Converting the flow key and mask back into netlink format during flow
dump is fairly expensive. By caching the netlink versions of these
during flow setup, and copying the memory directly during flow dump, we
are able to support up to 33% more flows in the datapath. Flow setup
rate decreases by aro
Split up ovs_flow_cmd_fill_info() to make it easier to cache parts of a
dump reply. This will be used to streamline flow_dump in the next patch.
Signed-off-by: Joe Stringer
---
datapath/datapath.c | 87 ---
1 file changed, 69 insertions(+), 18 de
We noticed that the unit tests sometimes fail on XenServer inside glibc's
memory checker, in the free_check() function. It turns out that the
glibc memory checker in glibc 2.11 and earlier had an internal race that
caused false positives in multithreaded programs.
This commit avoids the problem b
If megaflows are disabled, the userspace does not send the netlink attribute
OVS_FLOW_ATTR_MASK, and the kernel must create an exact match mask.
sw_flow_mask_set() sets every bytes (in 'range') of the mask to 0xff, even the
bytes that represent padding for struct sw_flow, or the bytes that represe
match_validate() enforce that a mask matching on NDP attributes has also an
exact match on ICMPv6 type.
The ICMPv6 type, which is 8-bit wide, is stored in the 'tp.src' field of
'struct sw_flow_key', which is 16-bit wide.
Therefore, an exact match on ICMPv6 type should only check the first 8 bits.
OK, I'll reply again when I have extended this forked-repo to export the
standard sFlow-LAG structure.
Neil
--
Neil McKee
InMon Corp.
http://www.inmon.com
On Mon, Jul 7, 2014 at 8:55 AM, Ben Pfaff wrote:
> I took a look at the patch. It's not necessary to submit a
> pull-request.
>
> Th
On 7/9/14, 4:50 PM, Thomas Graf wrote:
On 06/27/14 at 04:21pm, Lorand Jakab wrote:
diff --git a/lib/odp-execute.c b/lib/odp-execute.c
index cb89e72..b5add78 100644
--- a/lib/odp-execute.c
+++ b/lib/odp-execute.c
@@ -260,6 +260,25 @@ odp_execute_actions__(void *dp, struct dpif_packet
**packets,
On 06/27/14 at 04:21pm, Lorand Jakab wrote:
> diff --git a/lib/odp-execute.c b/lib/odp-execute.c
> index cb89e72..b5add78 100644
> --- a/lib/odp-execute.c
> +++ b/lib/odp-execute.c
> @@ -260,6 +260,25 @@ odp_execute_actions__(void *dp, struct dpif_packet
> **packets, int cnt,
> break;
Hi Ben,
Thanks for pointing this out.
On 08/07/14 18:10, Ben Pfaff wrote:
Hi Anoob. Guru noticed (and I confirmed) that recent versions of Open
vSwitch on master suffer a little, performance-wise, on XenServer. In
particular, the cmap code doesn't perform well. In the OVS testsuite,
the cmap
