Right, I didn't see that before.
So I'm a bit confused why the distcleancheck target is complaining about
these files. Are the ovsidl test files added to OVSIDL_BUILT after it is
added to EXTRA_DIST?
I see it complaining about lib/socket-util-unix.{lo,o} now, too.
On 28 May 2014 03:11, Ben Pfaf
Yeah, I'd like to get a gauge of this as well. I'll follow up on this.
On 28 May 2014 14:28, Ethan Jackson wrote:
> Could we measure if this actually causes a performance degradation?
> I'd honestly be surprised if it does . . . My guess is our bottlenecks
> are elsewhere, but perhaps they aren
Could we measure if this actually causes a performance degradation?
I'd honestly be surprised if it does . . . My guess is our bottlenecks
are elsewhere, but perhaps they aren't.
Ethan
On Tue, May 27, 2014 at 7:13 PM, Joe Stringer wrote:
> Thanks for the Ack, I pushed this to branch-2.3.
>
> I'd
hi,
> Yamamoto San, thanks for following up on this issue. Yes, you are
> right, both DP_HASH and RECIRC actions are not handled in
> SLOW_ACTION.
>
> RECIRC action can be handled like output action, by using the helper function.
>
> DP_HASH action, in theory, can also be handled in a helper f
Thanks for the Ack, I pushed this to branch-2.3.
I'd like to send a patch for master anyway, in part so that we ensure to
address this (fix the bug, and make the code more readable), and in part
because it still matters if newer userspace is used with an older kernel.
On 28 May 2014 13:52, Alex W
Acked-by: Alex Wang
For master, I'm about to refactor the datapath so that we never dump
duplicated flows. So, I think we should consider refactoring master after
that.
Thanks,
Alex Wang,
On Tue, May 27, 2014 at 6:19 PM, Joe Stringer wrote:
> A series of bugs have been identified recently
Due to the differences between master and branch-2.3 and pressure to get
this out, I've targetted this first at branch-2.3. I intend to prepare an
equivalent patch for master, separately.
On 28 May 2014 13:19, Joe Stringer wrote:
> A series of bugs have been identified recently that are caused
A series of bugs have been identified recently that are caused by a
combination of the awkward flow dump API, possibility of duplicate flows
in a flow dump, and premature optimisation of the revalidator logic.
This patch attempts to simplify the revalidator logic by combining
multiple critical sect
Thanks,
Daniele
On May 27, 2014, at 4:44 PM, Ben Pfaff wrote:
> On Tue, May 27, 2014 at 03:20:08PM -0700, Daniele Di Proietto wrote:
>> This commit replace a while loop in miniflow_equal() with a call to
>> memcmp() for performace reasons.
>>
>> Signed-off-by: Daniele Di Proietto
>> ---
>> Th
On Fri, May 23, 2014 at 03:26:20PM +0300, Lori Jakab wrote:
> On 5/23/14, 12:22 AM, Ben Pfaff wrote:
> >On Tue, May 13, 2014 at 05:02:16PM +0300, Lorand Jakab wrote:
> >>Implementation of the pop_eth and push_eth actions in the kernel, and
> >>layer 3 flow support.
> >>
> >>Signed-off-by: Lorand Ja
On Mon, May 19, 2014 at 12:52:26PM -0700, Gurucharan Shetty wrote:
> So far, we log the kernel assigned port number when the port number is
> not specified. On Windows, this happens multiple times because "unix"
> sockets are implemented internally via TCP ports. This means that many tests,
> speci
On Mon, May 19, 2014 at 12:52:25PM -0700, Gurucharan Shetty wrote:
> This is mostly for unit tests that use ovs-ofctl monitor.
> service_start() creates a new process and waits till
> daemonize_complete() is called by the child.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
_
On Mon, May 19, 2014 at 12:52:24PM -0700, Gurucharan Shetty wrote:
> Otherwise, on windows unit tests data sometimes is
> not seen in output files.
>
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
http:
On Mon, May 19, 2014 at 12:52:23PM -0700, Gurucharan Shetty wrote:
> Without it, garbage values make it to set_dscp function
> in Windows.
>
> Signed-off-by: Gurucharan Shetty
Ouch. Seems like an obvious bugfix, please backport.
