gmane.network.dns.c-ares

RELEASE: c-ares 1.10.0

Daniel Stenberg — 2013-05-12T13:55:21

Hi friends!

Almost eleven months since the previous release, I'm glad to announce a brand 
new c-ares tarball uploaded to the site: http://c-ares.haxx.se/

c-ares version 1.10.0

Changes:

  o Added ares_create_query(), to be used instead of ares_mkquery()
  o ares_inet_ntop() and ares_inet_pton() are now recognized c-ares functions

Bug fixes:

  o include the ares_parse_soa_reply.* files in the tarball
  o read_udp_packets: bail out loop on bad sockets
  o get_DNS_AdaptersAddresses: fix IPv6 parsing
  o adig: perror() doesn't work for socket errors on windows
  o ares_parse_aaaa_reply: fix memory leak
  o setup_once.h: HP-UX <sys/socket.h> issue workaround
  o configure: several fixes
  o config-dos.h: define strerror() to strerror_s_() for High-C
  o config-dos.h: define HAVE_CLOSE_S for MSDOS/Watt-32
  o ares_build.h.dist: enhance non-configure GCC ABI detection logic
  o ares.h: stricter CARES_EXTERN linkage decorations logic
  o ares_cancel(): cancel requests safely
  o protocol parsing: check input d

ares_set_servers_csv and compressed IPv6 string

Timothy J Fontaine — 2013-05-08T22:27:12

When passing in a compressed IPv6 string like "2001:4860:4860::8888" to
ares_set_servers_csv everything to the right of the first ":" in "::" is
stripped and interpreted as part of the port for the entry.

Is the intended behavior to require an IPv6 address in uncompressed
notation, or is this merely a bug?

On a tangent, is there any interest in actually supporting specifying the
port to be used? Conceptually I like the idea, but in practice I can't
imagine many people using it.

Fixing a few errors and warning detected by coverity

Patrick Valsecchi — 2013-05-02T06:25:35

Hi,

A colleague ran Coverity (a static analysis tool) against c-ares and 
fixed a few reported errors and warnings.

Please have a look at the attached patch.

Thanks.

Patch to build current head (3f0ec47) on msvc11

Alex Loukissas — 2013-05-01T17:44:13

Hi folks,

I'm a newcomer in the c-ares project and wanted to build it on my msvc11
(Visual C++ 2012), but I was failing to do so using the instructions in
https://github.com/bagder/c-ares/blob/master/Makefile.msvc

Since I figured out the issue, I thought I'd share the patch with you. Feel
free to use it.

Thanks
Alex

Proposal for unittests

Patrick Valsecchi — 2013-04-26T07:47:02

Hi,

In attachment, you'll find a small prototype of a unittest suite for 
c-ares. I didn't integrate it with the autotools stuff, since I don't 
know it (we use cmake). For the moment, it is only testing the 
ares_parse_txt_reply function.

I've used the simplest C testing framework I've found (CuTest) and I've 
fixed it a bit to remove a few memory leaks that were detected by valgrind.

What do you all think? Is the direction I'm taking OK with you?

Thanks and CU

soname

Jakub Hrozek — 2013-04-23T11:33:27

Hi,

I was notified[1] by another developer that the SONAME of libcares is
always reported as 2.0.0 despite the latest versions having a larger API.

For example on my RHEL5 test box the c-ares version is quite old but
still the soname is reported as 2.0.0:
# cat /etc/redhat-release 
Red Hat Enterprise Linux Server release 5.8 (Tikanga)
# rpm -qf /usr/lib64/libcares.so
c-ares-devel-1.6.0-5.el5
# objdump -p /usr/lib64/libcares.so | grep SONAME
SONAME      libcares.so.2
# ls /usr/lib64/libcares.so*
/usr/lib64/libcares.so  /usr/lib64/libcares.so.2 /usr/lib64/libcares.so.2.0.0

