1 <!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
2 "http://www.w3.org/TR/html4/loose.dtd">
6 <title>What's New in this Release</title>
7 <meta name="GENERATOR" content=
8 "Modular DocBook HTML Stylesheet Version 1.79">
9 <link rel="HOME" title="Privoxy 3.0.20 User Manual" href="index.html">
10 <link rel="PREVIOUS" title="Installation" href="installation.html">
11 <link rel="NEXT" title="Quickstart to Using Privoxy" href=
13 <link rel="STYLESHEET" type="text/css" href="../p_doc.css">
14 <meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
15 <link rel="STYLESHEET" type="text/css" href="p_doc.css">
16 <style type="text/css">
18 background-color: #EEEEEE;
21 :link { color: #0000FF }
22 :visited { color: #840084 }
23 :active { color: #0000FF }
24 hr.c1 {text-align: left}
29 <div class="NAVHEADER">
30 <table summary="Header navigation table" width="100%" border="0"
31 cellpadding="0" cellspacing="0">
33 <th colspan="3" align="center">Privoxy 3.0.20 User Manual</th>
37 <td width="10%" align="left" valign="bottom"><a href=
38 "installation.html" accesskey="P">Prev</a></td>
40 <td width="80%" align="center" valign="bottom"></td>
42 <td width="10%" align="right" valign="bottom"><a href=
43 "quickstart.html" accesskey="N">Next</a></td>
46 <hr class="c1" width="100%">
50 <h1 class="SECT1"><a name="WHATSNEW" id="WHATSNEW">3. What's New in this
53 <p><span class="APPLICATION">Privoxy 3.0.19</span> is a stable release.
54 The changes since 3.0.18 stable are:</p>
62 <p>Prevent a segmentation fault when de-chunking buffered
63 content. It could be triggered by malicious web servers if
64 Privoxy was configured to filter the content and running on a
65 platform where SIZE_T_MAX isn't larger than UINT_MAX, which
66 probably includes most 32-bit systems. On those platforms, all
67 Privoxy versions before 3.0.19 appear to be affected. To be on
68 the safe side, this bug should be presumed to allow code
69 execution as proving that it doesn't seems unrealistic.</p>
73 <p>Do not expect a response from the SOCKS4/4A server until it
74 got something to respond to. This regression was introduced in
75 3.0.18 and prevented the SOCKS4/4A negotiation from working.
76 Reported by qqqqqw in #3459781.</p>
82 <p>General improvements:</p>
86 <p>Fix an off-by-one in an error message about connect
91 <p>Use a GNUMakefile variable for the webserver root directory
92 and update the path. Sourceforge changed it which broke various
93 web-related targets.</p>
97 <p>Update the CODE_STATUS description.</p>
103 <p>The following changes were made between 3.0.17 and 3.0.18:</p>
111 <p>If a generated redirect URL contains characters RFC 3986
112 doesn't permit, they are (re)encoded. Not doing this makes
113 Privoxy versions from 3.0.5 to 3.0.17 susceptible to HTTP
114 response splitting (CWE-113) attacks if the
115 +fast-redirects{check-decoded-url} action is used.</p>
119 <p>Fix a logic bug that could cause Privoxy to reuse a server
120 socket after it got tainted by a server-header-tagger-induced
121 block that was triggered before the whole server response had
122 been read. If keep-alive was enabled and the request following
123 the blocked one was to the same host and using the same
124 forwarding settings, Privoxy would send it on the tainted server
125 socket. While the server would simply treat it as a pipelined
126 request, Privoxy would later on fail to properly parse the
127 server's response as it would try to parse the unread data from
128 the first response as server headers for the second one.
129 Regression introduced in 3.0.17.</p>
133 <p>When implying keep-alive in client_connection(), remember that
134 the client didn't. Fixes a regression introduced in 3.0.13 that
135 would cause Privoxy to wait for additional client requests after
136 receiving a HTTP/1.1 request with "Connection: close" set and
137 connection sharing enabled. With clients which terminates the
138 client connection after detecting that the whole body has been
139 received it doesn't really matter, but with clients that don't
140 the connection would be kept open until it timed out.</p>
144 <p>Fix a subtle race condition between
145 prepare_csp_for_next_request() and sweep(). A thread preparing
146 itself for the next client request could briefly appear to be
147 inactive. If all other threads were already using more recent
148 files, the thread could get its files swept away under its feet.
