> On Fri, Jul 11, 2014 at 04:55:55AM -0700, Jarno Rajahalme wrote:
>> This stylistic change makes the following patch a bit simpler.
>>
>> Signed-off-by: Jarno Rajahalme
>
> It's hard for me to see the difference between m_bits and n_bits in
> some of the fonts I use, and a 1-character differenc
Acked-by: Joe Stringer
On 16 July 2014 17:21, Alex Wang wrote:
> Commit 0d1cee123a84 (stp: Fix bpdu tx problem in listening state)
> makes ovs drop the stp bpdu packets if stp is not enabled on the
> input port. However, when pif bridge is used and stp is enabled
> on the integration bridge.
> On Fri, Jul 11, 2014 at 04:55:57AM -0700, Jarno Rajahalme wrote:
>> Clarify comments for trie-based subtable skipping.
>>
>> Perform the cheaper check first.
>>
>> Signed-off-by: Jarno Rajahalme
>
> Acked-by: Ben Pfaff
Acked-by: YAMAMOTO Takashi
Commit 0d1cee123a84 (stp: Fix bpdu tx problem in listening state)
makes ovs drop the stp bpdu packets if stp is not enabled on the
input port. However, when pif bridge is used and stp is enabled
on the integration bridge. The flow translation of stp bpdu
packets will go through a level of resubmi
> /* A flow classifier. */
> struct classifier {
> -struct cls_classifier *cls;
> +struct ovs_mutex mutex;
> +int n_rules OVS_GUARDED;/* Total number of rules. */
> +uint8_t n_flow_segments;
> +uint8_t flow_segments[CLS_MAX_INDICES]; /* Flow segment boundaries to use
>
On 16 July 2014 05:58, Ben Pfaff wrote:
> Commit 0791315e4d (netlink-socket: Work around kernel Netlink dump thread
> races.) introduced a simple workaround for Linux kernel races in Netlink
> dumps. However, the code remained more complicated than needed. This
> commit simplifies it.
>
> The m
On 16 July 2014 05:55, Ben Pfaff wrote:
>
> The kernel shouldn't ever return EAGAIN if there's more data to dump.
> When dumping is single-threaded (as in branch-1.11) or serialized (as
> post-commit 0791315e4d), the kernel implements that correctly. When
> dumping is multithreaded, the kernel ha
cmap_next_position() didn't update the node pointer while iterating through a
list of nodes with the same hash.
This commit fixes the bug and improve test-cmap to detect it.
Signed-off-by: Daniele Di Proietto
---
lib/cmap.c| 2 +-
tests/test-cmap.c | 31 +++
The message was not delivered due to the following reason:
Your message could not be delivered because the destination computer was
unreachable within the allowed queue period. The amount of time
a message is queued before it is returned depends on local configura-
tion parameters.
Most likely th
>
> @@ -685,12 +685,16 @@ stp_learn_in_state(enum stp_state state)
>> }
>>
>> /* Returns true if 'state' is one in which rx&tx bpdu should be done on
>> - * on a port, false otherwise. */
>> + * on a port, false otherwise.
>> + *
>> + * Returns true if 'state' is STP_DISABLED, since presumably in
Thx Ben, will add a new column FWD_BPDU as suggested by Joe,
Will add more comments to explain,
On Tue, Jul 15, 2014 at 9:35 PM, Ben Pfaff wrote:
> On Tue, Jul 15, 2014 at 07:19:53PM -0700, Alex Wang wrote:
> > Commit 0d1cee123a84 (stp: Fix bpdu tx problem in listening state)
> > makes ovs dro
On 16 July 2014 14:19, Alex Wang wrote:
> Commit 0d1cee123a84 (stp: Fix bpdu tx problem in listening state)
> makes ovs drop the stp bpdu packets if stp is not enabled on the
> input port. However, when pif bridge is used and stp is enabled
> on the integration bridge. The flow translation of s
On Tue, Jul 15, 2014 at 07:19:53PM -0700, Alex Wang wrote:
> Commit 0d1cee123a84 (stp: Fix bpdu tx problem in listening state)
> makes ovs drop the stp bpdu packets if stp is not enabled on the
> input port. However, when pif bridge is used and stp is enabled
> on the integration bridge. The flow
Converting the flow key and mask back into netlink format during flow
dump is fairly expensive. By caching the userspace-provided 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.
