RE: [PATCH 20.11 0/5] mlx5: some fixes

[Michael Baum <michaelba@nvidia.com>]

Hi

> -----Original Message-----
> From: Luca Boccassi <bluca@debian.org>
> Sent: Tuesday, February 22, 2022 4:33 PM
> To: Michael Baum <michaelba@nvidia.com>; stable@dpdk.org
> Cc: Matan Azrad <matan@nvidia.com>; Slava Ovsiienko
> <viacheslavo@nvidia.com>; Alexander Kozyrev <akozyrev@nvidia.com>
> Subject: Re: [PATCH 20.11 0/5] mlx5: some fixes
> 
> External email: Use caution opening links or attachments
> 
> 
> On Mon, 2022-02-21 at 21:46 +0200, Michael Baum wrote:
> > Backport some fixes to 20.11.5
> >
> > The upstream commits:
> >  - 10599cf83e  common/mlx5: add minimum WQE size for striding RQ
> >  - 8928997a13  common/mlx5: fix error handling in multi-class probe
> >  - 147f6fb42b  net/mlx5: fix memory socket selection in ASO management
> >  - 34776af600  net/mlx5: fix MPRQ stride devargs adjustment
> >  - 0947ed380f  net/mlx5: improve stride parameter names
> >
> > The following commit [1] sent by Alexander Kozyrev fixes one of them
> > [2], so I integrated it inside.
> >
> > [1] 728b6447e7  net/mlx5: fix MPRQ WQE size assertion
> >
> > [2] 34776af600  net/mlx5: fix MPRQ stride devargs adjustment
> >
> > Cc: akozyrev@nvidia.com
> >
> > Michael Baum (5):
> >   common/mlx5: add minimum WQE size for striding RQ
> >   net/mlx5: improve stride parameter names
> >   net/mlx5: fix MPRQ stride devargs adjustment
> >   net/mlx5: fix memory socket selection in ASO management
> >   common/mlx5: fix error handling in multi-class probe
> >
> >  drivers/common/mlx5/mlx5_common_pci.c |   7 +-
> >  drivers/common/mlx5/mlx5_devx_cmds.c  |  28 +++
> >  drivers/common/mlx5/mlx5_devx_cmds.h  |   1 +
> >  drivers/common/mlx5/mlx5_prm.h        |  38 +++-
> >  drivers/net/mlx5/linux/mlx5_os.c      |  48 ++---
> >  drivers/net/mlx5/linux/mlx5_verbs.c   |   4 +-
> >  drivers/net/mlx5/mlx5.c               |   4 +-
> >  drivers/net/mlx5/mlx5.h               |  12 +-
> >  drivers/net/mlx5/mlx5_defs.h          |   4 +-
> >  drivers/net/mlx5/mlx5_devx.c          |   4 +-
> >  drivers/net/mlx5/mlx5_flow_age.c      |  27 ++-
> >  drivers/net/mlx5/mlx5_rxq.c           | 249 +++++++++++++++++---------
> >  drivers/net/mlx5/mlx5_rxtx.c          |  22 +--
> >  drivers/net/mlx5/mlx5_rxtx.h          |  10 +-
> >  drivers/net/mlx5/mlx5_rxtx_vec.c      |   8 +-
> >  15 files changed, 290 insertions(+), 176 deletions(-)
> >
> 
> The second patch does not apply:
> 
> --- a/drivers/net/mlx5/mlx5_rxq.c
> +++ b/drivers/net/mlx5/mlx5_rxq.c
> @@@ -205,9 -205,9 +205,15 @@@ rxq_alloc_elts_sprq(struct mlx5_rxq_ctr
>   {
>         const unsigned int sges_n = 1 << rxq_ctrl->rxq.sges_n;
>         unsigned int elts_n = mlx5_rxq_mprq_enabled(&rxq_ctrl->rxq) ?
> ++<<<<<<< HEAD
>  +              (1 << rxq_ctrl->rxq.elts_n) * (1 << rxq_ctrl->rxq.strd_num_n) :
>  +              (1 << rxq_ctrl->rxq.elts_n);
>  +      bool has_vec_support = mlx5_rxq_check_vec_support(&rxq_ctrl->rxq)
> > 0;
> ++=======
> +                             RTE_BIT32(rxq_ctrl->rxq.elts_n) *
> +                             RTE_BIT32(rxq_ctrl->rxq.log_strd_num) :
> +                             RTE_BIT32(rxq_ctrl->rxq.elts_n);
> ++>>>>>>> net/mlx5: improve stride parameter names
>         unsigned int i;
>         int err;
> 
> Did you test it with the 20.11 branch from the staging area at
> https://github.com/bluca/dpdk-stable.git ?

I'm sorry, I made a mistake in the branch.
I'm sending v2.

> 
> --
> Kind regards,
> Luca Boccassi