On Thu, Oct 22, 2015 at 11:48:41AM -0400, Mike Holmes wrote:
> Ping, and progress on this ?
>
Waiting for review.
It needs a rebase now though, v4 coming up...
> On 14 October 2015 at 08:31, Stuart Haslam wrote:
>
> > Fixes bug:
Previously only atomic scheduler queues where supported.
This was a bottleneck when the number of worker threads was
increased. Added an option to choose scheduler queue sync
mode (none, atomic, ordered).
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 94
Application crashes if accuracy option is set to zero
(division by zero). Disable statistics printing if accuracy
<= 0.
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 59 +++-
1 file changed, 37 insertions(+), 22
By default every packet's source MAC address is filled to
match the output port. Add option to disable this to enable
testing packet forwarding without touching the packets.
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 17 +++--
1 file changed,
Previously only even numbers of ports were supported. Added
support for odd numbers of ports (including 1). Src-dst port
mappings are saved during initialization to minimize per
packet operations.
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 35
On 23.10.15 13:35, Savolainen, Petri (Nokia - FI/Espoo) wrote:
I think these are not generally needed or supported. It's better to add a
parameter into odp_pktio_param_t.
It's needed.
The main reason for that the same statistic module can be used by different
pktios, so if it's used for
User tells to the implementation on open() which interfaces needed statistics.
As long as implementation has enough stat resources open() succeeds. When all
stats are gone and user still ask for stats, open() fails. Implementation user
manual documents this limitation (how many interfaces can
Errors which occur while sending packets via odp_pktio_send() aren't
handled correctly or even consistently across the two socket based
implementations.
The problems being addressed are;
- calls may block indefinitely in certain error conditions (mmsg)
- packets may be freed and reported as
Avoid the need to pass the interface name to the create_pktio() function
as it can be derived from the index which must also be passed. This
means the caller can't make the mistake of passing a mismatched name and
index - there was one instance of this which is now fixed.
Signed-off-by: Stuart
> -Original Message-
> From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT Maxim
> Uvarov
> Sent: Friday, October 23, 2015 1:03 PM
> To: lng-odp@lists.linaro.org
> Subject: Re: [lng-odp] [PATCH v5 3/6] test: l2fwd: fix crash when accuracy is
> set to
> 0
>
> On
Signed-off-by: Mike Holmes
---
m4/ax_valgrind_check.m4 | 2 --
1 file changed, 2 deletions(-)
diff --git a/m4/ax_valgrind_check.m4 b/m4/ax_valgrind_check.m4
index ff05f56..ccee86b 100644
--- a/m4/ax_valgrind_check.m4
+++ b/m4/ax_valgrind_check.m4
@@ -59,7 +59,6 @@
Somehow that email inside previous version thread it's hard to follow it.
I have few notes:
1. version should be remove from long log and put under "---" in email.
2. doxygen:
/opt/Linaro/odp2.git/include/odp/api/hash.h:73: warning: argument
'crc_param' of command @param is not found in the
Add a couple of tests for receiving packets directly via
odp_pktio_recv().
Signed-off-by: Stuart Haslam
---
test/validation/pktio/pktio.c | 101 ++
test/validation/pktio/pktio.h | 2 +
2 files changed, 65 insertions(+), 38
https://bugs.linaro.org/show_bug.cgi?id=1851
--- Comment #6 from carl.wal...@nokia.com ---
I applied your patch and it seems to be working just fine. I did not test
performance though. Thanks a lot!
--
You are receiving this mail because:
You are on the CC list for the
Optimize ethernet address filling by using word copy.
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 57 ++--
1 file changed, 45 insertions(+), 12 deletions(-)
diff --git a/test/performance/odp_l2fwd.c
Enable doxygen to document platform-specific config values
Signed-off-by: Bill Fischofer
---
platform/linux-generic/include/odp/config.h | 9 +
1 file changed, 9 insertions(+)
diff --git a/platform/linux-generic/include/odp/config.h
https://bugs.linaro.org/show_bug.cgi?id=1864
Bug ID: 1864
Summary: Missing header file in install directory
Product: OpenDataPlane - linux- generic reference
Version: unspecified
Hardware: Other
OS: Linux
Status:
Previously only even numbers of ports were supported. Added
support for odd numbers of ports (including 1). Src-dst port
mappings are saved during initialization to minimize per
packet operations.
