On Tue, Jun 11, 2013 at 02:26:32PM -0700, Jesse Gross wrote:
> On Mon, Jun 10, 2013 at 1:07 AM, Simon Horman wrote:
> > diff --git a/datapath/actions.c b/datapath/actions.c
> > index 0dac658..197811a 100644
> > --- a/datapath/actions.c
> > +++ b/datapath/actions.c
> > +static int push_mpls(struct
Thanks for the quick review. I pushed this to master and branch-1.11.
--Justin
On Jun 11, 2013, at 11:03 PM, Ethan Jackson wrote:
> Acked-by: Ethan Jackson
>
> On Tue, Jun 11, 2013 at 11:00 PM, Justin Pettit wrote:
>> The handle_flow_miss_with_facet() function used the facet's flow
>> info
Acked-by: Ethan Jackson
On Tue, Jun 11, 2013 at 11:00 PM, Justin Pettit wrote:
> The handle_flow_miss_with_facet() function used the facet's flow
> information instead of the missed flow. This corrects that.
>
> Signed-off-by: Justin Pettit
> ---
> ofproto/ofproto-dpif.c |4 ++--
> 1 file
We really need to get rid of this compose_slow_path business, but for
now this seems fine to me. Thanks
Acked-by: Ethan Jackson
On Tue, Jun 11, 2013 at 11:00 PM, Justin Pettit wrote:
> It's often helpful to see what the slow path actions actual are. Print
> them when "ovs-appctl dpif/dump-meg
It's often helpful to see what the slow path actions actual are. Print
them when "ovs-appctl dpif/dump-megaflows" is called.
Signed-off-by: Justin Pettit
---
ofproto/ofproto-dpif.c | 15 +--
1 files changed, 13 insertions(+), 2 deletions(-)
diff --git a/ofproto/ofproto-dpif.c b/o
The handle_flow_miss_with_facet() function used the facet's flow
information instead of the missed flow. This corrects that.
Signed-off-by: Justin Pettit
---
ofproto/ofproto-dpif.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofpro
Reviewed-by: Simon Horman
Signed-off-by: Joe Stringer
---
v3: Rebase
v2: Add flowmod parse tests
Add set-field tests
Add NEWS item
Add sctp to ovs-ofctl.8
---
NEWS |1 +
OPENFLOW-1.1+|4 ---
lib/match.c |4 +++
lib/meta-flow.c
Reviewed-by: Simon Horman
Signed-off-by: Joe Stringer
---
v3: Rebase
v2: Fix broken test
Use the correct packet pointer
Calculate checksums as delta from incoming checksum
---
include/linux/openvswitch.h |6 +
include/sparse/netinet/in.h |1 +
lib/flow.c |
This implementation was derived from FreeBSD:
http://code.google.com/p/freebsd-head/source/browse/sys/libkern/crc32.c
Reviewed-by: Simon Horman
Signed-off-by: Joe Stringer
---
v3: Rebase
Return crc32c checksum in network byte-order
Add checksum tests based on Intel IPPs documentation
v2:
Signed-off-by: Joe Stringer
---
debian/copyright.in |4
1 file changed, 4 insertions(+)
diff --git a/debian/copyright.in b/debian/copyright.in
index a1a2c3c..3bcf332 100644
--- a/debian/copyright.in
+++ b/debian/copyright.in
@@ -101,6 +101,10 @@ License:
The full text of each license
This patchset introduces matching and rewriting support for sctp src,dst ports.
Round three adds further crc32c tests on the userspace side and adds support
for skb fragments on the kernel side. Various sparse errors and previous
feedback have also been addressed.
This series is also available at:
This patch adds support for rewriting SCTP src,dst ports similar to the
functionality already available for TCP/UDP.
Rewriting SCTP ports is expensive due to double-recalculation of the
SCTP checksums; this is performed to ensure that packets traversing OVS
with invalid checksums will continue to
Did not realize that the __CHECK_ENDIAN__ flags has to be passed by hand to
enable those checks. This is what I use to verify the fix. Any better way
to enable this check?
make C=2 CF="-D__CHECK_ENDIAN__"
The following incremental patch fixes those warnings.
Andy
diff --git a/datapath/flow.c
As suggested by Jesse in the comment for patch "gre: Restructure
tunneling", following patch keeps skb->csum correct across ovs.
Signed-off-by: Pravin B Shelar
---
v3-v4:
- use correct len for csum.
v2-v3:
- moved skb_postpull_rcsum() to private ovs header.
