On Mon, Nov 7, 2011 at 10:08 PM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 09:04:44PM -0800, Jesse Gross wrote:
>> Currently multicast and unicast tunnel ports share port pools but
>> there's no overlap between the two in the lookup, which means that
>> we can do a lookup that has no chance of ev
On Mon, Nov 07, 2011 at 09:04:44PM -0800, Jesse Gross wrote:
> Currently multicast and unicast tunnel ports share port pools but
> there's no overlap between the two in the lookup, which means that
> we can do a lookup that has no chance of ever finding a port. This
> separates them out.
>
> Sugg
This is less redundant.
---
lib/automake.mk |1 -
lib/classifier.c | 10 +-
lib/learn.c | 18 +-
lib/meta-flow.c | 310 +++
lib/meta-flow.h | 11 ++
lib/nx-match.c | 475 +++---
lib/nx-match.def |
---
lib/flow.c | 21 +
lib/flow.h |3 +++
2 files changed, 24 insertions(+), 0 deletions(-)
diff --git a/lib/flow.c b/lib/flow.c
index 1263734..bf7ba6d 100644
--- a/lib/flow.c
+++ b/lib/flow.c
@@ -985,6 +985,27 @@ flow_hash_fields_valid(enum nx_hash_fields fields)
NXM breaks ICMP into v4 and v6. An upcoming commit will drop all of the
NXM specific data in favor of mf_field, and so at that point we need to
have a separate mf_field for each NXM field. So, this commit splits
ICMP into v4 and v6 for meta-flow also.
---
lib/meta-flow.c | 74
These patches are the beginning of what was originally a larger series,
but the larger series is getting postponed and so I'm sending out the
initial patches that make sense on their own.
Plus, everyone loves meta-flow. (Well, actually, people regularly ask
me "WTF is meta-flow"?)
Ben Pfaff (3):
Currently multicast and unicast tunnel ports share port pools but
there's no overlap between the two in the lookup, which means that
we can do a lookup that has no chance of ever finding a port. This
separates them out.
Suggested-by: Ben Pfaff
Signed-off-by: Jesse Gross
---
datapath/tunnel.c |
On Mon, Nov 7, 2011 at 10:43 AM, Ben Pfaff wrote:
> On Sun, Nov 06, 2011 at 12:32:45PM -0800, Jesse Gross wrote:
>> It's possible to have an incoming packet that matches both a unicast
>> and multicast tunnel if the source address corresponds to the remote_ip
>> of a unicast tunnel and the destina
On Mon, Nov 7, 2011 at 6:52 PM, Pravin B Shelar wrote:
> Signed-off-by: Pravin B Shelar
> ---
> datapath/vport.c | 2 +-
> 1 files changed, 1 insertions(+), 1 deletions(-)
>
> diff --git a/datapath/vport.c b/datapath/vport.c
> index 6fe6042..b8eecf8 100644
> --- a/datapath/vport.c
> +++ b/dat
Signed-off-by: Pravin B Shelar
---
datapath/vport.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/datapath/vport.c b/datapath/vport.c
index 6fe6042..b8eecf8 100644
--- a/datapath/vport.c
+++ b/datapath/vport.c
@@ -465,7 +465,7 @@ int vport_send(struct vport *vport, str
On Mon, Nov 7, 2011 at 4:10 PM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 04:00:55PM -0800, Jesse Gross wrote:
