Re: [lng-odp] [PATCH] Define odp_cpumask_get_bitmap to retrieve the bitmap defined by odp_cpumask_t.

Hi, Actually, this kind of API is not possible since it depends on POSIX type cpu_set_t. ODP API may depend only on C99 definitions, so that it remains operating system agnostic. CPU ID values are system specific. For example, CPU ID 0 is Linux CPU 0 (when running on Linux). helper/linux.h

[lng-odp] [PATCH] helper: linux: correct cpumask usage

Use Linux cpu_set_t instead of odp_cpumask_t when passing cpus to Linux system calls. Signed-off-by: Petri Savolainen --- helper/linux.c | 16 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/helper/linux.c b/helper/linux.c index

Re: [lng-odp] ODP_SHM_SW_ONLY shmem flag

On 11 February 2016 at 10:06, Savolainen, Petri (Nokia - FI/Espoo) < petri.savolai...@nokia.com> wrote: > SW only means that the memory is accessed only from CPUs. Memory need to > be mapped into CPU TLBs only, no need to map it into system MMU, IOMMU, > etc. > > > > SHM used to be passed into

Re: [lng-odp] [API-NEXT PATCH 1/4] helper: eth: added mac address parse

Merged, Maxim. On 02/10/16 16:16, Tilli, Juha-Matti (Nokia - FI/Espoo) wrote: Hi, After having a discussion with Petri, I now agree that the scanf implementation without strspn is probably the best one, as the command-line arguments may be concatenated together in one string in some use

Re: [lng-odp] ODP_SHM_PROC shmem flag

On 02/11/16 12:10, Savolainen, Petri (Nokia - FI/Espoo) wrote: -Original Message- From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT Maxim Uvarov Sent: Wednesday, February 10, 2016 8:17 PM To: lng-odp@lists.linaro.org Subject: Re: [lng-odp] ODP_SHM_PROC shmem flag

[lng-odp] [PATCH v2 2/2] helper: linux: remove unnecessary cpumask copy

Copies are redundant since cpumask APIs are defined with const specifier. Signed-off-by: Petri Savolainen --- helper/linux.c | 20 1 file changed, 8 insertions(+), 12 deletions(-) diff --git a/helper/linux.c b/helper/linux.c index

Re: [lng-odp] ODP_SHM_PROC shmem flag

> -Original Message- > From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT > Maxim Uvarov > Sent: Wednesday, February 10, 2016 8:17 PM > To: lng-odp@lists.linaro.org > Subject: Re: [lng-odp] ODP_SHM_PROC shmem flag > > On 02/10/2016 20:43, Christophe Milard wrote: > >

[lng-odp] [PATCH v2 1/2] helper: linux: correct cpumask usage

Use Linux cpu_set_t instead of odp_cpumask_t when passing cpus to Linux system calls. Signed-off-by: Petri Savolainen --- helper/linux.c | 16 1 file changed, 8 insertions(+), 8 deletions(-) diff --git a/helper/linux.c b/helper/linux.c index

Re: [lng-odp] ODP_SHM_SW_ONLY shmem flag

SW only means that the memory is accessed only from CPUs. Memory need to be mapped into CPU TLBs only, no need to map it into system MMU, IOMMU, etc. SHM used to be passed into pool creation (shm for packet pool vs. SW internal pool). -Petri From: lng-odp

Re: [lng-odp] ODP_SHM_PROC shmem flag

Hi, Petri: Do you know which interface the non-ODP process is supposed to be using? shall it use odp_shm_lookup() and odp_shm_addr() or -in linux case- shm_open() +mmap? The first alternative implies that each plarform must document the method to be used. The second method implies that non-ODP

Re: [lng-odp] [API-NEXT PATCHv6 0/4] Preparing for new interfaces

Merged, Maxim. On 02/11/16 17:21, Christophe Milard wrote: since v5: rebased. since v4: removed the extern "C" wrapping from odp.h since v3: odp.h (just including odp_api.h) added to ease transition period for apps. double include protection macro name fixed in odp_api.h since v2: removed

