I've folded in the following incremental which causes us to only do the core
calculation once since it looks so expensive. I think this is fine for now, and
I intended to merge it, but I think we should revisit the assembler instruction.
It's probably cleaner, and would support platforms like Wind
Normally OVS userspace supplies a mask along with a flow key for each
new data path flow that should be created. OVS also provides an
option to disable the kernel wildcarding, in which case the flows are
created without a mask. When kernel wildcarding is disabled, the
datapath should use exact ma
On 10/12/2013 04:50, Jarno Rajahalme wrote:
On Dec 8, 2013, at 7:34 PM, Alexander Wu wrote:
Hi Jarno,
I get my gcc predefined __core2. But its performance seems to be worse when
I add '-O2'. Not sure if it's the reality.
From the numbers below it seems that performance is better with -O2
The only thing is I didn't follow is:
"""
Do we really still need the bfd_forwarding_if_rx_update() function?
It's so short and only called in this one place, could we just move
it's code there?
"""
Since there are two other places where bfd_forwarding_if_rx_update() are
called, and I think they
Currently, we update the forwarding flag in bfd_set_state() and
in bfd_forwarding_if_rx_update() if bfd_forwarding_if_rx is enabled.
However, these are not the exact places where the forwarding flag
needs to be updated. The exact places are in the bfd_process_packet()
where bfd status are changed
A Windows porter mentioned to me that these functions caused special
trouble in the Windows port. However, they are no longer used, so we
might as well remove them.
Signed-off-by: Ben Pfaff
---
lib/socket-util.c | 254 -
lib/socket-util.h |
On Mon, Dec 09, 2013 at 03:52:39PM -0800, Jarno Rajahalme wrote:
> On Dec 7, 2013, at 2:04 PM, Ben Pfaff wrote:
> > On Thu, Dec 05, 2013 at 04:27:26PM -0800, Jarno Rajahalme wrote:
> >> Add a prefix tree (trie) structure for tracking the used address
> >> space, enabling skipping classifier tables
On Dec 7, 2013, at 2:04 PM, Ben Pfaff wrote:
> On Thu, Dec 05, 2013 at 04:27:26PM -0800, Jarno Rajahalme wrote:
>> Add a prefix tree (trie) structure for tracking the used address
>> space, enabling skipping classifier tables containing longer masks
>> than necessary for an address field value i
Thanks Ethan for the comments, I'll adjust accordingly,
On Mon, Dec 9, 2013 at 3:38 PM, Ethan Jackson wrote:
> On Mon, Dec 9, 2013 at 1:40 PM, Alex Wang wrote:
> > Currently, we updates the forwarding flag in bfd_set_state() and
> > in bfd_forwarding_if_rx_update() if bfd_forwarding_if_rx is e
k sounds good.
Ethan
On Mon, Dec 9, 2013 at 3:21 PM, Alex Wang wrote:
> okay, this makes sense. we could drop this patch.
>
>
> On Mon, Dec 9, 2013 at 3:18 PM, Ethan Jackson wrote:
>>
>> Doesn't the next patch essentially resolve this issue by making the
>> forwarding_if_rx calculation more re
On Mon, Dec 9, 2013 at 1:40 PM, Alex Wang wrote:
> Currently, we updates the forwarding flag in bfd_set_state() and
> in bfd_forwarding_if_rx_update() if bfd_forwarding_if_rx is enabled.
> However, these are not the exact places where the forwarding flag
> needs to be updated. The exact places ar
okay, this makes sense. we could drop this patch.
On Mon, Dec 9, 2013 at 3:18 PM, Ethan Jackson wrote:
> Doesn't the next patch essentially resolve this issue by making the
> forwarding_if_rx calculation more responsive? With the new threaded
> put stuff that's coming, we can more or less gua
Doesn't the next patch essentially resolve this issue by making the
forwarding_if_rx calculation more responsive? With the new threaded
put stuff that's coming, we can more or less guarantee that stats will
be updated at worst once every 1.5 seconds which makes me think that
3000 seconds is overki
Currently, we updates the forwarding flag in bfd_set_state() and
in bfd_forwarding_if_rx_update() if bfd_forwarding_if_rx is enabled.
However, these are not the exact places where the forwarding flag
needs to be updated. The exact places are in the bfd_process_packet()
where bfd status are changed
Currently, the ovs updates the rx_packet counter approximately every
2000 ms. If there is rx_packet and forwarding_if_rx is enabled, the
forwarding_if_rx_detect_time will be incremented by 2000 ms. And as
long as forwarding_if_rx_detect_time is greater than current time,
the forwarding flag value
Acked-by: Jarno Rajahalme
On Dec 7, 2013, at 10:48 AM, Ben Pfaff wrote:
> Building with GCC 4.7 without __corei7:
>
>../lib/util.c:921:15: warning: symbol 'count_1bits_8' was not declared.
