To improve performance and avoid wasting resources for HW offloaded
flows, do not rewrite fields that are matched with the same value.
Signed-off-by: Eli Britstein
Reviewed-by: Roi Dayan
---
lib/odp-util.c| 122 +-
tests/mpls-xlate.at | 2 +-
Generate datapath ovs key fields macros as a pre-step of retrieving
field information, with no functional change.
Signed-off-by: Eli Britstein
Reviewed-by: Roi Dayan
---
.gitignore | 1 +
build-aux/extract-odp-netlink-xmacros-h | 56 +
inclu
This patch set avoids unnecessary rewrite actions to fields with the
same values as matched on.
Patch 1 is a pre-step of generating ovs key fields macros
Patch 2 avoids the unnecessary rewrites and adapts the tests accordingly
I see tests 1219-1220 fail, but also failed witout those patches.
The
Signed-off-by: Darrell Ball
---
lib/ipf.c | 6 ++
1 file changed, 2 insertions(+), 4 deletions(-)
diff --git a/lib/ipf.c b/lib/ipf.c
index df5196f..125a815 100644
--- a/lib/ipf.c
+++ b/lib/ipf.c
@@ -456,7 +456,7 @@ ipf_reassemble_v6_frags(struct ipf_list *ipf_list)
int rest_len = frag_l
ipf_reassemble_v4_frags() and ipf_reassemble_v6_frags() are
preallocating more than needed for the reassembled packet.
Signed-off-by: Darrell Ball
---
lib/ipf.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/ipf.c b/lib/ipf.c
index 125a815..ec7c889 100644
--- a/lib/i
Fixes: 4ea96698f667 ("Userspace datapath: Add fragmentation handling.")
Signed-off-by: Darrell Ball
---
lib/ipf.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/lib/ipf.c b/lib/ipf.c
index ec7c889..c386b69 100644
--- a/lib/ipf.c
+++ b/lib/ipf.c
@@ -612,7 +612,7 @@ ipf_is_
Bleep bloop. Greetings Eli Britstein, I am a robot and I have tried out your
patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Line is 83 characters long (recommended limit is 79)
#69 FILE: build-aux/extract-odp-
I resent this patch with an addition as part of a series
On Thu, Feb 14, 2019 at 5:44 PM Darrell Ball wrote:
> Signed-off-by: Darrell Ball
> ---
> lib/ipf.c | 4 +---
> 1 file changed, 1 insertion(+), 3 deletions(-)
>
> diff --git a/lib/ipf.c b/lib/ipf.c
> index df5196f..c91afd3 100644
> ---
Bleep bloop. Greetings Eli Britstein, I am a robot and I have tried out your
patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Line is 92 characters long (recommended limit is 79)
#87 FILE: lib/odp-util.c:7252:
#
Accidentially changed while updating conntrack support.
CC: Darrell Ball
Fixes: 4ea96698f667 ("Userspace datapath: Add fragmentation handling.")
Signed-off-by: Ilya Maximets
---
Documentation/faq/releases.rst | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/Documentation/faq/
should_steal is always true, not need to be passed
Signed-off-by: Li RongQing
---
lib/netdev-dpdk.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/lib/netdev-dpdk.c b/lib/netdev-dpdk.c
index f07b10c88..7ae29f3c2 100644
--- a/lib/netdev-dpdk.c
+++ b/lib/netdev-dpdk.c
Few more fixes + dummy implementation to enable unit testing
of this feature.
One more patch with refactoring:
* https://patchwork.ozlabs.org/patch/996328/
Version 4:
* Rebase on current master.
Version 3:
* Skip tests on non-Linux systems.
Version 2:
* Patch #3
It's a normal case for 'find' function, especially because this
happens for every first packet of flow that was not offloaded yet.
Should not warn about this. Dropped to DBG to avoid log trashing in
case of big number of new flows.
Fixes: 241bad15d99a ("dpif-netdev: associate flow with a mark id")
Getters should have const arguments.
