When invoking thp_vma_allowable_orders(), if the TVA_ENFORCE_SYSFS flag is not
specified, we will ignore the THP sysfs settings. And the MADV_COLLAPSE is an
example of such a case.
The MADV_COLLAPSE will ignore the system-wide shmem THP sysfs settings, which
means that even though we have disabled the shmem THP configuration, MADV_COLLAPSE
will still attempt to collapse into a shmem THP. This violates the rule we have
agreed upon: never means never.
Another rule for madvise, referring to David's suggestion: “allowing for collapsing
in a VM without VM_HUGEPAGE in the "madvise" mode would be fine".
To fix the MADV_COLLAPSE issue for shmem, then the current strategy should be:
For shmem, if none of always, madvise, within_size, and inherit have enabled
PMD-sized THP, then MADV_COLLAPSE will be prohibited from collapsing PMD-sized THP.
@@ -672,7 +672,7 @@ static unsigned int shmem_huge_global_enabled(struct inode *inode, pgoff_t index
fallthrough;
case SHMEM_HUGE_ADVISE:
- if (vm_flags & VM_HUGEPAGE)
+ if (shmem_huge_force || (vm_flags & VM_HUGEPAGE))
return maybe_pmd_order;
fallthrough;
default:
@@ -1806,7 +1806,7 @@ unsigned long shmem_allowable_huge_orders(struct inode *inode,
/* Allow mTHP that will be fully within i_size. */
mask |= shmem_get_orders_within_size(inode, within_size_orders, index, 0);
- if (vm_flags & VM_HUGEPAGE)
+ if (shmem_huge_force || (vm_flags & VM_HUGEPAGE))
mask |= READ_ONCE(huge_shmem_orders_madvise);
if (global_orders > 0)
diff --git a/tools/testing/selftests/mm/khugepaged.c b/tools/testing/selftests/mm/khugepaged.c
index 85bfff53dba6..9517ed99c382 100644
--- a/tools/testing/selftests/mm/khugepaged.c
+++ b/tools/testing/selftests/mm/khugepaged.c
@@ -502,7 +502,7 @@ static void __madvise_collapse(const char *msg, char *p, int nr_hpages,
printf("%s...", msg);
settings.thp_enabled = THP_ALWAYS;
- settings.shmem_enabled = SHMEM_NEVER;
+ settings.shmem_enabled = SHMEM_ALWAYS;
thp_push_settings(&settings);