Signed-off-by: Colin Kinloch <colin.kinloch@collabora.com>
Signed-off-by: Colin Kinloch <colin.kinloch@collabora.com>
Co-authored-by: Colin Kinloch <colin.kinloch@collabora.com>
* * Use flags for rarch_state
* Get rid of ^M linebreaks in retroarch_types.h
* Buildfixes for consoles
* (audio driver) use flags instead of bools
* (video) Use flags instead of bools
* Rewrite input driver state bools into flags
* Reduce snprintf and/or use them only for concatenating the
string formatting of numbers/values
* Reduce snprintfs
* Use snprintf for concatenation at parts
* * We don't need to NULL-terminate strings that get passed to strlcpy/strlcat
and friends
* Use snprintf for concatenation in certain instances
* Do away with some string intermediary copies where we can avoid it
* Fix warning unused variable
* * Reduce snprintf calls
* Rewrite snprintf calls into strlcpy where possible
* Use snprintf for concatenation when necessary
* Do away with some string intermediary copies in task_translation.c
* run_translation_service - make switch slightly smaller
when the previous title differs from the current title
(X11) Have to roll back the previous title optimization/less calls to XChangeProperty,
because the title is lost upon toggling between fullscreen/windowed and is no longer
set
at regular intervals even when the title output for the window title had not
changed at all, now we compare it properly against the last set title, and
only if it has changed will we make the call to XChangeProperty. Also, do
away with strlen(title) calculation and use return value of
video_driver_get_window_title instead
(menu_displaylist.c) Some optimizations:
* Make sure val_d is never bigger than 16, it should just
contain a number. 16 is more than plenty (and probably too big even)
* Hose val_d out of loops if its value never changes inside the loop
* Attempt to avoid calling msg_hash_to_str inside loops unnecessarily
(snprintf) Try to reduce or simplify snprintf calls, only tend to
use it for processing integers/numbers and avoid it for regular
string concatenation (NOTE: we try to be a bit safer about it to
address earlier cited criticism, although we don't consider concatenating
3 or 4 characters at the end to be insecure)
(msg_hash_to_str) Try to avoid duplicate calls to the same localized
string when we can just cache the results once instead locally
file path names instead of snprintf
(ctr_gfx.c) Use strlcpy instead of snprintf where possible
(ctr_gfx.c) Use snprintf instead of sprintf where possible
(ctr_gfx.c) Some general style nits
* steam.c - cleanups:
* Use string_to_lower from libretro-common/stdstring.c instead of
its own version
* Some stylistic changes
* Rewrite strncpy calls into strlcpy/strlcat/manual assignment
* Make it C89 compliant
* Some unused variables
Reasons:
1 - Just a macro for strcpy
1.2 - doesn't have a return type unlike strlcpy, can't be used
for string concatenation
1.3 - unsafe compared to strlcpy
Do either manual string assignment per character or strlcpy, no
inbetween by resorting to strcpy
(retroarch.c) retroarch_get_capabilities - get rid of strlcat
party in RARCH_CAPABILITIES_CPU, do simple manual assignment for
concatenation instead and NULL terminate at the end
(menu_displaylist.c) Do something about some of the unnecessarily huge
local strings of PATH_MAX_LENGTH size (can be as large as 8192)
(menu_displaylist.c) Replace some trivial strlcats / snprintfs
(D3D10/D3D11/D3D12) Strings don't need to be initialized -
strlcpy already null-terminates - get rid of some constants
- (database_info.c) Optimizations for database_info_build_query_enum() -
it would call strlcat at the very least 3 times, normally more than that.
Every strlcat call is at least one or more strlens. Reduce it all down
to manual string concatenation and one strlcat
- (database_info_build_query_enum) - always NULL terminates string so no need for NULL termination done by callee
* (menu_displaylist) use smaller strings for basename paths, remove one unnecessary string
in menu_displaylist.c function
- (disk_index_file.c) Avoid another trivial string concatenation by doing
it manually after fill_pathname_join_special
- (fill_pathname_join_delim) since fill_pathname_join_delim() now has a return value, use it for
string concatenation
- (platform_unix) - Use BUFSIZ as size for a buffer in system_property_get
- some other trivial strlcat usage replacements
(menu_contentless_cores) Fix mistake - needed to set len to return
value of fill_pathname_join_special
(configuration.c) Fix some unused variable warnings on Linux
CyberLab custom Blargg NTSC Filter Presets offer improved dedithering with less scrolling, shimmering and other artifacts than current core included Blargg NTSC Filter implementations.
They also work properly with several cores that would otherwise result in a squished image if using the Blargg NTSC SNES filters currently available in the Video Filters folder.
as deprecated.