>> On Mon, Nov 7, 2011 at 11:01 AM, Ben Pfaff wrote:
>> > On Sun, Nov 06, 2011 at 12:34:25PM -0800, Jesse Gross wrote:
>> >> The interfaces related to tunneling aren't finalized enough
On Mon, Nov 7, 2011 at 10:56 AM, Ben Pfaff wrote:
> On Sun, Nov 06, 2011 at 12:34:24PM -0800, Jesse Gross wrote:
>> Many of the function in vport.c are simply pass throughs to their
>> underlying vport implementation and, of these, many are used only
>> for bridge compatibility code. This allows
On Mon, Nov 07, 2011 at 04:00:55PM -0800, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 11:01 AM, Ben Pfaff wrote:
> > On Sun, Nov 06, 2011 at 12:34:25PM -0800, Jesse Gross wrote:
> >> The interfaces related to tunneling aren't finalized enough to be
> >> sent upstream but we also still want to reta
On Mon, Nov 07, 2011 at 03:54:40PM -0800, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 1:41 PM, Ben Pfaff wrote:
> > On Mon, Nov 07, 2011 at 11:18:04AM -0800, Jesse Gross wrote:
> >> On Mon, Nov 7, 2011 at 9:24 AM, Ben Pfaff wrote:
> >> > On Sun, Nov 06, 2011 at 09:56:10PM -0800, Jesse Gross wrote
On Mon, Nov 7, 2011 at 11:01 AM, Ben Pfaff wrote:
> On Sun, Nov 06, 2011 at 12:34:25PM -0800, Jesse Gross wrote:
>> The interfaces related to tunneling aren't finalized enough to be
>> sent upstream but we also still want to retain them in the OVS
>> repository. Since userspace should be compatib
On Mon, Nov 7, 2011 at 3:46 PM, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 3:07 PM, Pravin B Shelar wrote:
>> Fixed according to comments from Jesse.
>>
>> --8<--cut here-->8--
>>
>> Most of issue are reported by checkpatch.pl
>>
>> Signed-off-by:
On Mon, Nov 7, 2011 at 3:48 PM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 02:16:50PM -0800, Jesse Gross wrote:
>> On Mon, Nov 7, 2011 at 1:28 PM, Ben Pfaff wrote:
>> > On Mon, Nov 07, 2011 at 11:03:11AM -0800, Jesse Gross wrote:
>> >> On Mon, Nov 7, 2011 at 9:22 AM, Ben Pfaff wrote:
>> >> > The
On Mon, Nov 7, 2011 at 1:41 PM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 11:18:04AM -0800, Jesse Gross wrote:
>> On Mon, Nov 7, 2011 at 9:24 AM, Ben Pfaff wrote:
>> > On Sun, Nov 06, 2011 at 09:56:10PM -0800, Jesse Gross wrote:
>> >> On Fri, Nov 4, 2011 at 4:43 PM, Ben Pfaff wrote:
>> >> > Net
On Mon, Nov 07, 2011 at 02:16:50PM -0800, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 1:28 PM, Ben Pfaff wrote:
> > On Mon, Nov 07, 2011 at 11:03:11AM -0800, Jesse Gross wrote:
> >> On Mon, Nov 7, 2011 at 9:22 AM, Ben Pfaff wrote:
> >> > The kernel netlink code is not as picky as ours, BTW: gener
On Mon, Nov 7, 2011 at 3:07 PM, Pravin B Shelar wrote:
> Fixed according to comments from Jesse.
>
> --8<--cut here-->8--
>
> Most of issue are reported by checkpatch.pl
>
> Signed-off-by: Pravin B Shelar
> Bug #7771
Looks good, thanks.
