Commit Graph

245 Commits

Author SHA1 Message Date
Dan Fandrich
4adcf7e9ed Moved all signal-based name resolution timeout handling into a single new
Curl_resolv_timeout function to reduce coupling.
2008-09-29 21:02:22 +00:00
Yang Tse
4e909ee8b1 ntoa() and inet_ntoa_r() no longer used 2008-09-24 12:22:16 +00:00
Yang Tse
a622fd90b4 remove unnecessary typecasting of calloc() 2008-09-06 04:47:14 +00:00
Dan Fandrich
0994d7811f Removed some redundant type casts 2008-09-02 18:36:39 +00:00
Daniel Stenberg
0cd8840dba - Andreas Schuldei improved Phil Blundell's patch for IPv6 using c-ares, and I
edited it slightly. Now you should be able to use IPv6 addresses fine even
  with libcurl built to use c-ares.
2008-07-09 18:39:49 +00:00
Daniel Stenberg
b9d66dca51 - Dmitry Popov filed bug report #1911069
(http://curl.haxx.se/bug/view.cgi?id=1911069) that identified a race
  condition in the name resolver code when the DNS cache is shared between
  multiple easy handles, each running in simultaneous threads that could cause
  crashes.
2008-03-11 22:55:23 +00:00
Daniel Stenberg
991505e077 Woops, partly revert my previous commit and do it slightly differently instead.
The signalling of that a global DNS cache is wanted is done by setting the
option but the setting of the internal variable that it is in use must not be
done until it finally actually gets used!

NOTE and WARNING: I noticed that you can't actually switch off the global dns
cache with CURLOPT_DNS_USE_GLOBAL_CACHE but you couldn't do that previously
either and the option is very clearly and loudly documented as DO NOTE USE so
I won't bother to fix this bug now.
2008-01-15 22:44:12 +00:00
Daniel Stenberg
56f17d2c9f I made the torture test on test 530 go through. This was actually due to
silly code left from when we switched to let the multi handle "hold" the dns
cache when using the multi interface... Of course this only triggered when a
certain function call returned error at the correct moment.
2008-01-15 22:15:55 +00:00
Daniel Stenberg
cbd1a77ec2 if () => if()
while () => while()
and some other minor re-indentings
2007-11-07 09:21:35 +00:00
Dan Fandrich
8388366849 Renamed a variable to avoid shadowing a global declarations. 2007-09-28 18:47:59 +00:00
Dan Fandrich
523767660c Fixed some minor mismatched types found by splint. 2007-08-26 05:53:26 +00:00
Daniel Stenberg
54967d2a3a Thomas J. Moore provided a patch that introduces Kerberos5 support in
libcurl. This also makes the options change name to --krb (from --krb4) and
CURLOPT_KRBLEVEL (from CURLOPT_KRB4LEVEL) but the old names are still
2007-07-01 22:01:18 +00:00
Daniel Stenberg
62f0f5571d Robert Iakobashvili re-arranged the internal hash code to work with a custom
hash function for different hashes, and also expanded the default size for
the socket hash table used in multi handles to greatly enhance speed when
very many connections are added and the socket API is used.
2007-06-26 21:09:28 +00:00
Yang Tse
94b253fde7 Steve Little's fixes to allow compilation on VMS 64-bit mode 2007-04-25 03:00:10 +00:00
Gisle Vanem
c514a2a89a Removed inclusion of <sys/types.h> and <sys/stat.h> in .c-files
since they're already included through "setup.h".
2007-02-26 04:24:26 +00:00
Yang Tse
3a634a273a curlassert macro replaced with DEBUGASSERT macro defined in setup_once.h 2007-02-21 19:03:20 +00:00
Dan Fandrich
77516822f6 Workaround for Cray UNICOS 9.0 to fix ftp. 2006-08-21 22:28:19 +00:00
Dan Fandrich
839441e236 Minor portability fixes to get things running on UNICOS 9.0 on a Cray Y-MP 2006-08-16 18:48:27 +00:00
Yang Tse
ecfaa4f869 Fix warning: no newline at end of file 2006-07-25 18:48:12 +00:00
Yang Tse
13616f8f96 Simplify check for NEED_MALLOC_H, and make more explicit that NEED_MALLOC_H shall be defined if <malloc.h> header file must be included even when including <stdlib.h>. 2006-07-25 13:49:49 +00:00
Gisle Vanem
e4d6ade4b3 Moved functions common to IPv4 and C-ares to hostip.c;
Curl_freeaddrinfo() and Curl_ip2addr().
