From: "Paul E. McKenney" <paulmck@kernel.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: tglx@linutronix.de, x86@kernel.org, elver@google.com,
kasan-dev@googlegroups.com, linux-kernel@vger.kernel.org,
will@kernel.org, dvyukov@google.com, glider@google.com,
andreyknvl@google.com
Subject: Re: [PATCH 2/9] rcu: Fixup noinstr warnings
Date: Wed, 3 Jun 2020 20:34:09 -0700 [thread overview]
Message-ID: <20200604033409.GX29598@paulmck-ThinkPad-P72> (raw)
In-Reply-To: <20200603171320.GE2570@hirez.programming.kicks-ass.net>
On Wed, Jun 03, 2020 at 07:13:20PM +0200, Peter Zijlstra wrote:
> On Wed, Jun 03, 2020 at 09:46:00AM -0700, Paul E. McKenney wrote:
>
> > > --- a/kernel/rcu/tree.c
> > > +++ b/kernel/rcu/tree.c
> > > @@ -250,7 +250,7 @@ static noinstr void rcu_dynticks_eqs_ent
> > > * next idle sojourn.
> > > */
> > > rcu_dynticks_task_trace_enter(); // Before ->dynticks update!
> > > - seq = atomic_add_return(RCU_DYNTICK_CTRL_CTR, &rdp->dynticks);
> > > + seq = arch_atomic_add_return(RCU_DYNTICK_CTRL_CTR, &rdp->dynticks);
> >
> > To preserve KCSAN's ability to see this, there would be something like
> > instrument_atomic_write(&rdp->dynticks, sizeof(rdp->dynticks)) prior
> > to the instrumentation_end() invoked before rcu_dynticks_eqs_enter()
> > in each of rcu_eqs_enter() and rcu_nmi_exit(), correct?
>
> Yes.
>
> > > // RCU is no longer watching. Better be in extended quiescent state!
> > > WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) &&
> > > (seq & RCU_DYNTICK_CTRL_CTR));
> > > @@ -274,13 +274,13 @@ static noinstr void rcu_dynticks_eqs_exi
> > > * and we also must force ordering with the next RCU read-side
> > > * critical section.
> > > */
> > > - seq = atomic_add_return(RCU_DYNTICK_CTRL_CTR, &rdp->dynticks);
> > > + seq = arch_atomic_add_return(RCU_DYNTICK_CTRL_CTR, &rdp->dynticks);
> >
> > And same here, but after the instrumentation_begin() following
> > rcu_dynticks_eqs_exit() in both rcu_eqs_exit() and rcu_nmi_enter(),
> > correct?
>
> Yep.
>
> > > // RCU is now watching. Better not be in an extended quiescent state!
> > > rcu_dynticks_task_trace_exit(); // After ->dynticks update!
> > > WARN_ON_ONCE(IS_ENABLED(CONFIG_RCU_EQS_DEBUG) &&
> > > !(seq & RCU_DYNTICK_CTRL_CTR));
> > > if (seq & RCU_DYNTICK_CTRL_MASK) {
> > > - atomic_andnot(RCU_DYNTICK_CTRL_MASK, &rdp->dynticks);
> > > + arch_atomic_andnot(RCU_DYNTICK_CTRL_MASK, &rdp->dynticks);
> >
> > This one is gone in -rcu.
>
> Good, because that would make things 'complicated' with the external
> instrumentation call. And is actually the reason I didn't even attempt
> it this time around.
>
> > > smp_mb__after_atomic(); /* _exit after clearing mask. */
> > > }
> > > }
> > > @@ -313,7 +313,7 @@ static __always_inline bool rcu_dynticks
> > > {
> > > struct rcu_data *rdp = this_cpu_ptr(&rcu_data);
> > >
> > > - return !(atomic_read(&rdp->dynticks) & RCU_DYNTICK_CTRL_CTR);
> > > + return !(arch_atomic_read(&rdp->dynticks) & RCU_DYNTICK_CTRL_CTR);
>
> The above is actually instrumented by KCSAN, due to arch_atomic_read()
> being a READ_ONCE() and it now understanding volatile.