149 So far this has only been reproduced while stress testing in
150 valgrind while touching action files in a loop. It's unlikely to
151 have caused any actual problems in the real world.</p>
155 <p>Disable filters if SDCH compression is used unless filtering
156 is forced. If SDCH was combined with a supported compression
157 algorithm, Privoxy previously could try to decompress it and
158 ditch the Content-Encoding header even though the SDCH
159 compression wasn't dealt with. Reported by zebul666 in
164 <p>Make a copy of the --user value and only mess with that when
165 splitting user and group. On some operating systems modifying the
166 value directly is reflected in the output of ps and friends and
167 can be misleading. Reported by zepard in #3292710.</p>
171 <p>If forwarded-connect-retries is set, only retry if Privoxy is
172 actually forwarding the request. Previously direct connections
173 would be retried as well.</p>
177 <p>Fixed a small memory leak when retrying connections with IPv6
182 <p>Remove an incorrect assertion in
183 compile_dynamic_pcrs_job_list() It could be triggered by a pcrs
184 job with an invalid pcre pattern (for example one that contains a
185 lone quantifier).</p>
189 <p>If the --user argument user[.group] contains a dot, always
190 bail out if no group has been specified. Previously the intended,
191 but undocumented (and apparently untested), behaviour was to try
192 interpreting the whole argument as user name, but the detection
193 was flawed and checked for '0' instead of '\0', thus merely
194 preventing group names beginning with a zero.</p>
198 <p>In html_code_map[], use a numeric character reference instead
199 of ' which wasn't standardized before XHTML 1.0.</p>
203 <p>Fix an invalid free when compiled with
204 FEATURE_GRACEFUL_TERMINATION and shut down through
205 http://config.privoxy.org/die</p>
209 <p>In get_actions(), fix the "temporary" backwards compatibility
210 hack to accept block actions without reason. It also covered
211 other actions that should be rejected as invalid. Reported by
218 <p>General improvements:</p>
222 <p>Privoxy can (re)compress buffered content before delivering it
223 to the client. Disabled by default as most users wouldn't benefit
228 <p>The +fast-redirects{check-decoded-url} action checks URL
229 segments separately. If there are other parameters behind the
230 redirect URL, this makes it unnecessary to cut them off by
231 additionally using a +redirect{} pcrs command. Initial patch
232 submitted by Jamie Zawinski in #3429848.</p>
236 <p>When loading action sections, verify that the referenced
237 filters exist. Currently missing filters only result in an error
238 message, but eventually the severity will be upgraded to
243 <p>Allow to bind to multiple separate addresses. Patch set
244 submitted by Petr Pisar in #3354485.</p>
248 <p>Set socket_error to errno if connecting fails in
249 rfc2553_connect_to(). Previously rejected direct connections
250 could be incorrectly reported as DNS issues if Privoxy was
251 compiled with IPv6 support.</p>
255 <p>Adjust url_code_map[] so spaces are replaced with %20 instead
256 of '+' While '+' can be used by client's submitting form data,
257 this is not actually what Privoxy is using the lookups for. This
258 is more of a cosmetic issue and doesn't fix any known
263 <p>When compiled without FEATURE_FAST_REDIRECTS, do not silently
264 ignore +fast-redirect{} directives</p>
268 <p>Added a workaround for GNU libc's strptime() reporting
269 negative year values when the parsed year is only specified with
270 two digits. On affected systems cookies with such a date would