Re: [lng-odp] [API-NEXT PATCHv20 0/9] IPC (pktio)

Merged, Maxim. On 02/11/16 15:17, Savolainen, Petri (Nokia - FI/Espoo) wrote: -Original Message- From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT Maxim Uvarov Sent: Thursday, February 11, 2016 1:54 PM To: lng-odp-forward Subject: Re:

Re: [lng-odp] [API-NEXT PATCHv5 0/4] Preparing for new interfaces

Can you update patch to current api-next? Maxim. On 02/10/16 19:51, Bill Fischofer wrote: For this series: Reviewed-by: Bill Fischofer > On Wed, Feb 10, 2016 at 11:48 AM, Christophe Milard

Re: [lng-odp] building odp

Hello Thomas, which toolchain is there? tested with: http://releases.linaro.org/15.02/components/toolchain/binaries/aarch64-linux-gnu/gcc-linaro-4.9-2015.02-3-x86_64_aarch64-linux-gnu.tar.xz and odp compiles. Maxim. On 02/05/16 06:35, Thomas Gall wrote: I figured out my goof. Was "fixing"

Re: [lng-odp] [API-NEXT PATCHv20 0/9] IPC (pktio)

> -Original Message- > From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT > Maxim Uvarov > Sent: Thursday, February 11, 2016 1:54 PM > To: lng-odp-forward > Subject: Re: [lng-odp] [API-NEXT PATCHv20 0/9] IPC (pktio) > > If there is no

Re: [lng-odp] ODP_SHM_PROC shmem flag

> -Original Message- > From: EXT Maxim Uvarov [mailto:maxim.uva...@linaro.org] > Sent: Thursday, February 11, 2016 11:20 AM > To: Savolainen, Petri (Nokia - FI/Espoo) ; > lng-odp@lists.linaro.org > Subject: Re: [lng-odp] ODP_SHM_PROC shmem flag > > On

[lng-odp] [PATCH] validation: pktio: fix pktio_test_recv_queue() crash

pktio_test_recv_queue() would cause segfault if 'num_queue' > MAX_QUEUES. Signed-off-by: Matias Elo --- test/validation/pktio/pktio.c | 11 --- 1 file changed, 8 insertions(+), 3 deletions(-) diff --git a/test/validation/pktio/pktio.c

Re: [lng-odp] [API-NEXT PATCHv20 0/9] IPC (pktio)

On 02/11/16 15:17, Savolainen, Petri (Nokia - FI/Espoo) wrote: -Original Message- From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT Maxim Uvarov Sent: Thursday, February 11, 2016 1:54 PM To: lng-odp-forward Subject: Re: [lng-odp]

Re: [lng-odp] ODP_SHM_SW_ONLY shmem flag

ok. If that tells that the app will not even call any other ODP functions on this are, that make sense. I will try to document that. Thanks. Christophe On 11 February 2016 at 11:04, Savolainen, Petri (Nokia - FI/Espoo) < petri.savolai...@nokia.com> wrote: > > > > > *From:* EXT Christophe Milard

Re: [lng-odp] ODP_SHM_SW_ONLY shmem flag

From: EXT Christophe Milard [mailto:christophe.mil...@linaro.org] Sent: Thursday, February 11, 2016 11:13 AM To: Savolainen, Petri (Nokia - FI/Espoo) Cc: LNG ODP Mailman List Subject: Re: [lng-odp] ODP_SHM_SW_ONLY shmem flag On 11

Re: [lng-odp] [API-NEXT PATCHv20 0/9] IPC (pktio)

If there is no objection I merge that patches to api-next. Everybody agree? Maxim. On 02/10/16 16:08, Maxim Uvarov wrote: v20: fixed Stuart and Ivans comments. Plan is to include these patches to api-next branch first. Then define all use cases which we want to support (l2fwd,

