Hot-patching an already-live device's vtable methods via SetDevice races
the emulator's self-modifying-code handling (Windows-on-ARM's xtajit, or
Wine on FEX-Emu/box64). Detect that case with IsWow64Process2 and refuse
the attach instead, so callers fall back to loading gMod as d3d9.dll
before the game creates its device.
* remove more unnecessary code
* todo?
* 1.5.5.0
* change GetHash from in_out parameter to return the hash
* remove ForceReload bool for textures
* remove textureserver, place logic in textureclient, not yet finished
next up change FileToMod and shit to use modded_textures map
* fix debug compilation errors
* fix typo
* should_update boolean, replace the entire MergeUpdate logic later
* tidy up d3d9_dll.cpp and bits (#12)
* define `ASSERT`
Added minhook lib to tidy up hooks
Remove cruft from dx9_dll.cpp
* Bug fix
* Another typo
---------
Co-authored-by: Jon <>
* gitignore
* Proxy functions, cache dx9 device type
* simplify LoadModsFromFile
* move minhook dependency to fetch content
* Make it work
* Tweaked debug for file reads
* Fixed inverse check
* unordered_map
* set should_update (yikes)
* simplify libzippp.cmake
* remove unnecessary code
* rename project files (keep uMod_ prefix only for d3d9 classes)
* remove unused Nothing() function
* remove unnecessary fields
* remove some code duplication
refactor methods together
* refactor SwitchTextures and UnswitchTextures together
* move loaded_size into function
* enable libzippp encryption again
* copy data from opened zipentry and then delete the entry
* fix hash bug in volume textures
* const
* change uMod_TextureClient to TextureClient
* revert libzippp.cmake
* 1.5.6
GWToolbox Bot