Skip to contentSkip to conversation

mm: Check if section present during memory block (un)registering

PATCH Superseded Yinghai Lu 3 participants 4 messages Aug 25, 2015 – Aug 26, 2015 Driver core

v1 of 3 Superseded Latest is v3 →

lore
Share & tools
lore

Yinghai Lu

author
#0 lore raw

PATCH 1/1 mm: Check if section present during memory block (un)registering

Tony found on his setup, if memory block size 512M will cause crash during booting.

 BUG: unable to handle kernel paging request at ffffea0074000020
 IP: [<ffffffff81670527>] get_nid_for_pfn+0x17/0x40
 PGD 128ffcb067 PUD 128ffc9067 PMD 0 
 Oops: 0000 [#1] SMP 
 Modules linked in:
 CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.2.0-rc8 #1
...
 Call Trace:
  [<ffffffff81453b56>] ? register_mem_sect_under_node+0x66/0xe0
  [<ffffffff81453eeb>] register_one_node+0x17b/0x240
  [<ffffffff81b1f1ed>] ? pci_iommu_alloc+0x6e/0x6e
  [<ffffffff81b1f229>] topology_init+0x3c/0x95
  [<ffffffff8100213d>] do_one_initcall+0xcd/0x1f0

The system has non continuous RAM address:
BIOS-e820: [mem 0x0000001300000000-0x0000001cffffffff] usable
BIOS-e820: [mem 0x0000001d70000000-0x0000001ec7ffefff] usable
BIOS-e820: [mem 0x0000001f00000000-0x0000002bffffffff] usable
BIOS-e820: [mem 0x0000002c18000000-0x0000002d6fffefff] usable
BIOS-e820: [mem 0x0000002e00000000-0x00000039ffffffff] usable

So there are start sections in memory block not present. For example:
memory block : [0x2c18000000, 0x2c20000000) 512M first three sections are not present.

Current register_mem_sect_under_node() assume first section is present, but memory block section number range [start_section_nr, end_section_nr] would include not present section.

For arch that support vmemmap, we don't setup memmap for struct page area within not present sections area.

So skip the pfn range that belong to not present section.

Als fixes unregister_mem_sect_under_nodes().

drivers/base/node.c +18 -4
index 31df474d..cc910ad 100644
@@ -390,8 +390,14 @@ int register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
 	sect_end_pfn = section_nr_to_pfn(mem_blk->end_section_nr);
 	sect_end_pfn += PAGES_PER_SECTION - 1;
 	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
-		int page_nid;
+		int page_nid, scn_nr;
 
+		scn_nr = pfn_to_section_nr(pfn);
+		if (!present_section_nr(scn_nr)) {
+			pfn = round_down(pfn + PAGES_PER_SECTION,
+					 PAGES_PER_SECTION) - 1;
+			continue;
+		}
 		page_nid = get_nid_for_pfn(pfn);
 		if (page_nid < 0)
 			continue;
@@ -426,10 +432,18 @@ int unregister_mem_sect_under_nodes(struct memory_block *mem_blk,
 		return -ENOMEM;
 	nodes_clear(*unlinked_nodes);
 
-	sect_start_pfn = section_nr_to_pfn(phys_index);
-	sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
+	sect_start_pfn = section_nr_to_pfn(mem_blk->start_section_nr);
+	sect_end_pfn = section_nr_to_pfn(mem_blk->end_section_nr);
+	sect_end_pfn += PAGES_PER_SECTION - 1;
 	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
-		int nid;
+		int nid, scn_nr;
+
+		scn_nr = pfn_to_section_nr(pfn);
+		if (!present_section_nr(scn_nr)) {
+			pfn = round_down(pfn + PAGES_PER_SECTION,
+					 PAGES_PER_SECTION) - 1;
+			continue;
+		}
 
 		nid = get_nid_for_pfn(pfn);
 		if (nid < 0)
#1 lore raw
Quoted 33 lines · On Tue, 25 Aug 2015 15:41:16 -0700 Yinghai Lu <[email protected]> wrote:

Tony found on his setup, if memory block size 512M will cause crash during booting.

 BUG: unable to handle kernel paging request at ffffea0074000020
 IP: [<ffffffff81670527>] get_nid_for_pfn+0x17/0x40
 PGD 128ffcb067 PUD 128ffc9067 PMD 0 
 Oops: 0000 [#1] SMP 
 Modules linked in:
 CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.2.0-rc8 #1
...
 Call Trace:
  [<ffffffff81453b56>] ? register_mem_sect_under_node+0x66/0xe0
  [<ffffffff81453eeb>] register_one_node+0x17b/0x240
  [<ffffffff81b1f1ed>] ? pci_iommu_alloc+0x6e/0x6e
  [<ffffffff81b1f229>] topology_init+0x3c/0x95
  [<ffffffff8100213d>] do_one_initcall+0xcd/0x1f0

The system has non continuous RAM address:
BIOS-e820: [mem 0x0000001300000000-0x0000001cffffffff] usable
BIOS-e820: [mem 0x0000001d70000000-0x0000001ec7ffefff] usable
BIOS-e820: [mem 0x0000001f00000000-0x0000002bffffffff] usable
BIOS-e820: [mem 0x0000002c18000000-0x0000002d6fffefff] usable
BIOS-e820: [mem 0x0000002e00000000-0x00000039ffffffff] usable

So there are start sections in memory block not present. For example:
memory block : [0x2c18000000, 0x2c20000000) 512M first three sections are not present.

