Thanks, I added that change and pushed to branch-2.3.
A version of this series for master is available here:
http://mail.openvswitch.org/pipermail/dev/2014-July/042507.html
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/
The previous 'grep' logic would occasionally catch unintended flows.
There's no reason to check for these flows separately, so combine the
two checks.
Signed-off-by: Joe Stringer
---
tests/ofproto-dpif.at |4 +---
1 file changed, 1 insertion(+), 3 deletions(-)
diff --git
Thanks, I applied this to master and branch-2.3.
On 8 July 2014 13:17, Ansis Atteka wrote:
> On Mon, Jul 7, 2014 at 6:01 PM, Joe Stringer
> wrote:
> > The previous 'grep' logic would occasionally catch unintended flows.
> > There's no reason to check for these
On 8 July 2014 13:20, Alex Wang wrote:
>
> +/* Needs to check in case the socket pointer is changed in between
> + * of holding the locks. */
Might be worth mentioning here that a known case is when the port is
deleted while handling an upcall from that port.
Acke
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
-vswitchd [kernel.kallsyms] [k] mutex_unlock
1.52% ovs-vswitchd ovs-vswitchd[.] revalidator_sweep__.constprop.16
Suggested-by: Jesse Gross
Signed-off-by: Joe Stringer
---
I'm operating under the assumption that we can't just copy the version
of the netlink key/mask dir
Thanks for the review,
On 10 July 2014 21:13, Thomas Graf wrote:
> If I understand the code correctly the gain is only visible on
> consecutive dumps of the same flow. How about constructing the cache
> when you require it for the first time? That avoids the cost on flow
> setup.
Correct. My m
On 11 July 2014 04:53, Flavio Leitner wrote:
> Can we push the cache construction to later that it doesn't impact
> either flow setup or flow dump? I.e. like a workqueue? I don't know
> if we have such facility.
I'm open to the idea, but completely unfamiliar with how these work. I can
take a l
On 11 July 2014 11:29, Daniele Di Proietto wrote:
> -static void sw_flow_mask_set(struct sw_flow_mask *mask,
> -struct sw_flow_key_range *range, u8 val)
> +/* We expect the nlattr stream to be already validated */
> +static int nlattr_set(struct nlattr *attr, u8 val,
On 11 July 2014 21:03, Thomas Graf wrote:
> > Correct. My main concern with doing it the first time it is required, is
> > how it's synchronised. Flow dump is RCU. I don't really know what the
> > threadsafety requirements are for this code, or what options are
> available
> > to address this. Is
On 12 July 2014 00:55, Thomas Graf wrote:
> > Another sidenote:
>
> > Looking at the code, OVS does not handle NLM_F_DUMP_INTR in user space
> yet
> > > and the kernel dump does not call genl_dump_check_consistent() yet to
> > > provide
> > > the flag. So what can currently happen even without y
On 12 July 2014 01:03, Thomas Graf wrote:
> On 07/11/14 at 09:28am, Flavio Leitner wrote:
> > On Fri, Jul 11, 2014 at 11:27:09AM +1200, Joe Stringer wrote:
> > > On 11 July 2014 04:53, Flavio Leitner wrote:
> > >
> > > > Can we push the cache const
On 10 July 2014 20:41, Thomas Graf wrote:
> On 07/10/14 at 05:29pm, Joe Stringer wrote:
> > 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 S
On 14 July 2014 13:05, Pravin Shelar wrote:
> > @@ -734,9 +722,25 @@ static int ovs_flow_cmd_fill_info(struct datapath
> *dp,
> > err = ovs_nla_put_flow(dp, &flow->key, &flow->mask->key, skb);
> > if (err)
> > goto error;
> > -
> > nla_nest_end(skb, nla);
>
Isn't this here to cover the case where the datapath disappears when it is
returned to the caller?
(In which case, perhaps we could explain this in the comment)
On 15 July 2014 09:55, Andy Zhou wrote:
> Signed-off-by: Andy Zhou
> ---
> datapath/datapath.c | 1 -
> 1 file changed, 1 deletion(
On 15 July 2014 12:53, Jarno Rajahalme wrote:
>
> How about this:
>
> /* The caller must hold either rcu_read_lock or ovs_mutex to keep the
> returned dp pointer valid. */
>
This looks good. And I think it's better to have it at get_dp() rather than
the callers, to help prevent future misuse of
On 15 July 2014 12:11, Ben Pfaff wrote:
> dpif_netdev_flow_get() calls dp_netdev_flow_get_actions() twice. I
> did not check whether the actions could change in between. Please
> change the code to just retrieve the actions once.