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 35
By default every packet's source MAC address is filled to
match the output port. Add option to disable this to enable
testing packet forwarding without touching the packets.
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 17 +++--
1 file changed,
On 23 October 2015 at 03:00, Petri Savolainen
wrote:
> Refined and centralized comment about relaxed memory ordering.
> Removed in/out doxygen tags since 'atom' pointer to an object
> that application must not directly access (only through the API).
>
> Signed-off-by:
On 23 October 2015 at 08:40, Christophe Milard wrote:
> Well, I agree that the test environment should not prevent API
> improvement: The test environment should simply follow the API as it moves.
> I see no problem with this change as long as the test environment
Ping. Why this has not been merged yet.
> -Original Message-
> From: Savolainen, Petri (Nokia - FI/Espoo)
> Sent: Thursday, October 15, 2015 11:08 AM
> To: 'EXT Peng'; lng-odp@lists.linaro.org
> Cc: Peng
> Subject: RE: [PATCH] [API-NEXT PATCH v5] api: hash: Added crc32 and crc32c
> hash
On 23 October 2015 at 03:00, Petri Savolainen
wrote:
> Group level documentation can be used for API level general
> description. Layout file is needed to place detailed group
> description on the top of a page. By default doxygen shows
> only the brief description on
If the application is being run on a host connected to a
switch, not changing the ethernet destination may cause
loops. Added an option to change the destination ethernet
address. The destination addresses follow the format
02:00:00:00:00:XX where the final octet is the output
port number.
Added various improvements to the l2fwd test application:
- Supports using odd number of ports
- Scheduler queue type can be selected
- Added options for enabling/disabling eth address filling
New options:
-d, --dst_change <0/1>: Enable/disable changing packets' dst eth addresses
-s,
Previously only atomic scheduler queues where supported.
This was a bottleneck when the number of worker threads was
increased. Added an option to choose scheduler queue sync
mode (none, atomic, ordered).
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 94
On 23 October 2015 at 03:00, Petri Savolainen
wrote:
> Reference caused warning when creating doxygen documentation.
>
> Signed-off-by: Petri Savolainen
>
Reviewed-by: Mike Holmes
> ---
>
As long as the test-environment follows up (within a reasonable time), I am
fine.
On 23 October 2015 at 14:52, Savolainen, Petri (Nokia - FI/Espoo) <
petri.savolai...@nokia.com> wrote:
>
>
>
>
> *From:* EXT Mike Holmes [mailto:mike.hol...@linaro.org]
> *Sent:* Friday, October 23, 2015 3:43 PM
>
Application crashes if accuracy option is set to zero
(division by zero). Disable statistics printing if accuracy
<= 0.
Signed-off-by: Matias Elo
---
test/performance/odp_l2fwd.c | 37 +
1 file changed, 25 insertions(+), 12 deletions(-)
There's no need to enforce an artificial MTU for the loop interface.
Previously this value was reported but not enforced.
Signed-off-by: Stuart Haslam
---
platform/linux-generic/pktio/loop.c | 5 ++---
1 file changed, 2 insertions(+), 3 deletions(-)
diff --git
On 23.10.15 14:18, Savolainen, Petri (Nokia - FI/Espoo) wrote:
-Original Message-
From: EXT Ivan Khoronzhuk [mailto:ivan.khoronz...@linaro.org]
Sent: Friday, October 23, 2015 2:12 PM
To: Savolainen, Petri (Nokia - FI/Espoo); EXT Maxim Uvarov; lng-
o...@lists.linaro.org
Subject: Re:
Well, I agree that the test environment should not prevent API improvement:
The test environment should simply follow the API as it moves. I see no
problem with this change as long as the test environment follows it. I do
see a problem, though, if it does not... What structure should the test
Fixes bug: https://bugs.linaro.org/show_bug.cgi?id=1365
Changes since v3:
- Rebased
- Fix comment typo in 5/5
Changes since v2:
- Conditionally check for ability to run test and report as
inactive if not.
