"Fossies" - the Fresh Open Source Software Archive

Member "mod_fastcgi-2.4.7-0910052141/CHANGES" (10 Apr 2012, 37861 Bytes) of package /linux/www/apache_httpd_modules/old/mod_fastcgi-2.4.7-0910052141.tar.gz:


As a special service "Fossies" has tried to format the requested text file into HTML format (style: standard) with prefixed line numbers. Alternatively you can here view or download the uninterpreted source code file. See also the latest Fossies "Diffs" side-by-side code changes report for "CHANGES": 2.4.6_vs_2.4.7-0910052141.

    1 2.4.7
    2 
    3  *) *nix: On graceful restart, send SIGTERM, then wait 2sec before 
    4     removing any unix domain sockets.  Based on a patch by Bernd Wurst.
    5  
    6  *) Fix uid_t/gid_t compiler warnings.  [Artur Zaprzala <arturz ownmail.net>]
    7 
    8  *) A few more EINTR fixes. [Artur Zaprzala <arturz ownmail.net>]
    9 
   10  *) Don't let the Content-Length header propagate on errors or across
   11     redirects.  [Artur Zaprzala <arturz ownmail.net>]
   12 
   13  *) Fix pass-header handling (prefix with HTTP_).  Based on a patch by 
   14     [Christian Seiler <chris_se gmx.net>]
   15 
   16  *) Add an EOS bucket to the output filter chain.  Based on a patch by
   17     [Philipp Dunkel <p.dunkel i5invest.com>]
   18     
   19  *) Handle EINTR and EAGAIN in places AIX stress testing revealed 
   20     issues.  Based on a patch by [Rainer Jung <rainer.jung kippdata.de>]
   21 
   22  *) Fix process spawning on WIN under Apache 2.2.9 (an incompatible
   23     change was introduced in APR 1.3).  Patch by 
   24     [Tom Donovan <Tom tomdonovan.net>]
   25 
   26  *) Allow duplicate Status, Location and ContentType headers
   27     (consistent with mod_cgi).  Based on a patch by 
   28     [Christian Seiler <chris_se gmx.net>]
   29 
   30  *) Fix a spurious idle timeout error that occurred under unique 
   31     application timing and response size conditions.  Based on a 
   32     report and patch by [Joe Strout <joe strout.net>]
   33 
   34 2.4.6
   35 
   36  *) Fix a bug I introduced in 2.4.4 that broke dynamic application
   37     restarts.  Reported by [Yar <yarodin gmail.com>]
   38 
   39 2.4.4
   40 
   41  *) Allow FastCgiServer and FastCgiExternal server directives
   42     to be used within VirtualHosts (again).  Add docs to explain
   43     potential accessibility from other VirtualHosts.
   44     [Rob Saccoccio <robs fastcgi.com>]
   45     
   46  *) Check for a null filename in the request to prevent an NPE that
   47     was occurring when a WebSphere Apache module was also in use.
   48     [Fabian Pehla <fabian pehla.de>}
   49 
   50  *) [AP2] Call ap_set_content_type() rather than setting the content_type
   51     directly so that the AddOutputFilterByType directive can work correctly.
   52     [Thomas 'Freaky' Hurst <tom.hurst clara.net>]
   53 
   54  *) Don't use initializers for timeval structs because on 64bit
   55     MVS there is a padding field in between tv_sec and tv_usec.
   56     [Eric Covener <covener gmail.com>]
   57 
   58  *) [AP2] Add support for nph (non parsed header) scripts.
   59 	[Peter Zijlstra <peterz nedstat.com>]
   60 	
   61  *) Abort the request if the client connection ends prematurely.
   62     [Peter Zijlstra <peterz nedstat.com>]
   63 
   64  *) Introduce the apr_* backward compatibility macros removed 
   65     in Apache 2.2.  ["Daniel Smertnig" <daniel.smertnig gmail.com>]
   66     
   67  *) Fix a problem (remove install-modules) in Makefile.AP2 under 
   68     Apache 2.2.  ["Daniel Smertnig" <daniel.smertnig gmail.com>]
   69 
   70  *) Don't count an application exit towards the number of
   71     failures when doing restart backoff handling if the exit
   72     status is 0.  ["Rob Saccoccio" <robs fastcgi.com>]
   73 
   74  *) [*nix] Don't use suexec when there is no user/group in effect.
   75     This change is consistent with Apache2 handling.  Identified
   76     by ["Florian Effenberger" <floeff arcor.de>].
   77 
   78  *) Add a -min-server-life option to the FastCgiConfig and 
   79     FastCgiServer directives to provide better control of the 
   80     restart backoff feature.  ["Benjamin Osheroff" <ben gimbo.net>]
   81 
   82 2.4.2
   83 
   84  *) [WIN] Fix handle leaks in the process manager.  
   85     ["Sascha Schumann" <sascha schumann.cx>]
   86 
   87  *) [WIN] Use a permanent pool for allocating the SystemRoot 
   88     environment variable.  [Sakamoto <ringring zb4.so-net.ne.jp>]
   89 
   90  *) [WIN] Fix starting of scripts under AP2.
   91 
   92  *) Do the connect() to the application *after* collecting a chunk
   93     of client data.  This reinstates the pre-2.4.0 behaviour.
   94     ["James Jurach" <muaddib erf.net>]
   95 
   96  *) Remove an assert that was triggering on WIN when spawn() failed.