> Should it be static?
>
> Here is one possible fix, but perhaps you prefer a different one.
>
> d
On Dec 8, 2013, at 7:34 PM, Alexander Wu wrote:
> Hi Jarno,
>
> I get my gcc predefined __core2. But its performance seems to be worse when
> I add '-O2'. Not sure if it's the reality.
>
>From the numbers below it seems that performance is better with -O2 (1063893 <
>1317450), so I’m not sur
I would leave it up too you. An easier solution would be to put in the Vstudio
includes first(in the case for cccl) because there is no need for the string.h
wrapper.
Kind Regards,
Alin.
From: Ben Pfaff [b...@nicira.com]
Sent: Friday, December 06, 2013 10
Ben Pfaff schrieb:
>On Mon, Dec 09, 2013 at 10:32:06AM +0100, Helmut Schaa wrote:
>> Currently openvswitch builds all libraries static only. However,
>> libopenvswitch is linked into nearly all openvswitch executables
>> making it hardly possible to run openvswitch on embedded devices
>> (for
On Mon, Dec 09, 2013 at 10:32:06AM +0100, Helmut Schaa wrote:
> Currently openvswitch builds all libraries static only. However,
> libopenvswitch is linked into nearly all openvswitch executables
> making it hardly possible to run openvswitch on embedded devices
> (for example running OpenWrt).
>
On Mon, Dec 09, 2013 at 09:26:08AM +0100, Helmut Schaa wrote:
> On Mon, Dec 9, 2013 at 9:12 AM, Helmut Schaa
> wrote:
> > On Fri, Dec 6, 2013 at 7:58 PM, Ben Pfaff wrote:
> >> I think that your binaries must include debug symbols. I don't know why
> >> you would include debug symbols in binaries
On Mon, Dec 09, 2013 at 09:12:44AM +0100, Helmut Schaa wrote:
> On Fri, Dec 6, 2013 at 7:58 PM, Ben Pfaff wrote:
> > On Wed, Nov 27, 2013 at 10:35:39AM +0100, Helmut Schaa wrote:
> >> 7.5M./sbin/ovs-vswitchd
>
> > I think that your binaries must include debug symbols. I don't know why
> > you
Allows all socket operations to use ipv6 network addresses,
except in-band control which is not supported for ipv6
Signed-off-by: Nandan Nivgune
Signed-off-by: Abhijit Bhopatkar
Signed-off-by: Arun Sharma
---
lib/packets.h| 12 +++
lib/rconn.c | 39 ++-
Signed-off-by: Nandan Nivgune
Signed-off-by: Abhijit Bhopatkar
Signed-off-by: Arun Sharma
---
lib/vconn-active.man |6 +++---
lib/vconn-passive.man| 10 ++
ovsdb/remote-active.man |9 +
ovsdb/remote-passive.man | 13 -
4 files changed, 22 inserti
Also added tests for NetFlow and sFlow using IPv6 sockets.
Signed-off-by: Nandan Nivgune
Signed-off-by: Abhijit Bhopatkar
Signed-off-by: Arun Sharma
---
tests/ofproto-dpif.at | 872 ---
tests/ofproto-macros.at | 17 +
tests/ovs-vsctl.at |
v4.3:
Rollback table id 255 to 253 according to OpenFlow spec and current implement
Update AT.
v4.2:
Fix last table features id to 255.
Add more comments.
Fix wrong print of last table feature.
v4:
1. Delete duplication code.
2. Add new acts in *.def.
3. Update abstract table-feat
v4.1:
Update bitmap_count1 function: call count_1bits to improve
performance.
v4:
Add bitmap_count1 function to count all 1 bits.
---
lib/bitmap.c | 15 +++
lib/bitmap.h |1 +
2 files changed, 16 insertions(+), 0 deletions(-)
diff --git a/lib/bitmap.c b/lib/bitmap.c
index
v4:
Add enums of ofp13_action_type and make them work.
v3:
Add func to get instruction/action name by OpenFlow instruction type directly.
---
include/openflow/openflow-1.3.h | 20 ++--
lib/ofp-actions.c | 19 +++
lib/ofp-actions.h
v4.3:
Fix last table features id to 253 according to current
imlement and OpenFlow spec.
v4.2:
Fix last table features id to 255.
v4:
1. update abstract table-features to bitmap
2. update ofproto, reduce copy when reply table features.
3. Move max_table_feature to oftable.
v3:
1. D
v4:
No change.
v3:
No change.
v2:
No change
v1:
Now the cli we implement is very crude. Maybe it could
display better.
