linux-kernel - Re: [PATCH v5 00/18] Rework READ

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: <20200513111447.GE3001@hirez.programming.kicks-ass.net>
Date:   Wed, 13 May 2020 13:14:47 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Marco Elver <elver@...gle.com>
Cc:     Will Deacon <will@...nel.org>, LKML <linux-kernel@...r.kernel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        "Paul E. McKenney" <paulmck@...nel.org>,
        Ingo Molnar <mingo@...nel.org>, dvyukov@...gle.com
Subject: Re: [PATCH v5 00/18] Rework READ_ONCE() to improve codegen

On Wed, May 13, 2020 at 01:10:57PM +0200, Peter Zijlstra wrote:

> So then I end up with something like the below, and I've validated that
> does not generate instrumentation... HOWEVER, I now need ~10g of memory
> and many seconds to compile each file in arch/x86/kernel/.

> diff --git a/include/linux/compiler.h b/include/linux/compiler.h
> index 3bb962959d8b..48f85d1d2db6 100644
> --- a/include/linux/compiler.h
> +++ b/include/linux/compiler.h
> @@ -241,7 +241,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
>   * atomicity or dependency ordering guarantees. Note that this may result
>   * in tears!
>   */
> -#define __READ_ONCE(x)	(*(const volatile __unqual_scalar_typeof(x) *)&(x))
> +#define __READ_ONCE(x)	data_race((*(const volatile __unqual_scalar_typeof(x) *)&(x)))
>  
>  #define __READ_ONCE_SCALAR(x)						\
>  ({									\
> @@ -260,7 +260,7 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
>  
>  #define __WRITE_ONCE(x, val)						\
>  do {									\
> -	*(volatile typeof(x) *)&(x) = (val);				\
> +	data_race(*(volatile typeof(x) *)&(x) = (val));			\
>  } while (0)
>  
>  #define __WRITE_ONCE_SCALAR(x, val)					\

The above is responsible for that, the below variant is _MUCH_ better
again. It really doesn't like nested data_race(), as in _REALLY_ doesn't
like.

---

diff --git a/include/linux/compiler.h b/include/linux/compiler.h
index 3bb962959d8b..2ea532b19e75 100644
--- a/include/linux/compiler.h
+++ b/include/linux/compiler.h
@@ -241,12 +241,12 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
  * atomicity or dependency ordering guarantees. Note that this may result
  * in tears!
  */
-#define __READ_ONCE(x)	(*(const volatile __unqual_scalar_typeof(x) *)&(x))
+#define __READ_ONCE(x)	data_race((*(const volatile __unqual_scalar_typeof(x) *)&(x)))
 
 #define __READ_ONCE_SCALAR(x)						\
 ({									\
 	typeof(x) *__xp = &(x);						\
-	__unqual_scalar_typeof(x) __x = data_race(__READ_ONCE(*__xp));	\
+	__unqual_scalar_typeof(x) __x = __READ_ONCE(*__xp);		\
 	kcsan_check_atomic_read(__xp, sizeof(*__xp));			\
 	smp_read_barrier_depends();					\
 	(typeof(x))__x;							\
@@ -260,14 +260,14 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
 
 #define __WRITE_ONCE(x, val)						\
 do {									\
-	*(volatile typeof(x) *)&(x) = (val);				\
+	data_race(*(volatile typeof(x) *)&(x) = (val));			\
 } while (0)
 
 #define __WRITE_ONCE_SCALAR(x, val)					\
 do {									\
 	typeof(x) *__xp = &(x);						\
 	kcsan_check_atomic_write(__xp, sizeof(*__xp));			\
-	data_race(({ __WRITE_ONCE(*__xp, val); 0; }));			\
+	__WRITE_ONCE(*__xp, val);					\
 } while (0)
 
 #define WRITE_ONCE(x, val)						\