- dropped patch 1/7, not required after other test changes
- dropped patch 4/7
I think these are not generally needed or supported. It's better to add a
parameter into odp_pktio_param_t.
For example,
typedef enum odp_pktio_stats_mode_t {
/** Need basic statistics on this interface */
ODP_PKTIO_STATS_BASIC = 0,
/** Don't need any statistics on this interface */
> -Original Message-
> From: EXT Ivan Khoronzhuk [mailto:ivan.khoronz...@linaro.org]
> Sent: Friday, October 23, 2015 2:12 PM
> To: Savolainen, Petri (Nokia - FI/Espoo); EXT Maxim Uvarov; lng-
> o...@lists.linaro.org
> Subject: Re: [lng-odp] [PATCHv5 2/2] api: pktio statistics: define
On 23 October 2015 at 03:00, Petri Savolainen
wrote:
> Removed module synchronizer from doxygen documentation and
> introduced new modules for locks, atomics and barriers.
This ripples though the to the validation directory structure which is 1:1
in sync with the
From: EXT Mike Holmes [mailto:mike.hol...@linaro.org]
Sent: Friday, October 23, 2015 3:22 PM
To: Savolainen, Petri (Nokia - FI/Espoo)
Cc: lng-odp
Subject: Re: [lng-odp] [API-NEXT PATCH v3 3/7] api: doc: re-organize doxygen
doc for synchronizer
On 23 October 2015 at 03:00, Petri Savolainen
Test that transmit errors are handled correctly by attempting to send a
packet larger than the MTU of the interface.
Signed-off-by: Stuart Haslam
---
test/validation/pktio/pktio.c | 151 +-
1 file changed, 148 insertions(+), 3
Signed-off-by: Maxim Uvarov
---
v6: fix Petris comments
include/odp/api/packet_io.h | 2 +
include/odp/api/packet_io_stats.h | 141 ++
2 files changed, 143 insertions(+)
create mode 100644 include/odp/api/packet_io_stats.h
> -Original Message-
> From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT
> Maxim Uvarov
> Sent: Thursday, October 22, 2015 1:45 PM
> To: lng-odp@lists.linaro.org
> Subject: [lng-odp] [PATCHv5 1/2] api: define pktio statistics api
>
> Signed-off-by: Maxim Uvarov
Hi, Maxim
> 在 2015年10月23日,下午11:40,Maxim Uvarov 写道:
>
> Somehow that email inside previous version thread it's hard to follow it.
I am having troubles in using git send-email to reply to specific emails, even
I’ve
tried using —reply-to message-id, the email is sent as
https://bugs.linaro.org/show_bug.cgi?id=1449
--- Comment #11 from Ivan Khoronzhuk ---
Will send patch series soon to add fixes and corrections eliminating resolution
impact demonstrated by previous comment and decreasing impact of schedule
delays.
--
You are
Added atomic min and max operations. These can be used e.g.
to maintain high and low water marks of an another atomic counter.
These use relaxed memory order.
Signed-off-by: Petri Savolainen
---
include/odp/api/atomic.h| 44
Added 32 bit acquire load/cas and release store/add/sub calls.
These are the minimum set of non-relaxed calls that are needed
for building lock-free algorithms. 64 bit versions can be added
later.
Signed-off-by: Petri Savolainen
---
include/odp/api/atomic.h
Refined and centralized comment about relaxed memory ordering.
Removed in/out doxygen tags since 'atom' pointer to an object
that application must not directly access (only through the API).
Signed-off-by: Petri Savolainen
---
include/odp/api/atomic.h | 112
Sent out v3 which corrects these two issues.