   97 
   98  *) Provide the NO_SUEXEC_FOR_AP_USER_N_GROUP macro for building
   99     mod_fastcgi with the AP13 suexec behaviour (don't use suexec
  100     if httpd's user and group match that needed for the application).
  101 
  102  *) Prevent the use of all but the "auth" directives from being used
  103     anywhere but in global scope.  Prevent more than one instance
  104     of the FastCgiWrapper directive.
  105 
  106  *) Return NOT_FOUND (404) or FORBIDDEN (403) instead of 
  107     INTERNAL_SERVER_ERROR (500) when there are configuration issues
  108     or the script does't exist.  Suggested by ["Jeff Lawson" <bovine ud.com>].
  109 
  110 2.4.0
  111 
  112  *) [*nix Security] - When FastCgiWrapper (FastCgiSuexec) was in use 
  113     and a vhost configured to use the same uid/gid as the main
  114     server, mod_fastcgi would not bother using the wrapper (suexec)
  115     because its effective uid/gid was already appropriate.  This is
  116     consistent with Apache's v1.3 mod_cgi behaviour.  There are two
  117     problems with this approach:  1) when FastCgWrapper is in use 
  118     mod_fastcgi's process manager keeps its root privileges (as its
  119     real uid/gid) so it can terminate the applications its starts -
  120     this privilege was being passed to applications when the use of 
  121     the wrapper was bypassed  2) wrappers are often employed to
  122     perform functionality beyond setting the uid/gid - by not calling
  123     the wrapper under certain circumstances, application invocation
  124     environments were inconsistent.  With this change, the wrapper is
  125     always used (when enabled) under both Apache 1.3 and 2.
  126     Reported by ["Michael Richards" <michael fastmail.ca>].
  127 
  128  *) [*nix/AP2] Use the vhost uid/gid instead of the server uid/gid
  129     for dynamic application invocation when the FastCgiWrapper is in use.
  130     Reported by ["Michael Richards" <michael fastmail.ca>]
  131 
  132  *) [*nix] Fix handling of FastCgiWrapper when passed a real path,
  133     i.e. other than "on" or "off".  ["Michael Richards" <michael fastmail.ca>]
  134 
  135  *) Eliminate the logging of "incomplete headers (0 bytes) received from 
  136     server" when a client aborts. 
  137 
  138  *) [WIN32] Fix a delay in handling large POSTs to named pipe based
  139     servers. ["Philip Gladstone" <philip okena.com>]
  140 
  141  *) [*nix/AP2] Prevent the module from being initalized twice at startup
  142     (resulting in confusing error messages to the log).
  143 
  144  *) Eliminate the need for SetHandler or AddHandler with static or 
  145     external applications.
  146 
  147  *) Limit PM requests to start a dynamic application
  148     to 5sec to prevent endless spinning (this is a drop-dead
  149     limit that should only occur if the socket/named_pipe directory
  150     is removed out from under a running server).
  151 
  152  *) [*nix] Change the default socket directory from /tmp/fcgi to:
  153     Apache  - logs/fastcgi
  154     Apache2 - RUNTIMEDIR/fastcgi
  155 
  156  *) Add -user & -group args to FastCgiServer and FastCgiExternalServer
  157     for use with wrappers (in lieu of finding the user/group associated
  158     with a virtual host - under Apache2 this isn't accomodated).
  159 
  160  *) [WIN32] Under Apache2, require v2.0.41 or later in order to pickup my 
  161     apr_proc_create() changes.
  162 
  163  *) Log when invoking and restoring the restart backoff policy.
  164 
  165  *) [WIN32] Prevent intermittent ReadFile() failures (properly initialize the
  166     OVERLAPPED structure). 
  167 
  168  *) Eliminate need for dummy files for external servers under Apache2
  169 
  170  *) Fix auth compatibility mode handling for access checker and authorizer
  171 
  172  *) Fix HEAD request handling.  Based on a patch by 
  173     ["Chris Lightfoot" <chris ex-parrot.com>].
  174 
  175  *) [*nix] When autoupdate is enabled touch the socket when restarting
  176     the processes to prevent further requests.  
  177     ["Eckebrecht von Pappenheim" <evp heise.de>]
  178 
  179  *) Apache 2.0 support.
  180 
  181  *) [WIN32] Don't read from a potentially closed named pipe. 
  182     ["Philip Gladstone" <philip okena.com>]
  183 
  184  *) Require the Apache version 1.3.6 or later to eliminate some signal
  185     handling funk.
  186 
  187  *) [WIN32] Use asyncronous io with named pipes instead of polled
  188     nonblocking io.  This should eliminate the last of the npipe issues.
  189 
  190  *) Handle an application returning a complete and valid response without 
  191     having consumed all of the data sent to it.
  192 
  193  *) Consume remaining client data (RESPONDERs only) if any.
  194 
  195  *) Add support for backing off attempts to start applications that continuously
  196     fail to start.  Three new macros defined in mod_fastcgi.h control this 
  197     behaviour: MAX_FAILED_STARTS, RUNTIME_SUCCESS_INTERVAL, FAILED_STARTS_DELAY
  198 
  199  *) [WIN32] Add (back) support for use of TerminateProcess() to accomodate
  200     applications that do not (properly) support the shutdown event (this
  201     feature was introduced in fcgi2 2.2.2 and improved in 2.2.4).  The 
  202     new macro WIN32_SHUTDOWN_GRACEFUL_WAIT in mod_fastcgi.h conrols the 
  203     interval between signaling a proper shutdown and wacking the process(s)
  204     with a TerminateProcess().
