3 *) [*nix Security] - When FastCgiWrapper (FastCgiSuexec) was in use
4 and a vhost configured to use the same uid/gid as the main
5 server, mod_fastcgi would not bother using the wrapper (suexec)
6 because its effective uid/gid was already appropriate. This is
7 consistent with Apache's v1.3 mod_cgi behaviour. There are two
8 problems with this approach: 1) when FastCgWrapper is in use
9 mod_fastcgi's process manager keeps its root privileges (as its
10 real uid/gid) so it can terminate the applications its starts -
11 this privilege was being passed to applications when the use of
12 the wrapper was bypassed 2) wrappers are often employed to
13 perform functionality beyond setting the uid/gid - by not calling
14 the wrapper under certain circumstances, application invocation
15 environments were inconsistent. With this change, the wrapper is
16 always used (when enabled) under both Apache 1.3 and 2.
17 Reported by Michael Richards [michael@fastmail.ca].
19 *) [*nix/AP2] Use the vhost uid/gid instead of the server uid/gid
20 for dynamic application invocation when the FastCgiWrapper is in use.
21 Reported by Michael Richards [michael@fastmail.ca]
23 *) [*nix] Fix handling of FastCgiWrapper when passed a real path,
24 i.e. other than "on" or "off". Michael Richards [michael@fastmail.ca]
26 *) Eliminate the logging of "incomplete headers (0 bytes) received from
27 server" when a client aborts.
29 *) [WIN32] Fix a delay in handling large POSTs to named pipe based
30 servers. Philip Gladstone [philip@okena.com]
32 *) [*nix/AP2] Prevent the module from being initalized twice at startup
33 (resulting in confusing error messages to the log).
35 *) Eliminate the need for SetHandler or AddHandler with static or
36 external applications.
38 *) Limit PM requests to start a dynamic application
39 to 5sec to prevent endless spinning (this is a drop-dead
40 limit that should only occur if the socket/named_pipe directory
41 is removed out from under a running server).
43 *) [*nix] Change the default socket directory from /tmp/fcgi to:
45 Apache2 - RUNTIMEDIR/fastcgi
47 *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
48 for use with wrappers (in lieu of finding the user/group associated
49 with a virtual host - under Apache2 this isn't accomodated).
51 *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my
52 apr_proc_create() changes.
54 *) Log when invoking and restoring the restart backoff policy.
56 *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
57 OVERLAPPED structure).
59 *) Eliminate need for dummy files for external servers under Apache2
61 *) Fix auth compatibility mode handling for access checker and authorizer
63 *) Fix HEAD request handling. Based on a patch by
64 Chris Lightfoot [chris@ex-parrot.com]
66 *) [*nix] When autoupdate is enabled touch the socket when restarting
67 the processes to prevent further requests.
68 Eckebrecht von Pappenheim [evp@heise.de]
70 *) Apache 2.0 support.
72 *) [WIN32] Don't read from a potentially closed named pipe.
73 Philip Gladstone [philip@okena.com]
75 *) Require the Apache version 1.3.6 or later to eliminate some signal
78 *) [WIN32] Use asyncronous io with named pipes instead of polled
79 nonblocking io. This should eliminate the last of the npipe issues.
81 *) Handle an application returning a complete and valid response without
82 having consumed all of the data sent to it.
84 *) Consume remaining client data (RESPONDERs only) if any.
86 *) Add support for backing off attempts to start applications that continuously
87 fail to start. Three new macros defined in mod_fastcgi.h control this
88 behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
90 *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
91 applications that do not (properly) support the shutdown event (this
92 feature was introduced in fcgi2 2.2.2 and improved in 2.2.4). The
93 new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the
94 interval between signaling a proper shutdown and wacking the process(s)
95 with a TerminateProcess().
97 *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs -
98 setting it was just plain broken.
100 *) [WIN32] Fix the accept mutex - all applications were sharing one!?
102 *) Fix 'FastCgiConfig -autoUpdate'.
104 *) Fix 'FastCgiConfig -flush'.
106 *) Prevent silly maxProcesses and processSlack combinations.
107 Dmitry Dorofeev [dima@yasp.com]
109 *) Properly handle the killing of idle processes when one takes a long time
110 to exit once signaled down (or the config is funky).
