[lng-odp] Traffic Manager

Hi all, I'm currently moving odp-mppa to monarch_lts (we ware based on 1.6 until now). I was wondering if the implementation MUST support the TM API. The traffic_mngr.h is a bit hazy on that point. It looks to me as if odp_tm_capabilities returns 0, it announces that the TM API is not available

[lng-odp] [PATCH 0/4] Support non glibc in examples

Some common examples use glibc/Linux specific syscall. This patch series add check in configure.ac for these call and disables them if not present. You'll notice some subblocks have been added. They are there to avoid compilation warnings/errors about unused variables when these calls are not a

[lng-odp] [PATCH 2/4] performance: crypto: only call getrusage if available

Signed-off-by: Nicolas Morey-Chaisemartin --- test/common_plat/performance/odp_crypto.c | 5 + 1 file changed, 5 insertions(+) diff --git a/test/common_plat/performance/odp_crypto.c b/test/common_plat/performance/odp_crypto.c index 49a9f4b..5d74072 100644 --- a/test/common_plat/performance/

[lng-odp] [PATCH 1/4] configure: check for non standard function availability

Add check for sync, getrusage, getrlimit, setrlimit, sigaction, sched_setaffinity Signed-off-by: Nicolas Morey-Chaisemartin --- configure.ac | 5 - 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/configure.ac b/configure.ac index 3e89b0a..6e5125e 100644 --- a/configure.ac +++ b

[lng-odp] [PATCH 3/4] example: traffic_mgmt: only call glibc specific functions if available

Signed-off-by: Nicolas Morey-Chaisemartin --- example/traffic_mgmt/odp_traffic_mgmt.c | 38 + 1 file changed, 29 insertions(+), 9 deletions(-) diff --git a/example/traffic_mgmt/odp_traffic_mgmt.c b/example/traffic_mgmt/odp_traffic_mgmt.c index c4f5356..55add9d 10

[lng-odp] [PATCH 4/4] example: hello: only call sched_setaffinity if available

Signed-off-by: Nicolas Morey-Chaisemartin --- example/hello/odp_hello.c | 21 + 1 file changed, 13 insertions(+), 8 deletions(-) diff --git a/example/hello/odp_hello.c b/example/hello/odp_hello.c index 6d114ee..9cb3498 100644 --- a/example/hello/odp_hello.c +++ b/example/hell

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

Hi Bill, could you clarify if this "device" is related to devicetree as defined by Linaro (https://github.com/devicetree-org/devicetree-specification) or the devices that are in the scope of Christophe Millard, or some other device concept ? FF On 6 December 2016 at 04:43, Bill Fischofer wrot

Re: [lng-odp] [PATCH] linux-generic: move tm system barrier to tm group

Can you please elaborate on the use-case for this change? Regards, Bala On 2 December 2016 at 13:41, wrote: > From: Xuelin Shi > > since tm thread is handling tm group, move the thread based > barrier to tm group. otherwise, packet cannot get into the > second tm system in the same group. > >

Re: [lng-odp] Traffic Manager

On 6 December 2016 at 03:59, Nicolas Morey-Chaisemartin wrote: > Hi all, > > > I'm currently moving odp-mppa to monarch_lts (we ware based on 1.6 until > now). > I was wondering if the implementation MUST support the TM API. > ODP has no intentionally optional APIs becasue that impedes portabili

Re: [lng-odp] Traffic Manager

Le 12/06/2016 à 01:27 PM, Mike Holmes a écrit : > > > On 6 December 2016 at 03:59, Nicolas Morey-Chaisemartin > wrote: > > Hi all, > > > I'm currently moving odp-mppa to monarch_lts (we ware based on 1.6 until > now). > I was wondering if the implementation

Re: [lng-odp] [API-NEXT PATCH v3] api: ipsec: added IPSEC API

Hi Petri, 1. Will it worth it to add information (like relevant direction inbound/outbound or sync/async) in the name of some SA parameters? typedef struct odp_ipsec_sa_param_t { ………. /** Parameters for tunnel mode */ odp_ipsec_tunnel_param_t out_tunnel; /** F

