On Wed, Apr 30, 2025 at 2:34 PM Zi Yan ziy@nvidia.com wrote:
On 28 Apr 2025, at 14:29, Nico Pache wrote:
Now that we have defer to globally disable THPs at fault time, lets add a defer setting to the mTHP options. This will allow khugepaged to operate at that order, while avoiding it at PF time.
Signed-off-by: Nico Pache npache@redhat.com
include/linux/huge_mm.h | 5 +++++ mm/huge_memory.c | 38 +++++++++++++++++++++++++++++++++----- mm/khugepaged.c | 8 ++++---- 3 files changed, 42 insertions(+), 9 deletions(-)
diff --git a/include/linux/huge_mm.h b/include/linux/huge_mm.h index 57e6c962afb1..a877c59bea67 100644 --- a/include/linux/huge_mm.h +++ b/include/linux/huge_mm.h @@ -96,6 +96,7 @@ extern struct kobj_attribute thpsize_shmem_enabled_attr; #define TVA_SMAPS (1 << 0) /* Will be used for procfs */ #define TVA_IN_PF (1 << 1) /* Page fault handler */ #define TVA_ENFORCE_SYSFS (1 << 2) /* Obey sysfs configuration */ +#define TVA_IN_KHUGEPAGE ((1 << 2) | (1 << 3)) /* Khugepaged defer support */
Why is TVA_IN_KHUGEPAGE a superset of TVA_ENFORCE_SYSFS? Because khugepaged also obeys sysfs configuration?
Correct! The need for a TVA_IN_KHUGEPAGED is to isolate the "deferred" mTHPs from being "allowed" unless we are in khugepaged.
I wonder if explicitly coding the behavior is better. For example, in __thp_vma_allowable_orders(), enforce_sysfs = tva_flags & (TVA_ENFORCE_SYSFS | TVA_IN_KHUGEPAGE).
I'm rather indifferent about either approach. If you (or any others) have a strong preference for an explicit (none-supersetted) TVA flag I can make the change.
#define thp_vma_allowable_order(vma, vm_flags, tva_flags, order) \ (!!thp_vma_allowable_orders(vma, vm_flags, tva_flags, BIT(order))) @@ -182,6 +183,7 @@ extern unsigned long transparent_hugepage_flags; extern unsigned long huge_anon_orders_always; extern unsigned long huge_anon_orders_madvise; extern unsigned long huge_anon_orders_inherit; +extern unsigned long huge_anon_orders_defer;
static inline bool hugepage_global_enabled(void) { @@ -306,6 +308,9 @@ unsigned long thp_vma_allowable_orders(struct vm_area_struct *vma, /* Optimization to check if required orders are enabled early. */ if ((tva_flags & TVA_ENFORCE_SYSFS) && vma_is_anonymous(vma)) {
And code here becomes tva_flags & (TVA_ENFORCE_SYSFS | TVA_IN_KHUGEPAGE).
or just (enforce_sysfs & vma_is_anon) like you mentioned. Then we check for the TVA_IN_KHUGEPAGED before appending the defer bits.
Otherwise, LGTM. Reviewed-by: Zi Yan ziy@nvidia.com
Thanks !
-- Best Regards, Yan, Zi