On Thu, 2004-11-11 at 19:46, Roland Dreier wrote:
> In fact I don't see how Solaris can deduce the interface from
> a link local IPv6 address...
I don't see how this would work either (at least for Linux):
Here's my config:
eth1
inet6 addr: fe80::230:48ff:fe27:212f/64 Scope:Link
ib0
inet6 addr:
Roland Dreier wrote:
> Tom> Would you really bring both interfaces up? If this is a
> Tom> problem, the spec should have the pkey be part of the link
> Tom> local address.
>
> It actually seems to work fine to bring up multiple IPv6 interfaces
> that end up with the same link local ad
Tom> Would you really bring both interfaces up? If this is a
Tom> problem, the spec should have the pkey be part of the link
Tom> local address.
It actually seems to work fine to bring up multiple IPv6 interfaces
that end up with the same link local address (like ib0 and ib0.8001).
Th
On Thu, 2004-11-11 at 16:38 -0800, Roland Dreier wrote:
> I think we just need to copy our address to the child interface. This
> patch seems to fix it for me (already checked in).
Yup, this fixes it. You rock.
> (By the way, how does IPv6 handle autoconfig for VLAN interfaces?
> With this chan
Nitin> Based on some googling I found that for linux, since Link
Nitin> Local addresses are not routable, you need to provide the
Nitin> scope (by specifying an outgoing interface) to ssh in
Nitin> linux. This is very different from Solaris implementation
Nitin> where it still d
I think we just need to copy our address to the child interface. This
patch seems to fix it for me (already checked in).
(By the way, how does IPv6 handle autoconfig for VLAN interfaces?
With this change you can get duplicate autoconfig'ed addresses,
although they will be in different partitions.
Tom> With the updated patch (and with Nitin's original patch),
Tom> when I bring up ipv6, I am not getting the correct link local
Tom> address. I can assign it a global address and ping just
Tom> fine, but the lower 64 bits of the IPv6 address are NULL
Tom> (except for the set
Roland Dreier wrote:
> By the way, can anyone explain the following to me (an IPv6 rookie):
>
> # ping6 -I ib0 fe80::202:c901:78c:e461
> PING fe80::202:c901:78c:e461(fe80::202:c901:78c:e461) from
> fe80::202:c901:7fc:c711 ib0: 56 data bytes
> 64 bytes from fe80::202:c901:78c:e461: icm
On Thu, 2004-11-11 at 12:13 -0800, Roland Dreier wrote:
> OK, with the patch below all the correct IPv6 groups seem to be
> created and used. Ping works at least...
With the updated patch (and with Nitin's original patch), when I bring
up ipv6, I am not getting the correct link local address. I
lier fix defined
above.)
___
IPoverIB mailing list
IPoverIB at ietf.org
https://www1.ietf.org/mailman/listinfo/ipoverib
-Forwarded Message-
From: Hal Rosenstock <[EMAIL PROTECTED]>
To: Roland Dreier <[EMAIL PROTECTED]>
Cc: Nitin Hande &
Roland Dreier wrote:
> By the way, can anyone explain the following to me (an IPv6 rookie):
>
> # ping6 -I ib0 fe80::202:c901:78c:e461
> PING fe80::202:c901:78c:e461(fe80::202:c901:78c:e461) from
> fe80::202:c901:7fc:c711 ib0: 56 data bytes
> 64 bytes from fe80::202:c901:78c:e461: icm
By the way, can anyone explain the following to me (an IPv6 rookie):
# ping6 -I ib0 fe80::202:c901:78c:e461
PING fe80::202:c901:78c:e461(fe80::202:c901:78c:e461) from
fe80::202:c901:7fc:c711 ib0: 56 data bytes
64 bytes from fe80::202:c901:78c:e461: icmp_seq=1 ttl=64 time=32.2 ms
6
On Thu, 2004-11-11 at 15:13, Roland Dreier wrote:
> One question about IPv6 and IPoIB: currently the IPoIB driver joins
> the IPv4 broadcast group and then uses those parameters to join or
> create (as needed) the other groups, including all IPv6 multicast
> groups. Is this correct, or is there a
OK, with the patch below all the correct IPv6 groups seem to be
created and used. Ping works at least...
One question about IPv6 and IPoIB: currently the IPoIB driver joins
the IPv4 broadcast group and then uses those parameters to join or
create (as needed) the other groups, including all IPv6 m
I just tested, and the IPv6 ND packets are being sent to the MGID
ff12:401b::0:0:0::. This makes sense because
net/ipv6/ndisc.c uses dev->broadcast in ndisc_mc_map() if it doesn't
know about the interface type.
I'll see if creating ipv6_ib_mc_map() helps.
- R.
_
All,
Thanks for your comments,
Roland Dreier wrote:
> Nitin> I would appreciate if someone can review my patch to enable
> Nitin> inet6 address on ib interface. This is the first cut, will
> Nitin> like to hear from all. I plan to setup a bugzilla account
> Nitin> and append this
On Thu, 2004-11-11 at 13:46, Roland Dreier wrote:
> Hal> IBTA GUIDs are EUI-64. The only issue I recall was whether
> Hal> the polarity of the U/G bit was consistent with IEEE. This
> Hal> was updated at IBA 1.2. It now says "manufacturer assigns
> Hal> EUI-64 with global scope set.
Hal> IBTA GUIDs are EUI-64. The only issue I recall was whether
Hal> the polarity of the U/G bit was consistent with IEEE. This
Hal> was updated at IBA 1.2. It now says "manufacturer assigns
Hal> EUI-64 with global scope set. May also assign additional
Hal> EUI-64 with local sco
On Thu, 2004-11-11 at 13:27 -0500, Hal Rosenstock wrote:
> What's S10 ?
Solaris 10. Which has IPv6oIB.
-tduffy
signature.asc
Description: This is a digitally signed message part
___
openib-general mailing list
[EMAIL PROTECTED]
http://openib.org/mail
On Thu, 2004-11-11 at 13:11, Roland Dreier wrote:
> My only questions are:
>
> + eui[0] ^= 2;
>
> I remember some discussion about whether IBTA GUIDs are already
> modified EUI-64 or not. Is this the correct transformation or should
> we be doing something like "eui[0] |= 2;" (ie as
signed off by: Nitin Hande <[EMAIL PROTECTED]>
By the way, the proper format for
signed off by: Nitin Hande <[EMAIL PROTECTED]>
is really
Signed-off-by: Nitin Hande <[EMAIL PROTECTED]>
(see Documentation/SubmittingPatches).
- R.
___
openib-
Nitin> I would appreciate if someone can review my patch to enable
Nitin> inet6 address on ib interface. This is the first cut, will
Nitin> like to hear from all. I plan to setup a bugzilla account
Nitin> and append this patch to the bug that Tom has created for
Nitin> inet6.
T
signed off by: Nitin Hande <[EMAIL PROTECTED]>
I would appreciate if someone can review my patch to enable inet6
address on ib interface. This is the first cut, will like to hear from
all. I plan to setup a bugzilla account and append this patch to the bug
that Tom has created for inet6.
diff -N
23 matches
Mail list logo
