XLibre/xserver
1
1
Fork 0
mirror of https://github.com/X11Libre/xserver.git synced 2026-03-24 05:54:08 +00:00
Code Issues Packages Projects Releases Wiki Activity
22,349 Commits 314 Branches 33 Tags
e2c99a2e93c2438462ae46d6df281916ed8f8ccd
Commit Graph

22349 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
stefan11111
e2c99a2e93 glamor/glamor_egl.c: Fix misplaced #endif 
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-07 11:05:30 +01:00
stefan11111
8ed09861dc present: Call driver check_flip after we finish with our checks 
We don't really gain anything by calling `check_flip` early,
but it turns out that doing so causes problems.

Thanks to @jipok for finding this issue:
```
// gcc crash_x11libre.c -o crash_x11libre -lX11 -lXext  -lXpresent
// Code written by LLM!

// WARNING: THIS PROGRAM IS INTENDED TO CRASH THE X SERVER (via XPresent + Glamor bug)

void attempt_crash_via_present(Display* dpy, Window win, int width, int height, int depth) {
    XShmSegmentInfo shminfo;

    // 1. Calculate TIGHT stride (width * 4 bytes).
    // Misalignment here + XPresent is the killer combination.
    int stride = width * 4;
    size_t size = stride * height;

    // 2. Resize the window to match the buffer (simulating JTK resize behavior)
    XResizeWindow(dpy, win, width, height);
    // XSync(dpy, False); // Optional: let window manager catch up, but hammering triggers races better

    // 3. Allocate Shared Memory
    memset(&shminfo, 0, sizeof(shminfo));
    shminfo.shmid = shmget(IPC_PRIVATE, size, IPC_CREAT | 0777);
    if (shminfo.shmid < 0) { perror("shmget"); return; }
    shminfo.shmaddr = (char*)shmat(shminfo.shmid, 0, 0);
    shminfo.readOnly = False;
    shmctl(shminfo.shmid, IPC_RMID, 0); // Auto-cleanup

    if (!XShmAttach(dpy, &shminfo)) {
        fprintf(stderr, "XShmAttach failed\n");
        shmdt(shminfo.shmaddr);
        return;
    }

    // 4. Create the Misaligned Pixmap
    Pixmap pixmap = XShmCreatePixmap(dpy, win, shminfo.shmaddr, &shminfo, width, height, depth);

    // 5. THE TRIGGER: XPresentPixmap
    // This forces Glamor to treat the SHM segment as a presentable surface.
    // If strided alignment logic inside miCopyRegion/glamor is broken, this dies.
    XPresentPixmap(dpy, win, pixmap, 0, 0, 0, 0, 0, None, None, None, 0, 0, 0, 0, 0, 0);

    // Flush to make sure the server received the command
    XSync(dpy, False);

    // Cleanup for next iteration
    XFreePixmap(dpy, pixmap);
    XShmDetach(dpy, &shminfo);
    shmdt(shminfo.shmaddr);
}

int main() {
    Display* dpy = XOpenDisplay(NULL);
    if (!dpy) { fprintf(stderr, "No display\n"); return 1; }

    // Check extensions
    int major, minor, pixmaps;
    if (!XShmQueryVersion(dpy, &major, &minor, &pixmaps) || !pixmaps) {
        fprintf(stderr, "XShm not supported!\n"); return 1;
    }

    int present_opcode, event_base, error_base;
    if (!XQueryExtension(dpy, "Present", &present_opcode, &event_base, &error_base)) {
        fprintf(stderr, "XPresent extension not supported! (Can't repro bug)\n");
        return 1;
    }

    int screen = DefaultScreen(dpy);
    XVisualInfo vinfo;
    if (!XMatchVisualInfo(dpy, screen, 32, TrueColor, &vinfo)) {
        printf("32-bit visual not found, using default.\n");
        vinfo.visual = DefaultVisual(dpy, screen);
        vinfo.depth = DefaultDepth(dpy, screen);
    }

    XSetWindowAttributes attrs;
    attrs.colormap = XCreateColormap(dpy, RootWindow(dpy, screen), vinfo.visual, AllocNone);
    attrs.background_pixel = 0; // Black
    attrs.border_pixel = 0;

    Window win = XCreateWindow(dpy, RootWindow(dpy, screen),
                               0, 0, 100, 100, 0,
                               vinfo.depth, InputOutput, vinfo.visual,
                               CWColormap | CWBackPixel | CWBorderPixel, &attrs);

    XMapWindow(dpy, win);
    XSync(dpy, False);

    printf("Starting XPresent Stress Test...\n");
    sleep(1);

    // Fixed height (enough to have data), varying width
    int height = 500;

    // Iterate widths. We expect crashes on non-standard alignments.
    // XLibre often crashes at 1266 width (stride 5064).
    for (int w = 1000; w < 2000; w++) {
        int stride = w * 4;

        printf("\rTesting Width: %4d | Stride: %5d (mod 64: %2d)", w, stride, stride % 64);
        fflush(stdout);

        attempt_crash_via_present(dpy, win, w, height, vinfo.depth);

        // Small delay to let the server actually try to process the Present request
        // before we trash the memory, though XSync should handle it.
        usleep(1000);
    }

    printf("\nTest finished successfully (No crash).\n");

    XDestroyWindow(dpy, win);
    XCloseDisplay(dpy);
    return 0;
}
```

