On Thu, Jun 11, 2020 at 12:02 PM Thomas Monjalon wrote:
>
> Since dynamic fields and flags were added in 19.11,
> the idea was to use them for new features, not only PMD-specific.
>
> The guideline is made more explicit in doxygen, in the mbuf guide,
> and in the contribution design guidelines.
>
Since dynamic fields and flags were added in 19.11,
the idea was to use them for new features, not only PMD-specific.
The guideline is made more explicit in doxygen, in the mbuf guide,
and in the contribution design guidelines.
For more information about the original design, see the presentation
This patch support RSS action in flow query.
It can display the RSS configuration of the specified rule.
Signed-off-by: Chenxu Di
---
app/test-pmd/config.c | 55 +++
1 file changed, 55 insertions(+)
diff --git a/app/test-pmd/config.c b/app/test-pmd/config
The legacy filter API will be superseded by rte_flow.
There are also several small features which can not be
implemented in rte_flow. This patch re-implemented these
features as private API.
Two APIs are added:
rte_pmd_ixgbe_get_fdir_info.
rte_pmd_ixgbe_get_fdir_stats.
Signed-off-by: Chenxu Di
--
The legacy filter API will be superseded by rte_flow.
There are also several small features which can not be
implemented in rte_flow.
This patch re-implemented features get_fdir_info, get_fdir_stat and
set_gre_key_len as private API.
This patch also enable RSS action in flow query.
Chenxu Di (5):
This patch enables flow query function to get the
configuration ofthe specified rule.
Signed-off-by: Chenxu Di
---
drivers/net/i40e/i40e_flow.c | 51
1 file changed, 51 insertions(+)
diff --git a/drivers/net/i40e/i40e_flow.c b/drivers/net/i40e/i40e_flow.c
in
The legacy filter API will be superseded. This patch use
private api to change the implementation of commands
global_config gre-key-len and
show port fdir
Signed-off-by: Chenxu Di
---
app/test-pmd/cmdline.c | 23 -
app/test-pmd/config.c | 74 +++---
The legacy filter API will be superseded by rte_flow.
There are also several small features which can not be
implemented in rte_flow. This patch re-implemented these
features as private API.
Three APIs are added:
rte_pmd_i40e_get_fdir_info.
rte_pmd_i40e_get_fdir_stats.
rte_pmd_i40e_set_gre_key_len.
Ping
Anybody volunteer to make a v2 please?
24/05/2020 17:34, Thomas Monjalon:
> Somnath, why neither update, nor reply to reviews after February 7?
> It is discouraging for reviewers.
>
>
> 24/04/2020 20:24, Thomas Monjalon:
> > Please Somnath, it is waiting for a v2.
> >
> >
> > 26/02/2020
For SNOW and ZUC algos the offset value for enryption and decryption
is converted to bytes. Hence RTE_CRYPTODEV_FF_NON_BYTE_ALIGNED_DATA
feature is not supported by the octeontx2 crypto pmd.
Fixes: 51f3e107aca2 ("crypto/octeontx2: enable non-byte aligned data")
Signed-off-by: Ankur Dwivedi
---
From: Anoob Joseph
This patch enables the generic crypto tests for OCTEON TX PMDs.
Reomves the PMD specific tests.
Signed-off-by: Ankur Dwivedi
Signed-off-by: Anoob Joseph
---
app/test/test_cryptodev.c | 769 +-
1 file changed, 2 insertions(+), 767
For SNOW and ZUC algos the offset value for enryption and decryption
is converted to bytes. Hence RTE_CRYPTODEV_FF_NON_BYTE_ALIGNED_DATA
feature is not supported by the octeontx crypto pmd.
Fixes: 32b8f26adf8b ("crypto/octeontx: enable non-byte aligned data")
Signed-off-by: Ankur Dwivedi
---
do
This patch series enables the generic auto test crypto framework for the
OCTEON TX PMDs. The patch series also resolves the bugs observed while
testing these test cases.
Ankur Dwivedi (6):
common/cpt: fix encryption offset
crypto/octeontx: remove stats callback
crypto/octeontx: set session p
This patch sets the first 32 bytes of session private data
to zero. This prevents garbage data to be used in code logic.