  205 
  206  *) [WIN32] Don't set the OVERLAPPED_IO flag on NamedPipe listen HANDLEs - 
  207     setting it was just plain broken.
  208 
  209  *) [WIN32] Fix the accept mutex - all applications were sharing one!? 
  210 
  211  *) Fix 'FastCgiConfig -autoUpdate'.
  212 
  213  *) Fix 'FastCgiConfig -flush'.
  214 
  215  *) Prevent silly maxProcesses and processSlack combinations.
  216     ["Dmitry Dorofeev" <dima yasp.com>]
  217                                                                
  218  *) Properly handle the killing of idle processes when one takes a long time
  219     to exit once signaled down (or the config is funky).
  220     ["Dmitry Dorofeev" <dima yasp.com>]
  221 
  222  *) Always kill the youngest instance of an application.  Suggested by 
  223     ["Dmitry Dorofeev" <dima yasp.com>]
  224 
  225 2.2.12
  226 
  227  *) Delay the logging of write errors to the pm to account for shutdown/restart.
  228 
  229  *) (Win32) An assortment of fixes.
  230 
  231  *) Fix some broken casts that were likely the cause of an assert.
  232 
  233  *) Win32. Eliminate forward slashes from the named pipe path name.
  234     ["Gerald Richter" <richter ecos.de>]
  235 
  236  *) SIGUSR2 is no longer blocked in the process manager and the fastcgi 
  237     applications it spawns. [<ryans amazon.com>]
  238 
  239  *) Added support for the -flush argument to FastCgiConfig. 
  240     ["Eric Sit" <esit alum.mit.edu>]
  241 
  242  *) Change the "which call to module_init() is this" check to a more
  243     reliable approach.  ["Doru Petrescu" <pdoru kappa.ro>]
  244 
  245  *) Close the old pipe file descriptor in apache main on USR1/HUP 
  246     (elimnates a small leak).  
  247     ["James E. Jurach Jr." <jjurach fundsxpress.com>]
  248 
  249  *) Fix a bug in fcgi_config_set_authoritative_slot(). 
  250     ["Tetsuya Furukawa" <tetsuya secom-sis.co.jp>]
  251 
  252  *) Eliminate the use of locks to assist in the clean shutdown of
  253     applications.  Instead, it is assumed that applications handle
  254     termination signals properly (this is now embedded in the C
  255     application lib).
  256 
  257  *) Fix Win32 process termination.  Proper operation requires the use of
  258     an updated application lib (termination is now signalled with an
  259 	Event and handled by specialized thread).
  260 
  261  *) Docs cleanup.  ["Andrew Benham" <adsb bigfoot.com>]
  262 
  263  *) Added code so if the last instance of a dynamic application died without
  264     provocation, then don't restart it if singleThreshold > 0 (i.e. if the
  265     configuration allows the last instance to be killed, then allow it to die).
  266     ["Andrew Benham" <adsb bigfoot.com>]
  267 
  268  *) Fix the loadFactor calculation used to determine when dyanmic
  269     applications could be killed off due to low demand. [<adsb bigfoot.com>]
  270 
  271  *) Fix a deadlock condition that could occur with Win32 named 
  272     pipes (dynamic).
  273 
  274  *) Fix a potential deadlock condition when FastCGI application
  275     sent responses while still reading the client request (POST data).
  276 
  277 2.2.10
  278 
  279  *) Allow absolute pathnames in the -socket argument.  Suggested by 
  280     ["Christian Jaeger" <christian.jaeger sl.ethz.ch>].
  281 
  282  *) Don't invoke suexec when the user/group for the fastcgi application
  283     is the same as the apache main server.  This is consistent with 
  284     apache's suexec handling.  Suggested by ["Nikolaus Rath" 
  285     <Nikolaus rath.org>].
  286 
  287  *) Reset the apache drop dead timer upon successful read or writes
  288     to/from the client.  This eliminates timeouts that were occuring
  289     during the large file transfers to/from slow clients.
  290 
  291  *) Support generic wrappers such as cgiwrap by eliminating dependencies
  292     on Apache's SUEXEC, renaming the FastCgiSuexec directive 
  293     FastCgiWrapper and eliminating any checks regarding the target 
  294     application (this is the repsonibility of the wrapper).
  295 
  296  *) Fix a nasty bug that occurred when a client aborted a POST request 
  297     before the connection to a dynamic FastCGI application was opened.
  298     The application's lock file descriptor wasn't setup but was being 
  299     closed which resulted in FD0 being closed.  Normally this is open to 
  300     /dev/null and should pose no problems except that because the FD
  301     was available it was being returned by Apache's accept().  This 
  302     caused it to be registered for pool cleanup.  mod_cgi though moves
  303     the CGI stdin pipe to FD0 and thus it was getting waxed during pool
  304     cleanup.  Problem identified by [<checksum 163.net>].
  305 
  306 Changes with mod_fastcgi 2.2.8
  307 
  308  *) Eliminate the concept of disabled applications.  If a failure 
  309     occurs trying to setup an application (e.g. bind() error) its 
  310     tried repeatedly every init-start-delay seconds.
  311 
  312  *) Tweak to Makefile.tmpl to support DSOs.  
