23,567 commits

This branch

This branch

All branches

Author	SHA1	Message	Date
stefan11111	ce761d7101	kdrive/fbdev: Respect the user-provided refresh rate ... If the refresh rate is in the known mode list, we use that. If not, we generate a mode using the CVT formula. If we cannot set the mode we want, we try other modes, but don't exceed the requested refresh rate. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-07 11:27:04 +02:00
stefan11111	6a4b9633a0	kdrive: Implement version 1.2 of the CVT reduced blanking formula ... Like libxcvt, we are truncating instead of rounding to the nearest integer, for safety reasons. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-07 11:27:04 +02:00
Enrico Weigelt, metux IT consult	0229438f09	ci: run pyxtest under valgrind memcheck on Ubuntu ... Add a dedicated xserver-build-ubuntu-valgrind job that builds a debug Xvfb and runs the pyxtest suite (test/pyxtest/) with --valgrind, so memory errors (invalid reads/writes, use-after-free, uninitialised value use) are caught in CI, not just OOB bugs ASAN happens to catch locally. The pytest-based harness (test/pyxtest/conftest.py, xserver.py, valgrind.py) already implements valgrind support; it was simply never wired into CI. pytest itself was also not installed in the Ubuntu CI image, so the existing pyxtest suite has been silently skipped (not built, since meson's find_program('pytest', required: false) found nothing) on every other Ubuntu lane too. Installing python3-pytest fixes that as a side effect: the normal xserver-build-ubuntu* jobs now also actually run pyxtest (without valgrind) as part of "meson test", exercising the existing protocol-fuzzing regression tests that were previously dead weight. Kept as a separate lane rather than adding --valgrind to the existing run-xserver-build-and-test.sh: valgrind is slow, and pyxtest only needs Xvfb, not the full Xorg/Xephyr/Xnest/XTS/piglit build the other lanes carry. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-07 11:16:03 +02:00
Enrico Weigelt, metux IT consult	a959773be9	ci: freebsd: synthesize a bzip2.pc if the bzip2 package doesn't ship one ... Installing bzip2 (previous commit) wasn't enough: on this CI image the package provides libbz2 + headers but no bzip2.pc, so pkgconf's dependency resolution for freetype2 (Requires.private: bzip2, pulled in transitively by xfont2) still fails outright even though the library itself is present and usable - confirmed via the job log still showing 'Package bzip2 was not found in the pkg-config search path' with env[PKG_CONFIG_PATH] correctly unset (default search path), ruling out a path problem. This is missing package metadata, not a missing library: write a minimal bzip2.pc if pkg-config --exists bzip2 fails, locating the actual installed lib/header paths via 'pkg info -l bzip2' rather than hardcoding them. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-06 19:36:54 +02:00
Enrico Weigelt, metux IT consult	f95ab701fe	ci: freebsd: install bzip2 (freetype2's pkgconfig Requires) ... xfont2's pkg-config dependency lookup failed once the IGNORE_OSVERSION fix (previous commit) let the package-version check through to the actual build: freetype2's .pc file Requires: bzip2, but bzip2 wasn't in the install list, so pkgconfig couldn't generate cflags for xfont2 and meson bailed. Pre-existing gap, unmasked (not caused) by the previous fix - this lane never got far enough to hit it before. The 'ld: error: undefined symbol: main' lines in the same log are unrelated noise from meson's own linker-capability probe (compiling a main-less test program to detect LLD), not a real link failure. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-06 19:36:54 +02:00
Enrico Weigelt, metux IT consult	aaab164c0e	ci: freebsd: set IGNORE_OSVERSION=yes to survive pkg repo/VM release skew ... FreeBSD's official package repo tracks the '14' ABI branch, not the exact point release pinned for the CI VM (release: "14.3" in build-xserver.yml). When the repo rolls forward to packages built against a newer point release, pkg refuses to proceed non-interactively ("repository FreeBSD contains packages for wrong OS version", no TTY in CI to answer the y/N prompt), so pkg update -f fails every time and the job never reaches the actual build - currently breaking every FreeBSD CI run on master and on open PRs alike. IGNORE_OSVERSION=yes is pkg's own documented escape hatch for exactly this case; a one-point-release ABI skew is expected to be compatible in practice. Confirmed via job logs: 'pkg: ... wrong OS version: FreeBSD:14:amd64' / 'Newer FreeBSD version for package zycore-c: package: 1404000, running userland: 1403000' immediately precedes the generic 'pkg update -f' retry-loop failure that was previously misdiagnosed as a network flake. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-06 19:36:54 +02:00