Signed-off-by: Ankur Dwivedi
---
drivers/crypto/octeontx/otx_cryptodev_ops.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/crypto/octeontx/otx_cryptodev_ops.c
In case of gmac auth the encryption offset should be set to zero.
Fixes: b74652f3a91f ("common/cpt: add microcode interface for encryption")
Fixes: 177b41ceee61 ("common/cpt: add microcode interface for decryption")
Signed-off-by: Ankur Dwivedi
---
drivers/common/cpt/cpt_ucode.h | 24 ++
The stats get and reset functions for octeontx crypto PMD are
unimplemented. So removing them.
Signed-off-by: Ankur Dwivedi
---
drivers/crypto/octeontx/otx_cryptodev_ops.c | 17 ++---
1 file changed, 2 insertions(+), 15 deletions(-)
diff --git a/drivers/crypto/octeontx/otx_cryptodev
This patch sets the first 32 bytes of session private data
to zero. This prevents garbage data to be used in code logic.
Signed-off-by: Ankur Dwivedi
---
drivers/crypto/octeontx2/otx2_cryptodev_ops.c | 3 +++
1 file changed, 3 insertions(+)
diff --git a/drivers/crypto/octeontx2/otx2_cryptodev_o
This patch enables the filter that supports
to create following two rules for the same packet type:
One is to select source port only as input set and the
other is for destination port only.
Signed-off-by: Guinan Sun
---
doc/guides/rel_notes/release_20_08.rst | 7 +
drivers/net/i40e/i40e_ethde
This series is to optimize the virtio performance by using restrict
pointer aliasing, which allows the compiler to vectorize loops more
aggressively.
The patches were benchmarked by running PVP case on ThunderX2 platform
and showed positive performance results.
Joyce Kong (2):
net/virtio: restr
Restrict pointer aliasing to allow the compiler to vectorize
loops more aggressively.
With this patch, a 9.6% improvement is observed in throughput for the
packed virtio-net PVP case, and a 2.8% improvement in throughput for
the packed virtio-user PVP case. All performance data are measured
under
Restrict pointer aliasing to allow the compiler to vectorize loops
more aggressively.
With this patch, a 9.6% improvement is observed in throughput for
the virtio-net PVP case, and a 2.4% perf improvement in throughput
for the virtio-user PVP case. All performance data are measured
under the 0.001
Hi Maxime,
> -Original Message-
> From: dev On Behalf Of Maxime Coquelin
> Sent: Thursday, May 14, 2020 4:02 PM
> To: Ye, Xiaolong ; shah...@mellanox.com;
> ma...@mellanox.com; amore...@redhat.com; Wang, Xiao W
> ; viachesl...@mellanox.com; dev@dpdk.org
> Cc: jasow...@redhat.com; l...@red
From: Qi Zhang
Add support to allow the original VF actions in DCF.
Signed-off-by: Qi Zhang
Signed-off-by: Ting Xu
---
v1->v2: correct commit log.
---
drivers/net/ice/ice_switch_filter.c | 13 ++---
1 file changed, 10 insertions(+), 3 deletions(-)
diff --git a/drivers/net/ice/ice_sw
We should make sure off + size < sizeof(struct rte_mbuf) to avoid
possible out-of-bounds access of free_space array, there is no issue
currently due to the low bits of free_flags (which is adjacent to
free_space) are always set to 0. But we shouldn't rely on it since it's
fragile and layout of stru
On 06/09, Qi Zhang wrote:
>Upgrade base with below main changes
>
>1. add more protocol support - L2TPv3/ESP/AH/PFCP.
>2. Add IPv6 prefix RSS support.
>3. support GENEVE and VXLAN rules with VLAN.
>4. Add E823L device IDs
>5. couple bug fixes and code clean.
>
>v2:
>- fix a bug in patch that cause
This patch adds Chacha20-Poly1305 implementation to
cryptodev tests.
Signed-off-by: Arek Kusztal
---
v4:
- resent in 20.08 DPDK window
app/test/test_cryptodev.c | 18 ++-
app/test/test_cryptodev_aead_test_vectors.h | 75 +
2 files changed, 92 in
This patchset adds Chacha20-Poly1305 implementation to Intel
QuickAssist Technology pmd.
