W dniu 16.10.2020 o 17:42, Honnappa Nagarahalli pisze:
>
>
>> W dniu 16.10.2020 o 14:43, Lukasz Wojciechowski pisze:
>>> Hi Honnappa,
>>>
>>> Thank you for your answer.
>>> In the current v7 version I followed your advise and used RELAXED memory
>> model.
>>> And it works without any issues. I g
Hi Honnappa,
W dniu 16.10.2020 o 07:13, Honnappa Nagarahalli pisze:
>
>> During quit_workers function distributor's main core processes some packets
>> to wake up pending worker cores so they can quit.
>> As quit_workers acts also as a cleanup procedure for next test case it should
>> also collec
Hi Honnappa,
W dniu 16.10.2020 o 07:12, Honnappa Nagarahalli pisze:
>
>
>> Sanity tests with mbuf alloc and shutdown tests assume that mbufs passed
>> to worker cores are freed in handlers.
>> Such packets should not be returned to the distributor's main core. The only
>> packets that should be r
Hi Honnappa,
W dniu 16.10.2020 o 07:13, Honnappa Nagarahalli pisze:
> Hi Lukasz,
> I see that in commit 8/16, the same code is changed again (updating the
> counters using the RELAXED memory order). It is better to pull the statistics
> changes from 8/16 into this commit.
I reordered patc
All of the former tests analyzed only statistics
of packets processed by all workers.
The new test verifies also if packets are processed
on workers as expected.
Every packets processed by the worker is marked
and analyzed after it is returned to distributor.
This test allows finding issues in mat
All suggested changes applied and published in v8.
W dniu 16.10.2020 o 01:47, Honnappa Nagarahalli pisze:
>
>
>> rte_distributor_return_pkt function which is run on worker cores must wait
>> for distributor core to clear handshake on retptr64 before using those
>> buffers. While the handshake is
In all distributor tests there is a chance that tests
will send packets to distributor with rte_distributor_process()
before workers are started and requested for packets.
This patch ensures that all packets are delivered to workers
by calling rte_distributor_process() in loop until number
of succ
rte_distributor_flush() is using total_outstanding()
function to calculate if it should still wait
for processing packets. However in burst mode
only backlog packets were counted.
This patch fixes that issue by counting also in flight
packets. There are also sum fixes to properly keep
count of in
Instead of making delays in test code and waiting
for worker hopefully to reach proper states,
synchronize worker shutdown test cases with spin lock
on atomic variable.
Fixes: c0de0eb82e40 ("distributor: switch over to new API")
Cc: david.h...@intel.com
Cc: sta...@dpdk.org
Signed-off-by: Lukasz W
The patch clears distributors returns buffer
in clear_returns() by setting start and count to 0.
Fixes: 775003ad2f96 ("distributor: add new burst-capable library")
Cc: david.h...@intel.com
Cc: sta...@dpdk.org
Signed-off-by: Lukasz Wojciechowski
Acked-by: David Hunt
---
lib/librte_distributor/r
Statistics of handled packets are cleared and read on main lcore,
while they are increased in workers handlers on different lcores.
Without synchronization occasionally showed invalid values.
This patch uses atomic mechanisms to synchronize.
Relaxed memory model is used.
Fixes: c3eabff124e6 ("dis
Sending number of packets equal to number of workers isn't enough
to stop all workers in burst version of distributor as more than
one packet can be matched and consumed by a single worker. This way
some of workers might not be awaken from rte_distributor_get_pkt().
This patch fixes it by sending
After introducing burst API there were some artefacts in the
API documentation from legacy single API.
Also the rte_distributor_poll_pkt() function return values
mismatched the implementation.
Fixes: c0de0eb82e40 ("distributor: switch over to new API")
Cc: david.h...@intel.com
Cc: sta...@dpdk.org
The sanity test with worker shutdown delegates all bufs
to be processed by a single lcore worker, then it freezes
one of the lcore workers and continues to send more bufs.
The freezed core shuts down first by calling
rte_distributor_return_pkt().
