While we probably don't need to be signal safe here since we will never
return to the normal context, the logging signal context check will
cause unsafe logging to be unhandled. Using signal safe logging here
resolves the issue.
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Also, print out the offending message format. This will hopefully help
developers track down unsafe logging.
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Backtraces are often printed in signal context, such as when a segfault
occurs.
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
os: print offset as unsigned int, not long unsigned int
pnprintf() takes unsigned int for %u
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
ErrorF() is not signal safe. Use ErrorSigSafe() whenever an error
message may be logged in signal context.
[whot: edited to "ErrorFSigSafe"]
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
[whot: edited to use varargs, squashed commit below]
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
os: fix vararg length calculation
Make %u and %x sizeof(unsigned int), %p sizeof(void*). This is printf
behaviour and we can't guarantee that void* is uint64_t anyway.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Chase Douglas <chase.douglas@canonical.com>
The Solaris linker recently added a -z parent flag for easier checking
of symbol definitions in plugins against the program that loads them.
If that's present, this enables it, along with -z defs to error on
undefined symbols to alert us if any modules call symbols that won't
be found at runtime.
This builds upon, and requires, the recent Cygwin work to build Xorg.
It moves a couple more modules to be after the Xorg binary in the build
order so that they can find the binary to check against, much as the
Cygwin changes did (these would be modules built on Solaris but not
Cygwin).
v2: This version only sets the flags for the xorg-server build itself,
and does not yet export them in xorg-server.pc to the drivers, since
most of the drivers are not ready to build with -no-undefined yet.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
No other Xfont consumer used it, and this saves us from having to link
callers against libXfont for one simple function when doing
-no-undefined symbols builds.
The function is given a new name to avoid clashing with existing libXfont
binaries, but a #define is provided to preserve the API so we don't have
to fix all the callers at the same time.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
ProcRRGetScreenSizeRange uses REQUEST(xRRGetScreenSizeRangeReq) followed by
REQUEST_SIZE_MATCH(xRRGetScreenInfoReq). This happens to work out because both
requests have the same size, so this is not a functional change, just a cosmetic
one.
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
This code existed in 3 different forms, perhaps it should be
consolidated.
Reviewed-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
The legacy logic was embarassingly wrong; AuthMagic should return errno,
so returning FALSE only when AuthMagic returns nonzero is exactly wrong.
v2: Match drmAuthMagic by returning -EINVAL rather than EINVAL
Fix trailing whitespace
https://bugs.freedesktop.org/show_bug.cgi?id=51400
Signed-off-by: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Tested-by: Chris Wilson <chris@chris-wilson.co.uk>
Reviewed-by: Michel Dänzer <michel.daenzer@amd.com>
Tested-by: Knut Petersen <knut_petersen@t-online.de>
Signed-off-by: Keith Packard <keithp@keithp.com>
Ricardo Salveti <ricardo.salveti@linaro.org> found one place where the
randr code could use the randr screen private data without checking
for null first. This happens when the X server is running with
multiple screens, some of which are randr enabled and some of which
are not. Applications making protocol requests to the non-randr
screens can cause segfaults where the server touches the unset private
structure.
I audited the code and found two more possible problem spots; the
trick to auditing for this issue was to look for functions not taking
a RandR data structure and where there was no null screen private
check above them in the call graph.
Signed-off-by: Keith Packard <keithp@keithp.com>
Fix a seg fault in case pScrPriv is NULL at ProcRRGetScreenInfo,
which later calls RRFirstOutput.
Signed-off-by: Ricardo Salveti de Araujo <ricardo.salveti@linaro.org>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Link XWIN with GLX_SYS_LIBS, just like all the other DDXs
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>
Drivers call xf86InstallSIGIOHandler() for their fd on DEVICE_ON. That
function does not actually enable the signal if it was blocked to begin
with. As a result, if one vt-switches away from the server (SIGIO is
blocked) and then triggers a server regeneration, the signal remains
blocked and input devices are dead.
Avoid this by always unblocking SIGIO when we start the server.
X.Org Bug 50957 <http://bugs.freedesktop.org/show_bug.cgi?id=50957>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
set but not used variables
shadowing a previous local
A hidden problem was that the VERIFY_RR_* macros define local 'rc'
variables, any other local definitions for those would be shadowed and
generate warnings from gcc. I've renamed the other locals 'ret'
instead of 'rc'.
Signed-off-by: Keith Packard <keithp@keithp.com>
Why sending the number of the (implementation-dependent) error statement to
the client is a good idea is a bit beyond me, but at least document it so we
can all share the despair.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Daniel Stone <daniel@fooishbar.org>
libxservertest needs -lpthread from glxapi.c's pthread_once() call. Usually
this would be pulled in by the XORG_LIBS but not when building without Xorg.
