On Fri, Feb 25, 2022 at 10:30 AM wrote:
>
> From: Satha Rao
>
> cnxk platforms supports packet marking when TM enabled with
> valid shaper rates. VLAN DEI, IP ECN, or IP DSCP inside
> packet will be updated based on mark flags selected.
>
> Signed-off-by: Satha Rao
> ---
>
> v2:
> - rebased to m
Currently inline inbound device usage is not default for eventdev,
patch renames force_inl_dev dev arg to no_inl_dev and enables inline
inbound device by default.
Signed-off-by: Vamsi Attunuru
---
doc/guides/nics/cnxk.rst | 10 +-
drivers/event/cnxk/cnxk_eventdev_adptr.c
>
> 24/02/2022 12:06, Ananyev, Konstantin:
> > > > > > > > > > >> Or have a generic library for reading LPM entries.
> > > > > > > > > > >> L3fwd is supposed to be as small as possible (it no
> > > > > > > > > > >> longer is), and the real work should be done by
> > > > > > > > > > >> libraries
From: Satha Rao
Validate sq_node and parent before accessing their fields.
SQ was created without any associated TM node, this is valid negative
case, so return success while stopping TM without SQ node.
Signed-off-by: Satha Rao
---
drivers/common/cnxk/roc_nix_tm.c | 8 ++--
drivers/co
From: Satha Rao
cnxk platform supports red/yellow packet marking based on TM
configuration. This patch set hooks to enable/disable packet
marking for VLAN DEI, IP DSCP and IP ECN. Marking enabled only
in scalar mode.
Signed-off-by: Satha Rao
---
drivers/event/cnxk/cn10k_worker.h | 3 +-
dri
From: Satha Rao
cnxk platforms supports packet marking when TM enabled with
valid shaper rates. VLAN DEI, IP ECN, or IP DSCP inside
packet will be updated based on mark flags selected.
Signed-off-by: Satha Rao
---
v2:
- rebased to master, fixed conflicts
drivers/common/cnxk/meson.build
Hi Kai,
> Hi Akhil,
>
> This patch was intend to support Openssl 3.0 on symmetric crypto algorithms
> only, where the deprecated APIs, compile warnings and failing test cases were
> fixed.
> All the asymmetric crypto related issues stay untreated and will be fixed in
> the
> next patch.
>
How ca
Cast 1 to type uint64_t to avoid overflow.
CID 375812 (#1 of 1):
Unintentional integer overflow (OVERFLOW_BEFORE_WIDEN)
overflow_before_widen: Potentially overflowing expression 1 << 2 * i + 1
with type int (32 bits, signed) is evaluated using 32-bit arithmetic, and
then used in a context that exp
On Fri, 25 Feb 2022 09:00:37 +0800
Tianli Lai wrote:
> the mempool should be free when cleanup resources.
>
> Signed-off-by: Tianli Lai
Thanks for fixing this.
Acked-by: Stephen Hemminger
Add support for E824S and E825 family devices.
This will be documented later in release notes due to we don't have
mature product now.
Signed-off-by: Robin Zhang
---
v3:
- refine commit message.
drivers/net/ice/base/ice_common.c | 6 ++
drivers/net/ice/base/ice_devids.h | 13
> -Original Message-
> From: Jeff Daly
> Sent: Thursday, February 24, 2022 23:24
> To: dev@dpdk.org
> Cc: sta...@dpdk.org; Stephen Douthit ; Wang, Haiyue
>
> Subject: [PATCH v3 3/3] net/ixgbe: Fix SFP detection and linking on hotplug
>
> Currently the ixgbe driver does not ID any SFP ex
We need a separate fix for AVF which is ongoing.
And DPDK 20.11.4 LTS by default will only backport all the fix that during DPDK
21.11 dev cycle, so you may need to wait for 20.11.5 or just cherry pick by
yourself.
Regards
Qi
From: Navin Srinivas
Sent: Friday, February 25, 2022 2:00 AM
To: Zh
Hi Raslan,
Rebased and new version sent out.
