Re: [PATCH v3 09/10] mm: thp: always enable mTHP support

[Zi Yan <ziy@nvidia.com>]

On 8 Apr 2026, at 16:23, Luiz Capitulino wrote:

> If PMD-sized pages are not supported on an architecture (ie. the
> arch implements arch_has_pmd_leaves() and it returns false) then the
> current code disables all THP, including mTHP.
>
> This commit fixes this by allowing mTHP to be always enabled for all
> archs. When PMD-sized pages are not supported, its sysfs entry won't be
> created and their mapping will be disallowed at page-fault time.
>
> Similarly, this commit implements the following changes for shmem:
>
>  - In shmem_allowable_huge_orders(): drop the pgtable_has_pmd_leaves()
>    check so that mTHP sizes are considered
>  - In shmem_alloc_and_add_folio(): don't consider PMD and PUD orders
>    when PMD-sized pages are not supported by the CPU
>
> Signed-off-by: Luiz Capitulino <luizcap@redhat.com>
> ---
>  mm/huge_memory.c | 13 ++++++++-----
>  mm/shmem.c       |  4 +++-
>  2 files changed, 11 insertions(+), 6 deletions(-)
>
> diff --git a/mm/huge_memory.c b/mm/huge_memory.c
> index 86e489c0a150..6de3d8ebc35c 100644
> --- a/mm/huge_memory.c
> +++ b/mm/huge_memory.c
> @@ -118,6 +118,9 @@ unsigned long __thp_vma_allowable_orders(struct vm_area_struct *vma,
>  	else
>  		supported_orders = THP_ORDERS_ALL_FILE_DEFAULT;
>
> +	if (!pgtable_has_pmd_leaves())
> +		supported_orders &= ~(BIT(PMD_ORDER) | BIT(PUD_ORDER));

Why is BIT(PUD_ORDER) also removed? I thought PMD THP support and PUD THP support
are separate. Here the code implies PUD THP relies on PMD THP. Is that the case?

> +
>  	orders &= supported_orders;
>  	if (!orders)
>  		return 0;
> @@ -125,7 +128,7 @@ unsigned long __thp_vma_allowable_orders(struct vm_area_struct *vma,
>  	if (!vma->vm_mm)		/* vdso */
>  		return 0;
>
> -	if (!pgtable_has_pmd_leaves() || vma_thp_disabled(vma, vm_flags, forced_collapse))
> +	if (vma_thp_disabled(vma, vm_flags, forced_collapse))
>  		return 0;
>
>  	/* khugepaged doesn't collapse DAX vma, but page fault is fine. */
> @@ -787,7 +790,7 @@ static int __init hugepage_init_sysfs(struct kobject **hugepage_kobj)
>  	 * disable all other sizes. powerpc's PMD_ORDER isn't a compile-time
>  	 * constant so we have to do this here.
>  	 */
> -	if (!anon_orders_configured)
> +	if (!anon_orders_configured && pgtable_has_pmd_leaves())
>  		huge_anon_orders_inherit = BIT(PMD_ORDER);
>
>  	*hugepage_kobj = kobject_create_and_add("transparent_hugepage", mm_kobj);
> @@ -809,6 +812,9 @@ static int __init hugepage_init_sysfs(struct kobject **hugepage_kobj)
>  	}
>
>  	orders = THP_ORDERS_ALL_ANON | THP_ORDERS_ALL_FILE_DEFAULT;
> +	if (!pgtable_has_pmd_leaves())
> +		orders &= ~(BIT(PMD_ORDER) | BIT(PUD_ORDER));
> +

Ditto.

>  	order = highest_order(orders);
>  	while (orders) {
>  		thpsize = thpsize_create(order, *hugepage_kobj);
> @@ -908,9 +914,6 @@ static int __init hugepage_init(void)
>  	int err;
>  	struct kobject *hugepage_kobj;
>
> -	if (!pgtable_has_pmd_leaves())
> -		return -EINVAL;
> -
>  	/*
>  	 * hugepages can't be allocated by the buddy allocator
>  	 */
The code after is:

MAYBE_BUILD_BUG_ON(HPAGE_PMD_ORDER > MAX_PAGE_ORDER);

Should this check be removed or only performed when pgtable_has_pmd_leaves()?

I do not know if there is a possible Kconfig that lowers MAX_PAGE_ORDER
below PMD_ORDER and enables THP. After this patchset, that might be valid
if people do not want to use mTHP but not PMD THP.


> diff --git a/mm/shmem.c b/mm/shmem.c
> index 613393eae5a9..b49a30475cb0 100644
> --- a/mm/shmem.c
> +++ b/mm/shmem.c
> @@ -1839,7 +1839,7 @@ unsigned long shmem_allowable_huge_orders(struct inode *inode,
>  	vm_flags_t vm_flags = vma ? vma->vm_flags : 0;
>  	unsigned int global_orders;
>
> -	if (!pgtable_has_pmd_leaves() || (vma && vma_thp_disabled(vma, vm_flags, shmem_huge_force)))
> +	if (vma && vma_thp_disabled(vma, vm_flags, shmem_huge_force))
>  		return 0;
>
>  	global_orders = shmem_huge_global_enabled(inode, index, write_end,
> @@ -1947,6 +1947,8 @@ static struct folio *shmem_alloc_and_add_folio(struct vm_fault *vmf,
>
>  	if (!IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE))
>  		orders = 0;
> +	else if (!pgtable_has_pmd_leaves())
> +		orders &= ~(BIT(PMD_ORDER) | BIT(PUD_ORDER));

Same question as the first one.

>
>  	if (orders > 0) {
>  		suitable_orders = shmem_suitable_orders(inode, vmf,
> -- 
> 2.53.0


Best Regards,
Yan, Zi