Okay. I will incorporate these changes in the next version.
Regards,
Bala
On 28 January 2016 at 20:31, Savolainen, Petri (Nokia - FI/Espoo)
wrote:
>
> This is otherwise OK (and exactly the feature I was missing in the call
> yesterday), but I'm wondering the
Merged,
Maxim.
On 01/26/2016 14:38, Maxim Uvarov wrote:
Avoid seg. fault if invalid handle provided to queue destroy.
Reviewed-by: Zoltan Kiss
Signed-off-by: Maxim Uvarov
---
platform/linux-generic/odp_queue.c | 3 +++
1 file changed, 3
https://bugs.linaro.org/show_bug.cgi?id=1881
--- Comment #2 from Hongbo Zhang ---
patches sent out:
https://patches.linaro.org/60241/
https://patches.linaro.org/60242/
https://patches.linaro.org/60243/
https://patches.linaro.org/60244/
--
You are receiving this mail
On 29 January 2016 at 17:10, Savolainen, Petri (Nokia - FI/Espoo)
wrote:
>
>
>> -Original Message-
>> From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT
>> hongbo.zh...@linaro.org
>> Sent: Friday, January 29, 2016 10:50 AM
>> To:
Memory compare is a commonly used C library function on
data plane applications. This enables using HW offload
(e.g. vector unit) for compare operations.
Signed-off-by: Petri Savolainen
---
include/odp/api/std_clib.h| 18 ++
Added test for memory compare.
Signed-off-by: Petri Savolainen
---
test/validation/std_clib/std_clib.c | 38 +
1 file changed, 38 insertions(+)
diff --git a/test/validation/std_clib/std_clib.c
If two applications share a link (pktio), both can send up to MTU sized frames
… and if one of them wants send less than MTU sized frames, it’s free to do
that (without forcing the other app to the same limit). From ODP API point of
view, MTU is the limit of the local transmit buffer. Whereas
Add helper function dpdk_setup_port() for configuring
DPDK ports.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/pktio/dpdk.c | 62 -
1 file changed, 41 insertions(+), 21
ixgbe_pmd has a minimum supported RX burst size of 4. If
less than 4 packets are requested with
odp_pktio_recv_queue(), rte_eth_rx_burst() is called with
nb_pkts=4 and the possibly received extra packets are cached
for the next dpdk_recv_queue() call to use.
Reviewed-by: Petri Savolainen
Add locking support to dpdk_recv_queue() and
dpdk_send_queue().
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/include/odp_packet_dpdk.h | 5 +
platform/linux-generic/pktio/dpdk.c | 19
Synchronizers suite was split into atomic, lock and barrier
suites.
Signed-off-by: Petri Savolainen
---
test/validation/atomic/atomic.c | 16
test/validation/atomic/atomic.h | 12 ++--
test/validation/barrier/barrier.h | 4 ++--
Highlight that application must ensure that there's no
race while calling init functions. Also lock free call
will never set op.init flag, since init call will not need
to implement locking or other (lock free) synchronization.
Signed-off-by: Petri Savolainen
---
From: Hongbo Zhang
Currently all ODP_CACHE_LINE_SIZE macros for different architectures are
held in one header file, they should be moved to their own arch file.
This patch moves ODP_CACHE_LINE_SIZE for MIPS.
Signed-off-by: Hongbo Zhang
---
From: Hongbo Zhang
Currently all ODP_CACHE_LINE_SIZE macros for different architectures are
held in one header file, they should be moved to their own arch file.
This patch moves ODP_CACHE_LINE_SIZE for x86.
Signed-off-by: Hongbo Zhang
---
> -Original Message-
> From: EXT Maxim Uvarov [mailto:maxim.uva...@linaro.org]
> Sent: Friday, January 29, 2016 1:15 PM
> To: Elo, Matias (Nokia - FI/Espoo) ; lng-
> o...@lists.linaro.org
> Subject: Re: [lng-odp] [API-NEXT PATCH 00/11] DPDK pktio implementation
>
>
From: Hongbo Zhang
v1-v2 change:
Add macro __OCTEON for patch 2.
v1 notes:
This is for bug https://bugs.linaro.org/show_bug.cgi?id=1881
This is on top of latest api-next branch because the latest
platform/linux-generic/arch/ directory is needed, so I use "API-NEXT"
From: Hongbo Zhang
The arch/arm/odp_sysinfo_parse.c is currently a symlink to
arch/linux/odp_sysinfo_parse.c, but in fact there should be defferences
between them.