Re: [lng-odp] [PATCH] linux-generic: move tm system barrier to tm group

Hi Bala, For each pktio, I'm trying to create one odp_tm_t and also one tm thread. The TM thread is bound to an odp_tm_group_t , that means I will create odp_tm_group_t for each pktio. Currently, the group creation only available for big system with more than 24 cores. So I also reduce this numb

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

On Tue, Dec 6, 2016 at 5:03 AM, Francois Ozog wrote: > Hi Bill, > > could you clarify if this "device" is related to devicetree as defined by > Linaro (https://github.com/devicetree-org/devicetree-specification) or > the devices that are in the scope of Christophe Millard, or some other > device

[lng-odp] [API-NEXT PATCHv3 2/5] linux-gen: adding enum, devio and driver registration interface (stub)

The linux implementation for the enumerator class registration function, enumerator instance registration function, devio and driver registration functions (stub) Signed-off-by: Christophe Milard --- include/odp_drv.h | 1 + platform/linux-generic/Makefile.am

[lng-odp] [API-NEXT PATCHv3 1/5] drv: adding driver registration interface (stub)

The enumerator class, enumerator instance, devio and driver registration functions prototypes (and a draft of their parameters) are defined, the goal being to define the registration framework only. Signed-off-by: Christophe Milard --- include/odp/drv/spec/driver.h | 311

[lng-odp] [API-NEXT PATCHv3 0/5] driver initialisation framework

Since V2: -function odp_load_driver removed. replaced by config file. (Petri, FF) -configuration file "odp.conf" added. Configuration file is: 1) as specified in env variable ODP_SYSCONFIG_FILE (which can be "none"). 2) ./odp.conf 3) $(prefix)/etc/odp.conf -test removed: will be sent

[lng-odp] [API-NEXT PATCHv3 3/5] linux-gen: init: adding configuration file parsing

The parsing of the odp.conf configuration file is added. The file is searched first in the local directory (.) and then is the $prefix/etc directory. This requires libconfig (sudo apt-get install libconfig-dev) Signed-off-by: Christophe Milard --- DEPENDENCIES |

[lng-odp] [API-NEXT PATCHv3 4/5] test: preventing odp.conf loading for tests

The tests should not be affected by any system or user ODP configuration file. The ODP_SYSCONFIG_FILE environment variables is therefore set to "none" in TESTS_ENVIRONMENT. Tests which needs specific a configuration file will have to overwrite this setting. Note that tests ran manually (not using m

[lng-odp] [API-NEXT PATCHv3 5/5] linux-gen: drv_drivers: loading modules from config file

The shared objects listed in the ODP configuration files are loaded at init time. The odp configuration file list the shared objects to be loaded as shown in the following example: module = { modules = ["enumerator1.so", "driver1.so"]; }; Signed-off-by: Christophe Milard --- configure.ac

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

On Tue, Dec 6, 2016 at 7:23 AM, Bill Fischofer wrote: > > > On Tue, Dec 6, 2016 at 5:03 AM, Francois Ozog > wrote: > >> Hi Bill, >> >> could you clarify if this "device" is related to devicetree as defined by >> Linaro (https://github.com/devicetree-org/devicetree-specification) or >> the device

Re: [lng-odp] [PATCHv3] platform: linux-generic: reading cpu affinity from cpuset

On 2 December 2016 at 04:20, Yi He wrote: > The original code reads whole system cpuset, and provides no command line > parameters to specify cpuset resources while launching ODP application. > Thus can be an obstacle for the coexisting of multiple ODP applications. > > New code implies that ODP

Re: [lng-odp] [PATCH] linux-generic: move tm system barrier to tm group

On 6 December 2016 at 18:30, Forrest Shi wrote: > Hi Bala, > > For each pktio, I'm trying to create one odp_tm_t and also one tm thread. > > The TM thread is bound to an odp_tm_group_t , that means I will create > odp_tm_group_t for each pktio. > > Currently, the group creation only available for

[lng-odp] [PATCHv4] platform: linux-generic: reading cpu affinity from cpuset

With this new proposal cpu affinity is read correctly especially when using cgroups otherwise wrong cpu mask is set. Fixes bug: https://bugs.linaro.org/show_bug.cgi?id=2472 Signed-off-by: Balakrishna Garapati --- since v3: Renaming function call platform/linux-generic/odp_cpumask.c | 73

