[PATCH 0/5] net/mlx5: add BlueField socket direct support

[Bing Zhao]

Hi,

> -----Original Message-----
> From: Dariusz Sosnowski <dsosnowski at nvidia.com>
> Sent: Monday, March 2, 2026 7:35 PM
> To: Slava Ovsiienko <viacheslavo at nvidia.com>; Bing Zhao
> <bingz at nvidia.com>; Ori Kam <orika at nvidia.com>; Suanming Mou
> <suanmingm at nvidia.com>; Matan Azrad <matan at nvidia.com>
> Cc: dev at dpdk.org; Raslan Darawsheh <rasland at nvidia.com>
> Subject: [PATCH 0/5] net/mlx5: add BlueField socket direct support
> 
> Goal of this patchset is to prepare probing logic in mlx5 networking PMD
> for support of BlueField DPUs with Socket Direct.
> In such use case, BlueField DPU will be connected through PCI to 2
> different CPUs on the host.
> Each host CPU sees 2 PFs.
> Each PF is connected to one of the physical ports.
> 
>     +--------+               +--------+
>     |CPU 0   |               |CPU 1   |
>     |        |               |        |
>     |    pf0 |               |    pf0 |
>     |        |               |        |
>     |    pf1 |               |    pf1 |
>     |        |               |        |
>     +---+----+               +-+------+
>         |                      |
>         |                      |
>         |                      |
>         +----+           +-----+
>              |           |
>              |           |
>              |           |
>          +---+-----------+----+
>          |BF3 DPU             |
>          |                    |
>          | pf0hpf      pf1hpf |
>          |                    |
>          | pf2hpf      pf3hpf |
>          |                    |
>          |  p0          p1    |
>          +------+------+------+
>          | phy0 |      | phy1 |
>          +------+      +------+
> 
> 
> On BlueField DPU ARM Linux netdevs map to PFs/ports as follows:
> 
> - p0 and p1 to physical ports 0 and 1 respectively,
> - pf0hpf and pf2hpf to CPU0 pf0 and CPU1 pf0 respectively,
> - pf1hpf and pf3hpf to CPU0 pf1 and CPU1 pf1 respectively.
> 
> There are several possible ways to use such a setup:
> 
> - Single E-Switch (embedded switch) per each CPU PF to
>   physical port connection.
> - Shared E-Switch for related CPU PFs:
>     - For example, both pf0hpf and pf2hpf are in the same E-Switch domain.
> - Multiport E-Switch.
>     - All host PFs and physical ports are in the same E-Switch domain.
> 
> When a DPDK application would be run on BlueField ARM it should be
> possible for application to probe all the relevant representors
> (corresponding to available netdevs).
> Using testpmd syntax users will be able to do the following:
> 
>     # Probe both physical ports
>     port attach 03:00.0,dv_flow_en=2,representor=pf0-1
> 
>     # Probe both host PF 0 from CPU 0
>     # (VF representor index -1 is special encoding for host PF)
>     port attach 03:00.0,dv_flow_en=2,representor=pf0vf65535
>     # or with explicit controller index
>     port attach 03:00.0,dv_flow_en=2,representor=c1pf0vf65535
> 
>     # Probe both host PF 0 from CPU 1
>     port attach 03:00.0,dv_flow_en=2,representor=pf2vf65535
>     # or with explicit controller index
>     port attach 03:00.0,dv_flow_en=2,representor=c2pf2vf65535
> 
> Patches overview:
> 
> - Patch 1 and 2 - Fixes bond detection logic.
>   Previously mlx5 PMD relied on "bond" appearing in IB device name
>   which is not always the case. Moved to sysfs checks for bonding devices.
> - Patch 3 - Add calculation of number of physical ports and host PFs.
>   This information will be used to determine
>   how DPDK port name is generated, instead of relying on
>   specific setup type.
> - Patch 4 - Change "representor to IB port" matching logic to directly
>   compare ethdev devargs values to IB port info.
>   Added optional matching on controller index.
> - Patch 5 - Make DPDK port name generation dynamic and dependent on
>   types/number of ports, instead of specific setup type.
>   This allows more generic probing, independent of setup topology.
> 
> Dariusz Sosnowski (5):
>   common/mlx5: fix bond check
>   net/mlx5: fix bond check
>   net/mlx5: calculate number of uplinks and host PFs
>   net/mlx5: compare representors explicitly
>   net/mlx5: build port name dynamically
> 
>  drivers/common/mlx5/linux/mlx5_common_os.c |  86 ++++-
>  drivers/common/mlx5/linux/mlx5_common_os.h |   9 +
>  drivers/net/mlx5/linux/mlx5_os.c           | 356 ++++++++++++++-------
>  drivers/net/mlx5/mlx5.h                    |   2 +
>  4 files changed, 338 insertions(+), 115 deletions(-)
> 
> --
> 2.47.3

Series Acked-by: Bing Zhao <bingz at nvidia.com>