  313     ["Dave Hill" <ddhill zk3.dec.com>]
  314 
  315 Changes with mod_fastcgi 2.2.6
  316 
  317  *) Shutdown the PM when Apache appears to have disappeared.
  318 
  319  *) seteuid() tweak for HP-UX 11.  
  320     ["Milton L. Hankins" <mlh swl.msd.ray.com>]
  321 
  322  *) (Win32) More dynamic fixes.
  323 
  324  *) (Win32) Eliminate the per application named pipe mutex.  Its now 
  325     per process to keep the lib happy (the use _FCGI_MUTEX_ should
  326     removed from the lib).  This allows pipe-based applications to
  327     now handle multiple simoultanous requests (one per process).
  328  
  329  *) Increase the number of open FDs we look to close when spawning apps.
  330 
  331  *) Prevent an assert from popping unnecessarily.
  332 
  333  *) (Win32) Add support for interpretter scripts.
  334 
  335  *) (Win32) Fix named pipe handling (problems with large responses).
  336 
  337  *) (Win32) Remove the "can exec" check.
  338 
  339 Changes with mod_fastcgi 2.2.4
  340 
  341  *) Beta WinNT support.  ["David Allen" <djallen raleigh.ibm.com>] and 
  342     [Rob Saccoccio <robs chelsea.net>]
  343 
  344  *) Remove request type restriction (GET and POST only) in order to support
  345     Web DAV requests.  ["Scott Robertson" <sroberts codeit.com>] 
  346 
  347  *) Allow requests for URLs such as /server/some/other/qualifier to match a
  348     FastCGI server defined as /server.	This was done primarily for Java
  349 	Servlets, but is generically useful.
  350 
  351  *) Change the comm between the PM and the request handlers from a regular
  352     file to a pipe and an assortment of other dynamic fixes.
  353 
  354  *) Log dynamic process termination scheduling and the resulting 
  355     process exit notification.
  356 
  357  *) Change the default singleThreshhold from 10 to 0 to prevent the last dynamic
  358     application process from being killed off due to low demand.
  359 
  360  *) Clean up FastCGI application pathnames (e.g. remove duplicate slashes). 
  361  
  362  *) Fix bugs that prevented dynamic processes from being shutdown when
  363     the load subsided.  ["Lars Heete" <heete do.isst.fhg.de>]
  364 
  365  *) Prevent dynamic processes from being scheduled to be started before
  366     the init-start-delay or restart-delay period expires.  This prevents
  367 	the queuing of process start requests while an application which has
  368 	a long initialization period starts up.
  369 
  370  *) When suexec is enabled, allow the process manager to become root again in
  371     order to signal applications it spawned.
  372 
  373  *) Change the "server started" log message level from INFO to WARN.
  374 
  375  *) Fix dynamic server "Connection Refused" handling.
  376 
  377  *) Fix demand-based dynamic process spawing (uses a more portable approach).
  378 
  379  *) Add -idle-timeout arg to FastCgiServer, FastCgiExternalServer, and
  380     FastCgiConfig.  mod_fastcgi will now abort a connection if inactive for
  381 	longer than this period.  It applies to the initiation of connections as
  382 	well (and thus is similar to appConnTimeout).  Default is 30 seconds.
  383 
  384  *) appConnTimeout is 0 by default now resulting in blocking connect()s.  This
  385     is more platform portable/predicable.
  386 
  387  *) Leave STDOUT and STDERR open to the main server error_log.  This should
  388     eliminate a pile of protocol errors and having to track down 3rd party
  389     libs writing to stderr inadvertantly.  This doesn't mean these shouldn't
  390     be fixed in the application to use the FastCGI I/O, it just allows these
  391     applications to run when previously they'd crash and burn.
  392 
  393  *) Add the pass-header arg to server directives.
  394 
  395  *) Allow supplementary groups to be passed to spawned FastCGI applications (as
  396     is done for CGI) - at least until PR2580 is resolved.
  397 
  398  *) Initialize the default (empty) environment in a more socially acceptable
  399     way.
  400 
  401  *) Miscellaneous doc improvements based on notes I've collected up over the
  402     last few months.
  403 
  404  *) Fix a bug in the stderr handling introduced in 2.2.2.  There were conditions
  405     that resulted in not or improperly terminated strings.
  406 
  407  *) Fix the call to setsockopt() to disable Nagle.  Based on a bug report by
  408     ["Johannes Plassmann" <johannes.plassmann pdb.siemens.de>].
  409 
  410 Changes with mod_fastcgi 2.2.2
  411 
  412  *) Added support for blocking connect()s by setting appConnTimeout to 0.
  413     Non-blocking connect()s (the default) can be troublesome on some platforms.
  414     Its expected that blocking connect()s will become the default (and
  415     non-blocking connect()s will become optional) in the next release.
  416 
  417  *) Dump a compile time error if the version of Apache is too old.
  418 
  419  *) Wrap the SIGPIPE handler manipulation code such that it is only applicable
  420     to Apache releases prior to 1.3.6.
  421 
  422  *) Minor tweaks for RUSSIAN_APACHE.  ["Sergey Gershtein" <sg mplik.ru>]
  423 
  424  *) Dynamic updates:
  425 
  426     Always restart a failed dynamic application if it is the last instance.
  427     This means there once started there will always be at least one process
  428     instance of a dynamic application.
  429 
  430     Send PLEASE_START to the PM when a connect() results in an ECONNREFUSED.
  431     ECONNREFUSED means the listen queue is full (or there isn't one).  Asking
  432     the PM to start (another) application instance may help empty it faster.
