GL_CLAMP (clamp to border) was changed to GL_CLAMP_TO_EDGE in 2008
(f2baf359). In 2018 (ce1d5406) I made OpenGL 1.2 be required since
GL_CLAMP_TO_EDGE is used.
Restore support for GL_CLAMP in order to support OpenGL 1.1 like vanilla
Quake 3 does. This should allow using the default Microsoft Windows
GDI Generic OpenGL 1.1 driver (untested but it won't fail the version
check at least).
From gpuinfo.org, it looks like drivers stopped advertising support for
GL_SGIS_texture_edge_clamp so use a version check in addition to the
extension check.
r_allowExtensions 0 disables using GL_CLAMP_TO_EDGE in the opengl1
renderer. GL_CLAMP support wasn't added to the opengl2 renderer.
For lerped frames (refEntity_t frame not equal oldframe) IQM joint
matrices may have incorrect axis scale. This can cause significant model
distortion. The matrix lerp is linear causing each vector to move in a
straight line between frames instead of arcing like a circle. Each joint
frame can have a different scale so can't just normalize the joint
matrix.
Store joints as quaternions and spherical lerp between them and then
convert to a matrix. For my test model, setting up the skeleton is four
times slower now but it still seems to be fast enough to be usable.
r_cubeMapping requires textureCubeLod() which is only in OpenGL 3.0
(GLSL 1.30) and later. It's not in OpenGL ES 3.0 / GLSL ES 3.00.
This needs to be checked before R_InitImages() so can't just check in
GLSL_InitGPUShaders().
Override the video mode list in the Team Arena data files with detected
modes from SDL like in ioquake3's Q3 UI. Add the aspect ratio to the
end of the video resolution (i.e., "640x480 (4:3)"). Add the current
(custom) video mode to the list.
Before when using a custom resolution in the menu you could not change
the video mode using the mouse because the resolution text was blank.
Now custom video resolution is displayed and can be clicked.
Team Arena HUD's team member info box can be cycled through using
nextTeamMember and prevTeamMember commands. nextTeamMember command loops
around when end of list is reached but prevTeamMember command got stuck
at last entry in list (team overview).
My commit last month "Fix SDL audio playback with surround sound" broke
16-bit stereo sound. S_TransferStereo16() still assumed that dma.samples
was a power of two. I also cleaned up code related to the previously
mentioned commit.
If user has surround sound enabled, ioq3 would not play any sound.
Fix painting sound buffer for 4/5.1 audio channels. Extra channels
currently play no audio.
BotSetEntityNumForGoal() was checking all entities that are not
team_redobelisk (which is the obelisk visual entity) to find the
untitled obelisk collision entity. This may fail in rare cases where
there is an another entity within 10 units of the obelisk origin.
Failing to find the correct entity may cause bots to not attack the
obelisk.
Instead add BotSetEntityNumForGoalWithActivator() for looking for the
obelisk collision entity by activator classname (team_redobelisk) which
should be less likely to find the wrong entity.
This doesn't affect official Team Arena maps (unknown if it affects any
others).
Reversed strcmp check was reported by Thomas Köppe.
Bounds are optional for animated IQM models but are not possible to
include with unanimated models (seems intended for use with separate
model containing animations and bounds). Calculating bounds for
unanimated IQM models fixes culling and head model on HUD which
calculates position from model bounds.
The axis returned for IQM tag was the animation's joint rotation without
the base frame joint rotation. It only worked correct for models that
did not rotate the base frame joints.
Using GPU vertex skinning is significantly faster than CPU vertex
skinning. Especially since OpenGL2 has to run R_VaoPackNormal() and
R_VaoPackTangent() each vertex each frame which causes CPU vertex
skinning to be significantly slower than OpenGL1 renderer.
Only calculate vertex blend matrix for each unique bone indexes/weights
combination once per-surface instead of recalculating for each vertex.
For best performance the model surfaces needs to use few vertex bone
indexes and weights combinations.
Unroll loops so GCC better optimizes them.
In my tests drawing animated IQM may take 50% as long in opengl1 and
70% as long in opengl2. It will vary by model though and might not
help much at all.
Made unanimated IQM models skip matrix math altogether.
- Only allocate memory for vertex arrays that are present in the IQM
file and are actually used (may not have colors or blend index/weights,
don't load tangents in opengl1). (Colors is fixed to next commit.)
- Explicitly handle loading IQM files without meshes (bones only).
- Better IQM validation. Header data offset 0 mean data is not present
in file. Check if required vertex arrays are present.
This involved a lot of white space changes and moving code around.
