 -05:00
Lioncash
a1879ea099 GraphicsModGroup: Allow heterogenous lookup for GetMod()
Allows using keys that aren't directly std::string as the key. This lets
us use std::string_view for the incoming path name, making it more
flexible with other string types.
2024-01-31 12:05:17 -05:00
Admiral H. Curtiss
9a3e770c23
Migrate SConfig::bWii to System. 2024-01-31 12:54:07 +01:00
Admiral H. Curtiss
8482a50154
Merge pull request #12530 from iwubcode/custom_shaders_compilation_fixes
VideoCommon: fix some issues when compiling custom shaders
2024-01-28 14:30:37 +01:00