  433 
  434     Change two sleep() calls to select() based snoozes because alarm() is in
  435     effect and sleep() and alarm() don't always play nice together.
  436 
  437     Fixed a couple of error messages.
  438 
  439  *) Fix -listen-queue-depth arg on FastCgiConfig (dynamic).  Previously it
  440     was ignored and the default was always used.
  441 
  442  *) Allow the -initial-env argument to be used to pass variables from the
  443     Apache process environment to the FastCGI server (by specifying a
  444     variable name without the "=" or a value).  Suggested by
  445     ["Martin Lichtin" <lichtin oanda.com>]
  446 
  447  *) Cleanup some debug macros.
  448 
  449  *) Improved script stderr handling.  Based on suggestions by
  450     ["David Birnbaum" <davidb chelsea.net>]
  451 
  452  *) Added IRIX and FreeBSD to the list of supported platforms and other minor
  453     updates to the INSTALL doc.
  454 
  455  *) Changed the default listen-queue-depth (FCGI_DEFAULT_LISTEN_Q) from
  456     5 to 100.  Its still configurable with the -listen-queue-depth option.
  457     This should help eliminate the FAQ - Why do I see "Connection Refused"
  458     messages in the log?
  459 
  460  *) Fix a bug in FastCgiExternalServer that broke support for external servers
  461     on other hosts.  ["Dave Neuer" <dneuer futuristics.net>]
  462 
  463 Changes with mod_fastcgi 2.2.1
  464 
  465  *) Updates to the INSTALL doc to describe building as a DSO.
  466 
  467  *) If the FastCgiIpcDir directive was in use and "httpd -t" was issued while
  468     Apache was running ("apachectl restart" does this), the contents of the
  469     dynamic directory (dynamic sockets and the mbox) would be blown away
  470     (a 2.1b1 bug).
  471 
  472  *) Add an extern declaration for ap_sys_siglist (its exposed but not in any
  473     Apache header file) to prevent a compile error on systems without
  474     SYS_SIGLIST defined.
  475 
  476  *) During auth requests, the subprocess_env table was left holding variables
  477     sent to the auth FastCGI server (including REMOTE_PASSWD!) which means they
  478     were passed to other processing phases (such as CGI/FastCGI).  The
  479     subprocess_env is now restored to its pre-auth condition.
  480 
  481  *) Added a FastCGI Authorizer Role compatibility mode which implements the
  482     specification to the tee.  Use -compat arg with any of the Auth server
  483     directives.  This is intended for new and existing FastCGI authorizer
  484     applications that require compatibility with other server implementations.
  485 
  486  *) Fix logging from the process manager.  Previously, all calls to the
  487     logging routines used NULL for the server_rec.  This works fine in a dev
  488     environment, but results in most of these messages being tossed by
  489     log_error_core() because DEFAULT_LOGLEVEL is too low.  Odd logic.
  490 
  491  *) Always setup fr->header so that in the event the FastCGI server
  492     doesn't get a valid FastCGI protocol header over the wire, we
  493     can still print our error message to the log from do_work().
  494 
  495  *) Fix a compile error in open_connection_to_fs() on systems with TCP_NODELAY
  496     defined. ["Don Locrasto" <locrasto iceminer.com>]
  497 
  498 Changes with mod_fastcgi 2.2.0
  499 
  500  *) Serious rewrite of mod_fastcgi.html.  An example conf will have
  501     to wait until another time - I'm burned out on docs updates.
  502 
  503  *) Add to the mod_fastcgi to the server version string.
  504 
  505  *) SCO doesn't like the const in the arg to inet_addr() and
  506     gethostbyname().
  507 
  508  *) Use Apache's NET_SIZE_T to get the correct arg type for getsockopt().
  509 
  510  *) Deleted redundant header files from fcgi.h.
  511 
  512  *) Full path names are no longer required for directives which take
  513     file names as arguments.  Like other Apache directives you can
  514     specify paths from server_root (don't start with a "/").
  515 
  516  *) The fcgiKillMgr is gone.  This intermediate process sat between
  517     the Apache parent and the FastCGI Process Manager under Apache
  518     1.2.	 The process manager tries to name itself fcgi_pm (it used
  519     to be fcgiProcMgr).
  520 
  521  *) Authentication, Authorization, and Access phases are now
  522     supported. I've bent the FastCgi spec a bit: as many of the standard
  523     environment variables (that were easy) are sent (the spec says don't
  524     send some by name), all headers (except Status) sent by the auth
  525     FastCgi server are passed to subprocesses (CGI/FastCGI invocations)
  526     as environment variables rather than just those prefixed by
  527     "Variable-".  Custom responses for auth failures aren't supported
  528     (yet.. we'll see what the demand for this is like).  In addition to
  529     the FastCGI protocol defined environment variable "FCGI_ROLE" being
  530     set to "AUTHORIZER" an environment variable "FCGI_APACHE_ROLE" is
  531     set indicating which of the three phases is being processed.  See
  532     the docs for info on the new per-directory directives:
  533 
  534         FastCgiAuthenticator, FastCgiAuthenticatorAuthoritative,
  535         FastCgiAuthorizer, FastCgiAuthorizerAuthoritative,
  536         FastCgiAccessChecker, FastCgiAccessCheckerAuthoritative
  537 
  538  *) The code has been broken into logical chunks making figuring it
  539     out much easier.  All of the #defines a user may want to change are
  540     now in mod_fastcgi.h.
