On Jul 3, 2013, at 1:58 , ext Andy Zhou wrote:
> diff --git a/datapath/tunnel.c b/datapath/tunnel.c
> index 9102786..18c3622 100644
> --- a/datapath/tunnel.c
> +++ b/datapath/tunnel.c
> @@ -186,7 +186,9 @@ u16 ovs_tnl_get_src_port(struct sk_buff *skb)
> int low;
> int high;
> uns
On Jul 2, 2013, at 1:28 , ext Pravin B Shelar wrote:
> diff --git a/datapath/linux/compat/include/net/vxlan.h
> b/datapath/linux/compat/include/net/vxlan.h
> new file mode 100644
> index 000..76e0ddb
> --- /dev/null
> +++ b/datapath/linux/compat/include/net/vxlan.h
> @@ -0,0 +1,43 @@
> +#ifn
On Jun 29, 2013, at 0:59 , ext Jesse Gross wrote:
> On Fri, Jun 28, 2013 at 10:50 AM, Ben Pfaff wrote:
>> On Thu, Jun 27, 2013 at 01:39:52AM +0300, Jarno Rajahalme wrote:
>>>
>>> Signed-off-by: Jarno Rajahalme
>>> ---
>>> v2: Remove resetting of provider meter id on failure in dpif_meter_set()
On Jun 29, 2013, at 1:05 , ext Jesse Gross wrote:
> On Tue, Jun 25, 2013 at 1:08 AM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>> I also tried to figure out how the group table should be implemented. I
>> haven't done anything for it yet, but I'm rather convinced
On Jun 28, 2013, at 19:46 , ext Ben Pfaff wrote:
> On Fri, Jun 28, 2013 at 07:44:04PM +0300, Jarno Rajahalme wrote:
>> Any protocol version can match dl_type=0x86dd, but only NXM and OXM can match
>> IPv6 addresses, or nd_target.
>> Update unit tests to check for this.
>>
>> Signed-off-by: Jarno
On Jun 28, 2013, at 19:33 , ext Ben Pfaff wrote:
> On Fri, Jun 28, 2013 at 04:28:40PM +0000, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> On Jun 28, 2013, at 19:09 , ext Ben Pfaff wrote:
>>
>>> On Fri, Jun 28, 2013 at 07:17:09AM +, Rajahalme,
On Jun 28, 2013, at 19:09 , ext Ben Pfaff wrote:
> On Fri, Jun 28, 2013 at 07:17:09AM +0000, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> On Jun 28, 2013, at 1:19 , ext Ben Pfaff wrote:
>>> ofproto: Use another method to find meter_id.
>>
>> OK, wi
Now "make modules_install" works for me too, thanks.
Jarno
On Jun 28, 2013, at 3:26 , ext Jesse Gross wrote:
> The Open vSwitch kernel module now has dependencies on symbols
> exported by other kernel modules (currently just for GRE). In
> order for it to load, the dependencies must be correct
On Jun 28, 2013, at 4:43 , ext Jesse Gross wrote:
> It's a validation limitation: if you have a vlan packet then you need
> to have an exact match on a corresponding EtherType. If you only want
> to match on the Ethernet header then it's fully maskable.
>
On OpenFlow (& meta-flow etc.) ethertyp
On Jun 28, 2013, at 4:16 , ext Justin Pettit wrote:
...