Previous top-5 perf
Hey Ben, Joe,
Could I get a review of either or both of you, this evening?
I'm planning to push it later tonight,
Thanks,
Alex Wang
On Tue, Jul 15, 2014 at 7:19 PM, Alex Wang wrote:
> Commit 0d1cee123a84 (stp: Fix bpdu tx problem in listening state)
> makes ovs drop the stp bpdu packets if s
Commit 0d1cee123a84 (stp: Fix bpdu tx problem in listening state)
makes ovs drop the stp bpdu packets if stp is not enabled on the
input port. However, when pif bridge is used and stp is enabled
on the integration bridge. The flow translation of stp bpdu
packets will go through a level of resubmi
On 16 July 2014 12:15, Joe Stringer wrote:
>
> + ovs_nla_copy_flow(a[OVS_FLOW_ATTR_KEY], a[OVS_FLOW_ATTR_MASK],
> + new_flow->nl_match_cache);
I realised that this needs to be wrapped in a NULL check for
"new_flow->nl_match_cache", in case the skb_alloc() fails.
___
On 15 July 2014 19:19, Ben Pfaff wrote:
> Acked-by: Ben Pfaff
>
> s/quiesce period/quiescent period/ in two places though.
>
Thanks, I made this adjustment and applied the patch to master.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.or
Signed-off-by: Joe Stringer
---
v2: First post.
Side comment: OVS_KEY_ATTR_ETHERTYPE appears twice in this list. I guess
it's not harmful to reserve a bit of extra space, but does anyone know
why it's there twice?
---
datapath/datapath.c | 37 +++--
datapath
Converting the flow key and mask back into netlink format during flow
dump is fairly expensive. By caching the userspace-provided 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.
Previous top-5 perf
Here is a link to a followup patch that exports the new sFlow structures.
https://github.com/sflow/ovs/commit/e5a10227af2f9fc0a27bd52b40b616b22f3617ec
(There is another "empty" commit showing up in between that represents
the "git pull --rebase upstream master; git push" that I ran earlier
today
The dpif provider 'operate' call duplicates all of the features available
from the 'flow_put', 'flow_del', and 'execute' calls, yielding redundant
code in providers that support both mechanisms. This change drops the
latter calls in favor of making every dpif provider support 'operate'.
The result
I sent V2 under a different subject due to the patch content change.
Here is a link to the patch:
http://patchwork.openvswitch.org/patch/4979/
On Mon, Jul 14, 2014 at 5:58 PM, Joe Stringer wrote:
> On 15 July 2014 12:53, Jarno Rajahalme wrote:
>>
>>
>> How about this:
>>
>> /* The caller must h
The added get_dp_rcu() requires its caller to hold rcu lock.
get_dp() now requires caller to hold ovs lock.
Sepearate APIs makes it clear on the intended use and the requirements
of its caller.
Signed-off-by: Andy Zhou
---
datapath/datapath.c | 23 +--
1 file changed, 17 ins
On Tue, Jul 15, 2014 at 08:24:08AM -0700, Gurucharan Shetty wrote:
> commit 8661af798(timeval: Provide a variation for time/warp command)
> provided a variation of time/warp command to prevent multiple
> invocations of time/warp. That commit changed only the users in bfd.at
> and cfm.at as they use
On Mon, Jul 14, 2014 at 12:35:43PM -0700, Gurucharan Shetty wrote:
> In the 'interfaces' file, if an admin adds the openvswitch interface
> in 'auto', ifupdown will try to create OVS interfaces even before
> openvswitch has started. In a case like that, assume that the admin
> knows what he is doin
On Mon, Jul 14, 2014 at 12:35:42PM -0700, Gurucharan Shetty wrote:
> This is a port of commit d7aab661 ( rhel: Add Patch Port support to
> initscripts.) from rhel to debian's ifupdown script.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
__
On Fri, Jul 11, 2014 at 04:55:57AM -0700, Jarno Rajahalme wrote:
> Clarify comments for trie-based subtable skipping.
