On Fri, Jan 12, 2018 at 07:13:09PM +, Ferruh Yigit wrote:
> On 1/11/2018 3:35 PM, Tomasz Duszynski wrote:
> > Following commits provide fixes for a few recently
> > found NET MRVL PMD issues.
> >
> > Natalie Samsonov (5):
> > net/mrvl: fix multiple probe issue
> > net/mrvl: fix hif objects
> -Original Message-
> From: Zhang, Qi Z
> Sent: Sunday, January 14, 2018 12:02 PM
> To: Xing, Beilei
> Cc: dev@dpdk.org; sta...@dpdk.org
> Subject: RE: [PATCH v2] net/i40e: fix packet type parser issue
>
>
>
> > -Original Message-
> > From: Xing, Beilei
> > Sent: Friday, Janu
On 01/15/2018 04:57 AM, Qi Zhang wrote:
According to exist implementation, rte_eth_[rx|tx]_queue_setup will
always return fail if device is already started(rte_eth_dev_start).
This can't satisfied the usage when application want to deferred setup
part of the queues while keep traffic running on
Previously, mempool of multiple memory segments would lead to MR
registration error "mempool not virtually contiguous". This patch
support multi-segments mempool by registering each memory segment of
mempool.
Signed-off-by: Xueming Li
---
drivers/net/mlx5/mlx5.h | 4 +-
drivers/net/mlx
Secondary process are not alloed to access verbs resources, add check to
prevent MR registration in data path.
Signed-off-by: Xueming Li
---
doc/guides/nics/mlx5.rst | 1 +
drivers/net/mlx5/mlx5_mr.c | 3 +++
drivers/net/mlx5/mlx5_rxtx.h | 3 ---
3 files changed, 4 insertions(+), 3 deletio
MRs are registered to priv at device start or on the fly, destroyied
when device stop.
No mR registration to perticular TXQ, MR references cache in TXQ just
part of device MR list, no reason to keep MR refcnt.
Signed-off-by: Xueming Li
---
drivers/net/mlx5/mlx5.h | 1 -
drivers/net/mlx5/ml
This reverts commit 17fd4a504a4780455f79969803dc231521254ca8.
Fixes: 12ed59b702cc ("net/mlx5: fix Memory Region registration")
Signed-off-by: Xueming Li
---
drivers/net/mlx5/mlx5_rxtx.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/mlx5/mlx5_rxtx.h b/drivers/net
Thursday, January 11, 2018 9:58 AM, Yongseok Koh:
> Subject: [dpdk-stable] [PATCH] net/mlx5: fix calculation of flow ID flag
>
> PKT_RX_FDIR_ID should be set only if flow_tag is neither non-zero nor
> MLX5_FLOW_MARK_DEFAULT.
>
> Fixes: 6cb559d67b83 ("net/mlx5: add vectorized Rx/Tx burst for x86")
On Fri, Jan 12, 2018 at 07:13:22PM +, Ferruh Yigit wrote:
> On 1/11/2018 3:35 PM, Tomasz Duszynski wrote:
> > From: Natalie Samsonov
>
> Commit title updated to [1] assuming hif is an abbreviation (btw, what is
> hif?)
Thanks for fixing that. HIF stands for host-interface. You want to alloca
Wednesday, January 10, 2018 7:51 PM, Yongseok Koh:
> Subject: [dpdk-dev] [PATCH] net/mlx5: fix missing attribute size for drop
> action
>
> Adding the size of ibv_flow_spec_action_drop is missing. This can cause
> overflow when converting rte_flow to ibv_flow and thus corrupt memory.
>
> Fixes: 8
Wednesday, January 10, 2018 7:47 PM, Yongseok Koh:
> If mlx5_dev_link_status_handler() is executed while canceling the alarm,
> deadlock can happen because rte_eal_alarm_cancel() waits for all callbackes
> to finish execution and both calls are protected by priv->lock.
