Re: [PATCH v4 0/5] barrier: Add smp_cond_load_*_timewait()

["Okanovic, Haris" <harisokn@amazon.com>]

On Fri, 2025-08-29 at 01:07 -0700, Ankur Arora wrote:
> CAUTION: This email originated from outside of the organization. Do not click links or open attachments unless you can confirm the sender and know the content is safe.
> 
> 
> 
> Hi,
> 
> This series adds waited variants of the smp_cond_load() primitives:
> smp_cond_load_relaxed_timewait(), and smp_cond_load_acquire_timewait().
> 
> Why?: as the name suggests, the new interfaces are meant for contexts
> where you want to wait on a condition variable for a finite duration.
> This is easy enough to do with a loop around cpu_relax(). However,
> some architectures (ex. arm64) also allow waiting on a cacheline. So,
> these interfaces handle a mixture of spin/wait with a smp_cond_load()
> thrown in.
> 
> There are two known users for these interfaces:
> 
>  - poll_idle() [1]
>  - resilient queued spinlocks [2]
> 
> The interfaces are:
>    smp_cond_load_relaxed_timewait(ptr, cond_expr, time_check_expr)
>    smp_cond_load_acquire_spinwait(ptr, cond_expr, time_check_expr)
> 
> The added parameter, time_check_expr, determines the bail out condition.
> 
> Changelog:
>   v3 [3]:
>     - further interface simplifications (suggested by Catalin Marinas)
> 
>   v2 [4]:
>     - simplified the interface (suggested by Catalin Marinas)
>        - get rid of wait_policy, and a multitude of constants
>        - adds a slack parameter
>       This helped remove a fair amount of duplicated code duplication and in hindsight
>       unnecessary constants.
> 
>   v1 [5]:
>      - add wait_policy (coarse and fine)
>      - derive spin-count etc at runtime instead of using arbitrary
>        constants.
> 
> Haris Okanovic had tested an earlier version of this series with
> poll_idle()/haltpoll patches. [6]
> 
> Any comments appreciated!
> 
> Thanks!
> Ankur
> 
> [1] https://lore.kernel.org/lkml/20241107190818.522639-3-ankur.a.arora@oracle.com/
> [2] Uses the smp_cond_load_acquire_timewait() from v1
>     https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/arch/arm64/include/asm/rqspinlock.h
> [3] https://lore.kernel.org/lkml/20250627044805.945491-1-ankur.a.arora@oracle.com/
> [4] https://lore.kernel.org/lkml/20250502085223.1316925-1-ankur.a.arora@oracle.com/
> [5] https://lore.kernel.org/lkml/20250203214911.898276-1-ankur.a.arora@oracle.com/
> [6] https://lore.kernel.org/lkml/f2f5d09e79539754ced085ed89865787fa668695.camel@amazon.com
> 
> Cc: Arnd Bergmann <arnd@arndb.de>
> Cc: Will Deacon <will@kernel.org>
> Cc: Catalin Marinas <catalin.marinas@arm.com>
> Cc: Peter Zijlstra <peterz@infradead.org>
> Cc: Kumar Kartikeya Dwivedi <memxor@gmail.com>
> Cc: Alexei Starovoitov <ast@kernel.org>
> Cc: linux-arch@vger.kernel.org
> 
> Ankur Arora (5):
>   asm-generic: barrier: Add smp_cond_load_relaxed_timewait()
>   arm64: barrier: Add smp_cond_load_relaxed_timewait()
>   arm64: rqspinlock: Remove private copy of
>     smp_cond_load_acquire_timewait
>   asm-generic: barrier: Add smp_cond_load_acquire_timewait()
>   rqspinlock: use smp_cond_load_acquire_timewait()
> 
>  arch/arm64/include/asm/barrier.h    | 22 ++++++++
>  arch/arm64/include/asm/rqspinlock.h | 84 +----------------------------
>  include/asm-generic/barrier.h       | 57 ++++++++++++++++++++
>  include/asm-generic/rqspinlock.h    |  4 ++
>  kernel/bpf/rqspinlock.c             | 25 ++++-----
>  5 files changed, 93 insertions(+), 99 deletions(-)
> 
> --
> 2.31.1
> 

Tested on AWS Graviton 2, 3, and 4 (ARM64 Neoverse N1, V1, and V2) with
your V10 haltpoll changes, atop 6.17.0-rc3 (commit 07d9df8008).
Still seeing between 1.3x and 2.5x speedups in `perf bench sched pipe`
and `seccomp-notify`; no change in `messaging`.

Reviewed-by: Haris Okanovic <harisokn@amazon.com>
Tested-by: Haris Okanovic <harisokn@amazon.com>

Regards,
Haris Okanovic
AWS Graviton Software