Signed-off-by: Arek Kusztal
---
v4:
- resent in 20.08 DPDK window
This patch depends on "crypto/qat: add handling of multi process[1]."
[1] http://patchwork.dpdk.org/patch/71213/
doc/guides/cryptodevs/fea
This patch improves handling of multi process applications
using Intel QuickAssist Technology PMD.
Signed-off-by: Arek Kusztal
---
drivers/common/qat/qat_device.c | 103
drivers/common/qat/qat_device.h | 69 +---
drivers/common/qa
From: Fiona Trahe
This patch adds traces to some Cryptodev functions that are used
in primary/secondary context.
Signed-off-by: Fiona Trahe
---
lib/librte_cryptodev/rte_cryptodev.c | 22 --
1 file changed, 20 insertions(+), 2 deletions(-)
diff --git a/lib/librte_cryptodev/
From: Fiona Trahe
This patch adds function that can check if queue pair
was already setup. This may be useful when dealing with
multi process approach in cryptodev.
Signed-off-by: Fiona Trahe
---
lib/librte_cryptodev/rte_cryptodev.c | 28
lib/librte_cryptodev/rte_c
On 10/06/20 19:08 +0200, Maxime Coquelin wrote:
> Hi Gaëtan,
>
> On 6/10/20 2:06 PM, Gaëtan Rivet wrote:
> > Hello Maxime,
> >
> > On 08/06/20 17:53 +0200, Maxime Coquelin wrote:
> >> This patch makes rte_dev_probe() to return the
> >> rte_device pointer on success and NULL on error
> >> instead
This commit introduces mlx5 bus to support multiple class of devices
for a single PCI device.
Motivation and example
--
mlx5 PCI device supports multiple class of devices such as net, vdpa
and regex devices.
Currently only one pmd (either net or vdpa) can bind to this device.
Add mlx5 PCI bus which enables multiple mlx5 drivers to bind to single
pci device.
Signed-off-by: Parav Pandit
---
config/common_base| 6 ++
config/defconfig_arm64-bluefield-linuxapp-gcc | 6 ++
drivers/bus/meson.build | 2 +-
drivers/bus/mlx5
Currently mlx5_class_get() returns enabled single valid class.
To support multiple class and to improve readability of code, change it
to mlx5_class_enabled().
With this function, each class enablement can be checked, to load class
specific driver.
Signed-off-by: Parav Pandit
---
drivers/common/
Create a mlx5 bus driver framework for invoking drivers of
multiple classes who have registered with the mlx5_pci bus
driver.
Validate user class arguments for supported class combinations.
Signed-off-by: Parav Pandit
---
drivers/bus/mlx5_pci/Makefile | 1 +
drivers/bus/mlx5_pci/mes
mlx5 PCI Device supports multiple classes of devices such as net, vdpa,
and/or regex.
To support these multiple classes, change mlx5_class to a
bitmap values so that if users asks to enable multiple of them, all
supported classes can be returned by mlx5_class_supported().
Signed-off-by: Parav Pand
Enable class driver to match with the mlx5 pci devices.
Migrate mlx5 net PMD and vdpa PMD to start using mlx5 common class
driver.
Signed-off-by: Parav Pandit
---
drivers/bus/Makefile| 3 ++
drivers/bus/mlx5_pci/mlx5_pci_bus.c | 60 +
drivers/net/mlx5
There are several drivers which duplicate bit generation macro.
Introduce a generic bit macros so that such drivers avoid redefining
same in multiple drivers.
Signed-off-by: Parav Pandit
---
lib/librte_eal/include/rte_bits.h | 10 ++
1 file changed, 10 insertions(+)
create mode 100644 l
Hi Gaëtan,
On 6/10/20 2:06 PM, Gaëtan Rivet wrote:
> Hello Maxime,
>
> On 08/06/20 17:53 +0200, Maxime Coquelin wrote:
>> This patch makes rte_dev_probe() to return the
>> rte_device pointer on success and NULL on error
>> instead of returning 0 on success and negative
>> value on error.