Re: [lng-odp] [API-NEXT PATCH v4 0/5] new ordered queue implementation

Merged, it took some some to understand odp-check fails. It looks like there is hang of odp_generator process in background which is fixed with other patch by Christophe. Maxim. On 12/02/16 18:38, Bill Fischofer wrote: > For this series: > > Reviewed-and-tested-by: Bill Fischofer > > On Fri,

[lng-odp] [PATCH] configure: add option to enable/disable helper debug prints

As a side effect, debug print are turned off by default Signed-off-by: Nicolas Morey-Chaisemartin --- configure.ac | 14 ++ helper/Makefile.am | 3 ++- 2 files changed, 16 insertions(+), 1 deletion(-) diff --git a/configure.ac b/configure.ac index 3e89b0a..aa992fc 100644 ---

Re: [lng-odp] [API-NEXT PATCH] linux-gen: _fdserver: request sigterm if parent dies

Merged, Reviewed-by: Mike Holmes from email with API-NEXT tag. Maxim. On 11/30/16 15:48, Mike Holmes wrote: > Christophe > > Do you have a way I can create the problem and then prove to myself you > fixed it by applying this? > Can I CTRL C any of our apps ? > > Mike > > On 30 November 20

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

Devicetree has been formed by and for SoC vendors to represent what they have on their platforms (CPUs, Memory, interconnects, IOMMUs, ports, IRQ lines, firmware - http://free-electrons.com/pub/conferences/2013/elce/petazzoni-device-tree-dummies/petazzoni-device-tree-dummies.pdf). so it looks l

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

I have no problem deferring this if we want to have one of the SoC vendors drive a more comprehensive approach based on their capabilities which can then be generalized. BTW, your link doesn't resolve. Typo somewhere? On Tue, Dec 6, 2016 at 8:47 AM, Francois Ozog wrote: > Devicetree has been for

Re: [lng-odp] [PATCH] configure.ac: fix builds from raw git tar

Merged, Maxim. On 12/05/16 21:57, Mike Holmes wrote: > Ping, trivial fix for Monarch to actually work from the download page > > On 1 December 2016 at 14:28, Mike Holmes wrote: > >> I missed the prefix for MONARCH - re sending >> >> On 1 December 2016 at 14:21, Mike Holmes wrote: >> >>> Signe

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

OK. So let's defer the API but not the discussion. I think we need a "strategic architecture" call. FF On 6 December 2016 at 15:59, Bill Fischofer wrote: > I have no problem deferring this if we want to have one of the SoC vendors > drive a more comprehensive approach based on their capabilit

Re: [lng-odp] [PATCH] configure.ac: fix builds from raw git tar

Note, as Maxim caught there is another issue that results in a bad version string but all the test pass On 6 December 2016 at 10:03, Maxim Uvarov wrote: > Merged, > > Maxim. > > On 12/05/16 21:57, Mike Holmes wrote: >> Ping, trivial fix for Monarch to actually work from the download page >> >> On

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

For the link, I think it ate the following it. Here it is without decoration: http://free-electrons.com/pub/conferences/2013/elce/petazzoni-device-tree-dummies/petazzoni-device-tree-dummies.pdf On 6 December 2016 at 16:05, Francois Ozog wrote: > OK. So let's defer the API but not the disc

Re: [lng-odp] [MONARCH PATCH 1/2] validation: pktio: use PRIu32 to print uint32_t values

merged to master, also applied to monarch, both patches. Maxim. On 12/06/16 07:55, Nicolas Morey-Chaisemartin wrote: > I didn't submit there for master because I cannot build odp-mppa with master > yet (trying to merge monarch first). > And I'm pretty sure there'll be new patches like this to ap

Re: [lng-odp] [API-NEXT PATCH] linux-gen: _ishm: unlinking files asap for cleaner termination