  541 
  542  *) All the logging has been converted to the "supported" Apache
  543     logging routines, ap_log_error() and ap_log_rerror().  Log entries
  544     for request specific errors should now be directed to the correct
  545     server errorlog (if your running more than one).  Some attempt has
  546     been made at setting appropriate log levels (comments of course are
  547     welcome) and printing errno info when its of value.  All of log
  548     messages now have "FastCGI" in them (nice for grep).
  549 
  550  *) All of the calls to Apache routines have been updated to the ap_
  551     convention, eliminating the need for the compat header.
  552 
  553  *) By default we no longer flush every piece of data we get from the
  554     FastCGI server to the client.  This allows the FastCGI server to
  555     release without having to wait for the client flush to complete.
  556     The old behaviour can revived with the -flush argument to AppClass.
  557 
  558  *) I renamed some of the directives for consistency (the old names
  559     still work).
  560 
  561         AppClass -> FastCgiServer,
  562         FCGIConfig -> FastCgiConfig,
  563         ExternalAppClass -> FastCgiExternalServer
  564 
  565  *) Directive arguments are no longer case sensitive.
  566 
  567  *) The module now uses hard_timeout() rather than soft_timeout.  This
  568     means that the module will longjump out of the request if the
  569     drop-dead timeout expires (set with Apache's Timeout directive, the
  570     default is 5min) or if the client closes the connection (SIGPIPE).
  571     This is more typical for a Apache modules and I'm not convinced we
  572     properly handle all of the error cases well enough to use
  573     soft_timeout (I can think of one place we don't anyway). This means
  574     your FastCGI application can see SIGPIPE.
  575 
  576  *) SIGPIPE is now ignored by default in FastCGI servers spawned by
  577     Apache.  Without this the default behaviour is exit().
  578 
  579  *) It should (I think) run now under DSO now.
  580 
  581  *) The bug which prevented sending of binary data in 2.1b1 is fixed.
  582 
  583  *) The broken -initial-env bug ins 2.1b1 is fixed.
  584 
  585  *) Maybe some other stuff.
  586 
  587 
  588 
  589 *** Well there's a big gap here.  Maybe I'll go back and extract the CVS
  590 commit notes and stick 'em in here.. when I get some free time.	;)
  591 
  592     --robs  (8 Feb 99)
  593 
  594 
  595 *** Originally from docs/README..
  596 
  597 Apache[X1.03.01]/mod_fastcgi[02.00.05] 19970909 unsupported
  598 From: "Stanley Gambarin" <stanleyg cs.bu.edu>
  599 
  600  *) Yet more changes to the source distribution.  Separated out
  601     the Tcl dynamic string and buffer libraries into separate files.
  602     Did the same for the OS library, however, it is not really an
  603     abstraction... more like a bunch of wrappers.  Added to DEVNOTES
  604     a note about the problem of Apache not allowing including header
  605     file more than once.
  606 
  607 Apache[X1.03.01]/mod_fastcgi[02.00.04] 19970908 unsupported
  608 From: "Stanley Gambarin" <stanleyg cs.bu.edu>
  609 
  610  *) More changes to the installation scripts.  Theoretically we
  611     should be able to to use the same Makefile/installation script
  612     for both Apache 1.2.x and 1.3.x.  Need some sort of the abstraction
  613     layer to use the same source code for both 1.2.x and 1.3.x sources.
  614 
  615 Apache[X1.03.01]/mod_fastcgi[02.00.03] 19970905 unsupported
  616 From: "Stanley Gambarin" <stanleyg cs.bu.edu>
  617 
  618  *) Created Makefile to compile the module into the archive library,
  619     just like the proxy module.  Also initial draft of the installation
  620     script, which should work for both 1.2 and 1.3 sources of Apache.
  621 
  622 Apache[01.02.04]/mod_fastcgi[02.00.02] 19970903 unsupported
  623 From: "Stanley Gambarin" <stanleyg cs.bu.edu>
  624 
  625  *) More source reorg.  Separated out some header files.  Need much
  626     more work on this.  Updated the TODO file.  Added fcgivers.h and
  627     CHANGES	file to track the history.
  628 
  629 Apache[01.02.04]/mod_fastcgi[02.00.01] 19970902 unsupported
  630 From: "Stanley Gambarin" <stanleyg cs.bu.edu>
  631 
  632  *) Source reorganization.  This is done to provide a basis for later
  633     functionality implementation.  See TODO file for more information
  634     of what possible future enhancements are possible/desired.
  635 
  636 Apache[01.02.04]/mod_fastcgi[02.00.00] 19970902 unsupported
  637 From:  "David MacKenzie" <djm va.pubnix.com>
  638 
  639  *) Create the "dynamic" dir and "mbox" as the user specified in the
  640     User and Group directives instead of as root.  They are created with
  641     restrictive permissions, and then the module checks to see whether the
  642     user specified in the User and Group directives has read, write, and
  643     execute permissions on them.  Those permissions had been explicitly
  644     denied by creating them as "root".
  645 
  646  *) Some documentation was garbled or incomplete.
  647 
  648  *) When a FastCGI application can't be execl'd, the code used a
  649     value of errno that may have been stomped on by intervening system
  650     calls.  That happened on BSD/OS 2.1, where the error_log was
  651     reporting errno=25 (NOTTY) instead of the correct errno=13 (EACCES).