On Fedora 18 we are using pretty much the latest c-ares, but the SONAME
is still the same:
# cat /etc/redhat-release 
Fedora release 18 (Spherical Cow)
# rpm -qf /usr/lib64/libcares.so
c-ares-devel-1.9.1-2.fc18.x86_64
# objdump -p /usr/lib64/libcares.so | grep SONAME
SONAME               libcares.so.2
# ls /usr/lib64/libcares.so*
/usr/lib64/libcares.so  /usr/lib64/libcares.so.2
/usr/lib64/libcares.so.2.0.0

Even the version in git in Make

Parsing of 'Additional Records' in SRV response

ayush jain — 2013-04-22T05:32:56

Hi ,

I was just going through the c-ares API 'ares_parse_srv_reply' and was
wondering why it doesn't parse the 'additional record' section present in
the SRV response.
As in a multi-threaded architecture where you may want to minimize the
overall number of DNS queries per second for better performance, usage of
Additional Records can significantly reduce the need of 'A' queries or
reading the entries from /etc/hosts depending on the /etc/nsswitch.conf
file.
Any suggestions and/or comments are most welcome


Regards,
_Ayush

DNS servers on windows

Patrick Valsecchi — 2013-04-19T12:40:23

Hi,

I have a lot of trouble using c-ares on one of my windows boxes.

The machine has two interfaces:
   1) configured with a static IP address and it has DNS in automatic 
mode, but doesn't find any.
   2) configured by DHCP and it receives a valid DNS interface

My problem is that windows (at least 7), in all it's glory, invents DNS 
addesses for my first interface with 3 addresses taken from an outdated 
draft RFC [1]. And for some reasons, it shows them multiple times when 
GetAdaptersAddresses is called.

So I end up with 13 DNS server addresses:

  * fec0:0:0:ffff::1
  * fec0:0:0:ffff::2
  * fec0:0:0:ffff::3
  * 10.104.48.2               <- my actual server
  * fec0:0:0:ffff::1
  * fec0:0:0:ffff::2
  * fec0:0:0:ffff::3
  * 10.104.48.2
  * fec0:0:0:ffff::1
  * fec0:0:0:ffff::2
  * fec0:0:0:ffff::3

c-ares doesn't behave very well with such setup. At each query, it will 
start over with the first server, wait for a timeout of 5 seconds and 
then try the second one, ... . So I have to wait 15s for each q

ares_parse_txt_reply's output is not usable for DNS-SD

Patrick Valsecchi — 2013-04-15T08:29:29

Hi,

I'm using c-ares for doing DNS-SD. For that, I need to interpret TXT 
records formatted as specified in the DNS-SD spec [1]. Unfortunately, 
the ares_parse_txt_reply function just concatenates the sub-strings of 
the TXT record I get into a single string, loosing the information of 
the various key/value pairs contained in the TXT record.

To fix that, I see two solutions:

1) provide a second function that parses the TXT records that doesn't 
parse the length+value sub-strings and lets the caller do this part of 
the parsing.

2) change ares_parse_txt_reply to return another ares_txt_reply node for 
each sub-string.

I'd be happy to provide a patch for either solution.

While I'm writing to this mailing list, I've seen some discussions in 
the archives about doing an implementation for mDNS. What is the status 
on that?

Thanks for this nice library and have a nice day.

[1] http://tools.ietf.org/html/draft-cheshire-dnsext-dns-sd-11#section-6.1

Really fix memory leak when parsing AAAA replies. (#10) (fwd)

Daniel Stenberg — 2013-04-10T06:59:20

FYI

[Patch] ares_strcasecmp.h

Gisle Vanem — 2013-04-03T08:56:32

I think it would be best to decorate the following functions properly
in case they are to be exported from cares.dll:

--- Git-latest\ares_strcasecmp.h   Wed Jan 09 13:58:03 2013
+++ ares_strcasecmp.h        Thu Mar 10 22:31:33 2013
< at >< at > -17,14 +17,15 < at >< at >
  * without express or implied warranty.
  */

+#include "ares.h"
 #include "ares_setup.h"

 #ifndef HAVE_STRCASECMP
-extern int ares_strcasecmp(const char *a, const char *b);
+CARES_EXTERN int ares_strcasecmp(const char *a, const char *b);
 #endif

 #ifndef HAVE_STRNCASECMP
-extern int ares_strncasecmp(const char *a, const char *b, size_t n);
+CARES_EXTERN int ares_strncasecmp(const char *a, const char *b, size_t n);
 #endif

 #endif /* HEADER_CARES_STRCASECMP_H */

---------

All the a*.c samples could be using these functions.

