Re: [PATCH v2 net 2/2] net/sched: taprio: make qdisc_leaf() see the per-netdev-queue pfifo child qdiscs

[Vladimir Oltean <olteanv@gmail.com>]

+Greg, Sasha.

Hi Kurt,

On Wed, Feb 22, 2023 at 03:03:04PM +0100, Kurt Kanzenbach wrote:
> This commit was backported to v5.15-LTS which results in NULL pointer
> dereferences e.g., when attaching an ETF child qdisc to taprio.
> 
> From what I can see is, that the issue was reported back then and this
> commit was reverted [1]. However, the revert didn't make it into
> v5.15-LTS? Is there a reason for it? I'm testing 5.15.94-rt59 here.
> 
> Thanks,
> Kurt
> 
> [1] - https://lore.kernel.org/all/20221004220100.1650558-1-vladimir.oltean@nxp.com/

You are right; the patchwork-bot clearly says that the revert was
applied to net.git as commit af7b29b1deaa ("Revert "net/sched: taprio:
make qdisc_leaf() see the per-netdev-queue pfifo child qdiscs""), but
the revert never made it to stable.

OTOH, the original patch did make it to, and still is in, linux-stable.
I have backport notification emails of the original to 5.4, 5.10, 5.15, 5.19.

Greg, Sasha, can you please pick up and backport commit af7b29b1deaa
("Revert "net/sched: taprio: make qdisc_leaf() see the per-netdev-queue
pfifo child qdiscs"") to the currently maintained stable kernels?

Thanks.