  652     An intervening stdio call was setting errno as a side-effect, which
  653     it has the right to do.
  654 
  655  *) Erroneous fprintf arguments, a missing return value, an unused
  656     function, missing declarations, and other problems detected by gcc
  657     -Wall.
  658 
  659  *) Duplicated code to create the lock file name merged into a single
  660     function, closing several memory leaks.
  661 
  662  *) FCGIConfig -minProcesses didn't allow a value of 0, so the last
  663     FCGI app in a given dynamic class would live forever, even if it
  664     hadn't been requested for weeks.
  665 
  666  *) The test for whether to keep looking for dynamic app victims to
  667     kill was backwards, so no dynamic apps were ever selected as
  668     victims.
  669 
  670  *) The killInterval and updateInterval were being ignored; the
  671     sigsuspend() forced recalculations only when a child died,
  672     instead of at the intervals specified.
  673 
  674  *) When a dynamic app couldn't be started, the program SEGV'd when
  675     trying to free the ipcAddrPtr twice.
  676 
  677  *) Remove the dead lock file and socket when cleaning up after a
  678   	server whose last child has been killed, so the FastCGIHandler
  679     isn't fooled into thinking there is still a process serving
  680     that app.
  681 
  682  *) Make file locking robust in the presence of signals.
  683 
  684  *) Rename some badly named variables.
  685 
  686  *) Fix typos in many comments.
  687 
  688  *) Fix some memory and file descriptor leaks.
  689 
  690  *) Make the blocking kill of a server closer to working.
  691 
  692  *) More fixes to calculations in dynamic application management.
  693 
  694 
  695 *** Originally from docs/README.OMI..
  696 
  697 What's New: Version 2.0b1, 16 Apr 1997
  698 
  699  *) Implemented a mechanism by which FastCGI applications are started by
  700     the web server on the first request and continue running.  Also,  a
  701     heuristics have been implemented to allow for dynamic killing of the
  702     running FastCGI apps.  The configuration options are supplied via
  703 	the new FCGIConfig directive, that is described in the
  704 	mod_fastcgi.html
  705 
  706  *) When performing internal redirect, since the original request's body
  707     has already been read, do not allow the redirected request to think
  708     that is has one [body].
  709 
  710  *) More conditional compilation for OS/2.
  711 
  712  *) Fixed occassional  "Assertion failed: len > 0, file mod_fastcgi.c"
  713 
  714  *) Fixed "failed assertion `count >= 0 && count <= bufPtr->length'"
  715 
  716  *) Added bflush() call after bwrite() in DrainReqOutBuf function to
  717     force some output to be written in the period of inactivity. This
  718     circumvents the Apache's buffering during the server-push.
  719 
  720  *) Added a header-parser function placeholder for compatibility with an
  721     Apache source code.
  722 
  723  *) Additional checks have been placed with regard to permissions of the
  724     FastCGI processes.  These include the existence of the  file,
  725     ExecCGI and IncludesNOEXEC checks, disallowing of nph- scripts, etc.
  726 
  727  *) Implemented a restart cleanup, so that no parentlesses FastCGI
  728 	applications are left after Apache is restarted and/or terminated.
  729 	This was accomplished via two independent mechanisms, where the
  730 	first one prevents a start of the process manager on the first
  731 	reading of the configuration files.  The second mechanism deals away
  732 	from Apache's  implementation of fork() (via spawn_child()) and
  733 	implements its own forking, thus removing any dependencies on the
  734 	Apache to cleanup  processes during termination phase.
  735 
  736  *) Removed the definition of the Sigfunc, as Apache 2.0b10 defines it
  737     itself in the file conf.h
  738 
  739 What's New: Version 1.4.3, 15 Jan 1997
  740 
  741  *) Fixed compilation warnings for various platforms, as well as
  742 	conditional compilation for OS/2.
  743 
  744 What's New: Version 1.4.2, 12 Dec 1996
  745 
  746  *) mod_fastcgi.c is ported to Apache 1.2b.  Any further development
  747 	will proceed under this version of Apache web server.
  748 
  749  *) As the result of porting, the "include virtual" construct of SSI
  750 	will now work correctly using either <Location> or <Directory>
  751 	directive.
  752 
  753 What's New: Version 1.4.1, 4 Dec 1996
  754 
  755  *) Checks have been removed from the ScanCGIHeaders that provided
  756     for the presence of both Status and Location headers as being
  757     an error.  Contradictory to CGI/1.1 Internet Draft, both of
  758     these headers are used by the current CGI applications.
  759 
  760 What's New: Version 1.4, 22 Nov 1996
  761 
  762  *) Added the -port option to AppClass, allowing TCP/IP communication.
  763     Added the -socket option to AppClass, allowing Unix domain
  764     communication via a configurable pathname.
  765 
  766  *) Added the ExternalAppClass directive, allowing TCP/IP
  767     communication with remote FastCGI applications.
  768 
  769  *) The handler had its own code for generating HTTP response headers;
  770     now it uses Apache's.  This reduces the size of the module.
  771     More importantly, it fixes the bug in which "include virtual"
  772     sees the HTTP response headers.
  773 
  774  *) The response header parser performed very little checking.
  775     Now the parser enforces the guidelines in the CGI/1.1
  776     Internet-Draft: Status and Location are mutually exclusive,
  777     Location can only be a response to GET or HEAD,
  778     CGI response headers can't be repeated, etc.  (The CGI response
  779     headers are the ones the handler interprets: Status, Location,
  780     and Content-type.)
