On 06.08.15 15:08, Maxim Uvarov wrote:
On 08/06/15 14:30, Ivan Khoronzhuk wrote:
Hi, Maxim
On 06.08.15 11:28, Maxim Uvarov wrote:
On 08/06/15 00:01, Ivan Khoronzhuk wrote:
/**
- * Current time in CPU cycles
+ * Current time in counts of best hi-resolution timer available
*
- * @return
Hi, Maxim
On 06.08.15 11:28, Maxim Uvarov wrote:
On 08/06/15 00:01, Ivan Khoronzhuk wrote:
/**
- * Current time in CPU cycles
+ * Current time in counts of best hi-resolution timer available
*
- * @return Current time in CPU cycles
+ * @return Current time in timer counts
*/
-uint64_t
On 06.08.15 13:08, Stuart Haslam wrote:
On Tue, Aug 04, 2015 at 03:05:21PM +0300, Ivan Khoronzhuk wrote:
It's needed as pktio is not set in promisc mode.
Signed-off-by: Ivan Khoronzhuk
---
test/performance/odp_pktio_perf.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
The word "cycle" is left from old API time names. The "cycle" is
ambiguous word, especially when it can be used with software cycles.
So better to use "time" word or just "t" symbol, as no matter in
which units time is measured, in sec, hours, cycles or
The time API were changed from *_cycles* to *_counts*, so use
appropriate names.
Signed-off-by: Ivan Khoronzhuk
---
platform/linux-generic/arch/linux/odp_time.c | 2 +-
platform/linux-generic/arch/mips64/odp_time.c | 2 +-
platform/linux-generic/arch/x86/odp_time.c| 2 +-
platform/linux
The word "cycle" is left from old API time names. The "cycle" is
ambiguous word, especially when it can be used with software cycles.
So better to use "time" word or just "t" symbol, as no matter in
which units time is measured, in sec, hours, cycles or
which cycles to measure
time, the better hi-resolution timer the better measurements.
So, unbind CPU cycle counter from time API by eliminating word
"cycle" as it's believed to use with CPU.
Signed-off-by: Ivan Khoronzhuk
---
example/ipsec/odp_ipsec.c | 6
This seres is intedent to change time API names from *_cycles*
to *_counts*. Also remove usage of word "cycle" from appropriate
places as it's no more valid.
Based on api-next.
Ivan Khoronzhuk (4):
api: time: unbind CPU cycles from time API
test/example: avoid "cycle
The API names even can be leaved as is.
Only description change (remove CPU cycles)
On 05.08.15 15:01, Ivan Khoronzhuk wrote:
On 04.08.15 15:19, Mike Holmes wrote:
In the platform call just now the discussion came to this patch.
Ivan can you propose an API to get the timer resolution for the
o imitate the
"CPU counter" API needs redundant conversion to get real ns.
On 29 July 2015 at 08:39, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
Mike,
On 29.07.15 15:27, Mike Holmes wrote:
On 29 July 2015 at 08:17, Ivan Khoronzhuk
shure
On 04.08.15 15:19, Mike Holmes wrote:
In the platform call just now the discussion came to this patch.
Ivan can you propose an API to get the timer resolution for the
platform, then the test suite can call that and run the test appropriately.
On 29 July 2015 at 08:39, Ivan Khoronzhuk
It's needed as pktio is not set in promisc mode.
Signed-off-by: Ivan Khoronzhuk
---
test/performance/odp_pktio_perf.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)
diff --git a/test/performance/odp_pktio_perf.c
b/test/performance/odp_pktio_perf.c
index 08f13b6..fe4261e 1
Bala,
On 03.08.15 20:49, Bala Manoharan wrote:
Ivan,
On 3 August 2015 at 22:58, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
Bala,
On 03.08.15 20:16, Bala Manoharan wrote:
Ivan,
On 3 August 2015 at 22:33, Ivan Khoronzhuk
mailto:ivan.k
Bala,
On 03.08.15 20:16, Bala Manoharan wrote:
Ivan,
On 3 August 2015 at 22:33, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
Bala,
On 03.08.15 19:51, Bala Manoharan wrote:
On 3 August 2015 at 22:16, Ivan Khoronzhuk
mailto:ivan.khoronz...@lina
Bala,
On 03.08.15 19:51, Bala Manoharan wrote:
On 3 August 2015 at 22:16, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
Bala,
...
what in case of TCP? Incorrect seq num?
