This series adds missing read barriers after reading avail index
for split ring and desc flags for packed ring.
Also, it turns out that some descriptors prefetching are either
badly placed, or useless, last part of the series fixes that.
With the series applied, I get between 0 and 4% gain depend
A read barrier is required to ensure that the ordering between
descriptor's flags and content reads is enforced.
1. read flags = desc->flags
if (flags & AVAIL_BIT)
2. read desc->id
There is a control dependency between steps 1 and step 2.
2 could be speculatively executed before 1, which could
This patch moves the prefetch after the available index
is read to avoid prefetching a descriptor not available yet.
Signed-off-by: Maxime Coquelin
---
lib/librte_vhost/virtio_net.c | 7 ---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/lib/librte_vhost/virtio_net.c b/lib/lib
A read barrier is required to ensure the ordering between
available index and the descriptor reads is enforced.
1. read avail_head = avail->idx
2. read cur_idx = last_avail_idx
if (cur_idx != avail_head) {
3. read idx = avail->ring[cur_idx]
4. read desc[idx]
}
There is a control dependenc
This prefetch does not show any performance improvement.
Signed-off-by: Maxime Coquelin
---
lib/librte_vhost/virtio_net.c | 2 --
1 file changed, 2 deletions(-)
diff --git a/lib/librte_vhost/virtio_net.c b/lib/librte_vhost/virtio_net.c
index a9be633de..4d201a3fd 100644
--- a/lib/librte_vhost/vi
> -Original Message-
> From: dev On Behalf Of Konstantin Ananyev
> Sent: Wednesday, November 14, 2018 1:28 AM
> To: dev@dpdk.org
> Cc: Konstantin Ananyev
> Subject: [dpdk-dev] [PATCH 2/2] test: add new test-cases for rwlock autotest
>
> This patch targets 19.02 release.
>
> Add few fun
> -Original Message-
> From: Honnappa Nagarahalli
> Sent: Wednesday, December 19, 2018 2:38 PM
> To: Konstantin Ananyev ; dev@dpdk.org
> Cc: nd ; Honnappa Nagarahalli
> ; Gavin Hu (Arm Technology China)
> ; nd
> Subject: RE: [dpdk-dev] [PATCH 1/2] rwlock: introduce 'try' semantics for
Hi， Tianfei,
After checking the code, I found that opae_adapter_alloc() function is no long
called. So is it Ok to delete the function?
To:
I agree that the free is not necessary.
-Original Message-
From: Xu, Rosen
Sent: Wednesday, December 19, 2018 3:16 PM
To: Pei, Andy ; dev@dpdk.or
19/12/2018 08:41, Qi Zhang:
> When secondary process quit, the mp_socket* file still exist, that
> cause rte_mp_request_sync fail when try to send message on a floating
> socket.
>
> The patch fix the issue by introduce a function rte_mp_channel_fini.
> This function will be called by rte_eal_clea
On 19/12/2018 3:18 AM, Thomas Monjalon wrote:
14/12/2018 14:11, Liang Ma:
Previously, in order to use the power library, it was necessary
for the user to disable the intel_pstate driver by adding
“intel_pstate=disable” to the kernel command line for the system,
which causes the acpi_cpufreq dr
On Tue, Dec 18, 2018 at 1:27 PM David Marchand
wrote:
>
> On Tue, Dec 18, 2018 at 1:25 PM Neil Horman wrote:
>
>> On Tue, Dec 18, 2018 at 11:41:34AM +0100, David Marchand wrote:
>>
>> > The real issue in the end is that the __rte_experimental in headers is
>> the
>> > most important thing and ca
On 12/12/18 7:53 PM, Michael S. Tsirkin wrote:
On Wed, Dec 12, 2018 at 05:34:31PM +0100, Maxime Coquelin wrote:
Hi Ilya,
On 12/12/18 4:23 PM, Ilya Maximets wrote:
On 12.12.2018 11:24, Maxime Coquelin wrote:
Instead of writing back descriptors chains in order, let's
write the first chain fl
On Wed, Dec 19, 2018 at 01:37:18PM +0900, Hideyuki Yamashita wrote:
> The common data freeing has been moved to rte_eth_dev_release_port(),
> so freeing mac_addrs like this in eth_dev_close() is unnecessary and
> will cause double free.