* Use fill_pathname_join_special in the vast majority of cases where
we can ensure out_path is a new empty string
* Get rid of some extension concatenation with strlcat where encountered
* Some general cleanups with NULL termination of strings that get immediately
passed to strlcpy/strlcpy-adjacent functions
* struct was erroneously named 'catagory' - changed to 'category'
* Removed some strlcpys that were just setting the string to a fixed constant value
* Random nits
* (joypad_connection) Small optimizations -
* Turn functions static where possible
* Hose strlen call out of loop
* (input_driver.c) General cleanups:
* Some small code/style nits
task_screenshot.c:
* Move widget callback function for screenshots to task_screenshot.c
(file_path.c):
* Turn get_pathname_num_slashes into static function
* path_linked_list_free - always returns true, so get rid of return value
* path_linked_list_new - fix function signature
* path_get_archive_delim - do not NULL-terminate string, already done by strlcpy later on
General:
* Slight optimizations - use int/size_t for loop counter variable instead of unsigned
* Take advantage of fact that strlcpy already NULL-terminates, so don't do this explicitly
outside if we're just going to end up calling strlcpy/fill_pathname_join on it anyway
according to documentation note - 'You must not deallocate memory before calling
FT_Done_Face' - so store a pointer to the font data and then free it after calling
FT_Done_Face
- Get rid of memset, set the member variables to 0 manually instead. Note - rcDevice is behind a _MAC ifdef in the headers (some hypothetical Mac port
of Windows which never happened and therefore the ifdef is useless), so rcDevice is not set and will just be unused anyway
- Variable scoping, move variables to where they are needed
- get rid of unused gfx_display_driver_exists
- get rid of useless wrapper function gfx_display_font_init_first
- rewrite gfx_display_font_file somewhat
font from memory - first read it from file to memory beforehand -
this solves an asset extraction issue when selecting 'Update Assets' -
apparently FT_New_Face keeps an open file handle to the font file which
prevents it from being overwritten/deleted while the program is still
running.
TODO/FIXME - move file loading code out of the font renderer init function
and move it higher level
When a shader preset is loaded now when there are more than one #reference line the parameter values will be loaded from the references after the first one
syscall -
On older systems, you'll have to include linux/memfd.h for the MFD_ defines, and call memfd_create() via the the syscall(2) wrapper (and include unistd.h and sys/syscall.h for it work). We exclude linux/memfd.h header include because we already provide
the MFD_ defines in case they are missing
Multi reference settings working
Config fixes
Multi-Reference Loading Working, Checking for shader chain not working
Checking References is working
Improved path_linked_list_add_path
Simplify path list stuff
Fixed Static Definition
Fixes
Fixes
Fix to Trigger PR Update
Removed Legacy Orbis which got left in a merge
Memory Leak Fix
* Honor SUBOPTIMAL on non-Android since you'd want to recreate
swapchains then. On Android it can be promoted to SUCCESS.
SUBOPTIMAL_KHR can happen there when rotation (pre-rotate) is wrong.
* Small cleanups in general - less conditionals
Refresh rate in exclusive fullscreen mode was being incorrectly multiplied by vsync swap interval, breaking swap interval functionality at the gfx driver level
* CXX_BUILD buildfixes
* (D3D11) Buildfixes for CXX_BUILD
* (Linux/qb) Disable KMS for Linux when building for C89, headers use inline which is not
available for C89
Move frequently accessed member variables to the beginning of the
structure to improve cache locality. Having menu_texture at the end of
the structure also saves one instruction in the hot path of
sdl_rs90_gfx_frame() (for the ARMv5TE Miyoo) as well as a few other
functions.
* features_cpu: Fix clock for djgpp
* frontend_dos_set_fork: Fix wrong id names
* dos_input: Fix missing indexing
* dos_joypad: Fix wrong id name
* Fix menu drawing
Menu is already prescaled to VGA_WIDTH x VGA_HEIGHT in vga_menu_frame. We need
to always use those dimensions
* Change dos compilation from griffin to common
* Set vga menu size to 320x200
* Add DOS compilation in gitlab-ci.yml
* Support new elf2dol path in Makefile.griffin
* Fix NULL pointer dereference when no core info is available
* Fix double include of different versions of libogc
* Move ngc and wii to Makefile.common
Resulting binary is 60K *smaller*. This is surprising but probably due to
less inlining happening
* gx_gfx: Fix hang on gfx initialization
* Remove sthread_isself on NGC/Wii
It relies on pthread_equal that is not there for either legacy pthreads
or normal pthreads on those platforms
* Wii networking
* Missing includes in split
* Network fixes
* Fix logging of non-format string as a format
* Support for gx netlogger
* Network fix
* Network fixes
* Network info
* gx: Move config files to app directory
Otherwise it clashes between wiiU and vWii version with unpredictable
effects
* (Wayland) Show window early to get screen info
* Report current monitor size on initial get_video_size call
* Fullscreen to monitor selected in settings
* Fullscreen on current monitor when setting is set to automatic
* Fallback to shm_open when memfd_create is not supported
* Fix translations with messed-up string placement
* Fix Hebrew and Persian text display
Hebrew is now displayed properly as right-to-left.