>
> Fixes: 198a3c339a8f ("ml
Signed-off-by: Akhil Goyal
Acked-by: Hemant Agrawal
---
doc/guides/cryptodevs/features/dpaa_sec.ini | 1 +
doc/guides/rel_notes/release_18_02.rst | 5 +
drivers/crypto/dpaa_sec/dpaa_sec.c | 410 ++--
drivers/crypto/dpaa_sec/dpaa_sec.h | 62
From: Hemant Agrawal
Signed-off-by: Hemant Agrawal
Acked-by: Akhil Goyal
---
drivers/crypto/dpaa_sec/dpaa_sec.c | 154 +
drivers/crypto/dpaa_sec/dpaa_sec.h | 74 +-
2 files changed, 145 insertions(+), 83 deletions(-)
diff --git a/drivers/cr
From: Hemant Agrawal
Context memory is allocated from mempool. Ideally
it will get all memory from single segment, so simple offset
calculation is used for address conversion for such addresses
from context memory.
Signed-off-by: Hemant Agrawal
Acked-by: Akhil Goyal
---
drivers/crypto/dpaa_se
Following changes are added to improve performance.
1. optimize virtual to physical address conversion
2. support for multiple sessions in a single queue pair
3. support for ipsec protocol offload
changes in v3:
- updated release notes in patch 3/3 for ipsec protocol offload.
changes in v2:
- in
This patch add the user command line configured mempool ops name
API to librte_mbuf and sends a deprecation notice to remove the
similar API from eal.
Signed-off-by: Hemant Agrawal
---
doc/guides/rel_notes/deprecation.rst | 7 +++
lib/librte_mbuf/rte_mbuf.c | 8 +++-
lib/libr
W.r.t the multiple discussions in the past about the ability to
dynamically detect the HW mempool support. [1],[2] & [3]
This patchset helps in removing the current static mempool selection
model and provides a flexible model to select the pktmbuf mempool
in more dynamic way.
1) This patchset up
This patch prefix the mbuf pool ops name with "user" to indicate
that it is user defined.
This patch also change the logic to maintain the value of
user defined and compile time i.e. RTE_MBUF_DEFAULT_MEMPOOL_OPS.
The pktmbuf_create_pool is updated to reflect the same.
Signed-off-by: Hemant Agraw
Introduce a new helper for pktmbuf pool, which will allow
the application to optionally specify the mempool ops name
as well.
Signed-off-by: Hemant Agrawal
---
lib/librte_mbuf/rte_mbuf.c | 23 ++--
lib/librte_mbuf/rte_mbuf.h | 42 ++
With this patch the specific HW mempool are no longer required to be
specified in the config file at compile. A default platform hw mempool
can be detected dynamically and published to config at run time.
Only one type of HW mempool can be active default.
Signed-off-by: Hemant Agrawal
---
lib/li
Signed-off-by: Hemant Agrawal
---
lib/librte_eal/common/eal_common_options.c | 1 +
lib/librte_eal/common/eal_internal_cfg.h | 2 ++
lib/librte_eal/rte_eal_version.map | 1 +
3 files changed, 4 insertions(+)
diff --git a/lib/librte_eal/common/eal_common_options.c
b/lib/librte_eal/comm
Hi Anatoly,
Thanks for your comments. Updated. Please review the version 2.
Thanks,
Phil Yang
> -Original Message-
> From: Burakov, Anatoly [mailto:anatoly.bura...@intel.com]
> Sent: Saturday, January 13, 2018 1:44 AM
> To: Phil Yang ; dev@dpdk.org
> Cc: nd ; Jianbo Liu ; Herbert Guan
>
This series of patches are the fix for memzone autotest.
RTE>>memzone_autotest
test basic memzone
API Zone 0: name:, IO:0x177ffe9e00, len:0x80
virt:0x7ffe9e00, socket_id:0, flags:0
Zone 1: name:, IO:0x177ffe9980, len:0x400
virt:0x7ffe9980, socket_id:0, flags:0
Zone 2: name:, IO:0x177ffe950
No need to free a NULL memzone. It will cause test
termination.