On 12/02/16 12:42, Christophe Milard wrote: > _ishm now unlinks the created files as soon as possible, hence reducing > the chance to see left-overs, if ODP terminates abnormaly. > This does not provide 100% guaranty: if we are unlucky enough, > ODP may be killed between open() and unlink(). > Also

[lng-odp] [PATCH] example: traffic_mgmt: use PRIu32 instead of %u

Signed-off-by: Nicolas Morey-Chaisemartin --- This should be backported to monarch_lts too. However it requires an extra #include example/traffic_mgmt/odp_traffic_mgmt.c | 19 +++ 1 file changed, 11 insertions(+), 8 deletions(-) diff --git a/example/traffic_mgmt/odp_traffic_m

[lng-odp] [RFC API-NEXT PATCH] test: adding test for driver loading

Trying to fix a proper structure for driver tests in the hard world of autotools: Staticaly linked drivers fails with autotools as the linker would need the "--whole-archive" option which autotools cannot handle. This patch load an empty driver as a shared library (.so), and is the result of a long

[lng-odp] [API-NEXT PATCHv2] linux-gen: _ishm: unlinking files asap for cleaner termination

_ishm now unlinks the created files as soon as possible, hence reducing the chance to see left-overs, if ODP terminates abnormaly. This does not provide 100% guaranty: if we are unlucky enough, ODP may be killed between open() and unlink(). Also this method excludes exported files (flag _ODP_ISHM_E

Re: [lng-odp] [PATCH] configure: add option to enable/disable helper debug prints

On 6 December 2016 at 15:28, Nicolas Morey-Chaisemartin wrote: > As a side effect, debug print are turned off by default > > Signed-off-by: Nicolas Morey-Chaisemartin > --- > configure.ac | 14 ++ > helper/Makefile.am | 3 ++- > 2 files changed, 16 insertions(+), 1 deletion(-

Re: [lng-odp] [PATCH] configure: add option to enable/disable helper debug prints

Le 12/06/2016 à 05:35 PM, Christophe Milard a écrit : > On 6 December 2016 at 15:28, Nicolas Morey-Chaisemartin > wrote: > >> As a side effect, debug print are turned off by default >> >> Signed-off-by: Nicolas Morey-Chaisemartin >> --- >> configure.ac | 14 ++ >> helper/Make

[lng-odp] [PATCH v2 0/6] Make lidodp/libodp helper name platform specific

The lib suffix "linux" (libodp-linux, libodphelper-linux) is harcoded pretty much everywhere in the build system. The goal of this series is to allow each platform to define its library suffix so they don't have to touch the generic files but simply the platform specific files. v2: - Moved to m

[lng-odp] [PATCH v2 1/6] m4: add platform configurable library suffixes

This is the suffix to be used to compute the name of ODP and ODP helper libraries. platform can set a ODP_LIB_SUFFIX which is then exported as ODP_LIB_FLAVOR and ODP_LIB_FLAVOR_AM (for automake rules) Signed-off-by: Nicolas Morey-Chaisemartin --- configure.ac | 1 + p

[lng-odp] [PATCH v2 2/6] platform: move pkgconfig include to linux-generic

Signed-off-by: Nicolas Morey-Chaisemartin --- configure.ac | 2 -- platform/linux-generic/m4/configure.m4 | 4 +++- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/configure.ac b/configure.ac index 01d6448..0e1513f 100644 --- a/configure.ac +++ b/configure

[lng-odp] [PATCH v2 4/6] helper: use ODP_LIB_FLAVOR to generate libodp name

Signed-off-by: Nicolas Morey-Chaisemartin --- helper/Makefile.am | 6 +++--- helper/test/Makefile.am | 6 +++--- 2 files changed, 6 insertions(+), 6 deletions(-) diff --git a/helper/Makefile.am b/helper/Makefile.am index d09d900..c75db00 100644 --- a/helper/Makefile.am +++ b/helper/Makefile

[lng-odp] [PATCH v2 3/6] test: use ODP_LIB_FLAVOR to generate libodp name

Signed-off-by: Nicolas Morey-Chaisemartin --- test/Makefile.inc| 2 +- test/common_plat/validation/api/Makefile.inc | 2 +- test/linux-generic/Makefile.inc | 2 +- 3 files changed, 3 insertions(+), 3 deletions(-) diff --git a/test/Makefile.inc b/test/Make

