public inbox for libc-help@sourceware.org
 help / color / mirror / Atom feed
* getaddrinfo() returns two identical results
@ 2018-01-05 17:39 Yuri Kanivetsky
  2018-01-06 22:54 ` Yuri Kanivetsky
  0 siblings, 1 reply; 2+ messages in thread
From: Yuri Kanivetsky @ 2018-01-05 17:39 UTC (permalink / raw)
  To: libc-help

Hi,

Consider this test program:


    #include <string.h>
    #include <stdio.h>
    #include <stdlib.h>
    #include <sys/types.h>
    #include <sys/socket.h>
    #include <netdb.h>
    #include <netinet/in.h>

    const char* family_to_string(int family)
    {
        return family == AF_INET ? "AF_INET"
            : family == AF_INET6 ? "AF_INET6"
            : "<unknown>";
    }

    const char* socktype_to_string(int socktype)
    {
        return socktype == SOCK_STREAM ? "SOCK_STREAM"
            : socktype == SOCK_DGRAM ? "SOCK_DGRAM"
            : "<unknown>";
    }

    const char* protocol_to_string(int protocol)
    {
        return protocol == IPPROTO_TCP ? "IPPROTO_TCP"
            : protocol == IPPROTO_UDP ? "IPPROTO_UDP"
            : "<unknown>";
    }

    void print_addrinfo(struct addrinfo *ai)
    {
        int r;
        char hbuf[NI_MAXHOST], sbuf[NI_MAXSERV];
        r = getnameinfo(ai->ai_addr, ai->ai_addrlen, hbuf, sizeof(hbuf), sbuf,
                        sizeof(sbuf), NI_NUMERICHOST | NI_NUMERICSERV);
        if (r)
        {
            fputs("getnameinfo\n", stderr);
            exit(EXIT_FAILURE);
        }
        const char *family = family_to_string(ai->ai_family);
        const char *socktype = socktype_to_string(ai->ai_socktype);
        const char *protocol = protocol_to_string(ai->ai_protocol);
        const char *family2 = family_to_string(ai->ai_addr->sa_family);
        printf("flags=%i, family=%s, socktype=%s, protocol=%s,
family=%s, host=%s, serv=%s\n",
               ai->ai_flags, family, socktype, protocol, family2, hbuf, sbuf);
    }

    int main(int argc, char *argv[])
    {
        (void)argc;
        (void)argv;
        int r;
        struct addrinfo hints;
        struct addrinfo *result, *ai;

        memset(&hints, 0, sizeof(struct addrinfo));
        hints.ai_family = AF_INET;
        hints.ai_socktype = SOCK_STREAM;
        r = getaddrinfo("localhost", "8000", &hints, &result);
        if (r)
        {
            fputs("getaddrinfo\n", stderr);
            exit(EXIT_FAILURE);
        }

        ai = result;
        while (ai)
        {
            print_addrinfo(ai);
            ai = ai->ai_next;
        }

        exit(EXIT_SUCCESS);
    }


In /etc/hosts I have:


    127.0.0.1 localhost.localdomain localhost
    ::1 localhost.localdomain localhost


When I set `hints.ai_family = AF_INET`, I get:


    $ gcc -Wall -Wextra -pedantic -pedantic-errors -Werror -g
-gdwarf-2 -g3 % && ./a.out
    flags=0, family=AF_INET, socktype=SOCK_STREAM,
protocol=IPPROTO_TCP, family=AF_INET, host=127.0.0.1, serv=8000
    flags=0, family=AF_INET, socktype=SOCK_STREAM,
protocol=IPPROTO_TCP, family=AF_INET, host=127.0.0.1, serv=8000


Without it:


    flags=0, family=AF_INET6, socktype=SOCK_STREAM,
protocol=IPPROTO_TCP, family=AF_INET6, host=::1, serv=8000
    flags=0, family=AF_INET, socktype=SOCK_STREAM,
protocol=IPPROTO_TCP, family=AF_INET, host=127.0.0.1, serv=8000


From what I can see, `hints` parameter is supposed to filter results.
Which I cannot reproduce on my machine.

Not sure which info can be of help here. Feel free to ask. I'm running
Arch Linux and glibc-2.26. I haven't used systemd-resolved, and it's
not running.

/etc/resolv.conf:


    # Generated by resolvconf
    search Dlink
    nameserver 192.168.0.1


/etc/nsswitch.conf:


    # Begin /etc/nsswitch.conf

    passwd: compat mymachines systemd
    group: compat mymachines systemd
    shadow: compat

    publickey: files

    hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
    networks: files

    protocols: files
    services: files
    ethers: files
    rpc: files

    netgroup: files

    # End /etc/nsswitch.conf


I'm now building glibc with debugging information, hopefully it'll
shed some light on the matter. But if you've got any ideas what to
check, or possible explanation, feel free to reply.