Signed-off-by: Phil Yang
---
test/test/test_memzone.c | 4
1 file changed, 4 deletions(-)
diff --git a/test/test/test_memzone.c b/test/test/test_memzone.c
index 6e80977..9c20172 100644
--- a/test/test/test_memzone.c
+++ b/test
When reserving memzone for mz[], it will out of mz[RTE_MAX_MEMZONE] memory
bound after the counter reached to RTE_MAX_MEMZONE. It will flush the
counter's memory and lead to mz[] memory cannot be freed.
Fixd by extend to mz[RTE_MAX_MEMZONE + 1].
Signed-off-by: Phil Yang
---
test/test/test_memzo
When reserving memzone for mz[], it will out of mz[RTE_MAX_MEMZONE] memory
bound after the counter reached to RTE_MAX_MEMZONE. It will flush the counter's
memory and lead to mz[] memory cannot be freed.
Fixd by extend to mz[RTE_MAX_MEMZONE + 1].
Signed-off-by: Phil Yang
---
test/test/test_memzo
This series of patches are the fix for memzone autotest.
RTE>>memzone_autotest
test basic memzone API
Zone 0: name:, IO:0x177ffe9e00, len:0x80,
virt:0x7ffe9e00, socket_id:0, flags:0
Zone 1: name:, IO:0x177ffe9980, len:0x400,
virt:0x7ffe9980, socket_id:0, flags:0
Zone 2: name:, IO:0x177ffe9
No need to free a NULL memzone. It will cause test
termination.
Signed-off-by: Phil Yang
---
test/test/test_memzone.c | 4
1 file changed, 4 deletions(-)
diff --git a/test/test/test_memzone.c b/test/test/test_memzone.c
index 6e80977..9c20172 100644
--- a/test/test/test_memzone.c
+++ b/test
Hi Thomas,
> -Original Message-
> From: Thomas Monjalon [mailto:tho...@monjalon.net]
> Sent: Monday, January 15, 2018 4:31 AM
> To: Shreyansh Jain
> Cc: dev@dpdk.org; Hemant Agrawal ;
> fiona.tr...@intel.com; rosen...@intel.com
> Subject: Re: [dpdk-dev] [PATCH v1 0/5] Introduce generic 'r
other vfio based module e.g. fslmc will also need to use
the clear_group call.
So, exposing it and renaming it to *rte_vfio_clear_group*
Signed-off-by: Hemant Agrawal
Acked-by: Anatoly Burakov
---
lib/librte_eal/bsdapp/eal/eal.c| 5 +
lib/librte_eal/common/include/rte_vfio.
Signed-off-by: Hemant Agrawal
---
drivers/bus/fslmc/fslmc_vfio.c | 4
1 file changed, 4 insertions(+)
diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index fb2f403..12b254a 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -634,
> -Original Message-
> From: xiangxia.m@gmail.com [mailto:xiangxia.m@gmail.com]
> Sent: Sunday, January 14, 2018 6:04 PM
> To: Dai, Wei ; Xing, Beilei
> Cc: dev@dpdk.org; Tonghao Zhang
> Subject: [PATCH v3 1/6] net/ixgbevf: unregister irq handler when other
> interrupts not allowe
When vhost reallocate dev and vq for NUMA enabled case, it doesn't perform
deep copy, which lead to 1) zmbuf list not valid 2) remote memory access.
This patch is to re-initlize the zmbuf list and also do the deep copy.
Signed-off-by: Junjie Chen
---
lib/librte_vhost/vhost_user.c | 31 ++
Hi,
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Shahaf Shuler
> Sent: Wednesday, January 10, 2018 5:09 PM
> To: Wu, Jingjing ; tho...@monjalon.net; Yigit,
> Ferruh
> Cc: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH v4 11/11] app/testpmd: enable fast free Tx
>
Hi,
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Shahaf Shuler
> Sent: Wednesday, January 10, 2018 5:09 PM
> To: Wu, Jingjing ; tho...@monjalon.net; Yigit,
> Ferruh
> Cc: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH v4 10/11] app/testpmd: adjust on the flight V
Hi,
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Shahaf Shuler
> Sent: Wednesday, January 10, 2018 5:09 PM
> To: Wu, Jingjing ; tho...@monjalon.net; Yigit,
> Ferruh
> Cc: dev@dpdk.org
> Subject: [dpdk-dev] [PATCH v4 07/11] app/testpmd: add command line
> op
Thanks for providing idea to me! I will refer to that.
