aboutsummaryrefslogtreecommitdiff
path: root/src/microhttpd/daemon.c
Commit message (Collapse)AuthorAge
* Added more comments about data races.Evgeny Grin (Karlson2k)2016-04-18
|
* Fixed rare situations when suspended connection can't be resumed.Evgeny Grin (Karlson2k)2016-04-18
|
* Corrected pipes for worker thread for modes without MHD_USE_SUSPEND_RESUMEEvgeny Grin (Karlson2k)2016-04-14
|
* Don't create daemon if failed to make daemon control pipe non-blockingEvgeny Grin (Karlson2k)2016-04-14
|
* Free pipes if worker thread initialization failsEvgeny Grin (Karlson2k)2016-04-14
|
* Make pipe non-blocking for tread poolEvgeny Grin (Karlson2k)2016-04-14
|
* Fixed handling of caller-supplied socket with MHD_OPTION_LISTEN_SOCKET.Evgeny Grin (Karlson2k)2016-04-12
| | | | Was regression in 0.9.49.
* Corrected typos in commentsEvgeny Grin (Karlson2k)2016-04-10
|
* Use less locking with MHD_USE_THREAD_PER_CONNECTION.Evgeny Grin (Karlson2k)2016-04-10
| | | | | Do not maintain global list out timeouts as each thread use individual timeout. As result - global mutex is not acquired after each single send()/recv().
* Reworked calling shutdown() on connections:Evgeny Grin (Karlson2k)2016-04-08
| | | | | | Now called on all platforms (including W32), called only with SHUT_WR, except in close_all_connections() where shutdown() called with SHUT_RDWR. This should increase chances of graceful disconnection.
* daemon.c: remove unused variable in MHD_handle_connection()Evgeny Grin (Karlson2k)2016-04-08
|
* Dan Dedrick wrote:Christian Grothoff2016-03-15
| | | | | | | | | | | | | | | If pthread_create fails for some reason we need to not access the thread pointer as it will not be valid. Without this check a failed return code from pthread_create would cause a SIGSEGV to occur. An instance that pthread_create could fail is if enough connections were established that we ran out of space in our mapping to create another thread stack. Specifically I have seen this occur with systemd-journal-gatewayd where there was a bug with not releasing connections after they had disconnected. I submitted a fix for that here: https://github.com/systemd/systemd/pull/2287 but it would really be best if libmicrohttpd didn't SIGSEGV under these conditions.
* fix unnecessary and potentially problematic use of eready DLL in ↵Christian Grothoff2016-03-15
| | | | thread-per-connection mode
* bracket correctlyChristian Grothoff2016-02-05
|
* trying to fix full signal pipe issue reported on MLChristian Grothoff2016-02-03
|
* Reworked code for using accept4(), epoll_create1() and socket() with ↵Evgeny Grin (Karlson2k)2016-01-21
| | | | SOCK_CLOEXEC, SOCK_NONBLOCK and EPOLL_CLOEXEC. On some systems SOCK_CLOEXEC, SOCK_NONBLOCK and EPOLL_CLOEXEC are defined as enum members with the same names so precompiler assume zero value of such macros.
* Fixed making sockets non-blocking.Evgeny Grin (Karlson2k)2016-01-21
|
* daemon.c: refactored code making sockets non-blocking and non-inheritable: ↵Evgeny Grin (Karlson2k)2016-01-19
| | | | cleaned up, clarified and little bit optimized.
* Remove workaround for Cygwin and use non-blocking sockets everywhere.Evgeny Grin (Karlson2k)2016-01-19
| | | | Tested on Cygwin 1.7.28 and 2.4.0 x86/x86_64.
* make_nonblocking_noninheritable(): fixed checking for current non-blocking ↵Evgeny Grin (Karlson2k)2016-01-17
| | | | | | flags, improved error report.