This commit has no visible effect on the current tree, preparation for test
cleanups.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Much easier for scripts that try to read the display value off the file
descriptor. Plus, this restores the behaviour we had for this patch in
Fedora since server 1.6 (April 2009).
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Chase Douglas <chase.douglas@canonical.com>
Signed-off-by: Sergei Trofimovich <slyfox@gentoo.org>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Fix glxWinCreateContext() function signature to align with commit 96d74138
"glx: Extend __GLXscreen::createContext to take attributes", which added more
parameters to the screen createContext function for implementing GLX_ARB_create_context
indirect.c: In function 'glxWinScreenProbe':
indirect.c:683:36: warning: assignment from incompatible pointer type
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>
Fix the code generator for the dispatch table initialization after
the whitespace/coding style changes to glx/dispatch.h
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>
If the native GL renderer is the GDI generic renderer (as can happen
if we are in safe mode, or the video driver is VGA, or we have hybrid
graphics which hasn't noticed that xwin requires 3d acceleration), don't
use it. It's not accelerated and we will probably get better conformance
and perfomance from swrast.
Fix so we don't install screen function wrappers in glxWinScreenProbe
unless we are succesful.
Also, move fbConfig dumping to after GLX version has been determined
from extensions
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>
Create a new dispatch table rather than modifying the existing one
struct _glapi_table is not a complete type after including glapi.h, so we use
glapi_get_dispatch_table_size() to determine it's size (alternatively, we could
include glapitable.h, to complete the type)
This could possibly be written to use _glapi_create_table_from_handle() instead, but
that requires making all the wrapper functions exports
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Colin Harrison <colin.harrison@virgin.net>
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
None of the FILE based functions are signal safe, including fileno(), so
we need to save the file descriptor for when we are in signal context.
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
This will be used for checking for proper logging when in signal
context.
Signed-off-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Missing API update caused Xephyr to crash on startup, garbage memory
dereference when accessing timeout.
Introduced in 1f0e8bd5eb
kdrive.c:868:27: warning: assignment from incompatible pointer type
kdrive.c:869:28: warning: assignment from incompatible pointer type
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Now that FatalError is marked as _X_NORETURN, the compilers know we
can't get here, and the return statement added to make them happy in
the past now makes them unhappy.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
xwayland drivers need access to their screen private data to authenticate.
Now that drivers no longer have direct access to the global screen arrays,
this needs to be passed in as function context.
v2: Don't break ABI
v3: Paint the bikeshed blue; drop fd from AuthMagic2ProcPtr prototype
Signed-off-by: Christopher James Halse Rogers <christopher.halse.rogers@canonical.com>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
Call UpdateCurrentTimeIf(), not UpdateCurrentTime(), from RRTellChanged().
The latter calls ProcessInputEvents(), which can trigger a recursion
into mieqProcessInputEvents(). The former omits the call to
ProcessInputEvents().
Signed-off-by: Andy Ritger <aritger@nvidia.com>
Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Too much scrolling down may eventually trigger an overflow of the valuator.
If this happens, reset the valuator to 0 and skip this event for button
emulation. Clients will have to figure out a way to deal with this, but a
scroll event from (close to) INT_MAX to 0 is a hint of that it needs to be
ignored.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Chase Douglas <chase.douglas@canonical.com>
Rob Clark got an ARM tinderbox up and running, and this code is built there
but not here, this should fix it, though I hope that code never gets executed.
Signed-off-by: Dave Airlie <airlied@redhat.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Removed flags argument to match api rework in commit 1f0e8bd5eb
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Dave Airlie <airlied@redhat.com>
f3410b97cf introduced a regression on server
shutdown. If any button or key was held on shutdown (ctrl, alt, backspace
are usually still down) sending a raw event will segfault the server. The
the root windows are set to NULL before calling CloseDownDevices().
Avoid this by disabling all devices first when shutting down. Disabled
devices won't send events anymore.
Master keyboards must be disabled first, otherwise disabling the pointer
will trigger DisableDevice(keyboard) and the keyboard is removed from the
inputInfo.devices list and moved to inputInfo.off_devices. A regular loop
through inputInfo.devices would thus jump to off_devices and not recover.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Acked-by: Chase Douglas <chase.douglas@canonical.com>
Reviewed-by: Chase Douglas <chase.douglas@canonical.com>
Devices are unpaired as needed on DisableDevice now.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Chase Douglas <chase.douglas@canonical.com>