>>
>> Th
On Wed, 10 Jun 2020 16:48:58 +0100
"Burakov, Anatoly" wrote:
> On 10-Jun-20 3:31 PM, Dmitry Kozlyuk wrote:
> > On Wed, 10 Jun 2020 11:26:22 +0100
> > "Burakov, Anatoly" wrote:
> >
> > [snip]
> > + addr = eal_get_virtual_area(
> > + msl->base_va, &mem_sz, page_sz, 0
From: Kishore Padmanabha
Removed the mark and vfr flags and replaced it with the new
opcode and added the vfr flag process in mark database
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c| 74
From: Kishore Padmanabha
The ulp template database is updated to support the new
combined opcodes.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Venkat Duvvuru
Reviewed-by: Ajit Kumar Khaparde
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_templat
From: Kishore Padmanabha
The fields in the device params structure are renamed to reflect
the usage of those fields.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/bnxt_ulp.c | 12 ++--
drivers/net/bn
From: Mike Baucom
The fields that were setting a constant zero have now been switched to
use the new SET_TO_ZERO opcode as an optimization. The SET_TO_ZERO does
not copy data into the key/result/mask fields, but rather simply
increments the write pointer.
Signed-off-by: Mike Baucom
Reviewed-by
From: Kishore Padmanabha
Added support of internal exact match flows and the action
mark is supported for these flows.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/bnxt.h | 2 ++
drivers/net/bnxt/bnxt_rxr.c
From: Kishore Padmanabha
The implicit update of the egress vnic action bitset for a flow
that does not specify the forwarding port explicitly is removed.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_rte_parser.c |
From: Mike Baucom
The ADD_PAD opcode is now SET_TO_ZERO. This change better reflects the
intent of the opcode and allows it to be used in more circumstances
without overloading the term pad.
Signed-off-by: Mike Baucom
Reviewed-by: Ajit Kumar Khaparde
Signed-off-by: Somnath Kotur
---
drivers
From: Kishore Padmanabha
Added flow db api to get the vf representor action
record for a given flow.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_flow_db.c | 200 +++---
drivers/net/bnxt
From: Shuanglin Wang
User could set max flow count by passing a devarg
"-w :0d:00.0,max_num_kflows=64" to a DPDK application;
The value must be not less than 32K and be power-of-2;
the default value is 32K.
Signed-off-by: Shuanglin Wang
Reviewed-by: Kishore Padmanabha
Signed-off-by: Somnat
From: Venkat Duvvuru
ipv6 vtc_flow contains three fields
1. Version
2. Priority / Traffic Class
3. Flow Label
Currently, these are not parsed separately and also not set separately
in the field bitmap by the flow parser. However, the template treats
them separately. As a result, the flow matching
From: Kishore Padmanabha
The enums in the bnxt_ulp_resource_sub_type are renamed to reflect
the table types explicitly.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_template_db.c | 30 --
From: Kishore Padmanabha
The phy port action handler should get vport details and not
vnic id. The fix is to calculate the vport of the given
port.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_rte_parser.c | 10 +++
From: Kishore Padmanabha
The BNXT_ULP_REGFILE_INDEX_ACTION_PTR_MAIN is renamed to
BNXT_ULP_REGFILE_INDEX_MAIN_ACTION_PTR since it is the main
action pointer.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Somnath Kotur
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net
From: Kishore Padmanabha
The critical resource field in the template table is assigned
enumeration values instead of hard coded values.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c | 14 +++---
From: Kishore Padmanabha
The regfile index shall be used for both write and read operations.
Hence the field is renamed.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Venkat Duvvuru
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c |
From: Kishore Padmanabha
Removed the unsed enum BNXT_ULP_REGFILE_INDEX_CACHE_ENTRY_PTR
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_template_db.h | 7 +++
1 file changed, 3 insertions(+), 4 deletions(-)
diff -
From: Kishore Padmanabha
The index table processing is extended to address encapsulation fields
so that action template index table can be processed by a common index
processing function that can process both class and action index
tables.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael
From: Kishore Padmanabha
The vf representor flag is moved from computed field list to class
table so it can be set in the template details.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Somnath Kotur
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c | 14 ++-
From: Kishore Padmanabha
This is a fix to add the direction bit to the action bitmap during
flow parsing, so that egress flows can be matched to the
template signature.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Somnath Kotur
Reviewed-by: Ajit Kumar Khaparde
Reviewed-by: Michael Baucom
S
From: Kishore Padmanabha
This enables using the action bitmap to update the action result
fields in the flow creation instead of using computed header fields.