>
> Acked-by: Ben Pfaff
>
> As an additional optimization I thin
On 15 July 2014 15:26, Ben Pfaff wrote:
> > I notice that dpif_netdev_flow_dump_next() also does this optimization
> for
> > actions, but there's no mention of RCU in the dpif_flow_dump_next() API.
> Do
> > you think we should add a comment like this to dpif_flow_get() and
> > dpif_flow_dump_next
On 15 July 2014 12:13, Ben Pfaff wrote:
> On Tue, Jul 08, 2014 at 11:04:05AM +1200, Joe Stringer wrote:
> > If the datapath doesn't dump a flow for some reason, and the current
> > dump is expected to revalidate all flows in the datapath, then perform
> > revalidation f
these functions can be used.
Signed-off-by: Joe Stringer
---
lib/dpif-netdev.c |1 -
lib/dpif-provider.h |6 --
lib/dpif.c | 11 ---
3 files changed, 12 insertions(+), 6 deletions(-)
diff --git a/lib/dpif-netdev.c b/lib/dpif-netdev.c
index 21a4f56..9270873 100644
On 15 July 2014 15:40, Joe Stringer wrote:
> OK. I'll send a separate patch to add these comments.
>
The patch is here:
http://openvswitch.org/pipermail/dev/2014-July/042695.html
___
dev mailing list
dev@openvswitch.org
http://openvswitch.
to
> try to access element 0 of the set of channels, dereferencing a null
> pointer.
>
> This fixes the problem.
>
> I encountered this while looking at a bug that I had introduced during
> development that caused the port dump to always be empty. It would be
> difficult to encou
On 15 July 2014 09:10, Ben Pfaff wrote:
> Signed-off-by: Ben Pfaff
>
Acked-by: Joe Stringer
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
#x27;ve glanced sideways at this code a couple of times, but never seen the
failures. Looks much safer with this patch.
Acked-by: Joe Stringer
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
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 remained more complicated than needed. This
> commit simplifies it.
>
> The m
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
% ovs-vswitchd [kernel.kallsyms] [k] _raw_spin_lock_bh
+ 3.43% ovs-vswitchd [kernel.kallsyms] [k] ovs_flow_tbl_dump_next
Suggested-by: Jesse Gross
Signed-off-by: Joe Stringer
---
v2: Copy mask from userspace rather than constructing from sw_flow.
RFC: First post.
This version gets rid
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
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 c
% ovs-vswitchd [kernel.kallsyms] [k] _raw_spin_lock_bh
+ 3.43% ovs-vswitchd [kernel.kallsyms] [k] ovs_flow_tbl_dump_next
Suggested-by: Jesse Gross
Signed-off-by: Joe Stringer
---
v3: Add error checking to ovs_nla_copy_flow() and its users.
Allow nl_match_cache to fail allocation in
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 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
ntioning in the code
(might not be, given the kernel behaviour is decoupled from this code).
Acked-by: Joe Stringer
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
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 integr
U
+ * TX_BPDU/RX_BPDU: the port could generate/consume bpdus.
> + *
> + * FWD_BPDU: the port should should always forward the bpdus,
> + * whether they are generated by the port or received
> + * as incoming packets.
>
On 17 July 2014 04:48, Ben Pfaff wrote:
> On Wed, Jul 16, 2014 at 02:50:12PM +1200, Joe Stringer wrote:
> > 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 fl
On 12 July 2014 00:55, Thomas Graf wrote:
> On 07/11/14 at 11:29pm, Joe Stringer wrote:
> > I'm skeptical of taking the ovs_lock(). The current patch performs this
> > key/mask cache construction inside ovs_lock() as part of the critical
> > section for flow install.
On 18 July 2014 05:19, Ben Pfaff wrote:
>
> > From datapath/README:
> >
> > - If the kernel's flow key includes more fields than the userspace
> > version of the flow key, for example if the kernel decoded IPv6
> > headers but userspace stopped at the Ethernet type (because it
> >
On 18 July 2014 10:53, Joe Stringer wrote:
> Basically I'd be much more comfortable with the kernel generating its
>> own flow key the first time it dumps a flow, then caching that. How
>> much overhead does that have versus caching the userspace-provided
>> flow key?
