«/~t¼>èh¡ÝVxûé×ÓÙÈ/tóÙ
:Ùj·BÏÀ¸Y]pp²ô9xCÂf
µ¦Ôeqùú¿áñ[I5ó¥Í¿pZ[ æÈÇØ
t÷?xwKñëNË7ù°·ëÕ7òÂë,94'|3SowLlÊ
]eáz^äý¹¼·ÓßE¬IDíÉOW
±C3(ÏS4͹ϯáïÁ§ ,ZzU:D}ý:Ø¿\îë;j¥´!ËÜÉ
Ç3(wÃvV¾ÞTuf_§°íåb6Ynº¼yËÆÌ;ß
çú[»FfÝ2R¥bÂ5áe*~Òmcú
¹NN§DõEÒï¾7-7 ÙÂ)2ÇáUþ«ñ`ÐÎaÖ|&µqù
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
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 inspection.
>
> Signed-off-by: Ben Pfaff
>
I've glanced sideways at this code a cou
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
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 the port dump later in the function
> turns up no ports (which generally i
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.org/mailman/l
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 the dpif documentation more explicit about how the
results of these fu
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 for those flows by fetching
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 Tue, Jul 15, 2014 at 03:23:08PM +1200, Joe Stringer wrote:
> 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
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 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
How about this:
/* The caller must hold either rcu_read_lock or ovs_mutex to keep the returned
dp pointer valid. */
Jarno
> On Jul 15, 2014, at 3:45 AM, Andy Zhou wrote:
>
> Since rcu_read_lock() is taken within the function. Caller is not
> strictly required to take either rcu_read_lock()
Since rcu_read_lock() is taken within the function. Caller is not
strictly required to take either rcu_read_lock() or a mutex lock.
What you said is true in how we use this function. May be it should be
comments in the calling sits?
On Mon, Jul 14, 2014 at 3:55 PM, Joe Stringer wrote:
> Isn't t
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 for those flows by fetching them during the sweep phase.
> If revalidation is
On Tue, Jul 08, 2014 at 11:04:04AM +1200, Joe Stringer wrote:
> Change the interface to allow implementations to pass back a buffer, and
> allow callers to specify which of actions, mask, and stats they wish to
> receive. This will be used in the next commit.
>
> Signed-off-by: Joe Stringer
> ---
For all four:
Acked-by: Jarno Rajahalme
On Jul 14, 2014, at 3:00 PM, Ben Pfaff wrote:
> Ben Pfaff (4):
> ofp-util: Fix null pointer dereference in ofputil_pull_buckets().
> ofproto: Avoid theoretical double free of large rule collections.
> coverage: Move m_idx, h_idx to an inner scope in c
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(
Np, thx for the review,
On Fri, Jun 27, 2014 at 12:34 AM, Thomas Graf
wrote:
> On 06/26/14 at 03:19pm, Ben Pfaff wrote:
> > I would use readdir() instead of readdir_r(). See
> > http://elliotth.blogspot.com/2012/10/how-not-to-use-readdirr3.html for
> > more information.
>
> I wasn't aware of t
Thx Ben for the review,
I think that this should be dummied not just on Windows but on any
> non-Linux kernel, because I doubt that *BSD has the same NUMA interface
> as Linux.
>
>
Yeah, I'll do that.
> Usually, I prefer to write assertions based on predicate functions,
> rather than write sp
On Tue, Jul 08, 2014 at 05:35:57PM +0100, Zoltan Kiss wrote:
> This patch modifies the LACP selection logic by prefering a slaves with up and
> running partners when looking for a lead.
> That fixes the following scenario:
> - bond has 2 ports, A and B, their other ends are in separate chassis with
Found by clang-analyzer.
Signed-off-by: Ben Pfaff
---
lib/ofp-actions.c |6 +++---
lib/ofp-parse.c |6 +++---
lib/ofp-util.c|2 +-
ofproto/ofproto.c |1 -
4 files changed, 7 insertions(+), 8 deletions(-)
diff --git a/lib/ofp-actions.c b/lib/ofp-actions.c
index 9849771..e8f
These variables were initialized in an outer scope and then immediately
changed in an inner one, so they might as well be farther in.
Found by clang-analyzer.
Signed-off-by: Ben Pfaff
---
lib/coverage.c |8 +++-
1 file changed, 3 insertions(+), 5 deletions(-)
diff --git a/lib/coverage.
Found by clang-analyzer.
Signed-off-by: Ben Pfaff
---
lib/ofp-util.c |1 +
1 file changed, 1 insertion(+)
diff --git a/lib/ofp-util.c b/lib/ofp-util.c
index 6d551ad..a4ee2c2 100644
--- a/lib/ofp-util.c
+++ b/lib/ofp-util.c
@@ -6986,6 +6986,7 @@ ofputil_pull_buckets(struct ofpbuf *msg, size_
collect_rules_strict() and collect_rules_loose() destroy the rule
collections that they create if they return an error, and some of their
callers then go on to destroy them again. This could cause a double-free
in the case where rule_collection_destroy() actually calls free(). That
never happens
Ben Pfaff (4):
ofp-util: Fix null pointer dereference in ofputil_pull_buckets().
ofproto: Avoid theoretical double free of large rule collections.
coverage: Move m_idx, h_idx to an inner scope in coverage_run().