An example would be the usage of the vlan pop action bitmap that is
used to updated action result field as part of this commit.
Signed-of
From: Kishore Padmanabha
The cache table id is not needed anymore since the value can
be calculated from resource sub type and direction.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c | 21
From: Kishore Padmanabha
The bnxt_ulp_cache_tbl_id enums are not required any longer
since the index is now calculated using resource sub type
and direction.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Reviewed-by: Ccxsw Build
Signed-off-by: Somnath Kotur
---
drivers/net/b
From: Kishore Padmanabha
Added support for resource sub type to class and action tables
renamed table id to resource type.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c | 38 +--
From: Kishore Padmanabha
The ulp action bitmap enumeration values that
contain open flow string are renamed
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Reviewed-by: Ajit Kumar Khaparde
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_template_db.h | 12 ++--
From: Kishore Padmanabha
Added support for computed header fields in the result field
processing. The computed header fields are fields that are extracted
from header fields or derived from data that is not part of the flow
command but shall be used in setting up of the flow rule.
Signed-off-by:
From: Kishore Padmanabha
The compute field is extended to support action fields and not
just header fields, hence CHF is changed to CF. The access macro
for compute field is renamed to address this.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Somnath Kotur
Signed-off-by: Somnath Kotur
---
From: Kishore Padmanabha
The default identifier list in ulp mapper is extended to support
other truflow resource types and not just identifiers.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Somnath Kotur
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c | 231 +
From: Kishore Padmanabha
Created sub Makefile for tf_ulp and and tf_core
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/Makefile | 29 +
drivers/net/bnxt/tf_core/Makefile | 17 +++
From: Kishore Padmanabha
Remove the unused mem field in the ulp mapper class table structure
Signed-off-by: Kishore Padmanabha
Reviewed-by: Ajit Kumar Khaparde
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c | 2 +-
drivers/net/bnxt
From: Kishore Padmanabha
The ulp mapper changes to support mark actions for non
GFID entries that support only LFID rules.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ulp_mapper.c | 198 +++---
From: Kishore Padmanabha
The svif and vlan information are removed from header bitmap
signature so that the matching algorithm does not use these
fields to perform matching. So flows with or without vlan
tag could use the same flow template.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Kishor
From: Kishore Padmanabha
Add support for the vfr flag to the mark manager
Signed-off-by: Kishore Padmanabha
Reviewed-by: Michael Baucom
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/bnxt_rxr.c| 3 ++-
drivers/net/bnxt/tf_ulp/ulp_mark_mgr.c | 7 +++
drivers/net/bnxt/tf_ulp
From: Kishore Padmanabha
Added support to include more resource functions in the flow
database. The number of bits increased from 3 to 8 for storing
the resource function.
Signed-off-by: Kishore Padmanabha
Reviewed-by: Somnath Kotur
Signed-off-by: Somnath Kotur
---
drivers/net/bnxt/tf_ulp/ul
Max Rx Ring count could be < Max stat contexts. While accounting
for stat contexts, this should be also considered and
the max ring count adjusted accordingly.
Fixes: f03e66cb64ce ("net/bnxt: limit queue count for NS3/Stingray devices")
Cc: sta...@dpdk.org
Signed-off-by: Somnath Kotur
---
drive
On 10-Jun-20 3:31 PM, Dmitry Kozlyuk wrote:
On Wed, 10 Jun 2020 11:26:22 +0100
"Burakov, Anatoly" wrote:
[snip]
+ addr = eal_get_virtual_area(
+ msl->base_va, &mem_sz, page_sz, 0, reserve_flags);
+ if (addr == NULL) {
+#ifndef RTE_EXEC_ENV_WINDOWS
+ /* T
This patchset consists of enhancements for the host based flow table
management mechanism that was introduced in 20.05. Also has some
feature support for offloading vxlan encap/decap and on-chip exact match
flows.