>
> Perform the cheaper check first.
>
> Signed-off-by: Jarno Rajahalme
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
On Fri, Jul 11, 2014 at 04:55:56AM -0700, Jarno Rajahalme wrote:
> Previously we only returned the last matching prefix length
> encountered during a trie lookup, and skipped subtables that had
> prefixes longer than that. This patch changes the trie lookup
> functions to return all matching prefi
Good point on comments. I now agree it is necessary and should stay
with the function.
Comments aside, I am not happy with the nested rcu_read_lock(). It
seems unnecessary.
I will come up with a v2.
On Mon, Jul 14, 2014 at 5:58 PM, Joe Stringer wrote:
> On 15 July 2014 12:53, Jarno Rajahalme w
On Fri, Jul 11, 2014 at 04:55:54AM -0700, Jarno Rajahalme wrote:
> Now that it is clear that struct cls_classifier itself does not
> need RCU indirection and pvector is defined in its own header, it
> is possible get rid of the indirection from struct classifier to
> struct cls_classifier.
>
> Sug
On Fri, Jul 11, 2014 at 04:55:55AM -0700, Jarno Rajahalme wrote:
> This stylistic change makes the following patch a bit simpler.
>
> Signed-off-by: Jarno Rajahalme
It's hard for me to see the difference between m_bits and n_bits in
some of the fonts I use, and a 1-character difference isn't ide
On Fri, Jul 11, 2014 at 01:24:01PM +0200, Eric Sesterhenn wrote:
> on my debian installation, the ovsdb-server is running as root. Since I
> prefer to add additional mitigations for running services, I was looking
> into putting the ovsdb-server into a chroot and implemented it in the
> following t
On Jul 15, 2014, at 12:42 PM, Ben Pfaff wrote:
> On Tue, Jul 15, 2014 at 11:32:36AM -0700, Daniele Di Proietto wrote:
>> This removes a GCC 4.9 warning (unused-but-set-variable)
>>
>> Signed-off-by: Daniele Di Proietto
>
> Applied to master, thanks.
>
> If this warning shows up on other branc
On Tue, Jul 15, 2014 at 11:32:36AM -0700, Daniele Di Proietto wrote:
> This removes a GCC 4.9 warning (unused-but-set-variable)
>
> Signed-off-by: Daniele Di Proietto
Applied to master, thanks.
If this warning shows up on other branches then I'm willing to apply
it to them too.
If the userspace wants to match on a flow with some tunnel attributesset to 0,
it simply omits them in the netlink attributes stream.
Since our wildcarding logic (when megaflows are disabled) is based on the
attributes in the netlink stream, we set our mask incorrectly.
This commit adds a check to
This removes a GCC 4.9 warning (unused-but-set-variable)
Signed-off-by: Daniele Di Proietto
---
lib/coverage.c | 4
1 file changed, 4 deletions(-)
diff --git a/lib/coverage.c b/lib/coverage.c
index 5a20d6e..fe97d69 100644
--- a/lib/coverage.c
+++ b/lib/coverage.c
@@ -279,8 +279,6 @@ covera
On Jul 11, 2014, at 4:11 PM, Andy Zhou wrote:
> On Fri, Jul 11, 2014 at 3:05 PM, Daniele Di Proietto
> wrote:
>> If the userspace wants to match on a flow with tun_key.{tun_id, tun_flags,
>> ipv4_tos} set to 0 it simply omits them in the netlink attributes stream.
>> Since our wildcarding logic
On Tue, Jul 15, 2014 at 1:23 AM, Wenyu Zhang wrote:
> Hi Pravin, Thanks you for reviewing it.