I have another way feeling that it should works too. switchover active and
standby pointer on forward plane, rather than on control plane. when update ctx
in control plane(add rule, build on standby, meanwhile add rule on active not
build)
According to exist implementation, rte_eth_[rx|tx]_queue_setup will
always return fail if device is already started(rte_eth_dev_start).
This can't satisfied the usage when application want to deferred setup
part of the queues while keep traffic running on those queues already
be setup.
example:
r
Hi,
> -Original Message-
> From: dev [mailto:dev-boun...@dpdk.org] On Behalf Of Georgios Katsikas
> Sent: Saturday, January 13, 2018 5:01 AM
> To: olivier.m...@6wind.com
> Cc: dev@dpdk.org; Georgios Katsikas
> Subject: [dpdk-dev] [PATCH 1/3] app/testpmd: Moved cmdline_flow to
> librte_cmd
Hi Chas,
1, Would you like to use this CLI “lspci -nn | grep Eth” to get the device ID
of the NIC?
2, You mentioned “the doc of the platform”. Is it public? If so, could you tell
me the link? Just want to understand more about it.
Thanks.
Best regards
Wenzhuo Lu
From: Chas Williams [mailto:3c
10/01/2018 08:59, wei.guo.si...@gmail.com:
> This patch adds a new option "--ring-bind-lcpu"(no parameter). With
> this, testpmd can utilize the PCI-e bus bandwidth on another NUMA
> nodes.
Waiting for a review.
Please uese --in-reply-to when sending a new version of the patch,
so we can track m
11/01/2018 15:05, Jeff Guo:
> +enum rte_dev_state {
> + RTE_DEV_UNDEFINED, /**< unknown device state */
> + RTE_DEV_FAULT, /**< device fault or error */
> + RTE_DEV_PARSED, /**< device have been parsed on bus*/
> + RTE_DEV_PROBED, /**< devcie have been probed driver */
> +};
Hi,
11/01/2018 15:05, Jeff Guo:
> +/* A genaral callback for all registerd devices */
Typos: genaral, registerd
So the callback is only for registered devices?
What about hotplugged devices?
> +/**
> + * It registers the callback for the specific event. Multiple
> + * callbacks cal be registere
02/01/2018 13:57, Shreyansh Jain:
> This patchset introduces rawdevices or generic device support in DPDK.
This new library has no impact on the existing code,
so it may be integrated later in 18.02-rc2.
It will give you a bit more time for the last details.
02/01/2018 13:57, Shreyansh Jain:
> Patch introduces rawdev unit testcase for validation against the
> Skeleton rawdev dummy PMD implementation.
>
> Signed-off-by: Shreyansh Jain
> ---
> test/test/Makefile | 4 +
> test/test/test_rawdev.c | 376
> +
02/01/2018 13:57, Shreyansh Jain:
> Signed-off-by: Shreyansh Jain
> ---
> config/common_base | 1 +
> drivers/Makefile | 2 ++
> mk/rte.app.mk | 4
> 3 files changed, 7 insertions(+)
Please merge it with the previous patch.
02/01/2018 13:57, Shreyansh Jain:
> +struct skeleton_firmware {
> + struct skeleton_firmware_version_info firmware_version;
> + /**< Device firmware information */
> + enum skeleton_firmware_state firmware_state;
> + /**< Device state */
> +};
> +
> +#define SKELETON_MAX_ATTRIBUTES
02/01/2018 13:57, Shreyansh Jain:
> Add config option CONFIG_RTE_LIBRTE_RAWDEV for toggling rawdev
> library support. This patch also enables compilation of the library.