111 Dmitry Dorofeev [dima@yasp.com]
113 *) Always kill the youngest instance of an application. Suggested by
114 Dmitry Dorofeev [dima@yasp.com]
118 *) Delay the logging of write errors to the pm to account for shutdown/restart.
120 *) (Win32) An assortment of fixes.
122 *) Fix some broken casts that were likely the cause of an assert.
124 *) Win32. Eliminate forward slashes from the named pipe path name.
125 Gerald Richter [richter@ecos.de]
127 *) SIGUSR2 is no longer blocked in the process manager and the fastcgi
128 applications it spawns. [ryans@amazon.com]
130 *) Added support for the -flush argument to FastCgiConfig.
131 Eric Sit [esit@alum.mit.edu]
133 *) Change the "which call to module_init() is this" check to a more
134 reliable approach. Doru Petrescu [pdoru@kappa.ro]
136 *) Close the old pipe file descriptor in apache main on USR1/HUP
137 (elimnates a small leak). James E. Jurach Jr. jjurach@fundsxpress.com
139 *) Fix a bug in fcgi_config_set_authoritative_slot(). Tetsuya Furukawa
140 [tetsuya@secom-sis.co.jp]
142 *) Eliminate the use of locks to assist in the clean shutdown of
143 applications. Instead, it is assumed that applications handle
144 termination signals properly (this is now embedded in the C
147 *) Fix Win32 process termination. Proper operation requires the use of
148 an updated application lib (termination is now signalled with an
149 Event and handled by specialized thread).
151 *) Docs cleanup. Andrew Benham [adsb@bigfoot.com]
153 *) Added code so if the last instance of a dynamic application died without
154 provocation, then don't restart it if singleThreshold > 0 (i.e. if the
155 configuration allows the last instance to be killed, then allow it to die).
156 Andrew Benham [adsb@bigfoot.com]
158 *) Fix the loadFactor calculation used to determine when dyanmic
159 applications could be killed off due to low demand [adsb@bigfoot.com].
161 *) Fix a deadlock condition that could occur with Win32 named
164 *) Fix a potential deadlock condition when FastCGI application
165 sent responses while still reading the client request (POST data).
169 *) Allow absolute pathnames in the -socket argument. Suggested by
170 Christian Jaeger [christian.jaeger@sl.ethz.ch].
172 *) Don't invoke suexec when the user/group for the fastcgi application
173 is the same as the apache main server. This is consistent with
174 apache's suexec handling. Suggested by Nikolaus Rath [Nikolaus@rath.org].
176 *) Reset the apache drop dead timer upon successful read or writes
177 to/from the client. This eliminates timeouts that were occuring
178 during the large file transfers to/from slow clients.
180 *) Support generic wrappers such as cgiwrap by eliminating dependencies
181 on Apache's SUEXEC, renaming the FastCgiSuexec directive
182 FastCgiWrapper and eliminating any checks regarding the target
183 application (this is the repsonibility of the wrapper).
185 *) Fix a nasty bug that occurred when a client aborted a POST request
186 before the connection to a dynamic FastCGI application was opened.
187 The application's lock file descriptor wasn't setup but was being
188 closed which resulted in FD0 being closed. Normally this is open to
189 /dev/null and should pose no problems except that because the FD
190 was available it was being returned by Apache's accept(). This
191 caused it to be registered for pool cleanup. mod_cgi though moves
192 the CGI stdin pipe to FD0 and thus it was getting waxed during pool
193 cleanup. Problem identified by checksum@163.net.
195 Changes with mod_fastcgi 2.2.8
197 *) Eliminate the concept of disabled applications. If a failure
198 occurs trying to setup an application (e.g. bind() error) its
199 tried repeatedly every init-start-delay seconds.
201 *) Tweak to Makefile.tmpl to support DSOs. Dave Hill [ddhill@zk3.dec.com]
203 Changes with mod_fastcgi 2.2.6
205 *) Shutdown the PM when Apache appears to have disappeared.
207 *) seteuid() tweak for HP-UX 11. Milton L. Hankins [mlh@swl.msd.ray.com]
209 *) (Win32) More dynamic fixes.
