Hi, Thanks Bill
My title are more like: ODP User Cases in Cloud Profile Discussion (1 hour
with a google document).
Best Regards, Yi
On 9 February 2017 at 00:07, Bill Fischofer
wrote:
> As we discussed this morning, the LNG Design Summit to be held at
> BUD17 will
Hi, Maxim, please help check if this can be merged.
Thanks and Best Regards, Yi
On 7 February 2017 at 11:16, Bill Fischofer
wrote:
> On Thu, Dec 22, 2016 at 8:32 PM, Yi He wrote:
> > SP scheduler hangs in process mode performance test
> > due to
run checkpatch.pl to check code style for applied patches.
Do not check spelling now. Will check it only on merge requests.
Signed-off-by: Maxim Uvarov
---
v2: travis generates buggy $TRAVIS_COMMIT_RANGE env variable on git push.
That is situation when commit exist
https://travis-ci.org/mike-holmes-linaro/odp/jobs/199704885
I had trouble running this Maxim, I have not dug into why
0.00s$ echo $TRAVIS_COMMIT_RANGE
1c409144d6e6...9329b4025f3d
The command "echo $TRAVIS_COMMIT_RANGE" exited with 0.
0.01s$ ODP_PACHES=`echo $TRAVIS_COMMIT_RANGE | sed
For this series:
Reviewed-and-tested-by: Bill Fischofer
On Wed, Feb 8, 2017 at 6:19 AM, Petri Savolainen
wrote:
> Various fixes and development of packet related tests done during packet
> inline
> implementation. Better print outs, etc
As we discussed this morning, the LNG Design Summit to be held at
BUD17 will span 5 or possibly 6 slots from Monday through Wednesday.
Currently schedule looks like:
Session 1 - Monday 14:30 - 15:55 (1 hour 25 minutes)
Session 2 - Monday 16:10 - 17:05 (55 minutes)
Session 3 - Tuesday 10:00 -
On 01/30/17 13:32, Joe Savage wrote:
> Add an example application implementing lock-free IPv4 fragmentation
> and reassembly functionality using ODP's packet "concat" and "split".
>
> Signed-off-by: Joe Savage
> ---
> (This code contribution is provided under the terms of
I'm on 1.8.11 (standard with Ubuntu 16.10)
On Wed, Feb 8, 2017 at 6:53 AM, Maxim Uvarov wrote:
> On 02/08/17 15:48, Bill Fischofer wrote:
>> Those fields are documented but use the @private tag. Perhaps that
>> isn't supported in older versions of doxygen?
>>
>
> mine
On 02/08/17 14:03, Joe Savage wrote:
>> we use autotools and check to add or not add cflags or ldflags should be
>> there.
>> I think in that case it's better to make such check in that file:
>> ./platform/linux-generic/m4/configure.m4
>>
>> So check should be code compilation with libatomic and
On 02/08/17 15:48, Bill Fischofer wrote:
> Those fields are documented but use the @private tag. Perhaps that
> isn't supported in older versions of doxygen?
>
mine is:
doxygen --version
1.8.6
DEPENDENCIES says 1.8.8 is supported version.
which version do you use?
> On Wed, Feb 8, 2017 at
Those fields are documented but use the @private tag. Perhaps that
isn't supported in older versions of doxygen?
On Wed, Feb 8, 2017 at 1:46 AM, Maxim Uvarov wrote:
> just:
>
> ./configure
> make doxygen-doc
>
> On 8 February 2017 at 01:36, Bill Fischofer
Knowing the reason for suite init function failure helps in
debugging.
Signed-off-by: Petri Savolainen
---
test/common_plat/validation/api/packet/packet.c | 23 ++-
1 file changed, 18 insertions(+), 5 deletions(-)
diff --git
During 'make check' run generator on up to four cpus. Generator on
all cpus was an overkill compared to l2fwd on two cpus. Generator
and l2fwd still share cpus (on odp-linux) as cpumask_default_worker()
returns the same cpumask for both.
