steam-frame/FEX-Emu--FEX
1
0 Fork 0
mirror of https://github.com/FEX-Emu/FEX.git synced 2026-10-10 01:00:18 +02:00
Code Issues Packages Projects Releases 67 Wiki Activity
14,654 Commits 11 Branches 71 Tags
3e6b3c8c40ca093fa2b2e6edaade670a99e99a8a
Commit Graph
517 Commits
Author SHA1 Message Date
Ryan Houdek 341a5196ba Merge pull request #5701 from lioncash/thread 
Thread: Amend new thread handling in HandleNewClone()
 2026-07-11 17:09:59 -07:00
Ryan Houdek 48ce5bf6e6 Merge pull request #5714 from lioncash/readahead 
x32/FD: Fix readahead upper offset type
 2026-07-11 11:20:49 -07:00
Ryan Houdek 617a518714 Merge pull request #5713 from lioncash/select 
x32/FD: Correct total word calculation in select() variants
 2026-07-11 11:19:29 -07:00
Ryan Houdek 480f45f2f3 Merge pull request #5712 from lioncash/bpf 
BPFEmitter: Amend instruction class checking in HandleStore()
 2026-07-11 11:08:58 -07:00
Ryan Houdek 0e8b01c1b5 Merge pull request #5711 from lioncash/fault 
x64/Thread: Amend faulting copy handling related to LDTs
 2026-07-11 11:04:47 -07:00
LC a90a7dbec7 x32/FD: Fix readahead upper offset type 
This should be a uint32_t
 2026-07-11 13:33:11 -04:00
LC 2250bd58a2 x32/FD: Deduplicate guest and host fd set management 
Same behavior, but less copy pastey
 2026-07-11 13:23:19 -04:00
LC 126be45ed1 x32/FD: Correct total word calculation in select() variants 
Previously this would result in a larger amount of words specified than
necessary.

e.g. Given nfds = 1:

With AlignUp(1, 32) / 4, we'd end up with supposedly eight words, when it
should only be one word.

