cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-06-30 Thread manoj
manoj 99/06/30 12:01:10 Modified:.STATUS pthreads/src/main http_accept.c Log: Blah. Since poll probably shouldn't be called from multiple threads on the same pollfd strucuture at the same time, and also because restarts will be hosed otherwise, disable

cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-06-29 Thread manoj
manoj 99/06/28 17:18:04 Modified:pthreads/src/include httpd.h pthreads/src/main http_accept.c Log: Change the workings of get_connection in USE_MULTI_ACCEPT mode to search the pollfd array rather than the listener linked list. The code is a bit easier to follow

cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-05-28 Thread manoj
manoj 99/05/27 17:19:34 Modified:pthreads/src/main http_accept.c Log: Squash a potential thread-safeness bug. Every thread could touch the lr and head_listener pointers at the same time when accepts weren't serialized. So, put back the special condition for one listen socket

cvs commit: apache-apr/pthreads/src/main http_accept.c http_main.c

1999-05-24 Thread manoj
manoj 99/05/23 19:10:28 Modified:pthreads/src/include http_main.h pthreads/src/main http_accept.c http_main.c Log: Switch to using a pipe to notify children of graceful shutodwn instead of a signal. This also naturally solves a problem with graceful shutdown of

cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-05-20 Thread manoj
manoj 99/05/19 21:48:40 Modified:pthreads/src/include acceptlock.h pthreads/src/main http_accept.c Log: Minor cleanup. Revision ChangesPath 1.6 +0 -3 apache-apr/pthreads/src/include/acceptlock.h Index: acceptlock.h

cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-05-20 Thread manoj
manoj 99/05/19 22:00:42 Modified:pthreads/src/main http_accept.c Log: Another cleanup. Get rid of an if (1) that was left around to make a different patch look small. Revision ChangesPath 1.14 +49 -60apache-apr/pthreads/src/main/http_accept.c Index:

cvs commit: apache-apr/pthreads/src/main http_accept.c http_main.c

1999-04-29 Thread manoj
manoj 99/04/29 12:34:23 Modified:.STATUS pthreads/src/main http_accept.c http_main.c Log: Get rid of thread_starter_thread, and just start all of our worker threads from child_main. This makes the code simpler, and eliminates any race conditions

cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-04-22 Thread manoj
manoj 99/04/21 22:46:54 Modified:pthreads/src/main http_accept.c Log: Delete unused variable to eliminate a warning. Revision ChangesPath 1.10 +0 -1 apache-apr/pthreads/src/main/http_accept.c Index: http_accept.c

cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-04-17 Thread manoj
manoj 99/04/16 18:58:58 Modified:pthreads/src/main http_accept.c Log: Next time I'll read the whole man page. filedes[0] is for reading, filedes[1] is for writing. Revision ChangesPath 1.8 +2 -2 apache-apr/pthreads/src/main/http_accept.c Index:

cvs commit: apache-apr/pthreads/src/main http_accept.c http_main.c

1999-04-17 Thread manoj
manoj 99/04/16 21:25:57 Modified:pthreads/src/include http_accept.h pthreads/src/main http_accept.c http_main.c Log: An attempt to put SINGLE_LISTEN_UNSERIALIZED_ACCEPT support back in. This also moves the SAFE_ACCEPT definition from http_accept.h to

cvs commit: apache-apr/pthreads/src/main http_accept.c http_core.c http_main.c

1999-04-14 Thread stoddard
stoddard99/04/14 14:03:38 Modified:pthreads/src/include http_accept.h httpd.h pthreads/src/main http_accept.c http_core.c http_main.c Log: New accept loop logic: All worker threads accept. a.k.a. multi accept Revision ChangesPath 1.2 +5 -3

cvs commit: apache-apr/pthreads/src/main http_accept.c http_main.c

1999-04-14 Thread manoj
manoj 99/04/14 14:38:30 Modified:pthreads/src/include http_accept.h pthreads/src/main http_accept.c http_main.c Log: Some tweaking of our http_accept modularization. A call is added to perform any initialization needed in the parent. With this, we can take all

cvs commit: apache-apr/pthreads/src/main http_accept.c

1999-04-13 Thread manoj
manoj 99/04/12 16:47:27 Modified:pthreads/src/main http_accept.c Log: Ryan was right. We should only decrement requests_this_child after a connection is accepted; otherwise, we will accept too few connections (and once the child death code works properly, too few could mean

cvs commit: apache-apr/pthreads/src/main http_accept.c Makefile.tmpl fdqueue.c http_main.c

1999-04-07 Thread stoddard
stoddard99/04/07 15:52:19 Modified:pthreads/src/include fdqueue.h pthreads/src/main Makefile.tmpl fdqueue.c http_main.c Added: pthreads/src/include http_accept.h pthreads/src/main http_accept.c Log: Add accept loop abstraction. 4 new functions