P.S. Relevant Stack Overflow question:
https://stackoverflow.com/questions/48100853/getaddrinfo-returns-several-identical-results

Regards,
Yuri

^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: getaddrinfo() returns two identical results
  2018-01-05 17:39 getaddrinfo() returns two identical results Yuri Kanivetsky
@ 2018-01-06 22:54 ` Yuri Kanivetsky
  0 siblings, 0 replies; 2+ messages in thread
From: Yuri Kanivetsky @ 2018-01-06 22:54 UTC (permalink / raw)
  To: libc-help

Hi,

Okay, I more or less see now:


    int
    getaddrinfo (const char *name, const char *service,
                 const struct addrinfo *hints, struct addrinfo **pai)
    https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/posix/getaddrinfo.c;h=2c4b6d6793a4c3a96eb1e545570ac28867681751;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l2206

    last_i = gaih_inet (name, pservice, hints, end, &naddrs, &tmpbuf);
    https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/posix/getaddrinfo.c;h=2c4b6d6793a4c3a96eb1e545570ac28867681751;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l2304

    static int
    gaih_inet (const char *name, const struct gaih_service *service,
               const struct addrinfo *req, struct addrinfo **pai,
               unsigned int *naddrs, struct scratch_buffer *tmpbuf)
    https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/posix/getaddrinfo.c;h=2c4b6d6793a4c3a96eb1e545570ac28867681751;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l342

    rc = __gethostbyname2_r (name, AF_INET, &th,
    tmpbuf->data, tmpbuf->length,
    &h, &h_errno);
    https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/posix/getaddrinfo.c;h=2c4b6d6793a4c3a96eb1e545570ac28867681751;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l595

    int
    INTERNAL (REENTRANT_NAME) (ADD_PARAMS, LOOKUP_TYPE *resbuf, char *buffer,
                               size_t buflen, LOOKUP_TYPE **result H_ERRNO_PARM
                               EXTRA_PARAMS)
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/getXXbyYY_r.c;h=bce80e05dd2d176467809faf353828f50cf76eaf;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l189

    status = DL_CALL_FCT (fct.l, (ADD_VARIABLES, resbuf, buffer, buflen,
      &errno H_ERRNO_VAR EXTRA_VARIABLES));
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/getXXbyYY_r.c;h=bce80e05dd2d176467809faf353828f50cf76eaf;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l316

    enum nss_status
    _nss_files_gethostbyname2_r (const char *name, int af, struct
hostent *result,
                                 char *buffer, size_t buflen, int *errnop,
                                 int *herrnop)
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l385

    return _nss_files_gethostbyname3_r (name, af, result, buffer, buflen,
    errnop, herrnop, NULL, NULL);
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l389

    enum nss_status
    _nss_files_gethostbyname3_r (const char *name, int af, struct
hostent *result,
                                 char *buffer, size_t buflen, int *errnop,
                                 int *herrnop, int32_t *ttlp, char **canonp)
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l334

Here we get first match, and—if multi is on in /etc/host.conf—the rest of them:

    while ((status = internal_getent (stream, result, buffer, buflen, errnop,
                                      herrnop, af, flags))
           == NSS_STATUS_SUCCESS)
      {
        LOOKUP_NAME_CASE (h_name, h_aliases)
      }

    if (status == NSS_STATUS_SUCCESS
        && _res_hconf.flags & HCONF_FLAG_MULTI)
      status = gethostbyname3_multi
        (stream, name, af, result, buffer, buflen, errnop, herrnop, flags);
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l352

    static enum nss_status
    gethostbyname3_multi (FILE * stream, const char *name, int af,
                          struct hostent *result, char *buffer, size_t buflen,
                          int *errnop, int *herrnop, int flags)
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l127

    status = internal_getent (stream, &tmp_result_buf, tmp_buffer.data,
      tmp_buffer.length, errnop, herrnop, af,
      flags);
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l162

    static enum nss_status
    internal_getent (FILE *stream, struct STRUCTURE *result,
                     char *buffer, size_t buflen, int *errnop H_ERRNO_PROTO
                     EXTRA_ARGS_DECL)
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-XXX.c;h=265331ef21cd60ed301ab49eac4d6681d74bb388;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l178

    || ! (parse_result = parse_line (p, result, data, buflen, errnop
    EXTRA_ARGS)));
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-XXX.c;h=265331ef21cd60ed301ab49eac4d6681d74bb388;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l222

Here we parse "::1     localhost.localdomain   localhost" line. We try
to run inet_pton over "::1". That fails, since af == AF_INET. Then we
notice that "::1" is IPv6 loopback address, so we return IPv4 loopback
address:

    LINE_PARSER
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l51

    if (inet_pton (af == AF_UNSPEC ? AF_INET : af, addr, entdata->host_addr)
        > 0)
      af = af == AF_UNSPEC ? AF_INET : af;
    else
      {
        if (...)
      ...
        else if (af == AF_INET
                 && inet_pton (AF_INET6, addr, entdata->host_addr) > 0)
          {
            if (...)
        ...
            else if (IN6_IS_ADDR_LOOPBACK (entdata->host_addr))
              {
                in_addr_t localhost = htonl (INADDR_LOOPBACK);
                memcpy (entdata->host_addr, &localhost, sizeof (localhost));
              }
    https://sourceware.org/git/?p=glibc.git;a=blob;f=nss/nss_files/files-hosts.c;h=6f7cc4d94ba69ab58d684f63cc685f6fa5c6ad4a;hb=633e2f7f3d88df6427aa3a7a984d3a6b796d9611#l59


Can you explain why does not fail in this case? For what it's needed?

Also, it doesn't look like a major issue, but the behavior doesn't
seem to be expected. Should I report a bug?

Regards,
Yuri

^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2018-01-06 22:54 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-01-05 17:39 getaddrinfo() returns two identical results Yuri Kanivetsky
2018-01-06 22:54 ` Yuri Kanivetsky

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).