>
> Fixes: e16adf08e54d ("ethdev: free all common data when r
From: David Marchand [mailto:david.march...@redhat.com]
Sent: Tuesday 18 December 2018 10:49
To: Akhil Goyal ; Mokhtar, Amr
Cc: Neil Horman ; dev@dpdk.org; sta...@dpdk.org;
tredae...@redhat.com; Yigit, Ferruh
Subject: Re: [dpdk-dev] [PATCH] bbdev: add missing experimental tags
On Tue, Dec 18
On Tue, Dec 18, 2018 at 02:50:30PM +, Ferruh Yigit wrote:
> On 12/18/2018 10:26 AM, Gaetan Rivet wrote:
> > -Wmisleading-indentation was introduced in GCC 6.0.
>
> It seems '-Wmisleading-indentation' is part of -Wall, which we already set by
> default. If so no need to explicitly add it.
>
>
On Tue, Dec 11, 2018 at 02:48:02PM +0100, Maxime Coquelin wrote:
Signed-off-by: Maxime Coquelin
---
drivers/net/virtio/virtio_rxtx.c | 8
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/net/virtio/virtio_rxtx.c b/drivers/net/virtio/virtio_rxtx.c
index eb891433e..e1c
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> Add 'uint64_t opaque_data' inside struct rte_cryptodev_sym_session.
> That allows upper layer to easily associate some user defined
> data with the session.
>
> Signed-off-by: Konstantin Ananyev
> Acked-by: Fiona Trahe
> Acked-by: Mohammad Abdu
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> Add 'uint64_t opaque_data' inside struct rte_security_session.
> That allows upper layer to easily associate some user defined
> data with the session.
>
> Signed-off-by: Konstantin Ananyev
> Acked-by: Mohammad Abdul Awal
> Acked-by: Declan Doh
Instead of writing back descriptors chains in order, let's
write the first chain flags last in order to improve batching.
With Kernel's pktgen benchmark, ~3% performance gain is measured.
Signed-off-by: Maxime Coquelin
---
V2:
Revert back to initial implementation to have a write
barrier before
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> Signed-off-by: Konstantin Ananyev
> Acked-by: Mohammad Abdul Awal
> Acked-by: Declan Doherty
> ---
> lib/librte_net/rte_esp.h | 10 +-
> 1 file changed, 9 insertions(+), 1 deletion(-)
>
> diff --git a/lib/librte_net/rte_esp.h b/lib/
Hi,
I'm having a hard time compiling Pktgen on a Red Hat machine. The source of
the problem is the Lua library.
I'm using Pktgen 3.5.9, and it requires Lua 5.3. Standard installation of
Lua on RHEL ends up having Lua 5.1. So for having the latest release, I
need to compile Lua from its source. My
On Tue, Dec 11, 2018 at 02:48:04PM +0100, Maxime Coquelin wrote:
This patch improves both descriptors dequeue and refill,
by using the same bathing strategy as done in in-order path.
s/bathing/batching/
Signed-off-by: Maxime Coquelin
---
drivers/net/virtio/virtio_rxtx.c | 237 ++
- add missing APIs to map file
- add experimental tag to all bbdev APIs
Signed-off-by: Amr Mokhtar
---
lib/librte_bbdev/rte_bbdev.h | 8
lib/librte_bbdev/rte_bbdev_op.h| 12 ++--
lib/librte_bbdev/rte_bbdev_version.map | 4
3 files changed, 14 insertions(
On 19-Dec-18 9:24 AM, Gaëtan Rivet wrote:
On Tue, Dec 18, 2018 at 02:50:30PM +, Ferruh Yigit wrote:
On 12/18/2018 10:26 AM, Gaetan Rivet wrote:
-Wmisleading-indentation was introduced in GCC 6.0.
