From: "Paul E. McKenney" <paulmck@kernel.org>
To: rcu@vger.kernel.org
Cc: linux-kernel@vger.kernel.org, kernel-team@fb.com,
rostedt@goodmis.org, "Paul E. McKenney" <paulmck@kernel.org>,
Brian Foster <bfoster@redhat.com>,
Dave Chinner <david@fromorbit.com>,
Al Viro <viro@zeniv.linux.org.uk>, Ian Kent <raven@themaw.net>
Subject: [PATCH rcu 02/12] rcu: Provide a get_completed_synchronize_rcu() function
Date: Mon, 20 Jun 2022 15:51:18 -0700 [thread overview]
Message-ID: <20220620225128.3842050-2-paulmck@kernel.org> (raw)
In-Reply-To: <20220620224943.GA3841634@paulmck-ThinkPad-P17-Gen-1>
It is currently up to the caller to handle stale return values from
get_state_synchronize_rcu(). If poll_state_synchronize_rcu() returned
true once, a grace period has elapsed, regardless of the fact that counter
wrap might cause some future poll_state_synchronize_rcu() invocation to
return false. For example, the caller might store a separate flag that
indicates whether some previous call to poll_state_synchronize_rcu()
determined that the relevant grace period had already ended.
This approach works, but it requires extra storage and is easy to get
wrong. This commit therefore introduces a get_completed_synchronize_rcu()
that returns a cookie that causes poll_state_synchronize_rcu() to always
return true. This already-completed cookie can be stored in place of the
cookie that previously caused poll_state_synchronize_rcu() to return true.
It can also be used to flag a given structure as not having been exposed
to readers, and thus not requiring a grace period to elapse.
This commit is in preparation for polled expedited grace periods.
Link: https://lore.kernel.org/all/20220121142454.1994916-1-bfoster@redhat.com/
Link: https://docs.google.com/document/d/1RNKWW9jQyfjxw2E8dsXVTdvZYh0HnYeSHDKog9jhdN8/edit?usp=sharing
Cc: Brian Foster <bfoster@redhat.com>
Cc: Dave Chinner <david@fromorbit.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Ian Kent <raven@themaw.net>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
---
include/linux/rcupdate.h | 1 +
kernel/rcu/rcu.h | 3 +++
kernel/rcu/tiny.c | 4 ++--
kernel/rcu/tree.c | 3 ++-
kernel/rcu/update.c | 13 +++++++++++++
5 files changed, 21 insertions(+), 3 deletions(-)
diff --git a/include/linux/rcupdate.h b/include/linux/rcupdate.h
index 1a32036c918cd..7f12daa4708b1 100644
--- a/include/linux/rcupdate.h
+++ b/include/linux/rcupdate.h
@@ -41,6 +41,7 @@ void call_rcu(struct rcu_head *head, rcu_callback_t func);
void rcu_barrier_tasks(void);
void rcu_barrier_tasks_rude(void);
void synchronize_rcu(void);
+unsigned long get_completed_synchronize_rcu(void);
#ifdef CONFIG_PREEMPT_RCU
diff --git a/kernel/rcu/rcu.h b/kernel/rcu/rcu.h
index 0adb55941aeb3..32291f4eefde9 100644
--- a/kernel/rcu/rcu.h
+++ b/kernel/rcu/rcu.h
@@ -23,6 +23,9 @@
#define RCU_SEQ_CTR_SHIFT 2
#define RCU_SEQ_STATE_MASK ((1 << RCU_SEQ_CTR_SHIFT) - 1)
+/* Low-order bit definition for polled grace-period APIs. */
+#define RCU_GET_STATE_COMPLETED 0x1
+
extern int sysctl_sched_rt_runtime;
/*
diff --git a/kernel/rcu/tiny.c b/kernel/rcu/tiny.c
index 340b3f8b090d4..dbee6bea67269 100644
--- a/kernel/rcu/tiny.c
+++ b/kernel/rcu/tiny.c
@@ -58,7 +58,7 @@ void rcu_qs(void)
rcu_ctrlblk.donetail = rcu_ctrlblk.curtail;
raise_softirq_irqoff(RCU_SOFTIRQ);
}
- WRITE_ONCE(rcu_ctrlblk.gp_seq, rcu_ctrlblk.gp_seq + 1);
+ WRITE_ONCE(rcu_ctrlblk.gp_seq, rcu_ctrlblk.gp_seq + 2);
local_irq_restore(flags);
}
@@ -213,7 +213,7 @@ EXPORT_SYMBOL_GPL(start_poll_synchronize_rcu);
*/
bool poll_state_synchronize_rcu(unsigned long oldstate)
{
- return READ_ONCE(rcu_ctrlblk.gp_seq) != oldstate;