Signed-off-by: Ilya Maximets
---
lib/dp-packet.h | 12 ++--
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/lib/dp-packet.h b/lib/dp-packet.h
index 5b13329da..c6672f6be 100644
--- a/lib/dp-packet.h
+++ b/lib/dp-packet.h
@@ -482,7 +482,7
Dummy interfaces clones dp-packet while 'receive' appctl processing.
In general, we should do this anyway to avoid any possible issues in
the future with real interfaces.
Signed-off-by: Ilya Maximets
---
lib/dp-packet.c | 4
1 file changed, 4 insertions(+)
diff --git a/lib/dp-packet.c b/li
Additionally, new API call 'dp_packet_set_flow_mark' is needed
for packet clone. Mostly for dummy HWOL implementation.
Signed-off-by: Ilya Maximets
---
lib/dp-packet.h | 23 +--
1 file changed, 21 insertions(+), 2 deletions(-)
diff --git a/lib/dp-packet.h b/lib/dp-packet.h
i
1. No reason to have mbuf related APIs in a generic code.
2. Not only RSS/checksums should be invalidated in case of tunnel
decapsulation or sending to 'ring' ports.
In order to fix two above issues, new function
'dp_packet_offload_invalidate' introduced. In order to clean up/unify
the code and
Basic partial HWOL API for dummy interfaces.
Signed-off-by: Ilya Maximets
---
lib/netdev-dummy.c | 106 +++--
1 file changed, 103 insertions(+), 3 deletions(-)
diff --git a/lib/netdev-dummy.c b/lib/netdev-dummy.c
index 72b4f7adc..6ee3cbca8 100644
--- a/li
Add debug logging for partial HWOL for dummy interfaces for
the future using in tests.
Signed-off-by: Ilya Maximets
---
lib/netdev-dummy.c | 66 +++---
1 file changed, 63 insertions(+), 3 deletions(-)
diff --git a/lib/netdev-dummy.c b/lib/netdev-dummy.c
i
Match packets received on dummy interfaces with offloaded flows and
set up corresponding marks in dp-packet.
Signed-off-by: Ilya Maximets
---
lib/netdev-dummy.c | 40
1 file changed, 28 insertions(+), 12 deletions(-)
diff --git a/lib/netdev-dummy.c b/lib
Simple test for basic partial HWOL functionality.
Signed-off-by: Ilya Maximets
---
tests/dpif-netdev.at | 74
1 file changed, 74 insertions(+)
diff --git a/tests/dpif-netdev.at b/tests/dpif-netdev.at
index 6915d43ba..8b30a9573 100644
--- a/tests/dpif
thanks for the fix Ilya
On Fri, Feb 15, 2019 at 2:37 AM Ilya Maximets
wrote:
> Accidentially changed while updating conntrack support.
>
> CC: Darrell Ball
> Fixes: 4ea96698f667 ("Userspace datapath: Add fragmentation handling.")
> Signed-off-by: Ilya Maximets
> ---
> Documentation/faq/relea
On Fri, Feb 15, 2019 at 08:48:40AM +, Eli Britstein wrote:
> I see tests 1219-1220 fail, but also failed witout those patches.
> The difference from V3 is rebasing, adding struct ovs_key_nd_extensions
> support.
I don't see any test failures with or without the patch. Sometimes
there can be t
From: Han Zhou
Signed-off-by: Han Zhou
---
ovsdb/ovsdb-client.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/ovsdb/ovsdb-client.c b/ovsdb/ovsdb-client.c
index 83c3c12..0215357 100644
--- a/ovsdb/ovsdb-client.c
+++ b/ovsdb/ovsdb-client.c
@@ -1267,7 +1267,7 @@ ovsdb_client_
From: Han Zhou
Ommiting the parameter names in prototypes, as suggested by coding
style: Omit parameter names from function prototypes when the names
do not give useful information.
Adjust orders of parameters as suggested by coding style.
Signed-off-by: Han Zhou
---
ovsdb/jsonrpc-server.c |
From: Han Zhou
Maintaining last N (n = 100) transactions in memory, which will be
used for future patches for generating monitor data from any point
in this N transactions.
