Without this patch we end up with invalid C code if we've a
<pad align="n" /> between two variadic lists. Check for such a condition
and take the alignment pad into account.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=79808
Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>
This changes away from hard-coding the /tmp/launch-* path to now
supporting a generic <path to unix socket>[.<screen>] format for
$DISPLAY.
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
send_request may only write to out.queue if no other thread is busy
writing to the network (as that thread may be writing from out.queue).
send_request may only allocate request sequence numbers if XCB owns
the socket.
Therefore, send_request must make sure that both conditions are true
when it holds iolock, which can only be done by looping until both
conditions are true without having dropped the lock waiting for the
second condition.
We choose to get the socket back from Xlib first as get_socket_back
has a complicated test and checking for other threads writing is a
simple in-lined check.
This also changes the sequence number checks (64k requests with no
reply, 4M request wrapping) to ensure that both conditions are true
before queueing the request.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Uli Schlachter <psychon@znc.in>
This avoids having the nested header files also included at the top
level, which is more efficient.
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Uli Schlachter <psychon@znc.in>
Code can be simplified if the deallocation functions can always be called in
cleanup code. So if you have some code that does several things that can go
wrong, one of which is xcb_connect(), after this change, the xcb_connection_t*
variable can be initialized to NULL and xcb_disconnect() can always be called on
the connection object.
References: http://lists.freedesktop.org/archives/xcb/2013-September/008659.html
Signed-off-by: Uli Schlachter <psychon@znc.in>
Reviewed-by: Julien Cristau <jcristau@debian.org>
There are two kind of error connections in XCB. First, if something goes wrong
while the connection is being set up, _xcb_conn_ret_error() is used to return a
static connection in an error state. If something goes wrong later,
_xcb_conn_shutdown() is used to set c->has_error.
This is important, because the static object that _xcb_conn_ret_error() returns
must not be freed, while the dynamically allocated objects that go through
_xcb_conn_shutdown() must obviously be properly deallocated.
This used to work correctly, but in 769acff0da, xcb_disconnect() was made to
ignore all connections in an error state completely. Fix this by only ignoring
the few static error connections that we have.
This was tested with the following hack:
xcb_connection_t *c = xcb_connect(NULL, NULL);
close(xcb_get_file_descriptor(c));
xcb_discard_reply(c, xcb_get_input_focus(c).sequence);
xcb_flush(c);
xcb_disconnect(c);
Valgrind confirms that xcb has a memory leak before this patch that this patch
indeed fixes.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Reviewed-by: Julien Cristau <jcristau@debian.org>
If xcb_connect() fails, it doesn't return NULL. Instead, it always
returns an xcb_connection_t*, and the user should check for errors with
the xcb_connection_has_error() function. What this function does is
check if conn->has_error contains a non-zero error code, and returns it.
If an error did occur, xcb doesn't actually return a full
xcb_connection_t though, it just returns (xcb_connection_t *)
error_code. Since the 'has_error' field is the first, it is still
possible to check conn->has_error.
That last trick was not immediately obvious to me, so add some guiding
comments. This also ensures no one obliviously rearranges the struct.
Signed-off-by: Ran Benita <ran234@gmail.com>
Signed-off-by: Uli Schlachter <psychon@znc.in>
The section number is no longer hard-coded
The left footer is now "X Version 11".
The center footer is the package name with the version, "libxcb 1.9"
The three values above are provided through xorg-macros. They are passed-in
to the python c_client code.
Example of footer (last line, above dotted line)
[...]
AUTHOR
Generated from xproto.xml. Contact xcb@lists.freedesktop.org for cor‐
rections and improvements.
X Version 11 libxcb 1.9 xcb_send_event(3)
------------------------------------------------------------------------------
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
The automake MAN primary requires a hard coded extension to build
man pages. Let's avoid that as the extension number may vary by platform.
Take advantage of the fact that the man directory only contains man pages.
Wildcards are not supported by Automake but it happens to work
sufficiently well here.
Normally xorg build man pages by converting a source .man file to a
target file with the extension number. That would be too many files
in this case.
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
The section number is no longer hard-coded, supplied by xorg-macros.
The left footer is now "X Version 11".
The center footer is the package name with the version, "libxcb 1.9"
The man directory is a sibbling to the doc directory. One can build
or clean the man pages without disturbing the library code.
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
The BASE_CFLAGS variable contains only warnings, just like the XCB
version of CWARNFLAGS. This will result in no changes in the binaries
produced. Xorg was missing -fd for SUNCC so it has been added to util-macros
v 1.18.
Do not get confused with the xorg deprecated CWARNFLAGS variable which
contains an option that is not a warning, -fno-strict-aliasing. This
option, should it be needed, can be added using the XORG_TESTSET_CFLAG
macro.
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Gaetan Nadon <memsize@videotron.ca>
With the advent of the Present extension, some events (such as
PresentCompleteNotify) now use native 64-bit types on the wire.