Thanks,
Sean
> -Original Message-
> From: Raslan Darawsheh
> Sent: Thursday, February 24, 2022 9:18 PM
> To: Sean Zhang (Networking SW) ; NBU-Contact-
> Thomas Monjalon (EXTERNAL) ; Matan Azrad
> ; Slava Ovsiienko
> Cc: dev@dpdk.org
> Subjec
This patch adds matching on the optional fields (checksum/key/sequence)
of GRE header. The matching on checksum and sequence fields requests
support from rdma-core with the capability of misc5 and tunnel_header 0-3.
For patterns without checksum and sequence specified, keep using misc for
matching
the mempool should be free when cleanup resources.
Signed-off-by: Tianli Lai
---
app/pdump/main.c | 1 +
1 file changed, 1 insertion(+)
diff --git a/app/pdump/main.c b/app/pdump/main.c
index 04a38e8..3fe17ea 100644
--- a/app/pdump/main.c
+++ b/app/pdump/main.c
@@ -507,6 +507,7 @@ struct parse_
On 2/23/2022 11:48 PM, Ferruh Yigit wrote:
> On 2/23/2022 1:48 PM, Michael Baum wrote:
> > The "tx_db_nc" devarg forces doorbell register mapping to non-cached
> > region eliminating the extra write memory barrier. This argument was
> > used in creating the UAR for Tx and thus affected its perform
Hi,
> -Original Message-
> From: Ferruh Yigit
> Sent: Friday, February 25, 2022 6:58 AM
> To: Suanming Mou ; Slava Ovsiienko
> ; Matan Azrad
> Cc: Raslan Darawsheh ; Ori Kam ;
> dev@dpdk.org
> Subject: Re: [PATCH v4 02/14] net/mlx5: add HW steering low-level abstract
> code
>
> On 2/24/
Add option to probe common device using import CTX/PD functions instead
of create functions.
This option requires accepting the context FD and the PD handle as
devargs.
This sharing can be useful for applications that use PMD for only some
operations. For example, an app that generates queues itse
Add support queue/RSS action for external RxQ.
In indirection table creation, the queue index will be taken from
mapping array.
This feature supports neither LRO nor Hairpin.
Signed-off-by: Michael Baum
Acked-by: Matan Azrad
---
doc/guides/nics/mlx5.rst | 1 +
doc/guides/rel_no
The RxQ/TxQ control structure has a field named type. This type is enum
with values for standard and hairpin.
The use of this field is to check whether the queue is of the hairpin
type or standard.
This patch replaces it with a boolean variable that saves whether it is
a hairpin.
Signed-off-by: M
External queue is a queue that has been created and managed outside the
PMD. The queues owner might use PMD to generate flow rules using these
external queues.
When the queue is created in hardware it is given an ID represented by
32 bits. In contrast, the index of the queues in PMD is represented
Add support for rdma-core API to import device.
The API gets ibv_context file descriptor and returns an ibv_context
pointer that is associated with the given file descriptor.
Add also support for rdma-core API to import PD.
The API gets ibv_context and PD handle and returns a protection domain
(PD)
The functions which are not explicitly marked as internal
were exported because the local catch-all rule was missing in the
version script.
After adding the missing rule, all local functions are hidden.
The function mlx5_get_device_guid is used in another library,
so it needs to be exported (as int
These patches add support to external Rx queues.
External queue is a queue that is managed by a process external to PMD,
but uses PMD process to generate its flow rules.
For the hardware to allow the DPDK process to set rules for it, the
process needs to use the same PD of the external process. In
Hi Akhil,
This patch was intend to support Openssl 3.0 on symmetric crypto algorithms
only, where the deprecated APIs, compile warnings and failing test cases were
fixed.
All the asymmetric crypto related issues stay untreated and will be fixed in
the next patch.
Regards
Kai
> -Origina
On Thu, 24 Feb 2022 22:51:31 +0100
Morten Brørup wrote:
> > From: Tyler Retzlaff [mailto:roret...@linux.microsoft.com]
> > Sent: Thursday, 17 February 2022 08.42
> >
> > On Wed, Feb 16, 2022 at 10:10:01AM +, Bruce Richardson wrote:
> > > On Wed, Feb 16, 2022 at 11:05:09AM +0100, Morten Brø
On 2/24/2022 1:40 PM, Suanming Mou wrote:
The HW steering low-level implementation will be added later in another
patch series. To avoid the linkage issues the abstract stub replacement
is provided currently.