It seems '-Wmisleading-indentation' is part of -Wall, which we already set by
default. If so
> -Original Message-
> From: Chas Williams [mailto:3ch...@gmail.com]
> Sent: Tuesday, November 27, 2018 4:56 AM
> To: dev@dpdk.org
> Cc: Ananyev, Konstantin ; Chas Williams
>
> Subject: [PATCH] ip_frag: fix ipv6 when MTU sizes not aligned to 8 bytes
>
> From: Chas Williams
>
> The s
> -Original Message-
> From: dev On Behalf Of Jens Freimann
> Sent: Wednesday, December 19, 2018 5:25 PM
> To: Maxime Coquelin
> Cc: dev@dpdk.org; tiwei@intel.com; zhihong.w...@intel.com
> Subject: Re: [dpdk-dev] [PATCH v2 1/3] net/virtio: inline refill and offload
> helpers
>
> O
On 2018-12-19 07:37, Honnappa Nagarahalli wrote:
diff --git a/lib/librte_eal/rte_eal_version.map
b/lib/librte_eal/rte_eal_version.map
index 3fe78260d..8b1593dd8 100644
--- a/lib/librte_eal/rte_eal_version.map
+++ b/lib/librte_eal/rte_eal_version.map
@@ -355,6 +355,8 @@ EXPERIMENTAL {
rte_
> -Original Message-
> From: Thomas Monjalon [mailto:tho...@monjalon.net]
> Sent: Tuesday, December 18, 2018 8:20 PM
> To: Dumitrescu, Cristian
> Cc: Ananyev, Konstantin ; Pattan, Reshma
> ; dev@dpdk.org;
> jerin.ja...@caviumnetworks.com; Singh, Jasvinder
> ; david.march...@redhat.com;
Hi Cristian,
> > > 18/12/2018 14:19, Dumitrescu, Cristian:
> > > > From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > > > > 18/12/2018 12:18, Dumitrescu, Cristian:
> > > > > > > > I replied in v3 that it should stay in rte_meter.h.
> > > > > > > > You can include rte_meter.h in ethdev.
> > >
19/12/2018 11:47, Dumitrescu, Cristian:
> From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > 18/12/2018 20:34, Dumitrescu, Cristian:
> > > From: Ananyev, Konstantin
> > > > From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > > > > 18/12/2018 14:19, Dumitrescu, Cristian:
> > > > > > From: Tho
> 18/12/2018 20:34, Dumitrescu, Cristian:
> > From: Ananyev, Konstantin
> > > From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > > > 18/12/2018 14:19, Dumitrescu, Cristian:
> > > > > From: Thomas Monjalon [mailto:tho...@monjalon.net]
> > > > > > 18/12/2018 12:18, Dumitrescu, Cristian:
> > >
On Wed, Dec 19, 2018 at 10:26:10AM +, Gavin Hu (Arm Technology China) wrote:
-Original Message-
From: dev On Behalf Of Jens Freimann
Sent: Wednesday, December 19, 2018 5:25 PM
To: Maxime Coquelin
Cc: dev@dpdk.org; tiwei@intel.com; zhihong.w...@intel.com
Subject: Re: [dpdk-dev
> -Original Message-
> From: Mattias Rönnblom [mailto:hof...@lysator.liu.se]
> Sent: Wednesday, December 19, 2018 10:28 AM
> To: Honnappa Nagarahalli ; Ananyev, Konstantin
> ; dev@dpdk.org
> Cc: nd ; Gavin Hu (Arm Technology China)
> Subject: Re: [dpdk-dev] [PATCH 1/2] rwlock: introduce
On Wed, Dec 19, 2018 at 10:56:09AM +, Ananyev, Konstantin wrote:
>
>
> > -Original Message-
> > From: Mattias Rönnblom [mailto:hof...@lysator.liu.se]
> > Sent: Wednesday, December 19, 2018 10:28 AM
> > To: Honnappa Nagarahalli ; Ananyev,
> > Konstantin ; dev@dpdk.org
> > Cc: nd ; Gav
> -Original Message-
> From: Ananyev, Konstantin
> Sent: Wednesday, December 19, 2018 10:49 AM
> To: Dumitrescu, Cristian ; Thomas Monjalon
>
> Cc: Pattan, Reshma ; dev@dpdk.org;
> jerin.ja...@caviumnetworks.com; Singh, Jasvinder
> ; david.march...@redhat.com;
> olivier.m...@6wind.com
>
On Tue, Dec 11, 2018 at 02:48:04PM +0100, Maxime Coquelin wrote:
> This patch improves both descriptors dequeue and refill,
> by using the same bathing strategy as done in in-order path.
>
> Signed-off-by: Maxime Coquelin
> ---
> drivers/net/virtio/virtio_rxtx.c | 237 ---
This patch fixes I40E RxTx module to use the proper config setting,
CONFIG_RTE_LIBRTE_I40E_INC_VECTOR.