On 19 July 2014 06:13, Jesse Gross wrote:
> It's true, the field ordering is not significant if you understand all
> of the fields. The concern is that userspace may want to keep track of
> flows even if it doesn't understand all of the fields (for example, it
> can determine that the kernel prov
-vswitchd [kernel.kallsyms] [k] memset
3.41% ovs-vswitchd [kernel.kallsyms] [k] _raw_spin_lock_bh
...
Signed-off-by: Joe Stringer
---
v4: Construct the cache during the first flow dump.
v3: Add error checking to ovs_nla_copy_flow() and its users.
Allow nl_match_cache to fail allocation
On 21 July 2014 10:28, Joe Stringer wrote:
> On 19 July 2014 06:13, Jesse Gross wrote:
>
>> It's true, the field ordering is not significant if you understand all
>> of the fields. The concern is that userspace may want to keep track of
>> flows even if it doesn&#
On 16 July 2014 12:15, Joe Stringer wrote:
> Signed-off-by: Joe Stringer
>
This patch was previously required for the series "datapath: Cache netlink
flow key, mask on flow_put.", but is no longer required for newer versions
of that, as I've changed the approach.
On 18 September 2014 11:13, Ben Pfaff wrote:
> On Mon, Sep 15, 2014 at 02:25:14PM +1200, Joe Stringer wrote:
> > One of the limiting factors on the number of flows that can be supported
> > in the datapath is the overhead of assembling flow dump messages in the
> > datapa
Thanks for the review,
On 18 September 2014 10:33, Ben Pfaff wrote:
> On Mon, Sep 15, 2014 at 02:25:11PM +1200, Joe Stringer wrote:
> > Currently, when a revalidator thread first dumps a flow, it creates a
> > 'udpif_key' object and caches a copy of a kernel flow k
On 18 September 2014 10:53, Ben Pfaff wrote:
> On Mon, Sep 15, 2014 at 02:25:12PM +1200, Joe Stringer wrote:
> > This allows us to ignore most fields of a flow_dump, requiring only the
> > flow key for looking up the ukey. Fetching flows can also be avoided in
> > the corner
On 18 September 2014 11:03, Ben Pfaff wrote:
> On Mon, Sep 15, 2014 at 02:25:13PM +1200, Joe Stringer wrote:
> > Add the 128-bit murmurhash by Austin Appleby, for 32-bit systems from:
> > http://code.google.com/p/smhasher/source/browse/trunk/MurmurHash3.cpp
> >
> >
Thanks for the review, comments inline
On 17 September 2014 17:37, Pravin Shelar wrote:
> On Wed, Sep 10, 2014 at 3:16 AM, Joe Stringer
> wrote:
> > Reduce duplicate code by using nla_policy and nla_parse_strict().
> >
> Thanks for simplifying this code. I have couple of c
On 18 September 2014 13:02, Joe Stringer wrote:
>
>> > @@ -229,17 +244,19 @@ static bool match_validate(const struct
>> sw_flow_match *match,
>> > }
>> > }
>> >
>> > - if ((key_attrs & key_expected) != key_e
aces
- Various fixes
RFCv3:
- Add datapath implementation
- Minor fixes
- Rebased
RFCv2:
- Revised early patches from v1 feedback
- Add Acks from Ben
- Rebased
RFCv1:
http://openvswitch.org/pipermail/dev/2014-August/044383.html
Joe Stringer (14):
revalidator: Distinguish new and duplicate flows
We previously counted flows that have been installed during the current
dump as duplicates, rather than recognising them as new flows. This
patch separates the counters out for these two cases.
Signed-off-by: Joe Stringer
---
v5-v6: No change.
v4: Initial post.
---
ofproto/ofproto-dpif-upcall.c
these ukey
collections to be read or modified while a revalidator is garbage
collecting it. To protect the ukeys, this patch adds locking on the ukey
collection.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v4-v6: No change.
v3: Rebase.
v2: No change.
RFC: First post.
---
ofproto/ofproto
Future patches will make use of the 'struct dump_op' in a broader sense,
so this patch renames it to make things a bit clearer.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v4-v6: No change.
v3: Rebase.
v2: No change.
RFC: First post.
---
ofproto/ofproto-dpif-upcal
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v4-v6: No change.
v3: Rebase.
v2: Call ovsrcu_quiesce() unconditionally.
RFC: Initial Post.