Miscellaneous whitespaces, like the zero width non-joiner used in Persian, no longer break right-to-left texts.
* Report current monitor size on initial get_video_size call
* Fullscreen to monitor selected in settings
* Fullscreen on current monitor when setting is set to automatic
* Fix handling of big-endianness in gl1
* Enable gl1 on osx-ppc
* Force gl1 on apple if gfx supports only gl1
* Backport gl2 apple workaround to gl1
* Fix big-endianness handling in widgets
* Fix old osx condition
Current code assumes that osx < 10.12 is equivalent to ppc osx. It's not
true as Leopard x86 is still < 10.12 but not ppc. As xcode compiles fat
binaries it includes osx x86 and compilation fails.
* Disable crtswitchres when no c++11 is available
Crtswitchres altually needs c++11. Since it's not that important to make
it compatible with lower c++, just disable if no c++11 is available
* Don't use firstObject on old Mac OS X.
It was introduced in 10.6, so on old ones just implement it ourselves
* Compile osx-ppc frontend
* osx-ppc: Build a fat binary
On 10.6 i386 xcode apparently refuses to build a pure ppc.
Settle for a fat binary.
Currently each time a screen resolution or setting change occurs, fbdev_destroy, fbdev_init and fbdev_set_mode are called in sequence, trying to destroy context and surface (though context pointer seems to remain unchanged) and create again both of them.
However it seems something is wrong with egl_destroy, as after the call fps drops from 60 to 33fps in GUI with huge performance impact, at least with libmali.
Philosophy is changed with this commit to avoid destroying and creating context and surface each time (creation only occurs in fbdev_init and egl_destroy is called only if retroarch is shutting down).
As a minor modification, framebuffer is resetted to 0 on retroarch shutdown to avoid any chance of freezed screen effect.
* Fixed crash on boot with xmb and vulkan driver Fixed wrong blend state affecting the ribbon shader Fixed up C++ style comments with C style
* Fixed compilation issue - I made a last minute change and thought I had compiled it but obviously hadn't! No wonder it worked.
-In case yoffset is not set=0 by the driver (which seems to occur randomly on certain configurations), on next set_video_mode call screen freezes and cannot be updated. Condition to set yoffset to 0 if current value is >0 is added as fix.
-If setterm is not available, sh error occurs in the shell on gfx_ctx_mali_fbdev_destroy call. Condition based on setterm availability is added to avoid this.
-If pixclock from driver is not available (=0), divide by 0 occurs when trying to calculate refresh rate. Condition is added as a fix, if pixclock=0 refresh rate is set as 60.
* Add option for showing the overlay behind the menu
This commit lays the groundwork for this option. Support for this option
in the video drivers themselves is going to be added in later commits.
* gl1: Add overlay behind menu support
* gl2: Add overlay behind menu support
* gl3: Add overlay behind menu support
* vulkan: Add overlay behind menu support
* ctr: Add overlay behind menu support
* d3d9: Add overlay behind menu support
* d3d10: Add overlay behind menu support
* d3d11: Add overlay behind menu support
* d3d12: Add overlay behind menu support
* CHANGES.md: overlay behind menu
Co-authored-by: MrHuu <MrHuu@users.noreply.github.com>
Co-authored-by: Tony <45124675+sonninnos@users.noreply.github.com>
* force angle to render at 1080p regardless of screensize as the output is 1080p regardless of screensize.
This fixes an issue where at 4k any angle output would be zoomed into a corner
* set resolution based on display resolution (auto 4k)
* set driver to d3d11 if booting with opengl
* reset width and height of output on boot to match display
Co-authored-by: Tunip3 <tunip3@users.noreply.github.com>
* Netplay Stuff
## PROTOCOL FALLBACK
In order to support older clients a protocol fallback system was introduced.
The host will no longer send its header automatically after a TCP connection is established, instead, it awaits for the client to send his before determining which protocol this connection is going to operate on.
Netplay has now two protocols, a low protocol and a high protocol; the low protocol is the minimum protocol it supports, while the high protocol is the highest protocol it can operate on.