Drop assignments whose values are never used.
lib/coverage.c|8 +++---
Signed-off-by: Andy Zhou
---
datapath/datapath.c | 1 -
1 file changed, 1 deletion(-)
diff --git a/datapath/datapath.c b/datapath/datapath.c
index 065356f..1fd7bf3 100644
--- a/datapath/datapath.c
+++ b/datapath/datapath.c
@@ -140,7 +140,6 @@ static int queue_gso_packets(struct datapath *dp, str
Signed-off-by: Ben Pfaff
---
lib/netlink-socket.c | 16
1 file changed, 8 insertions(+), 8 deletions(-)
diff --git a/lib/netlink-socket.c b/lib/netlink-socket.c
index 8fd3933..378e719 100644
--- a/lib/netlink-socket.c
+++ b/lib/netlink-socket.c
@@ -721,15 +721,15 @@ nl_dump_st
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 the port dump later in the function
turns up no ports (which generally indicates a bug), then no channels will
be allocate
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
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 inspection.
Signed-off-by: Ben Pfaff
---
lib/netlink-socket.c | 13 ++---
1 file changed, 6 insertions(+), 7 deletions(-)
DPDK mempools rely on rte_lcore_id() to implement a thread-local cache.
Our non pmd threads had rte_lcore_id() == 0. This allowed concurrent access to
the "thread-local" cache, causing crashes.
This commit resolves the issue with the following changes:
- Every non pmd thread has the same lcore_id
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 doing and try to start openvswitch.
The negatives I see are
1. /usr is NFS mou
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
---
debian/ifupdown.sh |8 ++-
debian/openvswitch-switch.README.Debian | 38 +++
2
LGTM
Acked-by: Andy Zhou
On Tue, Jun 24, 2014 at 11:45 AM, Ben Pfaff wrote:
> Suggested-by: Gerald Rogers
> Signed-off-by: Ben Pfaff
> ---
> AUTHORS |1 +
> CodingStyle |7 +++
> 2 files changed, 8 insertions(+)
>
> diff --git a/AUTHORS b/AUTHORS
> index 64ede54..0a6c61a 1006
On Thu, Jul 10, 2014 at 01:59:18PM -0700, Ben Pfaff wrote:
> On Thu, Jul 10, 2014 at 05:57:03PM -0300, Flavio Leitner wrote:
> > It seems that the best way to create/destroy bridges or add/remove
> > ports is using OVSDB protocol, but as far as I know there is no
> > shared library yet for that.
>
Signed-off-by: Kirkland Spector
Acked-by: Andrey Falko
---
INSTALL.Debian | 15 ---
1 file changed, 8 insertions(+), 7 deletions(-)
diff --git a/INSTALL.Debian b/INSTALL.Debian
index 68c12ac..34251a1 100644
--- a/INSTALL.Debian
+++ b/INSTALL.Debian
@@ -34,20 +34,21 @@ You do not n
On Thu, Jul 10, 2014 at 02:38:00PM -0700, Gurucharan Shetty wrote:
> >> > What about when the controller is on network? You have to have that up
> >> > (networking-wise) before bring up the bridge, right?
> >> Can you elaborate with an example?
> >>
> >> I don't know of a use case yet where the con
OK. The commit message should therefore end with:
Signed-off-by: Kirkland Spector
Acked-by: Andrey Falko
It would be best to re-send it.
Thanks,
Ben.
On Mon, Jul 14, 2014 at 09:45:10AM -0700, Andrey Falko wrote:
> Kirkland wrote the patch, I reviewed it before he sent it over. We
>
Reported-by: Simon Jouet
Signed-off-by: Ben Pfaff
---
AUTHORS|1 +
lib/ofp-msgs.h |4 ++--
tests/ofp-print.at |4 ++--
3 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/AUTHORS b/AUTHORS
index 8296673..dbacdf8 100644
--- a/AUTHORS
+++ b/AUTHORS
@@ -259,6
Kirkland wrote the patch, I reviewed it before he sent it over. We
misunderstood what Signed-off-by really means. Now that we've read more
about it, should Kirk resend with the corrected field?
On Monday, July 14, 2014, Ben Pfaff wrote:
> On Fri, Jul 11, 2014 at 03:58:30PM -0700, Kirkland Specto
>>
> Yes, we can definitely do that as son as the porting takes its final shape,
> as some installer actions depend on the architectural choices that are in the
> making.
>
Great. Thanks.
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/m
Hi Guru,
> On 11.07.2014, at 23:06, "Gurucharan Shetty" wrote:
>
> Hello Alin, Alessandro,
> I see that you guys at Cloudbase have created an installer for Open
> vSwitch. Is it something you created as an one time effort? Or is it a
> script and one can create a new installer by running a simp
On Mon, Jul 14, 2014 at 08:01:42AM -0700, Gurucharan Shetty wrote:
> ovsdb-server's port on Windows does not support the "--run" option.
> The two tests skipped in this commit make use of "--run" option to
> test ovsdb-server's truncating of corrupt log or bad transaction.
> It looks a little trick
ovsdb-server's port on Windows does not support the "--run" option.
The two tests skipped in this commit make use of "--run" option to
test ovsdb-server's truncating of corrupt log or bad transaction.
It looks a little tricky to get this test running on Windows without
the "--run" option implemente
On Fri, Jul 11, 2014 at 03:58:30PM -0700, Kirkland Spector wrote:
> Signed-off-by: Andrey Falko
The content of this patch looks good, but the signoff is different from
the author. Who wrote the patch?
___
dev mailing list
dev@openvswitch.org
http://ope
46 matches
Mail list logo