Enrico Weigelt, metux IT consult	5f85319446	Xext: namespace: allow CreateColormap against the real root window ... The real-root-window allowlist in hookResourceAccess() covered CreateWindow/CreateGC/CreatePixmap/DestroyWindow/TranslateCoords/ QueryTree/ChangeWindowAttributes/QueryPointer/SendEvent/XInput, but missed CreateColormap - which, like CreateGC/CreatePixmap, only needs read-only access to the root window to find its screen/visual. Without it, any confined (non-superPower) client creating a colormap against the real root - which glxgears and most GL/visual-using clients do on startup - got BadAccess and could never render. Reproduced end-to-end: a FlyingTux-managed glxgears container confined to its own namespace failed with 'X Error: BadAccess, Request Major code 78 (CreateColormap)' until this was added; with it, the client renders normally while namespace refcounts confirm it stayed confined to its own namespace throughout. Signed-off-by: Enrico Weigelt, metux IT consult <enrico.weigelt@vnc.biz>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	97278df78a	Xext: namespace: fix NULL-deref crash for protocol-created namespaces ... A namespace created at runtime via the CreateNamespace protocol request never got a virtual root window assigned: XnsCreate() left rootWindow at its calloc'd NULL, since only hookInitRootWindow() (a one-shot PostInitRootWindowCallback fired once at boot, before any client can reach the extension) ever populated that field, for the fixed set of namespaces known from the config file at that point. Any client placed into such a namespace then crashed the whole server with a NULL-pointer segfault in hookWindowProperty() on its very first GetProperty request (Xext/namespace/hook-windowproperty.c:37, 'subj->ns->rootWindow->drawable.id') - which virtually every X11 client issues during normal startup. Reproduced end-to-end via a real FlyingTux-managed container app connecting through a freshly created namespace. Factor the per-namespace virtual-root-window creation out of hookInitRootWindow() into XnsCreateVirtualRoot(), and call it from XnsCreate() too, so a namespace created via the protocol gets the same virtual root a config-time one gets at boot. Signed-off-by: Enrico Weigelt, metux IT consult <enrico.weigelt@vnc.biz>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	0523619977	doc: Xnamespace-protocol: fix stale attribute names in CreateNamespace/§8 ... CreateNamespace's field comment and the §8 required-changes list still named the pre-implementation identifiers XnsAttrAutoRemove/XnsAttrBuiltin; the actual wire constants are XNS_ATTR_TRANSIENT/XNS_ATTR_IMMUTABLE. Also clean up the garbled §8 item 4 sentence. Flagged as an advisory (non-blocking) note in the bot review. Signed-off-by: Enrico Weigelt, metux IT consult <enrico.weigelt@vnc.biz>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	31336c8a4b	Xext: namespace: grant superPower bypass in resource-access hook ... Every other Xace hook in this extension already exempts superPower clients (hook-client.c, hook-ext-access.c, hook-ext-dispatch.c, hook-windowproperty.c, hook-send.c, hook-server.c, hook-device.c, hook-selection.c) but hookResourceAccess() did not, so a superPower client (e.g. a WM) could not resolve GetClientNamespace(windowID) for a window owned by a client in another namespace - exactly the gap probono asked about. Verified: xnamespace -s whoami <window-id-in-other-ns> now returns that namespace instead of BadAccess; go-x11proto conformance suite still 40/40 (both byte orders). Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	410e0ac3af	Xext: namespace: add Xnamespace management protocol extension ... Add a superPower-only X11 extension (X-NAMESPACE) for managing namespaces at runtime, the settings that were previously only provisionable from the namespace.conf file at startup. Requests: QueryVersion, {List,Create,Delete,Query}Namespace, SetNamespaceFlags, {Add,Remove,List}AuthToken and GetClientNamespace. Names are the wire identifier; capabilities/attributes are bitmasks. Transient namespaces auto-remove when their last client exits. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	e16cc89236	Xext: namespace: register auth tokens through a shared helper ... Introduce XnsAddToken(), which allocates an auth token, copies the protocol name and key, registers the authorization and assigns a stable per-namespace handle (struct auth_token.handle, allocated from a per-namespace counter). Route the config loader's "auth" parsing through it, so tokens loaded from the config now carry handles too and there is a single token-registration code path (reused by the upcoming management protocol). Also initialise the auth_tokens list for the built-in namespaces and for namespaces created by the config loader, fixing a latent crash when iterating the token list of a namespace that has no tokens. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	04e2f98f27	Xext: namespace: factor out length-aware namespace lookup ... Add XnsLookup(), which matches a namespace by an explicit byte length rather than requiring a NUL-terminated string, and route both the existing XnsFindByName() and select_ns() through it. This removes three copies of the same "walk ns_list comparing names" loop and gives a single lookup that also works on names taken straight from a request buffer (used by the upcoming management protocol). Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	7bb4869b45	dix: request_priv.h: some more request parsing macros ... Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 20:05:08 +02:00
Enrico Weigelt, metux IT consult	bf6b62648d	Xi: xiselectev: fix OOB read in ProcXISelectEvents' non-swapped path ... The per-mask loop read evmask->mask_len to compute the next mask's length before checking that the fixed xXIEventMask header itself was still within the request buffer - a request with num_masks >= 1 and insufficient trailing data reads one xXIEventMask-header's worth past the end of the request. The byte-swapped path just above already performs the equivalent pre-check ('len < bytes_to_int32(sizeof (xXIEventMask))') before touching evmask->mask_len; this brings the non-swapped path in line with it. Signed-off-by: Enrico Weigelt, metux IT consult <enrico.weigelt@vnc.biz>	2026-07-03 19:59:53 +02:00
Enrico Weigelt, metux IT consult	6a97f6d38e	render: filter: fix integer-overflow bypass of convolution size check ... convolutionFilterValidateParams() derives w/h from client-supplied xFixed values via the sign-extending xFixedToInt(), then checked 'w * h > nparams' as a plain (signed) int product. A client can choose negative w/h so the mathematical product is huge but the signed int product wraps to a small or negative value, defeating the bounds check that is supposed to reject a convolution kernel larger than the actual parameter list. Reject negative w/h outright and compare the product as int64_t so it cannot wrap before the comparison. Signed-off-by: Enrico Weigelt, metux IT consult <enrico.weigelt@vnc.biz>	2026-07-03 19:59:53 +02:00
Enrico Weigelt, metux IT consult	7fff115a93	panoramiX: zero-init xAllocColorReply to avoid leaking uninitialised bytes ... The reply struct was declared bare (uninitialised auto storage) and then filled field-by-field (red/green/blue/pixel only), leaving pad1/pad2 and any trailing padding uninitialised. Those bytes were then sent to the client via X_SEND_REPLY_SIMPLE, an information disclosure. Fix it with a designated initialiser so the pad/reserved bytes are zeroed. This block sits inside XINERAMA_FOR_EACH_SCREEN_BACKWARD(), which took a single macro parameter — so a top-level '{ .a = x, .b = y }' initialiser was split at its commas into multiple macro arguments (the old "static init would confuse preprocessor" caveat). Rather than work around that per call site, make the XINERAMA_FOR_EACH_SCREEN_* macros variadic (pass the body as __VA_ARGS__); commas in the body are then preserved, and an ordinary designated initialiser can be used with no compound-literal cast or extra parentheses. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 13:23:25 +02:00