Looks like a coding error. TCP should be added
)
+ return data.seq;
+ }
+
+ return TEST_SEQ_INVALID;
+}
+
...
--
Regards,
Ivan Khoronzhuk
Regards,
Bala
--
Re
On 03.08.15 19:24, Bala Manoharan wrote:
On 3 August 2015 at 21:44, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
One more issue
On 30.07.15 18:20, Balasubramanian Manoharan wrote:
Additional test suite is added to classification validation
su
return data.seq;
+ }
+
+ return TEST_SEQ_INVALID;
+}
+
...
--
Regards,
Ivan Khoronzhuk
___
lng-odp mailing list
lng-odp@lists.linaro.org
https://lists.linaro.org/mailman/listinfo/lng-odp
As a proposition, we had talk some time ago about,
You moved packet creation to common classification file.
Maybe it's time to assign correct src/dst MAC address, taken from pktio?
In separate patch.
--
Regards,
Ivan Khoronzhuk
___
lng-odp mailing
cket_t create_packet(bool vlan);
+odp_pktio_t create_pktio(odp_queue_type_t q_type);
+odp_queue_t create_default_inq(odp_pktio_t pktio, odp_queue_type_t qtype);
void configure_pktio_default_cos(void);
void test_pktio_default_cos(void);
void configure_pk
lidation/ver_abt_log_dbg/odp_system.h
delete mode 100644 test/validation/ver_abt_log_dbg/odp_ver_abt_log_dbg.c
___
lng-odp mailing list
lng-odp@lists.linaro.org
https://lists.linaro.org/mailman/listinfo/lng-odp
--
Regards,
Ivan Khoronzhuk
_
without
regard to its relative order with respect to other items originating
from the same ordered queue. In effect, after a release_ordered() call
the event behaves as if it came from a parallel queue rather than an
ordered one.
On Wed, Jul 29, 2015 at 10:36 AM, Ivan Khoronzhuk
mailto:ivan.khoronz
t
lng-odp@lists.linaro.org <mailto:lng-odp@lists.linaro.org>
https://lists.linaro.org/mailman/listinfo/lng-odp
_______
lng-odp mailing list
lng-odp@lists.linaro.org
https://lists.linaro.org/mailman/listinfo/lng-odp
--
Regards,
Ivan Khoronzhuk
___
lng-odp mailing list
lng-odp@lists.linaro.org
https://lists.linaro.org/mailman/listinfo/lng-odp
Mike,
On 29.07.15 15:27, Mike Holmes wrote:
On 29 July 2015 at 08:17, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
On 29.07.15 14:47, Bill Fischofer wrote:
Agreed. An ODP implementation should be for a platform, not for a
specific model within a pl
On 29.07.15 14:47, Bill Fischofer wrote:
Agreed. An ODP implementation should be for a platform, not for a
specific model within a platform family. Another reason why config.h
needs to be converted from #defines to APIs.
There is a little more sense under it. Each CPU, theoretically, (no
se
On 29.07.15 14:42, Ivan Khoronzhuk wrote:
On 29.07.15 14:34, Maxim Uvarov wrote:
Should we have
#define CONFIG_TIMER_RESOLUTION
in config.h?
And relay on it's value.
Maxim.
Maybe.
In case of CPU freq this is inversely proportional to odp_sys_cpu_hz(),
and can be determined. But in
s no
API to get it's freq, so it make sense.
On 07/28/15 18:23, Ivan Khoronzhuk wrote:
Current test supposes that timer counter resolution is very close
to 1ns. But that's not always true. The resoulution can be 8ns for
instance, it's not always close to CPU frequency.
Current test supposes that timer counter resolution is very close
to 1ns. But that's not always true. The resoulution can be 8ns for
instance, it's not always close to CPU frequency.
Signed-off-by: Ivan Khoronzhuk
---
test/validation/time/time.c | 2 +-
1 file changed, 1 inser
On 24.07.15 19:03, Ivan Khoronzhuk wrote:
Mike,
On 24.07.15 17:42, Mike Holmes wrote:
I had hoped this might fix the clang thread sanitizer warnings - see
https://bugs.linaro.org/show_bug.cgi?id=1457
But when applied it does not fix them, however this does not mean these
are bad patches
ple/timer/odp_timer_test.c:164 test_abs_timeouts
On 24 July 2015 at 09:31, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
This series fixes some races in the timer test termination path and
adds some new abilities. The races and drawbacks where found while
testing lin
In case if for some reason one of the TMOs was lost, the test will
be polling forever in order to take it and delete. It can be used to
indicate that event was lost. Just abort the test if it cannot receive
one of the "last" TMOs for some period.