Fixes: 9ed94e5bb04e ("i40e: add vector Rx")
Signed-off-by: Rami Rosen
---
drivers/net/i40e/i40e_rxtx.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/net/i40e/i40e
On Tue, Dec 11, 2018 at 02:48:03PM +0100, Maxime Coquelin wrote:
> This patch adds support for in-order path when meargeable buffers
> feature hasn't been negotiated.
>
> Signed-off-by: Maxime Coquelin
> ---
> drivers/net/virtio/virtio_ethdev.c | 11 +++
> drivers/net/virtio/virtio_ethde
Different NIC ports can have different numbers of xstats on them, which
means that we can't just use the xstats list from the first port registered
in the telemetry library. Instead, we need to check the type of each port -
by checking its ops structure pointer - and register each port type once
wi
On 12/19/18 12:18 PM, Tiwei Bie wrote:
On Tue, Dec 11, 2018 at 02:48:04PM +0100, Maxime Coquelin wrote:
This patch improves both descriptors dequeue and refill,
by using the same bathing strategy as done in in-order path.
Signed-off-by: Maxime Coquelin
---
drivers/net/virtio/virtio_rxtx.c
On 12/19/18 11:53 AM, Jens Freimann wrote:
On Wed, Dec 19, 2018 at 10:26:10AM +, Gavin Hu (Arm Technology
China) wrote:
-Original Message-
From: dev On Behalf Of Jens Freimann
Sent: Wednesday, December 19, 2018 5:25 PM
To: Maxime Coquelin
Cc: dev@dpdk.org; tiwei@intel.com
On 12/19/18 10:47 AM, Jens Freimann wrote:
On Tue, Dec 11, 2018 at 02:48:04PM +0100, Maxime Coquelin wrote:
This patch improves both descriptors dequeue and refill,
by using the same bathing strategy as done in in-order path.
s/bathing/batching/
Signed-off-by: Maxime Coquelin
---
driver
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> Introduce librte_ipsec library.
> The library is supposed to utilize existing DPDK crypto-dev and
> security API to provide application with transparent IPsec processing API.
> That initial commit provides some base API to manage
> IPsec Security
19/12/2018 13:08, Akhil Goyal:
> On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> > --- a/MAINTAINERS
> > +++ b/MAINTAINERS
> > @@ -1071,6 +1071,11 @@ F: doc/guides/prog_guide/pdump_lib.rst
> > F: app/pdump/
> > F: doc/guides/tools/pdump.rst
> >
> > +IPsec - EXPERIMENTAL
> > +M: Konstantin
On Wed, Dec 19, 2018 at 10:12:06AM +0100, David Marchand wrote:
> On Tue, Dec 18, 2018 at 1:27 PM David Marchand
> wrote:
>
> >
> > On Tue, Dec 18, 2018 at 1:25 PM Neil Horman wrote:
> >
> >> On Tue, Dec 18, 2018 at 11:41:34AM +0100, David Marchand wrote:
> >>
> >> > The real issue in the end is
On Tuesday, December 12/18/18, 2018 at 23:55:26 +0530, Ferruh Yigit wrote:
> On 12/13/2018 3:02 PM, Rahul Lakkireddy wrote:
> > Replace "args..." with "fmt, ..." and directly use __VA_ARGS__.
>
> What do you mean exactly by "for Windows"? Which compiler? Is there a
> specific C
> standard version
> -Original Message-
> From: Richardson, Bruce
> Sent: Wednesday, December 19, 2018 11:00 AM
> To: Ananyev, Konstantin
> Cc: Mattias Rönnblom ; Honnappa Nagarahalli
> ; dev@dpdk.org; nd
> ; Gavin Hu (Arm Technology China)
> Subject: Re: [dpdk-dev] [PATCH 1/2] rwlock: introduce 'try' s
On Tuesday, December 12/18/18, 2018 at 23:56:22 +0530, Ferruh Yigit wrote:
> On 12/13/2018 3:02 PM, Rahul Lakkireddy wrote:
> > Redefine symbols only when they are not available for Windows
> >
>
> Again "for Windows" is very vague, can you please add more context?
>
These patches are a pre-req
On Tuesday, December 12/18/18, 2018 at 23:53:38 +0530, Ferruh Yigit wrote:
> On 12/13/2018 3:02 PM, Rahul Lakkireddy wrote:
> > Explicitly state header file location using relative paths.
