[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20220712181457.41424-1-kuniyu@amazon.com>
Date: Tue, 12 Jul 2022 11:14:57 -0700
From: Kuniyuki Iwashima <kuniyu@...zon.com>
To: <edumazet@...gle.com>
CC: <davem@...emloft.net>, <kuba@...nel.org>, <kuni1840@...il.com>,
<kuniyu@...zon.com>, <netdev@...r.kernel.org>, <pabeni@...hat.com>,
<subashab@...eaurora.org>
Subject: Re: [PATCH v2 net] tcp/udp: Make early_demux back namespacified.
From: Eric Dumazet <edumazet@...gle.com>
Date: Tue, 12 Jul 2022 19:51:02 +0200
> On Tue, Jul 12, 2022 at 7:38 PM Kuniyuki Iwashima <kuniyu@...zon.com> wrote:
> >
> > Commit e21145a9871a ("ipv4: namespacify ip_early_demux sysctl knob") made
> > it possible to enable/disable early_demux on a per-netns basis. Then, we
> > introduced two knobs, tcp_early_demux and udp_early_demux, to switch it for
> > TCP/UDP in commit dddb64bcb346 ("net: Add sysctl to toggle early demux for
> > tcp and udp"). However, the .proc_handler() was wrong and actually
> > disabled us from changing the behaviour in each netns.
> >
>
> > static int proc_tfo_blackhole_detect_timeout(struct ctl_table *table,
> > int write, void *buffer,
> > size_t *lenp, loff_t *ppos)
> > @@ -695,14 +640,18 @@ static struct ctl_table ipv4_net_table[] = {
> > .data = &init_net.ipv4.sysctl_udp_early_demux,
> > .maxlen = sizeof(u8),
> > .mode = 0644,
> > - .proc_handler = proc_udp_early_demux
> > + .proc_handler = proc_dou8vec_minmax,
> > + .extra1 = SYSCTL_ZERO,
> > + .extra2 = SYSCTL_ONE,
>
> This does not belong to this patch.
>
> It is IMO too late, some users might use:
>
> echo 2 >/proc/sys/net/ipv4/udp_early_demux
Ok, I will drop these.
> > },
> > {
> > .procname = "tcp_early_demux",
> > .data = &init_net.ipv4.sysctl_tcp_early_demux,
> > .maxlen = sizeof(u8),
> > .mode = 0644,
> > - .proc_handler = proc_tcp_early_demux
> > + .proc_handler = proc_dou8vec_minmax,
> > + .extra1 = SYSCTL_ZERO,
> > + .extra2 = SYSCTL_ONE,
>
> Same here.
>
> Again, fix the bug, and only the bug. Do not hide 'fixes' in an innocent patch.
>
> There is a reason for that, we want each commit to have a clear description,
> and we want to be able to revert a patch without having to think about
> what needs
> to be re-written.
Sorry for bothering you.
But this makes my criteria explicit, thank you!
Powered by blists - more mailing lists