zmq_ppoll mostly mimics zmq_poll behavior, except for the added feature of being able to specify a signal mask. Signals in this mask will be blocked during execution of zmq_ppoll. Switching of the process' active signal mask happens atomically with the actual poll call, so that no race conditions can occur. This behavior is useful when one wants to gracefully handle POSIX signals without race conditions. See e.g. the discussion below https://250bpm.com/blog:12/ for an explanation.
Also includes two new tests:
1. test_zmq_ppoll_fd does the same thing as test_zmq_poll_fd, demonstrating backwards compatibility with zmq_poll when used with a default signal mask.
2. test_zmq_ppoll_signals demonstrates the use of zmq_ppoll with a signal mask, blocking out SIGTERM everywhere except in zmq_ppoll, allowing to handle the signal in one place without having to worry about race conditions.
* Problem: No direct support for setting socket priority
Solution: Add ZMQ_PRIORITY socket option, which sets the
SO_PRIORITY socket option on the underlying socket. This
socket option is not supported under Windows. Check option
and set socket option on creation of underlying socket.
Add -latomic to PKGCFG_LIBS_PRIVATE so applications linking statically
with libzmq (such czmq) will know that they have to link with -latomic
and the following build failure will be avoided:
CCLD src/czmq_selftest
/home/buildroot/autobuild/run/instance-3/output-1/host/opt/ext-toolchain/bin/../lib/gcc/sparc-buildroot-linux-uclibc/7.4.0/../../../../sparc-buildroot-linux-uclibc/bin/ld: /home/buildroot/autobuild/run/instance-3/output-1/host/sparc-buildroot-linux-uclibc/sysroot/usr/lib/libzmq.a(src_libzmq_la-ctx.o): in function `zmq::ctx_t::create_socket(int)':
ctx.cpp:(.text+0x1710): undefined reference to `__atomic_fetch_add_4'
Fixes:
- http://autobuild.buildroot.org/results/4a12f1ede260cd956a0b5ccb4eec6ca8b44cb04f
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Solution: detect cacheline size for aligment purposes at build time
instead of hard-coding it, so that PPC and S390 can align to a value
greater than the 64 bytes default.
Uses libc getconf program, and falls back to the previous value of 64
if not found.
Commit 2ec5a33f6e986661dc92f3585ea400f919a39cae added support to link
with -latomic if needed however using LDFLAGS doesn't work when
statically linking because LDFLAGS is added before LIBS
Detection of atomic fails with:
configure:23230: /accts/mlweber1/instance-2/output/host/bin/sparc-linux-g++ -o conftest -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -Os -static -static -pedantic -Werror -Wall -D_GNU_SOURCE -D_REENTRANT -D_THREAD_SAFE -Wno-long-long -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64 -static -latomic conftest.cpp -lrt -lpthread -lstdc++ >&5
/tmp/ccgrvVTg.o: In function `main':
conftest.cpp:(.text.startup+0x10): undefined reference to `__atomic_fetch_add_4'
collect2: error: ld returned 1 exit status
So use LIBS instead of LDFLAGS
Fixes:
- http://autobuild.buildroot.net/results/c471d6b1061a8516f7772735e471db68a32965aa
Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
On some cases, -latomic is needed for likning, and since the current
acinclude.m4 checks only compilation we can sometimes miss the need for -latomic
and the linking process will fail.
Therefore, the AC_CHECK_IFELSE was replaced with AC_LINK_IFELSE. If the first
try fails, we try to link again with -latomic and add LIBS="-latmoic" in case we
succeeded.
Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Linux now supports Virtual Routing and Forwarding (VRF) as per:
https://www.kernel.org/doc/Documentation/networking/vrf.txt
In order for an application to bind or connect to a socket with an
address in a VRF, they need to first bind the socket to the VRF device:
setsockopt(sd, SOL_SOCKET, SO_BINDTODEVICE, dev, strlen(dev)+1);
Note "dev" is the VRF device, eg. VRF "blue", rather than an interface
enslaved to the VRF.
Add a new socket option, ZMQ_BINDTODEVICE, to bind a socket to a device.
In general, if a socket is bound to a device, eg. an interface, only
packets received from that particular device are processed by the socket.