---
ofproto/ofproto-dpif-upcall.c | 61 ++---
1 file changed, 33 insertions(+), 28 deletions(-)
diff --git a
istics for flows inserted using
ovs-dpctl.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v6: Always call poll_block() in udpif_upcall_handler().
Reduce memory zeroing in upcall/ukey init code.
Fix up extraneous netflow_unref().
Don't fetch 'recirc' in paths that alw
maps from the revalidators, and increases
the number of maps used to store ukeys, to reduce contention.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v6: No change.
v4: Increase N_UMAPS to 512.
v3: Rebase.
v2: #define N_UMAPS.
Simplify upcall_unixctl_show() element counting.
Mention
This just makes it a bit easier for someone coming in fresh to do some
searches and figure out what the description means.
Signed-off-by: Joe Stringer
---
v6: First post.
---
tests/test-hash.c |4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/tests/test-hash.c b/tests
skip dumping these fields, so this cache will be used instead.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v6: Use atomic_read_relaxed() in ukey_create().
Simplify delete_op_init().
v5: Switch back to checking that the mask is more specific.
v4: No change.
v3: Rebase.
v2: Rebase.
RFC
Add the 128-bit murmurhash by Austin Appleby, r150 from:
http://code.google.com/p/smhasher/source/browse/trunk/MurmurHash3.cpp
Signed-off-by: Joe Stringer
---
The x64 version is approximately 2x faster than x86 in the tests I ran.
v6: Add version that uses 64-bit operations.
Add tests for
er patches will add support for datapaths to store and interpret this
UID, in which case the dpif has a responsibility to pass it through
transparently.
Signed-off-by: Joe Stringer
---
v6: Split out from "dpif: Add Unique flow identifiers."
Store flow hash seed staticly for all d
operation. For instance, OVS_UID_F_SKIP_KEY
allows the datapath to skip returning the key (eg, during dump to reduce
memory copy).
Signed-off-by: Joe Stringer
---
v6: Fix documentation for supporting UIDs between 32-128 bits.
Minor style fixes.
Rebase.
v5: No change.
v4: Fix memory leak
for newer datapaths that support
UID, while the latter is used for backwards compatibility.
Signed-off-by: Joe Stringer
---
v6: Split out from "dpif: Add Unique flow identifiers."
Make sure that actions are fetched if terse=false.
Add additional dpif.h documentation to de
.
Signed-off-by: Joe Stringer
---
v6: Split out from "dpif: Add Unique flow identifiers."
Rebase.
v5: Always pass flow_key down to dpif, to improve error logging.
Fix extraneous netflow_unref.
Rebase.
v4: Fix bug where UID-based terse dump wasn't enabled by defaul
and
simplifies mask comparison in revalidate_ukey().
Signed-off-by: Joe Stringer
---
For reference, prior to this patchset the size of a udpif_key is 656, but the
actions are not cached on master. This patch allows the entire patchset to have
a negligible impact on the memory footprint while
ps to reproduce: Unload/Reload kernel module && restart ovs-vswitchd.
Signed-off-by: Joe Stringer
---
I considered implementing this fix using ovsdb_idl_get_seqno() for the
entire logic block, but concluded that it doesn't make sense to update
the status whenever there is any kind of
This function allows netlink attributes to be parsed more strictly,
to check for additional constraints beyond those that nla_parse() checks
for. Passing flags=0 to nla_parse_strict() implies the same behaviour as
nla_parse().
Signed-off-by: Joe Stringer
---
v3: Shift nla_parse_strict() from
Reduce duplicate code by using nla_policy and nla_parse_strict().
Signed-off-by: Joe Stringer
---
v3: Fix match_validate()
Make ovs_key_from_nlattrs(),metadata_from_nlattrs() not change the
attrs they are given
Remove extraneous check for duplicate attributes
Use NLA_FLAG
September 2014 05:45, Alex Wang wrote:
> Thx a lot for the fix, that was a bad move~
>
> Acked-by: Alex Wang
>
> On Mon, Sep 29, 2014 at 3:09 AM, Joe Stringer
> wrote:
>
>> Commit 9c537baf613a16e (bridge: Refactor the stats and status update.)
>> inadvertently cha
un with the silent flags?
Acked-by: Joe Stringer
On 27 September 2014 06:02, Ben Pfaff wrote:
> On Wed, Sep 03, 2014 at 10:28:39AM -0700, Ben Pfaff wrote:
> > The Open vSwitch "make" output was still pretty verbose even when
> > configured with --enable-silent-rules. Th
veloper builds, where if something fails in a way I don't understand
> I can just rerun with V=1.