For XGE events, we insert an extra "uint32_t full_sequence" field
immediately after the first 32 bytes of data. Normally, this causes
the subsequent fields to be shifted over by 4 bytes, and the structure
to grow in size by 4 bytes. Everything works fine.
However, if event contains 64-bit extended fields, this may result in
the compiler adding an extra 4 bytes of padding so that those fields
remain aligned on 64-bit boundaries. This causes the structure to grow
by 8 bytes, not 4. Unfortunately, XCB doesn't realize this, and
always believes that the length only increased by 4. read_packet()
then fails to malloc enough memory to hold the event, and the event
processing code uses the wrong offsets.
To fix this, mark any event structures containing 64-bit extended
fields with __attribute__((__packed__)).
v2: Use any(...) instead of True in (...), as suggested by
Daniel Martin.
v3 (Alan Coopersmith): Fix build with Solaris Studio 12.3 by moving the
attribute to after the structure definition.
Signed-off-by: Kenneth Graunke <kenneth@whitecape.org>
Reviewed-by: Keith Packard <keithp@keithp.com> [v1]
Reviewed-by: Josh Triplett <josh@joshtriplett.org> [v1]
Reviewed-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
As RFC 2292 points out, some platforms (e.g. Darwin 9.8.0) provide
CMSG_FIRSTHDR(msg) which just returns msg.msg_control without first
checking if msg.msg_controllen is non-zero. We need a workaround for
such platforms not to let _xcb_in_read() segfault.
https://bugs.freedesktop.org/show_bug.cgi?id=72253
Signed-off-by: Julien Cristau <jcristau@debian.org>
Sync: Due to commit e6a246e50e62cbcba3 "sync: Change value list param of
CreateAlarm and ChangeAlarm into switch", various symbols disappeared,
for example xcb_sync_{change,create}_alarm_sizeof.
xinput: This extension was updated from version 1.4 to 2.3. This means
that lots of new things are generated. However, this change is
backwards-compatible and thus age gets set to 1.
xkb: In commit 37d0f55392d6 "xkb: Work around alignment problems in
GetNames and GetMap replies", some padding fields were introduced into
structures for which an _unpack() function is generated. This changed
the size of the struct and caused offsets into this struct to change.
https://bugs.freedesktop.org/show_bug.cgi?id=71507
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Cristau <jcristau@debian.org>
This reverts commit f4d5b84800.
The version of this struct that the code generator produces breaks the API,
because it gives the fields different (albeit better) names. Thus, we need to
restore the old version of this struct.
Additionally to the revert, this struct is documented as being deprecated. The
replacement was added to xcb-proto.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=71502
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Cristau <jcristau@debian.org>
Currently, it is not possible to correctly iterate over the replies of
some requests. For example, the list of XIDeviceInfo returned by
the XIQueryDevice request from xinput2 is read as garbage starting from
the second entry.
The culprits are the _sizeof() used by the iterators. In the above case:
int
xcb_input_xi_device_info_sizeof (const void *_buffer /**< */)
{
char *xcb_tmp = (char *)_buffer;
[...]
unsigned int xcb_block_len = 0;
[...]
xcb_block_len += sizeof(xcb_input_xi_device_info_t);
xcb_tmp += xcb_block_len;
/* name */
xcb_block_len += (((_aux->name_len + 3) / 4) * 4) * sizeof(char);
xcb_tmp += xcb_block_len;
[...]
}
The problem here is that `xcb_block_len` is not zero'd right above the
`/* name */` comment, causing `xcb_tmp` to be incremented by
`sizeof(xcb_input_xi_device_info_t)` twice. The returned size is too
large.
https://bugs.freedesktop.org/show_bug.cgi?id=68387
Tested-by: Ran Benita <ran234@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com>
Reviewed-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Ran Benita <ran234@gmail.com>
Signed-off-by: Julien Cristau <jcristau@debian.org>
This reverts commit 9ae84ad187.
After this patch was merged, there were complaints about it not being a good
idea. Revert this for now until we can agree on this.
References: http://lists.freedesktop.org/archives/xcb/2013-June/008340.html
Signed-off-by: Uli Schlachter <psychon@znc.in>
Conflicts:
src/xcbint.h
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 struct
alignment architectures like OpenBSD/sparc64. The canonical 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>
Signed-off-by: Uli Schlachter <psychon@znc.in>
Use these instead of computing the values directly so that it might
work on BSD or other non-Linux systems
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
This allows apps to peel off certain XGE events into separate queues
for custom handling. Designed to support the Present extension
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-By: Uli Schlachter <psychon@znc.in>
Requests signal which replies will have fds, and the replies report
how many fds they expect in byte 1.
