netdev - Re: [PATCH net v3 2/2] selftests/tc-testing: Check backlog stats in gso

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <eJjkeB42Nf7tzAJvIyM89SCEMP7FEZtZ6cp0LC0k-P2Ok4Io0UooUAwPMGqlZ0qa-IgYUe9Wpb1X2vJzKChin9r7CKVMo9LjrTXfbo5HDy0=@willsroot.io>
Date: Sun, 27 Jul 2025 23:03:11 +0000
From: William Liu <will@...lsroot.io>
To: Victor Nogueira <victor@...atatu.com>
Cc: netdev@...r.kernel.org, jhs@...atatu.com, xiyou.wangcong@...il.com, pabeni@...hat.com, kuba@...nel.org, jiri@...nulli.us, davem@...emloft.net, edumazet@...gle.com, horms@...nel.org, savy@...t3mfailure.io
Subject: Re: [PATCH net v3 2/2] selftests/tc-testing: Check backlog stats in gso_skb case

On Sunday, July 27th, 2025 at 2:17 PM, Victor Nogueira <victor@...atatu.com> wrote:

> 
> 
> On 7/26/25 20:50, William Liu wrote:
> 
> > Add tests to ensure proper backlog accounting in hhf, codel, pie, fq,
> > fq_pie, and fq_codel qdiscs. For hhf, codel, and pie, we check for
> > the correct packet count and packet size. For fq, fq_pie, and fq_codel,
> > we check to make sure the backlog statistics do not underflow in tbf
> > after removing those qdiscs, which was an original bug symptom.
> > 
> > Signed-off-by: William Liu will@...lsroot.io
> > Reviewed-by: Savino Dicanosa savy@...t3mfailure.io
> > ---
> > v2 -> v3:
> > - Simplify ping command in test cases
> > ---
> > .../tc-testing/tc-tests/infra/qdiscs.json | 195 ++++++++++++++++++
> > 1 file changed, 195 insertions(+)
> 
> 
> It seems like some test cases broke in this new iteration:
> 
> # not ok 670 34c0 - Test TBF with HHF Backlog Accounting in gso_skb case
> # Could not match regex pattern. Verify command output:
> # qdisc tbf 1: root refcnt 2 rate 8bit burst 100b lat 0us
> # Sent 98 bytes 1 pkt (dropped 0, overlimits 1 requeues 0)
> # backlog 98b 1p requeues 0
> # qdisc hhf 2: parent 1:1 limit 1p quantum 1514b hh_limit 2048
> reset_timeout 40ms admit_bytes 128Kb evict_timeout 1s non_hh_weight 2
> # Sent 196 bytes 2 pkt (dropped 0, overlimits 0 requeues 0)
> # backlog 98b 1p requeues 0
> # drop_overlimit 0 hh_overlimit 0 tot_hh 0 cur_hh 0
> #
> # not ok 671 fd68 - Test TBF with CODEL Backlog Accounting in gso_skb
> case# not ok 671 fd68 - Test TBF with CODEL Backlog Accounting in
> gso_skb case
> # Could not match regex pattern. Verify command output:
> # qdisc tbf 1: root refcnt 2 rate 8bit burst 100b lat 0us
> # Sent 98 bytes 1 pkt (dropped 0, overlimits 1 requeues 0)
> # backlog 98b 1p requeues 0
> # qdisc codel 2: parent 1:1 limit 1p target 5ms interval 100ms
> # Sent 196 bytes 2 pkt (dropped 0, overlimits 0 requeues 0)
> # backlog 98b 1p requeues 0
> # count 0 lastcount 0 ldelay 1us drop_next 0us
> # maxpacket 98 ecn_mark 0 drop_overlimit 0
> #
> # not ok 672 514e - Test TBF with PIE Backlog Accounting in gso_skb
> case# not ok 672 514e - Test TBF with PIE Backlog Accounting in gso_skb case
> # Could not match regex pattern. Verify command output:
> # qdisc tbf 1: root refcnt 2 rate 8bit burst 100b lat 0us
> # Sent 98 bytes 1 pkt (dropped 0, overlimits 1 requeues 0)
> # backlog 98b 1p requeues 0
> # qdisc pie 2: parent 1:1 limit 1p target 15ms tupdate 15ms alpha 2 beta 20
> # Sent 196 bytes 2 pkt (dropped 0, overlimits 0 requeues 0)
> # backlog 98b 1p requeues 0
> # prob 0 delay 0us
> # pkts_in 2 overlimit 0 dropped 0 maxq 0 ecn_mark 0
> 
> cheers,
> Victor

Hmm, that's strange. Those tests run fine on my machine, but it seems like the ping packet sizes are different (1 packet for me is 70b)? But running it manually gets me 98b too...

I will just make these tests follow the same pattern as the tests I added for fq, fq_pie, and fq_codel in this case.