211 *) (Win32) Eliminate the per application named pipe mutex. Its now
212 per process to keep the lib happy (the use _FCGI_MUTEX_ should
213 removed from the lib). This allows pipe-based applications to
214 now handle multiple simoultanous requests (one per process).
216 *) Increase the number of open FDs we look to close when spawning apps.
218 *) Prevent an assert from popping unnecessarily.
220 *) (Win32) Add support for interpretter scripts.
222 *) (Win32) Fix named pipe handling (problems with large responses).
224 *) (Win32) Remove the "can exec" check.
226 Changes with mod_fastcgi 2.2.4
228 *) Beta WinNT support. David Allen [djallen@raleigh.ibm.com] and
229 Rob Saccoccio [robs@chelsea.net]
231 *) Remove request type restriction (GET and POST only) in order to support
232 Web DAV requests. Scott Robertson [sroberts@codeit.com]
234 *) Allow requests for URLs such as /server/some/other/qualifier to match a
235 FastCGI server defined as /server. This was done primarily for Java
236 Servlets, but is generically useful.
238 *) Change the comm between the PM and the request handlers from a regular
239 file to a pipe and an assortment of other dynamic fixes.
241 *) Log dynamic process termination scheduling and the resulting
242 process exit notification.
244 *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
245 application process from being killed off due to low demand.
247 *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes).
249 *) Fix bugs that prevented dynamic processes from being shutdown when
250 the load subsided. Lars Heete [heete@do.isst.fhg.de]
252 *) Prevent dynamic processes from being scheduled to be started before
253 the init-start-delay or restart-delay period expires. This prevents
254 the queuing of process start requests while an application which has
255 a long initialization period starts up.
257 *) When suexec is enabled, allow the process manager to become root again in
258 order to signal applications it spawned.
260 *) Change the "server started" log message level from INFO to WARN.
262 *) Fix dynamic server "Connection Refused" handling.
264 *) Fix demand-based dynamic process spawing (uses a more portable approach).
266 *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
267 FastCgiConfig. mod_fastcgi will now abort a connection if inactive for
268 longer than this period. It applies to the initiation of connections as
269 well (and thus is similar to appConnTimeout). Default is 30 seconds.
271 *) appConnTimeout is 0 by default now resulting in blocking connect()s. This
272 is more platform portable/predicable.
274 *) Leave STDOUT and STDERR open to the main server error_log. This should
275 eliminate a pile of protocol errors and having to track down 3rd party
276 libs writing to stderr inadvertantly. This doesn't mean these shouldn't
277 be fixed in the application to use the FastCGI I/O, it just allows these
278 applications to run when previously they'd crash and burn.
280 *) Add the pass-header arg to server directives.
282 *) Allow supplementary groups to be passed to spawned FastCGI applications (as
283 is done for CGI) - at least until PR2580 is resolved.
285 *) Initialize the default (empty) environment in a more socially acceptable
288 *) Miscellaneous doc improvements based on notes I've collected up over the
291 *) Fix a bug in the stderr handling introduced in 2.2.2. There were conditions
292 that resulted in not or improperly terminated strings.
294 *) Fix the call to setsockopt() to disable Nagle. [based on a bug report by
295 Johannes Plassmann <johannes.plassmann@pdb.siemens.de>]
297 Changes with mod_fastcgi 2.2.2
299 *) Added support for blocking connect()s by setting appConnTimeout to 0.
300 Non-blocking connect()s (the default) can be troublesome on some platforms.
301 Its expected that blocking connect()s will become the default (and
302 non-blocking connect()s will become optional) in the next release.
304 *) Dump a compile time error if the version of Apache is too old.
306 *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
307 to Apache releases prior to 1.3.6.
309 *) Minor tweaks for RUSSIAN_APACHE. [Sergey Gershtein <sg@mplik.ru>]
313 Always restart a failed dynamic application if it is the last instance.
314 This means there once started there will always be at least one process
315 instance of a dynamic application.
317 Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
318 ECONNREFUSED means the listen queue is full (or there isn't one). Asking
319 the PM to start (another) application instance may help empty it faster.