Signed-off-by: Alexander Wu
Reviewed-by: Simon Horman
---
utilities/ovs-ofctl.c | 17 +
1 files changed, 17 insertions(+), 0 deletions(-)
diff -
v4.3:
Fix last table faetures id to 253 according to current implement
and OpenFlow specs.
Update table features AT.
v4.2:
Fix last table features id to 255.
Fix wrong prints of last table features.
Add more comments to describe the implement inner ovs:
The bitmaps stores table fea
Currently openvswitch builds all libraries static only. However,
libopenvswitch is linked into nearly all openvswitch executables
making it hardly possible to run openvswitch on embedded devices
(for example running OpenWrt).
Convert openvswitch to use libtool for building its internal libs.
This
v4.2:
Fix last table features id to 255.
Fix wrong prints of last table features.
Add more comments to describe the implement inner ovs:
The bitmaps stores table features OFPTFPT13_* properties.
Here are valid bits:
Instructions: OFPIT11_GOTO_TABLE ~ OFPIT13_METER
Next tab
v4:
Add enums of ofp13_action_type and make them work.
v3:
Add func to get instruction/action name by OpenFlow instruction type directly.
---
include/openflow/openflow-1.3.h | 20 ++--
lib/ofp-actions.c | 19 +++
lib/ofp-actions.h
v4.1:
Update bitmap_count1 function: call count_1bits to improve
performance.
v4:
Add bitmap_count1 function to count all 1 bits.
---
lib/bitmap.c | 15 +++
lib/bitmap.h |1 +
2 files changed, 16 insertions(+), 0 deletions(-)
diff --git a/lib/bitmap.c b/lib/bitmap.c
index
v4:
No change.
v3:
No change.
v2:
No change
v1:
Now the cli we implement is very crude. Maybe it could
display better.
Signed-off-by: Alexander Wu
Reviewed-by: Simon Horman
---
utilities/ovs-ofctl.c | 17 +
1 files changed, 17 insertions(+), 0 deletions(-)
diff -
v4.2:
Fix last table features id to 255.
Add more comments.
Fix wrong print of last table feature.
v4:
1. Delete duplication code.
2. Add new acts in *.def.
3. Update abstract table-features to bitmap.
4. Make decode_openflow13_props more general.
5. Update tests. Correct the wriab
v4.2:
Fix last table features id to 255.
v4:
1. update abstract table-features to bitmap
2. update ofproto, reduce copy when reply table features.
3. Move max_table_feature to oftable.
v3:
1. Delete variable OFPIT13_END, use OFPIT13_METER directly.
2. Fix implement, restructure with m
On Mon, Dec 9, 2013 at 9:12 AM, Helmut Schaa
wrote:
> On Fri, Dec 6, 2013 at 7:58 PM, Ben Pfaff wrote:
>> I think that your binaries must include debug symbols. I don't know why
>> you would include debug symbols in binaries meant for a
>> space-constrained application.
>
> Nope, I don't include
On Fri, Dec 6, 2013 at 8:04 PM, Ben Pfaff wrote:
> On Fri, Dec 06, 2013 at 08:24:03AM -0800, Jarno Rajahalme wrote:
>>
>> On Dec 6, 2013, at 1:59 AM, Helmut Schaa wrote:
>>
>> > On Wed, Nov 27, 2013 at 10:35 AM, Helmut Schaa
>> > wrote:
>> >> Currently openvswitch builds all libraries static onl
On Fri, Dec 6, 2013 at 7:58 PM, Ben Pfaff wrote:
> On Wed, Nov 27, 2013 at 10:35:39AM +0100, Helmut Schaa wrote:
>> Currently openvswitch builds all libraries static only. However,
>> libopenvswitch is linked into nearly all openvswitch executables
>> making it hardly possible to run openvswitch o
On Fri, Dec 6, 2013 at 5:24 PM, Jarno Rajahalme wrote:
>
> On Dec 6, 2013, at 1:59 AM, Helmut Schaa wrote:
>
>> On Wed, Nov 27, 2013 at 10:35 AM, Helmut Schaa
>> wrote:
>>> Currently openvswitch builds all libraries static only. However,
>>> libopenvswitch is linked into nearly all openvswitch e
On Sat, Dec 7, 2013 at 6:20 PM, Ben Pfaff wrote:
> On Fri, Dec 06, 2013 at 04:18:42PM +0100, Helmut Schaa wrote:
>> When adding a physical port to the main bridge the mac address
>> of the bridge is updated. We can do the same for fake bridges by
>> copying the mac address of the parent bridge.
>>
43 matches
Mail list logo
