Commit Graph

112 Commits

Author SHA1 Message Date
d3586380bd get_chain_data: reject invalid entries in proxylist section 2014-07-21 13:17:24 +02:00
5168bc4eed bail out when no proxy was configured 2014-07-21 12:37:01 +02:00
82d0b13b8f main.c: remove trailing whitespace noise 2014-07-09 19:16:33 +02:00
ed7c89072a main.c: append previously existing LD_PRELOAD contents rather than overwriting
some broken programs like pulseaudio rely on LD_PRELOAD hacks to function,
if we just override the environment variable, those will stop working.

simplified version of patch suggested by @hexchain

closes #35
2014-07-09 19:12:30 +02:00
84d9a97a08 main.c: remove code duplication around LD_PRELOAD 2014-07-09 18:07:24 +02:00
f669139c9e common.c: fix trailing whitespace 2014-07-02 10:10:53 +02:00
530fee89ce proxy_getaddrinfo: don't use strncpy
the strncpy function is both dangerous and slow.
dangerous because it doesn't do what the naive programmer expects
(bounded strcpy), and slow because it pads the entire bufsize
with zeroes.
2014-01-22 22:22:24 +01:00
b76aa653c9 proxy_getaddrinfo: fix case when func was used in lieu of inet_aton 2014-01-22 16:13:20 +01:00
f868928c4b fix incorrect poll.h inclusion
sys/poll.h is a glibc legacy alias for poll.h.
the latter is specified by POSIX, the former not.
on glibc one of them just includes the other so it doesnt hurt.
2014-01-08 14:57:13 +01:00
ecbd735508 Chromium tried to close our pipes and falls into infinity loop.
If return value of close() is -1, chromium will fall into infinity loop.
2014-01-08 13:38:59 +08:00
fa0f355ce8 add missing INIT() to close hook
this caused a crash when the gcc initializer was not called first.
2013-08-30 23:44:27 +02:00
8a84f980bb fix indentation of debug.c 2013-06-26 12:22:12 +02:00
354a4ce4e2 Add round_robin to example config. 2013-06-26 12:22:12 +02:00
f603e50cb3 Add support for round robin mode. 2013-06-26 12:22:11 +02:00
5c4c166802 Add DUMP_PROXY_CHAIN for debug builds and debug.c. 2013-06-25 12:56:41 +02:00
b9ca1cdefd proxychains.conf: add hint that proxy must be in ipv4 notation 2013-05-02 13:44:49 +02:00
64a7cd26dc fix case where proxy was using a DNS name or non-dotted ipv4
it is generally invalid to use a DNS name since DNS subsystem
is only available once connected to the proxy; because
DNS is done server-side.

closes #19
2013-05-02 13:44:49 +02:00
2aba83087e libproxychains.c: fix NULL pointer dereference after fopen().
if a configuration file can be accessed by proxychains but there is a
mandatory access control (or other) block on the target appication's
ability to read that file, fopen() creates a NULL pointer that will
cause a segfault in fgets().

closes #17

Signed-off-by: mancha <mancha1@hush.com>
2013-04-30 04:06:36 -04:00
f3af1239a9 better regex for obsolete proxyresolv script (thx yhzarcali)
closes #11
2013-01-29 18:43:21 +01:00
1da09e49e5 hook close() to prevent rude programs like ssh to close our pipes
those pipes are needed to talk with the dns-name allocator thread.

closes #9
2013-01-21 01:54:45 +01:00
173b90368e update proxyresolv 2013-01-06 19:20:44 +01:00
c8bfdc15e6 remove old proxychains launcher script 2013-01-06 19:12:15 +01:00
5526afb56d FreeBSD support 2012-12-25 19:08:05 +01:00
be4efc0fd5 fix no-newline warnings on old compilers 2012-12-25 18:01:11 +01:00
5ecd5ac51d fix compilation on musl libc 2012-12-18 09:38:32 +01:00
1c265b9628 getnameinfo: check size and family of salen 2012-12-17 23:21:58 +01:00
346474a43b getnameinfo: return error if buffers are too small 2012-12-17 22:41:51 +01:00
ce655fdac8 fix the never-ending issues with the wrong glibc prototype of getnameinfo
this bug was fixed shortly before 2.14 release, so we checked for that.
however some distros decided to backport this fix to earlier versions,
breaking our compiletime check.

http://sourceware.org/git/?p=glibc.git;a=commitdiff;h=e4ecafe004b3d4270b3a9dace8f970047400ed38

the portable solution is to stick the function into a separate comilation
unit that does not see the glibc prototype.

closes #7
2012-12-17 22:17:04 +01:00
af5c6f0c6a replace hostent lookup with better performing in-memory copy.
the central dns resolver function proxy_gethostbyname() used
to iterate over the gethostent() db (/etc/hosts) on each dns
request.
since this is not threadsafe, we synchronized access to it
previously using mutexes. the parsing of this file is slow,
and blocking all threads to do it even moreso.
since gethostent_r() is only available on a few platforms,
i decided to read the hostent db once and then use a quick
in-memory lookup on further usage.