> diff --git a/lib/flow.c b/lib/flow.c
> index a42fea1..1a5084b 100644
> --- a/lib/flow.c
> +++ b/lib/flow.c
> @@ -781,7 +781,8 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t
> basis)
>
> /* Masks the fields in 'wc' that are us
I sent some comments (mostly about comments it seems) earlier you may want to
revisit (or not):
http://openvswitch.org/pipermail/dev/2013-May/027297.html
Jarno
On Jun 28, 2013, at 1:30 , ext Ben Pfaff wrote:
> We have a controller that puts many rules with different metadata values
> into the
On Jun 28, 2013, at 1:19 , ext Ben Pfaff wrote:
> I spent some time looking at and playing with patch 4/5 "v2: ofproto:
> Implement OpenFlow 1.3 meter table." and decided that there was a fair bit
> that I wanted to tweak. This series represent what I came up with:
>
> IMPORTANT: In this series
On Jun 27, 2013, at 3:13 , ext Ben Pfaff wrote:
> On Thu, Jun 27, 2013 at 01:39:50AM +0300, Jarno Rajahalme wrote:
>> Validate claimed message length for meter stats in ofp-util.c.
>> Clean up meters in ofp-util.h.
>> Fix the impossible duration values in ofp-print.at.
>>
>> Signed-off-by: Jarno
On Jun 26, 2013, at 21:39 , ext Jesse Gross wrote:
>>
>> OK, so you need to use the kernel version from the "with-linux" option given
>> to configure then?
>> I have configured with --with-linux=/lib/modules/`uname -r`/build so that
>> happened to work for me.
>
> Would you mind taking a look
On Jun 25, 2013, at 20:35 , ext Ben Pfaff wrote:
> On Thu, Jun 20, 2013 at 05:26:18PM +0300, Jarno Rajahalme wrote:
>>
>> Signed-off-by: Jarno Rajahalme
>
> I'm applied this to master. I'm appending the incremental that I
> folded in. Most of the changes are minor improvements (some just
> s
On Jun 25, 2013, at 22:41 , ext Ben Pfaff wrote:
> On Thu, Jun 20, 2013 at 05:26:23PM +0300, Jarno Rajahalme wrote:
>> Adds commands add-meter, mod-meter, del-meter, del-meters, dump-meter,
>> dump-meters, meter-stats, and meter-features.
>>
>> Syntax is as follows:
>>
>> add-meter meter= (kbp
On Jun 25, 2013, at 22:38 , ext Ben Pfaff wrote:
...
> I'm not fond of the structure of the new ofpacts_check_ctx(). At a
> minimum, I think that ofpacts_check() should be a trivial wrapper
> around it (just change ofpacts_check_ctx() to ignore 'ctx' if it's
> NULL). But the inversion of control
Ben,
On a related note, the patch that you already applied added a new
ofp_flow_removed_reason code OFPRR_METER_DELETE (in openflow-common.h). I
intended to have some discussion on it, but forgot to point it out. There is a
corresponding OFPRR_GROUP_DELETE reason code, but the OF spec does not
On Jun 26, 2013, at 18:27 , ext Jesse Gross wrote:
> On Wed, Jun 26, 2013 at 8:05 AM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> On Jun 25, 2013, at 22:31 , ext Jesse Gross wrote:
>>
>>> The Open vSwitch kernel module now has dependencies on symbols
>
On Jun 26, 2013, at 0:47 , ext Roam Pune wrote:
...
i got same error, but then i found a way around it using following command:
$ sudo insmod datapath/linux/openvswitch.ko
insmod: error inserting 'datapath/linux/openvswitch.ko': -1 Unknown symbol in
module
$ sudo modprobe gre
$ sudo insmod data
On Jun 25, 2013, at 22:31 , ext Jesse Gross wrote:
> The Open vSwitch kernel module now has dependencies on symbols
> exported by other kernel modules (currently just for GRE). In
> order for it to load, the dependencies must be correctly resolved
> ahead of time. This runs depmod as part of the
On May 7, 2013, at 21:55 , ext Pravin Shelar wrote:
> On Tue, May 7, 2013 at 11:13 AM, Jarno Rajahalme
> wrote:
>> This reduces repeated code and makes it easier to add new UDP tunneling
>> protocols.
>>
> This approach is different from what tunnel upstream patches does.
