Moves the next nodes of packet classification node to a public
header files, so that it can be used from applications.
Signed-off-by: Ankur Dwivedi
---
lib/node/meson.build| 1 +
lib/node/pkt_cls.c | 33 +++--
lib/node/pkt_cls_priv.h | 7
The ip4_lookup_fib is added as next node in packet classification.
Signed-off-by: Ankur Dwivedi
---
lib/node/pkt_cls.c | 1 +
lib/node/rte_node_pkt_cls_api.h | 1 +
2 files changed, 2 insertions(+)
diff --git a/lib/node/pkt_cls.c b/lib/node/pkt_cls.c
index 5d6409fdbf..f6e76989bc 10
Adds a lookup FIB node for IP6.
Adds a public function to do fib configuration from application.
Signed-off-by: Ankur Dwivedi
---
lib/node/ip6_lookup_fib.c | 139
lib/node/meson.build| 1 +
lib/node/rte_node_ip6_api.h | 15
3 files changed,
The ip6_lookup_fib is added as next node in packet classification.
Signed-off-by: Ankur Dwivedi
---
lib/node/pkt_cls.c | 1 +
lib/node/rte_node_pkt_cls_api.h | 1 +
2 files changed, 2 insertions(+)
diff --git a/lib/node/pkt_cls.c b/lib/node/pkt_cls.c
index f6e76989bc..ca323ea573 10
Adds a command to setup lookup mode for IP6 packets. By default LPM
mode is used.
FIB can be given as a lookup mode by specifying the mode in CLI file.
When FIB is given as lookup mode, the next node of packet classification
node is updated to use IP6 lookup fib node.
Signed-off-by: Ankur Dwivedi
Adds the process callback function for ip6_lookup_fib node.
Signed-off-by: Ankur Dwivedi
---
lib/node/ip6_lookup_fib.c | 148 ++
1 file changed, 148 insertions(+)
diff --git a/lib/node/ip6_lookup_fib.c b/lib/node/ip6_lookup_fib.c
index b510452ad8..3df61890e9
Adds a command to setup lookup mode for IP4 packets. By default the
LPM mode is used.
FIB can be given as a lookup mode by specifying the mode in CLI file.
When FIB is given as lookup mode, the next node of packet classification
node is updated to use IP4 lookup fib node.
Signed-off-by: Ankur Dwiv
Adds a public function to add IP4 route to FIB. The applications should
call this function to add IP4 routes to FIB.
Signed-off-by: Ankur Dwivedi
---
lib/node/ip4_lookup_fib.c | 37 +
lib/node/rte_node_ip4_api.h | 19 +++
2 files changed, 56
Moves RTE_FIB_NAMESIZE to header file, in order to use it outside the fib
library.
Signed-off-by: Ankur Dwivedi
---
lib/fib/rte_fib.c | 3 ---
lib/fib/rte_fib.h | 3 +++
2 files changed, 3 insertions(+), 3 deletions(-)
diff --git a/lib/fib/rte_fib.c b/lib/fib/rte_fib.c
index 2c05d46500..184210f
Add support for setting application specified port history
Set HL equal to CQ depth when inflight control is enabled
Added command line parameters 'use_default_hl' (default: 1)
and 'alloc_hl_entries'
- When 'use_default_hl = 1'
* Per port HL is set to DLB2_FIXED_CQ_HL_SIZE (32)
* Recommen
When application configures a DIR port with CQ depth less than 8, DLB PMD
sets port's cq_depth as 8 and token reservation is used to make the
effective cq_depth smaller. However, while setting port's cq_depth_mask
application configured CQ depth was used resulting in reading incorrect
cachelines wh
This patchset introduces various fixes related to dlb2 hw resource
management. The dlb2 hw has limited resources, which are configurable
using command line options. This patch allows managing History list,
scheduling bandwidth and credits using command line options. It also
fixes some issues with r
From: Tirthendu Sarkar
dlb2_eventdev_info_get() that implements rte_event_dev_info_get() should
return the maximum available credits as supported by HW.
Set maximum credits before device probing by checking HW version.
Signed-off-by: Tirthendu Sarkar
---
drivers/event/dlb2/pf/dlb2_pf.c | 2 ++
update QID depth xstats counter in vector dequeue path
Signed-off-by: Pravin Pathak
---
drivers/event/dlb2/dlb2.c | 8
1 file changed, 8 insertions(+)
diff --git a/drivers/event/dlb2/dlb2.c b/drivers/event/dlb2/dlb2.c
index 5f3b816665..19fe973bff 100644
--- a/drivers/event/dlb2/dlb2.c
While providing port_cos as vdev/pf CLI argument, the port numbers should
take into account all ports (LDB and DIR) that are created by the
application and the same order should be provided for port_cos parameter.