Acked-by:
On Nov 7, 2011, at 2:10 PM, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 1:28 PM, Justin Pettit wrote:
>> diff --git a/acinclude.m4 b/acinclude.m4
>> index 648132a..fd89bea 100644
>> --- a/acinclude.m4
>> +++ b/acinclude.m4
>> @@ -248,6 +248,7 @@ AC_DEFUN([OVS_CHECK_LINUX_COMPAT], [
>>
On Mon, Nov 7, 2011 at 1:28 PM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 11:03:11AM -0800, Jesse Gross wrote:
>> On Mon, Nov 7, 2011 at 9:22 AM, Ben Pfaff wrote:
>> > The kernel netlink code is not as picky as ours, BTW: generally it
>> > only validates minimum lengths. ??Maybe we should only d
On Mon, Nov 7, 2011 at 1:28 PM, Justin Pettit wrote:
> diff --git a/acinclude.m4 b/acinclude.m4
> index 648132a..fd89bea 100644
> --- a/acinclude.m4
> +++ b/acinclude.m4
> @@ -248,6 +248,7 @@ AC_DEFUN([OVS_CHECK_LINUX_COMPAT], [
> [OVS_DEFINE([HAVE_CSUM_TYPES])])
>
> OVS_GREP_I
On Mon, Nov 07, 2011 at 01:45:59PM -0800, Justin Pettit wrote:
> On Nov 7, 2011, at 11:06 AM, Ben Pfaff wrote:
>
> > On Mon, Nov 07, 2011 at 10:33:06AM -0800, Justin Pettit wrote:
> >> On some platforms, the in6_addr structure is made up of 16- or 32-bit
> >> members. Depending on the members of
On Mon, Nov 7, 2011 at 1:53 PM, Pravin Shelar wrote:
> On Mon, Nov 7, 2011 at 1:35 PM, Jesse Gross wrote:
>> You remove a lot of included header files throughout this patch. How
>> did you come up with the list? It seems like the minimum set but
>> usually the rule is that if you use something
On Mon, Nov 07, 2011 at 01:43:43PM -0800, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 1:24 PM, Ben Pfaff wrote:
> > Here are redone versions of the initial patch and then of this one. ??I
> > can repost the whole thing if you'd prefer.
>
> The two patches look good to me.
Thanks.
___
On Mon, Nov 7, 2011 at 1:35 PM, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 12:49 PM, Pravin B Shelar wrote:
>> Most of issue are reported by checkpatch.pl
>>
>> Signed-off-by: Pravin B Shelar
>> Bug #7771
>
> I have a couple of questions but otherwise this looks very thorough to me.
>
>> diff -
On Nov 7, 2011, at 11:06 AM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 10:33:06AM -0800, Justin Pettit wrote:
>> On some platforms, the in6_addr structure is made up of 16- or 32-bit
>> members. Depending on the members of the "flow" structure, this can
>> lead to holes in the middle. Moving th
On Mon, Nov 7, 2011 at 1:24 PM, Ben Pfaff wrote:
> Here are redone versions of the initial patch and then of this one. I
> can repost the whole thing if you'd prefer.
The two patches look good to me.
___
dev mailing list
dev@openvswitch.org
http://open
On Mon, Nov 07, 2011 at 11:18:04AM -0800, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 9:24 AM, Ben Pfaff wrote:
> > On Sun, Nov 06, 2011 at 09:56:10PM -0800, Jesse Gross wrote:
> >> On Fri, Nov 4, 2011 at 4:43 PM, Ben Pfaff wrote:
> >> > NetFlow active timeouts were only mixed in with flow expira
On Mon, Nov 7, 2011 at 12:49 PM, Pravin B Shelar wrote:
> Most of issue are reported by checkpatch.pl
>
> Signed-off-by: Pravin B Shelar
> Bug #7771
I have a couple of questions but otherwise this looks very thorough to me.
> diff --git a/datapath/actions.c b/datapath/actions.c
> index b5b92ba.
On Mon, Nov 07, 2011 at 11:03:11AM -0800, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 9:22 AM, Ben Pfaff wrote:
> > The kernel netlink code is not as picky as ours, BTW: generally it
> > only validates minimum lengths. ??Maybe we should only do that in
> > userspace too; it would simplify a few th
On Nov 7, 2011, at 12:51 PM, Jesse Gross wrote:
> You're right that this is a bug (and one that I think has existed for
> much longer than just that commit from a year ago). I think the fix
> is correct but can you use csum_replace2() and then drop the casts?