> Since there is no li
Me too, on Debian 3.2.46-1 x86_64 GNU/Linux
Jarno
On Jun 24, 2013, at 20:09 , ext Ben Pfaff wrote:
> [oops, sorry, re-sending because I screwed up the list address on the
> first try]
>
> I'm getting the following warnings in my OVS master build tree (commit
> ede77a461fb "datapath: Fix a ke
On Jun 25, 2013, at 1:22 , ext Ben Pfaff wrote:
> What I really want to do is reconcile the two patch series, so that we
> can get the best features from both. Your code is always solid in my
> experience, so I'm kind of inclined to just review yours and apply it,
> and then go through the Cente
On Jun 20, 2013, at 23:36 , ext Ben Pfaff wrote:
> On Thu, Jun 20, 2013 at 05:26:17PM +0300, Jarno Rajahalme wrote:
>>
>> Signed-off-by: Jarno Rajahalme
>
> I have mixed feelings here. On one hand, it is nice to avoid a big
> copy in some cases (rare cases really). But on the other hand I am
All looks good and tests pass. However, I'd probably add a comment on the
hindex.mask (in hindex.h) stating that it has a number of lowest order bits
set, others cleared. After realizing that reading the code became a bit easier..
I also changed the cookie indexing to use this and it also passes
a
On Jun 18, 2013, at 0:18 , ext Ben Pfaff wrote:
> On Mon, Jun 17, 2013 at 11:45:49AM +0000, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>> On Jun 17, 2013, at 3:55 , Jarno Rajahalme wrote:
>>
>>> The simplest way for an OpenFlow controller to refer to a (set of) f
On Jun 17, 2013, at 3:55 , Jarno Rajahalme wrote:
> The simplest way for an OpenFlow controller to refer to a (set of) flows
> is by a controller-issued flow cookie. Make this fast by inserting
> flows with non-zero cookies to a hmap, and use that when flows are
> queried with a full cookie mask.
On Jun 13, 2013, at 1:16 , ext Ben Pfaff wrote:
> On Fri, May 31, 2013 at 02:35:10PM +0300, Jarno Rajahalme wrote:
>> The code in ofproto-dpif.c became harder on the eye after the reorg of
>> struct xlate. This series tries cleaning that up a bit, and while at it,
>> identifying and reducing unne
On Jun 10, 2013, at 20:04 , ext Jesse Gross wrote:
>
+static bool __flow_cmp_key_mask(struct sw_flow *flow, u32 hash, u8
*key,
+ int key_start, int key_len,
+ struct sw_flow_mask *mfm)
+{
+ return (flo
On Jun 12, 2013, at 2:10 , ext Alex Wang wrote:
> Thanks Ben,
>
> This makes sense. But one thing is that if we compare to "ofp_port_t"
> variables, e.g. "flow->in_port.ofp_port >= OFPP_MAX", we must call this
> function for both operands. this seems to make coding harder, really want to
> kn
07, 2013 at 08:28:20PM +0000, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>> It is in lib/util.h:
>>
>> lib/util.h:#define DIV_ROUND_UP(X, Y) (((X) + ((Y) - 1)) / (Y))
>> lib/util.h:#define ROUND_UP(X, Y) (DIV_ROUND_UP(X, Y) * (Y))
>>
>> Jarno
>>
>
dup(x, y) ( \
{ \
const typeof(y) __y = y;\
(((x) + (__y - 1)) / __y) * __y;\
}
On Fri, Jun 7, 2013 at 7:19 AM, Rajahalme, Jarno (NSN - FI/Espoo)
mailto:jarno.rajaha...@nsn.com>> wrote:
On Jun 7, 2013, at 13:35 , ext Andy Zhou wrote:
Is
On Jun 7, 2013, at 13:35 , ext Andy Zhou wrote:
Is this the same as DIV_ROUND_UP?
I did not use it because DIV_ROUND_UP gives the roundup of u32, I needed the
roundup in bytes. However, I could rewrite this function using DIV_ROUND_UP.
There is ROUND_UP doing that already.