271 not be turned into session cookies by the +session-cookies-only
272 action. Reported by Vaeinoe in #3403560</p>
276 <p>Fixed bind failures with certain GNU libc versions if no
277 non-loopback IP address has been configured on the system. This
278 is mainly an issue if the system is using DHCP and Privoxy is
279 started before the network is completely configured. Reported by
280 Raphael Marichez in #3349356. Additional insight from Petr
285 <p>Privoxy log messages now use the ISO 8601 date format
286 %Y-%m-%d. It's only slightly longer than the old format, but
287 contains the full date including the year and allows sorting by
288 date (when grepping in multiple log files) without hassle.</p>
292 <p>In get_last_url(), do not bother trying to decode URLs that do
293 not contain at least one '%' sign. It reduces the log noise and a
294 number of unnecessary memory allocations.</p>
298 <p>In case of SOCKS5 failures, dump the socks response in the log
303 <p>Simplify the signal setup in main().</p>
307 <p>Streamline socks5_connect() slightly.</p>
311 <p>In socks5_connect(), require a complete socks response from
312 the server. Previously Privoxy didn't care how much data the
313 server response contained as long as the first two bytes
314 contained the expected values. While at it, shrink the buffer
315 size so Privoxy can't read more than a whole socks response.</p>
319 <p>In chat(), do not bother to generate a client request in case
320 of direct CONNECT requests. It will not be used anyway.</p>
324 <p>Reduce server_last_modified()'s stack size.</p>
328 <p>Shorten get_http_time() by using strftime().</p>
332 <p>Constify the known_http_methods pointers in
333 unknown_method().</p>
337 <p>Constify the time_formats pointers in parse_header_time().</p>
341 <p>Constify the formerly_valid_actions pointers in
342 action_used_to_be_valid().</p>
346 <p>Introduce a GNUMakefile MAN_PAGE variable that defaults to
347 privoxy.1. The Debian package uses section 8 for the man page and
348 this should simplify the patch.</p>
352 <p>Deduplicate the INADDR_NONE definition for Solaris by moving
353 it to jbsockets.h</p>
357 <p>In block_url(), ditch the obsolete workaround for ancient
358 Netscape versions that supposedly couldn't properly deal with
363 <p>Remove a useless NULL pointer check in load_trustfile().</p>
367 <p>Remove two useless NULL pointer checks in
368 load_one_re_filterfile().</p>
372 <p>Change url_code_map[] from an array of pointers to an array of
373 arrays It removes an unnecessary layer of indirection and on
374 64bit system reduces the size of the binary a bit.</p>
378 <p>Fix various typos. Fixes taken from Debian's 29_typos.dpatch
379 by Roland Rosenfeld.</p>
383 <p>Add a dok-tidy GNUMakefile target to clean up the messy HTML
384 generated by the other dok targets.</p>
388 <p>GNUisms in the GNUMakefile have been removed.</p>
392 <p>Change the HTTP version in static responses to 1.1</p>
396 <p>Synced config.sub and config.guess with upstream
397 2011-11-11/386c7218162c145f5f9e1ff7f558a3fbb66c37c5.</p>
401 <p>Add a dedicated function to parse the values of toggles.
402 Reduces duplicated code in load_config() and provides better
403 error handling. Invalid or missing toggle values are now a fatal
404 error instead of being silently ignored.</p>
408 <p>Terminate HTML lines in static error messages with \n instead
413 <p>Simplify cgi_error_unknown() a bit.</p>
417 <p>In LogPutString(), don't bother looking at pszText when not
418 actually logging anything.</p>
422 <p>Change ssplit()'s fourth parameter from int to size_t. Fixes a
427 <p>Add a warning that the statistics currently can't be trusted.