Signed-off-by: Suanming Mou
Acked-by: Viacheslav Ovsiienko
---
drivers/net/mlx5/meson
23/02/2022 19:13, Morten Brørup:
> > From: Megha Ajmera [mailto:megha.ajm...@intel.com]
> > Sent: Wednesday, 23 February 2022 18.37
> >
> > Masking of core mask was incorrect. Instead of using 1U for shifting,
> > it
> > should be using 1LU as the result is assigned to uint64.
> >
> > CID 375859:
> > Megha Ajmera (4):
> > sched: remove code no longer needed
> > sched: move grinder configuration flag
> > sched: enable statistics unconditionally
> > sched: enable traffic class oversubscription unconditionally
>
> Series-acked-by: Cristian Dumitrescu
Applied, thanks.
> From: Tyler Retzlaff [mailto:roret...@linux.microsoft.com]
> Sent: Thursday, 17 February 2022 08.42
>
> On Wed, Feb 16, 2022 at 10:10:01AM +, Bruce Richardson wrote:
> > On Wed, Feb 16, 2022 at 11:05:09AM +0100, Morten Brørup wrote:
> > > > From: Bruce Richardson [mailto:bruce.richard...@int
Hi,
> -Original Message-
> From: Suanming Mou
> Sent: Thursday, February 24, 2022 3:41 PM
> To: Slava Ovsiienko ; Matan Azrad
>
> Cc: Raslan Darawsheh ; Ori Kam ;
> dev@dpdk.org
> Subject: [PATCH v4 00/14] net/mlx5: add hardware steering
>
> The Connect-X steering is a lookup hardware m
On Thu, Feb 24, 2022 at 11:59 PM Akhil Goyal wrote:
>
> cn10k platform can support IP reassembly offload for upto 4 fragments
> in the inline IPsec path.
> The feature is enabled and tested as per unit test app[1].
>
> The patchset is rebased over [2].
>
> [1]: http://patchwork.dpdk.org/user/todo/
On Fri, Feb 25, 2022 at 12:09 AM wrote:
>
> From: Pavan Nikhilesh
>
> In vWQE mode, the mbuf address is calculated without using the
> iova list.
>
> Packet length can also be calculated by using NIX_PARSE_S by
> which we can completely eliminate reading 2nd cache line
> depending on the offloads
On Thu, Feb 24, 2022 at 3:19 PM Vamsi Attunuru wrote:
>
> Patch implements soft expiry notification mechanism in outbound
> path by creating required number of ring buffers and a common poll
> thread which polls for soft expiry events enqueued by microcode.
>
> Signed-off-by: Vamsi Attunuru
> ---
On Thu, Feb 24, 2022 at 4:04 PM Tomasz Duszynski wrote:
>
> Model is uniquely identified by 4 numbers. Print them all in case
> model being populated is not on a list of known models. This makes
> debugging a bit easier.
>
> Signed-off-by: Tomasz Duszynski
> Reviewed-by: Jakub Palider
> Reviewed
On Thu, Feb 24, 2022 at 4:13 PM Tomasz Duszynski wrote:
>
> Add B0 variant to the list of supported models.
>
> Signed-off-by: Tomasz Duszynski
> Reviewed-by: Jerin Jacob Kollanukkaran
Updated the git commit as follows and applied to
dpdk-next-net-mrvl/for-next-net. Thanks
common/cnxk: sup
> -Original Message-
> From: Raja Zidane
> Sent: Wednesday, February 23, 2022 1:33 PM
> To: dev@dpdk.org
> Cc: ma...@nvidia.com; sta...@dpdk.org
> Subject: [PATCH] app/compress-perf: optimize ops pool allocation
>
> An array of the size of total operations needed for the de/compression is
> -Original Message-
> From: Raja Zidane
> Sent: Wednesday, February 23, 2022 1:33 PM
> To: dev@dpdk.org
> Cc: ma...@nvidia.com; sta...@dpdk.org
> Subject: [PATCH] app/compress-perf: optimize ops pool allocation
>
> An array of the size of total operations needed for the de/compression is
On Thu, Feb 24, 2022 at 3:28 PM wrote:
>
> From: Satha Rao
>
> cnxk platforms supports packet marking when TM enabled with
> valid shaper rates. VLAN DEI, IP ECN, or IP DSCP inside
> packet will be updated based on mark flags selected.
