mm: move pfn_to_online_page() out of line
authorDan Williams <dan.j.williams@intel.com>
Fri, 26 Feb 2021 01:16:57 +0000 (17:16 -0800)
committerLinus Torvalds <torvalds@linux-foundation.org>
Fri, 26 Feb 2021 17:41:00 +0000 (09:41 -0800)
Patch series "mm: Fix pfn_to_online_page() with respect to ZONE_DEVICE", v4.

A pfn-walker that uses pfn_to_online_page() may inadvertently translate a
pfn as online and in the page allocator, when it is offline managed by a
ZONE_DEVICE mapping (details in Patch 3: ("mm: Teach pfn_to_online_page()
about ZONE_DEVICE section collisions")).

The 2 proposals under consideration are teach pfn_to_online_page() to be
precise in the presence of mixed-zone sections, or teach the memory-add
code to drop the System RAM associated with ZONE_DEVICE collisions.  In
order to not regress memory capacity by a few 10s to 100s of MiB the
approach taken in this set is to add precision to pfn_to_online_page().

In the course of validating pfn_to_online_page() a couple other fixes
fell out:

1/ soft_offline_page() fails to drop the reference taken in the
   madvise(..., MADV_SOFT_OFFLINE) case.

2/ memory_failure() uses get_dev_pagemap() to lookup ZONE_DEVICE pages,
   however that mapping may contain data pages and metadata raw pfns.
   Introduce pgmap_pfn_valid() to delineate the 2 types and fail the
   handling of raw metadata pfns.

This patch (of 4);

pfn_to_online_page() is already too large to be a macro or an inline
function.  In anticipation of further logic changes / growth, move it out
of line.

No functional change, just code movement.

Link: https://lkml.kernel.org/r/161058499000.1840162.702316708443239771.stgit@dwillia2-desk3.amr.corp.intel.com
Link: https://lkml.kernel.org/r/161058499608.1840162.10165648147615238793.stgit@dwillia2-desk3.amr.corp.intel.com
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reported-by: Michal Hocko <mhocko@kernel.org>
Acked-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Oscar Salvador <osalvador@suse.de>
Cc: Naoya Horiguchi <naoya.horiguchi@nec.com>
Cc: Qian Cai <cai@lca.pw>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
include/linux/memory_hotplug.h
mm/memory_hotplug.c

index 15acce5..3d99de0 100644 (file)
@@ -16,22 +16,7 @@ struct resource;
 struct vmem_altmap;
 
 #ifdef CONFIG_MEMORY_HOTPLUG
-/*
- * Return page for the valid pfn only if the page is online. All pfn
- * walkers which rely on the fully initialized page->flags and others
- * should use this rather than pfn_valid && pfn_to_page
- */
-#define pfn_to_online_page(pfn)                                           \
-({                                                                \
-       struct page *___page = NULL;                               \
-       unsigned long ___pfn = pfn;                                \
-       unsigned long ___nr = pfn_to_section_nr(___pfn);           \
-                                                                  \
-       if (___nr < NR_MEM_SECTIONS && online_section_nr(___nr) && \
-           pfn_valid_within(___pfn))                              \
-               ___page = pfn_to_page(___pfn);                     \
-       ___page;                                                   \
-})
+struct page *pfn_to_online_page(unsigned long pfn);
 
 /*
  * Types for free bootmem stored in page->lru.next. These have to be in
index abe43c1..fc6cdd9 100644 (file)
@@ -301,6 +301,22 @@ static int check_hotplug_memory_addressable(unsigned long pfn,
 }
 
 /*
+ * Return page for the valid pfn only if the page is online. All pfn
+ * walkers which rely on the fully initialized page->flags and others
+ * should use this rather than pfn_valid && pfn_to_page
+ */
+struct page *pfn_to_online_page(unsigned long pfn)
+{
+       unsigned long nr = pfn_to_section_nr(pfn);
+
+       if (nr < NR_MEM_SECTIONS && online_section_nr(nr) &&
+           pfn_valid_within(pfn))
+               return pfn_to_page(pfn);
+       return NULL;
+}
+EXPORT_SYMBOL_GPL(pfn_to_online_page);
+
+/*
  * Reasonably generic function for adding memory.  It is
  * expected that archs that support memory hotplug will
  * call this function after deciding the zone to which to