The test intention is to verify if packets assigned
Sanity tests with mbuf alloc and shutdown tests assume that
mbufs passed to worker cores are freed in handlers.
Such packets should not be returned to the distributor's main
core. The only packets that should be returned are the packets
send after completion of the tests in quit_workers function.
In the single legacy version of the distributor synchronization
requires continues exchange of buffers between distributor
and workers. Empty buffers are sent if only handshake
synchronization is required.
However calls to the rte_distributor_return_pkt()
with 0 buffers in single mode were ignored
rte_distributor_request_pkt and rte_distributor_get_pkt dereferenced
oldpkt parameter when in RTE_DIST_ALG_SINGLE even if number
of returned buffers from worker to distributor was 0.
This patch passes NULL to the legacy API when number of returned
buffers is 0. This allows passing NULL as oldpkt p
The burst version of distributor implementation was missing proper
handling of worker shutdown. A worker processing packets received
from distributor can call rte_distributor_return_pkt() function
informing distributor that it want no more packets. Further calls to
rte_distributor_request_pkt() or
Fix improper indexes while comparing tags.
In the find_match_scalar() function:
* j iterates over flow tags of following packets;
* w iterates over backlog or in flight tags positions.
Fixes: 775003ad2f96 ("distributor: add new burst-capable library")
Cc: david.h...@intel.com
Cc: sta...@dpdk.org
During quit_workers function distributor's main core processes
some packets to wake up pending worker cores so they can quit.
As quit_workers acts also as a cleanup procedure for next test
case it should also collect these packets returned by workers'
handlers, so the cyclic buffer with returned pa
Synchronization of data exchange between distributor and worker cores
is based on 2 handshakes: retptr64 for returning mbufs from workers
to distributor and bufptr64 for passing mbufs to workers.
Without proper order of verifying those 2 handshakes a deadlock may
occur. This can happen when worker
During review and verification of the patch created by Sarosh Arif:
"test_distributor: prevent memory leakages from the pool" I found out
that running distributor unit tests multiple times in a row causes fails.
So I investigated all the issues I found.
There are few synchronization issues that mi
rte_distributor_return_pkt function which is run on worker cores
must wait for distributor core to clear handshake on retptr64
before using those buffers. While the handshake is set distributor
core controls buffers and any operations on worker side might overwrite
buffers which are unread yet.
Sam
Thanks
Reviewed-by: Nicolas Chautru
> -Original Message-
> From: wangyunjian
> Sent: Thursday, October 15, 2020 6:46 AM
> To: dev@dpdk.org
> Cc: Chautru, Nicolas ; jerry.lili...@huawei.com;
> xudin...@huawei.com; Yunjian Wang ;
> sta...@dpdk.org
> Subject: [dpdk-dev] [PATCH v2] baseband/
On 10/16/2020 10:58 PM, Ferruh Yigit wrote:
On 10/16/2020 1:05 PM, Ferruh Yigit wrote:
On 10/13/2020 4:32 PM, Ferruh Yigit wrote:
On 10/13/2020 3:53 PM, Andrew Rybchenko wrote:
Use ENODEV as the error code if specified port ID is invalid.
Signed-off-by: Andrew Rybchenko
---
lib/librte_ethd
On 10/14/2020 8:59 AM, Gaëtan Rivet wrote:
On 14/10/20 09:25 +0300, Andrew Rybchenko wrote:
On 10/13/20 7:56 PM, Ferruh Yigit wrote:
Prefix static function and variables with 'eth_dev'.
For some 'rte_' prefix dropped, and for others 'eth_dev' added.
This is useful to differentiate public and s
On 10/16/2020 1:05 PM, Ferruh Yigit wrote:
On 10/13/2020 4:32 PM, Ferruh Yigit wrote:
On 10/13/2020 3:53 PM, Andrew Rybchenko wrote:
Use ENODEV as the error code if specified port ID is invalid.