> >
> > Signed-off-by: Rahul Lakkireddy
> > ---
> > drivers/net/cxgbe/Makefile | 2 --
> > drivers/n
The original document written from the point of ABI versioning but later
additions make document confusing, convert document into a ABI/API
policy documentation and organize the document in subsections:
- ABI/API Deprecation
- Experimental APIs
- Library versioning
- ABI versioning
Aim to clarify
Define '__rte_deprecated' usage process.
Suggests keeping old API with '__rte_deprecated' marker until next LTS.
Signed-off-by: Ferruh Yigit
---
Cc: Luca Boccassi
Cc: Kevin Traynor
Cc: Yongseok Koh
Cc: Neil Horman
---
doc/guides/contributing/versioning.rst | 9 +
1 file changed, 9 i
18/12/2018 11:18, Akhil Goyal:
>
> On 11/15/2018 10:54 PM, Fan Zhang wrote:
> > This patch adds a sample power on self-test to fips_validate
> > sample application.
> >
> > Signed-off-by: Fan Zhang
> > ---
> >
> Applied to dpdk-next-crypto
Dropped from master because it does not compile with cla
Hi Konstantin,
Sorry for a late review. I was on unplanned leave for more than 2 weeks.
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> Introduce Security Association (SA-level) data-path API
> Operates at SA level, provides functions to:
> - initialize/teardown SA object
> - process
11/12/2018 13:29, Fan Zhang:
> From: Lukasz Krakowiak
>
> This patch updates the aesni_mb to use IMB_* arch independent
> macros to reduce the code size and future maintaining effort.
>
> Signed-off-by: Fan Zhang
> Signed-off-by: Lukasz Krakowiak
> ---
> v2:
> - making the PMD compatible with
On 12/19/2018 7:07 AM, Xiao Liang wrote:
> 'ndo_change_mtu_rh74' was changed to 'ndo_change_mtu' in RHEL8.
>
> Build error log:
> /home/dpdk-18.11/kernel/linux/kni/compat.h:107:24: error: ‘const struct
> net_device_ops’ has no member named ‘ndo_change_mtu_rh74’; did you mean
> ‘ndo_change_mtu’?
>
On 11/27/2018 1:38 PM, Ferruh Yigit wrote:
> On 11/27/2018 11:13 AM, Mohammed Gamal wrote:
>> On Tue, 2018-11-27 at 10:55 +, Ferruh Yigit wrote:
>>> On 11/26/2018 8:45 PM, Mohammed Gamal wrote:
As RHEL8 beta has been released and is based on recent upstream
kernels
which don't us
On 12/19/2018 1:26 PM, Ferruh Yigit wrote:
> On 12/19/2018 7:07 AM, Xiao Liang wrote:
>> 'ndo_change_mtu_rh74' was changed to 'ndo_change_mtu' in RHEL8.
>>
>> Build error log:
>> /home/dpdk-18.11/kernel/linux/kni/compat.h:107:24: error: ‘const struct
>> net_device_ops’ has no member named ‘ndo_chan
Add a little script to get the symbols of a given section looking at a
library map file, then, on installation, inspect the sources headers to
check that the declaration of the EXPERIMENTAL symbols are prefixed with
a __rte_experimental tag.
Signed-off-by: David Marchand
---
For now, the check i
On 12/19/2018 12:39 PM, Rahul Lakkireddy wrote:
> On Tuesday, December 12/18/18, 2018 at 23:55:26 +0530, Ferruh Yigit wrote:
>> On 12/13/2018 3:02 PM, Rahul Lakkireddy wrote:
>>> Replace "args..." with "fmt, ..." and directly use __VA_ARGS__.
>>
>> What do you mean exactly by "for Windows"? Which c
On Wed, Dec 19, 2018 at 1:39 PM Neil Horman wrote:
> On Wed, Dec 19, 2018 at 10:12:06AM +0100, David Marchand wrote:
> > On Tue, Dec 18, 2018 at 1:27 PM David Marchand <
> david.march...@redhat.com>
> > wrote:
> > But my script still needs some work to make it lighter...
> > A fresh build went fr
On 12/19/2018 7:07 AM, Xiao Liang wrote:
> 'ndo_change_mtu_rh74' was changed to 'ndo_change_mtu' in RHEL8.