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-By: Uli Schlachter <psychon@znc.in>
This uses sendmsg to transmit file descriptors from the application to
the X server
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-By: Uli Schlachter <psychon@znc.in>
When new XML files get installed, make sure the C files are regenerated
Signed-off-by: Keith Packard <keithp@keithp.com>
Reviewed-By: Uli Schlachter <psychon@znc.in>
It seems like POLLIN is specified as equivalent to POLLRDNORM | POLLRDBAND. Some
systems (e.g. QNX and HP-UX) take this literaly and have POLLIN defined as the
above bit combination. Other systems (e.g. Linux) have POLLIN as just a single
bit.
This means that if no out-of-band data is available (which should never be the
case), the result of poll() will not fulfil (fd.revents & POLLIN) == POLLIN on
QNX, because the POLLRDBAND bit is not set.
In other words, even though poll() signaled that the fd is readable, xcb would
not read from the file descriptor.
Fix this by checking if any bits from POLLIN are set in the result of poll(),
instead of all of them.
(This change was independently done by seanb@qnx.com as well)
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=38001
Acked-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Uli Schlachter <psychon@znc.in>
In commit 8eba8690ad, the API documentation for xcb_poll_for_event() was
fixed to remove an argument that was previously removed in commit 34168ab549.
However, that commit only removed the first line of the documentation, leaving
behind a spurious half-sentence. That commit happened seven years ago and now
finally someone noticed...
Thanks to Benjamin Herr for reporting this on IRC.
v2: Thanks again to Benjamin Herr for noticing that my commit message blamed the
wrong commit.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Do not create pointers in unions for fields of variadic length.
Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Ran Benita <ran234@gmail.com>
to get rid of:
warning: 'xcb_align_to' may be used uninitialized in this function
Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Peter Harris <pharris@opentext.com>
Fixes Solaris Studio compiler warning:
"xcb_list.c", line 50: warning: old style function definition
and gcc warning:
xcb_list.c: In function '_xcb_map_new':
xcb_list.c:50:11: warning: old-style function definition [-Wold-style-definition]
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
The generic event structure xcb_ge_event_t has the full_sequence field
at the 32byte boundary. That's why we've to inject this field into GE
events while generating the structure for them. Otherwise we would read
garbage (the internal full_sequence) when accessing normal event fields
there.
Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Reviewed-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Peter Harris <pharris@opentext.com>
To prevent different threads from stealing the socket from each other the
caller of "xcb_take_socket" must hold a lock that is also acquired in
"return_socket". Unfortunately xcb tries to prevent calling return_socket
from multiple threads and this can lead to a deadlock situation.
A simple example:
- X11 has taken the socket
- Thread A has locked the display.
- Thread B does xcb_no_operation() and thus ends up in libX11's return_socket(),
waiting for the display lock.
- Thread A calls e.g. xcb_no_operation(), too, ends up in return_socket() and
because socket_moving == 1, ends up waiting for thread B
=> Deadlock
This patch allows calling return_socket from different threads at the same time
an so resolves the deadlock situation.
Partially fixes: https://bugs.freedesktop.org/show_bug.cgi?id=20708
v2: fixes additional pthread_cond_wait dependencies,
rework comments and patch description
v3: separate pthread_cond_wait dependencies and unrelated whitespace
change into their own patch, use unsigned for socket_seq
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Uli Schlachter <psychon@znc.in>
Ensure that when calculating the size of the incoming response from the
Xserver, we don't overflow the integer used in the calculations when we
multiply the int32_t length by 4 and add it to the default response size.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Adopt a change from xcbgen. With that modification the expression in a
bitcase became a list of expressions to support multiple <enumref> in a
<bitcase>.
Signed-off-by: Daniel Martin <consume.noise@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>
Replace except statement with a PEP-3110 compliant one. This fixes a regression
introduced by c3deeaf714https://bugs.freedesktop.org/show_bug.cgi?id=55690
Reviewed-by: Peter Harris <pharris@opentext.com>
Signed-off-by: Uli Schlachter <psychon@znc.in>
This was found by distcheck. It tried to install src/man/xcb-examples.3 and
src/man/xcb-requests.3, but those files weren't in the distribution.
Fix this by explicitly telling automake to distribute those files.
Signed-off-by: Uli Schlachter <psychon@znc.in>
This fixes a deadlock which was seen in-the-wild with wine.
It could happen that two threads tried to read from the socket at the same time
and one of the thread got stuck inside of poll()/select().
The fix works by making sure that the writing thread doesn't steal the reading
thread's reply.
Debugged-by: Erich Hoover <ehoover@mines.edu>
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=54671
Signed-off-by: Uli Schlachter <psychon@znc.in>
This allows an application to do a scatter/gather operation on a large
image buffer to avoid the extra memcpy.
Use autoconf to use UIO_MAXIOV where IOV_MAX is not available (and the
POSIX minimum of 16 where neither are available).
Reviewed-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Peter Harris <pharris@opentext.com>
Matches the behaviour of Xlib - if you set DISPLAY to :0.1 but only have
one screen, closes connection and returns error.