Signed-off-by: Andrew Rybchenko
---
lib/librte_ethdev/rte_ethdev.c | 44
On Fri, Oct 16, 2020 at 8:07 AM Stephen Hemminger
wrote:
>
> On Fri, 16 Oct 2020 16:34:46 +0200
> Thomas Monjalon wrote:
>
> > 14/10/2020 10:43, Kinsella, Ray:
> > >
> > > On 14/10/2020 03:26, Ferruh Yigit wrote:
> > > > Queue stats will be removed from basic stats to xstats.
> > > >
> > > > It w
On 10/16/2020 3:32 PM, Thomas Monjalon wrote:
16/10/2020 14:16, Ferruh Yigit:
On 10/14/2020 3:26 AM, Ferruh Yigit wrote:
Queue stats are stored in 'struct rte_eth_stats' as array and array size
is defined by 'RTE_ETHDEV_QUEUE_STAT_CNTRS' compile time flag.
As a result of technical board discus
On 10/16/2020 8:22 PM, Ferruh Yigit wrote:
On 10/16/2020 5:44 PM, Viacheslav Ovsiienko wrote:
The DPDK datapath in the transmit direction is very flexible.
An application can build the multi-segment packet and manages
almost all data aspects - the memory pools where segments
are allocated from,
On 10/16/2020 5:44 PM, Viacheslav Ovsiienko wrote:
The DPDK datapath in the transmit direction is very flexible.
An application can build the multi-segment packet and manages
almost all data aspects - the memory pools where segments
are allocated from, the segment lengths, the memory attributes
l
On Fri, Oct 16, 2020 at 6:33 AM Thomas Monjalon wrote:
>
> The function rte_eth_dev_release_port() is partially resetting
> the struct rte_eth_dev. The drivers were completing this reset
> with more pointers set to NULL in the close or remove operations.
>
> More pointers are reset at ethdev level
On 10/15/2020 2:30 PM, Andrew Rybchenko wrote:
Very late attempt to close deprecation notice sent long time ago.
Together with Thomas's changes for dev_close it will do the job.
Ivan Ilchenko (11):
ethdev: change eth dev stop function to return int
test/event: check eth dev stop status
On 10/15/2020 2:30 PM, Andrew Rybchenko wrote:
From: Ivan Ilchenko
Change eth_dev_stop_t return value from void to int.
Make eth_dev_stop_t implementations across all drivers to return
negative errno values if case of error conditions.
Signed-off-by: Ivan Ilchenko
Signed-off-by: Andrew Rybche
On 10/16/2020 2:32 PM, Thomas Monjalon wrote:
Most of the patches from the v3 have been merged already.
The remaining 3 patches are reworked in v4:
- reset dev_started in some PMDs
- selectively reset pointers in rte_eth_dev_release_port()
- fix error handling
v5: change error code to E
Currently, users have to use external RCU mechanisms to free resources
when using lock free hash algorithm.
Integrate RCU QSBR process to make it easier for the applications to use
lock free algorithm.
Refer to RCU documentation to understand various aspects of
integrating RCU library into other
Replace rte_atomic APIs with C11 atomic APIs in
test_hash_readwrite_lf_perf.c
Signed-off-by: Dharmik Thakkar
Reviewed-by: Ruifeng Wang
---
app/test/test_hash_readwrite_lf_perf.c | 89 +++---
1 file changed, 36 insertions(+), 53 deletions(-)
diff --git a/app/test/test_hash_r
Integrate RCU QSBR to make it easier for the applications to use lock
free algorithm.
Resource reclamation implementation was split from the original
series, and has already been part of RCU library. Rework the series
to base hash integration on RCU reclamation APIs.
Refer 'Resource reclamation f
Add functional and performance tests for the integrated RCU QSBR.
Suggested-by: Honnappa Nagarahalli
Signed-off-by: Dharmik Thakkar
Reviewed-by: Ruifeng Wang
---
app/test/test_hash.c | 390 -
app/test/test_hash_readwrite_lf_perf.c | 171 ++-
2
> -Original Message-
> From: Thomas Monjalon
> Sent: Friday, October 16, 2020 18:14
> To: Slava Ovsiienko
> Cc: dev@dpdk.org; step...@networkplumber.org; ferruh.yi...@intel.com;
> olivier.m...@6wind.com; jerinjac...@gmail.com;
> maxime.coque...@redhat.com; david.march...@redhat.com;
> ary
On Fri, 2020-10-16 at 16:09 +0100, Bruce Richardson wrote:
> For users with 32-bit applications who wish to use DPDK we need to provide
> instructions on creating a 32-bit build of DPDK with meson. Therefore add a
> section with this information to the GSG.