stefan11111	2427dff73b	kdrive: Don't fixup the cursor position twice in KdCursorOffScreen ... The intent of the code is to take the coordinates of the cursor and adjust them so that they are where they are supposed to be on the new screen. If we're moving up on a screen with a larger height that the one we are on, *y < 0, so we fix it by adding the new screen's height. Since the height is larger, and the code was lacking an else branch, it thinks that we are now moving down, so it incorrectly subtracts the old screen height, making the cursor end up somewhere it's not supposed to be. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-03 10:24:22 +02:00
stefan11111	f8ee2e0734	kdrive/fbdev: Add some config examples to the man page ... We had no info on how to pass multi-screen config options to Xfbdev. We also had no example configs to guide people interested in configuring Xfbdev. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-03 10:24:22 +02:00
Jeremy Huddleston Sequoia	2a0fd97cdf	xquartz/GL: silence OpenGL deprecation warnings ... OpenGL.framework and CGL are annotated OPENGL_DEPRECATED(_, 10.14) in the SDK, but that can be silenced with -DGL_SILENCE_DEPRECATION. Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com> (cherry picked from commit 31060c9506)	2026-07-03 10:24:13 +02:00
Jeremy Huddleston Sequoia	73de6346ff	xquartz/GL: advertise GLX_ARB_create_context and _profile ... XQuartz never called __glXEnableExtension for GLX_ARB_create_context or its profile / robustness companions, so clients could not use glXCreateContextAttribsARB to ask for a specific OpenGL version or profile. Enable the three extensions in __glXAquaScreenProbe between the InitExtensionEnableBits call and __glXScreenInit so the extension string picks them up. The direct-rendering path (XQuartz's AppleDRI mode) does not go through __glXAquaScreenCreateContext -- createcontext.c routes direct requests to __glXdirectContextCreate, and the actual context is built on the client side. The server-side createContext hook stays capped at GL 1.4 for indirect clients by validate_GL_version and does not need changes. This is needed for OpenGL core profile support. Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com> (cherry picked from commit 609e7d9421)	2026-07-03 10:24:13 +02:00
Enrico Weigelt, metux IT consult	e113dab26f	dix: rpcbuf: write empty counted string for NULL str ... x_rpcbuf_write_counted_string_pad() wrote nothing at all when str was NULL. But a counted string is a fixed wire element - a 16-bit length, the bytes, then padding to a 4-byte unit. The matching reader (XKB's _GetCountedString) unconditionally consumes that 16-bit length (an empty string is length 0 + pad = 4 bytes), and XkbSizeCountedString() budgets 4 bytes for NULL too. Omitting the field for NULL therefore produced a short, malformed XkbGetGeometry reply whenever a geometry string field was NULL (label_font, a doodad text/font/logo_name, a colour spec, or a property name/value), desyncing every following field for the reading client. Write an empty counted string (length 0 + padding) for NULL instead, so the writer matches the reader and the size accounting. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 10:17:05 +02:00
Enrico Weigelt, metux IT consult	63a635b7bf	test: add XLIBRE_TEST env var to select which unit test suites run ... run_test_in_child() now checks XLIBRE_TEST (a comma-separated list of suite function names, e.g. "ddc_tests,xkb_test") and skips any suite not named in it. Unset/empty behaves exactly as before (run everything). Handy for iterating on a single new test suite without wading through the rest of test/tests' output every run, and for isolating one suite when an unrelated one in the sequence is flaky (run_test_in_child exit()s the whole binary on the first failing suite, so a single flaky suite ahead of the one you care about currently blocks everything after it). Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-03 10:14:24 +02:00
Enrico Weigelt, metux IT consult	f045ee09ad	glx: zero-init xGLXQueryVersionReply to avoid leaking uninitialised bytes ... The reply struct was declared bare (uninitialised auto storage) and then filled field-by-field (majorVersion/minorVersion only), leaving pad1 and any trailing padding uninitialised. Those bytes were then sent to the client via X_SEND_REPLY_SIMPLE, an information disclosure. Use a designated initialiser so all omitted fields and padding are zeroed; type/sequenceNumber/length are still set by X_SEND_REPLY_SIMPLE afterward. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-02 19:56:59 +02:00