321 Change two sleep() calls to select() based snoozes because alarm() is in
322 effect and sleep() and alarm() don't always play nice together.
324 Fixed a couple of error messages.
326 *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic). Previously it
327 was ignored and the default was always used.
329 *) Allow the -initial-env argument to be used to pass variables from the
330 Apache process environment to the FastCGI server (by specifying a
331 variable name without the "=" or a value). [suggested by
332 Martin Lichtin <lichtin@oanda.com>]
334 *) Cleanup some debug macros.
336 *) Improved script stderr handling. [based on suggestions by
337 David Birnbaum <davidb@chelsea.net>]
339 *) Added IRIX and FreeBSD to the list of supported platforms and other minor
340 updates to the INSTALL doc.
342 *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
343 5 to 100. Its still configurable with the -listen-queue-depth option.
344 This should help eliminate the FAQ - Why do I see "Connection Refused"
347 *) Fix a bug in FastCgiExternalServer that broke support for external servers
348 on other hosts. [Dave Neuer <dneuer@futuristics.net>]
350 Changes with mod_fastcgi 2.2.1
352 *) Updates to the INSTALL doc to describe building as a DSO.
354 *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
355 Apache was running ("apachectl restart" does this), the contents of the
356 dynamic directory (dynamic sockets and the mbox) would be blown away
359 *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
360 Apache header file) to prevent a compile error on systems without
363 *) During auth requests, the subprocess_env table was left holding variables
364 sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
365 were passed to other processing phases (such as CGI/FastCGI). The
366 subprocess_env is now restored to its pre-auth condition.
368 *) Added a FastCGI Authorizer Role compatibility mode which implements the
369 specification to the tee. Use -compat arg with any of the Auth server
370 directives. This is intended for new and existing FastCGI authorizer
371 applications that require compatibility with other server implementations.
373 *) Fix logging from the process manager. Previously, all calls to the
374 logging routines used NULL for the server_rec. This works fine in a dev
375 environment, but results in most of these messages being tossed by
376 log_error_core() because DEFAULT_LOGLEVEL is too low. Odd logic.
378 *) Always setup fr->header so that in the event the FastCGI server
379 doesn't get a valid FastCGI protocol header over the wire, we
380 can still print our error message to the log from do_work().
382 *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
383 defined (Don Locrasto [locrasto@iceminer.com]).
385 Changes with mod_fastcgi 2.2.0
387 *) Serious rewrite of mod_fastcgi.html. An example conf will have
388 to wait until another time - I'm burned out on docs updates.
390 *) Add to the mod_fastcgi to the server version string.
392 *) SCO doesn't like the const in the arg to inet_addr() and
395 *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
397 *) Deleted redundant header files from fcgi.h.
399 *) Full path names are no longer required for directives which take
400 file names as arguments. Like other Apache directives you can
401 specify paths from server_root (don't start with a "/").
403 *) The fcgiKillMgr is gone. This intermediate process sat between
404 the Apache parent and the FastCGI Process Manager under Apache
405 1.2. The process manager tries to name itself fcgi_pm (it used
408 *) Authentication, Authorization, and Access phases are now
409 supported. I've bent the FastCgi spec a bit: as many of the standard
410 environment variables (that were easy) are sent (the spec says don't
411 send some by name), all headers (except Status) sent by the auth
412 FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
413 as environment variables rather than just those prefixed by
414 "Variable-". Custom responses for auth failures aren't supported
415 (yet.. we'll see what the demand for this is like). In addition to
416 the FastCGI protocol defined environment variable "FCGI_ROLE" being
417 set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
418 set indicating which of the three phases is being processed. See
419 the docs for info on the new per-directory directives:
421 FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
422 FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
423 FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
425 *) The code has been broken into logical chunks making figuring it
426 out much easier. All of the #defines a user may want to change are
427 now in mod_fastcgi.h.
429 *) All the logging has been converted to the "supported" Apache
430 logging routines, ap_log_error() and ap_log_rerror(). Log entries
431 for request specific errors should now be directed to the correct
432 server errorlog (if your running more than one). Some attempt has
433 been made at setting appropriate log levels (comments of course are
434 welcome) and printing errno info when its of value. All of log
435 messages now have "FastCGI" in them (nice for grep).