On the other extreme, given a full fd set of 1024 fds, then we'd end up
with 256 words, when it should only be 32.
 2026-07-11 12:53:08 -04:00
LC e021a55abd BPFEmitter: Amend instruction class checking in HandleStore() 
This was previously checking for a load class, which would result in ST
clobbering the index register.
 2026-07-11 12:26:11 -04:00
LC 2ad6254894 x64/Thread: Amend faulting copy handling related to LDTs 
CopyToUser doesn't return the number of bytes copied, but rather returns
0 to indicate success, otherwise a fault has occurred (and the SIGSEGV
handler has set X0 to EFAULT)
 2026-07-11 11:57:38 -04:00
LC f16bc12b7d SignalDelegator: Remove odd double negation in GuestSigProcMask 
We can just reduce it to normal null comparisons.
 2026-07-11 05:42:48 -04:00
Ryan Houdek d2f096187c Merge pull request #5705 from lioncash/thread3 2026-07-10 22:34:04 -07:00
LC 8b0f07b2dc Syscalls: Remove unnecessary usages of namespace FEXCore::IR 
These aren't necessary.
 2026-07-10 21:02:12 -04:00
LC 017c898ed0 x64/Signals: Amend set size in rt_sigtimedwait 
We should be checking the size passed in, not the sizeof of it.
 2026-07-10 19:40:05 -04:00
LC 85b8e91b57 Thread: Amend new thread handling in HandleNewClone() 
Ensures that newly cloned threads get tracked properly.
 2026-07-10 18:35:58 -04:00
Ryan Houdek 0bdf15977f Merge pull request #5697 from lioncash/thread 
x32/Thread: Move writability check around in waitpid
 2026-07-10 10:34:56 -07:00
Ryan Houdek 8a844f9cca Merge pull request #5695 from lioncash/socket 
Socket: Pass size by reference in getsockopt
 2026-07-10 10:34:33 -07:00
Ryan Houdek 8fbde84380 Merge pull request #5696 from lioncash/time 
x32/Time: Correct sizeof expression in utimensat
 2026-07-10 10:31:47 -07:00
Ryan Houdek 00826d3327 Merge pull request #5694 from lioncash/rlimit 
x32/Info: Only modify output in getrlimit/ugetrlimit if successful
 2026-07-10 10:27:51 -07:00
Ryan Houdek 5dd0477440 Merge pull request #5689 from lioncash/msg 
x32/Msg: Fix result comparison in mq_getsetattr
 2026-07-10 10:15:15 -07:00
Ryan Houdek c6f823f855 Merge pull request #5688 from lioncash/pidfd 
Thread: Amend pidfd_open check
 2026-07-10 10:14:49 -07:00
Ryan Houdek 2af7c24e79 Merge pull request #5687 from lioncash/limit 
x32/Thread: Fix off-by-one in get_thread_area
 2026-07-10 10:14:29 -07:00
Ryan Houdek 94ccfefa84 Merge pull request #5686 from lioncash/fd 
x32/FD: Ensure sendfile updates offset if set
 2026-07-10 10:14:18 -07:00
LC f4c50105ba x32/Thread: Move writability check around in waitpid 
Same behavior, but catches the write before it actually occurs.
 2026-07-10 08:08:45 -04:00
LC 3395bedc2b x32/Time: Correct sizeof expression in utimensat 
Ensures we check the proper type.
 2026-07-10 08:04:56 -04:00
LC 7a14210e2b Socket: Pass size by reference in getsockopt 
Previously this was passing by value.
 2026-07-10 08:01:29 -04:00
LC d206b67ca8 x32/Info: Only modify output in getrlimit/ugetrlimit if successful 
Avoids trampling over input data.
 2026-07-10 07:49:44 -04:00
LC d5ae87b5ca Thread: Amend pidfd_open check 
Checks for success.
 2026-07-10 06:25:43 -04:00
LC 02028eb1ad x32/Msg: Fix result comparison in mq_getsetattr 
Checks against failure instead of 1.
 2026-07-10 06:19:11 -04:00
LC 37b1432514 x32/Thread: Fix off-by-one in get_thread_area 
12, 13, and 14 are the only valid TLS areas.
 2026-07-10 06:06:30 -04:00
LC 6e8bc337aa x32/FD: Ensure sendfile updates offset if set 2026-07-10 05:58:49 -04:00
LC 4069a9f7d5 Timer: Fix typo in timer_gettime 
This should be passed by reference rather than by value.
 2026-07-10 05:33:06 -04:00
Ryan Houdek 24647820d7 Linux: Ensure 64-bit vsyscall page is tracked 
It's a purely virtual page even on x86, so we need to manually add it to
tracking.
 2026-06-03 18:03:31 -07:00
Ryan Houdek 7e8aa711ef Linux: Pass gettimeofday through glibc 
This ensures that it hits the VDSO path if possible.
 2026-06-03 17:45:41 -07:00
Pepper Gray 92dce9a2ea add missing header <thread> 
building using musl fails due to missing defintions:

```
Source/Tools/LinuxEmulation/LinuxSyscalls/Syscalls.cpp:913:23: error: no member named 'sleep_for' in namespace 'std::this_thread'
  913 |     std::this_thread::sleep_for(std::chrono::milliseconds {10});
      |                       ^~~~~~~~~
1 error generated.
```

Signed-off-by: Pepper Gray <hello@peppergray.xyz>
 2026-05-29 13:19:24 +02:00
Tony Wasserka a040740974 CodeCache: Ensure atomicity of code page finalization 2026-05-13 22:53:09 +02:00
Tony Wasserka 5be0dc9fc5 CodeCache: Implement lazy code loading 2026-05-13 21:25:47 +02:00
Tony Wasserka 52ad434d24 LinuxSyscalls: Defer MappedResource deletion until after code invalidation 
This ensures that any code buffer memory owned by the MappedResource is
invalidated before being deallocated.
 2026-05-13 21:25:47 +02:00
Ryan Houdek 7506cb4ea1 Merge pull request #5483 from neobrain/fix_guest_wrapper_code_cache 
CodeCache: Delay cache loading for guest library wrappers until after LoadLib
 2026-05-06 18:04:55 -07:00
Tony Wasserka b13b46eefe CodeCache: Delay cache loading for guest library wrappers until after LoadLib 
These libraries need to be initialized before relocating their caches,
since the guest function hashes won't be registered before.
 2026-05-05 16:29:39 +02:00
Tony Wasserka 06d77c1c19 SignalDelegator: Simplify support for musl's sigset_t 2026-05-04 16:28:47 +02:00
Ryan Houdek 85773995e1 Merge pull request #5470 from peppergrayxyz/header_redirect 
fix include redirect for <poll.h> and <signal.h>
 2026-05-03 04:41:18 -07:00
Pepper Gray 00b4777290 fix include redirect for <poll.h> and <signal.h> 
building on musl/clang causes redirecting incorrect #includes warnings:

```
warning: redirecting incorrect #include <sys/poll.h> to <poll.h> [-W#warnings]
warning: redirecting incorrect #include <sys/signal.h> to <signal.h> [-W#warnings]
```