Enrico Weigelt, metux IT consult	9567d5fa8b	panoramiX: make XINERAMA_FOR_EACH_SCREEN_* macros variadic ... The XINERAMA_FOR_EACH_SCREEN_FORWARD / _FORWARD_SKIP0 / _BACKWARD macros took the loop body as a single macro parameter. Any top-level comma in the body was therefore parsed as an argument separator, so a body containing e.g. a designated initialiser (`Foo f = { .a = 1, .b = 2 };`) failed to compile — the reason for the existing "static init would confuse preprocessor" workarounds at the call sites. Pass the body as a variadic argument (`...` / `__VA_ARGS__`) instead. Commas in the body are then preserved, so ordinary designated initialisers (and other comma-containing statements) can be used in the loop body without compound-literal-cast or memset workarounds. Pure enabling change: all existing call sites pass a single brace block, which is unchanged as __VA_ARGS__. Build-verified (Xvfb/Xnest link). Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-02 17:59:46 +02:00
Enrico Weigelt, metux IT consult	6efb6f85de	xf86: drop dead #define termio termios ... The SysV `struct termio` type is used nowhere in the tree — the only `termio` occurrence is the unrelated `#include <termio.h>` in the SVR4/Sun block — so this BSD-only alias mapping `termio` to `termios` under `#ifdef CSRG_BASED` is dead code and can be removed. The rest of the `#ifdef CSRG_BASED` block is left intact: it is a legitimate structural BSD-header block, to be dealt with in a later xf86_OSlib.h restructure. The CSRG_BASED block isn't compiled on Linux; a `-Dwerror=true` Linux build links Xvfb and Xnest cleanly, confirming no accidental damage. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-02 13:23:11 +02:00
Gary T. Giesen	ccd005b2e3	config/udev: guard against NULL subsystem in fallback bus id ... config_udev_get_fallback_bus_id() passes the result of udev_device_get_subsystem() straight into strcmp(). That accessor can return NULL when the device has no subsystem, so strcmp(NULL, "pci") dereferences a NULL pointer and the server crashes with a SIGSEGV. The path is reached for any DRM device with no udev ID_PATH property, where config_udev_odev_setup_attribs() falls back to this function. DisplayLink/evdi virtual cards have no ID_PATH; a udev remove of such a card during teardown can present a parent whose subsystem is already gone (reported NULL), and the unchecked strcmp faults. The same NULL-subsystem-into-strcmp class was fixed for the four sibling callers in this file by commit 429ee86a; config_udev_get_fallback_bus_id() was added later (commit 2f53d1cf) without the guard. The preceding parent==NULL case is already guarded here; extend the same defensiveness to the subsystem lookup. Related to: https://gitlab.freedesktop.org/xorg/xserver/-/issues/1905 Signed-off-by: Gary T. Giesen <ggiesen@giesen.me> Part-of: <https://gitlab.freedesktop.org/xorg/xserver/-/merge_requests/2245> (cherry picked from commit 9babe7e7f6)	2026-07-02 13:22:41 +02:00
Enrico Weigelt, metux IT consult	26e24dba36	os/access.c: drop dead <sys/utsname.h> include ... The `#if defined(SVR4) || defined(__GNU__)` include of <sys/utsname.h> is unused: access.c calls neither uname() nor references struct utsname anywhere. Host-name retrieval was long ago factored out into xhostname() (os/xhostname.h), which owns the gethostname()/uname() logic; the include here is a leftover. The explanatory comment above the xhostname() call (mentioning uname()) is kept — it still documents why xhostname is preferred over gethostname(). No feature flag needed — there is nothing to gate. Verified: Linux -Dwerror=true build still links Xvfb and Xnest. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-02 13:21:09 +02:00
Enrico Weigelt, metux IT consult	b9d651f800	os: gate <sys/param.h> on HAVE_SYS_PARAM_H, not CSRG_BASED ... Whether <sys/param.h> is available is a header question, not an OS-family question. Both os/connection.c and os/access.c included it under `#ifdef CSRG_BASED` (i.e. "is this a BSD-derived system or Darwin?") — a grab-bag OS macro standing in for the concrete fact that actually matters. Replace it with a proper feature check. Add HAVE_SYS_PARAM_H via cc.has_header('sys/param.h') to conf_data (alongside the other HAVE_SYS_*_H probes) and switch both includes to it. This is behaviour-preserving: - On the BSD/Darwin systems where CSRG_BASED was set, sys/param.h