>
> On Tue, Sep 30, 2014 at 09:30:23AM +1300, Joe Stringer wrote:
> > From the linked travis build, it seems like there's still a bit of
> > verbosity on the testsuite bu
On 30 September 2014 09:30, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:13PM +1200, Joe Stringer wrote:
> > Add the 128-bit murmurhash by Austin Appleby, r150 from:
> > http://code.google.com/p/smhasher/source/browse/trunk/MurmurHash3.cpp
> >
> > Signed-off-by: Jo
On 30 September 2014 09:51, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:14PM +1200, Joe Stringer wrote:
> > This patch shifts the responsibility for determining the hash for a flow
> > from the revalidation logic down to the dpif layer. This assists in
> > handling b
On 30 September 2014 10:10, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:15PM +1200, Joe Stringer wrote:
> > If a datapath is created with the flag OVS_DP_F_INDEX_BY_UID, then an
> > additional table_instance is added to the flow_table, which is indexed
> > by uni
On 30 September 2014 10:15, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:16PM +1200, Joe Stringer wrote:
> > This patch modifies the dpif interface to allow flows to be manipulated
> > using a 128-bit identifier. This allows revalidator threads to perform
> > datapath
On 30 September 2014 10:24, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:17PM +1200, Joe Stringer wrote:
> > One of the limiting factors on the number of flows that can be supported
> > in the datapath is the overhead of assembling flow dump messages in the
> > datapath
On 30 September 2014 10:27, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:18PM +1200, Joe Stringer wrote:
> > Previously we stored the netlink-formatted version of a flow key and
> > mask in 'struct udpif_key'. This patch stores the original key,mask
> > in t
Thanks, applied to master.
On 30 September 2014 08:54, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:05PM +1200, Joe Stringer wrote:
> > We previously counted flows that have been installed during the current
> > dump as duplicates, rather than recognising them as new flows.
Thanks, applied to master.
On 30 September 2014 08:57, Ben Pfaff wrote:
> On Fri, Sep 26, 2014 at 09:28:12PM +1200, Joe Stringer wrote:
> > This just makes it a bit easier for someone coming in fresh to do some
> > searches and figure out what the description means.
> >
&
On 1 October 2014 06:56, Pravin Shelar wrote:
> On Mon, Sep 29, 2014 at 3:59 PM, Joe Stringer
> wrote:
> >
> >
> > On 30 September 2014 10:10, Ben Pfaff wrote:
> >>
> >> On Fri, Sep 26, 2014 at 09:28:15PM +1200, Joe Stringer wrote:
>
witchd --log-file=fakelog &
+pid=`echo $!`
+
+# sleep for a while
+sleep 5
5 seems high, is this just to make sure that vswitchd gets a chance to
fully initialize and go through the wait block once?
Acked-by: Joe Stringer
___
dev mailing list
dev@openv
On 1 October 2014 11:55, Pravin Shelar wrote:
> On Tue, Sep 30, 2014 at 3:15 PM, Joe Stringer
> wrote:
> >
> >
> > On 1 October 2014 06:56, Pravin Shelar wrote:
> >>
> >> On Mon, Sep 29, 2014 at 3:59 PM, Joe Stringer
> >> wrote:
> &g
gt;
>
> Yeah, also to maybe cumulate more logs, if there is something.
>
> How about changing it to 2 s?
>
>
>
>> Acked-by: Joe Stringer
>>
>
>
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
Maybe just stick with 5.
On 1 October 2014 13:10, Alex Wang wrote:
> Yeah, so should we use 5 seconds? or use ovs_wait() to wait on the number
> of lines which could wait up to 10 seconds...?
>
> On Tue, Sep 30, 2014 at 4:14 PM, Joe Stringer
> wrote:
>
>> I had a pl
On 1 October 2014 12:15, Pritesh Kothari wrote:
> diff --git a/README.md b/README.md
> new file mode 100644
> index 000..318f0e6
> --- /dev/null
> +++ b/README.md
> @@ -0,0 +1,114 @@
> +
> +http://openvswitch.org";>Open vSwitch
> +
> +
Markdown has syntax for headings and urls, could you us
On 1 October 2014 11:59, Joe Stringer wrote:
>
>
> On 1 October 2014 11:55, Pravin Shelar wrote:
>
>> On Tue, Sep 30, 2014 at 3:15 PM, Joe Stringer
>> wrote:
>> >
>> >
>> > On 1 October 2014 06:56, Pravin Shelar wrote:
>> >>
&
Thanks for fixing this.