Acked-by: Ben Pfaff
On Mon, May 19, 2014 at 12:52:22PM -0700, Gurucharan Shetty wrote:
> ovsdb-client's 'monitor' command works with --detach such
> that the parent detaches after printing initial transactions in
> the database. This is a little tricky to implement
> in windows. So for windows, send the process to bac
On Thu, May 22, 2014 at 05:37:42PM -0700, Jarno Rajahalme wrote:
> The documentation of the memory order argument of atomic operations
> states that memory loads after an atomic load with a
> memory_order_acquire cannot be moved before the atomic operation.
> This still leaves open the possibility
On Tue, May 27, 2014 at 03:20:08PM -0700, Daniele Di Proietto wrote:
> This commit replace a while loop in miniflow_equal() with a call to
> memcmp() for performace reasons.
>
> Signed-off-by: Daniele Di Proietto
> ---
> This seems to improve throughput in my benchmarks (forwading one flow betwee
On Tue, May 27, 2014 at 09:36:12AM -0700, Ben Pfaff wrote:
> On Tue, May 27, 2014 at 06:05:36PM +0900, Simon Horman wrote:
> > handle_port_request() already provides common code to handle
> > both port desc and port stats request messages once they have been decoded
> > by using a callback.
> >
>
On Tue, May 27, 2014 at 09:57:09AM -0700, Ben Pfaff wrote:
> On Tue, May 27, 2014 at 06:05:38PM +0900, Simon Horman wrote:
> > Buffer a multi-part requests until all its parts are received.
> >
> > This is achieved by initialising the list_node field of messages
> > and passing them to ofmp_req_fi
Addressed all these style comments in v5.
Ryan
On 5/27/14 3:00 PM, "Ethan Jackson" wrote:
>Couple of minor style comments on my end.
>
>I'd prefer if xlate_init_new_xcfg() and set_new_cfg() are renamed to
>something like xlate_txn_start() and xlate_txn_commit(). From the
>callers perspecdtive
Before, a global read-write lock protected the ofproto-dpif / ofproto-dpif-xlate
interface. Handler and revalidator threads had to wait while configuration was
being changed. This patch implements RCU locking which allows handlers and
revalidators to operate while configuration is being updated.
S
This commit replace a while loop in miniflow_equal() with a call to
memcmp() for performace reasons.
Signed-off-by: Daniele Di Proietto
---
This seems to improve throughput in my benchmarks (forwading one flow between
two 'netdev-dpdk's).
---
lib/flow.c | 6 +-
1 file changed, 1 insertion(+)
On Tue, May 27, 2014 at 09:09:10PM +0100, Zoltan Kiss wrote:
> Recently we have seen a pool-wide storage fence in a XenServer 6.2
> SP1 enviroment. OVS 1.4.6+ were in use, up until the commit
> e58f54a4aa1 on that branch.
> In a 20 minutes period all 4 hosts in the pool lost network
> connectivity
Couple of minor style comments on my end.
I'd prefer if xlate_init_new_xcfg() and set_new_cfg() are renamed to
something like xlate_txn_start() and xlate_txn_commit(). From the
callers perspecdtive we're creating a transaction with which we change
the configuration. It's a bit more obvious what'
Hi,
Recently we have seen a pool-wide storage fence in a XenServer 6.2 SP1
enviroment. OVS 1.4.6+ were in use, up until the commit e58f54a4aa1 on
that branch.
In a 20 minutes period all 4 hosts in the pool lost network connectivity
and HA restarted them automatically. First ovsdb-server logged
On Tue, May 27, 2014 at 12:32:50PM -0700, Ansis Atteka wrote:
> Between fork() and execvp() calls in the process_start()
> function both child and parent processes share the same
> file descriptors. This means that, if a child process
> received a signal during this time interval, then it could
>
Between fork() and execvp() calls in the process_start()
function both child and parent processes share the same
file descriptors. This means that, if a child process
received a signal during this time interval, then it could
potentially write data to a shared file descriptor.
One such example is
Hi Ben,
On Mon, May 19, 2014 at 03:20:27PM -0700, Ben Pfaff wrote:
> On Thu, Apr 3, 2014 at 12:45 PM, Karol Mroz wrote:
> > On 2014-04-03, Ben Pfaff wrote:
> >> On Thu, Apr 03, 2014 at 06:45:47PM +, Karol Mroz wrote:
> >>> On 2014-04-03, Ben Pfaff wrote:
> >>> > On Tue, Apr 01, 2014 at 11:1
Yamamoto San, thanks for following up on this issue. Yes, you are
right, both DP_HASH and RECIRC actions are not handled in
SLOW_ACTION.
RECIRC action can be handled like output action, by using the helper function.
