[020/192] mm/vmalloc: enable mapping of huge pages at pte level in vmalloc

Commit Message

Andrew Morton July 1, 2021, 1:48 a.m. UTC

From: Christophe Leroy <[e-mail]>
Subject: mm/vmalloc: enable mapping of huge pages at pte level in vmalloc

On some architectures like powerpc, there are huge pages that are mapped
at pte level.

Enable it in vmalloc.

For that, architectures can provide arch_vmap_pte_supported_shift() that
returns the shift for pages to map at pte level.

Link: https://lkml.kernel.org/r/2c717e3b[e-mail]
Signed-off-by: Christophe Leroy <[e-mail]>
Cc: Benjamin Herrenschmidt <[e-mail]>
Cc: Michael Ellerman <[e-mail]>
Cc: Mike Kravetz <[e-mail]>
Cc: Mike Rapoport <[e-mail]>
Cc: Nicholas Piggin <[e-mail]>
Cc: Paul Mackerras <[e-mail]>
Cc: Uladzislau Rezki <[e-mail]>
Signed-off-by: Andrew Morton <[e-mail]>
---

 include/linux/vmalloc.h |    7 +++++++
 mm/vmalloc.c            |   13 +++++++------
 2 files changed, 14 insertions(+), 6 deletions(-)

Patch

--- a/include/linux/vmalloc.h~mm-vmalloc-enable-mapping-of-huge-pages-at-pte-level-in-vmalloc
+++ a/include/linux/vmalloc.h
@@ -112,6 +112,13 @@  static inline unsigned long arch_vmap_pt
 }
 #endif
 
+#ifndef arch_vmap_pte_supported_shift
+static inline int arch_vmap_pte_supported_shift(unsigned long size)
+{
+	return PAGE_SHIFT;
+}
+#endif
+
 /*
  *	Highlevel APIs for driver use
  */
--- a/mm/vmalloc.c~mm-vmalloc-enable-mapping-of-huge-pages-at-pte-level-in-vmalloc
+++ a/mm/vmalloc.c
@@ -2927,8 +2927,7 @@  void *__vmalloc_node_range(unsigned long
 		return NULL;
 	}
 
-	if (vmap_allow_huge && !(vm_flags & VM_NO_HUGE_VMAP) &&
-			arch_vmap_pmd_supported(prot)) {
+	if (vmap_allow_huge && !(vm_flags & VM_NO_HUGE_VMAP)) {
 		unsigned long size_per_node;
 
 		/*
@@ -2941,11 +2940,13 @@  void *__vmalloc_node_range(unsigned long
 		size_per_node = size;
 		if (node == NUMA_NO_NODE)
 			size_per_node /= num_online_nodes();
-		if (size_per_node >= PMD_SIZE) {
+		if (arch_vmap_pmd_supported(prot) && size_per_node >= PMD_SIZE)
 			shift = PMD_SHIFT;
-			align = max(real_align, 1UL << shift);
-			size = ALIGN(real_size, 1UL << shift);
-		}
+		else
+			shift = arch_vmap_pte_supported_shift(size_per_node);
+
+		align = max(real_align, 1UL << shift);
+		size = ALIGN(real_size, 1UL << shift);
 	}
 
 again: