Re: [PATCH 1/1] prctl: allow overriding system THP policy to always per process

[Zi Yan <ziy@nvidia.com>]

On 7 May 2025, at 10:00, Usama Arif wrote:

> Allowing override of global THP policy per process allows workloads
> that have shown to benefit from hugepages to do so, without regressing
> workloads that wouldn't benefit. This will allow such types of workloads
> to be run/stacked on the same machine.
>
> It also helps in rolling out hugepages in hyperscaler configurations
> for workloads that benefit from them, where a single THP policy is likely
> to be used across the entire fleet, and prctl will help override it.
>
> Signed-off-by: Usama Arif <usamaarif642@gmail.com>
> ---
>  include/linux/huge_mm.h                          |  3 ++-
>  include/linux/mm_types.h                         |  7 ++-----
>  include/uapi/linux/prctl.h                       |  3 +++
>  kernel/sys.c                                     | 16 ++++++++++++++++
>  tools/include/uapi/linux/prctl.h                 |  3 +++
>  .../perf/trace/beauty/include/uapi/linux/prctl.h |  3 +++
>  6 files changed, 29 insertions(+), 6 deletions(-)
>
> diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h
> index 2f190c90192d..0587dc4b8e2d 100644
> --- a/include/linux/huge_mm.h
> +++ b/include/linux/huge_mm.h
> @@ -293,7 +293,8 @@ unsigned long thp_vma_allowable_orders(struct vm_area_struct *vma,
>  		if (vm_flags & VM_HUGEPAGE)
>  			mask |= READ_ONCE(huge_anon_orders_madvise);
>  		if (hugepage_global_always() ||
> -		    ((vm_flags & VM_HUGEPAGE) && hugepage_global_enabled()))
> +		    ((vm_flags & VM_HUGEPAGE) && hugepage_global_enabled()) ||
> +		    test_bit(MMF_THP_ALWAYS, &vma->vm_mm->flags))
>  			mask |= READ_ONCE(huge_anon_orders_inherit);
>
>  		orders &= mask;
> diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
> index e76bade9ebb1..9bcd72b2b191 100644
> --- a/include/linux/mm_types.h
> +++ b/include/linux/mm_types.h
> @@ -1704,11 +1704,8 @@ enum {
>  #define MMF_VM_MERGEABLE	16	/* KSM may merge identical pages */
>  #define MMF_VM_HUGEPAGE		17	/* set when mm is available for khugepaged */
>
> -/*
> - * This one-shot flag is dropped due to necessity of changing exe once again
> - * on NFS restore
> - */
> -//#define MMF_EXE_FILE_CHANGED	18	/* see prctl_set_mm_exe_file() */
> +/* override inherited page sizes to always for the entire process */
> + #define MMF_THP_ALWAYS	18

Could we have something like MMF_THP_POLICY_SET and another field
for "always"? Otherwise, how are we going to set MMF_THP_MADVISE if
we want it in the future?

>
>  #define MMF_HAS_UPROBES		19	/* has uprobes */
>  #define MMF_RECALC_UPROBES	20	/* MMF_HAS_UPROBES can be wrong */
> diff --git a/include/uapi/linux/prctl.h b/include/uapi/linux/prctl.h
> index 15c18ef4eb11..22c526681562 100644
> --- a/include/uapi/linux/prctl.h
> +++ b/include/uapi/linux/prctl.h
> @@ -364,4 +364,7 @@ struct prctl_mm_map {
>  # define PR_TIMER_CREATE_RESTORE_IDS_ON		1
>  # define PR_TIMER_CREATE_RESTORE_IDS_GET	2
>
> +#define PR_SET_THP_ALWAYS	78
> +#define PR_GET_THP_ALWAYS	79
> +
>  #endif /* _LINUX_PRCTL_H */
> diff --git a/kernel/sys.c b/kernel/sys.c
> index c434968e9f5d..ee56b059ff1f 100644
> --- a/kernel/sys.c
> +++ b/kernel/sys.c
> @@ -2658,6 +2658,22 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, unsigned long, arg3,
>  			clear_bit(MMF_DISABLE_THP, &me->mm->flags);
>  		mmap_write_unlock(me->mm);
>  		break;
> +	case PR_GET_THP_ALWAYS:
> +		if (arg2 || arg3 || arg4 || arg5)
> +			return -EINVAL;
> +		error = !!test_bit(MMF_THP_ALWAYS, &me->mm->flags);
> +		break;
> +	case PR_SET_THP_ALWAYS:
> +		if (arg3 || arg4 || arg5)
> +			return -EINVAL;
> +		if (mmap_write_lock_killable(me->mm))
> +			return -EINTR;
> +		if (arg2)
> +			set_bit(MMF_THP_ALWAYS, &me->mm->flags);
> +		else
> +			clear_bit(MMF_THP_ALWAYS, &me->mm->flags);
> +		mmap_write_unlock(me->mm);
> +		break;

prctl can take more than one arguments. Would it be better to add
PR_SET_THP_POLICY and PR_GET_THP_POLICY and specify PR_THP_POLICY_ALWAYS
in the second argument? So that in the future, if we want to add
more THP policies, we do not need to keep piling up PR_{GET,SET}_THP_*?

>  	case PR_MPX_ENABLE_MANAGEMENT:
>  	case PR_MPX_DISABLE_MANAGEMENT:
>  		/* No longer implemented: */
> diff --git a/tools/include/uapi/linux/prctl.h b/tools/include/uapi/linux/prctl.h
> index 35791791a879..f5f6cff42b3f 100644
> --- a/tools/include/uapi/linux/prctl.h
> +++ b/tools/include/uapi/linux/prctl.h
> @@ -328,4 +328,7 @@ struct prctl_mm_map {
>  # define PR_PPC_DEXCR_CTRL_CLEAR_ONEXEC	0x10 /* Clear the aspect on exec */
>  # define PR_PPC_DEXCR_CTRL_MASK		0x1f
>
> +#define PR_GET_THP_ALWAYS	78
> +#define PR_SET_THP_ALWAYS	79
> +
>  #endif /* _LINUX_PRCTL_H */
> diff --git a/tools/perf/trace/beauty/include/uapi/linux/prctl.h b/tools/perf/trace/beauty/include/uapi/linux/prctl.h
> index 15c18ef4eb11..680996d56faf 100644
> --- a/tools/perf/trace/beauty/include/uapi/linux/prctl.h
> +++ b/tools/perf/trace/beauty/include/uapi/linux/prctl.h
> @@ -364,4 +364,7 @@ struct prctl_mm_map {
>  # define PR_TIMER_CREATE_RESTORE_IDS_ON		1
>  # define PR_TIMER_CREATE_RESTORE_IDS_GET	2
>
> +#define PR_GET_THP_ALWAYS	78
> +#define PR_SET_THP_ALWAYS	79
> +
>  #endif /* _LINUX_PRCTL_H */
> -- 
> 2.47.1


--
Best Regards,
Yan, Zi