Content-Type: multipart/mixed; boundary="===============7077953738524869031==" MIME-Version: 1.0 From: Pablo Neira Ayuso To: lkp@lists.01.org Subject: Re: [PATCH nf] netfilter: nft_compat: remove flush counter optimization Date: Mon, 10 Aug 2020 13:03:51 +0200 Message-ID: <20200810110351.GA1613@salvia> In-Reply-To: <20200809182801.9315-1-fw@strlen.de> List-Id: --===============7077953738524869031== Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: quoted-printable On Sun, Aug 09, 2020 at 08:28:01PM +0200, Florian Westphal wrote: > WARNING: CPU: 1 PID: 16059 at lib/refcount.c:31 refcount_warn_saturate+0x= df/0xf > [..] > __nft_mt_tg_destroy+0x42/0x50 [nft_compat] > nft_target_destroy+0x63/0x80 [nft_compat] > nf_tables_expr_destroy+0x1b/0x30 [nf_tables] > nf_tables_rule_destroy+0x3a/0x70 [nf_tables] > nf_tables_exit_net+0x186/0x3d0 [nf_tables] > = > Happens when a compat expr is destoyed from abort path. > There is no functional impact; after this work queue is flushed > unconditionally if its pending. > = > This removes the waitcount optimization. Test of repeated > iptables-restore of a ~60k kubernetes ruleset doesn't indicate > a slowdown. In case the counter is needed after all for some workloads > we can revert this and increment the refcount for the > !=3D NFT_PREPARE_TRANS case to avoid the increment/decrement imbalance. > = > While at it, also flush for match case, this was an oversight > in the original patch. Applied, thanks. --===============7077953738524869031==--