Signed-off-by: Han Zhou
---
ovsdb/ovsdb-server.c | 11 +
ovsdb/ovsdb.c| 3 ++
ovsdb/ovsdb.h| 10 +
From: Han Zhou
Current ovsdb monitor maintains its own transaction version through an
incremental integer and use it to identify changes starting from different
version, and also use it to figure out if each set of changes should be
flushed. In particular, it uses number 0 to represent that the c
From: Han Zhou
Use monitor_cond_since to request changes after last version of local
data when connection to server is reset, without clearing the local
data. It falls back to clearing and repopulating all the data when
the requested id cannot be fulfilled by the server.
Test result at ovn-scale
From: Han Zhou
Use monitor_cond_since in C IDL. If it is not supported by server,
fall back to old method (monitor_cond, or monitor).
Signed-off-by: Han Zhou
---
lib/ovsdb-idl.c | 206 +++-
1 file changed, 158 insertions(+), 48 deletions(-)
From: Han Zhou
Support the new monitor method monitor_cond_since so that a client
can request monitoring start from a specific point instead of always
from beginning. This will reduce the cost at scenarios when server
is restarted/failed-over but client still has all existing data. In
these scena
On Fri, Feb 15, 2019 at 04:07:01PM +0300, Ilya Maximets wrote:
> Getters should have const arguments.
>
> Signed-off-by: Ilya Maximets
> ---
Acked-by: Flavio Leitner
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/list
On Fri, Feb 15, 2019 at 04:07:00PM +0300, Ilya Maximets wrote:
> It's a normal case for 'find' function, especially because this
> happens for every first packet of flow that was not offloaded yet.
> Should not warn about this. Dropped to DBG to avoid log trashing in
> case of big number of new flo
On Fri, Feb 15, 2019 at 04:07:02PM +0300, Ilya Maximets wrote:
> 1. No reason to have mbuf related APIs in a generic code.
> 2. Not only RSS/checksums should be invalidated in case of tunnel
>decapsulation or sending to 'ring' ports.
>
> In order to fix two above issues, new function
> 'dp_pac
On Fri, Feb 15, 2019 at 04:07:04PM +0300, Ilya Maximets wrote:
> Dummy interfaces clones dp-packet while 'receive' appctl processing.
> In general, we should do this anyway to avoid any possible issues in
> the future with real interfaces.
>
> Signed-off-by: Ilya Maximets
> ---
Acked-by: Flavio
Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Line is 98 characters long (recommended limit is 79)
#595 FILE: ovsdb/ovsdb-client.c:433:
Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
build:
depbase=`echo lib/ovsdb-data.lo | sed 's|[^/]*$|.deps/&|;s|\.lo$||'`;\
/bin/sh ./libtool --tag=CC --mo
On Fri, Feb 15, 2019 at 08:48:41AM +, Eli Britstein wrote:
> Generate datapath ovs key fields macros as a pre-step of retrieving
> field information, with no functional change.
>
> Signed-off-by: Eli Britstein
> Reviewed-by: Roi Dayan
Thanks for the patch.
The generated header file has a l
In scalability test with ovn-scale-test, ovsdb-server SB load is not a
problem at least with 1k HVs. However, if we restart the ovsdb-server,
depending on the number of HVs and scale of logical objects, e.g. the
number of logical ports, ovsdb-server of SB become an obvious bottleneck.
In our test
On Tue, Jan 29, 2019 at 12:01 PM Han Zhou wrote:
>
> v1 -> v2:
> - Fixed a bug in json cache handling in patch 3/7.
> - Fixed XXXs.
> - Other minor improvements.
>
> ---
> In scalability test with ovn-scale-test, ovsdb-server SB load is not a
> problem at least with 1k HVs. However, if we restart
On Fri, Feb 15, 2019 at 11:06 AM 0-day Robot wrote:
>
> Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your
> patch.
> Thanks for your contribution.
>
> I encountered some error that I wasn't expecting. See the details below.
>
>
> checkpatch:
> ERROR: Improper whitespace ar
On Fri, Feb 15, 2019 at 11:12 AM 0-day Robot wrote:
>
> Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your
> patch.