Fix two constants in GLSL shaders. Remove f suffix from float and fix
int to float assignment. They were causing shader compile errors in
OpenGL ES 2 context.
Remove disabling clip plane. Clip plane is unused and never enabled in
the opengl2 renderer. Remove disabling it to avoid causing a GL error
when using OpenGL 3.2 core profile or OpenGL ES.
Make VAO cache vertex stride be size of srfVert_t since that is what
is uploaded to the GPU. No behavior change. There is a disabled debug
id in srfVert_t though which if enabled changes srfVert_t size.
OpenGL ES is only required to support unsigned short for element buffer
values.
R_DrawElements() firstIndex argument was glIndex_t which caused element
indexes to wrap around to 0 when glIndex_t is an unsigned short.
(glIndex_t is an index into the vertexes buffer, not element buffer.)
Change it to 'int' like tess.firstIndex which is passed to
R_DrawElements().
World VAO cache buffer size allowed storing more vertexes than unsigned
short glIndex_t could reference. This resulted in the vertex indexes in
the element buffer wrapping around to 0.
Load functions procs supported by OpenGL ES 2.0, though there is not a
compatible renderer yet. Change argument for GLimp_Init from coreContext
to fixedFunction.
Also declare the GL functions in tr_local.h so there is compile error
for non-core GL functions instead of SEGFAULT from dereferencing a NULL
pointer.
Disable the non-functional stencil shadow code that hasn't been updated
to use OpenGL 3.2 core compatible drawing.
If renderer is compiled into client (USE_RENDERER_DLOPEN=0) and after
start up set r_allowExtension to 0 and run vid_restart, some extension
were still used.
Connecting to a server running a different fs_game and using a
autoexec.cfg containing in_restart would hit a fatal error in IN_Init().
IN_Init called before SDL_Init( SDL_INIT_VIDEO )
Reported by smokey2k on the ioquake3 forum.
In 2013 ioquake3 stopped referencing the pk3 file that qagame.qvm was
loaded from. This had the unintended side affect of causing
non-dedicated pure servers to no longer reference a pk3 that only
contains the three QVM files.
Non-dedicated pure servers did not reference the pk3 containing the
latest cgame.qvm so if client did not have the pk3 file they were kicked
as unpure instead of the client trying to download the pk3 file.
Also make server touch ui.qvm since it's required to pass pure check and
may be separate from cgame.qvm.
Moved all the code using Altivec intrinsics to separate files. This
means we can optionally use GCC's -maltivec on just these files, which
are chosen at runtime if the CPU supports Altivec, and compile the rest
without it, making a single binary that has Altivec optimizations but
can still work on G3.
Unlike SSE and similar extensions on x86, there does not seem to be
a way to enable conditional, targeted use of Altivec based on runtime
detection (which is what ioquake3 wants to do) without also giving the
compiler permission to use Altivec in code generation; so to not crash
on CPUs that do not implement Altivec, we'll have to turn it off
altogether, except in translation units that are only entered when
runtime Altivec detection is successful.
This has been tested on Linux PPC (on an Altivec-enabled CPU),
but we may need further work after testing trickles out to other
PowerPC devices and ancient Mac OS X builds.
I did a little work on this patch, but the majority of the effort belongs
to Simon McVittie (thanks!).
When you start recording using SDL pulseaudio driver the client sends
all audio captured while not recording. 240 milliseconds of audio is
sent each frame until the capture buffer is empty. This is a problem for
privacy and causes confusing to debug VoIP playback issues on other
clients connected to server and when playing back demos.
Fix specularScale <metallic> <smoothness> with r_pbr 1 which has been
broken since r_pbr was implemented in 2016.
Fix specularScale <r> <g> <b> <gloss> setting b to gloss and leaving
gloss as 0 since it was implemented in 2014.
The key handler allowed going 2 beyond the end of the bot list and the
display function clamped to 0 causing the first bot to be shown 3 times.
Attempting to add the bot in gametypes < GT_TEAM would fallback to
Sarge in UI_GetBotNameByNumber() (who isn't the first bot) and gametypes
>= GT_TEAM would access characterList past known values (typically NULL
but if teaminfo.txt contained 63 characters it would access out of
bounds memory).
Switching to dedicated camera follower with no possible players to
follow would spawn at the intermission point and display "connection
interrupted" HUD message. Pmove() was not run for the client so
ps.commandTime was too far behind. I made it so that dedicated camera
followers and scoreboard run Pmove() but cannot move (PM_FREEZE).
When all players possible to follow leave, the dedicated camera follower
would continue to display the old player state of the player they were
following (along with "connection interrupted" HUD message). Unlike the
regular case of a spectator following a specific player, dedicated
camera followers did not reset their player state to the intermission
point after the followed player was no longer valid.