This fix add checks to ensure that above is validated correctly.
Signed-off-by: Prav
While unlinking ports, all associated credits should be released.
This commit avoids race condition when main thread is unlinking
while workers are running.
Signed-off-by: Pravin Pathak
---
drivers/event/dlb2/dlb2.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/even
DLB 2.0 device has 64 single linked or directed ports.
DLB 2.5 device has 96 single linked ports.
This commit fixes issue of rte_event_dev_info_get returning 64
instead of 96 single link ports for DLB2.5
Signed-off-by: Pravin Pathak
---
drivers/event/dlb2/dlb2.c | 8
1 file changed, 4 i
This patchset introduces various fixes related to dlb2 hw resource
management. The dlb2 hw has limited resources, which are configurable
using command line options. This patch allows managing History list,
scheduling bandwidth and credits using command line options. It also
fixes some issues with r
When application configures a DIR port with CQ depth less than 8, DLB PMD
sets port's cq_depth as 8 and token reservation is used to make the
effective cq_depth smaller. However, while setting port's cq_depth_mask
application configured CQ depth was used resulting in reading incorrect
cachelines wh
It's 1ms, not 1 second.
It's a workaround, just to provide a fast and dirty fix for someone who needs
this.
From: "Stephen Hemminger"
> Date: Tue, Oct 8, 2024, 01:54
> Subject: [External] Re: [PATCH] mlx5: fix race at mlx5_dev_close
> To: "hepeng"
> Cc:
> On Thu, 11 Apr 2024 14:17:40 +0800
> From: Van Haaren, Harry
> Sent: Tuesday, May 6, 2025 12:39 PM
> To: Owen Hilyard ; Etelson, Gregory
> ; Richardson, Bruce
> Cc: dev@dpdk.org
> Subject: Re: [PATCH] rust: RFC/demo of safe API for Dpdk Eal, Eth and Rxq
> > From: Owen Hilyard
> > Sent: Saturday, May 03, 2025 6:13 PM
> > To: Va
Remove custom number parser and use C standard library instead. In order to
keep compatibility with earlier versions of the parser, we have to take
into account a couple of quirks:
- We did not consider "negative" numbers to be valid for anything other
than base-10 numbers, whereas C standard li
> I'm new to this project, and have no clue about the failure,
> could experts at this project provide
>
> some help about the following failure?
>
> + sudo babeltrace
> /home/runner/work/dpdk/dpdk/build/app/test/suites/rte-2025-04-30-AM-02-
> 25-21
>
>
Test-pmd already has a way to run a list of commands from file, but there
is no way to pause execution for a specified amount of time between two
commands. This may be necessary for simple automation, particularly for
waiting on some asynchronous operation such as link status update.
Add a simple
Add support for parsing floating point numbers in cmdline library, as well
as unit tests for the new functionality. Use C library for parsing.
Signed-off-by: Anatoly Burakov
---
Notes:
v7 -> v8:
- Fixed leftover comment about needing bigger epsilon value
- Fixed debug prints to avoid
On 5/6/2025 7:40 PM, Jake Freeland wrote:
Provide the EAL base address as a hint to mmap(2), so device resources
are not mapped where malloc(3) et al. make allocations.
This makes mapping conflicts less likely for secondary processes that
make memory allocations before initializing EAL.
Signed-
On 4/3/2025 4:57 AM, Dengdui Huang wrote:
When the process address space is insufficient,
mmap will fail, which will cause an infinite loop.
This patch stops attempting mmap if it fails and
the requested size cannot be reduced.
Fixes: b7cc54187ea4 ("mem: move virtual area function in common dire
On 5/6/2025 7:40 PM, Jake Freeland wrote:
When mapping PCI resources in secondary processes, use the
RTE_MAP_FORCE_ADDRESS_NOREPLACE flag to indicate that the
mapping must be made at the provided address.
Without this flag, the kernel might return a different address for
the mapping, even if the
On 5/6/2025 7:50 PM, Jake Freeland wrote:
It is possible for rte_fbarray_find_next_n_free() to misreport that there
are n contiguous open spots. If we need two contiguous entries for a
hole, make sure that we're not indexing out-of-bounds in the fbarray.
The `arr->len - arr->count < n` condition
On 5/6/2025 7:40 PM, Jake Freeland wrote:
Use open(2) instead of access(2) to check for the existence of the target
device. This avoids a possible race condition where the the device file is
removed after a successful call to access(2) but before open(2).
This also fixes any potential bugs assoc
On 5/6/2025 7:40 PM, Jake Freeland wrote:
When calling rte_pci_write_config(), use memcpy(3) to copy @len bytes
of @buf into local memory instead of casting it to a uint32_t pointer
and dereferencing it. This prevents us from reading data outside of
@buf in the case that @buf has a length less th
On 5/8/2025 7:19 AM, David Marchand wrote:
On systems with more cpu than RTE_MAX_LCORE, starting DPDK with
cores >= RTE_MAX_LCORE ends up with an ambiguous error log.
