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]
Message-ID: <ZSkD4RLAhJaW3VyB@gmail.com>
Date:   Fri, 13 Oct 2023 10:46:25 +0200
From:   Ingo Molnar <mingo@...nel.org>
To:     Linus Torvalds <torvalds@...ux-foundation.org>
Cc:     Oleg Nesterov <oleg@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Alexey Gladkov <legion@...nel.org>,
        Ingo Molnar <mingo@...hat.com>, Will Deacon <will@...nel.org>,
        Waiman Long <longman@...hat.com>,
        Boqun Feng <boqun.feng@...il.com>,
        linux-kernel@...r.kernel.org, Thomas Gleixner <tglx@...utronix.de>
Subject: [PATCH] locking/seqlock: Propagate 'const' pointers within read-only
 methods, remove forced type casts


* Linus Torvalds <torvalds@...ux-foundation.org> wrote:

> On Thu, 12 Oct 2023 at 11:21, Ingo Molnar <mingo@...nel.org> wrote:
> >
> > Okay, so dropping 'const' makes sense in terms of staying bug-compatible
> > with the  previous API and not build-breaking the world - but could we
> > perhaps follow this up with fixups of the type misuse and then a removal
> > of the forced type casts from these APIs?
> 
> No. The use of 'const' here is *not* a bug.
> 
> The thing is, 'const' doesn't mean what you seem to think it means. A
> 'const' pointer in C in no way means that the target is constant - it
> means that *THIS* use of the pointer will not write to the target!

Yeah, that is absolutely what I too think 'const' means - and sorry, I 
didn't expand: I meant something like the patch below, which clearly 
separates the 'const' from the non-const pointer uses within 
<linux/seqlock.h> and removes the two forced type casts I was unhappy 
about.

The 'bug' was that the __seqprop_*ptr() wrapper was used with both const 
and non-const pointers, and we forced a type and lost a tiny bit of 
potential const propagation. The code was fine and I should not have called 
it a 'bug', but I consider the dropping of 'const' a bad pattern, and I 
sometimes exaggerate problems to trick^W convince developers to continue 
working along a given path...

In hindsight my "break the world" expectation was overblown too: our const 
propagation through these methods was almost complete already, and the 
fixes stayed within <linux/seqlock.h>.

This patch could probably be split into two patches. Lightly tested only.

Does this work for you?

Thanks,

	Ingo

===================>
From: Ingo Molnar <mingo@...nel.org>
Date: Fri, 13 Oct 2023 10:15:46 +0200
Subject: [PATCH] locking/seqlock: Propagate 'const' pointers within read-only methods, remove forced type casts

Currently __seqprop_ptr() is an inline function that must chose to either
use 'const' or non-const seqcount related pointers - but this results in
the undesirable loss of 'const' propagation, via a forced type cast.

The easiest solution would be to turn the pointer wrappers into macros that 
pass through whatever type is passed to them - but the clever maze of 
seqlock API instantiation macros relies on the GCC CPP '##' macro 
extension, which isn't recursive, so inline functions must be used here.

So create two wrapper variants instead: 'ptr' and 'const_ptr', and pick the
right one for the codepaths that are const: read_seqcount_begin() and
read_seqcount_retry().

This cleans up type handling and allows the removal of all type forcing.

No change in functionality.

Signed-off-by: Ingo Molnar <mingo@...nel.org>
Cc: Oleg Nesterov <oleg@...hat.com>
Cc: Linus Torvalds <torvalds@...ux-foundation.org>
Cc: Peter Zijlstra <peterz@...radead.org>
Cc: Waiman Long <longman@...hat.com>
Cc: Will Deacon <will.deacon@....com>
Cc: Thomas Gleixner <tglx@...utronix.de>
Cc: Paul E. McKenney <paulmck@...nel.org>
---
 include/linux/seqlock.h | 26 +++++++++++++++++++-------
 1 file changed, 19 insertions(+), 7 deletions(-)

diff --git a/include/linux/seqlock.h b/include/linux/seqlock.h
index 4b8dcd3a0d93..80f21d2ca2aa 100644
--- a/include/linux/seqlock.h
+++ b/include/linux/seqlock.h
@@ -200,9 +200,15 @@ typedef struct seqcount_##lockname {					\
 } seqcount_##lockname##_t;						\
 									\
 static __always_inline seqcount_t *					\
-__seqprop_##lockname##_ptr(const seqcount_##lockname##_t *s)		\
+__seqprop_##lockname##_ptr(seqcount_##lockname##_t *s)			\
 {									\
-	return (void *)&s->seqcount; /* drop const */			\
+	return &s->seqcount;						\
+}									\
+									\
+static __always_inline const seqcount_t *				\
+__seqprop_##lockname##_const_ptr(const seqcount_##lockname##_t *s)	\
+{									\
+	return &s->seqcount;						\
 }									\
 									\
 static __always_inline unsigned						\
@@ -247,9 +253,14 @@ __seqprop_##lockname##_assert(const seqcount_##lockname##_t *s)		\
  * __seqprop() for seqcount_t
  */
 
-static inline seqcount_t *__seqprop_ptr(const seqcount_t *s)
+static inline seqcount_t *__seqprop_ptr(seqcount_t *s)
 {
-	return (void *)s; /* drop const */
+	return s;
+}
+
+static inline const seqcount_t *__seqprop_const_ptr(const seqcount_t *s)
+{
+	return s;
 }
 
 static inline unsigned __seqprop_sequence(const seqcount_t *s)
@@ -302,6 +313,7 @@ SEQCOUNT_LOCKNAME(mutex,        struct mutex,    true,     mutex)
 	__seqprop_case((s),	mutex,		prop))
 
 #define seqprop_ptr(s)			__seqprop(s, ptr)(s)
+#define seqprop_const_ptr(s)		__seqprop(s, const_ptr)(s)
 #define seqprop_sequence(s)		__seqprop(s, sequence)(s)
 #define seqprop_preemptible(s)		__seqprop(s, preemptible)(s)
 #define seqprop_assert(s)		__seqprop(s, assert)(s)
@@ -353,7 +365,7 @@ SEQCOUNT_LOCKNAME(mutex,        struct mutex,    true,     mutex)
  */
 #define read_seqcount_begin(s)						\
 ({									\
-	seqcount_lockdep_reader_access(seqprop_ptr(s));			\
+	seqcount_lockdep_reader_access(seqprop_const_ptr(s));		\
 	raw_read_seqcount_begin(s);					\
 })
 
@@ -419,7 +431,7 @@ SEQCOUNT_LOCKNAME(mutex,        struct mutex,    true,     mutex)
  * Return: true if a read section retry is required, else false
  */
 #define __read_seqcount_retry(s, start)					\
-	do___read_seqcount_retry(seqprop_ptr(s), start)
+	do___read_seqcount_retry(seqprop_const_ptr(s), start)
 
 static inline int do___read_seqcount_retry(const seqcount_t *s, unsigned start)
 {
@@ -439,7 +451,7 @@ static inline int do___read_seqcount_retry(const seqcount_t *s, unsigned start)
  * Return: true if a read section retry is required, else false
  */
 #define read_seqcount_retry(s, start)					\
-	do_read_seqcount_retry(seqprop_ptr(s), start)
+	do_read_seqcount_retry(seqprop_const_ptr(s), start)
 
 static inline int do_read_seqcount_retry(const seqcount_t *s, unsigned start)
 {

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