Now a client can be set as 'team follow1' to automatically switch
between displaying the intermission point and following a player when
possible.
SDL 2.0.5 dropped support for macOS 10.5 so target 10.6 instead. The
PPC build uses SDL 2.0.1 so it still targets 10.5. macOS 10.5 (x86,
x86_64) should automatically run the PPC build using Rosetta.
Revert MAN-AT-ARMS' change to SDL 2.0.8 SDL_platform.h that allowed
targeting macOS 10.5 for the sake of PPC. It also incorrectly allowed
x86 and x86_64 to target 10.5 as well. (Also macOS PPC uses separate
headers now.)
The version check is required for supporting macOS PPC with SDL 2.0.1
and Travis-CI (Ubuntu Trusty) with SDL 2.0.2.
The client now requires SDL 2.0.5 runtime if compiled against SDL 2.0.5
or newer.
code/libs/macosx/libSDL2-2.0.0.dylib has 2.0.8 for x86 and x86_64 and
2.0.1 for PPC. Add 2.0.1 headers for PPC with modifed SDL_platform.h to
allow compiling using macOS 10.5 SDK. Using separate headers allows the
engine to check the SDL version for enabling newer SDL features.
This is a little bit of future-proofing, but also gives us a little more
flexibility in general; now we can add in the cvars to open a specific
device, etc, that the OpenAL codepath does.
In SDL2, the initialized subsystems are referenced counted, so it's safe to
initialize them twice, and it makes the SDL_QuitSubSystem during our shutdown
correctly decrement the count. Before (as a probably-harmless bug), it would
not increment the refcount if the subsystem was already initialized, causing
problems when it decremented it later.
In September 2017 I moved loading arenas.txt/*.arena files from entering
start server menu to at startup to fix running out of memory in Team Arena
UI after opening the start server menu several times.
However, Team Arena completely replaces the uiInfo.mapList array when
switching between single player and start server menus. So after my
change, entering single player and then entering start server would only
display single player maps. It caused SP endofgame menu to use MP map
list for replay/next map since arenas were loaded after gameinfo.txt.
Continue loading arena info at start up to avoid reallocating arena info
but move setting up uiInfo.mapList to when entering the start server
menu.
I changed Color Depth options 'Default' to reset r_stencilbits instead
of 0 and '32 bit' to use r_stencilbits 8 instead of not changing the
value. (This matches my q3_ui changes in the previous commit.)
Set r_stencilbits when changing graphics presets like when changing
Color Depth.
In 2007 in ioquake3 unified-sdl branch (revision 1144) setting
r_colorbits in q3_ui was removed but the Color Depth menu option was
still kept. Setting r_colorbits was not removed from the Team Arena UI.
In 2011 I removed the Color Depth menu option from q3_ui as it did not
change any cvars. Yesterday I restored the option not realizing this
and thinking that requesting 16-bit color depth worked.
Add setting r_colorbits back to q3_ui so Color Depth menu option works
again. I changed Color Depth options 'Default' to reset r_stencilbits
instead of 0 and '32 bit' to use r_stencilbits 8 instead of not changing
the value.
However I discovered r_colorbits 16 does not actually work on my system
(Debian Jessie x86_64 nvidia). ioquake3 was reporting the requested
value instead of the actual obtained value. Fixed in my previous commit.
The values passed to SDL are just the minimum required. Get actually
values of colorbits, depthbits, and stencilbits from SDL instead of
assuming that the engine got exactly what it asked for.
The r_colorbits cvar still exists and gfxinfo reports it works on
Debian Jessie in both git master and building the commit where I
removed r_colorbits from q3_ui. So it does indeed control something.
Maybe I expected 16-bit color to look different? I don't know.
This reverts my commit 8e689739f4
from August 11 2011.
Loading a 1024-byte q3history file will fill the whole consoleSaveBuffer
leaving no space for a string terminator. Com_Parse will read at least
one byte beyond the end of consoleSaveBuffer. The written console
history file can only be 1023 bytes (enforced by Q_strcat) so don't
allow loading size of 1024.
If switching to a mod with a shorter q3history file, the data in
consoleSaveBuffer that isn't overwritten will be parsed. So always
add a string terminator.
String not terminated reported by David "devnexen" CARLIER.
cls.glconfig.isFullscreen was not updated when changing r_fullscreen
without a vid_restart. Starting in fullscreen and switching to
windowed mode would not release the mouse.