2006-07-25 10:31:31 +00:00
Gisle Vanem
c82e880f5b Remove comment about c-ares not supporting IPv6. 2006-07-25 10:23:15 +00:00
Gisle Vanem
5cdbd0cf4a Constify arguments to Curl_he2ai() and Curl_addrinfo_copy(). 2006-07-21 06:21:46 +00:00
Gisle Vanem
a55c70d4ae Constify 'hostname' and 'service' to various resolver functions. 2006-07-21 05:51:12 +00:00
Yang Tse
fe22872d14 include <malloc.h> only if HAVE_MALLOC_H and NEED_MALLOC_H are both defined. 2006-07-11 21:34:23 +00:00
Daniel Stenberg
ca319f63ad Ingmar Runge provided a source snippet that caused a crash. The reason for
the crash was that libcurl internally was a bit confused about who owned the
DNS cache at all times so if you created an easy handle that uses a shared
DNS cache and added that to a multi handle it would crash. Now we keep more
careful internal track of exactly what kind of DNS cache each easy handle
uses: None, Private (allocated for and used only by this single handle),
Shared (points to a cache held by a shared object), Global (points to the
global cache) or Multi (points to the cache within the multi handle that is
automatically shared between all easy handles that are added with private
caches).
2006-07-07 22:58:06 +00:00
Daniel Stenberg
e85e30546c Roland Blom filed bug report #1481217
(http://curl.haxx.se/bug/view.cgi?id=1481217), with follow-ups by Michele Bini
and David Byron. libcurl previously wrongly used GetLastError() on windows to
get error details after socket-related function calls, when it really should
use WSAGetLastError() instead.

When changing to this, the former function Curl_ourerrno() is now instead
called Curl_sockerrno() as it is necessary to only use it to get errno from
socket-related functions as otherwise it won't work as intended on Windows.
2006-05-04 22:39:47 +00:00
Gisle Vanem
3cbb1b2b64 Use the HAVE_MALLOC_H and HAVE_PROCESS_H defines
(more logical).
2006-04-26 17:23:28 +00:00
Daniel Stenberg
92009181af Shmulik Regev provided a fix for the DNS cache when using short life times,
as previously it could be holding on to old cached entries longer than
requested.
2006-02-16 23:42:32 +00:00
Daniel Stenberg
5f0366c2cb only define _REENTRANT if not already defined, and only in setup.h 2005-04-19 23:19:23 +00:00
Daniel Stenberg
6a2e21ec8c FTP code turned into state machine. Not completely yet, but a good start.
The tag 'before_ftp_statemachine' was set just before this commit in case
of future need.
2005-02-09 13:06:40 +00:00
Daniel Stenberg
043d70fcdf Use plain structs and not typedef'ed ones in the hash and linked-list code. 2005-01-25 00:06:29 +00:00
Daniel Stenberg
dcea109bb5 Dan Fandrich fix: eliminates some pedantic CodeWarrior compiler warnings and
errors.
2004-11-18 14:04:40 +00:00
Daniel Stenberg
6d1b37b3da Prevent a longjmp warning by moving the rc assign within Curl_resolv().
Andy Cedilnik reported. Warning on HP-UX?
2004-10-10 03:32:01 +00:00
Daniel Stenberg
39af394a1c removed tabs and trailing whitespace from source 2004-10-06 07:50:18 +00:00
Daniel Stenberg
95def48071 Made the dns entry remain locked while a connection to the host remains to
allow verbose output during this period. Bertrand Demiddelaer reported and
helped fixing.