> It's functionally exactly the same
On Mon, Nov 07, 2011 at 10:53:15AM -0800, Jesse Gross wrote:
> On Mon, Nov 7, 2011 at 9:16 AM, Ben Pfaff wrote:
> > diff --git a/lib/odp-util.c b/lib/odp-util.c
> > index 197b2d0..00d6f36 100644
> > --- a/lib/odp-util.c
> > +++ b/lib/odp-util.c
> > +static void
> > +format_ovs_key_attr_name(struct
On Mon, Nov 07, 2011 at 12:37:46PM -0800, Justin Pettit wrote:
> On Nov 7, 2011, at 11:42 AM, Ben Pfaff wrote:
>
> > On Mon, Nov 07, 2011 at 10:33:07AM -0800, Justin Pettit wrote:
> >> Signed-off-by: Justin Pettit
> >
> > Would you mind humoring me by writing htonl(0) below:
> >> +if (wc & F
On Mon, Nov 07, 2011 at 10:33:10AM -0800, Justin Pettit wrote:
> Add support matching the IPv4 TTL and IPv6 hop limit fields. This
> commit also adds support for modifying the IPv4 TTL. Modifying the IPv6
> hop limit isn't currently supported, since we don't support modifying
> IPv6 headers.
>
>
On Mon, Nov 7, 2011 at 10:33 AM, Justin Pettit wrote:
> Commit 8b6956 (datapath: Correctly update IP checksum with actions.)
> introduced an issue that caused the checksum to not be properly
> calculated on little endian systems when updating the IP TOS field.
> This commit fixes the issue.
>
> Si
On Mon, Nov 07, 2011 at 10:33:09AM -0800, Justin Pettit wrote:
> Signed-off-by: Justin Pettit
In nicira-ext.h, is there value in adding NXAST_SET_NW_ECN given that
it could be set with NXAST_REG_LOAD without adding a new action?
On the comments for NXAST_SET_NW_ECN and NXM_NX_IP_ECN, it'd be nic
On Nov 7, 2011, at 11:42 AM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 10:33:07AM -0800, Justin Pettit wrote:
>> Signed-off-by: Justin Pettit
>
> Would you mind humoring me by writing htonl(0) below:
>> +if (wc & FWW_IPV6_LABEL) {
>> +flow->ipv6_label = 0;
>> +}
>
> and here to
On Mon, Nov 07, 2011 at 10:33:08AM -0800, Justin Pettit wrote:
> This will be useful later when we add support for matching the ECN bits
> within the TOS field.
>
> Signed-off-by: Justin Pettit
kernel
--
Do you plan to use the top 6 bits of ip.frag for something later on?
If not, I think we
On Nov 7, 2011, at 11:27 AM, Ben Pfaff wrote:
> On Mon, Nov 07, 2011 at 10:33:07AM -0800, Justin Pettit wrote:
>> Signed-off-by: Justin Pettit
>
> "sparse" says:
>
> ../lib/nx-match.c:1216:26: warning: incorrect type in assignment (different
> base types)
> ../lib/nx-match.c:1216:26:expect
On Mon, Nov 07, 2011 at 10:33:07AM -0800, Justin Pettit wrote:
> Signed-off-by: Justin Pettit
Would you mind humoring me by writing htonl(0) below:
> +if (wc & FWW_IPV6_LABEL) {
> +flow->ipv6_label = 0;
> +}
and here too:
+case MFF_IPV6_LABEL:
+rule->wc.wildcards |= F
On Mon, Nov 07, 2011 at 10:33:07AM -0800, Justin Pettit wrote:
> Signed-off-by: Justin Pettit
"sparse" says:
../lib/nx-match.c:1216:26: warning: incorrect type in assignment (different
base types)
../lib/nx-match.c:1216:26:expected restricted __be32 [usertype] ipv6_label
../lib/nx-match.c:1
On Mon, Nov 7, 2011 at 9:24 AM, Ben Pfaff wrote:
> On Sun, Nov 06, 2011 at 09:56:10PM -0800, Jesse Gross wrote:
>> On Fri, Nov 4, 2011 at 4:43 PM, Ben Pfaff wrote:
>> > NetFlow active timeouts were only mixed in with flow expiration for
>> > convenience: both processes need to iterate all the fac
On Mon, Nov 07, 2011 at 10:33:06AM -0800, Justin Pettit wrote:
> On some platforms, the in6_addr structure is made up of 16- or 32-bit
> members. Depending on the members of the "flow" structure, this can
> lead to holes in the middle. Moving the IPv6 addresses to the middle
> fixes that on these
Looks good, thank you.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Mon, Nov 7, 2011 at 9:22 AM, Ben Pfaff wrote:
> The kernel netlink code is not as picky as ours, BTW: generally it
> only validates minimum lengths. Maybe we should only do that in
> userspace too; it would simplify a few things. Any thoughts on that?