On Jun 7, 2013, at 12:03 , ext Justin Pettit wrote:
...
> diff --git a/lib/flow.c b/lib/flow.c
> index 6476029..db4ce1d 100644
> --- a/lib/flow.c
> +++ b/lib/flow.c
> @@ -759,6 +759,32 @@ flow_hash_symmetric_l4(const struct flow *flow, uint32_t
> basis)
> return jhash_bytes(&fields, sizeof fi
On Jun 7, 2013, at 0:02 , ext Ben Pfaff wrote:
> handle_flow_miss_without_facet() previously translated the actions for
> each packet. Now, it translates the actions once (or not at all, if
> there's an xc already). Won't this prevent slow protocols (e.g. CFM,
> LACP, BFD) from working properly
On Jun 7, 2013, at 1:34 , ext Justin Pettit wrote:
> On Jun 5, 2013, at 10:43 PM, Ben Pfaff wrote:
>> I think that struct xout_cache could use a top-level comment
>> explaining its purpose and how it generally fits in with the rest of
>> the system.
>
> /* A cache of xlate_out (xout) translatio
On May 31, 2013, at 11:23 , ext Jesse Gross wrote:
> On Fri, May 31, 2013 at 4:53 PM, Fan Du wrote:
>> [ovs-dev] [PATCH] Use EEXIST as return error when adding vport
>>
>> Hi, Jesse
>>
>> Those two patches attempts to use EEXIST as return error when trying
>> to add an vport which is actually a
On May 30, 2013, at 0:24 , ext Ethan Jackson wrote:
> +/* Context used by xlate_actions() and its callees. */
> +struct xlate_ctx {
> +struct xlate_in *xin;
> +struct xlate_out *xout;
>
> -struct ofpbuf *odp_actions; /* Datapath actions. */
> -tag_type tags; /* Tags
On May 27, 2013, at 12:35 , ext Lorand Jakab wrote:
> Since commit 0ad90c8 it is possible to set tunnel destination IP address
> in the flow. This allows creating a LISP "map-cache" in the flow table.
> Update the LISP documentation to reflect these possibilities.
>
> Signed-off-by: Lorand Jaka
On May 24, 2013, at 19:26 , ext Jesse Gross wrote:
> On Thu, May 23, 2013 at 9:41 PM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> On May 23, 2013, at 20:28 , ext Jesse Gross wrote:
>>> The upstream commit is not really right - the check should depend on
>&g
Simon,
Here is some initial review of the ofproto-dpif.c changes,
Jarno
On May 24, 2013, at 10:18 , ext Simon Horman wrote:
> diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
> index 49f0270..8b1ccac 100644
> --- a/ofproto/ofproto-dpif.c
> +++ b/ofproto/ofproto-dpif.c
> @@ -43,6 +
Ben,
Tested, works, and looks good. I can now see port duration in "ovs-ofctl -O
OpenFlow13 dump-ports br0" :-)
Jarno
On May 24, 2013, at 2:56 , ext Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
> ---
> OPENFLOW-1.1+ |3 ++-
> lib/ofp-print.c|6 ++
> lib/ofp
Pravin,
Please find some review comments below,
Jarno
On May 23, 2013, at 23:01 , ext Pravin B Shelar wrote:
...
> diff --git a/datapath/linux/compat/gre.c b/datapath/linux/compat/gre.c
> new file mode 100644
> index 000..c352ff8
> --- /dev/null
> +++ b/datapath/linux/compat/gre.c
> @@ -0,
On May 23, 2013, at 21:21 , ext Alex Wang wrote:
Hey Ben,
Just found, I think there are cases that a whole packet is wanted.
