Cluster links were written with an additional padding as the struct is 11 in size, so it gets padded to 12. Cluster links now get correctly written to the disk.
Fixed writing of:
- Path nodes
- Node links
- Path clusters
- Cluster links
The previous globals 'g_pppAiNodeClusters' and 'g_pppAiNodeClusterLinks' were of type CUtlVector. These have been retyped and renamed accordingly.
Also mapped the CAI_Cluster class out slightly more.
Rebuild of path clusters and cluster links are still not correct; parsing a file in from r2, results in a different file on the disk, at the very offset the clusters are being written in (one field is off, this will be investigated soon).
Client's clock could be behind or ahead of the server's, take this into account. Note that this code has been tested over night and only had a few hits when new baseline snapshots were applied on the server, or when there was a clock drift. It still requires an actual playtest with several players on low and high latency to determine if this is even needed.
- Check and clamp if commandTime is above serverTime (too far in future).
- Check commandTime delta before checking if above serverTime or below lastCommandTime.
- Make sure cmdTime is never lower than last.
- Make sure delta doesn't exceed maxUnlag.
- Make sure latency gets clamped to maxUnlag (clamp appears to be missing in the engine).
- Only check for cmdTime if we are not discarding the cmd; previously, already processed queued cmd's got checked before being discarded.
Temporary lag compensation exploit fix by clamping the difference. Exploiter could carefully control the detonation time of a grenade, or alter the ability cool downs by altering the 'CUserCmd::command_time' field before sending it up.
The implementation in the engine always took/returned signed 64bit size types, but I made a mistake when reversing the virtual function table. All types have been changed to what they should be (mostly signed 64bit, very few unsigned), and usage in-SDK has been adjusted accordingly.
Work in progress; does not compile!
Moved script registration function to static gamedll libs instead, and used a pointer callback approach for calling them to avoid duplicate symbols during linkage.
Find regex pattern:
inline auto ([a-zA-Z0-9_]+) = ([a-zA-Z0-9_]+)\.RCast<([a-zA-Z0-9_:<>*]+) *\(\*\)\(([^)]*)\)>\(\);
Replace regex pattern:
inline $3(*$1)($4);
This commit also removes the unnecessary initialization (which was required to type the auto variables),
and therefore removed 6kb of unnecessary dynamic initialization code.
Member offset for 'm_bGamePaused' did not align with the executable, moving the member 'm_hPlayerViewEntity' before the first bool in this row results in correct code generation (also aligns properly in the disassembly of the game's code).
Rebuild of the 'CServerGameClients::ProcessUserCmds' and 'CPlayer::ProcessUserCmds' methods. The rebuild is pretty much identical to the original implementation, but with additional checks for the number of commands, and total commands in 'CServerGameClients::ProcessUserCmds'. In the future, additional clamps and checks between CPlayer and the recv'd UserCMD should be incorporated in these functions.
Global 'direct' usage of 'MemAllocSingleton()' has been jettisoned. Where possible, smart pointers were used instead. During the refactor, the following bugs were addressed and fixed:
- The virtual destructor of 'CCVarIteratorInternal' was NOT called on destruction.
- Class function 'KeyValues::MakeCopy' did NOT calculate the buffer size of the wide string correctly, the original calculation was 'len+1*sizeof(wchar_t)', but should've been '(len+1)*sizeof(wchar_t)'.
Some other code changes include:
- Tier0 include 'memstd.h' has been moved above all thirdparty includes, to make sure the memalloc functions get shadowed with ours in third party libraries as well.
- RPak file paths string literals are now defines.
- 'DestroyOverlay' has been refactored to match the assembly of the game.
Reason parameters are optional. For scripts, to use the default message, pass in an empty string. The function for the script system should be centralized soon to avoid more duplicate code.
Previously, it was all controlled from the global init (applied to all projects), but some projects need different options. With these changes, you can disable the common options applied in the 'add_module' macro, and set your own if desired.
Fix several exploitable bugs in the CUserCmd class. Some of these have been used to exploit/cheat in-game. Fixes contain:
- Camera position clamping (the only patch that hasn't been tested yet!)
- Weapon activity exploit, allowing player to infinitely throw ordnances, and perform other 'cheats'.
- Akimbo exploit + server crasher, allowing client to set multiple inventory weapons as active. The active weapon index bounds were also not checked, a properly crafter CUserCmd message would therefore be able to crash the server.
Note that this does not fix all issues related to the UserCmd class; further reversing and testing revealed there is more to be fixed, these fixes will get implemented with a future commit.
Check if the actual message type is 'TextMsg' before printing, since other data is binary and could be read as 'HUD_PRINTCONSOLE', 'HUD_PRINTCENTER', etc.. resulting in printing binary data. This fix permanently solves that problem. The check has also been applied to the 'ShouldReplayMessage' function.