>
> Signed-off-by: Satha Rao
Please rebase to next-net-mrvl
On Thu, Feb 24, 2022 at 3:07 AM Raslan Darawsheh wrote:
>
> ++ adding more people here,
>
> > -Original Message-
> > From: Thomas Monjalon
> > Sent: Thursday, February 24, 2022 11:36 AM
> > To: Raslan Darawsheh
> > Cc: dev@dpdk.org; ferruh.yi...@intel.com; Gal Cohen (ProdM)
> >
> > Subj
From: Elena Agostini
To enable the gpudev rte_gpu_mem_cpu_map feature to expose
GPU memory to the CPU, the GPU CUDA driver library needs
the GDRCopy library and driver.
If DPDK is built without GDRCopy, the GPU CUDA driver returns
error if the is invoked rte_gpu_mem_cpu_map.
All the others GPU
From: Elena Agostini
Signed-off-by: Elena Agostini
---
doc/guides/gpus/features/cuda.ini | 12
1 file changed, 12 insertions(+)
create mode 100644 doc/guides/gpus/features/cuda.ini
diff --git a/doc/guides/gpus/features/cuda.ini
b/doc/guides/gpus/features/cuda.ini
new file mode 1
> This patch update the symmetric EVP routine in crypto openssl pmd
> to adopt openssl 3.0 library.
>
> Signed-off-by: Kai Ji
>
> v5:
> - checkpatch fix
>
> v4:
> - code comments addressed
>
> v3:
> - rebase to 22.03-RC1
> - enable openssl 3.0 lagacy library of DES
> - remove local ctx in comb
From: Pavan Nikhilesh
Align perfetchs for CN10K cache model for vWQE in Rx and Tx.
Move mbuf->next NULL assignment to Tx path and enabled it only
when multi segments offload is enabled to reduce L1 pressure.
Add macros to detect corrupted mbuf->next values when
MEMPOOL_DEBUG is set.
Signed-off-b
From: Pavan Nikhilesh
In vWQE mode, the mbuf address is calculated without using the
iova list.
Packet length can also be calculated by using NIX_PARSE_S by
which we can completely eliminate reading 2nd cache line
depending on the offloads enabled.
Signed-off-by: Pavan Nikhilesh
Acked-by: Jeri
From: Vidya Sagar Velumuri
When reassembly is enabled by application, set corresponding
flags in SA during creation.
Provide roc API to configure reassembly unit with active and zombie limits
and step size
Signed-off-by: Vidya Sagar Velumuri
---
drivers/common/cnxk/cnxk_security.c | 10 ++
cn10k platform can support IP reassembly offload for upto 4 fragments
in the inline IPsec path.
The feature is enabled and tested as per unit test app[1].
The patchset is rebased over [2].
[1]: http://patchwork.dpdk.org/user/todo/dpdk/?series=21716
[2]: https://patches.dpdk.org/project/dpdk/list/
Hi,
Whether this fix is applicable for VF? I do not see this change ported to
VF in DPDK-20.11.4.
Thanks,
Navin Srinivas
On Wed, Oct 27, 2021 at 8:58 AM Zhang, Qi Z wrote:
>
>
> > -Original Message-
> > From: Yu, DapengX
> > Sent: Tuesday, October 26, 2021 9:56 AM
> > To: Yang, Qiming
On Thu, 24 Feb 2022 17:29:03 +
"Ananyev, Konstantin" wrote:
> Hi Dmitry,
>
> > 2022-02-21 00:56 (UTC+0300), Dmitry Kozlyuk:
> > > 2022-02-09 13:57 (UTC+), Ananyev, Konstantin:
> > > > > > Actually, please scrap that comment.
