The 'DirectX_Init()' call was performed late in code, shortly after the window has been created (at this point all device objects and window handles are valid), but the 'DirectX_Shutdown()' call was performed on DLL_DETACH, which was way too late, as the objects were already destroyed at this point. This wasn't an issue before, as we created our own objects in the old DX code. But due to optimizations, we were using the same pointers as the game (noticeable performance boost), but did not adjust the shutdown to accommodate the changes. The shutdown is now performed while the device objects and window handles are valid. Code has been tested on Nvidia and AMD systems, and has confirmed to fix the aforementioned issues.
Huge cleanup for dedicated. All patches in 'opcodes.cpp' are now directly applied to the executable, and kept as reference in the source file. The patch logic is commented. Any other patches for the dedicated server executable should be documented in the patch file found in 'r5dev/resource/patch'.
This change was planned for a long time. This moves all REGISTER calls to a single translation unit, this is required as we currently added a very dirty workaround for not registering duplicates by checking if VFTable pointer was already present in the vector... Registering from single translation unit prevents duplicate instances that gets created if header is included by more cpp files.
Reworking this reduced 100kb+ of compiled code. This commit also reworked the way functions/variables/constant gets logged with their addresses; the new code formats them on the fly, and allows for resize at any time. Formatting is no longer required by programmer.
TODO: currently there are some compile errors for dedicated and client dll's. These will be resolved very soon as they need to be properly worked out still (server & client only stuff needs to be properly split). Use the 'main' (stable) branch for the time being if you need to compile these dll's.
In Respawn's engine, the '-forceborder' command line parameter does the same thing as '-noborder'. '-forceborder' should remove the flags (nand) while '-noborder' should append the borderless flag. This code is an attempt to mitigate the problem. There was not enough bytes to assemble the nand operation for this code path without involving code caves or shifting.
server loads "vpk/server_mp_lobby.bsp" instead of "vpk/client_mp_common.bsp" and "vpk/client_frontend.bsp". Also removed the old hack we used to patch string constants to format VPK string paths as we have a proper implementation now using CEngineAPI::SetStartupInfo.
* Check if server is active in command callback before attempting to hot swap.
* Hook 'v_Detour_LevelInit', and log NavMeshes that failed to load.
* Split free/destroy logic into separate function.
* Created constants for NavMesh and AI Network paths/extensions.
* Added performance profiler for hot swap logic in command callback.
* Renamed "navmesh_reload" to "navmesh_hotswap".
* "navmesh_hotswap" is now development only.
* Fix all compiler error for GameSDK on S0 and S1.
* Remove some unused patterns and globals (launch performance gains of 100ms).
* Remove most duplicate patterns.
* Relocate globals from engine to a more appropriate location.
* Renamed some unknown vars to a more appropriate name (most of these where obtained a while back, but as we reversed more parts of the engine, some of these vars become known).
* Renamed some vars to fit the naming convention used in the SDK.
* Fixed pattern debug logging using the wrong pointer, thus displaying the wrong address.
Fixed rare crash in CMDLCache when buffer itteration returns invalid pointer (not null!) for handle.
Made all variables and functions used by BuildPropStaticFrustumCullMap searched dynamically (these should also work on s0, s1 and s2, they are compared but at the moment untested).
TODO: Fix deref crash properly (see bsplib.cpp l291).
Moved every pattern to IDetour interface. This allows for debugging patterns scans more easily, and create threads during pattern searching (Operation is now fired in APIENTRY).
Also cleaned up some unused code/extraneous comments.
Slightly increased performance by purging duplicate patterns.
Made variable search less dependent from other results (except if pattern-to-scan results is within the same header)
* Renamed 'r5apexsdkd64.dll' to 'gamesdk.dll'.
* Added required dedicated parameters to code instead.
* Bug fixes around CCommandLine class (fixed misaligned VTable indexes).
* SDK now supports being directly launched by the game executable.
The SDK launcher will pass '-launcher' to the game, which indicated its being launched by the launcher. If the game does not receive '-launcher', it assumes its being launched directly from the game executable, which will instead load 'startup_(dedi_)default.cfg'.