In function "dp_netdev_execute_actions()", there are calls to functions like
"push_mpls()" and "pop__mpls()", which requires the packet header pointers
(e.g. l2_5, l2 )
On May 23, 2013, at 20:28 , ext Jesse Gross wrote:
> The upstream commit is not really right - the check should depend on
> the origin of the packet, not the location in the networking stack
> processing it. Most of the time these are somewhat the same but
> obviously when bridging, received packe
>From 0aa52d88bdf6bc22fed80beb175a6dfe420dfabd Mon Sep 17 00:00:00 2001
From: Cong Wang mailto:amw...@redhat.com>>
Date: Wed, 6 Feb 2013 14:40:36 -0800
Subject: [PATCH] datapath: adjust skb_gso_segment() for calling in rx path
skb_gso_segment() is almost always called in tx path,
except for openvs
This good, and thanks for clarifying the policy on function output parameters.
Jarno
On May 20, 2013, at 21:31 , ext Ben Pfaff wrote:
> This makes life easier for a few callers, and it agrees with a general
> principle that a function should fill in its output parameters whether it
> succeeds
lean toward #2, but the world won't end if we don't do it.
>
> How do you feel about the first three patches? They are independent
> of this bigger decision.
>
> I need to look at the zeros issue too, I can't quite remember why we
> decided we didn't need them
Getting rid of mostly dummy function parameters and duplication is always nice,
but I'd like it better if there was a way not to have to see the "md."
everywhere (start to miss C++ here..). Also, since we care about the zeros at
the end of struct flow, maybe we should have them in struct flow_tn
Looks good on all three counts,
Jarno
On May 8, 2013, at 23:23 , ext Ben Pfaff wrote:
> The tunnel and non-tunnel paths were pretty much the same anyway, so this
> commit simplifies by merging them.
>
> Signed-off-by: Ben Pfaff
> ---
> ofproto/ofproto-dpif.c | 80 +--
On May 8, 2013, at 11:13 , ext Rajahalme, Jarno (NSN - FI/Espoo) wrote:
>>
>>old_rule = insert_rule(cls, table, rule);
>>if (!old_rule) {
>> +if (minimask_get_metadata_mask(&rule->match.mask) == OVS_BE64_MAX) {
>> +ovs_be64 m
I did not see this pushed yet, so here is a review:
On Mar 27, 2013, at 6:32 , ext Ben Pfaff wrote:
> We have a controller that puts many rules with different metadata values
> into the flow table, where metadata is used (by "resubmit"s) to distinguish
> stages in a pipeline. Thus, any given flo
Ethan,
Looking at the code, it seems to me that there is no link from BFD to OpenFlow
PORT_STATUS message. I might be wrong, but it seems possible that BFD indicated
port state changes could be communicated to the controller with the OFPPS_LIVE
(= OFPUTIL_PS_LIVE) status bit, using the ofproto_
On May 7, 2013, at 21:55 , ext Pravin Shelar wrote:
> On Tue, May 7, 2013 at 11:13 AM, Jarno Rajahalme
> wrote:
>> This reduces repeated code and makes it easier to add new UDP tunneling
>> protocols.
>>
> This approach is different from what tunnel upstream patches does.
> Since there is no li
Looks good, thanks :-)
Jarno
On May 7, 2013, at 1:38 , ext Ben Pfaff wrote:
> OpenFlow says that an "output" action to a flow's input port is ordinarily
> dropped, unless the flow explicitly outputs to OFPP_IN_PORT. We've
> occasionally been asked to implement some way to avoid this behavior
On May 2, 2013, at 21:19 , ext Ben Pfaff wrote:
> On Tue, Apr 30, 2013 at 05:21:28AM +0000, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> On Apr 29, 2013, at 20:49 , ext Jesse Gross wrote:
>>
>>> On Sun, Apr 28, 2013 at 11:29 AM, Rajahalme,
On May 2, 2013, at 21:21 , ext Ben Pfaff wrote:
> There has been a lot of discussion on patch 4. Do you plan to update
> and re-send the series, or do you expect further discussion or review
> of the rest of the series before we go on?