A separated real arch/arm/odp_sysinfo_parse.c is created for ARM, and
the model_str is set to a general
From: Hongbo Zhang
Currently there are two systemcpu() functions, one is for some specific
platforms and the other is for default dummy, but most of the contents
are same except for sysinfo->cache_line_size: one is true data from
calling systemcpu_cache_line_size() and
From: Hongbo Zhang
In the systemcpu() function, odp_global_data.system_info.huge_page_size
should be sysinfo->huge_page_size instead, because when systemcpu() is
called, the _global_data.system_info is passed as parameter, we should
operate the parameter instead of the
> -Original Message-
> From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT
> Zoltan Kiss
> Sent: Friday, January 29, 2016 2:13 PM
> To: Elo, Matias (Nokia - FI/Espoo); EXT Ola Liljedahl
> Cc: lng-odp@lists.linaro.org
> Subject: Re: [lng-odp] [API-NEXT PATCH 00/11]
On 01/28/2016 17:24, Elo, Matias (Nokia - FI/Espoo) wrote:
-Original Message-
From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT Maxim
Uvarov
Sent: Thursday, January 28, 2016 3:10 PM
To: lng-odp@lists.linaro.org
Subject: Re: [lng-odp] [API-NEXT PATCH 00/11] DPDK
Add dpdk_mtu_get(), dpdk_promisc_mode_set(),
dpdk_promisc_mode_get(), and dpdk_capability() functions.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/include/odp_packet_dpdk.h | 1 +
Implement odp_pktio_link_status() for dpdk pktio.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/pktio/dpdk.c | 13 -
1 file changed, 12 insertions(+), 1 deletion(-)
diff --git
Initial implementation of the DPDK pktio type.
Initialization code copied from the odp-dpdk branch.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/include/odp_internal.h| 5 +
Implement odp_pktio_input_queues_config() function and add
helper rss_conf_to_hash_proto() for converting
odp_pktin_hash_proto_t to struct rte_eth_rss_conf.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
Free/close open resources in odp_pktio_close().
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/include/odp_packet_dpdk.h | 1 +
platform/linux-generic/pktio/dpdk.c | 19 ++-
2
From: Hongbo Zhang
This patch set does some architecture clean-ups for linux-generic
sysinfo codes.
This patch set is based on my previous patch set
"separate ODP_CACHE_LINE_SIZE to arch files", so "API-NEXT" is uesed
although no API is touched.
Hongbo Zhang (4):
From: Hongbo Zhang
The dummy data of cpu_hz_max and model_str are used when platform is
unknown or data cannot be acquired, but these variables should be set
in function odp_cpuinfo_parser() instead of the systemcpu() which should
cover only the cpu_count, huge_page_size
On 29 January 2016 at 05:16, Ivan Khoronzhuk
wrote:
>
>
> On 29.01.16 00:54, Bill Fischofer wrote:
>
>> This is how you implement timers in HW as well.
>>
> A separate HW block operates a scan loop that constantly searches for
> timers to expire and creates events for
Add initial support for building ODP with DPDK pktio.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
DEPENDENCIES | 54
platform/linux-generic/Makefile.am | 2 +
Implement odp_pktio_output_queues_config() function.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/pktio/dpdk.c | 12 +++-
1 file changed, 11 insertions(+), 1 deletion(-)
diff --git
Implement odp_pktio_in_queues(), odp_pktio_pktin_queues(),
and odp_pktio_pktout_queues() functions.
Reviewed-by: Petri Savolainen
Signed-off-by: Matias Elo
---
platform/linux-generic/pktio/dpdk.c | 48 ++---
1
Test atomic_op bit fields and odp_atomic_lock_free_u64().
Signed-off-by: Petri Savolainen
---
test/validation/atomic/atomic.c | 107
test/validation/atomic/atomic.h | 1 +
2 files changed, 108 insertions(+)
diff --git
From: Hongbo Zhang
Currently all ODP_CACHE_LINE_SIZE macros for different architectures are
held in one header file, they should be moved to their own arch file.
This patch moves ODP_CACHE_LINE_SIZE for PowerPC.
The previous arch/powerpc/odp/cpu_arch.h was a symlink to
From: Hongbo Zhang
Currently all ODP_CACHE_LINE_SIZE macros for different architectures are
held in one header file, they should be moved to their own arch file.
And in the legacy codes there was no ARM architecture directory, so this
patch create it, the odp_cpu_arch.c
V2:
- Check the number of mbuf segments in mbuf_to_pkt() (Zoltan Kiss)
- Copy DPDK RSS hash to ODP packet header in mbuf_to_pkt() (Zoltan Kiss)
- Compare packet length to MTU value in pkt_to_mbuf() and drop too long packets
This patch set implements new DPDK pktio type, which operates in the same
On 29/01/16 08:19, Elo, Matias (Nokia - FI/Espoo) wrote:
*From:*EXT Ola Liljedahl [mailto:ola.liljed...@linaro.org]
*Sent:* Thursday, January 28, 2016 4:39 PM
*To:* Elo, Matias (Nokia - FI/Espoo)
*Cc:* EXT Zoltan Kiss ; lng-odp@lists.linaro.org
Merged,
api change reviewed by Petri,
other change for a long time in the list, all my comments fixed.
thank you,
Maxim.
On 01/27/2016 11:56, hongbo.zh...@linaro.org wrote:
From: Hongbo Zhang
v7->v8 changes:
- use file name odp_cpu_arch.c in this patch set
Merged, both patches.
Maxim.
On 01/27/2016 20:36, Zoltan Kiss wrote:
As per-thread caches might retain some elements, no particular thread
should assume that a certain amount of elements are available at any
time.