> > > > > > Obviously it wouldn't work for static variables
Hi Dmitry,
> 2022-02-21 00:56 (UTC+0300), Dmitry Kozlyuk:
> > 2022-02-09 13:57 (UTC+), Ananyev, Konstantin:
> > > > > Actually, please scrap that comment.
> > > > > Obviously it wouldn't work for static variables,
> > > > > and doesn't make much sense.
> > > > > Though few thoughts remain:
>
From: Vidya Sagar Velumuri
When reassembly is enabled by application, set corresponding
flags in SA during creation.
Provide roc API to configure reassembly unit with active and zombie limits
and step size
Signed-off-by: Vidya Sagar Velumuri
---
drivers/common/cnxk/cnxk_security.c | 10 ++
cn10k platform can support IP reassembly offload for upto 4 fragments
in the inline IPsec path.
The feature is enabled and tested as per unit test app[1].
The patchset is rebased over [2].
[1]: http://patchwork.dpdk.org/user/todo/dpdk/?series=21716
[2]: https://patches.dpdk.org/project/dpdk/list/
> -Original Message-
> From: De Lara Guarch, Pablo
> Sent: Thursday, February 24, 2022 7:58 AM
> To: Wang, Yipeng1 ; Marohn, Byron
>
> Cc: dev@dpdk.org; De Lara Guarch, Pablo ;
> sta...@dpdk.org
> Subject: [PATCH v2 1/2] test/efd: fix size of constant
>
> Constant value 1 has a size of 3
> -Original Message-
> From: De Lara Guarch, Pablo
> Sent: Thursday, February 24, 2022 7:58 AM
> To: Wang, Yipeng1 ; Marohn, Byron
>
> Cc: dev@dpdk.org; De Lara Guarch, Pablo ;
> sta...@dpdk.org
> Subject: [PATCH v2 2/2] efd: fix uninitialized structure
>
> Coverity flags that both eleme
https://bugs.dpdk.org/show_bug.cgi?id=938
Bug ID: 938
Summary: [TAP] testpmd crashed after stopping ports
Product: DPDK
Version: 21.11
Hardware: All
OS: All
Status: UNCONFIRMED
Severity: normal
P
From: Pavan Nikhilesh
Align perfetchs for CN10K cache model for vWQE in Rx and Tx.
Move mbuf->next NULL assignment to Tx path and enabled it only
when multi segments offload is enabled to reduce L1 pressure.
Add macros to detect corrupted mbuf->next values when
MEMPOOL_DEBUG is set.
Signed-off-b
From: Pavan Nikhilesh
In vWQE mode, the mbuf address is calculated without using the
iova list.
Packet length can also be calculated by using NIX_PARSE_S by
which we can completely eliminate reading 2nd cache line
depending on the offloads enabled.
Signed-off-by: Pavan Nikhilesh
---
v2 Changes
This patch adds mlx5 specifics description about
handling the Ethernet type by modify field action
for VLAN-ed traffic.
Fixes: 641dbe4fb053 ("net/mlx5: support modify field flow action")
Cc: sta...@dpdk.org
Signed-off-by: Viacheslav Ovsiienko
---
doc/guides/nics/mlx5.rst | 5 +
1 file chang
On Thu, 24 Feb 2022 20:14:58 +0800
Tianli Lai wrote:
> + if (pt->mp) {
> + rte_mempool_free(pt->mp);
> + pt->mp = NULL;
Check for null is unnecessary here, rte_mempool_free(NULL) is nop.
Coverity flags that both elements of efd_online_group_entry
are used uninitialized. This is OK because this structure
is initially used for starting values, so any value is OK.
Coverity ID: 375868
Fixes: 56b6ef874f80 ("efd: new Elastic Flow Distributor library")
Cc: pablo.de.lara.gua...@intel.com
Constant value 1 has a size of 32 bits, and shifting it more than 32 bits
to the left overflows. 1ULL is needed to be able to get a 64-bit value.
Coverity ID: 375846
Fixes: 8751a7e9832b ("efd: allow more CPU sockets in table creation")
Cc: pablo.de.lara.gua...@intel.com
Cc: sta...@dpdk.org
Signed
This patch adds mlx5 specifics description about
handling the Ethernet type by modify field action
for VLAN-ed traffic.
Fixes: 641dbe4fb053 ("net/mlx5: support modify field flow action")
Cc: sta...@dpdk.org
Signed-off-by: Viacheslav Ovsiienko
---
doc/guides/nics/mlx5.rst | 5 +
1 file chang
Fixes: a46bbb57605b4 ("net/bnxt: update multi device design")
In tf_session_create(), there is a case that with 'tfp->session' still
be NULL and run 'goto cleanup', which will leads to a null dereference
by 'tfp_free(tfp->session->core_data)' in the cleanup.
Signed-off-by: Weiguo Li
---
drivers
To speed the device suspend and resume time, make counter persitent
in reconfiguration until the device gets removed.
Signed-off-by: Xueming Li
---
doc/guides/vdpadevs/mlx5.rst| 6 ++
drivers/vdpa/mlx5/mlx5_vdpa.c | 19 +++--
drivers/vdpa/mlx5/mlx5_vdpa.h |
This patch supports device cleanup callback API which called when device
disconnected with VM. Cached resources like VM MR and VQ memory are
released.
Signed-off-by: Xueming Li
---
drivers/vdpa/mlx5/mlx5_vdpa.c | 23 +++
drivers/vdpa/mlx5/mlx5_vdpa.h | 1 +
2 files changed,
During device suspend and resume, resources are not changed normally.
When huge resources allocated to VM, like huge memory size or lots of
queues, time spent on release and recreate became significant.
To speed up, this patch reuse resoruces like VM MR and VirtQ memory if
not changed.
Signed-off
When Qemu suspend a VM, hw notifier is un-mmapped while vCPU thread may
still active and write notifier through kick socket.
PMD kick handler thread tries to install hw notifier through client
socket in such case will timeout and slow down device close.
This patch skips hw notifier install if VQ
To speed up device resume, create reuseable resources during device
probe state, release when device remove. Reused resources includes TIS,
TD, VAR Doorbell mmap, error handling event channel and interrupt
handler, UAR, Rx event channel, NULL MR, steer domain and table.
Signed-off-by: Xueming Li
In Ctrl+C handling, sometimes kick handling thread gets endless EGAIN
error and fall into dead lock.
Kick happens frequently in real system due to busy traffic or retry
mechanism. This patch simplifies kick firmware anyway and skip setting
hardware notifier due to potential device error, notifier
Disable interrupt unregister timeout to avoid invalid FD caused
interrupt thread segment fault.
Fixes: 62c813706e41 ("vdpa/mlx5: map doorbell")
Cc: ma...@mellanox.com
Cc: sta...@dpdk.org
Signed-off-by: Xueming Li
---
drivers/vdpa/mlx5/mlx5_vdpa_virtq.c | 20
1 file changed,
v1:
- rebase with latest upstream code
- fix coverity issues
v2:
- fix build issue on OS w/o flow DR API
Xueming Li (7):
vdpa/mlx5: fix interrupt trash that leads to segment fault
vdpa/mlx5: fix dead loop when process interrupted
vdpa/mlx5: no kick handling during shutdown
vdpa/mlx5: re
Coverity flags that both elements of efd_online_group_entry
are used unititialized. This is OK because this structure
is initially used for starting values, so any value is OK.
Coverity ID: 375868
Fixes: 56b6ef874f80 ("efd: new Elastic Flow Distributor library")
Cc: pablo.de.lara.gua...@intel.com
Constant value 1 has a size of 32 bits, and shifting it more than 32 bits
to the left overflows. 1ULL is needed to be able to get a 64-bit value.
Coverity ID: 375846
Fixes: 8751a7e9832b ("efd: allow more CPU sockets in table creation")
Cc: pablo.de.lara.gua...@intel.com
Cc: sta...@dpdk.org
Signed
Currently the ixgbe driver does not ID any SFP except for the first one
plugged in. This can lead to no-link, or incorrect speed conditions.