- used ETH_ALEN to replace rcsum.
v1-v2
Thanks. I pushed this to master.
--Justin
On Jun 11, 2013, at 5:30 PM, Ethan Jackson wrote:
> Acked-by: Ethan Jackson
>
> On Tue, Jun 11, 2013 at 5:21 PM, Justin Pettit wrote:
>> The is_gratuitous_arp() function is occasionally called when
>> processing the "normal" action. The previous c
Acked-by: Ethan Jackson
On Tue, Jun 11, 2013 at 5:21 PM, Justin Pettit wrote:
> The is_gratuitous_arp() function is occasionally called when
> processing the "normal" action. The previous code only disabled
> wildcarding the dl_type field when the function was called, but
> since it runs occasi
On Tue, Jun 11, 2013 at 4:45 PM, Andy Zhou wrote:
> Add mega flow support in kernel datapath.
>
> Pravin has made significant contributions to this patch. Including
> the mega flow id look up scheme, API clean ups, and bug fixes.
>
> Co-authored-by: Pravin B Shelar
> Signed-off-by: Pravin B Shela
The is_gratuitous_arp() function is occasionally called when
processing the "normal" action. The previous code only disabled
wildcarding the dl_type field when the function was called, but
since it runs occasionally, it could lead to inconsistencies in the
facet table. This commit causes the dl_t
Thanks!
On Tue, Jun 11, 2013 at 05:03:46PM -0700, Alex Wang wrote:
> Thanks Ben,
>
> The factors are all reasonable. Especially, the negation can be confusing.
>
> Also, with you confirmation, I'll choose the "oft_to_u16()", "odp_to_u32()"
> way. ;D
>
> I'll adjust the code accordingly and rese
Thanks Ben,
The factors are all reasonable. Especially, the negation can be confusing.
Also, with you confirmation, I'll choose the "oft_to_u16()", "odp_to_u32()"
way. ;D
I'll adjust the code accordingly and resend a new version.
Kind Regards,
Alex Wang,
On Tue, Jun 11, 2013 at 4:55 PM, Ben P
On Tue, Jun 11, 2013 at 03:17:17PM -0700, Justin Pettit wrote:
> This is a backport of the facet wildcarding code to 1.11. The patch that
> should be the most closely looked at is "ofproto-dpif: Consolidate facet
> stat logic.", since I had to modify the facet_flush_stats() and
> rule_get_stats()
On Mon, Jun 10, 2013 at 08:50:40AM -0700, Alex Wang wrote:
> Subject: [PATCH] This patch implements "ofp_is_less_than()" and use it
> to replace the PORT_COMPARE() function for port comparison.
> Date: Mon, 10 Jun 2013 09:05:50 -0700
> Message-Id: <1370880350-22000-1-git-send-email-al...@nicira.com
On Tue, Jun 11, 2013 at 04:10:34PM -0700, Alex Wang wrote:
> On Tue, Jun 11, 2013 at 3:59 PM, Ben Pfaff wrote:
>
> > On Fri, Jun 07, 2013 at 11:11:43AM -0700, Alex Wang wrote:
> > > Currently datapath ports and openflow ports are both represented by
> > unsigned
> > > integers of various sizes. W
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
On Tue, Jun 11, 2013 at 3:54 PM, Pravin B Shelar wrote:
> diff --git a/datapath/actions.c b/datapath/actions.c
> index 0dac658..f9da8a4 100644
> --- a/datapath/actions.c
> +++ b/datapath/actions.c
> @@ -132,9 +132,17 @@ static int set_eth_addr(struct sk_buff *skb,
> if (unlikely(err))
>
On Tue, Jun 11, 2013 at 3:59 PM, Ben Pfaff wrote:
> On Fri, Jun 07, 2013 at 11:11:43AM -0700, Alex Wang wrote:
> > Currently datapath ports and openflow ports are both represented by
> unsigned
> > integers of various sizes. With implicit casts, etc. it is easy to mix
> them
> > up and use one wh
On Fri, Jun 07, 2013 at 11:11:43AM -0700, Alex Wang wrote:
> Currently datapath ports and openflow ports are both represented by unsigned
> integers of various sizes. With implicit casts, etc. it is easy to mix them
> up and use one where the other is expected. This commit creates two typedefs
> of
On Tue, Jun 11, 2013 at 3:30 PM, Jesse Gross wrote:
> On Tue, Jun 11, 2013 at 1:20 PM, Pravin B Shelar wrote:
>> diff --git a/datapath/actions.c b/datapath/actions.c
>> index 0dac658..6d60cd0 100644
>> --- a/datapath/actions.c
>> +++ b/datapath/actions.c
>> @@ -132,9 +132,17 @@ static int set_eth
As suggested by Jesse in the comment for patch "gre: Restructure
tunneling", following patch keeps skb->csum correct across ovs.