Current register_mem_sect_under_node() assume first section is present, but memory block section number range [start_section_nr, end_section_nr] would include not present section.

Quoted 6 lines

For arch that support vmemmap, we don't setup memmap for struct page area within not present sections area.

So skip the pfn range that belong to not present section.

Als fixes unregister_mem_sect_under_nodes().

It appears this should be backported into -stable kernels, yes? Do you know which kernel versions need the fix?

Quoted 5 lines
drivers/base/node.c +0 -0
@@ -390,8 +390,14 @@ int register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
 	sect_end_pfn = section_nr_to_pfn(mem_blk->end_section_nr);
 	sect_end_pfn += PAGES_PER_SECTION - 1;

drivers/base/node.c int register_mem_sect_under_node(struct memory_block *mem_blk, int nid) · line 400

Quoted 10 lines
drivers/base/node.c +7 -1
 	for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
-		int page_nid;
+		int page_nid, scn_nr;
 
+		scn_nr = pfn_to_section_nr(pfn);
+		if (!present_section_nr(scn_nr)) {
+			pfn = round_down(pfn + PAGES_PER_SECTION,
+					 PAGES_PER_SECTION) - 1;
+			continue;
+		}

Can we please add a comment here telling readers why this is being done? What scenario is being detected and how it comes about.

Quoted 3 lines
drivers/base/node.c +0 -0
 		page_nid = get_nid_for_pfn(pfn);
 		if (page_nid < 0)
 			continue;
#2 lore raw
Quoted 2 lines

It appears this should be backported into -stable kernels, yes? Do you know which kernel versions need the fix?

For my setup the problem is first seen after:

commit bdee237c0343 " x86: mm: Use 2GB memory block size on large memory x86-64 systems"

which appeared in v3.19 and forced a 2GB memory block size. But it could happen on older systems depending on the block size picked based on the alignment of max_pfn.

Looking further back (to v3.15) ... we used a fixed MIN_MEMORY_BLOCK_SIZE for all systems prior to

commit 982792c782ef "x86, mm: probe memory block size for generic x86 64bit"

So maybe:

-Tony

author
#3 lore raw
In reply to Andrew Morton
Quoted 38 lines · Andrew Morton
Quoted 36 lines · On Tue, 25 Aug 2015 15:41:16 -0700 Yinghai Lu <[email protected]> wrote:

Tony found on his setup, if memory block size 512M will cause crash during booting.

 BUG: unable to handle kernel paging request at ffffea0074000020
 IP: [<ffffffff81670527>] get_nid_for_pfn+0x17/0x40
 PGD 128ffcb067 PUD 128ffc9067 PMD 0
 Oops: 0000 [#1] SMP
 Modules linked in:
 CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.2.0-rc8 #1
...
 Call Trace:
  [<ffffffff81453b56>] ? register_mem_sect_under_node+0x66/0xe0
  [<ffffffff81453eeb>] register_one_node+0x17b/0x240
  [<ffffffff81b1f1ed>] ? pci_iommu_alloc+0x6e/0x6e
  [<ffffffff81b1f229>] topology_init+0x3c/0x95
  [<ffffffff8100213d>] do_one_initcall+0xcd/0x1f0

The system has non continuous RAM address:
BIOS-e820: [mem 0x0000001300000000-0x0000001cffffffff] usable
BIOS-e820: [mem 0x0000001d70000000-0x0000001ec7ffefff] usable
BIOS-e820: [mem 0x0000001f00000000-0x0000002bffffffff] usable
BIOS-e820: [mem 0x0000002c18000000-0x0000002d6fffefff] usable
BIOS-e820: [mem 0x0000002e00000000-0x00000039ffffffff] usable

So there are start sections in memory block not present. For example:
memory block : [0x2c18000000, 0x2c20000000) 512M first three sections are not present.

Current register_mem_sect_under_node() assume first section is present, but memory block section number range [start_section_nr, end_section_nr] would include not present section.

For arch that support vmemmap, we don't setup memmap for struct page area within not present sections area.

Quoted 6 lines
Quoted 3 lines

So skip the pfn range that belong to not present section.

Als fixes unregister_mem_sect_under_nodes().

It appears this should be backported into -stable kernels, yes? Do you know which kernel versions need the fix?

should add following according to Tony's email.

Fixes bdee237c0343 ("x86: mm: Use 2GB memory block size on large

memory x86-64 systems")

Fixes 982792c782ef ("x86, mm: probe memory block size for generic x86 64bit")
Quoted 10 lines
Quoted 10 lines
drivers/base/node.c +1 -1
@@ -390,8 +390,14 @@ int register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
      sect_end_pfn = section_nr_to_pfn(mem_blk->end_section_nr);
      sect_end_pfn += PAGES_PER_SECTION - 1;
      for (pfn = sect_start_pfn; pfn <= sect_end_pfn; pfn++) {
-             int page_nid;
+             int page_nid, scn_nr;
 
Quoted 10 lines

drivers/base/node.c int register_mem_sect_under_node(struct memory_block *mem_blk, int nid) · line 400

Quoted 6 lines
drivers/base/node.c +6 -0
+             scn_nr = pfn_to_section_nr(pfn);
+             if (!present_section_nr(scn_nr)) {
+                     pfn = round_down(pfn + PAGES_PER_SECTION,
+                                      PAGES_PER_SECTION) - 1;
+                     continue;
+             }

Can we please add a comment here telling readers why this is being done? What scenario is being detected and how it comes about.

Yes, should add

/* skip pfn range from absent memory section */