--gv

[PATCH 1/4] ares_cancel(): ensure cancellation of all requests

Alexander Klauer — 2013-03-21T09:20:12

An invocation of ares_cancel() walks through the request list, calling
the callbacks of all pending requests on a channel. Previously, if such
a callback added a new request to the channel, the request list might
not end up empty, causing an abort by assertion failure. The present
commit ensures that all such newly added requests are cancelled
immediately and make it never into request list. Thus, the crash is
avoided, and it is made certain that upon return of ares_cancel(), there
are no requests whatsoever on the channel.
---
 ares.h               |    1 +
 ares_cancel.c        |    6 ++++++
 ares_gethostbyaddr.c |    6 ++++++
 ares_gethostbyname.c |    6 ++++++
 ares_getnameinfo.c   |    6 ++++++
 ares_private.h       |    3 +++
 ares_query.c         |    6 ++++++
 ares_search.c        |    6 ++++++
 ares_send.c          |    6 ++++++
 9 files changed, 46 insertions(+)

diff --git a/ares.h b/ares.h
index 9b3f376..8a9b995 100644
--- a/ares.h
+++ b/ares.h
< at >< at > -140,6 +140,7 < at >< at > extern "C" {
 #define ARES_FLAG_

Cosmetic bugs in http://c-ares.haxx.se/ares_parse_a_reply.html and http://c-ares.haxx.se/ares_parse_aaaa_reply.html

Alexander Klauer — 2013-03-21T08:41:31

Hi,

in the HTML pages mentioned in the subject, there are spurious '\fB' 
sequences in the synopses. Possibly they have to be changed to '\fP' in 
the source man pages, but I don't know enough roff to say for sure.

Best regards,
Alexander

[PATCH] library init: be recursive

Alexander Klauer — 2013-03-14T14:26:14

Previously, a single call to ares_library_cleanup() would deinitialise
the c-ares library, regardless of how many times ares_library_init() was
called. This behaviour may cause problems in programs linking two or
more libraries which, in turn, use c-ares. The present commit fixes this
problem, deinitializing the library only after a number of calls to
ares_library_cleanup() matching the number of calls to
ares_library_init().
---
 ares_library_init.c |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

diff --git a/ares_library_init.c b/ares_library_init.c
index f0137a1..770e7c2 100644
--- a/ares_library_init.c
+++ b/ares_library_init.c
< at >< at > -101,7 +101,10 < at >< at > int ares_library_init(int flags)
   int res;
 
   if (ares_initialized)
-    return ARES_SUCCESS;
+    {
+      ares_initialized++;
+      return ARES_SUCCESS;
+    }
   ares_initialized++;
 
   if (flags & ARES_LIB_INIT_WIN32)
< at >< at > -122,6 +125,8 < at >< at > void ares_library_cleanup(void)
   if (!ares_initialized)
     return;
   ares_initialized--;
+  i

where is ares_free_soa ?

Massimo — 2013-03-06T16:16:06

Hello all,

has anyone noticed that there is not ares_free_soa() anywhere in the code ?
There is the declaration in ares.h but the function is nowhere else.

I checked in the last release and in the last snapshot as well.

Am i missing something ?

Max

AVG false positive?

Gisle Vanem — 2013-02-24T13:40:56

Can anybody explain what happens in this NULL-case?

2013-02-22 19:49:05.226067 IP pcgv.2292 > google-public-dns-a.google.com.53: 64317+ A? 0.0.0.0.zz.countries.nerd.dk. (46)
2013-02-22 19:49:05.277156 IP google-public-dns-a.google.com.53 > pcgv.2292: 64317 NXDomain 0/1/0 (125)

My AVG-triggers an overzealous "treat warning". Attached picture from AVG.
What?

Win-XP SP3, MingW built acountry.exe from Git 3 days ago.