This introduces a new connection error code:
XCB_CONN_CLOSED_INVALID_SCREEN
Signed-off-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Jeremy Huddleston Sequoia <jeremyhu@apple.com>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Allows configure to set defines such as _POSIX_SOURCE in config.h
that affect functions exposed by system headers and get consistent
results across all the source files.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
On FreeBSD MSG_WAITALL on a non-blocking socket fails immediately if less bytes
than were asked for are available. This is different than the behavior on linux
where as many bytes as are available are returned in this case. Other OS
apparently follow the FreeBSD behavior.
_xcb_in_read() is used to fill xcb's read buffer, thus this function will call
recv() with a big length argument (xcb's read buffer is by default 16 KiB
large). That many bytes are highly unlikely to be available in the kernel
buffer.
This means that _xcb_in_read() always failed on FreeBSD. Since the socket was
still signaled as readable by poll(), this bug even resulted in a busy loop.
The same issue is present in read_block(), but here it is slightly different.
read_block() is called when we read the first few bytes of an event or a reply,
so that we already know its length. This means that we should be able to use
MSG_WAITALL here, because we know how many bytes there have to be.
However, that function could busy loop, too, when only the first few bytes of
the packet were sent while the rest is stuck somewhere on the way to us. Thus,
MSG_WAITALL should be removed here, too.
Thanks to Christoph Egger from Debian for noticing the problem, doing all the
necessary debugging and figuring out what the problem was! This patch is 99%
from debian. Thanks for all the work.
This bug was introduced in commit 2dcf8b025b.
This commit also reverts commit 9061ee45b8.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=45776
Signed-off-by: Uli Schlachter <psychon@znc.in>
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
2dcf8b025b was causing some regressions on
darwin, so go back to using read(2) there until I have time to investigate
further.
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Unfortunately, commit 31b57676 adding WSACleanup/WSAShutdown on Win32 adds a new use
of error_connection, which was removed in commit 769acff0, applied 5 minutes earlier.
src/xcb_util.c: In function 'xcb_connect_to_display_with_auth_info':
src/xcb_util.c:433:39: error: 'error_connection' undeclared (first use in this function)
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Arvind Umrao <arvind.umrao@oracle.com>
Signed-off-by: Uli Schlachter <psychon@znc.in>
Fix a redefinition problem which shows up when building for _WIN32 and
libXdmcp is installed, so HASXDMAUTH is enabled
It seems this is a special place in xcb as it uses other X11 library headers here
If HASXDMAUTH is defined, include the wrapped windows.h before any header which
includes it unwrapped, to avoid conflicts with types defined in X headers
We need to include config.h and check HASXDMAUTH to avoid an unconditional dependency
on x11proto headers
In file included from install/include/X11/Xdmcp.h:19:0,
from git/xcb/libxcb/src/xcb_auth.c:52:
install/include/X11/Xmd.h:120:14: error: conflicting types for 'INT32'
/usr/i686-pc-mingw32/sys-root/mingw/include/basetsd.h:54:13: note: previous declaration of 'INT32' was here
install/include/X11/Xmd.h:143:15: error: conflicting types for 'BOOL'
/usr/i686-pc-mingw32/sys-root/mingw/include/windef.h:234:17: note: previous declaration of 'BOOL' was here
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Jeremy Huddleston <jeremyhu@apple.com>
This reverts commit 0e9246def5.
This change caused build failures because <X11/Xdmcp.h> was never
included under any circumstance. This is because the check for
HASXDMAUTH was moved before the inclusion of config.h (via xcbint.h)
which defined it.
Found-by: Tinderbox
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Reviewed-by: Jon TURNEY <jon.turney@dronecode.org.uk>
The alternative is to use these in every WIN32 application which uses xcb. Doing
it this way should be safe, as, according to MSDN, "There must be a call to
WSACleanup for each successful call to WSAStartup. Only the final WSACleanup
function call performs the actual cleanup. The preceding calls simply decrement
an internal reference count"
(We should probably also include ws2_32 in Libs.private for libxcb, as anything
which links with libxcb will also need that, but there seems to be some pkg-config
issues to resolve first...)
v2: Check for errors so WSAStartup()/WSACleanup() uses are balanced
v3: Use same indentation style as surrounding code
Reviewed-by: Peter Harris <pharris@opentext.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
Fix a redefinition problem due to include order which shows up when
building for _WIN32 and libXdmcp is installed, so HASXDMAUTH is enabled
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Peter Harris <pharris@opentext.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
WIN32 does not have arpa/inet.h, so do not try to include it unless _WIN32 is
not defined
Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>
Reviewed-by: Peter Harris <pharris@opentext.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=41443
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=42304
I have added more xcb connection error states at xcb.h header.
Also I have removed global error_connection variable, and added
an interface that returns connection error state.
TBD:
I will segregate errors states in a separate header file and try to
provide more precise error states, in future. Also I will give patch
for libX11, in that patch xcb_connection_t::has_error will be passed
to default io handler of libX11. This value can then be used for
displaying error messages.