Signed-off-by: Ivan Khoronzhuk
---
exa
an assign event for any of the threads, so one thread can
receive two last events for example.
According to above, added several improvements:
- don't send more timeouts that supposed to receive
- free timer and tmo for a last received tmos = num of threads.
- leave the main loop only if a last
ments were fixed
Ivan Khoronzhuk (2):
example: timer: delete races while termination
example: timer: use ability to indicate that event was lost
example/timer/odp_timer_test.c | 45 ++
1 file changed, 28 insertions(+), 17 deletions(-)
--
value for an odp_cos_t
*
* @param hdl odp_cos_t handle to be printed
--
Regards,
Ivan Khoronzhuk
___
lng-odp mailing list
lng-odp@lists.linaro.org
https://lists.linaro.org/mailman/listinfo/lng-odp
NOTE: Ignored message types: DEPRECATED_VARIABLE NEW_TYPEDEFS
0001-example-timer-delete-races-while-termination.patch has style
problems, please review.
On 3 July 2015 at 08:38, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
Current implementation has at least two races that lead to s
Is someone going to pickup these obvious race condition fixes?
On 03.07.15 15:38, Ivan Khoronzhuk wrote:
This series fixes some races in the timer test termination path and
adds some new abilities. The races and drawbacks where found while
testing linux-generic timer API on Keystone
What about to pick up this patch?
I still need it. I'm not sure, but seems this issue was before this
patch.
On 01.07.15 19:03, Ola Liljedahl wrote:
On 30 June 2015 at 12:16, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
It's more generic implementation simpli
ctually it will be one PMR, but emulated by two. And implementation
cannot know a priori in what circumstances this entry will be used.
So, question is openBut, at least, this change allows to return
more accurate number.
Signed-off-by: Ivan Khoronzhuk
---
include/odp/api/classification.h | 6
ss through the list server.
It's good when the letter is seen by everyone with the same subject.
Envoyé depuis un mobile Samsung.
Message d'origine
De : Ivan Khoronzhuk
Date :15/07/2015 18:34 (GMT+01:00)
À : Bill Fischofer , Genis Riera
Cc : ODP mailing list
Objet :
On 15.07.15 19:43, Ivan Khoronzhuk wrote:
Hi, Nicolas
Maybe I'm wrong, but it's supposed that headroom is used to extend the
packet data, when it's required. And if you need some meta-data you
probably should add it in packet header or else, but under headroom.
I meant your
Hi, Nicolas
Maybe I'm wrong, but it's supposed that headroom is used to extend the
packet data, when it's required. And if you need some meta-data you
probably should add it in packet header or else, but under headroom.
On 15.07.15 19:28, Nicolas Morey-Chaisemartin wrote:
As this is a change of
gri...@starflownetworks.com <mailto:gri...@starflownetworks.com>
On Wed, Jul 15, 2015 at 6:04 PM, Ivan Khoronzhuk
mailto:ivan.khoronz...@linaro.org>> wrote:
Genis
On 15.07.15 19:00, Genis Riera wrote:
Hi, Ivan,
Not, but I adde
Genis,
On 15.07.15 19:11, Genis Riera wrote:
Ivan,
If you have this compilation issues I can send again the patch without
this check, assuming always positive values. Is it right for you?
Yes, but wait a while, maybe someone else have some propositions.
__
eer at StarFlow Networks
Edifici K2M, S103 c/ Jordi Girona 31
08034 Barcelona
E-mail: gri...@starflownetworks.com <mailto:gri...@starflownetworks.com>
On Wed, Jul 15, 2015 at 5:54 PM, Ivan Khoronzhuk
mailto:ivan.khoronz...@linaro.org>> wrote:
Hi, Genis
On 15.07.15 18:47, Ge
Hi, Genis
On 15.07.15 18:47, Genis Riera wrote:
A new headroom parameter is added to odp_pool_param_t when the pool is
type of ODP_POOL_PACKET, so that headroom size could be configured at
pool initialization time (ODP application runtime).
Signed-off-by: Genis Riera Perez mailto:gri...@starflo
resources for that, and in case if it cannot implement l2-l2-l3 it
has to return error, but not 2.
On 15.07.15 17:35, Bala wrote:
On 15-Jul-2015, at 7:58 pm, Ivan Khoronzhuk wrote:
Bala,
.