>
> Signed-off-by: Shreyansh Jain
> ---
> config/common_base | 7 +++
> lib/Makefile | 3 +++
> mk/rte.app.mk |
02/01/2018 13:57, Shreyansh Jain:
> --- /dev/null
> +++ b/lib/librte_rawdev/rte_rawdev_version.map
> @@ -0,0 +1,33 @@
> +EXPERIMENTAL {
> + global:
> +
> + rte_rawdev_close;
> + rte_rawdev_configure;
> + rte_rawdev_count;
> + rte_rawdev_dequeue_buffers;
> + rte_r
Hi,
02/01/2018 13:57, Shreyansh Jain:
> A device is DPDK has a flavor to it - ethernet, crypto, event etc.
I don't understand this sentence.
I guess it can be completely removed.
> A rawdevice represents a generic device map-able to a device flavour
> not being currently handled out-of-the-box b
On Sun, 14 Jan 2018 01:45:42 +0300
Aleksey Baulin wrote:
> Please see my comments inline.
>
> On Sun, Jan 14, 2018 at 1:24 AM, Thomas Monjalon
> wrote:
>
> > Hi,
> >
> > I moved your top-post below and did some comments inline.
> > More opinions are welcome.
> >
> > 13/01/2018 23:05, Aleksey B
On Thu, Jan 11, 2018 at 05:06:58PM +, Ferruh Yigit wrote:
> From: Stephen Hemminger
>
> This driver was not doing atomic update of link status information.
> And the return value was different than others.
> The hardware also does not do autonegotiation (at least on Linux).
>
> Signed-off-by
Hi Cristian,
Responses inline below.
Regards
Alan
> -Original Message-
> From: Dumitrescu, Cristian [mailto:cristian.dumitre...@intel.com]
> Sent: Thursday, January 11, 2018 1:11 PM
> To: alangordonde...@gmail.com
> Cc: dev@dpdk.org; Alan Dewar ; Kantecki, Tomasz
> ; Singh, Jasvinder
Hi Cristian,
> > +
> > + rte_red_scaling -= count - RTE_RED_SCALING;
>
> Why not simply: rte_red_scaling = count?
The RED code stores the moving average in a uint32_t as a pseudo floating point
number with a fixed sized fractional part of 10 bits.
This allows a maximum queue length of 1024.
14/01/2018 15:36, Neil Horman:
> On Sat, Jan 13, 2018 at 04:56:11PM +0100, Thomas Monjalon wrote:
> > 13/01/2018 01:28, Neil Horman:
> > > On Fri, Jan 12, 2018 at 03:55:10PM +, Ferruh Yigit wrote:
> > > > After this point agree to using EXPERIMENTAL tag in the version map as
> > > > standard,
On Sat, Jan 13, 2018 at 04:56:11PM +0100, Thomas Monjalon wrote:
> 13/01/2018 01:28, Neil Horman:
> > On Fri, Jan 12, 2018 at 03:55:10PM +, Ferruh Yigit wrote:
> > > After this point agree to using EXPERIMENTAL tag in the version map as
> > > standard,
> > > but it will be hard to maintain "AP
Add Cavium octeontx to meson build and change order of drivers built
from event->mempool->net to mempool->net->event to resolve dependency.
Signed-off-by: Pavan Nikhilesh
---
v2 Changes:
- Add SPDX license tags
- Modify driver build order similar to make build system.
drivers/event/meson.bu
Add Cavium octeontx and thunder nicvf to meson build infrastructure.
Signed-off-by: Pavan Nikhilesh
---
v2 Changes:
- Add SPDX license tags.
drivers/net/meson.build | 3 ++-
drivers/net/octeontx/base/meson.build | 21 +
drivers/n
Signed-off-by: Pavan Nikhilesh
---
v2 Changes:
- Add SPDX license tags.
app/meson.build | 1 +
app/test-eventdev/meson.build | 28
2 files changed, 29 insertions(+)
create mode 100644 app/test-eventdev/meson.build
diff --git a/app/meson.build b/ap
Add octeontx hardware mempool driver to meson build.
