netdev - Re: [PATCH v3 net 05/17] ratelimit: Fix data-races in __

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

Date:   Mon, 22 Aug 2022 12:22:35 -0700
From:   Eric Dumazet <edumazet@...gle.com>
To:     Kuniyuki Iwashima <kuniyu@...zon.com>
Cc:     David Miller <davem@...emloft.net>,
        Jakub Kicinski <kuba@...nel.org>,
        Kuniyuki Iwashima <kuni1840@...il.com>,
        netdev <netdev@...r.kernel.org>, Paolo Abeni <pabeni@...hat.com>
Subject: Re: [PATCH v3 net 05/17] ratelimit: Fix data-races in ___ratelimit().

On Mon, Aug 22, 2022 at 12:15 PM Kuniyuki Iwashima <kuniyu@...zon.com> wrote:
>
> From:   Eric Dumazet <edumazet@...gle.com>
> Date:   Mon, 22 Aug 2022 12:00:11 -0700
> > On Thu, Aug 18, 2022 at 11:29 AM Kuniyuki Iwashima <kuniyu@...zon.com> wrote:
> > >
> > > While reading rs->interval and rs->burst, they can be changed
> > > concurrently.  Thus, we need to add READ_ONCE() to their readers.
> > >
> > > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> > > Signed-off-by: Kuniyuki Iwashima <kuniyu@...zon.com>
> > > ---
> > >  lib/ratelimit.c | 8 +++++---
> > >  1 file changed, 5 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/lib/ratelimit.c b/lib/ratelimit.c
> > > index e01a93f46f83..b59a1d3d0cc3 100644
> > > --- a/lib/ratelimit.c
> > > +++ b/lib/ratelimit.c
> > > @@ -26,10 +26,12 @@
> > >   */
> > >  int ___ratelimit(struct ratelimit_state *rs, const char *func)
> > >  {
> > > +       int interval = READ_ONCE(rs->interval);
> > > +       int burst = READ_ONCE(rs->burst);
> >
> > I thought rs->interval and rs->burst were constants...
> >
> > Can you point to the part where they are changed ?
> >
> > Ideally such a patch should also add corresponding WRITE_ONCE(), and
> > comments to pair them,
> >  this would really help reviewing it.
>
> In this case, &net_ratelimit_state.(burst|interval) are directly
> passed to proc_handlers, and exactly the relation is unclear.
>
> As Jakub pointed out, two reads can be inconsistent, so I'll add
> a spin lock in struct ratelimit_state and two dedicated proc
> handlers for each member.

This seems overkill to me... Adding a comment explaining why a race
(or inconsistency) is acceptable is enough I think.

Otherwise, we will have to review all other 'struct ratelimit_state'
which expose
in r/w mode their @interval or @burst field.


>  Then, I'll add few more comments to
> make that relation clear.
>
> Thanks for feedback!
>
>
> > >         unsigned long flags;
> > >         int ret;
> > >
> > > -       if (!rs->interval)
> > > +       if (!interval)
> > >                 return 1;
> > >
> > >         /*
> > > @@ -44,7 +46,7 @@ int ___ratelimit(struct ratelimit_state *rs, const char *func)
> > >         if (!rs->begin)
> > >                 rs->begin = jiffies;
> > >
> > > -       if (time_is_before_jiffies(rs->begin + rs->interval)) {
> > > +       if (time_is_before_jiffies(rs->begin + interval)) {
> > >                 if (rs->missed) {
> > >                         if (!(rs->flags & RATELIMIT_MSG_ON_RELEASE)) {
> > >                                 printk_deferred(KERN_WARNING
> > > @@ -56,7 +58,7 @@ int ___ratelimit(struct ratelimit_state *rs, const char *func)
> > >                 rs->begin   = jiffies;
> > >                 rs->printed = 0;
> > >         }
> > > -       if (rs->burst && rs->burst > rs->printed) {
> > > +       if (burst && burst > rs->printed) {
> > >                 rs->printed++;
> > >                 ret = 1;
> > >         } else {
> > > --
> > > 2.30.2
> > >