Re: [PATCH 11/11] exportfs: only do glibc specific hackery on glibc

[Steve Dickson <SteveD@redhat.com>]

On 07/30/2014 07:23 AM, Natanael Copa wrote:
> We should not depend on the libc do free(3) on ai_canonname as that is
> completely up to implementation and known o break things on uclibc and
> musl libc.
> 
> Signed-off-by: Natanael Copa <ncopa@alpinelinux.org>
> ---
>  support/export/hostname.c | 3 +++
>  1 file changed, 3 insertions(+)
> 
> diff --git a/support/export/hostname.c b/support/export/hostname.c
> index d9153e1..30584b4 100644
> --- a/support/export/hostname.c
> +++ b/support/export/hostname.c
> @@ -382,6 +382,7 @@ host_numeric_addrinfo(const struct sockaddr *sap)
>  
>  	ai = host_pton(buf);
>  
> +#if !definded(__UCLIBC__) && defined(__GLIBC__)
You still have this typo here... and the only reason
it compiled is HAVE_GETNAMEINFO is not defined 
in your world....

How well were these change tested against glibc? I'm 
concern about eliminating chunks of need code with
all these new  defines.... 

steved.

>  	/*
>  	 * getaddrinfo(AI_NUMERICHOST) never fills in ai_canonname
>  	 */
> @@ -392,7 +393,9 @@ host_numeric_addrinfo(const struct sockaddr *sap)
>  			ai = NULL;
>  		}
>  	}
> +#endif
>  
>  	return ai;
>  }
> +
>  #endif	/* !HAVE_GETNAMEINFO */
>