Skip to contentSkip to conversation

x86: fix the initialization of physnode_map

PATCH In review Petr Tesarik 3 participants 4 messages Feb 1, 2014 – Feb 2, 2014

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 x86: fix the initialization of physnode_map

With DISCONTIGMEM, the mapping between a pfn and its owning node is initialized using data provided by the BIOS. However, the initialization may fail if the extents are not aligned to section boundary (64M).

The symptom of this bug is an early boot failure in pfn_to_page(), as it tries to access NODE_DATA(__nid) using index from an unitialized element of the physnode_map[] array.

While the bug is always present, it is more likely to be hit in kdump kernels on large machines, because:

1. The memory map for a kdump kernel is specified as exactmap, and exactmap is more likely to be unaligned.

2. Large reservations are more likely to span across a 64M boundary.

---
 arch/x86/mm/numa_32.c | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/mm/numa_32.c +2 -0
index 0342d27..8c62ec6 100644
@@ -52,6 +52,8 @@ void memory_present(int nid, unsigned long start, unsigned long end)
 			nid, start, end);
 	printk(KERN_DEBUG "  Setting physnode_map array to node %d for pfns:\n", nid);
 	printk(KERN_DEBUG "  ");
+	pfn = round_down(pfn, PAGES_PER_SECTION);
+	end = round_up(end, PAGES_PER_SECTION);
 	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION) {
 		physnode_map[pfn / PAGES_PER_SECTION] = nid;
 		printk(KERN_CONT "%lx ", pfn);
Signature · 1 line
1.8.4.5
Acked-by on v2 1/1
#1 lore raw
Quoted 11 lines · Petr Tesarik

With DISCONTIGMEM, the mapping between a pfn and its owning node is initialized using data provided by the BIOS. However, the initialization may fail if the extents are not aligned to section boundary (64M).

The symptom of this bug is an early boot failure in pfn_to_page(), as it tries to access NODE_DATA(__nid) using index from an unitialized element of the physnode_map[] array.

While the bug is always present, it is more likely to be hit in kdump kernels on large machines, because:

Quoted 6 lines

1. The memory map for a kdump kernel is specified as exactmap, and exactmap is more likely to be unaligned.

2. Large reservations are more likely to span across a 64M boundary.

#2 lore raw

[tip:x86/urgent] x86: Fix the initialization of physnode_map

In reply to Petr Tesarik
x86: Fix the initialization of physnode_map
Commit-ID:  170750c108bb9382f32a2a99d097aa07d551a89e
Gitweb:     http://git.kernel.org/tip/170750c108bb9382f32a2a99d097aa07d551a89e
Author:     Petr Tesarik <[email protected]>
AuthorDate: Sat, 1 Feb 2014 13:30:19 +0100
Committer:  H. Peter Anvin <[email protected]>
CommitDate: Sat, 1 Feb 2014 21:57:48 -0800

x86: Fix the initialization of physnode_map

With DISCONTIGMEM, the mapping between a pfn and its owning node is initialized using data provided by the BIOS. However, the initialization may fail if the extents are not aligned to section boundary (64M).

The symptom of this bug is an early boot failure in pfn_to_page(), as it tries to access NODE_DATA(__nid) using index from an unitialized element of the physnode_map[] array.

While the bug is always present, it is more likely to be hit in kdump kernels on large machines, because:

1. The memory map for a kdump kernel is specified as exactmap, and exactmap is more likely to be unaligned.

2. Large reservations are more likely to span across a 64M boundary.

---
 arch/x86/mm/numa_32.c | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/mm/numa_32.c +2 -0
index 0342d27..8c62ec6 100644
@@ -52,6 +52,8 @@ void memory_present(int nid, unsigned long start, unsigned long end)
 			nid, start, end);
 	printk(KERN_DEBUG "  Setting physnode_map array to node %d for pfns:\n", nid);
 	printk(KERN_DEBUG "  ");
+	pfn = round_down(pfn, PAGES_PER_SECTION);
+	end = round_up(end, PAGES_PER_SECTION);
 	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION) {
 		physnode_map[pfn / PAGES_PER_SECTION] = nid;
 		printk(KERN_CONT "%lx ", pfn);
#3 lore raw

[tip:x86/urgent] x86: Fix the initialization of physnode_map

In reply to Petr Tesarik
x86: Fix the initialization of physnode_map
Commit-ID:  85fc73a2cdf10cf42bc36fb3bca3896b2095a1c2
Gitweb:     http://git.kernel.org/tip/85fc73a2cdf10cf42bc36fb3bca3896b2095a1c2
Author:     Petr Tesarik <[email protected]>
AuthorDate: Sat, 1 Feb 2014 13:30:19 +0100
Committer:  H. Peter Anvin <[email protected]>
CommitDate: Sat, 1 Feb 2014 22:15:51 -0800

x86: Fix the initialization of physnode_map

With DISCONTIGMEM, the mapping between a pfn and its owning node is initialized using data provided by the BIOS. However, the initialization may fail if the extents are not aligned to section boundary (64M).

The symptom of this bug is an early boot failure in pfn_to_page(), as it tries to access NODE_DATA(__nid) using index from an unitialized element of the physnode_map[] array.

While the bug is always present, it is more likely to be hit in kdump kernels on large machines, because:

1. The memory map for a kdump kernel is specified as exactmap, and exactmap is more likely to be unaligned.

2. Large reservations are more likely to span across a 64M boundary.

[ hpa: fixed incorrect use of "pfn" instead of "start" ]

---
 arch/x86/mm/numa_32.c | 2 ++
 1 file changed, 2 insertions(+)
arch/x86/mm/numa_32.c +2 -0
index 0342d27..47b6436 100644
@@ -52,6 +52,8 @@ void memory_present(int nid, unsigned long start, unsigned long end)
 			nid, start, end);
 	printk(KERN_DEBUG "  Setting physnode_map array to node %d for pfns:\n", nid);
 	printk(KERN_DEBUG "  ");
+	start = round_down(start, PAGES_PER_SECTION);
+	end = round_up(end, PAGES_PER_SECTION);
 	for (pfn = start; pfn < end; pfn += PAGES_PER_SECTION) {
 		physnode_map[pfn / PAGES_PER_SECTION] = nid;
 		printk(KERN_CONT "%lx ", pfn);