Kishore Padmanabha (30):
bnxt/tf_ulp: Created sub Makefile for tf_ulp and tf_core
On 10-Jun-20 4:28 PM, Stephen Hemminger wrote:
On Wed, 10 Jun 2020 10:24:44 +0100
"Burakov, Anatoly" wrote:
On 09-Jun-20 4:35 PM, Stephen Hemminger wrote:
On Tue, 9 Jun 2020 14:39:54 +0100
"Burakov, Anatoly" wrote:
On 09-Jun-20 2:13 PM, Ferruh Yigit wrote:
On 6/9/2020 1:46 PM, Burakov,
On Wed, Jun 10, 2020 at 8:48 PM David Marchand
wrote:
>
> On Wed, Jun 10, 2020 at 5:09 PM Jerin Jacob wrote:
> >
> > On Wed, Jun 10, 2020 at 8:15 PM David Marchand
> > wrote:
> > >
> > > OVS and some other applications have been hacking into DPDK internals to
> > > fake EAL threads and avoid per
On Wed, 10 Jun 2020 10:24:44 +0100
"Burakov, Anatoly" wrote:
> On 09-Jun-20 4:35 PM, Stephen Hemminger wrote:
> > On Tue, 9 Jun 2020 14:39:54 +0100
> > "Burakov, Anatoly" wrote:
> >
> >> On 09-Jun-20 2:13 PM, Ferruh Yigit wrote:
> >>> On 6/9/2020 1:46 PM, Burakov, Anatoly wrote:
> On
We should make sure off + size < sizeof(struct rte_mbuf) to avoid
possible out-of-bounds access of free_space array, there is no issue
currently due to the low bits of free_flags (which is adjacent to
free_space) are always set to 0. But we shouldn't rely on it since it's
fragile and layout of stru
On Wed, Jun 10, 2020 at 5:09 PM Jerin Jacob wrote:
>
> On Wed, Jun 10, 2020 at 8:15 PM David Marchand
> wrote:
> >
> > OVS and some other applications have been hacking into DPDK internals to
> > fake EAL threads and avoid performance penalty of only having non-EAL
> > threads.
> >
> > This serie
Hi, Harman
> -Original Message-
> From: Harman Kalra
> Sent: Wednesday, June 10, 2020 16:34
> To: Slava Ovsiienko
> Cc: dev@dpdk.org; Thomas Monjalon ; Matan
> Azrad ; Raslan Darawsheh
> ; Ori Kam ;
> olivier.m...@6wind.com; Shahaf Shuler
> Subject: Re: [dpdk-dev] [RFC] mbuf: accurate p
On Wed, Jun 10, 2020 at 08:39:30PM +0530, Jerin Jacob wrote:
> On Wed, Jun 10, 2020 at 8:15 PM David Marchand
> wrote:
> >
> > OVS and some other applications have been hacking into DPDK internals to
> > fake EAL threads and avoid performance penalty of only having non-EAL
> > threads.
> >
> > Thi
On Wed, Jun 10, 2020 at 8:15 PM David Marchand
wrote:
>
> OVS and some other applications have been hacking into DPDK internals to
> fake EAL threads and avoid performance penalty of only having non-EAL
> threads.
>
> This series proposes to add a new type of lcores and maps those external
> threa
Add a little helper to dump all lcores.
Signed-off-by: David Marchand
---
lib/librte_eal/common/eal_common_lcore.c | 32 +++
lib/librte_eal/common/eal_common_thread.c | 16
lib/librte_eal/common/eal_thread.h| 13 +++--
lib/librte_eal/freebsd/eal.c
DPDK allows calling some part of its API from a non-EAL thread but this
has some limitations.
OVS (and other applications) has its own thread management but still
want to avoid such limitations by hacking RTE_PER_LCORE(_lcore_id) and
faking EAL threads potentially unknown of some DPDK component.
