mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Bart Van Assche <bvanassche@acm.org>
To: Peter Zijlstra <peterz@infradead.org>
Cc: "Ingo Molnar" <mingo@redhat.com>, "Will Deacon" <will@kernel.org>,
	"Boqun Feng" <boqun@kernel.org>,
	"Waiman Long" <longman@redhat.com>,
	linux-kernel@vger.kernel.org, "Marco Elver" <elver@google.com>,
	"Paul E . McKenney" <paulmck@kernel.org>,
	"Christoph Hellwig" <hch@lst.de>,
	"Bart Van Assche" <bvanassche@acm.org>,
	"Lai Jiangshan" <jiangshanlai@gmail.com>,
	"Josh Triplett" <josh@joshtriplett.org>,
	"Onur Özkan" <work@onurozkan.dev>,
	"Nathan Chancellor" <nathan@kernel.org>
Subject: [PATCH v4 3/3] srcu: Make the srcu_dereference_check() reader lock check conditional
Date: Wed, 18 Nov 2026 15:09:59 -0800	[thread overview]
Message-ID: <459250d4f2a6aa38ebab3fb51f6d4a63815bb4a1.1795043161.git.bvanassche@acm.org> (raw)
In-Reply-To: <cover.1795043161.git.bvanassche@acm.org>

Users of srcu_dereference_check() must either hold an SRCU reader lock
or satisfy the lockdep condition 'c', e.g. by holding the update-side
lock. However, srcu_dereference_check() calls
__srcu_read_lock_must_hold(), which tells the Clang context analyzer
that the SRCU reader lock must be held. That requirement is too strict
for update-side callers. Hence, only call __srcu_read_lock_must_hold()
from srcu_dereference_check() if 'c' is the compile-time constant 0,
i.e. if there is no update-side protection. Keep the fake RCU acquire
and release around __rcu_dereference_check() and also keep the runtime
lockdep check in srcu_dereference_check().

Use const_true() instead of __builtin_constant_p() to test whether 'c'
is the constant 0. const_true() is based on __is_constexpr(), a
sizeof() trick that the compiler front end always evaluates, so the
Clang context analysis sees an integer constant expression. If 'c' is
a plain variable, function argument or structure member, Clang may
leave __builtin_constant_p(c) for the optimizer to decide. The context
analysis then considers the __srcu_read_lock_must_hold() call
reachable and reports a false positive. Additionally, const_true()
never evaluates a non-constant 'c' at runtime, so the runtime behavior
does not change.

srcu_dereference() passes 0 as 'c' and hence gets the compile-time
check automatically. srcu_dereference_notrace() does not accept a
lockdep condition and hence may only be used by SRCU readers. Since it
passes 1 as 'c', call __srcu_read_lock_must_hold() explicitly from
srcu_dereference_notrace() such that the Clang context analysis keeps
verifying at compile time that its callers hold the SRCU reader lock.

Cc: Marco Elver <elver@google.com>
Suggested-by: Paul E. McKenney <paulmck@kernel.org>
Fixes: f0b7ce22d718 ("srcu: Support Clang's context analysis")
Signed-off-by: Bart Van Assche <bvanassche@acm.org>
---
 include/linux/srcu.h | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/include/linux/srcu.h b/include/linux/srcu.h
index d4dcaeb83e2b..380b5964fcb4 100644
--- a/include/linux/srcu.h
+++ b/include/linux/srcu.h
@@ -280,6 +280,10 @@ static inline int srcu_read_lock_held(const struct srcu_struct *ssp)
  * with __guarded_by(ssp) as it would complicate annotations for writers, we
  * choose the following strategy: srcu_dereference_check() calls this helper
  * that checks that the passed ssp is held, and then fake-acquires 'RCU'.
+ * srcu_dereference_check() only calls this helper if @c is the compile-time
+ * constant 0 because otherwise its callers may satisfy the condition @c
+ * instead of holding ssp, e.g. on the update side. srcu_dereference_notrace()
+ * passes 1 as @c and hence calls this helper explicitly.
  */
 static inline void __srcu_read_lock_must_hold(const struct srcu_struct *ssp) __must_hold_shared(ssp) { }
 
@@ -297,7 +301,8 @@ static inline void __srcu_read_lock_must_hold(const struct srcu_struct *ssp) __m
  */
 #define srcu_dereference_check(p, ssp, c)					\
 ({										\
-	__srcu_read_lock_must_hold(ssp);					\
+	if (const_true(!(c)))							\
+		__srcu_read_lock_must_hold(ssp);				\
 	__acquire_shared_ctx_lock(RCU);					\
 	__auto_type __v = __rcu_dereference_check((p), __UNIQUE_ID(rcu),	\
 				(c) || srcu_read_lock_held(ssp), __rcu);	\
@@ -323,7 +328,11 @@ static inline void __srcu_read_lock_must_hold(const struct srcu_struct *ssp) __m
  * @ssp: pointer to the srcu_struct, which is used to check that we
  *	really are in an SRCU read-side critical section.
  */
-#define srcu_dereference_notrace(p, ssp) srcu_dereference_check((p), (ssp), 1)
+#define srcu_dereference_notrace(p, ssp)				\
+({									\
+	__srcu_read_lock_must_hold(ssp);				\
+	srcu_dereference_check((p), (ssp), 1);				\
+})
 
 /**
  * srcu_read_lock - register a new reader for an SRCU-protected structure.

  parent reply	other threads:[~2026-11-18 23:10 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-18 23:09 [PATCH v4 0/3] Fix three Clang context analysis issues Bart Van Assche
2026-11-18 23:09 ` [PATCH v4 1/3] bit_spinlock: Fix the __bitlock() definition Bart Van Assche
2026-11-18 23:09 ` [PATCH v4 2/3] cleanup: Disable context analysis for __free() helpers Bart Van Assche
2026-11-18 23:09 ` Bart Van Assche [this message]
2026-11-18 23:47   ` [PATCH v4 3/3] srcu: Make the srcu_dereference_check() reader lock check conditional 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=459250d4f2a6aa38ebab3fb51f6d4a63815bb4a1.1795043161.git.bvanassche@acm.org \
    --to=bvanassche@acm.org \
    --cc=boqun@kernel.org \
    --cc=elver@google.com \
    --cc=hch@lst.de \
    --cc=jiangshanlai@gmail.com \
    --cc=josh@joshtriplett.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=longman@redhat.com \
    --cc=mingo@redhat.com \
    --cc=nathan@kernel.org \
    --cc=paulmck@kernel.org \
    --cc=peterz@infradead.org \
    --cc=will@kernel.org \
    --cc=work@onurozkan.dev \
    /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

all inboxes | Powered by JetHome®