Does anyone ever try to send extended st
On Sun, Nov 06, 2011 at 12:34:25PM -0800, Jesse Gross wrote:
> The interfaces related to tunneling aren't finalized enough to be
> sent upstream but we also still want to retain them in the OVS
> repository. Since userspace should be compatible with both versions
> of the kernel, this renumbers th
On Sun, Nov 06, 2011 at 12:34:24PM -0800, Jesse Gross wrote:
> Many of the function in vport.c are simply pass throughs to their
> underlying vport implementation and, of these, many are used only
> for bridge compatibility code. This allows users of these functions
> to directly call through the
On Mon, Nov 7, 2011 at 9:16 AM, Ben Pfaff wrote:
> diff --git a/lib/odp-util.c b/lib/odp-util.c
> index 197b2d0..00d6f36 100644
> --- a/lib/odp-util.c
> +++ b/lib/odp-util.c
> +static void
> +format_ovs_key_attr_name(struct ds *ds, enum ovs_key_attr attr)
> +{
> + const char *name = ovs_key_att
On Mon, Nov 7, 2011 at 9:16 AM, Ben Pfaff wrote:
> Usually we don't put a new-line in prototypes unless we have to. I also
> think that this comment deserves to be higher up in the file, so I moved
> it.
Looks good.
___
dev mailing list
dev@openvswitch
On Mon, Nov 7, 2011 at 9:16 AM, Ben Pfaff wrote:
> I usually do things this way but must have overlooked this one a while ago.
Looks good.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Sun, Nov 06, 2011 at 12:34:23PM -0800, Jesse Gross wrote:
> Now that most fix function logic (like sFlow) has been moved to
> userspace, the vport member of OVS_CB is no longer used by anything,
> so drop it.
>
> Signed-off-by: Jesse Gross
Acked-by: Ben Pfaff
Good spotting.
On Sun, Nov 06, 2011 at 12:32:45PM -0800, Jesse Gross wrote:
> It's possible to have an incoming packet that matches both a unicast
> and multicast tunnel if the source address corresponds to the remote_ip
> of a unicast tunnel and the destination is multicast. Currently this
> will match the mult
Add support matching the IPv4 TTL and IPv6 hop limit fields. This
commit also adds support for modifying the IPv4 TTL. Modifying the IPv6
hop limit isn't currently supported, since we don't support modifying
IPv6 headers.
We will likely want to change the user-space interface, since basic
matchi
Signed-off-by: Justin Pettit
---
NEWS |2 +
datapath/actions.c| 12 ++--
datapath/datapath.c |4 ---
datapath/flow.c |9 +++---
include/openflow/nicira-ext.h | 25 +-
lib/classifier.c |
This will be useful later when we add support for matching the ECN bits
within the TOS field.
Signed-off-by: Justin Pettit
---
datapath/datapath.c |6 ++--
datapath/flow.c | 57 +++
datapath/flow.h |8 +++---
lib/classifie
Signed-off-by: Justin Pettit
---
NEWS |3 ++-
datapath/flow.c |7 +--
datapath/flow.h |7 ---
include/linux/openvswitch.h |1 +
include/openflow/nicira-ext.h |9 +
lib/classifier.c | 16 +
On some platforms, the in6_addr structure is made up of 16- or 32-bit
members. Depending on the members of the "flow" structure, this can
lead to holes in the middle. Moving the IPv6 addresses to the middle
fixes that on these platforms.