Reviewed-by: Rami Ylimäki <rami.ylimaki@vincit.fi>
Reviewed-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Arvind Umrao <arvind.umrao@oracle.com>
_xcb_out_flush_to will drop the iolock in pthread_cond_wait allowing
other threads to queue new requests. When this happened,
there would be requests queued for the socket after _xcb_out_flush_to
returned, and xcb_take_socket would throw an assert.
Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=29875
Signed-off-by: Keith Packard <keithp@keithp.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
Imagine two threads:
Thread#1: for(;;) { xcb_get_input_focus_reply(c, xcb_get_input_focus(c), 0); }
Thread#2: for(;;) { xcb_poll_for_event(c); }
Since xcb_poll_for_event() calls _xcb_in_read() directly without synchronizing
with any other readers, this causes two threads to end up calling recv() at the
same time. We now have a race because any of these two threads could get read
the GetInputFocus reply.
If thread#2 reads this reply, it will be put in the appropriate queue and
thread#1 will still be stuck in recv(), although its reply was already received.
If no other reply or event causes this thread to wake up, the process deadlocks.
To fix this, we have to make sure that there is only ever one thread reading
from the connection. The obvious solution is to check in poll_for_next_event()
if another thread is already reading (in which case c->in.reading != 0) and not
to read from the wire in this case.
This solution is actually correct if we assume that the other thread is blocked
in poll() which means there isn't any data which can be read. Since we already
checked that there is no event in the queue this means that
poll_for_next_event() didn't find any event to return.
There might be a small race here where the other thread already determined that
there is data to read, but it still has to wait for c->iolock. However, this
means that the next poll_for_next_event() will be able to read the event, so
this shouldn't cause any problems.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=40372
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Peter Harris <pharris@opentext.com>
Uli's patch is an excellent solution; I just want to keep the new
ALIGNOF macro hidden from XCB's users, as they don't need it to call
XCB.
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Some of these systems (eg. Interix on XP) are still in use.
Reviewed-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Peter Harris <pharris@opentext.com>
The code previously assumed that everything has to be aligned to a 4 byte
boundary. This assumption is wrong as e.g. the STR struct from xproto shows.
Instead, each type has to be aligned to its natural alignment. So a char doesn't
need any alignment, a INT16 gets aligned to a 2-byte-boundary and a INT32 gets
the old 4 byte alignment.
I'm not 100% sure that this commit is correct, but some quick tests with awesome
and cairo-xcb went well.
This commit causes lots of dead assignments to xcb_align_to since only the last
field's alignment is actually used, but this simplified this patch a lot.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=34037
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Peter Harris <pharris@opentext.com>
When a system is completely offline (no interface has an IP address but 'lo'),
xcb could not connect to localhost via TCP, e.g. connections with
DISPLAY=127.0.0.1:0 fail.
AI_ADDRCONFIG will only return IPv4 addresses if the system has an IPv4
address configured (likewise for IPv6). This also takes place when
resolving localhost (or 127.0.0.0/8 or ::1). Also, as per RFC 3493,
loopback addresses are not considered as valid addresses when
determining whether to return IPv4 or IPv6 addresses.
As per mailing-list discussion on the xcb list started with message
20110813215405.5818a0c1@x200, the AI_ADDRCONFIG flag is there for historical
reasons:
In the old days, the "default on-link" assumption in IPv6 made the flag vey
much indispensable for dual-stack hosts on IPv4-only networks. Without it,
there would be long timeouts trying non-existent IPv6 connectivity. Nowadays,
this assumption has been flagged as historic bad practice by IETF, and hosts
should have been updated to not make it anymore.
Then AI_ADDRCONFIG became mostly cosmetic: it avoids phony "Protocol family
not supported" or "Host unreachable" errors while trying to connect to a dual-
stack mode from a host with no support for source address selection.
Nowadays, on up-to-date systems, this flag is completely useless. Then again,
I understood only the very latest MacOS release is "up-to-date" with this
definition.
If a the path to the xcb python generate libs is
explicitly specified to c_client.py, insert it in
the python path list just after the local dir entry,
rather than appending it to the existing paths.
This keeps a global/distro install of xcb from
overriding a local build of the xcb proto files.
Signed-off-by: James Jones <jajones@nvidia.com>
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Python 3 introduces some language changes that cause issues when running
c_client.py. This also breaks compatibility with Python 2.5 since it does not
support the "as" statement in try/except blocks and does not have reduce() in
the functools package.
The main changes are:
* try/except blocks require `except ... as ...:` to resolve syntactical ambiguity
* map() and filter() return iterators rather than lists in Python 3
* reduce() is now in functools package (and not built-in in Python 3)
* Dictionaries don't have a has_key() method in Python 3
* None and int types can't be directly compared in Python 3
* print() is a statement in Python 3
See http://diveintopython3.org/porting-code-to-python-3-with-2to3.html and
PEP-3110 for details.