Signed-off-by: Petri Savolainen
User may select number of worker threads (-w) or cpumask (-c)
(but not both) to limit number of worker thread.
Increased pool size since many threads empty the pool easily and
result some threads to give up (on the first time those see pool
empty).
Added EXAMPLE_ERR() before abort() to make
Print out pktio driver name in start up. Driver name (e.g. dpdk or
netmap) helps in checking that correct pktio device started.
Signed-off-by: Petri Savolainen
---
test/common_plat/performance/odp_l2fwd.c | 10 --
1 file changed, 8 insertions(+), 2
Since l2fwd can assume that all packets are Ethernet. Also
odp_packet_data() points to the first byte of the frame directly
after packet input.
Signed-off-by: Petri Savolainen
---
test/common_plat/performance/odp_l2fwd.c | 12 +---
1 file changed, 5
Various fixes and development of packet related tests done during packet inline
implementation. Better print outs, etc make it easier to debug packet
implementation bugs.
Petri Savolainen (5):
test: l2fwd: add pktio driver print out
validation: packet: print reason for suite init failure
Added first inlined packet functions. Functions are allways used
inlined within the implementation. Applications see inlined
versions only when ABI compatibility is disabled.
Signed-off-by: Petri Savolainen
---
platform/linux-generic/Makefile.am |
This define is visible to applications through API headers. Use _ODP prefix
to avoid name space collision. Also "inline" describes better the purpose
of the macro than "static".
Signed-off-by: Petri Savolainen
---
.../include/odp/api/plat/atomic_inlines.h |
Added first inlined packet flag functions. Functions are allways used
inlined within the implementation. Applications see inlined
versions only when ABI compatibility is disabled.
Signed-off-by: Petri Savolainen
---
platform/linux-generic/Makefile.am
Changed packet handle (odp_packet_t) to store pointer to the packet
header, instead a bit field structure of pool and index. Buffer and
event handles are not changed.
Pointer as handle improves packet accessor function performance and
enable accessor function inlining as mapping to pool/index is
First set of packet and packet flag functions are inlined. Inline functions are
used (can be used) internally always, but are exposed through API only when
builing for non-ABI compatibility.
Packet handle value is changed from index(es) to a packet header pointer.
Packet/buffer header structure
Enable ticketlock function inlining in applications when
not building for ABI compatibility.
Signed-off-by: Petri Savolainen
---
platform/linux-generic/Makefile.am | 1 +
.../include/odp/api/plat/ticketlock_inlines.h | 19 +++-
> we use autotools and check to add or not add cflags or ldflags should be
> there.
> I think in that case it's better to make such check in that file:
> ./platform/linux-generic/m4/configure.m4
>
> So check should be code compilation with libatomic and without and set or
> not set it in ldflags.
On 8 February 2017 at 12:36, Joe Savage wrote:
> > No, only which I wrote before. Main comment is that we need it compile
> in our CI env,
> > so you need to add atomic library as Ola wrote. And few places where
> variables need go
> > on top.
>
> Okay then. There were some
> No, only which I wrote before. Main comment is that we need it compile in our
> CI env,
> so you need to add atomic library as Ola wrote. And few places where
> variables need go
> on top.
Okay then. There were some unresolved issues in the email to which I replied
to prompt you, but if you
On 8 February 2017 at 12:15, Joe Savage wrote:
> Any further thoughts on these issues, Maxim? I'm keen to get this patch
> merged already!
>
>
No, only which I wrote before. Main comment is that we need it compile in
our CI env,
so you need to add atomic library as Ola
Any further thoughts on these issues, Maxim? I'm keen to get this patch
merged already!
> > > I raised question about coding style question on today’s arch call
> > > discussion. And agreement was:
> > >
> > > 1. variables are on top. (actually we discussed that but looks like
> > > forget to
28 matches
Mail list logo