2004-10-04 10:36:51 +00:00
Daniel Stenberg
f2a99d7d74 the _num_chars() function is not used, removing 2004-06-24 15:05:39 +00:00
Daniel Stenberg
feb2dd2835 Replaced all uses of sprintf() with the safer snprintf(). It is just a
precaution to prevent mistakes to lead to buffer overflows.
2004-06-24 11:54:11 +00:00
Daniel Stenberg
5e34f3dc01 made the Curl_he2ai() take the port number as an int intead, to avoid lots
of typecasts all over
2004-06-24 10:43:50 +00:00
Daniel Stenberg
560c257bd0 prevent warning 2004-06-24 08:08:28 +00:00
Daniel Stenberg
c39858aac0 Source cleanups. The major one being that we now _always_ use a Curl_addrinfo
linked list for name resolved data, even on hosts/systems with only IPv4
stacks as this simplifies a lot of code.
2004-06-24 07:43:48 +00:00
Daniel Stenberg
8d76d4016d removed trailing whitespace 2004-06-10 11:56:02 +00:00
Daniel Stenberg
9f92657f76 Gisle corrected two comments 2004-06-10 11:55:37 +00:00
Daniel Stenberg
9f341f9ce5 Gisle Vanem's improved verbose output and timeout handling when connecting to
a host name that resolves to multiple IP addresses.
2004-06-10 11:06:21 +00:00
Daniel Stenberg
e9056f5f95 if shrinking the buffer fails, use the older larger one 2004-05-17 22:07:43 +00:00
Daniel Stenberg
bbafb2eb27 curl_global_init_mem() allows the memory functions to be replaced.
memory.h is included everywhere for this.
2004-05-11 11:30:23 +00:00
Daniel Stenberg
4250637e7d improved the cleaning up of memory when we fail to resolve names due to
out of memory (thanks to 'runtests.pl -t')
2004-05-04 13:40:30 +00:00
Daniel Stenberg
245e3122df Made defines instead of plain numbers for the Curl_resolv() return code to
make the code easier to read
2004-04-26 14:18:00 +00:00
Daniel Stenberg
648e82f05d Major hostip.c cleanup and split into multiple files and easier #ifdef
usage.
2004-04-26 07:20:11 +00:00
Daniel Stenberg
3755bffcc2 Gisle Vanem found and fixed a memory leak when doing (failing) Windows
threaded name resolves.
2004-04-23 14:04:30 +00:00
Daniel Stenberg
62b7c08bb3 include unistd.h as well for the close() proto on some platforms (like Tru64) 2004-04-21 11:18:42 +00:00
Daniel Stenberg
dc25cd6f3a cleanup leftovers 2004-04-20 09:46:10 +00:00
Daniel Stenberg
68e8a0f0d9 asking for CURL_IPRESOLVE_V6 when ipv6 addresses can't be resolved will
now cause the resolve function to return NULL immediately
2004-04-14 12:10:44 +00:00
Daniel Stenberg
1d0b5b507a Curl_wait_for_resolv() could hang due to the bad timeout timer resolution and
some bad thinking on my part.
2004-04-14 11:43:26 +00:00
Daniel Stenberg
31a693b99a Gisle Vanem's fix that makes the multi interface work on Windows again even
when not using ares.
2004-04-13 13:59:12 +00:00
Daniel Stenberg
eb6345de60 somewhat safer typecasting in case sizeof(long) != sizeof(void *) (is there
even such platforms?)
2004-04-12 06:55:25 +00:00
Daniel Stenberg
5ca47f19d7 the pack_hostent() proto isn't used/needed with ipv6 is enabled.
time to restructure this source file!
2004-04-06 15:09:43 +00:00
Daniel Stenberg
9044fcbb5f Gisle Vanem's fix for bug item #927979 reported by Nathan O'Sullivan.
Good enough?
2004-04-06 14:51:14 +00:00
Daniel Stenberg
ad9e83a90f Dirk Manske's fix that makes sure we cancel the ares resolve when we time out
from a name resolve. Without this, we leak memory!