>
> Build error log:
> /home/dpdk-18.11/kernel/linux/kni/compat.h:107:24: error: ‘const struct
> net_device_ops’ has no member named ‘ndo_change_mtu_rh74’; did you mean
> ‘ndo_change_mtu’?
>
Hi Thomas,
Sorry the patch caused the problem.
I have tested the patch with intel-ipsec-mb 0.50/0.52 library with make and did
not find problem.
Once switching between versions of ipsec-mb, a necessary "make uninstall" has
to be done to clear old in /usr.
However I didn't test meson in 0.50, s
Add rte_eth_read_clock to read the raw clock of a devide.
The main use is to get the device clock conversion co-efficients to be
able to translate the raw clock of the timestamp field of the pkt mbuf
to a local synced time value.
This function was missing to allow users to convert the RX timestam
Some NICs allow to timestamp packets, but do not support the full
PTP synchronization process. Hence, the value set in the mbuf
timestamp field is only the raw value of an internal clock.
To make sense of this value, one at least needs to be able to query
the current hardware clock value. As with
Signed-off-by: Tom Barbette
---
drivers/net/mlx5/Makefile | 2 +-
drivers/net/mlx5/meson.build | 2 +-
drivers/net/mlx5/mlx5.c| 1 +
drivers/net/mlx5/mlx5.h| 1 +
drivers/net/mlx5/mlx5_ethdev.c | 30 ++
drivers/net/mlx5/mlx5_glue.c | 8 +++
Use rxtx callback to demonstrate a way to use rte_eth_read_clock to
convert the hardware timestamps to an amount of cycles.
This allows to get the amount of time the packet spent since its entry
in the device. While the regular latency only shows the latency from
when it entered the software stack
> On Dec 18, 2018, at 7:53 PM, Hossein Golestani wrote:
>
> Hi,
>
> I'm having a hard time compiling Pktgen on a Red Hat machine. The source of
> the problem is the Lua library.
>
> I'm using Pktgen 3.5.9, and it requires Lua 5.3. Standard installation of
> Lua on RHEL ends up having Lua 5.1.
On Wednesday, December 12/19/18, 2018 at 19:15:39 +0530, Ferruh Yigit wrote:
> On 12/19/2018 12:39 PM, Rahul Lakkireddy wrote:
> > On Tuesday, December 12/18/18, 2018 at 23:55:26 +0530, Ferruh Yigit wrote:
> >> On 12/13/2018 3:02 PM, Rahul Lakkireddy wrote:
> >>> Replace "args..." with "fmt, ..." a
>
> >
> > This patch targets 19.02 release.
> >
> > Introduce rte_rwlock_read_trylock() and rte_rwlock_write_trylock().
> > Signed-off-by: Konstantin Ananyev
> > ---
> > .../common/include/generic/rte_rwlock.h | 54 +++
> > lib/librte_eal/rte_eal_version.map| 2
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> Provide implementation for rte_ipsec_pkt_crypto_prepare() and
> rte_ipsec_pkt_process().
> Current implementation:
> - supports ESP protocol tunnel mode.
> - supports ESP protocol transport mode.
> - supports ESN and replay window.
> - su
Added new rte_color definition in librte_meter to
consolidate color definition which is currently replicated
in various places such as rte_meter.h, rte_tm.h and rte_mtr.h
Created aliases for rte_tm_color, rte_mtr_color and rte_meter_color
to use new rte_color values.
The definitions of rte_tm_col
This patch implements the changes proposed in the deprecation
notes [1][2].
librte_mbuf changes:
The mbuf::hash::sched field is updated to support generic
definition in line with the ethdev TM and MTR APIs. The new generic
format contains: queue ID, traffic class, color.
Added public APIs to set
Added new rte_color definition in librte_meter to
consolidate color definition which is currently replicated
in various places such as rte_meter.h, rte_tm.h and rte_mtr.h
Created aliases for rte_tm_color, rte_mtr_color and rte_meter_color
to use new rte_color values.
The definitions of rte_tm_col
This patch implements the changes proposed in the deprecation
notes [1][2].
librte_mbuf changes:
The mbuf::hash::sched field is updated to support generic
definition in line with the ethdev TM and MTR APIs. The new generic
format contains: queue ID, traffic class, color.