Mods calling trap_GetGlconfig() after a fullscreen toggle now get
the correct value for isFullscreen. (Note: Mods already got the
correct value at start up and after vid_restart.)
Reported by Mickaël "mickael9" Thomas.
Allows higher download throughput (from ~2 MB/s to ~60 MB/s at 120FPS)
This has no effect for curl versions older than 7.53.0 (for which the
buffer can only be shrunk below 16k)
Fix the output/behaviour of CVAR_CHEAT flagged cvars in case they are
also of type CVAR_LATCH (avoid the early latch case return to make it
work as intended).
This is exact root of q3msgboom bug http://aluigi.altervista.org/adv/q3msgboom-adv.txt
Unfortunately, server still need this ugly '1022 char limit' hack to support unfixed clients in some degree.
And as it affects MSG_ReadBigString() - unfixed clients can still be crashed by 8191-chars long configstrings that comes with gamestate
If client sends wrong serverId but is already active in the world
(CS_ACTIVE) don't resend initial gamestate for the map. This isn't a
valid situation. The player should be CS_CONNECTED or CS_PRIMED.
Resending gamestate to an active player will cause them to respawn
without dying or disconnecting. If the player had a CTF flag it gets
lost until the map is changed or restarted.
Reported by Ensiform at:
https://bugzilla.icculus.org/show_bug.cgi?id=6324
In May 2017 the library loading was limited to *.dylib but the macOS
system OpenAL framework does not have dylib extension. So allow loading
files from /System/Library/Frameworks/ as libraries even without dylib
extension.
This is checked in Sys_DllExtension() so that QVM filesystem access will
not allow writing files to /System/Library/Frameworks/ even if homepath
is changed to include it. (Admittedly it doesn't fit the function name
but fits the function description and current usage.)
Allow listing about 273 .arena filenames for loading in q3_ui instead
of only about 136 that fit in a 2048 byte buffer (average 15 bytes per
file name).
The buffer for filename list runs out of space long before the buffer
for arena file content does. There is no warning for file list out of
space but there is a warning for arena file content.
This was requested by a user with many maps.
Team Arena's mpteam6 map has a shader textures/base_wall2/space_concrete
that contains an opaque stage, two non-lightmap blendfunc filter stages,
a blendfunc add stage, and a lightmap stage. The lightmap was attached
to all four of the non-lightmap stages causing the filter stages to
darken the lightmap multiple times.
Change setting up the lightall GLSL shader to only use lightmap if it's
the first stage or not a blendfunc filter stage. Now only the opaque
and blendfunc add stages of the mpteam6 shader use the lightmap.
Reported by Alexander Nadeau (wareya).
Newer versions of BSPC such as 2.1h included with the Quake 3 GPL source
code create AAS files containing areas in cluster 0 if the area has no
reachabilities.
The AAS files included with Quake 3 and Team Arena do not contain areas
in cluster 0. It's apparent that BSPC would not create them. Instead it
created clusters with no reachability areas.
It seems the intention was to check if the areanum and goalareanum have
reachable areas using AAS_AreaReachability(areanum) everywhere before
calling AAS_AreaRouteToGoalArea(). This prevents adding cluster 0 to
the routing cache and portal cache. However, it is not checked
everywhere and including some places in the Game VM.
Fix AAS_AreaRouteToGoalArea() instead of trying to wack-a-mole with all
the places that call it.
Cluster 0 access reported by Thomas Köppe (github @tkoeppe) as causing
crashes in rare cases.
A negative timelimit value or a value that would overflow the
multiplication by 60000 caused an endless map change/reload.
Based on patch and description by @vloup.
By default mingw-w64 uses Microsoft's broken _vsnprintf() in msvcrt.dll.
It can be overriden by defining __USE_MINGW_ANSI_STDIO but let's just
use the same behavior for both MSVC and mingw-w64.
Reported by @birdstakes.
Windows API ignores all trailing spaces and periods which can get around
Quake 3 file system restrictions. QVM opening 'uix86.dll.' actually
opens 'uix86.dll' which allows QVM to write native code.
This is done in the low-level Sys_FOpen() instead of the function
directly used by VMs ( FS_FOpenFileByMode() ) in case there are engine
commands now or in the future that can read or write arbitrary files.
Reported by Noah Metzger (Chomenor).
My cvar latch system changes prevent the Game VM from changing
g_gametype when the value is out of range due to it being registed in
the engine. It's been pointed out as fragile method of security, which
was still exploitable, by Noah Metzger (Chomenor). It doesn't seem like
this is working out to be a good solution.
The issue of fs_game '..' on server being relicated on client via
systeminfo exploit is still fixed as it's not affected by latch.