On 15.07.15 12:25, Bala Manoharan wrote:
.
It's only a proposition to add.
Bala,
.
On 15.07.15 12:25, Bala Manoharan wrote:
.
It's only a proposition to add.
You can add this as separate test. I just want to show that current test
is not enough to test this primitive.
The current validation suite is just used to test the acceptance of
differ
Bala,
On 15.07.15 11:31, Bala Manoharan wrote:
Hi Ivan,
Comments Inline...
On 15 July 2015 at 02:48, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
It's simple improvement is intended to open eyes on possible
hidden issues when a packet can be lost (or sent
used only one.
It's not formated patch and is only for demonstration.
Signed-off-by: Ivan Khoronzhuk
---
.../classification/odp_classification_tests.c | 42 ++
1 file changed, 42 insertions(+)
diff --git a/test/validation/classification/odp_classification_tests
I suppose, the intention was to check only ODP_PMR_IPPROTO capability.
Signed-off-by: Ivan Khoronzhuk
---
test/validation/classification/odp_classification_tests.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/test/validation/classification/odp_classification_tests.c
b
Maxim,
On 03.07.15 18:13, Maxim Uvarov wrote:
On 3 July 2015 at 17:59, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
Maxim,
I've never used such kind of lock.
They are not used in LK for a lot of reasons.
It's not the same as rwlocks, it'
07/03/15 17:00, Ivan Khoronzhuk wrote:
Maxim,
mutex and spinlock is not the same ...
yes, I know.
What is the use case of that locks? Defend some data between control
thread and worker thread? I.e. odp worker holds write lock and updates
something. While control thread needs do periodical re
For my opinion, using lock in lock its bad design. It shouldn't be
masked. So it can be useful only for legacy applications and should be
avoided in new apps or even disabled somehow.
On 03.07.15 17:00, Ivan Khoronzhuk wrote:
Maxim,
mutex and spinlock is not the same ...
On 03.07.15
Maxim,
mutex and spinlock is not the same ...
On 03.07.15 16:54, Maxim Uvarov wrote:
Petri, looks like you invented mutex (mtx_recursive) :
https://www-01.ibm.com/support/knowledgecenter/ssw_aix_71/com.ibm.aix.basetrf1/mtx_destroy.htm
Thanks,
Maxim.
On 07/03/15 16:25, Petri Savolainen wrot
In case if for some reason one of the TMOs was lost, the test will
be polling forever in order to take it and delete. It can be used to
indicate that event was lost. Just abort the test if it cannot receive
one of the "last" TMOs for some period.
Signed-off-by: Ivan Khoronzhuk
---
exa
an assign event for any of the threads, so one thread can
receive two last events for example.
According to above, added several improvements:
- don't send more timeouts that supposed to receive
- free timer and tmo for a last received tmos = num of threads.
- leave the main loop only if a last
This series fixes some races in the timer test termination path and
adds some new abilities. The races and drawbacks where found while
testing linux-generic timer API on Keystone implementation, so it's
tested.
Since v1:
- only comments were fixed
Ivan Khoronzhuk (2):
example:
Just decided to add the ability to indicate that some TMO was lost.
At least it already helped me to figure out the issue in odp_schedule of
keystone implementation.
So see this patch inside new series:
"[lng-odp] [Patch 0/2] example: timer: fix/improve test"
On 03.07.15 12:56, Ivan
leave the main loop only if the last tmo/timer is free.
Signed-off-by: Ivan Khoronzhuk
---
example/timer/odp_timer_test.c | 38 +-
1 file changed, 21 insertions(+), 17 deletions(-)
diff --git a/example/timer/odp_timer_test.c b/example/timer/odp_timer_test.c
ind
In case if for some reason one of the TMOs was lost, the test will
be polling forever in order to take it and delete. It can be used to
indicate that event was lost. Just abort the test if it cannot receive
one of the last TMOs for some period.
Signed-off-by: Ivan Khoronzhuk
---
example/timer
This series fixes some races in the timer test and adds some
new abilities. The races and drawbacks where found while testing
linux-generic timer API on keystone implementation, so it tested.
Ivan Khoronzhuk (2):
example: timer: delete races while termination
example: timer: use ability to
On 03.07.15 12:56, Ivan Khoronzhuk wrote:
Current implementation has at least two races that lead to several issues:
- gbls->remain can overflow. One thread can decrement remain counter to 0.