Example with RTE_MAX_LCORE=8:
$ taskset -c 8 ./build/app/dpdk-testpmd
EAL: Detected CPU lcores: 8
EAL: Detected NUMA nodes: 1
EAL
Add support for parsing floating point numbers in cmdline library, as well
as unit tests for the new functionality. Use C library for parsing.
Signed-off-by: Anatoly Burakov
---
Notes:
v5 -> v6:
- Small refactor to reduce amount of noise
- Use strtof for parsing single precision floa
On 5/6/2025 7:50 PM, Jake Freeland wrote:
Use rte_fbarray_is_used() to check if the previous fbarray entry is
already empty.
Using prev_ms_idx to do this is flawed in cases where we loop through
multiple memseg lists. Each memseg list has its own count and length,
so using a prev_ms_idx from one
On 5/6/2025 7:50 PM, Jake Freeland wrote:
Use rte_fbarray_is_used() to check if the previous fbarray entry is
already empty.
Using prev_ms_idx to do this is flawed in cases where we loop through
multiple memseg lists. Each memseg list has its own count and length,
so using a prev_ms_idx from one
On 5/8/2025 12:01 PM, Anatoly Burakov wrote:
Add support for parsing floating point numbers in cmdline library, as well
as unit tests for the new functionality. Use C library for parsing.
Signed-off-by: Anatoly Burakov
---
Notes:
v6 -> v7:
- Fixed a bug in float compare in unit tests where a
Add support for parsing floating point numbers in cmdline library, as well
as unit tests for the new functionality. Use C library for parsing.
Signed-off-by: Anatoly Burakov
---
Notes:
v5 -> v6:
- Small refactor to reduce amount of noise
- Use strtof for parsing single precision floa
Remove custom number parser and use C standard library instead. In order to
keep compatibility with earlier versions of the parser, we have to take
into account a couple of quirks:
- We did not consider "negative" numbers to be valid for anything other
than base-10 numbers, whereas C standard li
Test-pmd already has a way to run a list of commands from file, but there
is no way to pause execution for a specified amount of time between two
commands. This may be necessary for simple automation, particularly for
waiting on some asynchronous operation such as link status update.
Add a simple
Test-pmd already has a way to run a list of commands from file, but there
is no way to pause execution for a specified amount of time between two
commands. This may be necessary for simple automation, particularly for
waiting on some asynchronous operation such as link status update.
Add a simple
Remove custom number parser and use C standard library instead. In order to
keep compatibility with earlier versions of the parser, we have to take
into account a couple of quirks:
- We did not consider "negative" numbers to be valid for anything other
than base-10 numbers, whereas C standard li
On 5/8/2025 9:27 AM, Bruce Richardson wrote:
On Wed, May 07, 2025 at 04:22:10PM +0100, Anatoly Burakov wrote:
Remove custom number parser and use C standard library instead. In order to
keep compatibility with earlier versions of the parser, we have to take
into account a few quirks:
- We do no
Wouldn't the correct fix assume the change has no effect if the operation
failed, like the _enable_ functions do, i.e.:
int
rte_eth_promiscuous_disable(uint16_t port_id)
{
struct rte_eth_dev *dev;
int diag = 0;
RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
de
To get better performance, using async mode to replace sync mode
However, case UADK_CHAIN_CIPHER_AUTH and UADK_CHAIN_AUTH_CIPHER
still use sync mode for the first operation and async mode for
the second operation since the dependence.
Also RTE_CRYPTO_AUTH_OP_VERIFY will hold the generated auth in
To get better performance, using async mode to replace sync mode
Signed-off-by: Zhangfei Gao
---
drivers/compress/uadk/uadk_compress_pmd.c | 107 +-
.../compress/uadk/uadk_compress_pmd_private.h | 2 +-
2 files changed, 82 insertions(+), 27 deletions(-)
diff --git a/driver
Realize async mode to replace sync mode for better performance
Zhangfei Gao (2):
compress/uadk: use async mode to replace sync mode
crypto/uadk: use async mode to replace sync mode
drivers/compress/uadk/uadk_compress_pmd.c | 107 --
.../compress/uadk/uadk_compress_pmd_private.h | 2
On Wed, May 07, 2025 at 04:22:10PM +0100, Anatoly Burakov wrote:
> Remove custom number parser and use C standard library instead. In order to
> keep compatibility with earlier versions of the parser, we have to take
> into account a few quirks:
>
> - We do not consider "negative" numbers to be va
47 matches
Mail list logo