Signed-off-by: Pravin B Shelar
---
v2-v3:
- moved skb_postpull_rcsum() to private ovs header.
- used ETH_ALEN to replace rcsum.
v1-v2:
- Fixed lisp and vlan_deaccel_tag
On Jun 11, 2013, at 12:52 PM, Ben Pfaff wrote:
> On Tue, Jun 11, 2013 at 12:01:24PM -0700, Justin Pettit wrote:
>
>> An incremental follows.
>
> This seems reasonable.
Thanks for the reviews! I pushed this to master and sent out a backported
version for branch-1.11.
--Justin
_
On Tue, Jun 11, 2013 at 1:20 PM, Pravin B Shelar wrote:
> diff --git a/datapath/actions.c b/datapath/actions.c
> index 0dac658..6d60cd0 100644
> --- a/datapath/actions.c
> +++ b/datapath/actions.c
> @@ -132,9 +132,17 @@ static int set_eth_addr(struct sk_buff *skb,
> if (unlikely(err))
>
From: Ethan Jackson
The logic for updating statistics at the facet level had been
spread through ofproto-dpif in a rather confusing manner. This
patch consolidates as much of this logic as is reasonable into
facet_push_stats().
On a side note, I'd expect this patch to have a marginal positive
p
From: Ethan Jackson
This function will be useful in a future commit.
Signed-off-by: Ethan Jackson
Co-authored-by: Justin Pettit
Signed-off-by: Justin Pettit
---
lib/flow.c | 43 +--
lib/flow.h |3 +++
2 files changed, 32 insertions(+), 14 deletio
This is a backport of the facet wildcarding code to 1.11. The patch that
should be the most closely looked at is "ofproto-dpif: Consolidate facet
stat logic.", since I had to modify the facet_flush_stats() and
rule_get_stats() to prevent double-counting. The others were very
straight-forward.
Et
From: Ethan Jackson
Rename the function flow_wildcards_combine() to flow_wildcards_and().
Add new flow_wildcards_or() and flow_hash_in_wildcards() functions.
These will be useful in a future patch.
Signed-off-by: Ethan Jackson
Signed-off-by: Justin Pettit
---
lib/flow.c | 48 ++
From: Ethan Jackson
A future commit will want to know what bits were significant during the
classifier lookup.
Signed-off-by: Ethan Jackson
Co-authored-by: Justin Pettit
Signed-off-by: Justin Pettit
---
lib/classifier.c| 17 +++--
lib/classifier.h|3 ++-
ofp
On Mon, Jun 10, 2013 at 1:07 AM, Simon Horman wrote:
> diff --git a/datapath/actions.c b/datapath/actions.c
> index 0dac658..197811a 100644
> --- a/datapath/actions.c
> +++ b/datapath/actions.c
> +static int push_mpls(struct sk_buff *skb,
> +const struct ovs_action_push_mpls *m
Thanks, applied.
On Tue, Jun 11, 2013 at 01:40:52PM -0700, Justin Pettit wrote:
> Acked-by: Justin Pettit
>
>
> On Jun 11, 2013, at 1:38 PM, Ben Pfaff wrote:
>
> > list_size() returns a size_t, not a uint64_t.
> >
> > Found by GCC.
> >
> > Signed-off-by: Ben Pfaff
> > ---
> > ofproto/ofpro
Acked-by: Justin Pettit
On Jun 11, 2013, at 1:38 PM, Ben Pfaff wrote:
> list_size() returns a size_t, not a uint64_t.
>
> Found by GCC.
>
> Signed-off-by: Ben Pfaff
> ---
> ofproto/ofproto-dpif.c |3 +--
> 1 files changed, 1 insertions(+), 2 deletions(-)
>
> diff --git a/ofproto/ofproto
list_size() returns a size_t, not a uint64_t.
Found by GCC.
Signed-off-by: Ben Pfaff
---
ofproto/ofproto-dpif.c |3 +--
1 files changed, 1 insertions(+), 2 deletions(-)
diff --git a/ofproto/ofproto-dpif.c b/ofproto/ofproto-dpif.c
index 8df2679..8bc3352 100644
--- a/ofproto/ofproto-dpif.c
+
On Mon, Jun 10, 2013 at 02:13:26PM -0700, Gurucharan Shetty wrote:
> On Mon, Apr 29, 2013 at 3:23 PM, Ben Pfaff wrote:
>
> > I don't see any reason for this variable to be static.