Verified on Python 2.6.5 and 3.1.3.
Signed-off-by: David Coles <dcoles@gaikai.com>
Signed-off-by: Julien Danjou <julien@danjou.info>
This function was intended to allow libX11 to fix a multi-threaded hang,
but the corresponding libX11 patch caused single-threaded apps to spin
sometimes. Since I've retracted that patch, this patch has no users and
shouldn't go into a release unless/until that changes.
This reverts commit 2415c11dec.
Conflicts:
src/xcb.h
src/xcb_in.c
Signed-off-by: Jamey Sharp <jamey@minilop.net>
In some circumstances using xcb_poll_for_event is suboptimal because
it checks the connection for new events. This may lead to a lot of
failed nonblocking read system calls.
Signed-off-by: Rami Ylimäki <rami.ylimaki@vincit.fi>
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Fixes fallback to local connections from Xlib's XOpenDisplay(), which
will try with protocol "unix" if a hostname is specified and tcp fails
(as it usually will now that most OS'es ship with -nolisten tcp enabled).
Also fixes explicitly specifying DISPLAY as "unix/foo:0", which Xlib
previously accepted for Unix domain sockets.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
This patch is necessary so xcb reads the payload after the message
for GenericEvents with the 0x80 flag turned on.
Signed-off-by: Carlos Garnacho <carlosg@gnome.org>
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Solves compiler warning on Solaris:
"xcb_conn.c", line 304: warning: implicit function declaration: shutdown
Also provides system definition of SHUT_RDWR on Solaris 11.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Jamey Sharp <jamey@minilop.net>
"unifdef -UDNETCONN src/xcb_util.c" plus re-indenting code that was
formerly in the else clause after a DECnet check.
DECnet support has been removed from most of the X.Org code base for
several years, and it appears DNETCONN was never defined in XCB.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Jamey Sharp <jamey@minilop.net>
Solaris Trusted Extensions puts the endpoints for the X server's Unix
domain sockets in a special directory shared from the global zone to
each of the labeled zones, since each labeled zone has a separate /tmp.
Signed-off-by: Alan Coopersmith <alan.coopersmith@oracle.com>
Reviewed-by: Peter Harris <pharris@opentext.com>
Variable new_sockname will leak and sockname will be double freed if
both of the cases shown below are true.
1. realloc succeeds and doesn't return the original pointer
2. calling socket_func fails
Signed-off-by: Rami Ylimäki <rami.ylimaki@vincit.fi>
Signed-off-by: Erkki Seppälä <erkki.seppala@vincit.fi>
Reviewed-by: Arnaud Fontaine <arnau@debian.org>
Signed-off-by: Peter Harris <pharris@opentext.com>
Later patches will insert reader_list entries from other entry points.
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Reviewed-by: Josh Triplett <josh@freedesktop.org>
It's possible to call xcb_wait_for_reply more than once for a single
request. In this case we are nice and let reply waiters continue so
that they can notice that the reply is not available
anymore. Otherwise an event waiter could just signal the reply waiter
that got its reply to continue but leave a waiter for an earlier reply
blocked.
Below is an example sequence for reproducing this problem.
thread #1 (XNextEvent)
- waits for events
thread #2 (XSync)
- executes request #2
- waits for reply #2
thread #1
- reads reply #2
- signals waiter of reply #2 to continue
- waits for events
thread #2
- handles reply #2
thread #3 (XCloseDisplay)
- executes request #3
- waits for reply #2
thread #1
- reads reply #3
- nobody is waiting for reply #3 so don't signal
- wait for events
Of course it may be questionable to wait for a reply twice, but XCB
should be smart enough to let clients continue if they choose to do
so.
Signed-off-by: Rami Ylimäki <rami.ylimaki@vincit.fi>
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Windows' file handles have never been small or consecutive, so Windows'
select has always been implemented the same way as everyone else's poll.
On Windows, FD_SETSIZE is the size of the poll array, not the maximum
SOCKET number.
Signed-off-by: Peter Harris <git@peter.is-a-geek.org>
If lots of requests are send without one causing a reply, xcb can get confused
about the current sequence number of a reply. Document the requirements of an
external socket owner to avoid this problem.
The return_socket callback for xcb_take_socket() originally was supposed to
return the last sequence number used, but the version committed to libxcb never
actually had this signature. This fixes the function's documentation not to
mention this non-existent return value.
Signed-off-by: Uli Schlachter <psychon@znc.in>
Signed-off-by: Julien Danjou <julien@danjou.info>
If the initial get_peer_sock_name(getpeername ...) succeeds, the
pointer to allocated memory is overwritten by the later call to
get_peer_sock_name(getsockname ...). Fix that up by freeing
the allocated memory before overwriting the pointer.