>
> About the output tunnel info sent to userland, all the tunnel parameters in
> ovs_key_ipv4_tunnel
> have been sent to userland, except two parameters in geneve_opt, which is
> only for GENEVE.
>
Commit 0791315e4d (netlink-socket: Work around kernel Netlink dump thread
races.) introduced a simple workaround for Linux kernel races in Netlink
dumps. However, the code remained more complicated than needed. This
commit simplifies it.
The main reason for complication in the code was 'status_s
On Tue, Jul 15, 2014 at 04:52:21PM +1200, Joe Stringer wrote:
> On 15 July 2014 09:10, Ben Pfaff wrote:
>
> > Commit 0791315e4d (netlink-socket: Work around kernel Netlink dump thread
> > races.) introduced a simple workaround for Linux kernel races in Netlink
> > dumps. However, the code remain
On Tue, Jul 15, 2014 at 04:18:19PM +1200, Joe Stringer wrote:
> On 15 July 2014 09:10, Ben Pfaff wrote:
>
> > If nl_pool_alloc() failed, then 'dump' was not initialized at all and
> > further use of the dump would access uninitialized data, probably causing
> > a crash.
> >
> > Found by inspectio
On Tue, Jul 15, 2014 at 04:13:05PM +1200, Joe Stringer wrote:
> On 15 July 2014 09:10, Ben Pfaff wrote:
>
> > When dpif_linux_refresh_channels() refreshes the set of channels when
> > the number of handlers changes, it destroys all the dpif's channels and
> > sets dpif->uc_array_size to 0. If th
On Tue, Jul 15, 2014 at 04:14:10PM +1200, Joe Stringer wrote:
> On 15 July 2014 09:10, Ben Pfaff wrote:
>
> > Signed-off-by: Ben Pfaff
> >
>
> Acked-by: Joe Stringer
Thanks. I pushed this to master.
___
dev mailing list
dev@openvswitch.org
http://o
On Mon, Jul 14, 2014 at 10:37:38AM -0700, Ben Pfaff wrote:
> Reported-by: Simon Jouet
> Signed-off-by: Ben Pfaff
I applied this to master because Simon confirmed off-list that it
fixes the problem.
___
dev mailing list
dev@openvswitch.org
http://openvs
On Fri, Jul 11, 2014 at 01:08:48AM +, Alin Serdean wrote:
> Add the structure sockaddr_nl to netlink-protocol.h for MSVC. Also remove
> some warnings in netlink-socket.c when using MSVC.
>
> Signed-off-by: Alin Gabriel Serdean
I think that nl_sock always uses 'fd' as a HANDLE on Windows and
On Fri, Jul 11, 2014 at 01:07:29AM +, Alin Serdean wrote:
> Add the newly created headers in include/windows/* to
> include/windows/automake.mk.
>
> Signed-off-by: Alin Gabriel Serdean
This should be squashed with the patch that adds the headers,
otherwise it breaks the build if you apply th
On Mon, Jul 14, 2014 at 11:32:52AM -0700, Kirkland Spector wrote:
> Signed-off-by: Kirkland Spector
> Acked-by: Andrey Falko
Thanks! I applied this to master.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Thanks a lot! I must have good karma lately, because all my patches
are getting reviewed quickly.
I pushed all of these to master, and backported the first patch (the
crasher bug) to all relevant branches (branch-2.[321]).
On Mon, Jul 14, 2014 at 04:47:29PM -0700, Jarno Rajahalme wrote:
> For al
commit 8661af798(timeval: Provide a variation for time/warp command)
provided a variation of time/warp command to prevent multiple
invocations of time/warp. That commit changed only the users in bfd.at
and cfm.at as they used it the most. Since we haven't had any negative
confequences because of th
On Tue, Jul 15, 2014 at 04:04:52PM +1200, Joe Stringer wrote:
> The userspace datapath returns RCU-protected actions from flow_get() and
> flow_dump_next(). This doesn't cause any trouble for current users of
> these functions, but it imposes additional constraints on their use.
> This patch makes
51 matches
Mail list logo