For example:
* If link is initially established with a 1G SFP, and later a 1G/10G
multispeed part is later installed, then the MAC link setup functions are
1ca05831b9b added a check that SDP3 (used as a TX_DISABLE output to the
SFP cage on these cards) is not asserted to avoid incorrectly reporting
link up when the SFP's laser is turned off.
ff8162cb957 limited this workaround to fiber ports
This patch:
* Adds additional check for 82599 type, not al
From: Stephen Douthit
Currently all X500EM* MAC types fallthrough to the default case and get
reported as non-SFP regardless of media type, which isn't correct.
Fixes: 0790adeb567 ("ixgbe/base: support X550em_a device")
Cc: sta...@dpdk.org
Signed-off-by: Stephen Douthit
Signed-off-by: Jeff Dal
Hello all,
We have several platforms based on Intel's C3000 series of SoCs that
have integrated ixgbe devices (X550EM) operating in the "Native SFI"
mode (the 0x15c4 device ID).
This set of patches address issues with detection and hotplug of
SPFs. This is the first of a series of patches to ref
To speed the device suspend and resume time, make counter persitent
in reconfiguration until the device gets removed.
Signed-off-by: Xueming Li
---
doc/guides/vdpadevs/mlx5.rst| 6 ++
drivers/vdpa/mlx5/mlx5_vdpa.c | 19 +++--
drivers/vdpa/mlx5/mlx5_vdpa.h |
This patch supports device cleanup callback API which called when device
disconnected with VM. Cached resources like VM MR and VQ memory are
released.
Signed-off-by: Xueming Li
---
drivers/vdpa/mlx5/mlx5_vdpa.c | 23 +++
drivers/vdpa/mlx5/mlx5_vdpa.h | 1 +
2 files changed,
During device suspend and resume, resources are not changed normally.
When huge resources allocated to VM, like huge memory size or lots of
queues, time spent on release and recreate became significant.
To speed up, this patch reuse resoruces like VM MR and VirtQ memory if
not changed.
Signed-off
When Qemu suspend a VM, hw notifier is un-mmapped while vCPU thread may
still active and write notifier through kick socket.
PMD kick handler thread tries to install hw notifier through client
socket in such case will timeout and slow down device close.
This patch skips hw notifier install if VQ
To speed up device resume, create reuseable resources during device
probe state, release when device remove. Reused resources includes TIS,
TD, VAR Doorbell mmap, error handling event channel and interrupt
handler, UAR, Rx event channel, NULL MR, steer domain and table.
Signed-off-by: Xueming Li
Disable interrupt unregister timeout to avoid invalid FD caused
interrupt thread segment fault.
Fixes: 62c813706e41 ("vdpa/mlx5: map doorbell")
Cc: ma...@mellanox.com
Cc: sta...@dpdk.org
Signed-off-by: Xueming Li
---
drivers/vdpa/mlx5/mlx5_vdpa_virtq.c | 20
1 file changed,
In Ctrl+C handling, sometimes kick handling thread gets endless EGAIN
error and fall into dead lock.
Kick happens frequently in real system due to busy traffic or retry
mechanism. This patch simplifies kick firmware anyway and skip setting
hardware notifier due to potential device error, notifier
v1:
- rebase with latest upstream code
- fix coverity issues
Xueming Li (7):
vdpa/mlx5: fix interrupt trash that leads to segment fault
vdpa/mlx5: fix dead loop when process interrupted
vdpa/mlx5: no kick handling during shutdown
vdpa/mlx5: reuse resources in reconfiguration
vdpa/mlx5:
śr., 19 sty 2022 o 15:50 Michal Krawczyk napisał(a):
>
> Patch changing the way of accessing interrupt handle also changed order
> of the rte_pci_map_device() call and rte_pci_device:driver assignment.
> It was causing issues with Write Combine mapping on the Linux platform
> if it was used with t
On Thu, Feb 24, 2022 at 3:44 PM Nithin Dabilpuram
wrote:
>
> Remove unused files that were left over after Rx and Tx template
> function rework.