2004-04-01 08:40:36 +00:00
Daniel Stenberg
675db3a211 removed my previously attempted fix for ares timeouts, not needed 2004-04-01 07:04:58 +00:00
Daniel Stenberg
ee1595dcd5 Roy Shan fixed a case that prevented ares name resolve timeouts to occur. 2004-03-31 20:50:01 +00:00
Daniel Stenberg
9d0330d5bd Remove the elapsed time from the most recent select() only. 2004-03-31 13:19:41 +00:00
Daniel Stenberg
d5074f74bb The asynch name resolve methods now all use CURL_TIMEOUT_RESOLVE for
the specific time to wait for a resolve. The definition is at the top of
this source file.
2004-03-31 12:55:24 +00:00
Daniel Stenberg
ea0cf7c87b Dirk Manske found out the Curl_wait_for_resolv() timed out too early. 2004-03-31 12:45:26 +00:00
Daniel Stenberg
b8b8473b6d Lots of comments added an clarified. Added timeout for the ares version
of Curl_is_resolved() to address Roy Shan's reported problem.
2004-03-30 13:02:07 +00:00
Daniel Stenberg
dc46f535ae The select() timeout is better not static since some implementation actually
might change it. I don't *think* it does it when the timeout is 0,0 but it
is better to be sure...
2004-03-29 21:29:24 +00:00
Daniel Stenberg
5b55f9ecb3 Gnter Knauf's NetWare changes. 2004-03-17 12:46:42 +00:00
Daniel Stenberg
47afa058a8 store times in time_t 2004-03-10 09:36:43 +00:00
Daniel Stenberg
52d85341ae only build with the windows threading trace code if
DEBUG_THREADING_GETHOSTBYNAME is defined
2004-03-09 21:42:43 +00:00
Daniel Stenberg
9116dd41bd tv_sec is an int, so we explicitly typecast the result of long - long to
an int when we assign it.
2004-03-03 13:07:32 +00:00
Daniel Stenberg
f35c28214e use size_t to keep strlen() results 2004-02-23 08:38:49 +00:00
Daniel Stenberg
35fd1365ae simplied how create_hostcache_id() is used, and also its function somewhat
cleared up some ssize_t/size_t mixups
2004-02-23 08:35:57 +00:00
Daniel Stenberg
7d844c430c Gisle Vanem brings name resolving timeout possibilities to windows people.
This works by magicly starting up a new thread that can be killed when the
timeout is reached. testtesttest!
2004-02-20 16:41:50 +00:00
Daniel Stenberg
62a12b7faf No longer uses the 'ret' variable in the plain ipv4-version of
my_getaddrinfo() (caused a warning by the IRIX MIPSPro compiler). Also
clarified the situation for the 3-arg version of gethostbyname_r() with a huge
comment.
2004-02-18 10:05:17 +00:00
Daniel Stenberg
e2b6a45cfa Make sure dns cache timeout -1 really means forever, as it is documented to
be. Simply skip the pruning.
2004-02-18 07:56:18 +00:00
Daniel Stenberg
ccdaa0b51f Make the 'areschannel' get created in the curl_easy_init() and re-use that
same channel during the whole curl handle's life until curl_easy_cleanup().
2004-02-16 15:24:22 +00:00
Daniel Stenberg
a7e04cac78 Mark the dns entry 'inuse' properly even when used from the cache. This
seems to correct some host cache screw-ups I could reproduce.
2004-02-15 16:57:53 +00:00
Daniel Stenberg
4674f3f115 Make hostcache_fixoffset() take a long for offset, to fully work with 64bit
archs, also no longer typecast pointers to ints as that is a nono on 64bit
systems.
2004-02-12 16:02:55 +00:00
Daniel Stenberg
791e49134b hide the pack_hostent proto if ipv6 is enabled, as figured out by Tor Arntsen 2004-02-05 08:34:31 +00:00
Daniel Stenberg
886391d145 adjusted to the modified ares_strerror() function
NOTE that this breaks ares-compatibility, we have now officially taken the
turn into the c-ares path. We will now officially depend on c-ares for asynch
name resolves.
