Re: IPv6 enabled Squid - how?

I have now done a "blind" ...

Henrik Nordstrom wrote:
>
> I have not done a "blind" update of the patch to current HEAD, but some
> work remains in IPv6 enabling the new additions before it will build
> correctly again..
>
> Anyone wanting to take a stab at it?
>
> Regards
> Henrik Nordström
> Squid Hacker
>
> Henrik Nordstrom wrote:
> >
> > Peter Bieringer wrote:
> >
> > > Looks like there are some existing buchs in this branch
> >
> > It should, as it is not fully finished yet.
> >
> > > 1) configure without "--disable-wccp" breaks compiling in
> > > src/cf_parser.c
> >
> > And so is documented for the ipv6 branch
> > <http://devel.squid-cache.org/ipv6/>
> >
> > > 2) src/acl.c has a not really C conform source part (using gcc-2.96):
> > >
> > > --- src/acl.c.cvsorig Thu Oct 25 10:00:25 2001
> > > +++ src/acl.c Thu Oct 25 10:03:27 2001
> > > @@ -2252,8 +2252,8 @@
> > > if (ADDR_IS_ANYADDR(q->addr2)) {
> > > return memcmp(&A, &q->addr1, sizeof(struct IN_ADDR));
> > > } else {
> > > - if (memcmp(&A, &q->addr1, sizeof(struct IN_ADDR)) => 0) &
> > > - (memcmp(&A, &q->addr2, sizeof(struct IN_ADDR)) <= 0)
> > > + if ((memcmp(&A, &q->addr1, sizeof(struct IN_ADDR)) >= 0) &&
> > > + (memcmp(&A, &q->addr2, sizeof(struct IN_ADDR)) <= 0))
> > > return 0; /* valid */
> > > else
> > > return -1; /* outside of range, 'less than' */
> >
> > Looks like a bug in the ipv6 code.. fixed.
> >
> > > BTW: a warning was also displayed here:
> > > acl.c:1723: warning: enumeration value `ACL_SRC_ASN' not handled in
> > > switch
> > > acl.c:1723: warning: enumeration value `ACL_DST_ASN' not handled in
> > > switch
> >
> > This is fixed in the more current Squid code I think, and will thus be
> > fixed in the ipv6 branch once someone cares to maintain it.
> >
> > > 3) Ooops:
> > > client_side.c: In function `parseHttpRequest':
> > > client_side.c:2652: structure has no member named `sin_addr'
> > > client_side.c:2654: structure has no member named `sin_port'
> > > client_side.c:2654: structure has no member named `sin_port'
> > > client_side.c:2654: structure has no member named `sin_port'
> > > client_side.c:2654: structure has no member named `sin_port'
> > > client_side.c:2654: warning: `__v' might be used uninitialized in
> > > this function
> > > make[2]: *** [client_side.o] Error 1
> > > make[2]: Leaving directory `/usr/local/src/squid/squid-ipv6/src'
> > > make[1]: *** [all] Error 2
> > > make[1]: Leaving directory `/usr/local/src/squid/squid-ipv6/src'
> > > make: *** [all] Error 1
> >
> > Hmm.. appears the --enable-linux-netfilter code is not ipv6 adoped yet.
> >
> > > BTW: because of RHL 7.2 is now good IPv6 enabled would be imho a big
> > > step further on if a working IPv6 squid is available, too.
> > > Reason: browsers like Mozilla or Konqueror cannot be setup (by
> > > configuration) to use for IPv4 a proxy, but not for IPv6). So if for
> > > IPv4 a proxy is required, it will be also used for IPv6 requests -
> > > but without success.
> >
> > I fully agree.
> >
> > The browser when using a proxy should not diverse between IPv4 and IPv6.
> > It is the proxy's job to figure out how to reach the requested service.
> >
> > Regards
> > Henrik Nordström
> > Squid Hacker
Received on Thu Oct 25 2001 - 13:06:17 MDT