437 *) All of the calls to Apache routines have been updated to the ap_
438 convention, eliminating the need for the compat header.
440 *) By default we no longer flush every piece of data we get from the
441 FastCGI server to the client. This allows the FastCGI server to
442 release without having to wait for the client flush to complete.
443 The old behaviour can revived with the -flush argument to AppClass.
445 *) I renamed some of the directives for consistency (the old names
448 AppClass -> FastCgiServer,
449 FCGIConfig -> FastCgiConfig,
450 ExternalAppClass -> FastCgiExternalServer
452 *) Directive arguments are no longer case sensitive.
454 *) The module now uses hard_timeout() rather than soft_timeout. This
455 means that the module will longjump out of the request if the
456 drop-dead timeout expires (set with Apache's Timeout directive, the
457 default is 5min) or if the client closes the connection (SIGPIPE).
458 This is more typical for a Apache modules and I'm not convinced we
459 properly handle all of the error cases well enough to use
460 soft_timeout (I can think of one place we don't anyway). This means
461 your FastCGI application can see SIGPIPE.
463 *) SIGPIPE is now ignored by default in FastCGI servers spawned by
464 Apache. Without this the default behaviour is exit().
466 *) It should (I think) run now under DSO now.
468 *) The bug which prevented sending of binary data in 2.1b1 is fixed.
470 *) The broken -initial-env bug ins 2.1b1 is fixed.
472 *) Maybe some other stuff.
476 *** Well there's a big gap here. Maybe I'll go back and extract the CVS
477 commit notes and stick 'em in here.. when I get some free time. ;)
482 *** Originally from docs/README..
484 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
485 From: Stanley Gambarin <stanleyg@cs.bu.edu>
487 *) Yet more changes to the source distribution. Separated out
488 the Tcl dynamic string and buffer libraries into separate files.
489 Did the same for the OS library, however, it is not really an
490 abstraction... more like a bunch of wrappers. Added to DEVNOTES
491 a note about the problem of Apache not allowing including header
494 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
495 From: Stanley Gambarin <stanleyg@cs.bu.edu>
497 *) More changes to the installation scripts. Theoretically we
498 should be able to to use the same Makefile/installation script
499 for both Apache 1.2.x and 1.3.x. Need some sort of the abstraction
500 layer to use the same source code for both 1.2.x and 1.3.x sources.
502 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
503 From: Stanley Gambarin <stanleyg@cs.bu.edu>
505 *) Created Makefile to compile the module into the archive library,
506 just like the proxy module. Also initial draft of the installation
507 script, which should work for both 1.2 and 1.3 sources of Apache.
509 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
510 From: Stanley Gambarin <stanleyg@cs.bu.edu>
512 *) More source reorg. Separated out some header files. Need much
513 more work on this. Updated the TODO file. Added fcgivers.h and
514 CHANGES file to track the history.
516 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
517 From: Stanley Gambarin <stanleyg@cs.bu.edu>
519 *) Source reorganization. This is done to provide a basis for later
520 functionality implementation. See TODO file for more information
521 of what possible future enhancements are possible/desired.
523 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
524 From: David MacKenzie <djm@va.pubnix.com>
526 *) Create the "dynamic" dir and "mbox" as the user specified in the
527 User and Group directives instead of as root. They are created with
528 restrictive permissions, and then the module checks to see whether the
529 user specified in the User and Group directives has read, write, and
530 execute permissions on them. Those permissions had been explicitly
531 denied by creating them as "root".
533 *) Some documentation was garbled or incomplete.
535 *) When a FastCGI application can't be execl'd, the code used a
536 value of errno that may have been stomped on by intervening system
537 calls. That happened on BSD/OS 2.1, where the error_log was
538 reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
539 An intervening stdio call was setting errno as a side-effect, which
540 it has the right to do.
542 *) Erroneous fprintf arguments, a missing return value, an unused
543 function, missing declarations, and other problems detected by gcc
546 *) Duplicated code to create the lock file name merged into a single
547 function, closing several memory leaks.
549 *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
550 FCGI app in a given dynamic class would live forever, even if it
551 hadn't been requested for weeks.
