This reverts commit d0339a5c66.
seriously, what the fuck? Are we making xstrdup() return a const char now too?
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
These are generated in code which uses sprintf as a convenient way to
construct strings from various pieces.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Solaris <sys/errno.h> has:
#define EWOULDBLOCK EAGAIN
so checking (errno == EAGAIN || errno == EWOULDBLOCK) is overkill.
This leads cppcheck 1.62 to complain:
[xserver/os/io.c:365] -> [xserver/os/io.c:365]: (style) Same expression on both sides of '||'.
[xserver/os/io.c:941] -> [xserver/os/io.c:941]: (style) Same expression on both sides of '||'.
This quiets it, and reduces the number of calls Solaris Studio cc
generates to the __errno() function to get the thread-specific errno value.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Keith Packard <keithp@keithp.com>
We call atoi() on the server's display to get the socket but otherwise use the
unmodified display for log file name, xkb paths, etc. This results in
Xorg :banana being the equivalent of Xorg :0, except for the log files being
in /var/log/Xorg.banana.log. I'm not sure there's a good use-case for this
behaviour.
Check the display for something that looks reasonable, i.e. digits only, but
do allow for :0.0 (i.e. digits, followed by a period, followed by one or two
digits).
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
This just removes the comment markers from around the formals in
several function prototypes near where pointer -> void * changes were
made. There are plenty more of these to fix.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
This lets us stop using the 'pointer' typedef in Xdefs.h as 'pointer'
is used throughout the X server for other things, and having duplicate
names generates compiler warnings.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Eric Anholt <eric@anholt.net>
Applications may end up allocating a bunch of shmfence objects, each
of which uses a file descriptor, which must be kept open lest some
other client ask for a copy of it later on.
Lacking an API that can turn a memory mapping back into a file
descriptor, about the best we can do is push the file descriptors out
of the way of other X clients so that we don't run out of the ability
to accept new connections.
This uses fcntl F_GETFD to push the FD up above MAXCLIENTS.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Julien Cristau <jcristau@debian.org>
The former doesn't exist on BSD and the latter is available everywhere
AFAIK (checked Solaris and Linux).
You also might want to wrap that line ;).
Reported-by: Mark Kettenis <mark.kettenis@xs4all.nl>
Signed-off-by: Keith Packard <keithp@keithp.com>
req_fds and SetReqFds in include/dixstruct.h
ReadFdFromClient, WriteFdToClient and the FD flushing in os/io.c
Signed-off-by: Keith Packard <keithp@keithp.com>
If a client passes a section of memory via file descriptor and then
subsequently truncates that file, the underlying pages will be freed
and the addresses invalidated. Subsequent accesses to the page will
fail with a SIGBUS error.
Trap that SIGBUS, figure out which segment was causing the error and
then allocate new pages to fill in for that region. Mark the offending
shared segment as invalid and free the resource ID so that the client
will be able to tell when subsequently attempting to use the segment.
Signed-off-by: Keith Packard <keithp@keithp.com>
v2: Use MAP_FIXED to simplify the recovery logic (Mark Kettenis)
v3: Also catch errors in ShmCreateSegment
Conflicts:
include/dix-config.h.in
include/xorg-config.h.in
The selection of which clock to use for this function was not actually
getting used when fetching the final clock value.
Reported-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Keith Packard <keithp@keithp.com>
This passes a file descriptor from the client to the server, which is
then mmap'd
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
This adds two interfaces:
void SetReqFds(ClientPtr client, int req_fds)
Marks the number of file descriptors expected for this
request. Call this before any request processing so that
any un-retrieved file descriptors will be closed
automatically.
int ReadFdFromClient(ClientPtr client)
Reads the next queued file descriptor from the connection. If
this request is not expecting any more file descriptors, or
if there are no more file descriptors available from the
connection, then this will return -1.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Adam Jackson <ajax@redhat.com>
And now that we have the accessors, localize it. No functional changes, just
preparing for a future change.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Adam Jackson <ajax@redhat.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
It is needed in IPv6 configurations (for inet_pton) also when
SIOCGIFCONF is not defined.
Signed-off-by: Pino Toscano <toscano.pino@tiscali.it>
Acked-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Signed-off-by: Julien Cristau <jcristau@debian.org>
It's already not optional at configure time, this just makes it so at
build time too.
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Adam Jackson <ajax@redhat.com>
If we immediately put the WriteToClient() buffer into the socket's write
queue, not only do we benefit from sending the response back to client
earlier, but we also avoid the overhead of copying the data into our own
staging buffer and causing extra work in the next select(). The write is
effectively free as typically we may only send one reply per client per
select() call, so the cost of the FlushClient() is the same.
shmget10: 26400 -> 110000
getimage10: 25000 -> 108000
shmget500: 3160 -> 13500
getimage500: 1000 -> 1010
The knock-on effect is that on a mostly idle composited desktop, the CPU
overhead is dominated by the memmove in WriteToClient, which is in turn
eliminated by this patch.