--gv

undocumented but used functions

Daniel Stenberg — 2013-02-13T14:00:13

Hi all,

I noticed that acountry.c and ahost.c, two of our demo examples, use 
ares_inet_ntop() and ares_inet_pton().

None of these functions are part of our documented public API and I would like 
to clean this up.

Anyone who feels strongly enough about this subject to speak up for either 
making them part of the API or for properly removing the use of them from the 
examples?

Assertion in ares_cancel.c on channel->queries_by_qid check

Evgeny Feoktistov — 2013-02-13T11:08:51

Hi,

I've got similar assertion fails during ares_cancel and ares_destroy calls:
Assertion `ares__is_list_empty(&(channel->queries_by_qid[i]))' failed
(../c-ares-1.9.1/ares_cancel.c:48)
Assertion `ares__is_list_empty(&(channel->queries_by_qid[i]))' failed
(../c-ares-1.9.1/ares_destroy.c:65)

Could it be due c-ares API misuse? I didn't find any problems in
queries_by_qid cache related logic. I use c-ares 1.9.1 with curl-7.19.6.

Problem on free() method on Windows 7

2013-02-13T09:48:55

Hello all,

I use c-ares 1.9.1 library in a program compiled with Visual C++ 6.0.
When I run the program on Windows XP, it's OK.
But, when I run the program on Windows 7, I have an error : "the program has
stopped working"...

By adding traces in my program, I noticed that the problem was on free() method.

For example:
_____________________________________________________________________
char *hostname = NULL;
...
int status = ares_expand_name (aptr, abuf, alen, &hostname, &len);
if (status != ARES_SUCCESS)
  return status;

if (hostname)
  free (hostname);<==> ERROR <==>
_____________________________________________________________________

Before doing free() on hostname, hostname is allocated in ares_expand_name()
method.
When I do a printf of "hostname" before calling free() method, hostname is OK,
it contain the good value.

Does someone has already encountered that problem ?
Or does someone has a solution to that problem ?

Thanks in advance,
Regards

c-ares patch

David Stuart — 2013-01-08T18:04:32

Hello,

In the method "get_iphlpapi_dns_info" inside of "ares_init.c", instead
of this:

    /* process the results */
    for( pEntry = pFirstEntry ; pEntry != NULL ; pEntry = pEntry->Next )
    {
      IP_ADAPTER_DNS_SERVER_ADDRESS* pDNSAddr = pEntry->FirstDnsServerAddress;
      for( ; pDNSAddr != NULL ; pDNSAddr = pDNSAddr->Next )
      {
        struct sockaddr *pGenericAddr = pDNSAddr->Address.lpSockaddr;
        int stringlen = 0;


Shouldn't it be:

    /* process the results */
    for( pEntry = pFirstEntry ; pEntry != NULL ; pEntry = pEntry->Next )
    {
      IP_ADAPTER_DNS_SERVER_ADDRESS* pDNSAddr = NULL;
      if( pEntry->OperStatus != IfOperStatusUp )
         continue;

      pDNSAddr = pEntry->FirstDnsServerAddress;
      for( ; pDNSAddr != NULL ; pDNSAddr = pDNSAddr->Next )
      {
        struct sockaddr *pGenericAddr = pDNSAddr->Address.lpSockaddr;
        int stringlen = 0;


We have a customer complaining about enabled (but not connected)
interfaces being used for their DNS configuration

'options single-request' support in c-ares?

Trey Duskin — 2012-12-10T21:37:31

Hello,

Is there currently any support in c-ares for the 'options single-request'
option in resolver.conf? (
http://man7.org/linux/man-pages/man5/resolver.5.html).  We are trying to
use libcurl for our system which could potentially be deployed behind some
restrictive firewalls, and the way that glibc sends parallel A and AAAA
queries from the same source port will cause an eventual timeout.  The
'option single-request' config option is supposed to fix this (sending the
two requests from two different source ports) for the system resolver, and
it does, but for anything using libcurl (e.g. curl command-line, pycurl),
the parallel queries come from the same source port, and we're stuck with
the timeout.  Can an option be set in c-ares to control this behavior?

Thanks in advance,
*
*
*Trey Duskin*