Fixes: https://github.com/X11Libre/xserver/issues/1754
Fixes: 9108a2bf55

Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-07 11:01:37 +01:00
stefan11111
0d93b942ad kdrive/src: Remove all SIGIO code from kdrive 
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-07 10:59:39 +01:00
stefan11111
fb16cef942 kdrive/src: Use threaded input in KdReleaseAllKeys 
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-07 10:59:39 +01:00
stefan11111
77f15917d4 kdrive/linux: Fix typo in keyboard.c 
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-07 10:59:39 +01:00
stefan11111
9afc6b1345 kdrive/src: revert c85ef2a6b5 
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-07 10:59:39 +01:00
stefan11111
7a50b1dc3d kdrive/src: revert fd2483745b 
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-07 10:59:39 +01:00
Enrico Weigelt, metux IT consult
885ba96ff6 os: xtrans: drop some unused defines 
The TRANS_LOCAL_*_INDEX defines aren't used anymore.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-07 10:56:57 +01:00
b-aaz
d2dcac6eca Revert "xkb: unexport XkbGetCoreMap() and XkbSetRepeatKeys()" 
This reverts commit 25d99353ee.
These symbols are required by the xorgxrdp driver.

Signed-off-by: b-aaz <b-aazbsd@proton.me>
 2026-01-07 10:56:33 +01:00
stefan11111
65d4e5e081 Xext/dri2: Fix message logging 
Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-06 19:23:34 +01:00
Enrico Weigelt, metux IT consult
1189bbdb40 xfree86: little documentation on XF86ModuleData 
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-05 16:12:05 +01:00
Enrico Weigelt, metux IT consult
0b1e8a759c stop defining _XF86DRI_SERVER_ 
This once was needed on including xf86driproto.h, but these day
have gone now for aeons.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-05 14:42:01 +01:00
stefan11111
186e269a9d glamor/glamor_egl.c: Skip modifiers that do not result in images 
that can be rendered to

According to https://registry.khronos.org/EGL/extensions/EXT/EGL_EXT_image_dma_buf_import_modifiers.txt ,
there is an interface that allows us to detect if a given format modifier is supported.

For example, nvidia uses this to report that pitch-linear images can't be
rendered to.

Since we are only interested in formats that we can actually render to,
we should strip these formats from the modifier list.

See:
https://github.com/elFarto/nvidia-vaapi-driver/issues/15#issuecomment-1015827050
https://gitlab.freedesktop.org/xorg/xserver/-/issues/1444#note_2450902

v2: Faster and simpler filter thanks to @algrid

Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-05 14:18:08 +01:00
Enrico Weigelt, metux IT consult
6890d547f2 dri: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-05 14:16:31 +01:00
stefan11111
dc39eda42b treewide: Move the dri2 extension to Xext and use it for all X servers 
This is needed to get proton working in Xfbdev, but is probably useful in other places too.

Xephyr has some unrelated issues regaring Xinput, so steam doesn't work there.

Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-02 17:51:43 +01:00
Alan Coopersmith
2a41522349 os: make FormatInt64() handle LONG_MIN correctly 
When compiling with gcc 15.2.0 using -O3 -m64 on Solaris SPARC & x64,
we'd get a test failure of:

Assertion failed: strcmp(logmsg, expected) == 0,
 file ../test/signal-logging.c, line 339, function logging_format

because 'num *= -1' produced a value that was out of the range of the
int64_t it was being stored in.  (Compiling with -O2 worked fine with
the same compiler/configuration/platform though.)

Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Part-of: <https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/2117>
 2026-01-01 13:23:22 +01:00
Oleh Nykyforchyn
3a367c5a47 xfree86/common: omit unnecessary search for matching devices 
It is patch 3/3 of a series that makes adding GPU screens
more controllable.

If AutoAddGPU is "off", matching devices for autoconfigured
drivers are sought for anyway, and then the unused list is
freed. This patch cancels an unnecessary search.

Signed-off-by: Oleh Nykyforchyn <oleh.nyk@gmail.com>
 2026-01-01 13:23:22 +01:00
Oleh Nykyforchyn
1dea31ce5c xfree86/man: add SingleDriver server flag to man page 
It is patch 2/3 of a series that makes adding GPU screens
more controllable.

Signed-off-by: Oleh Nykyforchyn <oleh.nyk@gmail.com>
 2026-01-01 13:23:22 +01:00
Oleh Nykyforchyn
c79f376594 xfree86/common: introduce SingleDriver server flag 
It is patch 1/3 of a series that makes adding GPU screens
more controllable.

If SingleDriver option is set to "on", then only the first
successfully probed driver adds non-GPU screens, others
may add secondary GPU screens only.

Fixes github.com/X11Libre/xserver/issues/1669

Signed-off-by: Oleh Nykyforchyn <oleh.nyk@gmail.com>
 2026-01-01 13:23:22 +01:00
stefan11111
8cb69cd810 modesetting: clear bo in drmmode_create_front_bo 
Fixes: https://github.com/X11Libre/xserver/issues/1740

Signed-off-by: stefan11111 <stefan11111@shitposting.expert>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
05fe0b519f Xext: selinux: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
8ff401cde1 Xext: xres: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
0616d9b325 composite: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
a0cd40d5af Xext: xvmc: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
d5e3eaeba2 xfixes: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
f3b783ec3d dri3: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Aggelos Tselios
4bf912f92b hw: xfree86: seatd: remove unnecessary semicolons 
Functions don't need to be terminated by semicolons, so this only
generates warnings.

Signed-off-by: Aggelos Tselios <aggelostselios777@gmail.com>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
095ea152b4 xfree86: dri2: use rpcbuf and sender macros 
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
callmetango
db9b427f3e .github: 01-bug-report: add version 25.1.X 
Signed-off-by: callmetango <callmetango@users.noreply.github.com>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
f122e2bb86 Xext: shm: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
dcb9329b37 Xext: saver: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
2a6b886563 Xext: dpms: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2026-01-01 13:23:22 +01:00
polaris
7b72ce3a23 Only call xf86platformVTProbe() when defined 
Signed-off-by: 9olaris <251053665+9olaris@users.noreply.github.com>
 2026-01-01 13:23:22 +01:00
Enrico Weigelt, metux IT consult
86a0ae295a .github: move signed-off check into main xserver build workflow 
Reduce the number of total workflows, so the list isn't so crowded.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
Signed-off-by: artist <artist@artixlinux.org>
 2025-12-22 23:16:14 +01:00
Enrico Weigelt, metux IT consult
32050edf76 .github: cygwin: skip setup signature check 
Cygwin currently has some mirror problem: it's missing the signature file
for the setup program, thus our build job is failing.

https://github.com/cygwin/cygwin-install-action/issues/39

Temporary workaround: just skip the signature check.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-22 13:34:44 +01:00
Enrico Weigelt, metux IT consult
2ada6f5178 .github: downgrade vmactions/dragonflybsd-vm to 1.1.4 
Latest vmactions/dragonflybsd-vm release (1.1.4) broke several things,
eg. not chdir'ing into the working tree.