>
> Signed-off-by: Bruce Richardson
> -
On 10/16/20 2:20 PM, Kinsella, Ray wrote:
On 15/10/2020 14:30, Andrew Rybchenko wrote:
From: Ivan Ilchenko
Change rte_eth_dev_stop() return value from void to int
and return negative errno values in case of error conditions.
Also update the usage of the function in ethdev according to
the new
> -Original Message-
> From: Ferruh Yigit
> Sent: Friday, October 16, 2020 20:05
> To: Slava Ovsiienko ; dev@dpdk.org
> Cc: NBU-Contact-Thomas Monjalon ;
> step...@networkplumber.org; olivier.m...@6wind.com;
> jerinjac...@gmail.com; maxime.coque...@redhat.com;
> david.march...@redhat.com;
On 10/16/2020 5:44 PM, Viacheslav Ovsiienko wrote:
The DPDK datapath in the transmit direction is very flexible.
An application can build the multi-segment packet and manages
almost all data aspects - the memory pools where segments
are allocated from, the segment lengths, the memory attributes
l
On 10/16/2020 4:55 PM, Slava Ovsiienko wrote:
-Original Message-
From: Ferruh Yigit
Sent: Friday, October 16, 2020 18:52
To: Slava Ovsiienko ; dev@dpdk.org
Cc: NBU-Contact-Thomas Monjalon ;
step...@networkplumber.org; olivier.m...@6wind.com;
jerinjac...@gmail.com; maxime.coque...@redhat.
16/10/2020 17:56, Ferruh Yigit:
> On 10/6/2020 4:03 PM, Konstantin Ananyev wrote:
> > @@ -120,7 +161,7 @@ _m256_mask_gather_epi8x4(__m256i pdata, __mmask8 mask)
> >
> > static const uint32_t zero;
>
> icc complains about this, although it is static [1].
> Would it be acceptable to initializ
On 10/15/2020 2:30 PM, Andrew Rybchenko wrote:
From: Ivan Ilchenko
Change rte_eth_dev_stop() return value from void to int
and return negative errno values in case of error conditions.
Also update the usage of the function in ethdev according to
the new return type.
Signed-off-by: Ivan Ilchenk
On Fri, Oct 16, 2020 at 5:56 PM Bruce Richardson
wrote:
> > librte_eal.so is indeed built with the 64 value:
> > $ pahole -C rte_mem_config build/install/lib64/librte_eal.so |grep memsegs
> > die__process_function: tag not supported (INVALID)!
> > struct rte_memseg_list memsegs[64];
Add command line parameter:
--rxpkts=X[,Y]
Sets the length of segments to scatter packets on receiving if split
feature is engaged. Affects only the queues configured with split
offloads (currently BUFFER_SPLIT is supported only).
Add interactive mode command:
testpmd> set rxpkts (x[,y]*)
Wher
16/10/2020 14:43, Bruce Richardson:
> On Fri, Oct 16, 2020 at 02:39:06PM +0200, Thomas Monjalon wrote:
> > When running Sphinx through ninja, the wrapper configured in meson
> > redirects stdout to a log file.
> > It makes more important to print issues on stderr.
> >
> > Some warnings generated b
This patch add support for RTE_ETH_RX_OFFLOAD_BUFFER_SPLIT
providing per queue configuration for this offload.
Signed-off-by: Viacheslav Ovsiienko
---
app/test-pmd/cmdline.c | 21 +++--
app/test-pmd/config.c | 9 +
2 files changed, 20 insertions(+), 10 deletions(-)
dif
If Rx queue is configured with split feature the extended
setup with specified segment sizes and pool will be performed.