>
> > Also instrument_atomic_write(&rdp->dynticks, sizeof(rdp->dynticks)) as
Right, this should instead be instrument_read(...).
Though if KCSAN is unconditionally instrumenting volatile, how does
this help? Or does KCSAN's instrumentation of volatile somehow avoid
causing trouble?
> > follows:
> >
> > o rcu_nmi_exit(): After each following instrumentation_begin().
>
> Yes
>
> > o In theory in rcu_irq_exit_preempt(), but as this generates code
> > only in lockdep builds, it might not be worth worrying about.
> >
> > o Ditto for rcu_irq_exit_check_preempt().
> >
> > o Ditto for __rcu_irq_enter_check_tick().
>
> Not these, afaict they're all the above arch_atomic_read(), which is
> instrumented due to volatile in these cases.
>
> > o rcu_nmi_enter(): After each following instrumentation_begin().
>
> Yes
>
> > o __rcu_is_watching() is itself noinstr:
> >
> > o idtentry_enter_cond_rcu(): After each following
> > instrumentation_begin().
> >
> > o rcu_is_watching(): Either before or after the call to
> > rcu_dynticks_curr_cpu_in_eqs().
>
> Something like that yes.
>
> > > }
> > >
> > > /*
> > > @@ -692,6 +692,7 @@ noinstr void rcu_nmi_exit(void)
> > > {
> > > struct rcu_data *rdp = this_cpu_ptr(&rcu_data);
> > >
> > > + instrumentation_begin();
> > > /*
> > > * Check for ->dynticks_nmi_nesting underflow and bad ->dynticks.
> > > * (We are exiting an NMI handler, so RCU better be paying attention
> > > @@ -705,7 +706,6 @@ noinstr void rcu_nmi_exit(void)
> > > * leave it in non-RCU-idle state.
> > > */
> > > if (rdp->dynticks_nmi_nesting != 1) {
> > > - instrumentation_begin();
> > > trace_rcu_dyntick(TPS("--="), rdp->dynticks_nmi_nesting, rdp->dynticks_nmi_nesting - 2,
> > > atomic_read(&rdp->dynticks));
> > > WRITE_ONCE(rdp->dynticks_nmi_nesting, /* No store tearing. */
> > > @@ -714,7 +714,6 @@ noinstr void rcu_nmi_exit(void)
> > > return;
> > > }
> > >
> > > - instrumentation_begin();
> > > /* This NMI interrupted an RCU-idle CPU, restore RCU-idleness. */
> > > trace_rcu_dyntick(TPS("Startirq"), rdp->dynticks_nmi_nesting, 0, atomic_read(&rdp->dynticks));
> > > WRITE_ONCE(rdp->dynticks_nmi_nesting, 0); /* Avoid store tearing. */
> >
> > This one looks to be having no effect on instrumentation of atomics, but
> > rather coalescing a pair of instrumentation_begin() into one.
> >
> > Do I understand correctly?
>
> Almost, it puts the WARN_ON_ONCE()s under instrumentation_begin() too,
> and that makes a differnce, iirc it was the
> rcu_dynticks_curr_cpu_in_eqs() call that stood out. But that could've
> been before I switched it to arch_atomic_read(). In any case, I find
> this form a lot clearer.
Got it, thank you.