If device is a VRF device, then subsequent binds/connects to that socket
use addresses in the VRF routing table.
(msys building is buggy, please be aware, it fails to compile on my
machine) also I modified the buildall.bat/buildbase.bat to use correct
MSVC versions instead of "visual studio 2017"
Solution: even though epoll() semantics are supported in modern illumos, it may differ from what Linux developers are used to expect. Skip epoll and use poll or select semantics (both were checked to work well).
Solution: keep the new --without-docs option, but also keep the old
--without-documentation with an added deprecation warning.
We can then remove it in the next major release, to leave enough time
for users and maintainers to change it without disruptions.
It's especially annoying to see this:
--enable-perf Build performance measurement tools [default=yes].
--disable-eventfd disable eventfd [default=no]
--enable-curve-keygen Build curve key-generation tool [default=yes].
Solution: all options should explain the non-default case. Also
the language should be enable/disable, with/without, rather than
yes/no. E.g. '--without-docs'.
Specifically, the poller detection code does not set macros in
platform.hpp. The configure script passed them as -D on the command
line.
Solution: rewrite the poller detection code.
Specifically when cross-compiling, one can be willing to force these
variable values using the environment of a config.cache file. This
commit makes this possible.
The affected variables are:
* libzmq_cv_sock_cloexec
* libzmq_cv_so_keepalive
* libzmq_cv_tcp_keepcnt
* libzmq_cv_tcp_keepidle
* libzmq_cv_tcp_keepintvl
* libzmq_cv_tcp_keepalive
Signed-off-by: Lionel Orry <lionel.orry@gmail.com>
The decision about the poller mechanism to use (select, poll, ...)
was done twice: once by the build system and once by the code in
poller.hpp. As the build-system can actually detect the mechanisms
available, prefer that result to the hard coded defaults in
poller.hpp.
At the same time, remove the duplicate detection of select() vs.
poll()-variant from proxy.cpp, signaler.cpp and zmq.cpp.
This patch has not been tested on many build platforms: especially
the cmake build needs testing / patching. For the other builds,
hard code the result as these these are all Windows platforms.
- Added a macro for checking clang compiler
- Moved basic compiler checks to a macro
- Added a macro for checking if compiler supports a flag
- Added --enable-debug flag
- Added a macro for running normal autoconf check with compiler flags
- Added a macro for checking for verbose flag for different compilers (-Wall)
- Added a macro for turning on strict standards compliance
- Added a macro for turning warnings to errors
- Added a macro for checking if compiler supports given pragma
- Most of the flags now go through checks if the compilers supports them
rather than enumerating different compilers
- Added DSO symbol visibility for sun compiler
- Enabled verbose mode for sun compiler
- Fixed build for ICC 12.x by adding -wd279 to size_t checks
- Removed pkg-config checks as those don't seem to be used anywhere
Signed-off-by: Mikko Koppanen <mkoppanen@php.net>
The patch tests that the C and C++ compilers actually exist
and work. autoconf seems to default to 'g++' when C++ compiler is not
found, which causes the following error (when the compiler isn't
there):
checking for uuid_generate in -luuid... no
configure: error: cannot link with -luuid, install uuid-dev.
The config.log contains the real error message "g++ command not found"
but the error message shown to the user is misleading.
Signed-off-by: Mikko Koppanen <mkoppanen@php.net>
I ran autoupdate on the configure.in, which generated most of the
patch attached. There is also a small manual fix in which removes the
warning "Remember to add LT_INIT to configure.in" which I assume is
because AC_PROG_LIBTOOL was called inside a macro.
Signed-off-by: Mikko Koppanen <mkoppanen@php.net>
Allows building without documentation even if asciidoc and xmlto are
installed. This eases up working with the build system because doc
generation take quite a while on repeated builds.
Signed-off-by: Mikko Koppanen <mkoppanen@php.net>
After long discussions in IRC we came to conclusion that there is no
elegant way to handle static builds with MinGW32. This patch sets
AC_DISABLE_STATIC by default on MinGW32 and Cygwin
Signed-off-by: Mikko Koppanen <mkoppanen@php.net>