Signed-off-by: Viacheslav Ovsiienko
---
app/test-pmd/cmdline.c | 12 ++--
app/test-pmd/testpmd.c | 49 +++--
app/test-pmd/t
Add command line parameter:
--rxoffs=X[,Y]
Sets the offsets of packet segments from the beginning of the
receiving buffer if split feature is engaged. Affects only the
queues configured with split offloads (currently BUFFER_SPLIT
is supported only).
Add interactive mode command, providing the sa
The command line parameter --mbuf-size is updated, it can handle
the multiple values like the following:
--mbuf-size=2176,512,768,4096
specifying the creation the extra memory pools with the requested
mbuf data buffer sizes. If some buffer split feature is engaged
the extra memory pools can be us
The DPDK datapath in the transmit direction is very flexible.
An application can build the multi-segment packet and manages
almost all data aspects - the memory pools where segments
are allocated from, the segment lengths, the memory attributes
like external buffers, registered for DMA, etc.
In th
The DPDK datapath in the transmit direction is very flexible.
An application can build the multi-segment packet and manages
almost all data aspects - the memory pools where segments
are allocated from, the segment lengths, the memory attributes
like external buffers, registered for DMA, etc.
In th
On Fri, Oct 16, 2020 at 05:47:45PM +0200, David Marchand wrote:
> Hello Bruce,
>
> On Thu, Sep 3, 2020 at 4:50 PM Bruce Richardson
> wrote:
> >
> > A number of the more advanced DPDK build settings which are not expected to
> > be user modified are stored in config/rte_config.h. In some cases, fo
On 10/16/2020 5:05 PM, Thomas Monjalon wrote:
16/10/2020 17:47, Ferruh Yigit:
On 10/16/2020 2:39 PM, Viacheslav Ovsiienko wrote:
+__rte_experimental
+struct rte_eth_rxseg_split {
+ struct rte_mempool *mp; /**< Memory pool to allocate segment from. */
+ uint16_t length; /**< Segment
16/10/2020 17:47, Ferruh Yigit:
> On 10/16/2020 2:39 PM, Viacheslav Ovsiienko wrote:
> > +__rte_experimental
> > +struct rte_eth_rxseg_split {
> > + struct rte_mempool *mp; /**< Memory pool to allocate segment from. */
> > + uint16_t length; /**< Segment data length, configures split point. */
16/10/2020 17:48, Kinsella, Ray:
> On 16/10/2020 12:22, David Marchand wrote:
> > On Fri, Oct 16, 2020 at 1:16 PM Kinsella, Ray wrote:
> >>
> >> The windows exports and the map files, feels like duplication of effort.
> >> Could we massage one into the other during the build?
> >
> > That's what
On 10/6/2020 4:03 PM, Konstantin Ananyev wrote:
Current rte_acl_classify_avx512x32() and rte_acl_classify_avx512x16()
code paths are very similar. The only differences are due to
256/512 register/instrincts naming conventions.
So to deduplicate the code:
- Move common code into “acl_run_avx512
16/10/2020 04:50, Thomas Monjalon:
> The DPDK version should appear in the top left corner of the HTML guides.
> When dropping make, the variable version has been removed,
> so Sphinx stopped integrating the version number.
>
> Fixes: a4362f150204 ("doc: build without using make")
>
> Signed-off-
> -Original Message-
> From: Ferruh Yigit
> Sent: Friday, October 16, 2020 18:52
> To: Slava Ovsiienko ; dev@dpdk.org
> Cc: NBU-Contact-Thomas Monjalon ;
> step...@networkplumber.org; olivier.m...@6wind.com;
> jerinjac...@gmail.com; maxime.coque...@redhat.com;
> david.march...@redhat.com;
On Fri, Oct 16, 2020 at 04:48:59PM +0100, Kinsella, Ray wrote:
>
> On 16/10/2020 12:22, David Marchand wrote:
> > On Fri, Oct 16, 2020 at 1:16 PM Kinsella, Ray wrote:
> >>
> >>
> >> The windows exports and the map files, feels like duplication of effort.