Added public APIs to set
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> Introduce helper functions to process completed crypto-ops
> and group related packets by sessions they belong to.
>
> Signed-off-by: Konstantin Ananyev
> Acked-by: Declan Doherty
> ---
> lib/librte_ipsec/Makefile | 1 +
> lib
On Wed, Dec 19, 2018 at 09:21:10AM +0100, Maxime Coquelin wrote:
> A read barrier is required to ensure the ordering between
> available index and the descriptor reads is enforced.
>
> 1. read avail_head = avail->idx
> 2. read cur_idx = last_avail_idx
> if (cur_idx != avail_head) {
> 3. read i
This patch adds checksum support in the ISA-L PMD for both compression
and decompression.
CRC32 is supported as well as Adler32.
V2:
Documentation Changes
V3:
Added Release note
V4:
Removed dependency on offset unit test
V5:
If checksum type is incorrect, compression will continue without checksu
This patch adds a test which examines what type of checksum the PMD
supports, adler, crc32 or alder32_crc32
and tests that feature if the PMD supports it.
V2:
Documentation Changes
V3:
Added Release note
V4:
Removed dependency on offset unit test
V5:
If checksum type is incorrect, compression will
This updates the ISA-L compression driver guide on how to enable and use
checksums and release notes.
Signed-off-by: Lee Daly
Acked-by: Fiona Trahe
---
doc/guides/compressdevs/features/isal.ini | 2 ++
doc/guides/compressdevs/isal.rst | 30 --
doc/guides/re
On Wed, Dec 19, 2018 at 09:21:13AM +0100, Maxime Coquelin wrote:
> This prefetch does not show any performance improvement.
>
> Signed-off-by: Maxime Coquelin
Likely because of the RMB.
Try prefetching the *next* descriptor maybe?
> ---
> lib/librte_vhost/virtio_net.c | 2 --
> 1 file changed,
On 12/19/18 4:47 PM, Michael S. Tsirkin wrote:
On Wed, Dec 19, 2018 at 09:21:10AM +0100, Maxime Coquelin wrote:
A read barrier is required to ensure the ordering between
available index and the descriptor reads is enforced.
1. read avail_head = avail->idx
2. read cur_idx = last_avail_idx
if
On Wed, Dec 19, 2018 at 09:21:09AM +0100, Maxime Coquelin wrote:
> This series adds missing read barriers after reading avail index
> for split ring and desc flags for packed ring.
>
> Also, it turns out that some descriptors prefetching are either
> badly placed, or useless, last part of the seri
On 12/14/2018 9:53 PM, Konstantin Ananyev wrote:
> +static struct unit_test_suite ipsec_testsuite = {
> + .suite_name = "IPsec NULL Unit Test Suite",
> + .setup = testsuite_setup,
> + .teardown = testsuite_teardown,
> + .unit_test_cases = {
> + TEST_CASE_ST(ut_setup,
On 12/14/2018 9:57 PM, Konstantin Ananyev wrote:
> Add IPsec library guide and update release notes.
>
> Signed-off-by: Bernard Iremonger
> Signed-off-by: Konstantin Ananyev
> ---
> doc/guides/prog_guide/index.rst| 1 +
> doc/guides/prog_guide/ipsec_lib.rst| 74 +
rte_timer_manage() adds expired timers to a "run list", and walks the
list, transitioning each timer from the PENDING to the RUNNING state.
If another lcore resets or stops the timer at precisely this
moment, the timer state would instead be set to CONFIG by that other
lcore, which would cause time
On Wed, Dec 19, 2018 at 10:16:24AM +0100, Maxime Coquelin wrote:
>
>
> On 12/12/18 7:53 PM, Michael S. Tsirkin wrote:
> > On Wed, Dec 12, 2018 at 05:34:31PM +0100, Maxime Coquelin wrote:
> > > Hi Ilya,
> > >
> > > On 12/12/18 4:23 PM, Ilya Maximets wrote:
> > > > On 12.12.2018 11:24, Maxime Coqu
> > > However, if a timer expires quickly enough, there exists the
> > > following race condition that causes the timer_manage() routine to
> > > misinterpret a timer in CONFIG state, resulting in lost timers:
> > >
> > > - Thread A:
> > > - starts a timer with rte_timer_reset()
> > > - the tim
> > > + * try to take a write lock.