[lng-odp] [PATCH v2 6/6] platform: use ODP_LIB_FLAVOR to generate libodp name

Signed-off-by: Nicolas Morey-Chaisemartin --- platform/Makefile.inc | 6 +++--- platform/linux-generic/Makefile.am | 4 ++-- 2 files changed, 5 insertions(+), 5 deletions(-) diff --git a/platform/Makefile.inc b/platform/Makefile.inc index 2722946..a37fd10 100644 --- a/platform/Makef

[lng-odp] [PATCH v2 5/6] example: use ODP_LIB_FLAVOR to generate libodp name

Signed-off-by: Nicolas Morey-Chaisemartin --- example/Makefile.inc | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/example/Makefile.inc b/example/Makefile.inc index 19d3994..62ead02 100644 --- a/example/Makefile.inc +++ b/example/Makefile.inc @@ -1,6 +1,6 @@ include $(top_src

Re: [lng-odp] [API-NEXT PATCHv3 1/3] api: dev: add device apis for numa support

Hi Bill, Do you have any slides/docs on Petri's suggestion? Thank you, Honnappa On 6 December 2016 at 09:10, Francois Ozog wrote: > For the link, I think it ate the following it. > Here it is without decoration: > http://free-electrons.com/pub/conferences/2013/elce/petazzoni-device-tree

[lng-odp] [Bug 2704] New: Builds TAR have no .scmversion

https://bugs.linaro.org/show_bug.cgi?id=2704 Bug ID: 2704 Summary: Builds TAR have no .scmversion Product: OpenDataPlane - linux- generic reference Version: v1.11.0.0 Hardware: Other OS: Linux Status: UNCONFIRMED

Re: [lng-odp] [APO-NEXT PATCHv6 2/3] doc: userguide: move crypto documentation to its own sub-document

On 5 December 2016 at 16:28, Bill Fischofer wrote: > Signed-off-by: Bill Fischofer Reviewed-by: Mike Holmes > --- > doc/users-guide/Makefile.am | 1 + > doc/users-guide/users-guide-crypto.adoc | 71 > doc/users-guide/users-guide.adoc| 72

Re: [lng-odp] [PATCH v2 1/6] m4: add platform configurable library suffixes

On 6 December 2016 at 12:05, Nicolas Morey-Chaisemartin wrote: > This is the suffix to be used to compute the name of ODP > and ODP helper libraries. > platform can set a ODP_LIB_SUFFIX which is then exported as > ODP_LIB_FLAVOR and ODP_LIB_FLAVOR_AM (for automake rules) I think this should be

Re: [lng-odp] [PATCH v2 1/6] m4: add platform configurable library suffixes

On 6 December 2016 at 15:25, Mike Holmes wrote: > On 6 December 2016 at 12:05, Nicolas Morey-Chaisemartin > wrote: >> This is the suffix to be used to compute the name of ODP >> and ODP helper libraries. >> platform can set a ODP_LIB_SUFFIX which is then exported as >> ODP_LIB_FLAVOR and ODP_LI

Re: [lng-odp] [PATCH v2 4/6] helper: use ODP_LIB_FLAVOR to generate libodp name

On 6 December 2016 at 12:08, Nicolas Morey-Chaisemartin wrote: > Signed-off-by: Nicolas Morey-Chaisemartin > --- > helper/Makefile.am | 6 +++--- > helper/test/Makefile.am | 6 +++--- > 2 files changed, 6 insertions(+), 6 deletions(-) > > diff --git a/helper/Makefile.am b/helper/Makefile.am

Re: [lng-odp] [PATCH v2 4/6] helper: use ODP_LIB_FLAVOR to generate libodp name

On 12/06/16 23:37, Mike Holmes wrote: > On 6 December 2016 at 12:08, Nicolas Morey-Chaisemartin > wrote: >> Signed-off-by: Nicolas Morey-Chaisemartin >> --- >> helper/Makefile.am | 6 +++--- >> helper/test/Makefile.am | 6 +++--- >> 2 files changed, 6 insertions(+), 6 deletions(-) >> >> dif

Re: [lng-odp] [API-NEXT PATCHv3 0/5] driver initialisation framework

After applying this series ./configure fails: ./configure checking for GCC atomic builtins... yes checking libconfig.h usability... no checking libconfig.h presence... no checking for libconfig.h... no checking for pkg-config... /usr/bin/pkg-config checking pkg-config is at least version 0.9.