+ return oldstate == RCU_GET_STATE_COMPLETED || READ_ONCE(rcu_ctrlblk.gp_seq) != oldstate;
}
EXPORT_SYMBOL_GPL(poll_state_synchronize_rcu);
diff --git a/kernel/rcu/tree.c b/kernel/rcu/tree.c
index ec28e259774e7..46cfceea87847 100644
--- a/kernel/rcu/tree.c
+++ b/kernel/rcu/tree.c
@@ -3924,7 +3924,8 @@ EXPORT_SYMBOL_GPL(start_poll_synchronize_rcu);
*/
bool poll_state_synchronize_rcu(unsigned long oldstate)
{
- if (rcu_seq_done_exact(&rcu_state.gp_seq, oldstate)) {
+ if (oldstate == RCU_GET_STATE_COMPLETED ||
+ rcu_seq_done_exact(&rcu_state.gp_seq, oldstate)) {
smp_mb(); /* Ensure GP ends before subsequent accesses. */
return true;
}
diff --git a/kernel/rcu/update.c b/kernel/rcu/update.c
index fc7fef5756064..2e93acad1e31f 100644
--- a/kernel/rcu/update.c
+++ b/kernel/rcu/update.c
@@ -516,6 +516,19 @@ int rcu_cpu_stall_suppress_at_boot __read_mostly; // !0 = suppress boot stalls.
EXPORT_SYMBOL_GPL(rcu_cpu_stall_suppress_at_boot);
module_param(rcu_cpu_stall_suppress_at_boot, int, 0444);
+/**
+ * get_completed_synchronize_rcu - Return a pre-completed polled state cookie
+ *
+ * Returns a value that will always be treated by functions like
+ * poll_state_synchronize_rcu() as a cookie whose grace period has already
+ * completed.
+ */
+unsigned long get_completed_synchronize_rcu(void)
+{
+ return RCU_GET_STATE_COMPLETED;
+}
+EXPORT_SYMBOL_GPL(get_completed_synchronize_rcu);
+
#ifdef CONFIG_PROVE_RCU
/*
--
2.31.1.189.g2e36527f23
next prev parent reply other threads:[~2022-06-20 22:51 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-06-20 22:49 [PATCH rcu 0/12] Polled grace-period updates for v5.20 Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 01/12] rcu: Make normal polling GP be more precise about sequence numbers Paul E. McKenney
2022-06-20 22:51 ` Paul E. McKenney [this message]
2022-06-20 22:51 ` [PATCH rcu 03/12] rcutorture: Validate get_completed_synchronize_rcu() Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 04/12] rcu: Switch polled grace-period APIs to ->gp_seq_polled Paul E. McKenney
2022-07-21 0:53 ` Boqun Feng
2022-07-21 1:04 ` Paul E. McKenney
2022-07-21 1:51 ` Boqun Feng
2022-07-21 4:47 ` Paul E. McKenney
2022-07-21 5:49 ` Boqun Feng
2022-07-22 1:03 ` Paul E. McKenney
2022-07-21 1:56 ` Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 05/12] rcu: Make polled grace-period API account for expedited grace periods Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 06/12] rcu: Make Tiny RCU grace periods visible to polled APIs Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 07/12] rcutorture: Verify that polled GP API sees synchronous grace periods Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 08/12] rcu: Add polled expedited grace-period primitives Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 09/12] rcutorture: Test " Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 10/12] rcu: Put panic_on_rcu_stall() after expedited RCU CPU stall warnings Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 11/12] rcu: Diagnose extended sync_rcu_do_polled_gp() loops Paul E. McKenney
2022-06-20 22:51 ` [PATCH rcu 12/12] rcu: Add irqs-disabled indicator to expedited RCU CPU stall warnings Paul E. McKenney
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=20220620225128.3842050-2-paulmck@kernel.org \
--to=paulmck@kernel.org \
--cc=bfoster@redhat.com \
--cc=david@fromorbit.com \
--cc=kernel-team@fb.com \
--cc=linux-kernel@vger.kernel.org \
--cc=raven@themaw.net \
--cc=rcu@vger.kernel.org \
--cc=rostedt@goodmis.org \
--cc=viro@zeniv.linux.org.uk \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.