exists, so the header is still included. - In os/access.c the inner `#if (BSD >= 199103)` guard is unchanged: BSD is only defined by sys/param.h itself (on real BSD), so on other platforms it stays undefined and VARIABLE_IFREQ is not defined — exactly as before. - On platforms that always have sys/param.h (e.g. glibc/Linux) the header is now also included, which is harmless. Verified: Linux build with -Dwerror=true configures and links Xvfb and Xnest cleanly (no MIN/MAX or other redefinition from the widened include). First step of replacing the tree's grab-bag OS-family #ifdefs (CSRG_BASED, SVR4, ...) with concrete, meson-detectable feature flags. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-02 11:42:50 +02:00
Enrico Weigelt, metux IT consult	7103c74da1	xf86: gate consType on CONFIG_BSD_CONSOLE, not CSRG_BASED ... The xf86Info.consType field and the BSD console-driver-type constants (PCCONS/SYSCONS/PCVT/WSCONS/...) are the BSD console machinery consumed by hw/xfree86/os-support/bsd/console_*.c and bsd_VTsw.c. They were gated on the opaque grab-bag macro CSRG_BASED ("Computer Systems Research Group", i.e. BSD-derived) OR'd with __FreeBSD_kernel__. Introduce a self-documenting meson flag CONFIG_BSD_CONSOLE (following the existing CONFIG_WSCONS/CONFIG_UDEV/... convention) set to exactly `csrg_based or host_machine.system() == 'kfreebsd'` — i.e. the same set of platforms the old `CSRG_BASED || __FreeBSD_kernel__` guard matched — and use it at all three sites (struct field, constants, initializer). Behaviour / ABI: - The flag is true on precisely the same platforms as before, so the xf86InfoRec layout is unchanged everywhere. On Linux (and every non-BSD target) CONFIG_BSD_CONSOLE is undefined just as CSRG_BASED was, so consType stays absent and the struct offsets — hence the driver ABI, incl. the nvidia blob — are untouched. - Fixes a latent inconsistency on GNU/kFreeBSD: the struct field was present there (via __FreeBSD_kernel__) but xf86Globals.c initialised .consType only under #ifdef CSRG_BASED, so it defaulted to 0 instead of -1. Both are now driven by the single CONFIG_BSD_CONSOLE flag, so field presence and initialiser always agree. Verified: Linux -Dwerror=true build configures (dix-config.h gets `#undef CONFIG_BSD_CONSOLE`) and links Xorg cleanly. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-02 11:42:00 +02:00
Enrico Weigelt, metux IT consult	657b337e71	xf86: drop dead HAS_SETPRIORITY / <sys/resource.h> block ... xf86Helper.c defined HAS_SETPRIORITY and included <sys/resource.h> under `#if defined(CSRG_BASED) || defined(__GNU__)`, commented "For xf86GetClocks". But nothing in the tree uses any of it: - `setpriority()` is called nowhere, - the HAS_SETPRIORITY macro is referenced nowhere, - xf86GetClocks no longer touches process priority at all. It is a leftover from when xf86GetClocks temporarily raised its priority while measuring dot clocks; that code is long gone. Remove the orphaned block rather than convert its OS-family guard to a feature flag — there is no feature left to gate. Verified: Linux build with -Dwerror=true compiles xf86Helper.c and links Xorg cleanly. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-02 11:39:27 +02:00
Enrico Weigelt, metux IT consult	812c0238ad	ci: bump go-x11proto (go-xts) pin v0.0.4 -> v0.0.7 ... The go-xts CI suite is built against the go-x11proto client library, pinned via PKG_GOXPROTO_REF/GOXPROTO_REF. Raise it from v0.0.4 to the current release v0.0.7 (fresh tag pushed upstream). Both pin sites must move together: install-prereq.sh consumes PKG_GOXPROTO_REF (conf.sh), which defaults to $GOXPROTO_REF from the workflow env. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 20:12:07 +02:00
stefan11111	3f5d7f1a87	kdrive/fbdev: From option -noaccel ... It does the same as `-dumb` now. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:48 +02:00
stefan11111	243d972875	kdrive/fbdev: Modify screen parsing/declaration ... Before this, screens were parsed like so: ``` xinit -- /path/to/Xfbdev -screen <geom> -<screen_0_args> \ -screen <geom> -<screen_1_args> ... \ -screen <geom> -<screen_n_args> ``` This somewhat worked, but it was against the general kdrive design. Notably, `-dumb` and `-origin` (and probably others) didn't work right. This patch changes the screen parsing like so: ``` xinit -- /path/to/Xfbdev -<screen_0_args> -screem <geom> \ -<screen_1_args> -screen <geom> ... \ -<screen_n_args> -screen <geom> -<screen_n_args> ``` This fixes the above issues. This also e.g. makes: `xinit -- /path/to/Xfbdev -fb /dev/fb0 -screen 1920x1080` do the more "intuitive" thing of declaring one screen, instead of what it currently does, which is declare 2 screens, one implicit, one explicit. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:48 +02:00
