Re: [patch net-next v2 00/11] devlink: introduce dump selector attr and use it for per-instance dumps

[Jiri Pirko <jiri@resnulli.us>]

I see that this patchset got moved to "changes requested" in patchwork.
Why exacly? There was no comment so far. Petr's splat is clearly not
caused by this patchset.

Should I resubmit?

Thanks!

Thu, Jul 20, 2023 at 02:18:18PM CEST, jiri@resnulli.us wrote:
>From: Jiri Pirko <jiri@nvidia.com>
>
>Motivation:
>
>For SFs, one devlink instance per SF is created. There might be
>thousands of these on a single host. When a user needs to know port
>handle for specific SF, he needs to dump all devlink ports on the host
>which does not scale good.
>
>Solution:
>
>Allow user to pass devlink handle alongside the dump command
>and dump only objects which are under selected devlink instance.
>
>Introduce new attr DEVLINK_ATTR_DUMP_SELECTOR to nest the selection
>attributes. This way the userspace can use maxattr to tell if dump
>selector is supported by kernel or not.
>
>Assemble netlink policy for selector attribute. If user passes attr
>unknown to kernel, netlink validation errors out.
>
>Example:
>$ devlink port show
>auxiliary/mlx5_core.eth.0/65535: type eth netdev eth2 flavour physical port 0 splittable false
>auxiliary/mlx5_core.eth.1/131071: type eth netdev eth3 flavour physical port 1 splittable false
>
>$ devlink port show auxiliary/mlx5_core.eth.0
>auxiliary/mlx5_core.eth.0/65535: type eth netdev eth2 flavour physical port 0 splittable false
>
>$ devlink port show auxiliary/mlx5_core.eth.1
>auxiliary/mlx5_core.eth.1/131071: type eth netdev eth3 flavour physical port 1 splittable false
>
>This is done in patch #10
>
>Dependency:
>
>The DEVLINK_ATTR_DUMP_SELECTOR parsing is very suitable to be done
>once at the beginning of the dumping. Unfortunatelly, it is not possible
>to define start() and done() callbacks for netlink small ops.
>So all commands that use instance iterator for dumpit are converted to
>split ops. This is done in patch #1-9
>
>Extension:
>
>patch #11 extends the selector by port index for health reporter
>dumping.
>
>v1->v2:
>- the original single patch (patch #10) was extended to a patchset
>
>Jiri Pirko (11):
>  devlink: parse linecard attr in doit() callbacks
>  devlink: parse rate attrs in doit() callbacks
>  devlink: introduce __devlink_nl_pre_doit() with internal flags as
>    function arg
>  devlink: convert port get command to split ops
>  devlink: convert health reporter get command to split ops
>  devlink: convert param get command to split ops
>  devlink: convert trap get command to split ops
>  devlink: introduce set of macros and use it for split ops definitions
>  devlink: convert rest of the iterator dumpit commands to split ops
>  devlink: introduce dump selector attr and use it for per-instance
>    dumps
>  devlink: extend health reporter dump selector by port index
>
> include/uapi/linux/devlink.h |   2 +
> net/devlink/devl_internal.h  |  42 +++---
> net/devlink/health.c         |  21 ++-
> net/devlink/leftover.c       | 211 ++++++++--------------------
> net/devlink/netlink.c        | 263 ++++++++++++++++++++++++++++++-----
> 5 files changed, 333 insertions(+), 206 deletions(-)
>
>-- 
>2.41.0
>