553 *) The test for whether to keep looking for dynamic app victims to
554 kill was backwards, so no dynamic apps were ever selected as
557 *) The killInterval and updateInterval were being ignored; the
558 sigsuspend() forced recalculations only when a child died,
559 instead of at the intervals specified.
561 *) When a dynamic app couldn't be started, the program SEGV'd when
562 trying to free the ipcAddrPtr twice.
564 *) Remove the dead lock file and socket when cleaning up after a
565 server whose last child has been killed, so the FastCGIHandler
566 isn't fooled into thinking there is still a process serving
569 *) Make file locking robust in the presence of signals.
571 *) Rename some badly named variables.
573 *) Fix typos in many comments.
575 *) Fix some memory and file descriptor leaks.
577 *) Make the blocking kill of a server closer to working.
579 *) More fixes to calculations in dynamic application management.
582 *** Originally from docs/README.OMI..
584 What's New: Version 2.0b1, 16 Apr 1997
586 *) Implemented a mechanism by which FastCGI applications are started by
587 the web server on the first request and continue running. Also, a
588 heuristics have been implemented to allow for dynamic killing of the
589 running FastCGI apps. The configuration options are supplied via
590 the new FCGIConfig directive, that is described in the
593 *) When performing internal redirect, since the original request's body
594 has already been read, do not allow the redirected request to think
595 that is has one [body].
597 *) More conditional compilation for OS/2.
599 *) Fixed occassional "Assertion failed: len > 0, file mod_fastcgi.c"
601 *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
603 *) Added bflush() call after bwrite() in DrainReqOutBuf function to
604 force some output to be written in the period of inactivity. This
605 circumvents the Apache's buffering during the server-push.
607 *) Added a header-parser function placeholder for compatibility with an
610 *) Additional checks have been placed with regard to permissions of the
611 FastCGI processes. These include the existence of the file,
612 ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
614 *) Implemented a restart cleanup, so that no parentlesses FastCGI
615 applications are left after Apache is restarted and/or terminated.
616 This was accomplished via two independent mechanisms, where the
617 first one prevents a start of the process manager on the first
618 reading of the configuration files. The second mechanism deals away
619 from Apache's implementation of fork() (via spawn_child()) and
620 implements its own forking, thus removing any dependencies on the
621 Apache to cleanup processes during termination phase.
623 *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
624 itself in the file conf.h
626 What's New: Version 1.4.3, 15 Jan 1997
628 *) Fixed compilation warnings for various platforms, as well as
629 conditional compilation for OS/2.
631 What's New: Version 1.4.2, 12 Dec 1996
633 *) mod_fastcgi.c is ported to Apache 1.2b. Any further development
634 will proceed under this version of Apache web server.
636 *) As the result of porting, the "include virtual" construct of SSI
637 will now work correctly using either <Location> or <Directory>
640 What's New: Version 1.4.1, 4 Dec 1996
642 *) Checks have been removed from the ScanCGIHeaders that provided
643 for the presence of both Status and Location headers as being
644 an error. Contradictory to CGI/1.1 Internet Draft, both of
645 these headers are used by the current CGI applications.
647 What's New: Version 1.4, 22 Nov 1996
649 *) Added the -port option to AppClass, allowing TCP/IP communication.
650 Added the -socket option to AppClass, allowing Unix domain
651 communication via a configurable pathname.
653 *) Added the ExternalAppClass directive, allowing TCP/IP
654 communication with remote FastCGI applications.
656 *) The handler had its own code for generating HTTP response headers;
657 now it uses Apache's. This reduces the size of the module.
658 More importantly, it fixes the bug in which "include virtual"
659 sees the HTTP response headers.
661 *) The response header parser performed very little checking.
662 Now the parser enforces the guidelines in the CGI/1.1
663 Internet-Draft: Status and Location are mutually exclusive,
664 Location can only be a response to GET or HEAD,
665 CGI response headers can't be repeated, etc. (The CGI response
666 headers are the ones the handler interprets: Status, Location,
669 *) The response header parser used to miss CGI headers
670 with no whitespace after the colon, e.g. "Status:200 OK".
672 *) The response header parser sometimes interpreted the first line
673 of content as an RFC822 continuation line.
