Re: [PATCH net v2 1/3] net/sched: taprio: catch up in bounded time when the schedule falls behind

[Junjie Cao <junjie.cao@intel.com>]

On Mon, 24 Aug 2026 11:53:35 -0700, Jakub Kicinski <kuba@kernel.org> wrote:
> No Link: tags present in the commit header.

v3 will carry the syzbot tags, including a fourth bucket akpm had syzbot
test this patch alone against:
https://lore.kernel.org/all/6a936ec2.4d659fcc.734b4.0032.GAE@google.com/

> has a second exit condition (--budget), so taprio_catch_up() can return
[...]
> jump makes end == now, and the strict ktime_before(end, now) test then skips

Right. v3 treats end == now as behind, in the walk and its caller, and
the comment mentions the cap.

> case, so a large gate control list turns a previously O(1) critical section
> into an O(num_entries) walk over separately allocated entries.

One lap at most per hold now, versus no bound at all before. Leaving it
as is.

> inoperative. Adding N * cycle_time then shifts the timeline by
> (N * cycle_time) modulo the interval sum, and the gate_close_time[] values

Yes. v3 jumps whole laps instead: min(cycle_time, sum of intervals) on
the entry timeline, cycle_time on cycle_end_time, i.e. exactly what N
restarts of the current code do.

> then terminates on budget exhaustion with end < now, and behind stays below
[...]
> changelog describes does not converge at all. Is the intent that this patch
> depends on the minimum interval patch for the bound to hold?

No. With the jump in laps the residual is under one lap, so the walk
finishes within num_entries steps for any schedule. v3 drops 2/3 anyway.

> be observed torn, which is also pre-existing. Is it worth a comment that
> advance_sched() may legitimately update the currently published entry?

Yes, will add.