stefan11111	aae5c0f428	kdrive: Move the KdDPMS call up in KdEnableScreen ... Make it match the reverse of `KdDisableScreen` Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:31 +02:00
stefan11111	44c2356df6	kdrive: Only change the dpms state if the screen is enabled. ... Keeps the same behavior as xf86 Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:31 +02:00
stefan11111	408de3f941	kdrive: Call the card's dpms function regardless of the current dpms state. ... xf86 has no such checks for the dpms state Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:31 +02:00
stefan11111	3b8f8e8272	kdrive: Add support for the DPMS extension ... Add the dpms related code was already there for other purposes. This just exposes it to the dpms extension code. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:31 +02:00
stefan11111	b054fad8a4	kdrive: Add a helper for dpms and related state tracking ... Before this, dpms state tracking had to be done on every call, and Xfbdev needed to duplicate the state tracking. With this, all state tracking is done by the helper, and kdrive X server don't need to do any state tracking. Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:31 +02:00
stefan11111	e29ad4cc1a	kdrive.h: Add some comments ... Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 19:41:31 +02:00
Enrico Weigelt, metux IT consult	98b538dadc	xfree86: vbe: fix memory leak on VideoModePtr allocation failure ... In VBEGetVBEInfo(), when the calloc() for block->VideoModePtr fails, the already-allocated VbeInfoBlock was leaked on the error path. By that point block->OEMStringPtr has also been strdup()'d, so a bare free(block) would still leak that string. Use VBEFreeVBEInfo() to release the partially-built block together with all strings it already owns (the fields not yet set are NULL from the initial calloc(), so freeing them is safe). Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 19:41:22 +02:00
Enrico Weigelt, metux IT consult	1e51bbbac5	Xext/xf86bigfont: drop redundant SHM page-rounding and pagesize ... shmalloc() rounded the SHM segment size up to a page boundary before shmget() (`size = (size + pagesize - 1) & -pagesize`). That is redundant: SysV shmget() already rounds the segment up to a page, and the rounded size is never sent to the client (it receives the shmid and reads the real glyph-metrics size). So the rounding — and the `pagesize` global computed solely for it via SHMLBA / sysconf(_SC_PAGESIZE) (the only CSRG_BASED user in this file) — served no purpose. pagesize's only other role was the `pagesize == 0` sentinel meaning "the SHM syscall probe failed" on MUST_CHECK_FOR_SHM_SYSCALL platforms. Replace it with a dedicated `bool shmSupported` (stdbool.h), set once CheckForShmSyscall() succeeds. This also drops the `#if !defined(CSRG_BASED) && !defined(__CYGWIN__)` block. No behavioural change: shmget() allocates the same page-rounded segment regardless. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 18:51:50 +02:00
Enrico Weigelt, metux IT consult	c7843a9d39	dix: guard ARRAY_SIZE against a platform-provided definition ... include/dix.h defined ARRAY_SIZE() unconditionally. On platforms whose system headers already provide it — notably illumos/Solaris <sys/sysmacros.h> — that is a redefinition, fatal under -Werror ("ARRAY_SIZE redefined"). Wrap it in #ifndef, matching what Xext/dri2/pci_ids/pci_id_driver_map.h already does, so any translation unit that pulls in a system ARRAY_SIZE before dix.h keeps building. (Robustness fix: the concrete xf86bigfont Solaris break that exposed this is separately fixed by dropping its <sys/sysmacros.h> include, but dix.h should not be fragile for the next such file.) Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 18:25:14 +02:00