-Petri
From: EXT Mike Holmes [mailto:mike.hol...@linaro.org]
Sent: Thursday, October 22, 2015 5:15 PM
To: Savolainen, Petri (Nokia - FI/Espoo)
Cc: lng-odp
Subject: Re: [lng-odp] [API-NEXT PATCH v2 2/7] doc: add doxygen layout file to
control group
Reference caused warning when creating doxygen documentation.
Signed-off-by: Petri Savolainen
---
include/odp/api/packet_io.h| 2 +-
platform/linux-generic/include/odp/debug.h | 8
2 files changed, 1 insertion(+), 9 deletions(-)
diff --git
Removed module synchronizer from doxygen documentation and
introduced new modules for locks, atomics and barriers. Removed
unnecessary group tagging from internal headers, which are not
visible to doxygen anyway.
Signed-off-by: Petri Savolainen
---
Group level documentation can be used for API level general
description. Layout file is needed to place detailed group
description on the top of a page. By default doxygen shows
only the brief description on top (the first sentence).
Signed-off-by: Petri Savolainen
Added cas operations for 32 and 64 bit atomic variables. These
use relaxed memory order (as all other operations).
Signed-off-by: Petri Savolainen
---
include/odp/api/atomic.h | 37 ++
On 22 October 2015 at 18:06, Nikita Kalyazin wrote:
> Hi,
>
> > [Alex] ODP queues are neither software nor hardware by definition, each
> > implementation is free to implement them as they see fit. Also PacketI/O
> > abstraction is not an abstraction for a NIC device. The
> -Original Message-
> From: EXT Ivan Khoronzhuk [mailto:ivan.khoronz...@linaro.org]
> Sent: Thursday, October 22, 2015 3:02 PM
> To: lng-odp@lists.linaro.org; Savolainen, Petri (Nokia - FI/Espoo)
> Cc: Ivan Khoronzhuk
> Subject: [lng-odp] [API-NEXT PATCH v5 3/5] api: time: unbind CPU
Netmap interface takes a few seconds to become active after
calling nm_open(). This caused several test applications to
fail. Check link status at the end of netmap_open() to fix
this.
Signed-off-by: Matias Elo
---
v3:
- Rebased to master
- Added sleep comment to
Reviewed-by: Petri Savolainen
Bill, did you run 'make doxygen-html'?
Not sure if @def ODP_PACKET_GREEN works when it's actually an enum. Anyway,
it's in the correct file now.
-Petri
> -Original Message-
> From: lng-odp
Reviewed-by: Petri Savolainen
> -Original Message-
> From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT
> Bill Fischofer
> Sent: Friday, October 23, 2015 12:19 AM
> To: lng-odp@lists.linaro.org
> Subject: [lng-odp] [PATCH] linux-generic:
Hi, Petri
Thanks for the reply.
On 23.10.15 10:57, Savolainen, Petri (Nokia - FI/Espoo) wrote:
-Original Message-
From: EXT Ivan Khoronzhuk [mailto:ivan.khoronz...@linaro.org]
Sent: Thursday, October 22, 2015 3:02 PM
To: lng-odp@lists.linaro.org; Savolainen, Petri (Nokia - FI/Espoo)
Using separate file descriptors enables using rx and tx
functions simultaneously from different threads.
Previously netmap tx queues were flushed only after no free
slots were left which caused long delays on slow packet
rates. Now queues are flushed after each packet
burst.
Signed-off-by:
Added various improvements to the l2fwd test application:
- Supports using odd number of ports
- Scheduler queue type can be selected
- Added options for enabling/disabling eth address filling
New options:
-d, --dst_change <0/1>: Enable/disable changing packets' dst eth addresses
-s,
If the application is being run on a host connected to a
switch, not changing the ethernet destination may cause
loops. Added an option to change the destination ethernet
address. The destination addresses follow the format
02:00:00:00:00:XX where the final octet is the output
port number.
59 matches
Mail list logo