> >> Could we massage one into the other d
On 10/16/2020 4:48 PM, Slava Ovsiienko wrote:
-Original Message-
From: Ferruh Yigit
Sent: Friday, October 16, 2020 18:39
To: Slava Ovsiienko ; dev@dpdk.org
Cc: NBU-Contact-Thomas Monjalon ;
step...@networkplumber.org; olivier.m...@6wind.com;
jerinjac...@gmail.com; maxime.coque...@redhat.
On 16/10/2020 12:22, David Marchand wrote:
> On Fri, Oct 16, 2020 at 1:16 PM Kinsella, Ray wrote:
>>
>>
>> The windows exports and the map files, feels like duplication of effort.
>> Could we massage one into the other during the build?
>
> That's what is done with map-to-win.py, unless we have
> -Original Message-
> From: Ferruh Yigit
> Sent: Friday, October 16, 2020 18:39
> To: Slava Ovsiienko ; dev@dpdk.org
> Cc: NBU-Contact-Thomas Monjalon ;
> step...@networkplumber.org; olivier.m...@6wind.com;
> jerinjac...@gmail.com; maxime.coque...@redhat.com;
> david.march...@redhat.com;
Hello Bruce,
On Thu, Sep 3, 2020 at 4:50 PM Bruce Richardson
wrote:
>
> A number of the more advanced DPDK build settings which are not expected to
> be user modified are stored in config/rte_config.h. In some cases, for a
> custom build a user may want to override those settings via CFLAGS, so w
On 10/16/2020 2:39 PM, Viacheslav Ovsiienko wrote:
The DPDK datapath in the transmit direction is very flexible.
An application can build the multi-segment packet and manages
almost all data aspects - the memory pools where segments
are allocated from, the segment lengths, the memory attributes
l
On 16/10/2020 15:27, Ciara Power wrote:
> This patch adds a max SIMD bitwidth EAL configuration. The API allows
> for an app to set this value. It can also be set using EAL argument
> --force-max-simd-bitwidth, which will lock the value and override any
> modifications made by the app.
>
> Each
>
> W dniu 16.10.2020 o 14:43, Lukasz Wojciechowski pisze:
> > Hi Honnappa,
> >
> > Thank you for your answer.
> > In the current v7 version I followed your advise and used RELAXED memory
> model.
> > And it works without any issues. I guess after fixing other issues found
> since v4 the distrib
Added -v option to switch between different lookup implementations
to measure their performance and correctness.
Signed-off-by: Vladimir Medvedkin
---
app/test-fib/main.c | 65 ++---
1 file changed, 62 insertions(+), 3 deletions(-)
diff --git a/ap
Add new lookup implementation for FIB6 trie algorithm using
AVX512 instruction set
Signed-off-by: Vladimir Medvedkin
Acked-by: Konstantin Ananyev
---
doc/guides/rel_notes/release_20_11.rst | 2 +-
lib/librte_fib/meson.build | 17 +++
lib/librte_fib/rte_fib6.c | 2 +
Move trie table layout and lookup definition into the
private header file. This is necessary for implementing a
vectorized lookup function in a separate .с file.
Signed-off-by: Vladimir Medvedkin
Acked-by: Konstantin Ananyev
---
lib/librte_fib/trie.c | 121 --
Add type argument to trie_get_lookup_fn()
Now it only supports RTE_FIB6_TRIE_SCALAR
Add new rte_fib6_set_lookup_fn() - user can change lookup
function type runtime.
Signed-off-by: Vladimir Medvedkin
Acked-by: Konstantin Ananyev
---
lib/librte_fib/rte_fib6.c | 20 +++-
lib/
Add new lookup implementation for DIR24_8 algorithm using
AVX512 instruction set
Signed-off-by: Vladimir Medvedkin
Acked-by: Konstantin Ananyev
---
doc/guides/rel_notes/release_20_11.rst | 3 +
lib/librte_fib/dir24_8.c | 38
lib/librte_fib/dir24_8_avx512.c| 165
Move dir24_8 table layout and lookup definition into the
private header file. This is necessary for implementing a
vectorized lookup function in a separate .с file.