Acked-by: Joe Stringer
On 1 October 2014 18:10, YAMAMOTO Takashi wrote:
> NetBSD implementation of wc command outputs extra whitespaces
> like the following. Tweak the test to success on such environments.
>
> % echo hoge|wc -l|hexdump -C
> 000
On 30 September 2014 10:24, Ben Pfaff wrote:
> I suspect that check_recirc() and check_uid() should pass
> DPIF_FP_MODIFY as well as DPIF_FP_CREATE, just in case a previous run
> of OVS was killed at just the right time to leave a straggler probe
> flow in the datapath.
>
I checked on this, and
Hi Gur, looks straightforward. I have a few broad feedback points though:
If you haven't read it yet, please take a look at the CONTRIBUTING file in
the top of the source tree.
I'm not sure what format this patch is in, but git won't accept it in my
setup. I suspect it's due to the footer message
Sounds good to me. Nice to hear confirmation from the overnight test. Might
be worthwhile adding a short note above the line to explain that reducing
the update interval prevents particular race conditions.
Did you consider applying the same change to the CFM flap_count test?
Acked-by: Joe
Acked-by: Joe Stringer
On 2 October 2014 07:44, Alex Wang wrote:
> Test the switch over to inactive ovs-vswitchd process when user
> kill the currently active ovs-vswitchd.
>
> Signed-off-by: Alex Wang
> ---
> tests/ovs-vswitchd.at | 41 +
Hmm. This time the patch is littered with extra spaces all through the
email.
If you're able to use them, "git format-patch" and "git send-email" is a
good way to send your patches out reliably without mail clients interfering
with the presentation of the patch.
I took a look at the group deletio
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v4-v7: No change.
v3: Rebase.
v2: Call ovsrcu_quiesce() unconditionally.
RFC: Initial Post.
---
ofproto/ofproto-dpif-upcall.c | 61 ++---
1 file changed, 33 insertions(+), 28 deletions(-)
diff --git a
s fixes
RFCv3:
- Add datapath implementation
- Minor fixes
- Rebased
RFCv2:
- Revised early patches from v1 feedback
- Add Acks from Ben
- Rebased
RFCv1:
http://openvswitch.org/pipermail/dev/2014-August/044383.html
Joe Stringer (11):
revalidator: Use 'cmap' for storing ukeys.
revalida
these ukey
collections to be read or modified while a revalidator is garbage
collecting it. To protect the ukeys, this patch adds locking on the ukey
collection.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v4-v7: No change.
v3: Rebase.
v2: No change.
RFC: First post.
---
ofproto/ofproto
maps from the revalidators, and increases
the number of maps used to store ukeys, to reduce contention.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v6-v7: No change.
v4: Increase N_UMAPS to 512.
v3: Rebase.
v2: #define N_UMAPS.
Simplify upcall_unixctl_show() element counting
Future patches will make use of the 'struct dump_op' in a broader sense,
so this patch renames it to make things a bit clearer.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v7: Rebase.
v4-v6: No change.
v3: Rebase.
v2: No change.
RFC: First post.
---
ofproto/ofproto-dpif-upcal
skip dumping these fields, so this cache will be used instead.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v7: Rebase against ukey_create() changes in previous patch.
v6: Use atomic_read_relaxed() in ukey_create().
Simplify delete_op_init().
v5: Switch back to checking that the mask
happen in corner cases such as when
ovs-vswitchd is restarted (and flows remain in the datapath) or a user
installs a flow using ovs-dpctl.
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v7: Create ukeys in revalidator threads in corner cases.
v6: Always call poll_block() in udpif_upca
Add the 128-bit murmurhash by Austin Appleby, r150 from:
http://code.google.com/p/smhasher/source/browse/trunk/MurmurHash3.cpp
Signed-off-by: Joe Stringer
Acked-by: Ben Pfaff
---
v7: Replace getblock*() with get_unaligned_u*().
Add ovs_u128_equal() function rather than using memcmp
for newer datapaths that support
UID, while the latter is used for backwards compatibility.
Signed-off-by: Joe Stringer
---
v7: Shift UID terseness flag below the dpif layer
Remove OVS_DP_F_INDEX_BY_UID
Use ovs_u128_equal().
Don't use 'static bool' for dpif_n
901 - 1000 of 2028 matches
Mail list logo
