Save 2 bytes from segment padding by performing zero test at original address followed by the long jump to patch, there were 5 bytes, zero test only requires 2, thus only 3 bytes were nopped with these changes.
This commit fixes an engine bug where netmessages are getting copied into the replay buffer, while these messages should never be replayed. The engine performs an internal check on 'CNetMessage::m_nGroup', and if its NOT 2, the message is getting copied into the replay buffer. All messages returning false in 'ShouldReplayMessage' are not getting copied into the replay buffer anymore. This exploit has been used in the past to route clients that were watching a replay to an arbitrary server, which essentially forms an info leak as the client attempts to connect to the arbitrary server on its own. The exploit also allows for some form of remote code execution, depending on if the client was launched in developer mode or not.
* Add method for finding a substring ('CUtlString::Find').
* Add method for appending a slash ('CUtlString::AppendSlash').
* Add 'CUtlString' overload for 'CUtlString::IsEqual_CaseSensitive'.
* Add 'CUtlString' overload for 'CUtlString::IsEqual_CaseInsensitive'.
* Add optional parameter to strip trailing slashes in 'CUtlString::DirName'.
* Add optional parameter to strip trailing slashes in 'CUtlString::StripFilename'.
If the chunk if uncompressed, continue the loop instead of breaking out of it. Only 2 files were affected by this bug from original VPK's: 'notosansjp-regular.vfont' and 'notosanstc-regular.vfont'.
Only certain lumps can be cached; see switch case in 'IsLumpIdxCachable'. Attempting to load anything else from cache results in a code crash or invalid data.
This commit adds support for loading cubemap lumps from the disk, several patches and hooks had to be implemented to make this work. Since the game was trying to read the pakfile lump from the BSP file (even when absent), it would read EOF (heavily reduces down map loading performance).
This commit allows for loading all lumps from disk, if the m_bExternal var is set, it also attempts to load the lumps listed in the function 'IsLumpTypeExternal' from the cache, but through a different code path, which causes a crash. Setting this to false allows for loading it in the ctor.
bsp_lump loading was missing a check for if the lump is supposed to be able to load from disk
also removes VERTEX lumps from loading from disk due to crashes that occur when they are kept
The 'CMapLoadHelper' constructor has been rebuild to load lump files too (if they exist). Currently, the game only reads the packed BSP file, or lumps if they exist in the filesystem cache. The newly added logic also reads them from a file. This is currently broken, most likely due to alignment problems. Experimental.
Added option: /Zf (Faster PDB generation); this was required as we enabled /MP. Without this, the compiler regularly fails to create a PDB.
Added option: /EHsc (Standard C++ exception handling).
This should fix a rare crash that happens when the playlists file is reloaded during disconnect. The client fetches the playlists from the server on connect. On disconnect, we want to reload the one from the client so that if the player want's to create his own game, that he would have his own playlists. The fix is to delay the reload of the playlists file to the beginning of the next frame.
Properly fix the aligned memalloc singleton in the SDK; the implementation now uses a callback based approach for calling the allocator and deallocator.
Use 'std::lock_guard' for mutex management. Also fixed a compiler warning that got triggered by having a mutex unlock placed at an unreachable code path in 'CGaussianRandomStream::RandomFloat'.
Set the persistence fields to 'ready' in 'CClient::ActivatePlayer', before executing the rest of the function. Previously, it was set in 'CVEngineServer::PersistenceAvailable', but this is too late. The function 'FairFight_Init' was actually 'CClient::ActivatePlayer', and thus it has been moved to the correct file, and the old file defining it previously has been removed.
Added all public headers to CMake projects, also moved some files around in the public directory. Translation units have been moved to the libraries that were responsible for implementing them, as this game is monolithic.
On some compile runs, the 'LoadTextureBuffer' call is getting optimized out as 'ret' is only used for the assert statement, but on release, the assert statements does nothing and thus the compiler got rid of the call. This code should not be fired in this function, but rather in the initialization routine anyways. The code has been moved to 'CBrowser::Init', and the issue has been fixed.
2 warnings:
Operation 'lkup[arr[i] & 0xF] | 0x8000;' returns an int, but should return a unsigned short; added explicit cast.
Operation 'arr[i] & 0xF' could read up to 64 bytes, while the stack buffer is only 16 bytes. Switched to operation: ''arr[i] & 0x3'.
The static buffer has also been declare const now.
