Re: [libvirt] [PATCH v2 8/9] virsocketaddr: Introduce virSocketAddrIsSameTCPHost
On 04/24/2015 04:25 AM, Peter Krempa wrote:
On Thu, Apr 23, 2015 at 16:12:16 -0400, John Ferlan wrote:
> Add new API to be able to compare two TCP host name strings or numeric IP
> Addresses in order to determine if the resolved and translated names match.
>
> Signed-off-by: John Ferlan <jferlan(a)redhat.com>
> ---
> src/libvirt_private.syms | 1 +
> src/util/virsocketaddr.c | 62 ++++++++++++++++++++++++++++++++++++++++++++++++
> src/util/virsocketaddr.h | 2 ++
> 3 files changed, 65 insertions(+)
>
> diff --git a/src/libvirt_private.syms b/src/libvirt_private.syms
> index 848b44a..34cb871 100644
> --- a/src/libvirt_private.syms
> +++ b/src/libvirt_private.syms
> @@ -2058,6 +2058,7 @@ virSocketAddrGetRange;
> virSocketAddrIsNetmask;
> virSocketAddrIsNumericLocalhost;
> virSocketAddrIsPrivate;
> +virSocketAddrIsSameTCPHost;
> virSocketAddrIsWildcard;
> virSocketAddrMask;
> virSocketAddrMaskByPrefix;
> diff --git a/src/util/virsocketaddr.c b/src/util/virsocketaddr.c
> index 993d460..c602604 100644
> --- a/src/util/virsocketaddr.c
> +++ b/src/util/virsocketaddr.c
> @@ -23,6 +23,7 @@
>
> #include <config.h>
>
> +#include "viralloc.h"
> #include "virsocketaddr.h"
> #include "virerror.h"
> #include "virstring.h"
> @@ -980,3 +981,64 @@ virSocketAddrIsNumericLocalhost(const char *addr)
>
> return false;
> }
> +
> +
> +/**
> + * virSocketAddrIsSameTCPHost:
> + * @host1: either a numeric network address IPv4 or IPv6 or a network hostname
> + * @host2: either a numeric network address IPv4 or IPv6 or a network hostname
> + *
> + * For each hostname, get the array of resolved addresses and in a loop make
> + * compare the translated host IP Address looking for duplicates.
> + *
> + * Returns 1 if the hosts match, 0 if there is no match, and -1 on error
> + */
> +int
> +virSocketAddrIsSameTCPHost(const char *host1,
> + const char *host2)
> +{
> + int ret = -1;
> + struct addrinfo *reshost1 = NULL, *reshost2 = NULL;
> + struct addrinfo *curhost1, *curhost2;
> + char *host1rslv = NULL, *host2rslv = NULL;
> +
> + if (virSocketAddrParseInternal(&reshost1, host1, AF_UNSPEC, IPPROTO_TCP,
> + false, true) < 0)
> + goto cleanup;
> +
> + if (virSocketAddrParseInternal(&reshost2, host2, AF_UNSPEC, IPPROTO_TCP,
> + false, true) < 0)
> + goto cleanup;
> +
> + for (curhost1 = reshost1; curhost1; curhost1 = curhost1->ai_next) {
> + if (!(host1rslv = virSocketAddrGetNumericHost(curhost1->ai_addr,
> + curhost1->ai_addrlen,
> + false, NULL)))
> + goto cleanup;
> + for (curhost2 = reshost2; curhost2; curhost2 = curhost2->ai_next) {
> + if (curhost1->ai_family != curhost2->ai_family)
> + continue;
> +
> + if (!(host2rslv = virSocketAddrGetNumericHost(curhost2->ai_addr,
> + curhost2->ai_addrlen,
> + false, NULL)))
> + goto cleanup;
> +
> + if (STREQ(host1rslv, host2rslv)) {
I'm still failing to see why the host address check portion of
virSocketAddrEqual() isn't good for this job and you have to format the
address as string.
Because I was more focused on the implementation that I had in my mind
that did the getaddrinfo/getnameinfo loops. Many times the mind is a
very difficult thing to convince to do something a different way - about
to send an adjustment,
> + ret = 1;
> + goto cleanup;
> + }
> + }
Next iteration leaks host2rslv.
> + }
Next iteration leaks host1rslv.
Oh yeah ... right, but they won't matter with the adjustment.
John
> +
> + ret = 0;
> +
> + cleanup:
> + if (reshost1)
> + freeaddrinfo(reshost1);
> + if (reshost2)
> + freeaddrinfo(reshost2);
> + VIR_FREE(host1rslv);
> + VIR_FREE(host2rslv);
> + return ret;
> +}
Peter