I have the rest of the series done, I was waiting for some
On Apr 30, 2013, at 18:54 , ext Jesse Gross wrote:
> On Mon, Apr 29, 2013 at 10:21 PM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> On Apr 29, 2013, at 20:49 , ext Jesse Gross wrote:
>>
>>> On Sun, Apr 28, 2013 at 11:29 AM, Rajahalme, Jarno (NSN - FI/Espo
On Apr 29, 2013, at 20:49 , ext Jesse Gross wrote:
> On Sun, Apr 28, 2013 at 11:29 AM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> Another thing that I came to think only when reading Ben's new tutorial:
>> Output to the input port is skipped. This would be
On Apr 28, 2013, at 21:29 , ext Rajahalme, Jarno (NSN - FI/Espoo) wrote:
>
> On Apr 27, 2013, at 0:43 , ext Jesse Gross wrote:
>
>> On Thu, Apr 18, 2013 at 8:07 AM, Jarno Rajahalme
>> wrote:
>>> diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
>
On Apr 27, 2013, at 0:43 , ext Jesse Gross wrote:
> On Thu, Apr 18, 2013 at 8:07 AM, Jarno Rajahalme
> wrote:
>> diff --git a/lib/match.c b/lib/match.c
>> index 2aa4e89..222e5d7 100644
>> --- a/lib/match.c
>> +++ b/lib/match.c
>> @@ -133,13 +133,9 @@ match_wc_init(struct match *match, const stru
On Apr 25, 2013, at 14:02 , ext Simon Horman wrote:
>
> Yes, stack_len is supposed to advance by one for each iteration of the loop.
Better make it by four instead ;-)
Jarno
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/lis
On Apr 25, 2013, at 10:58 , ext Simon Horman wrote:
> @@ -648,6 +650,7 @@ int ovs_flow_extract(struct sk_buff *skb, u16 in_port,
> struct sw_flow_key *key,
> return -ENOMEM;
>
> skb_reset_network_header(skb);
> + skb_reset_mac_len(skb);
> __skb_push(skb, skb->data
On Apr 23, 2013, at 4:51 , ext Simon Horman wrote:
> On Mon, Apr 22, 2013 at 01:55:43PM +0000, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> Here the skb_network_header is changed to point to the L3 header. Is it
>> significant that in some cases (?) mpls_stack_
On Apr 22, 2013, at 19:16 , ext Ben Pfaff wrote:
> On Mon, Apr 22, 2013 at 09:12:23AM -0700, Ben Pfaff wrote:
>> On Thu, Apr 18, 2013 at 06:07:41PM +0300, Jarno Rajahalme wrote:
>>> This makes sure that output actions leave no changes to any flow fields,
>>> while all explicit set_field actions ar
On Apr 22, 2013, at 5:12 , ext Simon Horman wrote:
> diff --git a/datapath/actions.c b/datapath/actions.c
...
> +static int push_mpls(struct sk_buff *skb,
> + const struct ovs_action_push_mpls *mpls,
> + unsigned *mpls_stack_depth)
> +{
> + __be32 *new_mpls_ls
On Apr 19, 2013, at 10:41 , ext Simon Horman wrote:
> diff --git a/datapath/actions.c b/datapath/actions.c
> index 0dac658..2c923be 100644
> --- a/datapath/actions.c
> +++ b/datapath/actions.c
> @@ -38,6 +38,7 @@
> #include "vport.h"
>
> static int do_execute_actions(struct datapath *dp, struct
On Apr 3, 2013, at 8:24 , ext Simon Horman wrote:
> In the case where a non-MPLS packet is recieved and an MPLS stack is
> added it may well be the case that the original skb is GSO but the
> NIC used for transmit does not support GSO of MPLS packets.
>
...