> >
> Looks good to me, thanks.
Thanks, applied to master.
___
dev maili
On Wed, Jun 05, 2013 at 05:27:32PM -0700, Ethan Jackson wrote:
> By detecting that a port is a vlan splinter realdev, we can force
> xlate_actions() to emit the appropriate vlan push action. This
> allows as to ditch struct initial_vals. It will not be missed.
>
> Signed-off-by: Ethan Jackson
>
CC: Ethan Jackson
Signed-off-by: Ben Pfaff
---
lib/dummy.c |3 +-
lib/dummy.h |3 +-
lib/vlandev.c | 346 +++
lib/vlandev.h |8 +-
tests/automake.mk |1 +
tests/ofproto-macros.at |9
As suggested by Jesse in the comment for patch "gre: Restructure
tunneling", following patch keeps skb->csum correct across ovs.
Signed-off-by: Pravin B Shelar
---
v1-v2:
- Fixed lisp and vlan_deaccel_tag()
---
datapath/actions.c |8
datapath/flow.c
On Tue, Jun 11, 2013 at 12:01:24PM -0700, Justin Pettit wrote:
> On Jun 11, 2013, at 9:41 AM, Ben Pfaff wrote:
>
> > On Tue, Jun 11, 2013 at 01:20:32AM -0700, Justin Pettit wrote:
> >> Dynamically determines the flow fields that were relevant in
> >> processing flows based on the OpenFlow flow ta
On Jun 11, 2013, at 9:41 AM, Ben Pfaff wrote:
> On Tue, Jun 11, 2013 at 01:20:32AM -0700, Justin Pettit wrote:
>> Dynamically determines the flow fields that were relevant in
>> processing flows based on the OpenFlow flow table and switch
>> configuration. The immediate use for this functionali
On Fri, Jun 07, 2013 at 01:35:22PM -0700, Andy Zhou wrote:
> Added --mega option to ovs-dpctl command to allow mega flow to be
> specified and displayed. ovs-dpctl tool is mainly used as debugging
> tool.
>
> This patch also implements the low level user space routines to send
> and receive mega
On Tue, Jun 11, 2013 at 11:05:55AM +0900, Joe Stringer wrote:
> Signed-off-by: Joe Stringer
Applied, thanks.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Tue, Jun 11, 2013 at 01:20:32AM -0700, Justin Pettit wrote:
> Dynamically determines the flow fields that were relevant in
> processing flows based on the OpenFlow flow table and switch
> configuration. The immediate use for this functionality is to
> cache action translations for similar flows
On Tue, Jun 11, 2013 at 01:20:31AM -0700, Justin Pettit wrote:
> From: Ethan Jackson
>
> Rename the function flow_wildcards_combine() to flow_wildcards_and().
> Add new flow_wildcards_or() and flow_hash_in_wildcards() functions.
> These will be useful in a future patch.
>
> Signed-off-by: Ethan
This patch series is a replacement for the final patch in my "xc"
series. Instead of keeping an external cache of xout translations, we
store wildcards directly in facets and create exact-match flows in
subfacets. This is a less intrusive change and less brittle than the
xout cache method. Thank
From: Ethan Jackson
Rename the function flow_wildcards_combine() to flow_wildcards_and().
Add new flow_wildcards_or() and flow_hash_in_wildcards() functions.
These will be useful in a future patch.
Signed-off-by: Ethan Jackson
Signed-off-by: Justin Pettit
---
lib/flow.c | 48 ++
On Jun 7, 2013, at 1:04 PM, Ben Pfaff wrote:
> I think that nxm_execute_reg_move() and nxm_execute_stack_push() could
> just mask out the subfield that is read, not the whole field that
> contains the subfield.
Fair enough. Should be addressed in next patch.
> The 'wc' member of xlate_out is
On Jun 7, 2013, at 2:51 AM, "Rajahalme, Jarno (NSN - FI/Espoo)"
wrote:
>> @@ -3764,7 +3880,7 @@ handle_flow_miss_with_facet(struct flow_miss *miss,
>> struct facet *facet,
>>struct ofpbuf *packet;
>>
>>subfacet = subfacet_create(facet, miss, now);
>> -want_path = subfacet->facet->x
54 matches
Mail list logo