Enrico Weigelt, metux IT consult	5c9ba0bd89	.github: build xf86bigfont on the CI lanes, keep it off by default ... PR #3202 both fixed the extension's cross-platform build *and* flipped the meson default on. Keep the compile fixes, but restore the meson default to off: xf86bigfont is a low-relevance legacy optimisation (server-side core fonts; modern toolkits render client-side) that was deliberately off, and re-enabling it for all downstreams isn't wanted. Instead pass -Dxf86bigfont=true on the CI build lanes, so the extension stays compiled and validated across all platforms in CI without changing the shipped default. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 17:50:11 +02:00
Enrico Weigelt, metux IT consult	43143d82e9	Xext/xf86bigfont: guard the Unix-only bits for mingw (Windows) ... Building the extension on mingw (Windows, no SysV SHM → CONFIG_MITSHM off) failed on two counts: - QueryVersion filled the reply's uid/gid via geteuid()/getegid(), which Windows has no equivalent for (implicit-declaration error). Those fields only matter for the CONFIG_MITSHM local-shm path (off on Windows), so guard them (`#if !defined(WIN32) || defined(__CYGWIN__)` — Cygwin does have them); on native Windows uid/gid stay 0. - stuff_flags is only consumed under CONFIG_MITSHM, so with SHM off it was set-but-unused (-Werror=unused-but-set-variable). Scope its declaration and assignment under CONFIG_MITSHM too, matching how pDesc is already handled in this function. Verified the Linux (CONFIG_MITSHM, non-WIN32) path is unchanged and builds clean; the Windows path is validated by the mingw32 CI lane. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 16:55:08 +02:00
Enrico Weigelt, metux IT consult	c504e16903	Xext/xf86bigfont: drop the Linux-only <sys/sysmacros.h> include ... xf86bigfont.c unconditionally included <sys/sysmacros.h> inside the CONFIG_MITSHM block. That header is glibc/Linux-specific (major()/minor()/ makedev()) and does not exist on the BSDs or macOS, so building the extension there fails: xf86bigfont.c:49:10: fatal error: 'sys/sysmacros.h' file not found None of those macros are used in this file, so the include is simply removed. Fixes the compile break on macOS, FreeBSD, DragonFly, NetBSD and OpenBSD (surfaced once the extension was built in CI). Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 16:55:08 +02:00
Enrico Weigelt, metux IT consult	d7f258b3bf	meson: build the XF86-BigFont extension by default ... xf86bigfont was option(value: false), so nothing — not a single CI lane — ever compiled Xext/xf86bigfont/. Flip the default to true so the extension is part of the standard build and gains CI coverage on every platform. When enabled it pulls in xf86bigfontproto (>= 1.2.0, with the xorgproto subproject fallback) as a build dependency. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 16:55:08 +02:00
Enrico Weigelt, metux IT consult	9c4a3a9301	kdrive/ephyr: fix NULL-deref crash on lost host-display connection during color setup ... hostx_init() calls xcb_lookup_color_reply() and xcb_alloc_color_reply() without checking for a NULL reply, then immediately dereferences it. If the connection to the host X display dies in between (e.g. a transport error on the socket Xephyr just connected through), both functions return NULL and Xephyr crashes with a SIGSEGV instead of failing cleanly, the same way the adjacent xcb_connect() failure a few lines above already does. Found while investigating a CI segfault in the Xephyr -glamor test that traced back to exactly this NULL deref (offset 0x8 into a NULL xcb_lookup_color_reply_t), exposed by a local-transport listener race under concurrent test load. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 16:37:12 +02:00
Enrico Weigelt, metux IT consult	9049e6ca59	os: drop the never-defined XTHREADS_NEEDS_BYNAMEPARAMS blocks ... XTHREADS_NEEDS_BYNAMEPARAMS is defined nowhere in the tree — not by the meson build, not by any header. Its six `#ifdef XTHREADS_NEEDS_BYNAMEPARAMS` blocks (in access.c, Xtranssock.c, utils.c, xdmcp.c) each only declared a _Xget{host,serv}bynameparams buffer for the old reentrant _XGethostbyname / _XGetservbyname wrappers. The declaration was therefore never compiled, and the buffer is only referenced inside the equally-uncompiled pre-getaddrinfo gethostbyname fallbacks. Pure dead code; remove it. Signed-off-by: Enrico Weigelt, metux IT consult <info@metux.net>	2026-07-01 15:50:50 +02:00
stefan11111	5c55dbc6fb	glamor/glamor_egl: Fix build without libdrm or dri3 ... We were missing some `ifdef`s Signed-off-by: stefan11111 <stefan11111github@gmail.com>	2026-07-01 14:00:42 +02:00