There are a few cases from current values of fs_game are used which
ideally should use fs_gamedir char array which has been validated.
Revert "Don't let VMs change engine latch cvars immediately"
Partially revert "Fix fs_game '..' reading outside of home and base path"
Revert "Fix VMs forcing engine latch cvar to update to latched value"
A few commits ago I stopped VM's Cvar_Set() from instantly updating
latched values. Now VM can't call Cvar_Register() afterword to force
latched value to be used.
Reported by Noah Metzger (Chomenor).
VMs could set fs_game to '..' at anytime to access files outside of home
and base path. fs_game sent by server to clients could also be '..' to
access files outside of home and base path.
'..' was not caught by FS_CheckDirTraversal() as it expects filenames
not a single directory.
I've made fs_game be latched to prevent VMs from changing it with no
good way to validate it before it's used. com_basegame and fs_basegame
are now latched as well.
Additionally, it's now possible to change com_basegame while the engine
is running. game_restart or vid_restart will make it take affect.
com_homepath is now CVAR_PROTECTED to prevent VMs from changing it
to a directory traversal.
This requires my two previous commits for preventing VMs from changing
engine latch cvars and only Cvar_Get fs_game in FS_Startup (so CVAR_INIT
isn't added in serveral other places).
Reported by Noah Metzger (Chomenor).
If a VM increases sv_maxclients while a server is running the engine
will crash. The value should be latched until engine decides to update
the cvar; the same as when a user sets it.
VM could use Cvar_Register to set a protected cvar as user created and
was then able to use Cvar_Register with CVAR_ROM to change the value.
Don't allow Cvar_Register to affect protected cvars and prevent VMs
from adding internal flags to any cvars (creator, modified, protected,
nonexistent).
Reported by Noah Metzger (Chomenor).
There was an extra plus sign in Huff_Compress(). It wasn't causing any
issues as it does not affect the generated code. Removing it makes the
source code the same as Huff_Decompress().
The odd source code was brought to my attention by Tobias Kuehnhammer.
Use opengl1 renderer behavior of adding fixed amount of ambient light
to all models regardless of HDR setting. It fixes the view weapon
having zero ambient light on pillcity map.
Models for Team Arena's holdable medkit and invulnerability effects use
the frames numbers from the player's torso but the actual models only
have one frame (0).
Gauntlet and Grappling Hook use -1 ammo. Gauntlet is excluded from
the check ammo loop but Grappling Hook causes ammo time remaining
to decrease 200 milliseconds.
The out of ammo check tests time remaining is equal to zero. This
means carrying Grappling Hook and out of ammo will have negative
time remaining which results in the low ammo message being displayed
instead of out of ammo.
The game world is too dark when r_autoExposure is disabled. It can be
fixed by setting (cheat) r_cameraExposure to 1 but then the game is
too bright when r_autoExposure is enabled. So default r_cameraExposure
to 1 and make auto exposure subtract 1 from r_cameraExposure value.
The use of signed types in these expressions lead to overflow, hence undefined behaviour. The "sum" aggregator in Com_TouchMemory isn't even used (and presumbably just exists to inhibit optimizations from removing the memory access).
- Parse OpenGL version in sdl_glimp.c to share with both renderers.
- Add GL_VERSION_ATLEAST(major, minor) macro.
- Get address of glGetStringi if using OpenGL 3.
- Fix glConfig.extensions_string when using GL3 core context in
opengl2 renderer.
- Make opengl1 renderer's gfxinfo support qglGetStringi too.
Get all OpenGL functions using SDL_GL_GetProcAddress(). This makes it
easier to cross-arch compile on Linux and add support for OpenGL ES
in the future.
Users still have to supply their own libSDL2 for cross-arch compiling
on Linux. But now the user does not have to re-install libgl1-mesa-dev
package for i386 or amd64 on Debian when switching between compiling
ioquake3 for x86 and x86_64.
If a lightning bolt killed a player or the first shotgun pellet that
hit a player killed them, the shot was not counted as accurate.
Check if shot player is alive for hit accuracy before dealing damage.
Running a server using the client engine would send server snapshots
to loopback and LAN clients each client frame (com_maxfps).
This causes excessive network traffic and likely does not provide new
information to the clients because SV_Frame() won't have run between
the extra snapshots.
This commit restores the original behavior of respecting local/LAN
client's snaps userinfo and sv_fps. The issue was introduced by the
following commit:
Commit ac30d86db0
From: Thilo Schulz
Date: Wed, 13 Jul 2011 17:11:30 +0000
Subject: - Improve snapshot rate and data rate control
Reported by Eugene C.