I
Now that lcores can be dynamically allocated/freed, we will have to
notify DPDK components and applications of such events for cases where
per lcore context must be allocated/initialised.
Signed-off-by: David Marchand
---
lib/librte_eal/common/eal_common_lcore.c | 91 +++
li
We have per lcore thread symbols scattered in OS implementations but
common code relies on them.
Move all of them in common.
RTE_PER_LCORE(_socket_id) and RTE_PER_LCORE(_cpuset) have public
accessors and are not exported through the library map, they can be
made static.
Signed-off-by: David March
Introduce a helper responsible for initialising the per thread context.
We can then have a unified context for EAL and non-EAL threads and
remove copy/paste'd OS-specific helpers.
Per EAL thread CPU affinity setting is separated from the thread init.
It is to accommodate with Windows EAL where CPU
This is a preparation step for dynamically unregistering external
threads.
Since we explicitly allocate a per thread trace buffer in
rte_thread_init, add an internal helper to free this buffer.
Note: I preferred renaming the current internal function to free all
threads trace buffers (new name tr
Because of the inline accessor + static declaration in rte_gettid(),
we end up with multiple symbols for RTE_PER_LCORE(_thread_id).
Each compilation unit will pay a cost when accessing this information
for the first time.
$ nm build/app/dpdk-testpmd | grep per_lcore__thread_id
0054 d p
OVS and some other applications have been hacking into DPDK internals to
fake EAL threads and avoid performance penalty of only having non-EAL
threads.
This series proposes to add a new type of lcores and maps those external
threads to such lcores.
Those threads won't run the DPDK eal mainloop and
On 6/10/2020 3:30 PM, David Marchand wrote:
> rte_eal_init_alert() already appends a newline.
>
> Fixes: 0a529578f162 ("eal: clean up unused files on initialization")
> Cc: sta...@dpdk.org
>
> Signed-off-by: David Marchand
Reviewed-by: Ferruh Yigit
rte_eal_get_configuration() has been made private in 19.11, remove
leftover in Windows export list.
Fixes: f58cef079b05 ("eal: make the global configuration private")
Signed-off-by: David Marchand
---
lib/librte_eal/rte_eal_exports.def | 1 -
1 file changed, 1 deletion(-)
diff --git a/lib/libr
On Wed, 10 Jun 2020 11:26:22 +0100
"Burakov, Anatoly" wrote:
[snip]
> >>> + addr = eal_get_virtual_area(
> >>> + msl->base_va, &mem_sz, page_sz, 0, reserve_flags);
> >>> + if (addr == NULL) {
> >>> +#ifndef RTE_EXEC_ENV_WINDOWS
> >>> + /* The hint would be misleading on Windows, b
rte_eal_init_alert() already appends a newline.
Fixes: 0a529578f162 ("eal: clean up unused files on initialization")
Cc: sta...@dpdk.org
Signed-off-by: David Marchand
---
lib/librte_eal/freebsd/eal.c | 2 +-
lib/librte_eal/linux/eal.c | 2 +-
2 files changed, 2 insertions(+), 2 deletions(-)
Add hugepages discovery ("large pages" in Windows terminology)
and update documentation for required privilege setup. Only 2MB
hugepages are supported and their number is estimated roughly
due to the lack or unstable status of suitable OS APIs.
Assign myself as maintainer for the implementation fil
Basic memory management supports core libraries and PMDs operating in
IOVA as PA mode. It uses a kernel-mode driver, virt2phys, to obtain
IOVAs of hugepages allocated from user-mode. Multi-process mode is not
implemented and is forcefully disabled at startup. Assign myself as a
maintainer for Windo
1. Map CPU cores to their respective NUMA nodes as reported by system.
2. Support systems with more than 64 cores (multiple processor groups).
3. Fix magic constants, styling issues, and compiler warnings.
4. Add EAL private function to map DPDK socket ID to NUMA node number.
Fixes: 53ffd9f080fc (
EAL common code depends on tracepoint calls, but generic implementation
cannot be enabled on Windows due to missing standard library facilities.
Add stub functions to support tracepoint compilation, so that common
code does not have to conditionally include tracepoints until proper
support is added
1 - 100 of 200 matches
Mail list logo