> Thanks for your contribution.
>
> I encountered some error that I wasn't expecting. See the details below.
>
>
> checkpatch:
> WARNING: Line is 98 character
On Fri, Feb 15, 2019 at 11:15 AM 0-day Robot wrote:
>
> Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your
> patch.
> Thanks for your contribution.
>
> I encountered some error that I wasn't expecting. See the details below.
>
>
> build:
> depbase=`echo lib/ovsdb-data.lo |
From: Han Zhou
Signed-off-by: Han Zhou
---
ovsdb/ovsdb-client.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/ovsdb/ovsdb-client.c b/ovsdb/ovsdb-client.c
index 83c3c12..0215357 100644
--- a/ovsdb/ovsdb-client.c
+++ b/ovsdb/ovsdb-client.c
@@ -1267,7 +1267,7 @@ ovsdb_client_
From: Han Zhou
Ommiting the parameter names in prototypes, as suggested by coding
style: Omit parameter names from function prototypes when the names
do not give useful information.
Adjust orders of parameters as suggested by coding style.
Signed-off-by: Han Zhou
---
ovsdb/jsonrpc-server.c |
From: Han Zhou
Maintaining last N (n = 100) transactions in memory, which will be
used for future patches for generating monitor data from any point
in this N transactions.
Signed-off-by: Han Zhou
---
ovsdb/ovsdb-server.c | 11 +
ovsdb/ovsdb.c| 3 ++
ovsdb/ovsdb.h| 10 +
From: Han Zhou
Current ovsdb monitor maintains its own transaction version through an
incremental integer and use it to identify changes starting from different
version, and also use it to figure out if each set of changes should be
flushed. In particular, it uses number 0 to represent that the c
From: Han Zhou
Use monitor_cond_since to request changes after last version of local
data when connection to server is reset, without clearing the local
data. It falls back to clearing and repopulating all the data when
the requested id cannot be fulfilled by the server.
Test result at ovn-scale
From: Han Zhou
Use monitor_cond_since in C IDL. If it is not supported by server,
fall back to old method (monitor_cond, or monitor).
Signed-off-by: Han Zhou
---
lib/ovsdb-idl.c | 203 ++--
1 file changed, 155 insertions(+), 48 deletions(-)
From: Han Zhou
Support the new monitor method monitor_cond_since so that a client
can request monitoring start from a specific point instead of always
from beginning. This will reduce the cost at scenarios when server
is restarted/failed-over but client still has all existing data. In
these scena
In scalability test with ovn-scale-test, ovsdb-server SB load is not a
problem at least with 1k HVs. However, if we restart the ovsdb-server,
depending on the number of HVs and scale of logical objects, e.g. the
number of logical ports, ovsdb-server of SB become an obvious bottleneck.
In our test
Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
ERROR: Improper whitespace around control block
#40 FILE: include/openvswitch/hmap.h:149:
#define HM
Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
WARNING: Line is 98 characters long (recommended limit is 79)
#595 FILE: ovsdb/ovsdb-client.c:433:
Sent from my iPhone
___
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
On Fri, Feb 15, 2019 at 04:07:03PM +0300, Ilya Maximets wrote:
> Additionally, new API call 'dp_packet_set_flow_mark' is needed
> for packet clone. Mostly for dummy HWOL implementation.
>
> Signed-off-by: Ilya Maximets
> ---
> lib/dp-packet.h | 23 +--
> 1 file changed, 21 in
Bleep bloop. Greetings Han Zhou, I am a robot and I have tried out your patch.
Thanks for your contribution.
I encountered some error that I wasn't expecting. See the details below.
checkpatch:
ERROR: Improper whitespace around control block
#40 FILE: include/openvswitch/hmap.h:149:
#define HM
From: Han Zhou
When creating IDL, "retry" was set to false. However, in daemon
mode, reconnecting upon DB server failure should be transparent
to user. This even impacts HA mode. E.g. in clustered mode, although
IDL tries to connect to next server, but at the first retry the
server fail-over may
56 matches
Mail list logo