include headers instead of sys/headers.

fix: #5469
Signed-off-by: Pepper Gray <hello@peppergray.xyz>
 2026-05-03 13:22:32 +02:00
Ryan Houdek 197e6de194 Merge pull request #5462 from peppergrayxyz/uc_sigmask 
determine sigset_t fieldname to enhance portability (musl)
 2026-05-03 03:48:53 -07:00
Ryan Houdek c402b15bd3 Merge pull request #5464 from peppergrayxyz/tgkill 
add header and classpath for tgkill
 2026-05-03 03:46:07 -07:00
Pepper Gray b1aab0e498 determine sigset_t fieldname to enhance portability (musl) 
musl build fails due to access to internal glibc member:

```
FEX/Source/Tools/LinuxEmulation/LinuxSyscalls/SignalDelegator.cpp:655:39: error: no member named '__val' in '__sigset_t'
  655 |       .SigMask = _context->uc_sigmask.__val[0],
      |                  ~~~~~~~~~~~~~~~~~~~~ ^
1 error generated.
```

add check to determine private glibc or musl member name or throw an error.

fixes: #5461
Signed-off-by: Pepper Gray <hello@peppergray.xyz>
 2026-05-03 12:17:09 +02:00
Pepper Gray 57d9dc037d add header and classpath for tgkill 
building on clang/musl fails with:

```
FEX/Source/Tools/LinuxEmulation/LinuxSyscalls/GdbServer.cpp:1174:5: error: use of undeclared identifier 'tgkill'
 1174 |     tgkill(::getpid(), ::getpid(), SIGKILL);
      |     ^~~~~~
1 error generated.
```

include and use `FHU::Syscalls::tgkill`.

fixes: #5463
Signed-off-by: Pepper Gray <hello@peppergray.xyz>
 2026-05-03 11:50:23 +02:00
Pepper Gray e3cfe28848 use <signal.h> instead of glibc header to enhance portability (musl) 
building using musl fails with:

```
FEX/Source/Tools/LinuxEmulation/LinuxSyscalls/ThreadManager.h:33:10: fatal error: 'bits/types/sigset_t.h' file not found
   33 | #include <bits/types/sigset_t.h>
      |          ^~~~~~~~~~~~~~~~~~~~~~~
1 error generated.
```

`<bits/types/sigset_t.h>` is a glibc internal header, use <signal.h> instead.

fixes: #5459
Signed-off-by: Pepper Gray <hello@peppergray.xyz>
 2026-05-03 11:03:38 +02:00
Ryan Houdek 01a3ab6ca7 FEXCore: Use WritePriorityMutex for Code Invalidation Mutex 
The previous `ForkableSharedMutex` using `std::shared_mutex` was showing
up as significant CPU time on arm64ec. In particular it was showing up
upwards of 700ms/S of CPU time for read-contented workloads at only
~2700 locks per second. The libc++ implementation for arm64ec must be
particularly gnarly for this to be so slow.

With this swapped over, it's now only spending around 56ms/S on the
contended shared lock, but at ~8000 locks per second. So a significant
uplift.
 2026-03-31 19:02:51 -07:00
LC 5b4a5969cc Merge pull request #5401 from Sonicadvance1/126 
Linux/GuestFrames: Ensure MXCSR is saved and restored on signal
 2026-03-28 22:25:59 -04:00