Signed-off-by: Nick Bowler <nbowler@draconx.ca>
Signed-off-by: Julien Danjou <julien@danjou.info>
If you discard a sequence number that has multiple responses already
read, this will do more allocations than necessary. But nobody cares
about ListFontsWithInfo.
Signed-off-by: Jamey Sharp <jamey@minilop.net>
This simplifies the critical section of xcb_send_request and fixes a
couple of subtle bugs:
- It's possible for xcb_send_request to need to issue two sync requests
before it can issue the real request. Previously, we counted sequence
numbers as if both were issued, but only one went out on the wire.
- The test for whether to sync at 32-bit sequence number wrap has been
incorrect since we switched to 64-bit sequence numbers internally.
This change means that if the output queue was already full and the
current request is bigger than the output queue, XCB will do one more
write syscall than it did before. But syncs are rare and small requests
are the norm, so this shouldn't be a measurable difference.
Signed-off-by: Jamey Sharp <jamey@minilop.net>
By calling memmove instead of memcpy, and walking the buffer backward
from the end, *_unserialize is safe to use in-place.
Signed-off-by: Peter Harris <pharris@opentext.com>
In support of this, consolidate the two static error_connection
definitions into one so we don't try to free the static out-of-memory
error_connection.
Commit by Josh Triplett and Jamey Sharp.
Signed-off-by: Josh Triplett <josh@joshtriplett.org>
Signed-off-by: Jamey Sharp <jamey@minilop.net>
If a client calls close(2) on the connection's file descriptor and then
flushes writes, libxcb causes a hang in the client.
Any flush eventually calls _xcb_out_send() with has the following loop:
while(ret && *count)
ret = _xcb_conn_wait(c, &c->out.cond, vector, count);
_xcb_conn_wait(), if built with USE_POLL, gets the POLLNVAL error. It only
checks for POLLIN and POLLOUT though, ignoring the error. Return value is 1,
count is unmodified, leaving us with an endless loop and a client hang.
XTS testcase Xlib3/XConnectionNumber triggers this bug. It creates a display
connection, closes its file descriptor, tries to send a no-op, and then expects
an error.
http://cgit.freedesktop.org/xorg/test/xts/tree/xts5/Xlib3/XConnectionNumber.m
If poll returned POLLHUP or POLLERR, we might see the same result.
If poll returns any event we didn't ask for, this patch causes
_xcb_conn_shutdown() to be invoked and an error returned. This matches the
behaviour if select(2) is used instead of poll(2): select(2) returns -1 and
EBADF for an already closed file descriptor.
I believe this fix both is safe and will handle any similar error. POSIX says
that the only bits poll is permitted to set in revents are those bits that were
set in events, plus POLLHUP, POLLERR, and POLLNVAL. So if we see any flags we
didn't ask for then something has gone wrong.
Patch inspired by earlier proposals from Peter Hutterer and Aaron
Plattner--thanks!
Reported-by: Peter Hutterer <peter.hutterer@who-t.net>
Reported-by: Aaron Plattner <aplattner@nvidia.com>
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Reviewed-by: Aaron Plattner <aplattner@nvidia.com>
Reviewed-by: Peter Hutterer <peter.hutterer@who-t.net>
Tested-by: Aaron Plattner <aplattner@nvidia.com>
Cc: Peter Hutterer <peter.hutterer@who-t.net>
Cc: Dan Nicholson <dbn.lists@gmail.com>
Signed-off-by: Peter Harris <pharris@opentext.com>
Fixes the X Test Suite's XCloseDisplay-6 test, which has this (admittedly
ridiculous) behavior:
1. Create a window w.
2. Open two display connections, dpy1, and dpy2.
3. Grab the server using dpy1.
4. Fork.
5 (child). XSetProperty on w using dpy2.
5 (parent). Verify that no event was recieved on dpy1.
6 (parent). XCloseDisplay(dpy1).
6 (child). Verify that an event was received on dpy2.
It was failing because at step 6 (child), the server had not actually ungrabbed
yet because the file descriptor for dpy1 was still open in the child process.
Shutting down the socket during XCloseDisplay matches the behavior of non-XCB
Xlib, which calls shutdown() from _X11TransSocketDisconnect.
Thanks to Julien Cristau for noticing this.
Signed-off-by: Aaron Plattner <aplattner at nvidia.com>
Reviewed-by: Julien Cristau <jcristau@debian.org>
Signed-off-by: Peter Harris <pharris@opentext.com>
This fixes the test case I have so far for Havoc's report that
xcb_request_check hangs.
Rationale: Since we have a void cookie, request_expected can't have been
set equal to this sequence number when the request was sent; it can only
have become equal due to the arrival of an event or error. If it became
equal due to an event then we still need to sync. If it became equal due
to an error, then request_completed will have been updated, which means
we correctly won't sync.
Fixes: https://bugs.freedesktop.org/show_bug.cgi?id=29599
However, Havoc reports that he can still reproduce the problem, so we
may be revisiting this later.