> diff --git a/include/linux/netdev_fe
On Mar 30, 2013, at 18:20 , ext Pravin B Shelar wrote:
> diff --git a/datapath/linux/compat/include/linux/netdevice.h
> b/datapath/linux/compat/include/linux/netdevice.h
> index 71aad87..4ffdac4 100644
> --- a/datapath/linux/compat/include/linux/netdevice.h
> +++ b/datapath/linux/compat/include/
On Mar 30, 2013, at 18:20 , ext Pravin B Shelar wrote:
> Following patch restructure ovs tunneling to make use of kernel
> api. Doing this tunneling code is simplified as most of protocol
> processing on send and recv is pushed to kernel tunneling. This
> way we can share most protocol related co
On Mar 27, 2013, at 1:47 , ext Jesse Gross wrote:
> On Mon, Mar 25, 2013 at 12:03 PM, Jarno Rajahalme
> wrote:
>> Changes the default tunnel dont_fragment from "true" (don't
>> fragment) to "false" (allow fragmentation). Tunnel outer headers
>> will not have the DF bit set by default, and if "d
On Mar 9, 2013, at 16:54 , ext Simon Horman wrote:
> It is not obvious to me how the behaviour relating to get_priority() as
> described above can sensibly be reconciled with a desire to preserve the
> order of actions. Or even reconciled with a looser constraint that actions
> that need l3+ data
On Mar 9, 2013, at 16:54 , ext Simon Horman wrote:
> On Thu, Mar 07, 2013 at 11:12:31AM -0800, Ben Pfaff wrote:
>> On Thu, Mar 07, 2013 at 04:12:22AM +0100, Simon Horman wrote:
>>> Perhaps Ben has some ideas, but it seems to me that the
>>> way that execute_* and compose_* interact needs to be ch
Ben,
Thanks for fixing that up, looks good :-)
Jarno
On Mar 8, 2013, at 18:52 , ext Ben Pfaff wrote:
> On Fri, Mar 08, 2013 at 03:50:12PM +0200, Jarno Rajahalme wrote:
>> Remove recursion from GOTO_TABLE processing in do_xlate_actions().
>> This allows packet processing pipelines built with g
I recall someone mentioning on this list that the only planned use for skb_mark
is for ipsec tunneling. At least currently this seems to be the case, as the
only place where the skb_mark is set to a potentially non-zero value in
userspace is in tnl_port_send(). The kernel datapath may also provi
On Feb 28, 2013, at 11:15 , ext Simon Horman wrote:
> * Implement recirculation and a host of more sophisticated MPLS features
> including:
> - copy_ttl_in
> - copy_ttl_out
> - L3+ actions on MPLS frames (e.g. pop_mpls(0x800),set_ttl)
> - multi-level mpls_pop and mpls_push
> * Other more speci
On Feb 20, 2013, at 11:00 , ext Lorand Jakab wrote:
> +static int lisp_tnl_send(struct vport *vport, struct sk_buff *skb)
> +{
> + int network_offset = skb_network_offset(skb);
> +
> + /* We only encapsulate IPv4 and IPv6 packets */
> + switch (skb->protocol) {
> + case htons(ETH_
On Feb 20, 2013, at 18:00 , ext Kyle Mestery (kmestery) wrote:
> On Feb 19, 2013, at 7:35 PM, Pravin B Shelar wrote:
>> Since userspace flow based tunneling code is checked in, the kernel
>> port based tunneling code can be removed. Following patch removes
>> this tunnel compatibility code and s
On Feb 19, 2013, at 21:38 , ext Jesse Gross wrote:
> On Mon, Feb 18, 2013 at 12:13 AM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> I'd be happy to break it apart.
>>
>> In datapath.c ovs_flow_cmd_build_info(), there is first a call to
>> ovs_fl
On Feb 16, 2013, at 1:19 , ext Jesse Gross wrote:
> On Mon, Feb 11, 2013 at 6:46 AM, Jarno Rajahalme
> wrote:
>> Store key_len and hash fields at the end of struct sw_flow_key, past the
>> area being hashed.