2004-02-02 16:00:31 +00:00
Daniel Stenberg
e1d541086c Dirk Manske fixed the ares usage even more. We could get a timeout from ares
as well, and when failing and not getting a timeout we now include the
error message ares can provide us with.
2004-02-02 15:53:50 +00:00
Daniel Stenberg
12fa877f00 Timeout slow ares name lookups. This is based on the patch brought by
Dirk Manske, but modified by me.
2004-02-02 10:15:22 +00:00
Daniel Stenberg
2d567af069 fixed the Curl_resolv()'s return code when the looked up host was already
in the cache (Vincent Bronner)
2004-01-14 09:05:54 +00:00
Daniel Stenberg
053f6c85ef updated year in the copyright string 2004-01-07 09:19:33 +00:00
Daniel Stenberg
3788f25eac added a library-wide interface for doing dns cache pruning, and no longer
make the pruning at each name lookup, only in Curl_done().
2003-12-15 17:33:49 +00:00
Daniel Stenberg
cfe53768dd if Curl_hash_add() returns NULL, we shall not free the addrinfo field as that
is made in the hash function in the case of failure (using the already setup
'dtor' function).
2003-12-15 15:22:10 +00:00
Daniel Stenberg
06c38330ba Steve Green fixed Curl_resolv() 2003-12-03 07:55:51 +00:00
Daniel Stenberg
a0a5668dbd do ares_destroy() on the ares handle when we're done in Curl_is_resolved() 2003-11-15 10:29:43 +00:00
Daniel Stenberg
a701f3dddf Siddhartha Prakash Jain found a case with a bad resolve that we didn't
properly bail out from, when using ares.
2003-11-14 11:56:15 +00:00
Daniel Stenberg
25613503cb Another glibc resolve name fix 2003-10-28 13:06:15 +00:00
Daniel Stenberg
017a78d0c5 better bailing-out cleanup if a malloc fails in the DNS cache 2003-10-24 12:57:23 +00:00
Daniel Stenberg
31b6bb6a41 lock the DNS cache properly before adding an entry when using asynch DNS 2003-10-20 08:45:33 +00:00
Daniel Stenberg
749f5387c1 Gisle Vanem's IPv6-on-Windows patch applied! 2003-10-14 12:00:45 +00:00
Daniel Stenberg
022996e5c7 Dirk Manske made the share-locking around DNS lookups a bit "looser" so that
multiple DNS lookups can run simultaneously faster. The downside is that
resolving the same host name now can be made at once from multiple threads,
but the upside is that threads now don't alwys have to wait for the others'
resolves. Test case 506 updated accordingly.
2003-10-12 13:57:41 +00:00
Daniel Stenberg
a0118fb390 just re-indented some code 2003-10-05 15:03:21 +00:00
Daniel Stenberg
343291ce37 Based on a patch provided by Siddhartha Prakash Jain. In Curl_resolv() when
my_getaddrinfo() has been called (and wait has been set to TRUE), we check
if the name already is resolved and if so don't return wait status to the
parent. This can happen with IP-only names.
2003-10-04 14:50:45 +00:00
Daniel Stenberg
2297eca103 Added CURLOPT_IPRESOLVE support 2003-09-19 15:16:47 +00:00
Daniel Stenberg
14597475b1 Jeff Pohlmeyer did some marvelous debugging to track this one down. We MUST
NOT free the existing hash entry when we try to add a new one that matches
an existing entry. We now instead free the new one, and make the parent
function use the old entry's struct instead.
2003-09-14 21:17:54 +00:00
Daniel Stenberg
c1f31180a2 Curl_is_resolved(): FD_ZERO the file descriptors before we call ares_fds().
Problem tracked down by Bjorn Reese.
2003-09-11 21:27:19 +00:00
Daniel Stenberg
5629eeaec2 The error buffer was not getting filled when Curl_wait_for_resolv() fails.
Jeff Pohlmeyer fixed.
2003-09-01 08:21:08 +00:00