The sdk dll's are now added to the game's IAT by their dummy exports allowing for them to be loaded when the exe is loaded (the dll's do everything on init).
* Dedicated SDK now supports ASLR enabled executables.
* Complete removal of CEngineClient on the dedicated server.
* Complete removal of gHLClient and g_pHLClient on the dedicated server.
* Performance improvements for the dedicated server frame loop.
* ConVar 'CVCallback_t' structure rebuild.
* Reduced overhead throughout SDK.
Migrating to this to initialize all patterns and prototypes in Systems_Init() instead.
This should make debugging missing/not found patterns easier and allow for opting out variable/constant search (some of these require other patterns to be found, thus resulting in seg faults..).
Also added check to detect if user has a eligible CPU to run this SDK.
The game requires SSE and SSE2 instruction sets. Our SDK requires this too due to the use of SSE intrinsics, so we cannot let the game handle this. We have to check it ourselves.
* Move most definitions to implementation file to avoid recompiling whole program for small changes
* Pass strings by reference for where possible.
* Split Module class to dedicated file.
* Add const qualifiers to all eligible methods for address/module class
* Some renaming
Inline every signature defined in the SDK (previously each translation unit had its own copy of the signature and function prototype). DLL init is near instant now (85% speed improvements).
Game will check scripts/levels/mapname.json and load all rpaks in the rpak field.
Game now also unloads pak files loaded by the SDK preventing crashes and unnecessary memory usage.
VPK's and STBSP's are planned to be added as well.
* Partially rebuild CGlobalVarsBase and CGlobalVars struct.
* Partially implement BuildAINFile (disk builder) written by BobTheBob9. All that needs to be found are the addresses/offsets for 'pUnkStruct0Count' and 'pUnkLinkStruct1Count' (see ai_node.h). I didn't have time for it yet.
* Patch call which restarts the engine once AIN file has been build (so it doesn't get cleared from memory).
* Temporarily patch write operation in 'Detours_LevelInit()' which writes out of scope since the navmesh structure is misaligned somewhere still with a few bytes.
NOTE: 'CAI_NetworkBuilder::BuildFile()' is for now commented out until the 2 other structs are found in the game dll. Once found then we could start fixing the struct offsets and hope it works as-is.
* Add Cbuf_AddText and Cbuf_Execute signatures.
* Remove additional VGui stuff from dedicated statemachine.
* Move signatures to their respective headers.
* De-inline 'CHostState::FrameUpdate()'
* Ported al new patters to GAMEDLL_S1, CEngine still recreation needs extra work
* Load server VPK's instead of client VPK's on standalone servers
* Dynamically search for 'g_bDedicated' global boolean instead of hardcoded address
* Add 'PatchString' to memory class
* IConsole overlay now clears the oldest log entries to stay under the vector limit instead of clearing the whole vector.
* IConsole code cleanup.
* IBrowser code cleanup, reordered class methods.
* Use ConVar utilities to get/set values for strings in IBrowser and IConsole.
* New ConVar's for RUI console overlay to tweak colors.
* Since the material mode is set to '2', all of the render loops and shader loading functions are no longer called. These have all been removed.
* Added description for patches that missed them
* Traced root caller/cause for certain patches to skip even more unnecessary code and save more memory.
The dedicated server still has directx dll imports which will cause issues with ReShade or 3DMigoto, or trying to load it on a headless machine with no directx installed. The imports have to be cleared to avoid issues. The most convenient approach is to do this with a separate exe rather then patching it in runtime.
The instruction at 'CalcPropStaticFrustumCull' [0x14028F3B0 + 0x5C7] moves RSP + 0x70 into the R13 register.
RSP + 0x70 seems to contain a pointer to collission data for that particular prop model.
When running NoShaderApi() and passing the dedicated server the '-noshaderapi' command line parameter, RSP + 0x70 will be a nullptr.
This has to be fixed to have prop static collissions on the server.