* make_nonblocking_noninheritable(): removed one system callEvgeny Grin (Karlson2k)2016-01-17
|
* daemon.c: refactored code around accept4() for clarity and to remove ↵Evgeny Grin (Karlson2k)2016-01-17
| | | | unneeded variable
* Do not give up if sendfile() failed with EINVALEvgeny Grin (Karlson2k)2016-01-17
|
* configure.ac: cleanup and fixes related to optional features, corresponding ↵Evgeny Grin (Karlson2k)2016-01-11
| | | | changes in source files
* Be more compatible with older compilersEvgeny Grin (Karlson2k)2015-12-15
|
* rev 36743 missed a spotDavid Barksdale2015-12-08
|
* Change default listen backlog size from 32 to SOMAXCONN, allow to specify ↵Evgeny Grin (Karlson2k)2015-12-08
| | | | required listen backlog size
* MHD_select(): handle at least some of pending connections instead of failing ↵Evgeny Grin (Karlson2k)2015-12-08
| | | | without any data processing
* daemon.c: rename variables 'max' to prevent conflicts with system macrosEvgeny Grin (Karlson2k)2015-12-08
|
* MHD_get_fdset() & MHD_get_fdset2(): continue to add FDs to sets in case of ↵Evgeny Grin (Karlson2k)2015-12-08
| | | | error.
* daemon.c: print log messages when dealing with fdsets on all platforms;Evgeny Grin (Karlson2k)2015-12-08
| | | | replace listen socket in fdset with pipe FD instead of failing if no fdset space is available on W32.
* Use real system default value of FD_SETSIZE instead of guessingEvgeny Grin (Karlson2k)2015-12-08
|
* Handle redefined FD_SETSIZE and really redefine FD_SETSIZE for W32Evgeny Grin (Karlson2k)2015-12-07
|
* aggressively close sockets in multi-threaded operationChristian Grothoff2015-12-05
|
* Add and use _MHD_SOCKOPT_BOOL_TYPE for simplify usage of setsockopt() on ↵Evgeny Grin (Karlson2k)2015-11-21
| | | | W32/POSIX sockets.
* Hi folks,Christian Grothoff2015-11-10
| | | | | | | | | | | | | | | | I guess i found a problem with suspended connections. There is a scenarion where one can resume a connection and then stop the daemon without gives the daemon the oportunity to really resume the connection. In this case, microhttpd will abort telling one that there is suspended connection (because the connection is still in the daemon's suspended list). The following patch solves the problem (i've this problem using external mainloop). Regards, Flavio Ceolin
* -fix assertion failure from race on shutdown and buffer shrinkage with ↵Christian Grothoff2015-10-25
| | | | pipelining
* deduplicate arg parsing logic between connection.c and digestauth.c (now in ↵Christian Grothoff2015-10-05
| | | | internal.c)
* daemon.c: silent compiler warningEvgeny Grin (Karlson2k)2015-09-30
|
* help systemd by not calling shutdown() on listen socket if ↵Christian Grothoff2015-09-22
| | | | MHD_USE_PIPE_FOR_SHUTDOWN is given
* daemon.c: use correct type for setsockopt() on W32, do not use same var name ↵Evgeny Grin (Karlson2k)2015-09-06
| | | | for nested branches
* Fix mixed code/var declarationEvgeny Grin (Karlson2k)2015-09-06
|
* Fix compiler warnings; use correct variable typeEvgeny Grin (Karlson2k)2015-09-06
|
* daemon.c: MHD_ip_limit_add(): small clarificationEvgeny Grin (Karlson2k)2015-09-06
|
* patch from FC to fix use of resume in combination with external selectChristian Grothoff2015-09-02
|
* correct documentation as to when MHD_USE_EPOLL_LINUX_ONLY is allowed ↵Christian Grothoff2015-08-30
| | | | (previous formulation was wrong)
* Reimplement monotonic clock with wide range of platforms supportEvgeny Grin (Karlson2k)2015-08-27
|
* fix file names, include missing files to MakefileEvgeny Grin (Karlson2k)2015-08-22
|
* daemon.c: recv_param_adapter(): fix VS compiler warning for send()/recv()Evgeny Grin (Karlson2k)2015-08-08
|
* daemon.c: MHD_select(): check for timeout value overflowEvgeny Grin (Karlson2k)2015-08-08
|