Re: [lng-odp] [PATCHv4] platform: linux-generic: reading cpu affinity from cpuset

Reviewed-and-tested-by: Yi He On 6 December 2016 at 22:15, Balakrishna Garapati < balakrishna.garap...@linaro.org> wrote: > With this new proposal cpu affinity is read correctly especially > when using cgroups otherwise wrong cpu mask is set. > > Fixes bug: https://bugs.linaro.org/show_bug.cgi?

Re: [lng-odp] [PATCH] linux-generic: move tm system barrier to tm group

Hi Bala, I'm testing example/traffic_mgmt with 20Gbps connections, not see any issue here. Thanks, Forrest On 6 December 2016 at 21:43, Bala Manoharan wrote: > On 6 December 2016 at 18:30, Forrest Shi wrote: > > Hi Bala, > > > > For each pktio, I'm trying to create one odp_tm_t and also one t

Re: [lng-odp] [PATCH v2 4/6] helper: use ODP_LIB_FLAVOR to generate libodp name

Le 12/06/2016 à 10:17 PM, Maxim Uvarov a écrit : > On 12/06/16 23:37, Mike Holmes wrote: >> On 6 December 2016 at 12:08, Nicolas Morey-Chaisemartin >> wrote: >>> Signed-off-by: Nicolas Morey-Chaisemartin >>> --- >>> helper/Makefile.am | 6 +++--- >>> helper/test/Makefile.am | 6 +++--- >>>

Re: [lng-odp] [PATCH v2 1/6] m4: add platform configurable library suffixes

Le 12/06/2016 à 09:25 PM, Mike Holmes a écrit : > On 6 December 2016 at 12:05, Nicolas Morey-Chaisemartin > wrote: >> This is the suffix to be used to compute the name of ODP >> and ODP helper libraries. >> platform can set a ODP_LIB_SUFFIX which is then exported as >> ODP_LIB_FLAVOR and ODP_L

Re: [lng-odp] [PATCH v2 4/6] helper: use ODP_LIB_FLAVOR to generate libodp name

Le 12/06/2016 à 09:37 PM, Mike Holmes a écrit : > On 6 December 2016 at 12:08, Nicolas Morey-Chaisemartin > wrote: >> Signed-off-by: Nicolas Morey-Chaisemartin >> --- >> helper/Makefile.am | 6 +++--- >> helper/test/Makefile.am | 6 +++--- >> 2 files changed, 6 insertions(+), 6 deletions(

Re: [lng-odp] [PATCH v2 4/6] helper: use ODP_LIB_FLAVOR to generate libodp name

Nicolas, what do you do to solve coherency problem? I think that maybe we are missing some api in odp, like read/write barriers. Maxim. On 7 December 2016 at 09:54, Nicolas Morey-Chaisemartin wrote: > > > Le 12/06/2016 à 09:37 PM, Mike Holmes a écrit : > > On 6 December 2016 at 12:08, Nicolas M

Re: [lng-odp] [API-NEXT PATCHv3 0/5] driver initialisation framework

On 7 December 2016 at 01:19, Bill Fischofer wrote: > After applying this series ./configure fails: > > ./configure > > checking for GCC atomic builtins... yes > checking libconfig.h usability... no > checking libconfig.h presence... no > checking for libconfig.h... no > checking for pkg-conf

Re: [lng-odp] [PATCH v2 4/6] helper: use ODP_LIB_FLAVOR to generate libodp name

I usually force somme access to be uncached. A whole barrier would work, but the performance cost is way too big. Just as an example: static inline void odp_atomic_init_u64(odp_atomic_u64_t *atom, uint64_t val) { atom->v = val; #if __GCC_ATOMIC_LLONG_LOCK_FREE < 2 __atomic_clear(&atom->