  781 
  782  *) The response header parser used to miss CGI headers
  783     with no whitespace after the colon, e.g. "Status:200 OK".
  784 
  785  *) The response header parser sometimes interpreted the first line
  786     of content as an RFC822 continuation line.
  787 
  788  *) The handler implemented a nonstandard version of Location
  789     which never used internal redirects.  The handler
  790     now attempts to implement Location as specified in the
  791     CGI/1.1 Internet-Draft.  The module documentation explains
  792     the new behavior.
  793 
  794  *) Error log entries from the response header parser were
  795     pretty uninformative; they are better now.  If a header is
  796     malformed, the log entry includes it.  If the headers are
  797     unterminated rather than malformed, the log entry says that,
  798     and says how many bytes were received from the app.
  799 
  800  *) When the application manager forked a new process, and that
  801     process ran into trouble before executing the first instruction
  802     of application code, the process used to exit with status =
  803     errno.  This made certain configuration problems (e.g. incorrect
  804     file permissions when server parent is root) quite difficult
  805     to pin down.  Now the failing child process opens up the error log
  806     and writes an informative entry before exiting.
  807 
  808  *) The module now correctly handles a slash at the end of the
  809     DocumentRoot directive.  This was a one-line fix.
  810 
  811 What's New: Version 1.3.3, 17 Oct 1996
  812 
  813  *) The module now registers its request handler under the name
  814     fastcgi-script in addition to the name application/x-httpd-fcgi.
  815     This was a one-line addition to the module, but had quite
  816     a large impact on the documentation and sample configuration.
  817 
  818 What's New: Version 1.3.2, 27 Sept 1996
  819 
  820  *) On some systems (SunOS, Linux?), fopen for append has a bug
  821     that strikes when two processes append to the same file.
  822     This bug causes the process manager to corrupt the error log.
  823     Work around the bug by calling open, then fdopen.  A patch
  824     to Apache 1.1.1 is also required, as described in mod_fastcgi.html.
  825     (Reported by Bob Ramstad.)
  826 
  827 What's New: Version 1.3.1, 17 Sept 1996
  828 
  829  *) On Linux and SunOS, the Apache default user_id (-1) is not a
  830     legal value of uid_t, so cast it.  (Reported by Bob Ramstad,
  831     Scott Langley, others.)
  832 
  833  *) On some systems (Linux, some Solaris?, Irix?), connect requires
  834     write access to a Unix Domain socket, so provide it.  (Reported
  835     by Scott Langley, <freeform wired.com>, others.)
  836 
  837  *) If you hit a .fcg file, but there's no AppClass defined, the
  838     error should be NOT_FOUND instead of SERVER_ERROR.  (Reported
  839     by Michael Smith.)
  840 
  841 What's New: Version 1.3, 4 Sept 1996
  842 
  843  *) Module sometimes busy-waited in FastCgiDoWork.  Fixed.
  844 
  845  *) Module violated the Apache buff abstraction.
  846     Fixed.  Might now work with SSL (not tested).
  847     As part of the fix, eliminated support for
  848     Apache 1.0x versions.
  849 
  850  *) Module failed to chmod the Unix domain listening sockets
  851     it created, so protections were set according to the current
  852     umask.  Fixed.
  853 
  854  *) Module forked too many process manager processes: One
  855     per AppClass.  When the Apache parent ran as root, these
  856     process manager processes ran as root.  New process
  857     manager is a single process (only started if AppClass is
  858     used) and runs with same privileges as other children.
  859     The new process manager doesn't do any polling, so
  860     there's less system overhead than before.
  861 
  862  *) AppClass insisted on getting at least two arguments.  Fixed.
  863 
  864  *) Module used setjmp/longjmp, causing compiler warnings
  865     on some platforms (e.g. Linux.)  The module no longer
  866     uses setjmp/longjmp.
  867 
  868  *) Module created listening sockets in /tmp, where they were
  869     sometimes wiped out by cleanup scripts.  Added FastCgiIpcDir
  870     directive to give control over the location of listening
  871     sockets.
  872 
  873  *) Module wrote error log entries without a timestamp.  Fixed.
  874 
  875  *) AppClass directive wrote error messages to stderr in addition
  876     to returning a char * message to the Apache core.  Fixed.
  877 
  878 What's New: Version 1.2, 3 June 1996
  879 
  880  *) Ported from Apache-1.0-based code to Apache-1.1b2 internals by
  881     ["Ralf S. Engelschall" <rse en.muc.de>]
  882     Add version string: APACHEVERSION macro to mod_fastcgi.c
  883 
  884  *) chown FastCGI socket to user_id and setuid to user_id for app
  885     class processes
  886 
  887  *) Modify GetFromStream() by having it call an OS dependent
  888     function GetStreamSize(FILE *) which uses FILE internal data member.
  889     Linux users might need to modify GetStreamSize().
  890 
  891 What's New: Version 1.1, 10 May 1996
  892 
  893  *) If you specify a non-existent executable in the AppClass directive,
  894 	or if the file exists and it does not have execute permission you
  895 	get a constant stream of error messages telling you that "program
  896 	terminated due to a signal".
  897 
  898  *) The mod_fastcgi module should use the standard Apache error logging
  899 	facility instead of writing to stderr.
  900 
  901 Version 1.0, 30 April 1996
  902 
  903