This will allow a server to disable listeners by default and then
let later configuration re-enable them. In particular, this lets the X
server disable inet and unix listen sockets by default while still
providing a '-listen' command line option to re-enable them later on.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Required also changing receive_listening_fds to specify port as const,
which we can do now that TRANS(ReopenCOTSServer) takes it as const.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Also required constifying UnixHostReallyLocal, since SocketUNIXConnect
passes the host arg through to it.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Required also adding const to static TRANS(ParseAddress) function which
they pass the address arg to for parsing.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Some systems provide a really small default buffer size for UNIX sockets.
Bump it up to 64k if necessary such that large transfers (such as
XGetImage() on a 8-megapixel image) don't take tens of seconds.
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
The latest glibc considers _BSD_SOURCE deprecated, leading to the following
warning being issued for pretty much every C-file in the xserver:
In file included from /usr/include/stdint.h:25:0,
from /usr/lib/gcc/x86_64-redhat-linux/4.8.2/include/stdint.h:9,
from ../include/misc.h:81,
from miexpose.c:82:
/usr/include/features.h:145:3: warning: #warning "_BSD_SOURCE and _SVID_SOURCE are deprecated, use _DEFAULT_SOURCE" [-Wcpp]
# warning "_BSD_SOURCE and _SVID_SOURCE are deprecated, use _DEFAULT_SOURCE"
^
I've discussed this with the glibc developers and the prefered way of fixing
this is by also defining _DEFAULT_SOURCE which is the new way of stating
_BSD_SOURCE / _SVID_SOURCE .
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Mark Kettenis <kettenis@openbsd.org>
Tested-by: Kristian Høgsberg <krh@bitplanet.net>
Reviewed-by: Kristian Høgsberg <krh@bitplanet.net>
libxtrans provides TransNoListen() to set the 'don't listen' flag for a
particular transport, but there is no interface to query the state of that flag
This is a bit of a problem for the XWin server, as it wants to start some helper
clients (for clipboard integration and integrated window management), so needs
to know what transports the server is listening on to construct appropriate
display names for those clients.
Add TransIsListening() to discover if TransNoListen() has been called for a
particular protocol or not
HdG:
-Invert the final check so that TransIsListening returns True when
TRANS_NOLISTEN is not set, as one would expect of it.
-Make the protocol argument a const char * as similar functions do
-Fix "warning: too many arguments for format" warning
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Assume signal handlers return void, as C89 requires
Drops use of autoconf's obsolete AC_TYPE_SIGNAL and Imake's even more
obsolete SIGNALRETURNSINT.
None of the modules including xtrans.m4 uses RETSIGTYPE from autoconf.
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
A char array on the stack is not guaranteed to have more than byte
alignment. This means that casting it to a 'struct cmsghdr' and
accessing its members may result in unaligned access. This will
generate SIGBUS on strict alignment architectures like OpenBSD/sparc64.
The solution is to use a union to force proper alignment.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Matthieu Herrb <matthieu@herrb.eu>
Required to expose the structure members in Solaris headers, since it
was an XPG4/UNIX95 addition to the Solaris ABI.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Leave it up to the consumer to request this functionality by defining
XTRANS_SEND_FDS.
Signed-off-by: Mark Kettenis <kettenis@openbsd.org>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Keith Packard <keithp@keithp.com>
v2 - make sure this is only defined on systems where the code actually
works (Linux for now)
Now that we've found that libFS uses readv, we can test whether this
readv implementation works correctly.
Signed-off-by: Keith Packard <keithp@keithp.com>
Has never been converted to build in modular builds, so has been unusable
since X11R7.0 release in 2005. All known platforms with TLI/XTI support
that X11R7 & later releases run on also have (and mostly prefer) BSD
socket support for their networking API.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
On certain tables, add top and bottom borders to table
header and a bottom border to the table. This matches
what those tables in the old pdfs looked like.
the <?dbfo keep-together='always'> prevents tables from
splitting across pages. Useful for tiny tables.
Converting the colwidth to a floating point,
IE, 1* -> 1.0* cleans up these build errors:
WARNING: table-layout="fixed" and column-width unspecified
=> falling back to proportional-column-width(1)
Signed-off-by: Matt Dew <marcoz@osource.org>
This function was constructing an address from a port string allocating
a buffer according to the size of the string but then later copying
the address according to sizeof(struct sockaddr).
This patch ensures that we allocate a struct sockaddr buffer with enough
space for the port string to be copied into sa_data[] and uses that
combined length to determine how much should be copied at the end of the
function.
This fixes a crash when using xwayland which uses ListenOnOpenFD() that
will call _XSERVTransReopenCOTSServer() with a short port string like
":1".
Signed-off-by: Robert Bragg <robert@linux.intel.com>
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