675 *) The handler implemented a nonstandard version of Location
676 which never used internal redirects. The handler
677 now attempts to implement Location as specified in the
678 CGI/1.1 Internet-Draft. The module documentation explains
681 *) Error log entries from the response header parser were
682 pretty uninformative; they are better now. If a header is
683 malformed, the log entry includes it. If the headers are
684 unterminated rather than malformed, the log entry says that,
685 and says how many bytes were received from the app.
687 *) When the application manager forked a new process, and that
688 process ran into trouble before executing the first instruction
689 of application code, the process used to exit with status =
690 errno. This made certain configuration problems (e.g. incorrect
691 file permissions when server parent is root) quite difficult
692 to pin down. Now the failing child process opens up the error log
693 and writes an informative entry before exiting.
695 *) The module now correctly handles a slash at the end of the
696 DocumentRoot directive. This was a one-line fix.
698 What's New: Version 1.3.3, 17 Oct 1996
700 *) The module now registers its request handler under the name
701 fastcgi-script in addition to the name application/x-httpd-fcgi.
702 This was a one-line addition to the module, but had quite
703 a large impact on the documentation and sample configuration.
705 What's New: Version 1.3.2, 27 Sept 1996
707 *) On some systems (SunOS, Linux?), fopen for append has a bug
708 that strikes when two processes append to the same file.
709 This bug causes the process manager to corrupt the error log.
710 Work around the bug by calling open, then fdopen. A patch
711 to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
712 (Reported by Bob Ramstad.)
714 What's New: Version 1.3.1, 17 Sept 1996
716 *) On Linux and SunOS, the Apache default user_id (-1) is not a
717 legal value of uid_t, so cast it. (Reported by Bob Ramstad,
718 Scott Langley, others.)
720 *) On some systems (Linux, some Solaris?, Irix?), connect requires
721 write access to a Unix Domain socket, so provide it. (Reported
722 by Scott Langley, freeform@wired.com, others.)
724 *) If you hit a .fcg file, but there's no AppClass defined, the
725 error should be NOT_FOUND instead of SERVER_ERROR. (Reported
728 What's New: Version 1.3, 4 Sept 1996
730 *) Module sometimes busy-waited in FastCgiDoWork. Fixed.
732 *) Module violated the Apache buff abstraction.
733 Fixed. Might now work with SSL (not tested).
734 As part of the fix, eliminated support for
735 Apache 1.0x versions.
737 *) Module failed to chmod the Unix domain listening sockets
738 it created, so protections were set according to the current
741 *) Module forked too many process manager processes: One
742 per AppClass. When the Apache parent ran as root, these
743 process manager processes ran as root. New process
744 manager is a single process (only started if AppClass is
745 used) and runs with same privileges as other children.
746 The new process manager doesn't do any polling, so
747 there's less system overhead than before.
749 *) AppClass insisted on getting at least two arguments. Fixed.
751 *) Module used setjmp/longjmp, causing compiler warnings
752 on some platforms (e.g. Linux.) The module no longer
755 *) Module created listening sockets in /tmp, where they were
756 sometimes wiped out by cleanup scripts. Added FastCgiIpcDir
757 directive to give control over the location of listening
760 *) Module wrote error log entries without a timestamp. Fixed.
762 *) AppClass directive wrote error messages to stderr in addition
763 to returning a char * message to the Apache core. Fixed.
765 What's New: Version 1.2, 3 June 1996
767 *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
770 Add version string: APACHEVERSION macro to mod_fastcgi.c
772 *) chown FastCGI socket to user_id and setuid to user_id for app
775 *) Modify GetFromStream() by having it call an OS dependent
776 function GetStreamSize(FILE *) which uses FILE internal data member.
777 Linux users might need to modify GetStreamSize().
779 What's New: Version 1.1, 10 May 1996
781 *) If you specify a non-existent executable in the AppClass directive,
782 or if the file exists and it does not have execute permission you
783 get a constant stream of error messages telling you that "program
784 terminated due to a signal".
786 *) The mod_fastcgi module should use the standard Apache error logging
787 facility instead of writing to stderr.
789 Version 1.0, 30 April 1996