While another can decrement it once again and it will be > 0 once again. After
what the threa
leave the main loop only if the last tmo/timer is free.
Signed-off-by: Ivan Khoronzhuk
---
example/timer/odp_timer_test.c | 38 +-
1 file changed, 21 insertions(+), 17 deletions(-)
diff --git a/example/timer/odp_timer_test.c b/example/timer/odp_timer_test
Ola,
I need this change.
Could you please comment on this if you have some objections.
or maybe send your version.
On 30.06.15 13:16, Ivan Khoronzhuk wrote:
It's more generic implementation simplify reusing timer API
for other platforms.
Signed-off-by: Ivan Khoronzhuk
---
I've ch
Hi, Bill
On 01.07.15 05:12, Bill Fischofer wrote:
Signed-off-by: Bill Fischofer
---
include/odp/api/classification.h | 106 +++
1 file changed, 106 insertions(+)
diff --git a/include/odp/api/classification.h b/include/odp/api/classification.h
index f597b2
It's more generic implementation simplify reusing timer API
for other platforms.
Signed-off-by: Ivan Khoronzhuk
---
I've checked this patch with Keystone implementation.
.../linux-generic/include/odp_timer_internal.h | 9 ---
platform/linux-generic/odp_timer.c
POSIX model and provides for that implementation flexibility. In this
case for linux-generic it's implementated as a memset, but that detail
is now hidden from the user as other platforms are free to implement the
init functions in a manner that work best for them.
On Thu, Jun 25, 2015 at 12:06
+ lng-odp. Unintentionally removed
On 25.06.15 20:06, Ivan Khoronzhuk wrote:
Hi Bill,
On 25.06.15 15:09, Bill Fischofer wrote:
Signed-off-by: Bill Fischofer
---
include/odp/api/pool.h | 9 +
1 file changed, 9 insertions(+)
diff --git a/include/odp/api/pool.h b/include/odp/api
Hi Bill,
On 25.06.15 16:52, Bill Fischofer wrote:
This RFC adds routines that permit classification rules to be manipulated
in a stateful manner. The odp_pktio_pmr_cos() routine is intended to be
deprecated and in its place the new odp_pktio_cls_context_swap() routine
is to be used instead.
Th
On 24.06.15 20:09, Bill Fischofer wrote:
Answers inline
On Wed, Jun 24, 2015 at 11:13 AM, Ivan Khoronzhuk
mailto:ivan.khoronz...@linaro.org>> wrote:
Guys, sorry I didn't ask the following questions during the ODP meeting.
I had an issue with my microphone and it seems t
On 24.06.15 19:57, Bala Manoharan wrote:
Hi Ivan,
Pls see my comments inline.
On 24 June 2015 at 09:13, Ivan Khoronzhuk mailto:ivan.khoronz...@linaro.org>> wrote:
Guys, sorry I didn't ask the following questions during the ODP meeting.
I had an issue with my microphone a
generates error.
It's not correct, the code shouldn't have such kind dependency from
macro definition, and it prevents to re-use this code for another
platforms with different ODP_QUEUE_INVALID. So, fix it.
Reviewed-by: Ola Liljedahl
Signed-off-by: Ivan Khoronzhuk
---
platform/lin
Guys, sorry I didn't ask the following questions during the ODP meeting.
I had an issue with my microphone and it seems the call was ended quickly.
But I need to ask. Maybe it's better, it requires some illustration.
For now, I need the answer on the following questions:
- Are we going to use th
. It's not correct, the code shouldn't
have such kind dependency from macro definition, and it prevents to
re-use this code for another platforms with different
ODP_QUEUE_INVALID. So, fix it.
Signed-off-by: Ivan Khoronzhuk
---
platform/linux-generic/odp_timer.c | 1 +
1 file changed, 1
. It's not correct, the code shouldn't
have such kind dependency from macro definition, and it prevents to
re-use this code for another platforms with different
ODP_QUEUE_INVALID. So, fix it.
Signed-off-by: Ivan Khoronzhuk
---
platform/linux-generic/odp_timer.c | 1 +
1 file changed, 1
Hi Ola,
Maybe I'm wrong, but let me try to explain change in question.
The main reason in this change is the lack of possibility to convert
buf_hdr to
buf_handle. The function odp_timeout_to_event requires this conversion.
By and
large it's not required the tmo_handle to be the tmo_header for
601 - 676 of 676 matches
Mail list logo