[v2,1/2] iopoll: Call cpu_relax() in busy loops

It is considered good practice to call cpu_relax() in busy loops, see
Documentation/process/volatile-considered-harmful.rst.  This can not
only lower CPU power consumption or yield to a hyperthreaded twin
processor, but also allows an architecture to mitigate hardware issues
(e.g. ARM Erratum 754327 for Cortex-A9 prior to r2p0) in the
architecture-specific cpu_relax() implementation.

In addition, cpu_relax() is also a compiler barrier.  It is not
immediately obvious that the @op argument "function" will result in an
actual function call (e.g. in case of inlining).

Where a function call is a C sequence point, this is lost on inlining.
Therefore, with agressive enough optimization it might be possible for
the compiler to hoist the:

        (val) = op(args);

"load" out of the loop because it doesn't see the value changing. The
addition of cpu_relax() would inhibit this.

As the iopoll helpers lack calls to cpu_relax(), people are sometimes
reluctant to use them, and may fall back to open-coded polling loops
(including cpu_relax() calls) instead.

Fix this by adding calls to cpu_relax() to the iopoll helpers:
  - For the non-atomic case, it is sufficient to call cpu_relax() in
    case of a zero sleep-between-reads value, as a call to
    usleep_range() is a safe barrier otherwise.  However, it doesn't
    hurt to add the call regardless, for simplicity, and for similarity
    with the atomic case below.
  - For the atomic case, cpu_relax() must be called regardless of the
    sleep-between-reads value, as there is no guarantee all
    architecture-specific implementations of udelay() handle this.

Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Acked-by: Arnd Bergmann <arnd@arndb.de>
---
v2:
  - Add Acked-by,
  - Add compiler barrier and inlining explanation (thanks, Peter!).
---
 include/linux/iopoll.h | 2 ++
 1 file changed, 2 insertions(+)

* Geert Uytterhoeven <geert+renesas@glider.be> [230510 13:23]:
> It is considered good practice to call cpu_relax() in busy loops, see
> Documentation/process/volatile-considered-harmful.rst.  This can not
> only lower CPU power consumption or yield to a hyperthreaded twin
> processor, but also allows an architecture to mitigate hardware issues
> (e.g. ARM Erratum 754327 for Cortex-A9 prior to r2p0) in the
> architecture-specific cpu_relax() implementation.
> 
> In addition, cpu_relax() is also a compiler barrier.  It is not
> immediately obvious that the @op argument "function" will result in an
> actual function call (e.g. in case of inlining).
> 
> Where a function call is a C sequence point, this is lost on inlining.
> Therefore, with agressive enough optimization it might be possible for
> the compiler to hoist the:
> 
>         (val) = op(args);
> 
> "load" out of the loop because it doesn't see the value changing. The
> addition of cpu_relax() would inhibit this.
> 
> As the iopoll helpers lack calls to cpu_relax(), people are sometimes
> reluctant to use them, and may fall back to open-coded polling loops
> (including cpu_relax() calls) instead.
> 
> Fix this by adding calls to cpu_relax() to the iopoll helpers:
>   - For the non-atomic case, it is sufficient to call cpu_relax() in
>     case of a zero sleep-between-reads value, as a call to
>     usleep_range() is a safe barrier otherwise.  However, it doesn't
>     hurt to add the call regardless, for simplicity, and for similarity
>     with the atomic case below.
>   - For the atomic case, cpu_relax() must be called regardless of the
>     sleep-between-reads value, as there is no guarantee all
>     architecture-specific implementations of udelay() handle this.

Reviewed-by: Tony Lindgren <tony@atomide.com>

On Wed, 10 May 2023 at 15:23, Geert Uytterhoeven
<geert+renesas@glider.be> wrote:
>
> It is considered good practice to call cpu_relax() in busy loops, see
> Documentation/process/volatile-considered-harmful.rst.  This can not
> only lower CPU power consumption or yield to a hyperthreaded twin
> processor, but also allows an architecture to mitigate hardware issues
> (e.g. ARM Erratum 754327 for Cortex-A9 prior to r2p0) in the
> architecture-specific cpu_relax() implementation.
>
> In addition, cpu_relax() is also a compiler barrier.  It is not
> immediately obvious that the @op argument "function" will result in an
> actual function call (e.g. in case of inlining).
>
> Where a function call is a C sequence point, this is lost on inlining.
> Therefore, with agressive enough optimization it might be possible for
> the compiler to hoist the:
>
>         (val) = op(args);
>
> "load" out of the loop because it doesn't see the value changing. The
> addition of cpu_relax() would inhibit this.
>
> As the iopoll helpers lack calls to cpu_relax(), people are sometimes
> reluctant to use them, and may fall back to open-coded polling loops
> (including cpu_relax() calls) instead.
>
> Fix this by adding calls to cpu_relax() to the iopoll helpers:
>   - For the non-atomic case, it is sufficient to call cpu_relax() in
>     case of a zero sleep-between-reads value, as a call to
>     usleep_range() is a safe barrier otherwise.  However, it doesn't
>     hurt to add the call regardless, for simplicity, and for similarity
>     with the atomic case below.
>   - For the atomic case, cpu_relax() must be called regardless of the
>     sleep-between-reads value, as there is no guarantee all
>     architecture-specific implementations of udelay() handle this.
>
> Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
> Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
> Acked-by: Arnd Bergmann <arnd@arndb.de>

Makes sense to me! Feel free to add:

Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>

Kind regards
Uffe

> ---
> v2:
>   - Add Acked-by,
>   - Add compiler barrier and inlining explanation (thanks, Peter!).
> ---
>  include/linux/iopoll.h | 2 ++
>  1 file changed, 2 insertions(+)
>
> diff --git a/include/linux/iopoll.h b/include/linux/iopoll.h
> index 2c8860e406bd8cae..0417360a6db9b0d6 100644
> --- a/include/linux/iopoll.h
> +++ b/include/linux/iopoll.h
> @@ -53,6 +53,7 @@
>                 } \
>                 if (__sleep_us) \
>                         usleep_range((__sleep_us >> 2) + 1, __sleep_us); \
> +               cpu_relax(); \
>         } \
>         (cond) ? 0 : -ETIMEDOUT; \
>  })
> @@ -95,6 +96,7 @@
>                 } \
>                 if (__delay_us) \
>                         udelay(__delay_us); \
> +               cpu_relax(); \
>         } \
>         (cond) ? 0 : -ETIMEDOUT; \
>  })
> --
> 2.34.1
>