Reported-by: Havoc Pennington <hp@pobox.com>
Signed-off-by: Jamey Sharp <jamey@minilop.net>
- API compatibility with valueparam
- request _aux() auxiliary functions
- _serialize() and _unserialize() auxiliary functions
- new data type that allows mixing of fixed and variable size members
protocol and host are allocated in _xcb_parse_display but ownership of
them is passed to the caller. They have to be freed in
xcb_connect_to_display_with_auth_info.
Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com>
Signed-off-by: Peter Harris <pharris@opentext.com>
xcb_parse_display claims that there is no side effects when failing.
That requires _xcb_parse_display to free the memory in failure case.
Signed-off-by: Pauli Nieminen <ext-pauli.nieminen@nokia.com>
Signed-off-by: Peter Harris <pharris@opentext.com>
Support scenarios where host is not set and protocol is. eg:
DISPLAY=tcp/:0
as well as the "inet" and "inet6" alias for "tcp" for compatability
with Xlib
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Reviewed-by: Jamey Sharp <jamey@minilop.net>
This is a regression found by tinderbox in previous commit:
xcb_util.c: In function '_xcb_open':
xcb_util.c:213: error: 'fd' undeclared (first use in this function)
There could be no upper limit on the length of a path according
to POSIX, therefore these macros may not be defined at all on
some systems (such as GNU Hurd).
Signed-off-by: Arnaud Fontaine <arnau@debian.org>
Reviewed-by: Peter Harris <pharris@opentext.com>
Regression found by tinderbox in 89b3485dad
xcb_util.c:31:27: error: sys/syslimits.h: No such file or directory
xcb_util.c: In function '_xcb_open':
xcb_util.c:148: error: 'PATH_MAX' undeclared (first use in this function)
Signed-off-by: Jeremy Huddleston <jeremyhu@apple.com>
Since writers must make sure they read as well, threads may have gone to
sleep waiting for the opportunity to read. The writer must wake up one
of those readers or the application can hang.
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Reviewed-by: Josh Triplett <josh@freedesktop.org>
xcb_ge_event_t has its length field in the same place that
xcb_generic_reply_t does, so there's no need to cast the generic reply.
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Cc: Peter Hutterer <peter.hutterer@who-t.net>
Reviewed-by: Julien Danjou <julien@danjou.info>
_xcb_out_send needs _xcb_conn_wait to store back its progress so it can
be reinvoked to pick up where it left off---but then _xcb_out_send
guarantees that it leaves either an empty output vector or a shut-down
connection, so *its* callers never care how much progress was made.
Signed-off-by: Jamey Sharp <jamey@minilop.net>
Reviewed-by: Josh Triplett <josh@freedesktop.org>
Depending on the process file limit, a file descriptor can be larger
than the capacity of fd_set. There is no portable way to create a
large enough fd_set at run-time. So we just fail if the file descriptor
number is too high and poll() is not available.
Reviewed-by: Jamey Sharp <jamey@minilop.net>
Signed-off-by: Julien Danjou <julien@danjou.info>
libxcb's 010e5661 (Fix XDM-AUTHORIZATION-1 (bug #14202)) mistakenly
inverted a few lines of code, making local socket authentication fail on
hpux and Hurd: when getpeername fails, sockname needs to be initialized
by getsockname before its address family can be checked.
Signed-off-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Signed-off-by: Julien Danjou <julien@danjou.info>
This saves the X11 connection from leaking into children processes.
On Linux, this is fully thread-safe using SOCK_CLOEXEC. On other
systems, there is a small race condition.
Signed-off-by: Julien Danjou <julien@danjou.info>
This function is useful for dynamic language garbage collectors. Frequently
a GC cycle may run before you want to block wainting for a reply.
This function is also marginally useful for libxcb apps that issue
speculative requests (eg. xlsclients).
Reviewed-by: Jamey Sharp <jamey@minilop.net>
Tested-by: Eamon Walsh <efw@eamonwalsh.com>
Signed-off-by: Peter Harris <pharris@opentext.com>
Daniel Martin
Jeremy Huddleston Sequoia
Keith Packard
Uli Schlachter
Ran Benita
Gaetan Nadon
Peter Harris
Kenneth Graunke
PHO
Mark Kettenis
Daphne Pfister
Alan Coopersmith
Christian König
Chí-Thanh Christopher Nguyễn
Colin Walters
Julien Danjou
Michael Stapelberg
Julien Cristau
Jon TURNEY
Jeremy Huddleston
Ryan Pavlik
Arvind Umrao
Jamey Sharp
Markus Duft
James Jones
David Coles
Rami Ylimäki
Carlos Garnacho
Vincent Torri
Jeetu Golani
Nick Bowler
Josh Triplett
Aaron Plattner
Christoph Reimann
Pauli Nieminen
Marcin Kościelnicki
Arnaud Fontaine
Rémi Denis-Courmont
Samuel Thibault
Rémi Denis-Courmont