Signed-off-by: Vladimir Medvedkin
Acked-by: Konstantin Ananyev
---
lib/librte_fib/dir24_8.c | 225 +---
Add type argument to dir24_8_get_lookup_fn()
Now it supports 3 different lookup implementations:
RTE_FIB_DIR24_8_SCALAR_MACRO
RTE_FIB_DIR24_8_SCALAR_INLINE
RTE_FIB_DIR24_8_SCALAR_UNI
Add new rte_fib_set_lookup_fn() - user can change lookup
function type runtime.
Signed-off-by: Vladimir Medvedk
This patch series implements vectorized lookup using AVX512 for
ipv4 dir24_8 and ipv6 trie algorithms.
Also introduced rte_fib_set_lookup_fn() to change lookup function type.
Added option to select lookup function type in testfib application.
Depends-on: patch-81125 ("eal: add max SIMD bitwidth")
New data type to manipulate 512 bit AVX values.
Signed-off-by: Vladimir Medvedkin
Acked-by: Konstantin Ananyev
---
lib/librte_eal/x86/include/rte_vect.h | 19 +++
1 file changed, 19 insertions(+)
diff --git a/lib/librte_eal/x86/include/rte_vect.h
b/lib/librte_eal/x86/include/r
On 16/10/2020 13:51, Gregory Etelson wrote:
> From: Eli Britstein
>
> rte_flow API provides the building blocks for vendor-agnostic flow
> classification offloads. The rte_flow "patterns" and "actions"
> primitives are fine-grained, thus enabling DPDK applications the
> flexibility to offload
On 10/16/2020 4:05 PM, Ferruh Yigit wrote:
On 10/16/2020 2:39 PM, Viacheslav Ovsiienko wrote:
The command line parameter --mbuf-size is updated, it can handle
the multiple values like the following:
--mbuf-size=2176,512,768,4096
specifying the creation the extra memory pools with the requested
Hello,
On 16/10/2020 16:15, David Marchand wrote:
On Tue, Oct 13, 2020 at 3:14 PM Vladimir Medvedkin
wrote:
This patch series implements vectorized lookup using AVX512 for
ipv4 dir24_8 and ipv6 trie algorithms.
Also introduced rte_fib_set_lookup_fn() to change lookup function type.
Added opti
On Tue, Oct 13, 2020 at 3:14 PM Vladimir Medvedkin
wrote:
>
> This patch series implements vectorized lookup using AVX512 for
> ipv4 dir24_8 and ipv6 trie algorithms.
> Also introduced rte_fib_set_lookup_fn() to change lookup function type.
> Added option to select lookup function type in testfib
16/10/2020 15:39, Viacheslav Ovsiienko:
> + if (offset != 0) {
> + if (seg_capa->offset_allowed == 0) {
> + RTE_ETHDEV_LOG(ERR, "Rx segmentation with
> offset is not supported\n");
> + return -ENOTSUP;
> +
For users with 32-bit applications who wish to use DPDK we need to provide
instructions on creating a 32-bit build of DPDK with meson. Therefore add a
section with this information to the GSG.
Signed-off-by: Bruce Richardson
---
doc/guides/linux_gsg/build_dpdk.rst | 25 +
On Fri, 16 Oct 2020 16:34:46 +0200
Thomas Monjalon wrote:
> 14/10/2020 10:43, Kinsella, Ray:
> >
> > On 14/10/2020 03:26, Ferruh Yigit wrote:
> > > Queue stats will be removed from basic stats to xstats.
> > >
> > > It will be PMDs responsibility to fill queue stats based on number of
> > > q
On 10/16/2020 2:39 PM, Viacheslav Ovsiienko wrote:
The command line parameter --mbuf-size is updated, it can handle
the multiple values like the following:
--mbuf-size=2176,512,768,4096
specifying the creation the extra memory pools with the requested
mbuf data buffer sizes. If some buffer spli
Hi David,
> On Oct 16, 2020, at 8:53 AM, David Marchand wrote:
>
> Hello Dharmik,
>
> On Wed, Sep 2, 2020 at 12:01 AM Dharmik Thakkar
> wrote:
>>> [Wang, Yipeng]
>>> Hi, Dharmik,
>>> Thanks for the patch. It generally looks good to me.