---
lib/flow.h | 10 +-
1 files changed, 5 inser
---
utilities/ovs-ofctl.8.in |6 +++---
1 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/utilities/ovs-ofctl.8.in b/utilities/ovs-ofctl.8.in
index 007b54a..ee2caf2 100644
--- a/utilities/ovs-ofctl.8.in
+++ b/utilities/ovs-ofctl.8.in
@@ -404,9 +404,9 @@ Matches IP ToS/DSCP or IPv6
Commit 8b6956 (datapath: Correctly update IP checksum with actions.)
introduced an issue that caused the checksum to not be properly
calculated on little endian systems when updating the IP TOS field.
This commit fixes the issue.
Signed-off-by: Justin Pettit
---
datapath/actions.c |4 ++--
1
I'm still having a really hard time spotting the problem. Can you
pass along an "ovs-dpctl dump-flows" output that includes a
set(eth(...)) action? An "ofproto/trace" output that includes such an
action might also be helpful.
On Sun, Nov 06, 2011 at 12:00:41PM +0900, Jari Sundell wrote:
> Packet
On Sun, Nov 06, 2011 at 09:56:10PM -0800, Jesse Gross wrote:
> On Fri, Nov 4, 2011 at 4:43 PM, Ben Pfaff wrote:
> > NetFlow active timeouts were only mixed in with flow expiration for
> > convenience: both processes need to iterate all the facets. ??But
> > an upcoming commit will change flow expi
On Sat, Nov 05, 2011 at 10:44:58AM -0700, Jesse Gross wrote:
> On Fri, Nov 4, 2011 at 4:43 PM, Ben Pfaff wrote:
> > diff --git a/lib/netlink.h b/lib/netlink.h
> > index b7a87c6..7eced42 100644
> > --- a/lib/netlink.h
> > +++ b/lib/netlink.h
> > +bool nl_attr_validate(const struct nlattr *, const s
On Sat, Nov 05, 2011 at 10:34:07AM -0700, Jesse Gross wrote:
> On Fri, Nov 4, 2011 at 4:43 PM, Ben Pfaff wrote:
> > ---
> > ??lib/netlink.c | ??105
> > +++-
> > ??1 files changed, 58 insertions(+), 47 deletions(-)
>
> Looks good, I was wonderin
On Fri, Nov 04, 2011 at 07:54:44PM -0700, Jesse Gross wrote:
> On Fri, Nov 4, 2011 at 4:43 PM, Ben Pfaff wrote:
> > This seems like a worthwhile improvement in itself, but it will also see
> > additional users in upcoming commits.
>
> I briefly wondered whether it was worth combining this with
>
---
lib/odp-util.c | 55 +--
1 files changed, 33 insertions(+), 22 deletions(-)
diff --git a/lib/odp-util.c b/lib/odp-util.c
index 197b2d0..00d6f36 100644
--- a/lib/odp-util.c
+++ b/lib/odp-util.c
@@ -40,6 +40,9 @@
* interactions with the da
Usually we don't put a new-line in prototypes unless we have to. I also
think that this comment deserves to be higher up in the file, so I moved
it.
---
lib/odp-util.c |5 ++---
1 files changed, 2 insertions(+), 3 deletions(-)
diff --git a/lib/odp-util.c b/lib/odp-util.c
index fb7aabb..197b2
I usually do things this way but must have overlooked this one a while ago.
---
lib/odp-util.c |5 -
1 files changed, 4 insertions(+), 1 deletions(-)
diff --git a/lib/odp-util.c b/lib/odp-util.c
index f348e58..fb7aabb 100644
--- a/lib/odp-util.c
+++ b/lib/odp-util.c
@@ -353,6 +353,7 @@ fo
69 matches
Mail list logo