+ some further refactoring.
2012-11-08 01:18:19 +01:00
c698d48c03 move stringfromipv4 into common.c 2012-11-07 22:00:03 +01:00
03ee84060e remove THREAD_SAFE ifdefs. from now on, pthreads are required.
additionally we have some explicit init and deinit routines for
core.c now, so that we dont need to share variables with
libproxychains.c.
2012-11-07 21:31:19 +01:00
bd07ca49b9 use pipe instead of pipe2 2012-11-07 21:31:18 +01:00
f570a66902 use mutexes also to protect the pipe communication
wasn't threadsafe before.
2012-11-07 21:31:18 +01:00
d0abc93c74 preliminary first fork-safe version
instead of allocating memory in the child, we now use the allocator
thread to do all the necessary allocations himself.
additionally we provide a clean API to query the ip <-> dns mapping.
these functions connect via a pipe to the allocator thread, and
exchange messages.

further cleanup is needed, but it seems to work so far.
thread-safety is not yet guaranteed.

closes #1
2012-11-07 21:31:11 +01:00
25afe98b20 failed attempt to use shared memory for the ip <-> dns mapping
this is in order to get irssi, which forks for DNS lookups,
and similar programs, to work as intended.

in a previous attempt i learned that shared memory created in a
child process is not visible to the parent;
in this attempt i spin off a thread from the parent which listens
on a pipe and manages the shared memory allocation from the parent
address-space. however this doesnt work as expected:
memory allocated in the parent after the child forked is not visi-
ble to the child as well.

so what happens is: irssi starts a child process, the thread allocs
memory and hands it to the child, the child attempts to write and
segfaults. however irssi doesnt crash. since now the memory is
already allocated, doing the dns lookup again will succeed.

i.e. the dns lookup works now in irssi by luck.
all but the first dns lookups will suceed.

however this is not good enough for me to be satisfied, i commit
this only for documentation purposes.
2012-11-07 16:49:14 +01:00
7bca3ba5ef main.c: fixed forgotten initialization for second loop using i
this caused the combination of -q -f somefile to not find the dll
in the current dir, because it started iterating the directory
list with 2 instead of 0.
2012-11-04 06:14:33 +01:00
e05cafc8e2 put a mutex lock around gethostent() usage, to prevent internal
races. the external usage was covered by the latest commit.
2012-11-04 05:23:51 +01:00
2d58820635 proxy_gethostbyname: fix thread safety issues arising from ...
gethostent() usage.

also set hostent.h_aliases member to a valid pointer.
2012-11-04 05:02:57 +01:00
b851b39a37 add RcB tags to core.h
this is handy as it allows me to do quick testprograms using my RcB program,
without specifying dozens of things on the command line.
2012-11-04 05:01:38 +01:00
e8d49b02e8 gethostbyaddr hook: also set aliases.
i inspected behaviour of libc's and they all seem to set a valid
h_aliases pointer, of which the first one should be NULL, if no
aliases exist.
2012-11-04 04:58:48 +01:00
859dabb302 common.h: supply a default for SYSCONFDIR if not passed via CFLAGS 2012-11-04 04:57:45 +01:00
bddb79a286 fix bug which lead to segfault in ubuntu 12.04 telnet
it called gethostbyname and expected that the h->h_name was set.

this code path here which gets only active if the passed name
equals the gethostname() result failed to set a couple of values.

additionally fixed usage of strncpy, which causes the entire
memory buffer to be written, even when the string is much shorter.

a similar bugfix was independently discovered by semion laptev, but my
version is threadsafe - his version introduces a new static buffer,
even though we have plenty of thread-safe space reserved exactly
for this usage case, thus causing a regression.
2012-11-04 01:00:58 +01:00
bbf8b7aab0 fix typo in help message 2012-08-27 23:30:45 +08:00
01b8c81e90 add some default localnets (commented) to config 2012-08-16 16:44:51 +02:00
923c4f1026 fix core.h to be usable on its own 2012-08-06 01:26:34 +02:00
108610e30e core.c: add getservbyname_r replacement function for mac 2012-08-06 01:26:03 +02:00
0f1bc719fd put base64 variable to the place it gets used 2012-08-04 17:55:37 +02:00
068e088a8b tunnel_to: use better suited type 2012-08-03 10:43:59 +02:00
40bb78b53f fix omission in common.c 2012-07-16 02:42:09 +02:00
80e58726e2 correct copyright of main.c 2012-07-16 02:36:35 +02:00