Also, to make the high watermark test reliable we should avoid releasing
timers.
From: Hongbo Zhang
This is for bug https://bugs.linaro.org/show_bug.cgi?id=1881
This is on top of latest api-next branch because the latest
platform/linux-generic/arch/ directory is needed, so I use "API-NEXT" tag
There should be some check patch warnings for patch 4,
> -Original Message-
> From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT
> hongbo.zh...@linaro.org
> Sent: Friday, January 29, 2016 10:50 AM
> To: lng-odp@lists.linaro.org
> Subject: [lng-odp] [PATCH API-NEXT 2/4] linux-generic: separate MIPS
> ODP_CACHE_LINE_SIZE
From: Hongbo Zhang
Currently all ODP_CACHE_LINE_SIZE macros for different architectures are
held in one header file, they should be moved to their own arch file.
And in the legacy codes there was no ARM architecture directory, so this
patch create it, the odp_cpu_arch.c
On 28.01.16 20:31, Maxim Uvarov wrote:
I have some thoughts and questions about timer implementation in linux-generic.
Current implementation:
sigev.sigev_notify = SIGEV_THREAD;
sigev.sigev_notify_function = timer_notify;
sigev.sigev_value.sival_ptr = tp;
On 29 January 2016 at 15:40, Hongbo Zhang wrote:
> On 29 January 2016 at 17:10, Savolainen, Petri (Nokia - FI/Espoo)
> wrote:
>>
>>
>>> -Original Message-
>>> From: lng-odp [mailto:lng-odp-boun...@lists.linaro.org] On Behalf Of EXT
>>>
I have considered unsched also, but really don’t like that since it would
define the queue type by what it is not (instead of what it is). It would be
also problematic term, if we’d need to define a third type in the future (e.g.
sorted queue, etc).
This queue type provides the basic (or
From: Hongbo Zhang
Currently all ODP_CACHE_LINE_SIZE macros for different architectures are
held in one header file, they should be moved to their own arch file.
This patch moves ODP_CACHE_LINE_SIZE for x86.
Signed-off-by: Hongbo Zhang
---
From: Hongbo Zhang
Currently all ODP_CACHE_LINE_SIZE macros for different architectures are
held in one header file, they should be moved to their own arch file.
This patch moves ODP_CACHE_LINE_SIZE for MIPS.
Signed-off-by: Hongbo Zhang
---
On 28.01.16 20:31, Maxim Uvarov wrote:
I have some thoughts and questions about timer implementation in linux-generic.
Current implementation:
sigev.sigev_notify = SIGEV_THREAD;
sigev.sigev_notify_function = timer_notify;
sigev.sigev_value.sival_ptr = tp;
From: EXT Ola Liljedahl [mailto:ola.liljed...@linaro.org]
Sent: Thursday, January 28, 2016 4:39 PM
To: Elo, Matias (Nokia - FI/Espoo)
Cc: EXT Zoltan Kiss ; lng-odp@lists.linaro.org
Subject: Re: [lng-odp] [API-NEXT PATCH 00/11] DPDK pktio
On 29.01.16 00:54, Bill Fischofer wrote:
This is how you implement timers in HW as well.
A separate HW block operates a scan loop that constantly searches for timers to
expire and creates events for those who do.
The rest of the system operates undisturbed. For a SW analog in manycore
On 29/01/16 02:44, Bill Fischofer wrote:
Add odp_cpu_socket_id() and odp_cpu_socket_id_cpu() routines
Signed-off-by: Bill Fischofer
---
include/odp/api/cpu.h | 23 +++
1 file changed, 23 insertions(+)
diff --git a/include/odp/api/cpu.h
On 29/01/16 08:26, Savolainen, Petri (Nokia - FI/Espoo) wrote:
If two applications share a link (pktio),
A diverging question, but: how does that work? Who calls
odp_init_global()? Who calls pktio_open? And how do they share the same
pktio?
both can send up to MTU sized
frames … and
Thanks for the further explanation. Yes, UNSCHED implies a dichotomy that
may be limiting. I guess the real distinction is that certain queues are
managed by ODP while others are managed by the application itself
(alternately, queues are either system managed, with various ODP-defined
On 28 January 2016 at 10:05, Bill Fischofer
wrote:
> Signed-off-by: Bill Fischofer
>
Reviewed-by Mike Holmes
> ---
> doc/users-guide/users-guide.adoc | 121
> +++
> 1 file
You can always pretend the MTU is smaller than it actually is and do
segmenting yourself, however this defeats the purpose of Large Segment
Offload (LSO), which is a key HW offload capability of most modern I/O
interfaces. With LSO the application can send arbitrary sized packets (up
to at least
We'll be discussing this RFC on Monday's ARCH call. I'm certainly open to
adding this if it's deemed useful. I do note that DPDK defines SOCK_ID_ANY
to be -1 so I'd like to minimize translation overhead for odp-dpdk in this
area as well.
On Fri, Jan 29, 2016 at 10:27 AM, Zoltan Kiss
56 matches
Mail list logo