4f1587c57eIf client connection aborts during parse of "\r\n\r\n", then lighttpd never cleans up the connection and keeps trying to read from the died connection. lighttpd still responses to other requests, it just eats up all the cpu trying to handle the died connection.
Robert Jakabosky
2006-12-15 06:13:43 +0000
14d40b3d66replaced SEGFAULT() with return HANDLER_ERROR
Robert Jakabosky
2006-12-14 22:42:22 +0000
288dca4b28added error message.
Robert Jakabosky
2006-12-14 20:29:21 +0000
7a330fc067fixed bug with file uploads(large request content)
Robert Jakabosky
2006-12-14 05:18:45 +0000
fd8fe29e47updated fastcgi/proxy settings to use mod_proxy_core
Robert Jakabosky
2006-12-13 06:18:03 +0000
a2585f63b8fixed compile error for gcc 2.96
Robert Jakabosky
2006-12-13 01:44:01 +0000
0eef996951- fix a small signed comparison warning
Marcus Rückert
2006-12-12 23:32:25 +0000
0c24f423a9- fix a few warnings with unused variables
Marcus Rückert
2006-12-12 23:23:58 +0000
bf6f03f410converted mod_proxy_backend_xxx's into loadable modules. added backend scgi rewrote http-chunked and fastcgi decoding added support for X-LIGHTTPD-send-tempfile
Robert Jakabosky
2006-12-12 22:13:38 +0000
61f5045cbd- disabled aio_write() on the socket and use the classic non-blocking write() instead - fallback to blocking read() in case we are out of iocbsJan Kneschke2006-11-28 14:55:47 +0000
2414c9c1b5added a hack to get the current number of open fdsJan Kneschke2006-11-15 21:48:13 +0000
6f824d7aafcheck if we have place for a iocb before we start getting a mmap(), a tempfile and so onJan Kneschke2006-11-15 21:46:39 +0000
da92511683handle connection close from a HTTP/1.0 backendJan Kneschke2006-11-15 21:24:47 +0000
af802be9ffclose the connection after we had a parse error in the HTTP-requestJan Kneschke2006-11-15 18:27:10 +0000
fdd23a2175include the invalid hostname in the error-msgJan Kneschke2006-11-15 18:26:46 +0000
07164d2d22use a pool for the unused buffers and reuse them along the lineJan Kneschke2006-11-15 18:26:26 +0000
37640146d9- moved buffer-pool() functions to buffer.c - reset the header array between parsing runs - free last tokenJan Kneschke2006-11-15 18:25:26 +0000
f21b733a6dmoved buffer_pool() functions to buffer.hJan Kneschke2006-11-15 18:24:30 +0000
e5fb34d7eafixed error-msgs for invalid proxy-core.balancer and .protocolJan Kneschke2006-11-15 17:13:40 +0000
89a4941384- added support for server.use-noatime - added alignment check in case we are unaligned after a fallback to sendfile()Jan Kneschke2006-11-15 17:01:31 +0000
499076c293added more debug-info in case io-getevents() returns a failure for a io-submit() callJan Kneschke2006-11-15 17:00:17 +0000
23068466bcreport out-of-fds (EMFILE) nicely and handle it in the mainloop as usualJan Kneschke2006-11-13 23:48:40 +0000
7b44b01839fallback the synchronous sendfile() in case of async fails for some reasonJan Kneschke2006-11-13 23:48:01 +0000
4fca8ae425provide a useful error-msg on socket-fd outJan Kneschke2006-11-13 23:46:59 +0000
62e7b8664fbe a nice fdevent-handler and use the joblist-handler of the mainloop, just call joblist_append()Jan Kneschke2006-11-13 23:46:34 +0000
930c6c8086fixed bug where we tried to send a chunk which was already send, a second time [and crashed]Jan Kneschke2006-11-13 22:40:08 +0000
3adce80778don't send the data to /dev/null if another module wants to handle the contentJan Kneschke2006-11-13 20:29:37 +0000
422310a966steal the request-body tempfile-chunk from the incoming queueJan Kneschke2006-11-13 20:28:59 +0000
7e4b188475POST requests might have tempfiles which we want to send with aio-sendfile. make the switch() aware of itJan Kneschke2006-11-13 20:28:28 +0000
97ace079cdr1435@h2o: darix | 2006-11-10 18:25:42 +0100 "www.example.com." is equivalent to "www.example.com". The trailing dot just tells the resolver to script the searchlist.
Marcus Rückert
2006-11-10 17:37:19 +0000
4a609547a0r1435@h2o: darix | 2006-11-10 18:25:42 +0100 "www.example.com." is equivalent to "www.example.com". The trailing dot just tells the resolver to script the searchlist.
Marcus Rückert
2006-11-10 17:37:19 +0000
8f53b41de2fixed off-by-one when send the FastCGI headerJan Kneschke2006-11-10 14:56:34 +0000
b608f1cab8- disconnect from the terminal and point stdout and stderr to /dev/null - close only the fds which are really open (and don't try to brute force 3..256)Jan Kneschke2006-11-10 14:23:50 +0000
1508ca0d94don't report a "400 Bad Request" on connection shutdown after keep-aliveJan Kneschke2006-11-10 13:57:10 +0000
d38c026b5afixed crash when aio is used in a daemon-modeJan Kneschke2006-11-10 12:17:41 +0000
2462fffeafgot it working again (same problem as network_writev)Jan Kneschke2006-11-09 15:11:27 +0000
c2ee69856d- merged -r1090:1091 => add --with-webdav-locks - open modules with 'RTLD_NOW|RTLD_GLOBAL' this fixes the problems with luafilesystems and missing symbols - added AC_SUBST(UUID_LIBS)
Marcus Rückert
2006-10-09 13:50:29 +0000
53ab644ead- merged -r1090:1091 => add --with-webdav-locks - open modules with 'RTLD_NOW|RTLD_GLOBAL' this fixes the problems with luafilesystems and missing symbols - added AC_SUBST(UUID_LIBS)
Marcus Rückert
2006-10-09 13:50:29 +0000
2c0e987fb7save the old errno of the sendfile() call in case that the attr_get() fails in stat-cache-get-entryJan Kneschke2006-10-08 18:24:14 +0000
a0aafd9a24save the old errno of the sendfile() call in case that the attr_get() fails in stat-cache-get-entryJan Kneschke2006-10-08 18:24:14 +0000
9b16cb637fthe patch attached to #119 was right. If we have content on a HEAD request, take it as the Content-Length and discard the content (as it is a HEAD request)Jan Kneschke2006-10-07 17:47:49 +0000
d6bfa8aaa8the patch attached to #119 was right. If we have content on a HEAD request, take it as the Content-Length and discard the content (as it is a HEAD request)Jan Kneschke2006-10-07 17:47:49 +0000
08cd3daa2fupdated fixed tickets for the pre-releaseJan Kneschke2006-10-07 17:35:24 +0000
9a49269981updated fixed tickets for the pre-releaseJan Kneschke2006-10-07 17:35:24 +0000
1ee7b454dcfixed writev() on FreeBSD 4.x and older (FreeBSD-port: PR103758)Jan Kneschke2006-10-07 17:27:29 +0000
Robert Jakabosky
Jan Kneschke
Xuefer