Signed-off-by: Pavan Nikhilesh
---
v2:
- remove unneeded include variable.
- include SPDX license tags
drivers/mempool/meson.build | 2 +-
drivers/mempool/octeontx/meson.build | 10 ++
2 files changed, 11 insertions(+),
On 1/12/2018 10:06 PM, Thomas Monjalon wrote:
07/12/2017 11:28, Hemant Agrawal:
other vfio based module e.g. fslmc will also need to use
the clear_group call.
So, exposing it and renaming it to *rte_vfio_clear_group*
Signed-off-by: Hemant Agrawal
---
--- a/lib/librte_eal/linuxapp/eal/eal_vfio.
Date: 20 Dec 2017
Attendees:
Bruce, Hemant, Konstantin, Olivier, Thomas, Yuanhan, Stephen
Topic: Licensing progress
* Discussion about the format of SPDX tag line. Should we choose "/*" or "//"
as used by kernel?
* We concluded to use "/*" for SPDX tag in first line. If any issue will c
Friday, January 12, 2018 3:31 PM, Ferruh Yigit:
> On 12/26/2017 9:23 AM, Shahaf Shuler wrote:
> > Ethdev offloads API has changed since:
> >
> > commit ce17eddefc20 ("ethdev: introduce Rx queue offloads API") commit
> > cba7f53b717d ("ethdev: introduce Tx queue offloads API")
> >
> > This commit su
From: Tonghao Zhang
ixgbe doesn't write the ITR counter, disable it now.
Signed-off-by: Tonghao Zhang
---
drivers/net/ixgbe/ixgbe_ethdev.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/net/ixgbe/ixgbe_ethdev.c b/drivers/net/ixgbe/ixgbe_ethdev.c
index 604ee6d..a4
From: Tonghao Zhang
If dpdk APPs call the rte_eth_dev_rx_intr_enable or
rte_eth_dev_rx_intr_disable frequently, and ixgbe vf will read
the IXGBE_VTEIMS register everytime. The patch saves the IXGBE_VTEIMS
to mask to avoid read frequently.
Signed-off-by: Tonghao Zhang
Acked-by: Beilei Xing
Acke
From: Tonghao Zhang
When binding the ixgbe pf or vf to vfio and call the
rte_eth_dev_rx_intr_enable frequently, the interrupt setting
(msi_set_mask_bit) will take more CPU as show below. rte_intr_enable
calls the ioctl to map the fd to interrupts frequently.
perf top:
5.45% [kernel] [k] msi_s
From: Tonghao Zhang
With this patch, the ITR value of ixgbe PF, can be
configured as wanted.
Signed-off-by: Tonghao Zhang
---
drivers/net/ixgbe/ixgbe_ethdev.c | 5 +
drivers/net/ixgbe/ixgbe_rxtx.c | 3 ++-
2 files changed, 3 insertions(+), 5 deletions(-)
diff --git a/drivers/net/ixgbe/i
From: Tonghao Zhang
When bind the ixgbe VF (e.g 82599 card) to igb_uio and enable the
rx-interrupt, there will be more than one epoll_wait on intr_handle.fd.
One is in "eal-intr-thread" thread, and the others are in the thread
which call the "rte_epoll_wait". The problem is that sometimes
"eal-in
From: Tonghao Zhang
Set EITR interval as default. This patch can improve the
performance when we enable the rx-intrrupt to process the
packets because we hope rx-intrrupt reduce CPU. For example,
the 200us value of EITR makes the performance better with
the low CPU.
Users can configure the value
This command will simulate the process of setting the
eth-peer from command line.
It will be useful to preform extra testing.
usage:
testpmd> set eth-peer .
Signed-off-by: Wisam Jaddo
Acked-by: Wenzhuo Lu
---
changes in v2:
added Ack from Wenzhuo, and sending to ML again
due
70 matches
Mail list logo