>>
>> That’s great. I will convert it to a patch.
>>
>>
Ferruh Yigit wrote:
> Sent: Friday, October 16, 2020 1:57 PM
> >> Lijun's idea can work. There was a problem in implementation related
> >> to the key size assumption, which can be fixed.
> >>
> >> Even it is fixed, when user gives a rss rule without a key, we are
> >> getting key from device an
On Fri, Oct 16, 2020 at 8:05 PM Thomas Monjalon wrote:
>
> 14/10/2020 10:43, Kinsella, Ray:
> >
> > On 14/10/2020 03:26, Ferruh Yigit wrote:
> > > Queue stats will be removed from basic stats to xstats.
> > >
> > > It will be PMDs responsibility to fill queue stats based on number of
> > > queues
On Fri, Oct 16, 2020 at 04:34:46PM +0200, Thomas Monjalon wrote:
> 14/10/2020 10:43, Kinsella, Ray:
> >
> > On 14/10/2020 03:26, Ferruh Yigit wrote:
> > > Queue stats will be removed from basic stats to xstats.
> > >
> > > It will be PMDs responsibility to fill queue stats based on number of
> >
14/10/2020 10:43, Kinsella, Ray:
>
> On 14/10/2020 03:26, Ferruh Yigit wrote:
> > Queue stats will be removed from basic stats to xstats.
> >
> > It will be PMDs responsibility to fill queue stats based on number of
> > queues they have.
> >
> > Until all PMDs implement the xstats, a temporary
>
When choosing a vector path to take, an extra condition must be
satisfied to ensure the max SIMD bitwidth allows for the CPU enabled
path. These checks are added in the check alg helper functions.
Cc: Konstantin Ananyev
Signed-off-by: Ciara Power
Acked-by: Konstantin Ananyev
Tested-by: Konstan
16/10/2020 14:16, Ferruh Yigit:
> On 10/14/2020 3:26 AM, Ferruh Yigit wrote:
> > Queue stats are stored in 'struct rte_eth_stats' as array and array size
> > is defined by 'RTE_ETHDEV_QUEUE_STAT_CNTRS' compile time flag.
> >
> > As a result of technical board discussion, decided to remove the queu
When choosing the vector path, max SIMD bitwidth is now checked to
ensure the vector path is suitable. To do this, the scalar function is
chosen by default in the struct, but at node initialisation time, this
function pointer is updated to the vector version if supported, and
if it is within the ma
When choosing a vector path to take, an extra condition must be
satisfied to ensure the max SIMD bitwidth allows for the CPU enabled
path.
Cc: Byron Marohn
Cc: Yipeng Wang
Signed-off-by: Ciara Power
Acked-by: Yipeng Wang
---
v4: Updated enum name.
---
lib/librte_efd/rte_efd.c | 7 +--
1
When choosing a vector path to take, an extra condition must be
satisfied to ensure the max SIMD bitwidth allows for the CPU enabled
path.
The vector path was initially chosen in RTE_INIT, however this is no
longer suitable as we cannot check the max SIMD bitwidth at that time.
Default handlers ar
When choosing a vector path to take, an extra condition must be
satisfied to ensure the max SIMD bitwidth allows for the CPU enabled
path.
Cc: David Hunt
Signed-off-by: Ciara Power
Acked-by: David Hunt
---
v4: Updated enum name.
---
lib/librte_distributor/rte_distributor.c | 3 ++-
1 file ch
When choosing a vector path to take, an extra condition must be
satisfied to ensure the max SIMD bitwidth allows for the CPU
enabled path.
Cc: Yipeng Wang
Cc: Sameh Gobriel
Signed-off-by: Ciara Power
Acked-by: Yipeng Wang
---
v4: Updated enum name.
---
lib/librte_member/rte_member_ht.c | 3
1 - 100 of 284 matches
Mail list logo