Reviewed-by: Adam Jackson <ajax@redhat.com>
Signed-off-by: Chris Wilson <chris@chris-wilson.co.uk>
If we are not backing up logfiles, remove the old logfile before trying to write
a new logfile, as otherwise the operation may fail if the previous logfile was
created by a different user.
This change is useful when:
- The DDX doesn't use the logfile backup mechanism (i.e. not Xorg)
- The DDX is run by a non-root user, and then by a different non-root user
- The logfile directory doesn't have the restricted-deletion flag set
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Acked-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
If a client sends a request larger than maxBigRequestSize, the server is
supposed to ignore it.
Before commit cf88363d, the server would simply disconnect the client. After
that commit, it attempts to gracefully ignore the request by remembering how
long the client specified the request to be, and ignoring that many bytes.
However, if a client sends a BigReq header with a large size and disconnects
before actually sending the rest of the specified request, the server will
reuse the ConnectionInput buffer without resetting the ignoreBytes field. This
makes the server ignore new X clients' requests.
This fixes that behavior by resetting the ignoreBytes field when putting the
ConnectionInput buffer back on the FreeInputs list.
Signed-off-by: Robert Morell <rmorell@nvidia.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Backtrace logging etc. is already sigsafe, but the actual FatalError message
in response is not yet, leading to amusing logs like this:
(EE) Segmentation fault at address 0x0
(EE) BUG: triggered 'if (inSignalContext)'
(EE) BUG: log.c:499 in LogVMessageVerb()
(EE) Warning: attempting to log data in a signal unsafe manner while in
signal context.
Please update to check inSignalContext and/or use LogMessageVerbSigSafe() or
ErrorFSigSafe().
The offending log format message is:
Fatal server error:
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Mainly for %ld, smaller than int is propagated anyway, and %lld isn't really
used.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Format strings with length modifiers but missing format specifier like "%0"
will read one byte past the array size.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
If we're about to abort, we're already in the signal handler and cannot call
down to the default device cleanup routines (which reset, free, alloc, and
do a bunch of other things).
Add a new DEVICE_ABORT mode to signal a driver's DeviceProc that it must
reset the hardware if needed but do nothing else. An actual HW reset is only
required for some drivers dealing with the HW directly.
This is largely backwards-compatible, hence the input ABI minor bump only.
Drivers we care about either return BadValue on a mode that's not
DEVICE_{INIT|ON|OFF|CLOSE} or print an error and return BadValue. Exception
here is vmmouse, which currently ignores it and would not reset anything.
This should be fixed if the reset is required.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
Truncating the fraction part leads to a test failure where -1203.30 is
printed as -1203.29. Round this to the nearest value instead by adding
0.5 before converting to an integer
Signed-off-by: Keith Packard <keithp@keithp.com>
This is the lazy man's %f support. Print the decimal part of the number,
then append a decimal point, then print the first two digits of the
fractional part. So %f in sigsafe printing is really %.2f.
No boundary checks in place here.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Until we have support for them, ignore any length modifiers so we don't need
to update all callers.
Signed-off-by: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
The formatter confused address operators preceded by casts with
bitwise-and expressions, placing spaces on either side of both.
That syntax isn't used by ordinary address operators, however,
so fix them for consistency.
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
setitimer() and SIGALRM aren't available on WIN32, so smart scheduler
code cannot be built. Provide only stubs for smart scheduler timer
code, and disable smart scheduler by default.
Signed-off-by: Ryan Pavlik <rpavlik@iastate.edu>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
Fix compilation of OsBlockSIGIO with -Werror=return-type when SIGIO isn't
defined.
/jhbuild/checkout/xorg/xserver/os/utils.c: In function 'OsBlockSIGIO':
/jhbuild/checkout/xorg/xserver/os/utils.c:1248:1: error: control reaches end of non-void function [-Wreturn-type]
v2: Shuffle around to avoid writing unreachable code
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
MinGW doesn't have sigaction, so this patch is needed for building.
No attempt is made to actually install the fatal error signal handler, as MinGW
will simply terminate the process rather than deliver a fatal signal.
Also avoid using strsignal
Signed-off-by: Ryan Pavlik <rpavlik@iastate.edu>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Tested-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Keith Packard <keithp@keithp.com>
libnettle is smaller than libgcrypt, currently being released more
frequently, and has replaced the latter in gnutls-3.x (which is used
by TigerVNC, so they can avoid pulling in two crypto libraries
simultaneously).
Signed-off-by: Yaakov Selkowitz <yselkowitz@users.sourceforge.net>
Reviewed-by: Julien Cristau <jcristau@debian.org>