DP_HASH action, in theory, can also be handled in a helper function so
that the
On Fri, May 23, 2014 at 09:40:26AM -0700, Gurucharan Shetty wrote:
> Signed-off-by: Gurucharan Shetty
Acked-by: Ben Pfaff
___
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
On Tue, May 27, 2014 at 06:05:38PM +0900, Simon Horman wrote:
> Buffer a multi-part requests until all its parts are received.
>
> This is achieved by initialising the list_node field of messages
> and passing them to ofmp_req_filter().
>
> * If the message is not recognised as part of a multi-pa
On Tue, May 27, 2014 at 06:05:37PM +0900, Simon Horman wrote:
> Merge ofp_print_stats_reply() and ofp_print_stats_request()
> into a single new function, ofp_print_stats().
>
> For stats replies there should be no run-time change in behaviour.
>
> For pre-OpenFlow1.3 stats requests there should a
On Tue, May 27, 2014 at 06:05:36PM +0900, Simon Horman wrote:
> handle_port_request() already provides common code to handle
> both port desc and port stats request messages once they have been decoded
> by using a callback.
>
> This patch extends this common handling code to also perform decoding
On Tue, May 27, 2014 at 06:05:39PM +0900, Simon Horman wrote:
> Signed-off-by: Simon Horman
A port stats request consists of exactly one struct
ofp_port_stats_request. A port desc stats request is empty. Neither
one has any call to be broken into multiple parts.
Hi Ben,
Do you have comments on the scenario in which the assert might happen ?
Thanks,
Anoob.
From: Anoob Soman
Sent: 20 May 2014 20:06
To: Ben Pfaff
Cc: dev@openvswitch.org
Subject: RE: [ovs-dev] [PATCH 1/1] netflow: Modify netflow_flow_clear() to do
ne
On Sat, May 24, 2014 at 12:54:04PM +0100, Thomas Graf wrote:
> On 05/23/14 at 08:51am, Ben Pfaff wrote:
> > On Fri, May 23, 2014 at 09:03:54AM +0100, Thomas Graf wrote:
> > > > Using nl_sock_send_linear() can cause messages to be reordered. Do we
> > > > need to wait for the tx ring to empty befor
On Sat, May 24, 2014 at 12:44:40PM +0100, Thomas Graf wrote:
> On 05/23/14 at 08:47am, Ben Pfaff wrote:
> > On Fri, May 23, 2014 at 01:15:58AM +0200, Thomas Graf wrote:
> > > Signed-off-by: Thomas Graf
> >
> > nl_sock_recv__() compares its 'events' member against EPOLLERR with ==,
> > but EPOLL*
On Tue, May 27, 2014 at 01:46:59PM +1200, Joe Stringer wrote:
> This was causing failures in 'make distcleancheck'.
>
> Signed-off-by: Joe Stringer
OVSIDL_BUILT is already in EXTRA_DIST. It doesn't make sense to have
files be both distributed and distcleaned.
___
I don't know anything about these files and it's in the datapath
directory so I'll leave it to Pravin and Jesse.
On Tue, May 27, 2014 at 01:47:00PM +1200, Joe Stringer wrote:
> This was causing failures in 'make distcleancheck'.
>
> Signed-off-by: Joe Stringer
> ---
> datapath/linux/Makefile.ma
handle_port_request() already provides common code to handle
both port desc and port stats request messages once they have been decoded
by using a callback.
This patch extends this common handling code to also perform decoding
of both port desc and port stats request messages by using one more cal
Signed-off-by: Simon Horman
---
OPENFLOW-1.1+ | 6 ++-
ofproto/ofproto.c | 74 +---
tests/ofproto-macros.at | 1 +
tests/ofproto.at| 109
4 files changed, 163 insertions(+), 27 deletions(-)
Merge ofp_print_stats_reply() and ofp_print_stats_request()
into a single new function, ofp_print_stats().
For stats replies there should be no run-time change in behaviour.
For pre-OpenFlow1.3 stats requests there should also be
no run-time change in behaviour.
For OpenFlow1.3+ stats requests t
This purpose of this patchset is to provide initial multipart request
support and provide a basis for discussion of how such support should be
implemented.
There are 4 patches in this series:
1. ofproto: Further factor out handling of port requests
A clean up for ofproto port request handli
Buffer a multi-part requests until all its parts are received.
This is achieved by initialising the list_node field of messages
and passing them to ofmp_req_filter().
* If the message is not recognised as part of a multi-part requests it is
simply returned and processing continues as before.
*
44 matches
Mail list logo