To fully support older clients, a hack was necessary: sending the high protocol in the unused client's header salt field, while keeping the protocol field to the low protocol. Without this hack we would only be able to support older clients if a newer client was the host.
Any future system can make use of this system by checking connection->netplay_protocol, which is available for both the client and host.
## NETPLAY CHAT
Starting with protocol 6, netplay chat is available through the new NETPLAY_CMD_PLAYER_CHAT command.
Limitations of the command code, which causes a disconnection on unknown commands, makes this system not possible on protocol 5.
Protocol 5 connections can neither send nor receive chat, but other netplay operations are unaffected.
Clients send chat as a string to the server, and it's the server's sole responsability to relay chat messages.
As of now, sending chat uses RetroArch's input menu, while the display of on-screen chat uses a widget overlay and RetroArch's notifications as a fallback.
If a new overlay and/or input system is desired, no backwards compatibility changes need to be made.
Only clients in playing mode (as opposed to spectating mode) can send and receive chat.
## SETTINGS SHARING
Some settings are better used when both host and clients share the same configuration.
As of protocol 6, the following settings will be shared from host to clients (without altering a client's configuration file): input latency frames and allow pausing.
## NETPLAY TUNNEL/MITM
With the current MITM system being defunct (at least as of 1.9.X), a new system was in order to solve most if not all of the problems with the current system.
This new system uses a tunneling approach, which is similar to most VPN and tunneling services around.
Tunnel commands:
RATS[unique id] (RetroArch Tunnel Session) - 16 bytes -> When this command is sent with a zeroed unique id, the tunnel server interprets this as a netplay host wanting to create a new session, in this case, the same command is returned to the host, but now with its unique session id. When a client needs to connect to a host, this command is sent with the unique session id of the host, causing the tunnel server to send a RATL command to the host.
RATL[unique id] (RetroArch Tunnel Link) - 16 bytes -> The tunnel server sends this command to the host when a client wants to connect to the host. Once the host receives this command, it establishes a new connection to the tunnel server, sending this command together with the client's unique id through this new connection, causing the tunnel server to link this connection to the connection of the client.
RATP (RetroArch Tunnel Ping) - 4 bytes -> The tunnel server sends this command to verify that the host, whom the session belongs to, is still around. The host replies with the same command. A session is closed if the tunnel server can not verify that the host is alive.
Operations:
Host -> Instead of listening and accepting connections, it connects to the tunnel server, requests a new session and then monitor this connection for new linking requests. Once a request is received, it establishes a new connection to the tunnel server for linking with a client. The tunnel server's address and port are obtained by querying the lobby server. The host will publish its session id together with the rest of its info to the lobby server.
Client -> It connects to the tunnel server and then sends the session id of the host it wants to connect to. A host's session id is obtained from the json data sent by the lobby server.
Improvements (from current MITM system):
No longer a risk of TCP port exhaustion; we only use one port now at the tunnel server.
Very little cpu usage. About 95% net I/O bound now.
Future backwards compatible with any and all changes to netplay as it no longer runs any netplay logic at MITM servers.
No longer operates the host in client mode, which was a source of many of the current problems.
Cleaner and more maintainable system and code.
Notable functions:
netplay_mitm_query -> Grabs the tunnel's address and port from the lobby server.
init_tcp_socket -> Handles the creation and operation mode of the TCP socket based on whether it's host, host+MITM or client.
handle_mitm_connection -> Creates and completes linking connections and replies to ping commands (only 1 of each per call to not affect performance).
## MISC
Ping Limiter: If a client's estimated latency to the server is higher than this value, connection will be dropped just before finishing the netplay handshake.
Ping Counter: A ping counter (similar to the FPS one) can be shown in the bottom right corner of the screen, if you are connected to a host.
LAN Discovery: Refactored and moved to its own "Refresh Netplay LAN List" button.
## FIXES
Many minor fixes to the current netplay implementation are also included.
* Remove NETPLAY_TEST_BUILD
* Allow parallel compilation in PS2
* Allow to compile with griffin or common compilation in PS2
* Enable dummy core to be used in other platforms
* Use threads in YML config
* Add the compilation to PS2 in GitHub Actions
The emulating mailbox infrastructure is very hacky and needs a rethink
to be able to properly support screenshots. emulating mailbox is pretty
nasty to begin with and should be considered a hack.
* CRT Fractal scalling fix for SR2
Added comment to remind me to add video core to SR2
fixed typo
* Addedd only x fractal scaling for super resolutions
* Properly handle fractional scale factors and fix some typos.
Co-authored-by: Antonio Giner <calamity15khz@gmail.com>
LibretroAdmin