428 Mention Privoxy-Log-Parser's --statistics option as an
429 alternative for the time being.</p>
433 <p>In rfc2553_connect_to(), start setting cgi->error_message
438 <p>Change the expected status code returned for http://p.p/die
439 depending on whether or not FEATURE_GRACEFUL_TERMINATION is
444 <p>In cgi_die(), mark the client connection for closing. If the
445 client will fetch the style sheet through another connection it
446 gets the main thread out of the accept() state and should thus
447 trigger the actual shutdown.</p>
451 <p>Add a proper CGI message for cgi_die().</p>
455 <p>Don't enforce a logical line length limit in
456 read_config_line().</p>
460 <p>Slightly refactor server_last_modified() to remove useless
465 <p>In get_content_type(), also recognize '.jpeg' as JPEG
470 <p>Add '.png' to the list of recognized file extensions in
471 get_content_type().</p>
475 <p>In block_url(), consistently use the block reason "Request
476 blocked by Privoxy" In two places the reason was "Request for
477 blocked URL" which hides the fact that the request got blocked by
478 Privoxy and isn't necessarily correct as the block may be due to
483 <p>In listen_loop(), reload the configuration files after
484 accepting a new connection instead of before. Previously the
485 first connection that arrived after a configuration change would
486 still be handled with the old configuration.</p>
490 <p>In chat()'s receive-data loop, skip a client socket check if
491 the socket will be written to right away anyway. This can
492 increase the transfer speed for unfiltered content on fast
493 network connections.</p>
497 <p>The socket timeout is used for SOCKS negotiations as well
498 which previously couldn't timeout.</p>
502 <p>Don't keep the client connection alive if any configuration
503 file changed since the time the connection came in. This is
504 closer to Privoxy's behaviour before keep-alive support for
505 client connection has been added and also less confusing in
510 <p>Treat all Content-Type header values containing the pattern
511 'script' as a sign of text. Reported by pribog in #3134970.</p>
517 <p>Action file improvements:</p>
521 <p>Moved the site-specific block pattern section below the one
522 for the generic patterns so for requests that are matched in
523 both, the block reason for the domain is shown which is usually
524 more useful than showing the one for the generic pattern.</p>
528 <p>Remove -prevent-compression from the fragile alias. It's no
529 longer used anywhere by default and isn't known to break stuff
534 <p>Add a (disabled) section to block various Facebook tracking
535 URLs. Reported by Dan Stahlke in #3421764.</p>
539 <p>Add a (disabled) section to rewrite and redirect
540 click-tracking URLs used on news.google.com. Reported by Dan
541 Stahlke in #3421755.</p>
545 <p>Unblock linuxcounter.net/. Reported by Dan Stahlke in
550 <p>Block 'www91.intel.com/' which is used by Omniture. Reported
551 by Adam Piggott in #3167370.</p>
555 <p>Disable the handle-as-empty-doc-returns-ok option and mark it
556 as deprecated. Reminded by tceverling in #2790091.</p>
560 <p>Add ".ivwbox.de/" to the "Cross-site user tracking" section.
561 Reported by Nettozahler in #3172525.</p>
565 <p>Unblock and fast-redirect ".awin1.com/.*=http://". Reported by
566 Adam Piggott in #3170921.</p>
570 <p>Block "b.collective-media.net/".</p>
574 <p>Widen the Debian popcon exception to "qa.debian.org/popcon".
575 Seen in Debian's 05_default_action.dpatch by Roland
580 <p>Block ".gemius.pl/" which only seems to be used for user
581 tracking. Reported by johnd16 in #3002731. Additional input from
586 <p>Disable banners-by-size filters for '.thinkgeek.com/'. The
587 filter only seems to catch pictures of the inventory.</p>
591 <p>Block requests for 'go.idmnet.bbelements.com/please/showit/'.
592 Reported by kacperdominik in #3372959.</p>
596 <p>Unblock adainitiative.org/.</p>
600 <p>Add a fast-redirects exception for
601 '.googleusercontent.com/.*=cache'.</p>
605 <p>Add a fast-redirects exception for
606 webcache.googleusercontent.com/.</p>
610 <p>Unblock http://adassier.wordpress.com/ and
611 http://adassier.files.wordpress.com/.</p>
617 <p>Filter file improvements:</p>
621 <p>Let the yahoo filter hide '.ads'.</p>
625 <p>Let the msn filter hide overlay ads for Facebook 'likes' in
626 search results and elements with the id 's_notf_div'. They only
627 seem to be used to advertise site 'enhancements'.</p>
631 <p>Let the js-events filter additionally disarm setInterval().
632 Suggested by dg1727 in #3423775.</p>
638 <p>Documentation improvements:</p>
642 <p>Clarify the effect of compiling Privoxy with zlib support.
643 Suggested by dg1727 in #3423782.</p>
647 <p>Point out that the SourceForge messaging system works like a
648 black hole and should thus not be used to contact individual
653 <p>Mention some of the problems one can experience when not
654 explicitly configuring an IP addresses as listen address.</p>
658 <p>Explicitly mention that hostnames can be used instead of IP
659 addresses for the listen-address, that only the first address
660 returned will be used and what happens if the address is invalid.
