Re: [RFC PATCH] compiler.h: Introduce ptr_eq() to preserve address dependency

[Boqun Feng]

[Cc Gary]

On Fri, Sep 27, 2024 at 04:33:34PM -0400, Mathieu Desnoyers wrote:
> Compiler CSE and SSA GVN optimizations can cause the address dependency
> of addresses returned by rcu_dereference to be lost when comparing those
> pointers with either constants or previously loaded pointers.
> 
> Introduce ptr_eq() to compare two addresses while preserving the address
> dependencies for later use of the address. It should be used when
> comparing an address returned by rcu_dereference().
> 
> This is needed to prevent the compiler CSE and SSA GVN optimizations
> from replacing the registers holding @a or @b based on their
> equality, which does not preserve address dependencies and allows the
> following misordering speculations:
> 
> - If @b is a constant, the compiler can issue the loads which depend
>   on @a before loading @a.
> - If @b is a register populated by a prior load, weakly-ordered
>   CPUs can speculate loads which depend on @a before loading @a.
> 
> The same logic applies with @a and @b swapped.
> 
> The compiler barrier() is ineffective at fixing this issue.
> It does not prevent the compiler CSE from losing the address dependency:
> 
> int fct_2_volatile_barriers(void)
> {
>     int *a, *b;
> 
>     do {
>         a = READ_ONCE(p);
>         asm volatile ("" : : : "memory");
>         b = READ_ONCE(p);
>     } while (a != b);
>     asm volatile ("" : : : "memory");  <----- barrier()
>     return *b;
> }
> 
> With gcc 14.2 (arm64):
> 
> fct_2_volatile_barriers:
>         adrp    x0, .LANCHOR0
>         add     x0, x0, :lo12:.LANCHOR0
> .L2:
>         ldr     x1, [x0]    <------ x1 populated by first load.
>         ldr     x2, [x0]
>         cmp     x1, x2
>         bne     .L2
>         ldr     w0, [x1]    <------ x1 is used for access which should depend on b.
>         ret
> 
> On weakly-ordered architectures, this lets CPU speculation use the
> result from the first load to speculate "ldr w0, [x1]" before
> "ldr x2, [x0]".
> Based on the RCU documentation, the control dependency does not prevent
> the CPU from speculating loads.
> 
> Suggested-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> Suggested-by: Boqun Feng <boqun.feng@xxxxxxxxx>
> Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@xxxxxxxxxxxx>
> Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> Cc: Sebastian Andrzej Siewior <bigeasy@xxxxxxxxxxxxx>
> Cc: "Paul E. McKenney" <paulmck@xxxxxxxxxx>
> Cc: Will Deacon <will@xxxxxxxxxx>
> Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
> Cc: Boqun Feng <boqun.feng@xxxxxxxxx>
> Cc: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx>
> Cc: John Stultz <jstultz@xxxxxxxxxx>
> Cc: Neeraj Upadhyay <Neeraj.Upadhyay@xxxxxxx>
> Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
> Cc: Boqun Feng <boqun.feng@xxxxxxxxx>
> Cc: Frederic Weisbecker <frederic@xxxxxxxxxx>
> Cc: Joel Fernandes <joel@xxxxxxxxxxxxxxxxx>
> Cc: Josh Triplett <josh@xxxxxxxxxxxxxxxx>
> Cc: Uladzislau Rezki <urezki@xxxxxxxxx>
> Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
> Cc: Lai Jiangshan <jiangshanlai@xxxxxxxxx>
> Cc: Zqiang <qiang.zhang1211@xxxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: Waiman Long <longman@xxxxxxxxxx>
> Cc: Mark Rutland <mark.rutland@xxxxxxx>
> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> Cc: Vlastimil Babka <vbabka@xxxxxxx>
> Cc: maged.michael@xxxxxxxxx
> Cc: Mateusz Guzik <mjguzik@xxxxxxxxx>
> Cc: rcu@xxxxxxxxxxxxxxx
> Cc: linux-mm@xxxxxxxxx
> Cc: lkmm@xxxxxxxxxxxxxxx
> ---
>  include/linux/compiler.h | 62 ++++++++++++++++++++++++++++++++++++++++
>  1 file changed, 62 insertions(+)
> 
> diff --git a/include/linux/compiler.h b/include/linux/compiler.h
> index 2df665fa2964..f26705c267e8 100644
> --- a/include/linux/compiler.h
> +++ b/include/linux/compiler.h
> @@ -186,6 +186,68 @@ void ftrace_likely_update(struct ftrace_likely_data *f, int val,
>  	__asm__ ("" : "=r" (var) : "0" (var))
>  #endif
>  
> +/*
> + * Compare two addresses while preserving the address dependencies for
> + * later use of the address. It should be used when comparing an address
> + * returned by rcu_dereference().
> + *
> + * This is needed to prevent the compiler CSE and SSA GVN optimizations
> + * from replacing the registers holding @a or @b based on their
> + * equality, which does not preserve address dependencies and allows the
> + * following misordering speculations:
> + *
> + * - If @b is a constant, the compiler can issue the loads which depend
> + *   on @a before loading @a.
> + * - If @b is a register populated by a prior load, weakly-ordered
> + *   CPUs can speculate loads which depend on @a before loading @a.
> + *
> + * The same logic applies with @a and @b swapped.
> + *
> + * Return value: true if pointers are equal, false otherwise.
> + *
> + * The compiler barrier() is ineffective at fixing this issue. It does
> + * not prevent the compiler CSE from losing the address dependency:
> + *
> + * int fct_2_volatile_barriers(void)
> + * {
> + *     int *a, *b;
> + *
> + *     do {
> + *         a = READ_ONCE(p);
> + *         asm volatile ("" : : : "memory");
> + *         b = READ_ONCE(p);
> + *     } while (a != b);
> + *     asm volatile ("" : : : "memory");  <-- barrier()
> + *     return *b;
> + * }
> + *
> + * With gcc 14.2 (arm64):
> + *
> + * fct_2_volatile_barriers:
> + *         adrp    x0, .LANCHOR0
> + *         add     x0, x0, :lo12:.LANCHOR0
> + * .L2:
> + *         ldr     x1, [x0]  <-- x1 populated by first load.
> + *         ldr     x2, [x0]
> + *         cmp     x1, x2
> + *         bne     .L2
> + *         ldr     w0, [x1]  <-- x1 is used for access which should depend on b.
> + *         ret
> + *
> + * On weakly-ordered architectures, this lets CPU speculation use the
> + * result from the first load to speculate "ldr w0, [x1]" before
> + * "ldr x2, [x0]".
> + * Based on the RCU documentation, the control dependency does not
> + * prevent the CPU from speculating loads.
> + */
> +static __always_inline
> +int ptr_eq(const volatile void *a, const volatile void *b)
> +{
> +	OPTIMIZER_HIDE_VAR(a);
> +	OPTIMIZER_HIDE_VAR(b);
> +	return a == b;
> +}
> +

This is better than what I proposed, thank you!

Reviewed-by: Boqun Feng <boqun.feng@xxxxxxxxx>

Could you also make some documentation changes at the "Be very careful
about comparing pointers obtained from..." paragraph in
Documentation/RCU/rcu_dereference.rst? Since 'ptr_eq' is a good tool in
those cases mentioned there. Thanks.

Regards,
Boqun

>  #define __UNIQUE_ID(prefix) __PASTE(__PASTE(__UNIQUE_ID_, prefix), __COUNTER__)
>  
>  /**
> -- 
> 2.39.2
>