> > > + *
> > > + * @param rwl
> > > + * A pointer to a rwlock structure.
> > > + * @return
> > > + * - zero if the lock is successfully taken
> > > + * - -EBUSY if lock could not be acquired for writing because
> > > + * it was already locked for re
On 19.12.2018 18:50, Michael S. Tsirkin wrote:
> On Wed, Dec 19, 2018 at 09:21:09AM +0100, Maxime Coquelin wrote:
>> This series adds missing read barriers after reading avail index
>> for split ring and desc flags for packed ring.
>>
>> Also, it turns out that some descriptors prefetching are eith
This series of patches fix compilation warning and errors seen while
building CXGBE PMD in preparation for Microsoft Windows OS support.
Patch 1 uses relative paths for including header files.
Patch 2 fixes macros related to rte_log().
Patch 3 only redefines symbols when they are not available.
Pa
The Intel C++ compiler is not able to locate the header files without
relative paths in Windows build. Following errors are seen for these
header files.
C:\> base\common.h(9): error : cannot open source file "cxgbe_compat.h"
#include "cxgbe_compat.h"
^
[...
Replace "args..." with "fmt, ..." and directly use __VA_ARGS__.
This fixes following errors reported by Intel C++ compiler in Windows
build.
C:\> cxgbe_compat.h(28): error : expected a ")"
#define dev_printf(level, fmt, args...) \
^
C:\> cxgbe_compa
Define symbols only when they are not available.
This fixes following types of issues reported by Intel C++ compiler
in Windows build.
C:\> cxgbe_compat.h(154): warning #47: incompatible redefinition of
macro "min"
#define min(a, b) RTE_MIN(a, b)
^
C:\> t4_hw.c(338): warn
Fix following build errors reported by Intel C++ compiler in Windows
build.
C:\> t4_hw.c(5105): warning #147: declaration is incompatible with
"int t4_bar2_sge_qregs(struct adapter *, unsigned int, unsigned int,
u64={uint64_t={unsigned __int64}} *, unsigned int *)"
(declared at line 524 of "..\..\
On Wed, Dec 19, 2018 at 10:29:52AM +0100, Maxime Coquelin wrote:
> Instead of writing back descriptors chains in order, let's
> write the first chain flags last in order to improve batching.
>
> With Kernel's pktgen benchmark, ~3% performance gain is measured.
>
> Signed-off-by: Maxime Coquelin
Hi Shally,
> -Original Message-
> From: Verma, Shally [mailto:shally.ve...@cavium.com]
> Sent: Tuesday, December 18, 2018 10:48 PM
> To: Trahe, Fiona ; Stephen Hemminger
>
> Cc: dev@dpdk.org; akhil.go...@nxp.com; Jozwiak, TomaszX
> ; Gupta,
> Ashish ; Daly, Lee ; Luse, Paul E
>
> Subje
> On Dec 17, 2018, at 9:01 AM, Wiles, Keith wrote:
>
>
>
>> On Dec 17, 2018, at 5:45 AM, Thomas Monjalon wrote:
>>
>> Hi Keith,
>>
>> 16/12/2018 18:46, Keith Wiles:
>>> DFS stands for DPDK Filesystem, which helps expose data
>>> and control files in a FUSE based filesystem. The dfs require
Introduce rte_rwlock_read_trylock() and rte_rwlock_write_trylock().
Signed-off-by: Konstantin Ananyev
Reviewed-by: Honnappa Nagarahalli
Reviewed-by: Gavin Hu
---
.../common/include/generic/rte_rwlock.h | 56 +++
1 file changed, 56 insertions(+)
diff --git a/lib/librte_ea
Introduce rte_rwlock_read_trylock() and rte_rwlock_write_trylock()
and new UT test-case for it.
v1 -> v2
Changes per Honnappa and Gavin comments:
- remove cycle in rte_rwlock_write_trylock()
- remove static inline functions from .map file
- update comments
Konstantin Ananyev (2):
rwlock: in
Add few functional and perfomance tests
for rte_rwlock_read_trylock() and rte_rwlock_write_trylock().
Signed-off-by: Konstantin Ananyev
Reviewed-by: Gavin Hu
---
test/test/test_rwlock.c | 409 ++--
1 file changed, 390 insertions(+), 19 deletions(-)
diff --gi
1 - 100 of 168 matches
Mail list logo