661 Requested by Calestyo in #3302213.</p>
667 <p>Log message improvements:</p>
671 <p>If only the server connection is kept alive, do not pretend to
672 wait for a new client request.</p>
676 <p>Remove a superfluous log message in forget_connection().</p>
680 <p>In chat(), properly report missing server responses as such
681 instead of calling them empty.</p>
685 <p>In forwarded_connect(), fix a log message nobody should ever
690 <p>Fix a log message in socks5_connect(), a failed write
691 operation was logged as failed read operation.</p>
695 <p>Let load_one_actions_file() properly complain about a missing
696 '{' at the beginning of the file. Simply stating that a line is
697 invalid isn't particularly helpful.</p>
701 <p>Do not claim to listen on a socket until Privoxy actually
702 does. Patch submitted by Petr Pisar #3354485</p>
706 <p>Prevent a duplicated LOG_LEVEL_CLF message when sending out
707 the "no-server-data" response.</p>
711 <p>Also log the client socket when dropping a connection.</p>
715 <p>Include the destination host in the 'Request ... marked for
716 blocking. limit-connect{...} doesn't allow CONNECT ...' message
717 Patch submitted by Saperski in #3296250.</p>
721 <p>Prevent a duplicated log message if none of the resolved IP
722 addresses were reachable.</p>
726 <p>In connect_to(), do not pretend to retry if
727 forwarded-connect-retries is zero or unset.</p>
731 <p>When a specified user or group can't be found, put the name in
732 single-quotes when logging it.</p>
736 <p>In rfc2553_connect_to(), explain getnameinfo() errors
741 <p>Remove a useless log message in chat().</p>
745 <p>When retrying to connect, also log the maximum number of
746 connection attempts.</p>
750 <p>Rephrase a log message in compile_dynamic_pcrs_job_list().
751 Divide the error code and its meaning with a colon. Call the pcrs
752 job dynamic and not the filter. Filters may contain dynamic and
753 non-dynamic pcrs jobs at the same time. Only mention the name of
754 the filter or tagger, but don't claim it's a filter when it could
759 <p>In a fatal error message in load_one_actions_file(), cover
760 both URL and TAG patterns.</p>
764 <p>In pcrs_strerror(), properly report unknown positive error
765 code values as such. Previously they were handled like 0 (no
770 <p>In compile_dynamic_pcrs_job_list(), also log the actual error
771 code as pcrs_strerror() doesn't handle all errors reported by
776 <p>Don't bother trying to continue chatting if the client didn't
777 ask for it. Reduces log noise a bit.</p>
781 <p>Make two fatal error message in load_one_actions_file() more
786 <p>In cgi_send_user_manual(), log when rejecting a file name due
791 <p>In load_file(), log a message if opening a file failed. The
792 CGI error message alone isn't too helpful.</p>
796 <p>In connection_destination_matches(), improve two log messages
797 to help understand why the destinations don't match.</p>
801 <p>Rephrase a log message in serve(). Client request arrival
802 should be differentiated from closed client connections now.</p>
806 <p>In serve(), log if a client connection isn't reused due to a
807 configuration file change.</p>
811 <p>Let mark_server_socket_tainted() always mark the server socket
812 tainted, just don't talk about it in cases where it has no
813 effect. It doesn't change Privoxy's behaviour, but makes
814 understanding the log file easier.</p>
824 <p>Added a --disable-ipv6-support switch for platforms where
825 support is detected but doesn't actually work.</p>
829 <p>Do not check for the existence of strerror() and memmove()
834 <p>Remove a useless test for setpgrp(2). Privoxy doesn't need it
835 and it can cause problems when cross-compiling.</p>
839 <p>Rename the --disable-acl-files switch to
840 --disable-acl-support. Since about 2001, ACL directives are
841 specified in the standard config file.</p>
845 <p>Update the URL of the 'Removing outdated PCRE version after
846 the next stable release' posting. The old URL stopped working
847 after one of SF's recent site "optimizations". Reported by Han
854 <p>Privoxy-Regression-Test:</p>
858 <p>Added --shuffle-tests option to increase the chances of
859 detection race conditions.</p>
863 <p>Added a --local-test-file option that allows to use
864 Privoxy-Regression-Test without Privoxy.</p>