[lng-odp] [API-NEXT PATCHv6 0/4] Preparing for new interfaces

since v5: rebased. since v4: removed the extern "C" wrapping from odp.h since v3: odp.h (just including odp_api.h) added to ease transition period for apps. double include protection macro name fixed in odp_api.h since v2: removed strange characters from docs since v1: odp.h renamed to

[lng-odp] [Bug 2025] odp/test/api_tests dir is not used and the code is stale

https://bugs.linaro.org/show_bug.cgi?id=2025 Mike Holmes changed: What|Removed |Added Status|UNCONFIRMED |CONFIRMED Ever

[lng-odp] [Bug 2033] New: odp_cpuinfo_parser() causes segfault on systems where cpu speed info is missing in sysinfo->model_str

https://bugs.linaro.org/show_bug.cgi?id=2033 Bug ID: 2033 Summary: odp_cpuinfo_parser() causes segfault on systems where cpu speed info is missing in sysinfo->model_str Product: OpenDataPlane - linux- generic reference Version: 1.7

[lng-odp] [API-NEXT PATCHv6 2/4] api: move include/odp.h to include/odp_api.h

odp.h actually describes the applicatiion interface of ODP and is logicaly moved to odp_api.h. Applications now includes . This simplifies the addition of other ODP interfaces (e.g. drivers may include odp_drv.h in the future.) Signed-off-by: Christophe Milard

[lng-odp] [API-NEXT PATCHv6 4/4] api: odp.h to ease transition period

The former odp.h (now called odp_api.h) is recreated here and just includes odp_api.h. This is to ease the transition period for ODP applications, but this file (odp.h) is meant to be removed in the future. ODP applications should include odp_api.h, but those including odp.h will keep working

[lng-odp] [API-NEXT PATCHv6 3/4] doc: descr of structure for new interfaces

updates of the documentation to reflect the new structure allowing new interface addition. Signed-off-by: Christophe Milard Reviewed-by: Bill Fischofer --- doc/implementers-guide/implementers-guide.adoc | 72 +++---

Re: [lng-odp] ODP_SHM_PROC shmem flag

OK. So my first understanding was correct. So I am back to my original question: through which interface should the non-ODP process reach the shared mem? - the ODP API? (which means that non ODP processes/thread could be linked with the ODP lib and however remain non ODP) - or the native OS (e.g.

Re: [lng-odp] [PATCH] linux-generic: netmap: improve single RX queue performance

Hi, I was the one that originally proposed this change. My original problem was that netmap_start() and netmap_close() run too slowly. Great that it was such a simple change! I didn't know that performance would be increased too, but it seems to be a good added bonus. However, when actually

Re: [lng-odp] ODP_SHM_PROC shmem flag

I don't understand here... My understanding is that shmem_reserve would always allocate SHARED memory, i.e. at least visible by all ODP processes (even with no flag). Are you saying, Maxim, that shmem_reserve() should allocate thread-local memory when no flag is set and that the ODP_SHM_PROC is

Re: [lng-odp] [PATCH] linux-generic: netmap: improve single RX queue performance

Hi, All it took was increasing the link timeout from 5 seconds to 10 seconds to get it to work. It is visibly faster to start and to stop now after this change, thus ensuring minimal user annoyance. I don't recall right now what perf was before this patch, but after this patch it's 8 Gbps for

[lng-odp] Kalray release their ODP enabled Processor

Congratulations Kalray! http://www.design-reuse.com/news/39210/kalray-high-speed-i-o-processor.html -- Mike Holmes Technical Manager - Linaro Networking Group Linaro.org *│ *Open source software for ARM SoCs "Work should be fun and collborative, the rest follows"

[lng-odp] [PATCH] linux-generic: netmap: improve single RX queue performance

Improve netmap pktio performance by using a single netmap descriptor if only one RX queue is configured. Also, makes netmap_start()/netmap_close() run faster in single queue cases. Signed-off-by: Matias Elo --- platform/linux-generic/pktio/netmap.c | 93