Thanx, Paul
next prev parent reply other threads:[~2020-06-04 3:34 UTC|newest]
Thread overview: 51+ messages / expand[flat|nested] mbox.gz Atom feed top
2020-06-03 11:40 [PATCH 0/9] x86/entry fixes Peter Zijlstra
2020-06-03 11:40 ` [PATCH 1/9] x86/entry: Fix irq_exit() Peter Zijlstra
2020-06-03 11:40 ` [PATCH 2/9] rcu: Fixup noinstr warnings Peter Zijlstra
2020-06-03 16:46 ` Paul E. McKenney
2020-06-03 17:13 ` Peter Zijlstra
2020-06-04 3:34 ` Paul E. McKenney [this message]
2020-06-04 6:02 ` Marco Elver
2020-06-04 14:14 ` Paul E. McKenney
2020-06-04 8:05 ` Peter Zijlstra
2020-06-04 14:17 ` Paul E. McKenney
2020-06-15 15:30 ` Peter Zijlstra
2020-06-15 15:52 ` Paul E. McKenney
2020-06-15 16:06 ` Peter Zijlstra
2020-06-15 15:49 ` Peter Zijlstra
2020-06-15 15:55 ` Peter Zijlstra
2020-06-15 16:24 ` Peter Zijlstra
2020-06-15 17:14 ` Paul E. McKenney
2020-06-15 18:33 ` Peter Zijlstra
2020-06-15 18:59 ` Paul E. McKenney
2020-06-15 20:00 ` Paul E. McKenney
2020-06-19 22:15 ` Paul E. McKenney
2020-06-23 20:46 ` Peter Zijlstra
2020-06-23 21:44 ` Paul E. McKenney
2020-06-24 7:52 ` Peter Zijlstra
2020-06-24 13:03 ` Paul E. McKenney
2020-06-03 11:40 ` [PATCH 3/9] x86/entry: __always_inline debugreg for noinstr Peter Zijlstra
2020-06-03 17:50 ` [tip: x86/entry] " tip-bot2 for Peter Zijlstra
2020-06-03 11:40 ` [PATCH 4/9] x86/entry: __always_inline irqflags " Peter Zijlstra
2020-06-03 17:50 ` [tip: x86/entry] " tip-bot2 for Peter Zijlstra
2020-06-03 11:40 ` [PATCH 5/9] x86/entry: __always_inline arch_atomic_* " Peter Zijlstra
2020-06-03 17:50 ` [tip: x86/entry] " tip-bot2 for Peter Zijlstra
2020-06-03 11:40 ` [PATCH 6/9] x86/entry: Re-order #DB handler to avoid *SAN instrumentation Peter Zijlstra
2020-06-03 17:50 ` [tip: x86/entry] " tip-bot2 for Peter Zijlstra
2020-06-03 11:40 ` [PATCH 7/9] lockdep: __always_inline more for noinstr Peter Zijlstra
2020-06-03 17:50 ` [tip: x86/entry] " tip-bot2 for Peter Zijlstra
2020-06-03 11:40 ` [PATCH 8/9] x86/entry: __always_inline CR2 " Peter Zijlstra
2020-06-03 17:50 ` [tip: x86/entry] " tip-bot2 for Peter Zijlstra
2020-06-03 11:40 ` [PATCH 9/9] x86/entry, cpumask: Provide non-instrumented variant of cpu_is_offline() Peter Zijlstra
2020-06-03 17:50 ` [tip: x86/entry] " tip-bot2 for Peter Zijlstra
2020-06-03 12:00 ` [PATCH 0/9] x86/entry fixes Peter Zijlstra
2020-06-03 12:08 ` Peter Zijlstra
2020-06-03 12:08 ` Marco Elver
2020-06-03 12:18 ` Peter Zijlstra
2020-06-03 13:32 ` Marco Elver
2020-06-03 14:47 ` Marco Elver
2020-06-03 16:07 ` Peter Zijlstra
2020-06-03 17:26 ` Marco Elver
2020-06-03 18:16 ` Peter Zijlstra
2020-06-03 19:10 ` Marco Elver
2020-06-04 6:00 ` Marco Elver
2020-06-04 9:52 ` Marco Elver
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20200604033409.GX29598@paulmck-ThinkPad-P72 \
--to=paulmck@kernel.org \
--cc=andreyknvl@google.com \
--cc=dvyukov@google.com \
--cc=elver@google.com \
--cc=glider@google.com \
--cc=kasan-dev@googlegroups.com \
--cc=linux-kernel@vger.kernel.org \
--cc=peterz@infradead.org \
--cc=tglx@linutronix.de \
--cc=will@kernel.org \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).