>
> Fixes: 5169508a68fa ("net/cnxk: add cn9k template Rx functions to build")
> Fixes: dd8c20eee472 ("net/cnxk: add cn9k template Tx functions to build")
On Thu, Feb 24, 2022 at 02:46:24PM +0100, Thomas Monjalon wrote:
> 24/02/2022 12:06, Ananyev, Konstantin:
> > > > > > > > > > >> Or have a generic library for reading LPM entries.
> > > > > > > > > > >> L3fwd is supposed
> > > > > > > > > > >> to be as small as possible (it no longer is), and the
From: Pavan Nikhilesh
Align perfetchs for CN10K cache model for vWQE in Rx and Tx.
Move mbuf->next NULL assignment to Tx path and enabled it only
when multi segments offload is enabled to reduce L1 pressure.
Add macros to detect corrupted mbuf->next values when
MEMPOOL_DEBUG is set.
Signed-off-b
From: Pavan Nikhilesh
In vWQE mode, the mbuf address is calculated without using the
iova list.
Packet length can also be calculated by using NIX_PARSE_S by
which we can completely eliminate reading 2nd cache line
depending on the offloads enabled.
Signed-off-by: Pavan Nikhilesh
---
drivers/ne
24/02/2022 12:06, Ananyev, Konstantin:
> > > > > > > > > >> Or have a generic library for reading LPM entries. L3fwd
> > > > > > > > > >> is supposed
> > > > > > > > > >> to be as small as possible (it no longer is), and the real
> > > > > > > > > >> work should
> > > > > > > > > >> be done by l
On Thu, Feb 24, 2022 at 2:05 PM Rakesh Kudurumalla
wrote:
>
> rss reta table is corrupted during
> rte_eth_dev_rss_reta_update.This fix restores
> previous table entries before updating.
>
> Fixes: 00242a687de ("net/cnxk: support RETA and RSS hash")
> Cc: sta...@dpdk.org
>
> Signed-off-by: Rakesh
HW steering can support indirect action as well. With indirect action,
the flow can be created with more flexible shared RSS action selection.
This will can save the action template with different RSS actions.
This commit adds the flow queue operation callback for:
rte_flow_async_action_handle_cre
HW steering header reformat action can work under bulk mode. In
this case, when create the table, bulk size of header reformat
actions will be allocated in low level. Afterwards, when create
flow, just simply specify the action index in the bulk and the
encapsulation data to the action will be enou
Flow table is a group of flows with the same matching criteria
and the same actions defined for them. The table defines rules
that have the same matching fields but with different matching
values. For example, matching on 5 tuple, the table will be
(IPv4 source + IPv4 dest + s_port + d_port + next_
In case port is being stopped, all created flows should be flushed.
This commit adds the flow flush helper function.
Signed-off-by: Suanming Mou
Acked-by: Viacheslav Ovsiienko
---
drivers/net/mlx5/mlx5_flow.c| 8 ++
drivers/net/mlx5/mlx5_flow_hw.c | 129
2
The mark action is covered by tag action internally. While it is added
the HW will add a tag to the packet. The mark value can be set as fixed
or dynamic as the action mask indicates.
Signed-off-by: Suanming Mou
Acked-by: Viacheslav Ovsiienko
---
drivers/net/mlx5/mlx5.h | 2 +
drivers/
This commit adds the queue and RSS action. Similar to the jump action,
dynamic ones will be added to the action construct list.
Due to the queue and RSS action in template should not be destroyed
during port restart, the actions are created with standalone indirect
table as indirect action does. W
The HW steering uses async queue-based flow rules management
mechanism. The matcher and part of the actions have been
prepared during flow table creation. Some remaining actions
will be constructed during flow creation if needed.
A flow postpone attribute bit describes if flow management
should be
Jump action connects different level of flow tables and allows packet
handling in the chain of flows.
A new action construct data struct is also added in this commit to help
to handle not only the dynamic jump action but also for the other generic
dynamic actions. The actions with empty mask confi
1 - 100 of 158 matches
Mail list logo