For the time being, it's better to go back to 1.1.4.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-22 13:20:28 +01:00
Enrico Weigelt, metux IT consult
32210c1990 .github: downgrade vmactions/netbsd-vm to 1.2.3 
Latest vmactions/netbsd-vm release (1.2.4) broke several things,
eg. not chdir'ing into the working tree and NetBSD became *extremely*
slow (5mins job now slowed down to hours).

For the time being, it's better to just go back to 1.2.3.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-22 13:17:45 +01:00
Enrico Weigelt, metux IT consult
306e4a8a3f release 25.1 
Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
xlibre-xserver-25.1.0 		2025-12-21 15:53:30 +01:00
b-aaz
b4f0ee1b59 Added Cygwin builds to CI. 
Bringing back Cygwin support!
This will add Cygwin builds on windows-latest in our CI.

Signed-off-by: b-aaz <b-aazbsd@proton.me>
 2025-12-21 12:59:03 +01:00
b-aaz
8368df64d1 Xext: xf86bigfont: Include osdep.h for OsSignal. 
osdep.h holds the definition for OsSignal, and it was not included in
xf86bigfont which used it. This lead to build errors on Cygwin.

Signed-off-by: b-aaz <b-aazbsd@proton.me>
 2025-12-21 12:59:03 +01:00
b-aaz
d334e01e31 os: osdep: Disabled a section on Cygwin that was redundant. 
The section does some mappings between POSIX and windows APIs that are
already handled by Cygwin. This caused build errors on Cygwin due to
redefinition of things like the sigset_t struct.

Signed-off-by: b-aaz <b-aazbsd@proton.me>
 2025-12-21 12:59:03 +01:00
b-aaz
2c9d6b38d1 xwin: ddraw: Moved the include guard to its correct location. 
The include guard in this file has been incorrectly moved up breaking
the #include_next behavior on MS Windows builds.
Moved it to its correct location.

Signed-off-by: b-aaz <b-aazbsd@proton.me>
 2025-12-21 12:59:03 +01:00
b-aaz
7f7efe8195 Revert 9a82f5c, bring back Cygwin support. 
Brought back the code removed in that commit so that we can bring back
Cygwin support.
Small changes are done in the process of resolving conflicts against the
current head.

Some checks have not been reverted, because they were not necessary
anymore.

Signed-off-by: b-aaz <b-aazbsd@proton.me>
 2025-12-21 12:59:03 +01:00
Enrico Weigelt, metux IT consult
9bd66fc97d Xext: security: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-21 12:04:33 +01:00
Enrico Weigelt, metux IT consult
e85db86ced Xext: damage: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-21 12:01:33 +01:00
Enrico Weigelt, metux IT consult
ec973853c9 Xext: xtest: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-21 11:57:12 +01:00
Enrico Weigelt, metux IT consult
6f7aa46c51 Xext: geext: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-21 11:54:29 +01:00
Enrico Weigelt, metux IT consult
2e6a9a1fbc Xext: bigreq: use REQUEST_HEAD_STRUCT and REQUEST_FIELD_* macros 
Use the new macros to make request struct parsing / field swapping
much easier.

Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>
 2025-12-21 11:54:12 +01:00
b-aaz
a53697e70e xfree86: meson: Use install_symlink function instead of a hack. 
This was a old hack used before Meson 0.61.0 added install_symlink, we
are at the meson release 1.10.0 so it is safe to use the new proper
function for this task.

Also used the X server's executable name as the link target name, and as
the SUID wrapper's name.

Moved the minimum Meson version up to 0.61.0.

Signed-off-by: b-aaz <b-aazbsd@proton.me>
 2025-12-21 11:39:23 +01:00
Mikhail Dmitrichenko
252dda933a os: avoid closing null fd at Fopen 
In `Fopen` function variable `iop` may store NULL as a result of `fopen`
call. In this case, if later privileges couldn't be restored (`seteuid`
call fails), further `fclose(iop)` call will cause runtime error.

This commit adds check `iop` for NULL before calling `fclose` to prevent
potential NULL pointer dereference.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

Signed-off-by: Mikhail Dmitrichenko <m.dmitrichenko222@gmail.com>
Part-of: <https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/2115>
 2025-12-19 18:31:23 +01:00