mbox series

[iproute2-next,0/9] Add tc-mqprio and tc-taprio support for preemptible traffic classes

Message ID 20230403105245.2902376-1-vladimir.oltean@nxp.com (mailing list archive)
Headers show
Series Add tc-mqprio and tc-taprio support for preemptible traffic classes | expand

Message

Vladimir Oltean April 3, 2023, 10:52 a.m. UTC
This is the iproute2 support for the tc program to make use of the
kernel feature for preemptible traffic classes described here:
https://patchwork.kernel.org/project/netdevbpf/cover/20230403103440.2895683-1-vladimir.oltean@nxp.com/

The state of the man pages prior to this work was a bit unsatisfactory,
so patches 03-07 contain some man page cleanup in tc-taprio(8) and
tc-mqprio(8).

Vladimir Oltean (9):
  uapi: add definitions for preemptible traffic classes in mqprio and
    taprio
  utils: add max() definition
  tc/taprio: add max-sdu to the man page SYNOPSIS section
  tc/taprio: add a size table to the examples from the man page
  tc/mqprio: fix stray ] in man page synopsis
  tc/mqprio: use words in man page to express min_rate/max_rate
    dependency on bw_rlimit
  tc/mqprio: break up synopsis into multiple lines
  tc/mqprio: add support for preemptible traffic classes
  tc/taprio: add support for preemptible traffic classes

 include/uapi/linux/pkt_sched.h | 17 ++++++
 include/utils.h                |  8 +++
 man/man8/tc-mqprio.8           | 92 ++++++++++++++++++++++---------
 man/man8/tc-taprio.8           | 27 ++++++++--
 tc/q_mqprio.c                  | 98 +++++++++++++++++++++++++++++++++
 tc/q_taprio.c                  | 99 +++++++++++++++++++++++++---------
 6 files changed, 289 insertions(+), 52 deletions(-)

Comments

Ferenc Fejes April 3, 2023, 11:18 a.m. UTC | #1
Hi Vladimir!

On Mon, 2023-04-03 at 13:52 +0300, Vladimir Oltean wrote:
> This is the iproute2 support for the tc program to make use of the
> kernel feature for preemptible traffic classes described here:
> https://patchwork.kernel.org/project/netdevbpf/cover/20230403103440.2895683-1-vladimir.oltean@nxp.com/
> 
> The state of the man pages prior to this work was a bit
> unsatisfactory,
> so patches 03-07 contain some man page cleanup in tc-taprio(8) and
> tc-mqprio(8).
> 
> Vladimir Oltean (9):
>   uapi: add definitions for preemptible traffic classes in mqprio and
>     taprio
>   utils: add max() definition
>   tc/taprio: add max-sdu to the man page SYNOPSIS section
>   tc/taprio: add a size table to the examples from the man page

Seems like Stephen merged Péter's manpages patch [1] but IMO your
version [2] is a better overhaul of that, also Péter ACK-ed to go
forward with that version. Looks like you rebased this work on the new
manpages, you have any plan to submit the changes from [2] separately?
Probably Stephen missed the whole discussion and about [2] and I'm
admit that putting acked/reviewed into a mail inside the discussion
might be misleading (probably thats show up for the original patch in
patchwork). Sorry for making it complicated.

>   tc/mqprio: fix stray ] in man page synopsis
>   tc/mqprio: use words in man page to express min_rate/max_rate
>     dependency on bw_rlimit
>   tc/mqprio: break up synopsis into multiple lines
>   tc/mqprio: add support for preemptible traffic classes
>   tc/taprio: add support for preemptible traffic classes
> 
>  include/uapi/linux/pkt_sched.h | 17 ++++++
>  include/utils.h                |  8 +++
>  man/man8/tc-mqprio.8           | 92 ++++++++++++++++++++++---------
>  man/man8/tc-taprio.8           | 27 ++++++++--
>  tc/q_mqprio.c                  | 98
> +++++++++++++++++++++++++++++++++
>  tc/q_taprio.c                  | 99 +++++++++++++++++++++++++-------
> --
>  6 files changed, 289 insertions(+), 52 deletions(-)
> 

[1]
https://lore.kernel.org/netdev/167789641838.26474.2747633103367439718.git-patchwork-notify@kernel.org/

[2]
https://lore.kernel.org/netdev/20230220161809.t2vj6daixio7uzbw@skbuf/

Best,
Ferenc
Vladimir Oltean April 3, 2023, 11:23 a.m. UTC | #2
Hi Ferenc,

On Mon, Apr 03, 2023 at 01:18:07PM +0200, Ferenc Fejes wrote:
> Seems like Stephen merged Péter's manpages patch [1] but IMO your
> version [2] is a better overhaul of that, also Péter ACK-ed to go
> forward with that version. Looks like you rebased this work on the new
> manpages, you have any plan to submit the changes from [2] separately?
> Probably Stephen missed the whole discussion and about [2] and I'm
> admit that putting acked/reviewed into a mail inside the discussion
> might be misleading (probably thats show up for the original patch in
> patchwork). Sorry for making it complicated.
> [1]
> https://lore.kernel.org/netdev/167789641838.26474.2747633103367439718.git-patchwork-notify@kernel.org/
> 
> [2]
> https://lore.kernel.org/netdev/20230220161809.t2vj6daixio7uzbw@skbuf/

Yes, this is true. I still have the delta between Péter's merged version
and my suggested changes, but it needs to be broken up into a gazillion
smaller patches which I haven't done yet. I now also doubt the value of
some of those changes as standalone patches. I wanted to get the preemption
stuff over with first, and this is why I've submitted only what I have.
Ferenc Fejes April 3, 2023, 11:30 a.m. UTC | #3
Hi Vladimir!

On Mon, 2023-04-03 at 14:23 +0300, Vladimir Oltean wrote:
> Hi Ferenc,
> 
> On Mon, Apr 03, 2023 at 01:18:07PM +0200, Ferenc Fejes wrote:
> > Seems like Stephen merged Péter's manpages patch [1] but IMO your
> > version [2] is a better overhaul of that, also Péter ACK-ed to go
> > forward with that version. Looks like you rebased this work on the
> > new
> > manpages, you have any plan to submit the changes from [2]
> > separately?
> > Probably Stephen missed the whole discussion and about [2] and I'm
> > admit that putting acked/reviewed into a mail inside the discussion
> > might be misleading (probably thats show up for the original patch
> > in
> > patchwork). Sorry for making it complicated.
> > [1]
> > https://lore.kernel.org/netdev/167789641838.26474.2747633103367439718.git-patchwork-notify@kernel.org/
> > 
> > [2]
> > https://lore.kernel.org/netdev/20230220161809.t2vj6daixio7uzbw@skbuf/
> 
> Yes, this is true. I still have the delta between Péter's merged
> version
> and my suggested changes, but it needs to be broken up into a
> gazillion
> smaller patches which I haven't done yet. I now also doubt the value
> of
> some of those changes as standalone patches. I wanted to get the
> preemption
> stuff over with first, and this is why I've submitted only what I
> have.

Understandable, thanks for the info. With all respect of your time, it
would be nice if those changes make their way into the manpages in one
way or another. I think it has enough value in its own too, but I agree
it has much lower prio than having FP merged and running :-)

Ferenc