>> Rename functions operating on keys from "_flow_" to "_flow_key_".
>> Shift the respon
On Feb 13, 2013, at 16:44 , ext Lorand Jakab wrote:
> +static int lisp_tnl_send(struct vport *vport, struct sk_buff *skb)
> +{
> + int network_offset = skb_network_offset(skb);
> +
> + /* We only encapsulate IPv4 and IPv6 packets */
> + switch (ntohs(skb->protocol)) {
> + case ETH_
On Feb 12, 2013, at 18:52 , ext Ben Pfaff wrote:
> On Tue, Feb 12, 2013 at 03:31:16PM +0000, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>> The patches look good, but as I haven't seen the EXT-272, I don't know
>> if there is a possible conflict with the all-one
On Feb 13, 2013, at 0:03 , ext Ben Pfaff wrote:
> On Mon, Feb 11, 2013 at 04:46:16PM +0200, Jarno Rajahalme wrote:
>> - 'udpping' sends 1000 UDP packets with a unique source and destination
>> port pair to br0 in bursts of 97 packets. Between bursts it nanosleeps
>> for 50 nanoseconds. Th
The patches look good, but as I haven't seen the EXT-272, I don't know if there
is a possible conflict with the all-ones value returned when the generation id
is undefined. Have you proposed to exclude the all-ones value from the set of
valid generation ID values? Or does it matter?
Jarno
On
On Feb 12, 2013, at 1:54 , ext Ben Pfaff wrote:
> On Mon, Feb 11, 2013 at 04:46:17PM +0200, Jarno Rajahalme wrote:
>> Take ofproto-dpif upcall recv pooling down to the system call interface.
>>
>> Signed-off-by: Jarno Rajahalme
>
> I tried this out with my test case. I found that it yields a
On Feb 11, 2013, at 20:35 , ext Ben Pfaff wrote:
> I find this organization clearer.
>
Agree, but see the minor comments below,
Jarno
> CC: Jarno Rajahalme
> Signed-off-by: Ben Pfaff
> ---
> lib/classifier.c | 190 --
> 1 files changed, 1
On Feb 7, 2013, at 22:36 , ext Ben Pfaff wrote:
> With complicated flow tables and multiple levels of resubmit, I see
> flow setup performance improvements of up to 5-6% with this patch.
>
I played around with the same issue last week and came up with a
little bit different solution. I think I
On Feb 6, 2013, at 2:57 , ext Jesse Gross wrote:
>>
>> While I would not see a problem in passing a key hash computed with an
>> undisclosed algorithm up to userspace, I see your point.
>>
>> In my tests it seems that about 1/3 of the benefit is attainable with
>> deferred
>> layer pointer comp
On Jan 31, 2013, at 3:19 , ext Jesse Gross wrote:
> On Wed, Jan 30, 2013 at 3:14 AM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> Otherwise not so much use of the layer pointers, but
>> dpif_flow_stats_extract()
>> does packet_get_tcp_flags(),
On Jan 30, 2013, at 2:59 , ext Jesse Gross wrote:
> On Tue, Jan 29, 2013 at 7:10 AM, Rajahalme, Jarno (NSN - FI/Espoo)
> wrote:
>>
>> On Jan 24, 2013, at 19:41 , ext Jesse Gross wrote:
>>
>>> On Thu, Jan 24, 2013 at 7:34 AM, Jarno Rajahalme
>>> w
On Jan 24, 2013, at 19:41 , ext Jesse Gross wrote:
> On Thu, Jan 24, 2013 at 7:34 AM, Jarno Rajahalme
> wrote:
>>
>> On Jan 23, 2013, at 19:30 , ext Jesse Gross wrote:
>>
>>> On Tue, Jan 22, 2013 at 9:48 PM, Jarno Rajahalme
>>> wrote:
Add OVS_PACKET_ATTR_KEY_INFO to relieve userspace fro
92 matches
Mail list logo