868 <p>Added tests for missing socks4 and socks4a forwarders.</p>
872 <p>The --privoxy-address option now works with IPv6 addresses
873 containing brackets, too.</p>
877 <p>Perform limited sanity checks for parameters that are supposed
878 to have numerical values.</p>
882 <p>Added a --sleep-time option to specify a number of seconds to
883 sleep between tests, defaults to 0.</p>
887 <p>Disable the range-requests tagger for tests that break if it's
892 <p>Log messages use the ISO 8601 date format %Y-%m-%d.</p>
896 <p>Fix spelling in two error messages.</p>
900 <p>In the --help output, include a list of supported tests and
901 their default levels.</p>
905 <p>Adjust the tests to properly deal with FEATURE_TOGGLE being
912 <p>Privoxy-Log-Parser:</p>
916 <p>Perform limited sanity checks for command line parameters that
917 are supposed to have numerical values.</p>
921 <p>Implement a --unbreak-lines-only option to try to revert MUA
926 <p>Accept and highlight: Added header: Content-Encoding:
931 <p>Accept and highlight: Compressed content from 29258 to 8630
936 <p>Accept and highlight: Client request arrived in time on socket
941 <p>Highlight: Didn't receive data in time: a.fsdn.com:443</p>
945 <p>Accept log messages with ISO 8601 time stamps, too.</p>
955 <p>Bump generated Firefox version to 8.0.</p>
959 <p>Only randomize the release date if the new
960 --randomize-release-date option is enabled. Firefox versions
961 after 4 use a fixed date string without meaning.</p>
968 <h2 class="SECT2"><a name="UPGRADERSNOTE" id="UPGRADERSNOTE">3.1. Note
969 to Upgraders</a></h2>
971 <p>A quick list of things to be aware of before upgrading from earlier
972 versions of <span class="APPLICATION">Privoxy</span>:</p>
976 <p>The recommended way to upgrade <span class=
977 "APPLICATION">Privoxy</span> is to backup your old configuration
978 files, install the new ones, verify that <span class=
979 "APPLICATION">Privoxy</span> is working correctly and finally merge
980 back your changes using <span class="APPLICATION">diff</span> and
981 maybe <span class="APPLICATION">patch</span>.</p>
983 <p>There are a number of new features in each <span class=
984 "APPLICATION">Privoxy</span> release and most of them have to be
985 explicitly enabled in the configuration files. Old configuration
986 files obviously don't do that and due to syntax changes using old
987 configuration files with a new <span class=
988 "APPLICATION">Privoxy</span> isn't always possible anyway.</p>
992 <p>Note that some installers remove earlier versions completely,
993 including configuration files, therefore you should really save any
994 important configuration files!</p>
998 <p>On the other hand, other installers don't overwrite existing
999 configuration files, thinking you will want to do that
1004 <p><tt class="FILENAME">standard.action</tt> has been merged into
1005 the <tt class="FILENAME">default.action</tt> file.</p>
1009 <p>In the default configuration only fatal errors are logged now.
1010 You can change that in the <a href="config.html#DEBUG">debug
1011 section</a> of the configuration file. You may also want to enable
1012 more verbose logging until you verified that the new <span class=
1013 "APPLICATION">Privoxy</span> version is working as expected.</p>
1017 <p>Three other config file settings are now off by default:
1018 <a href="config.html#ENABLE-REMOTE-TOGGLE">enable-remote-toggle</a>,
1020 "config.html#ENABLE-REMOTE-HTTP-TOGGLE">enable-remote-http-toggle</a>,
1022 "config.html#ENABLE-EDIT-ACTIONS">enable-edit-actions</a>. If you
1023 use or want these, you will need to explicitly enable them, and be
1024 aware of the security issues involved.</p>
1030 <div class="NAVFOOTER">
1031 <hr class="c1" width="100%">
1033 <table summary="Footer navigation table" width="100%" border="0"
1034 cellpadding="0" cellspacing="0">
1036 <td width="33%" align="left" valign="top"><a href="installation.html"
1037 accesskey="P">Prev</a></td>
1039 <td width="34%" align="center" valign="top"><a href="index.html"
1040 accesskey="H">Home</a></td>
1042 <td width="33%" align="right" valign="top"><a href="quickstart.html"
1043 accesskey="N">Next</a></td>
1047 <td width="33%" align="left" valign="top">Installation</td>
1049 <td width="34%" align="center" valign="top"> </td>
1051 <td width="33%" align="right" valign="top">Quickstart to Using