Skip to contentSkip to conversation

Cleanup and fixups for vmemmap handling

PATCH Superseded Oscar Salvador 3 participants 7 messages Feb 2, 2021

Subsystems (3)

v1 of 6 Superseded Latest is v6 →

lore
Share & tools
lore
#0 lore raw

cover Cleanup and fixups for vmemmap handling

Hi,

During the review of patch#2, David noticed that we could remove the !PAGE_ALIGNED handling in remove_pte_table, as we should never get there. Therefore, let us place that cleanup before the sub-pmd-populated fixup.

More details can be found in the respective changelogs.

[1] https://lkml.org/lkml/2021/1/29/82

Oscar Salvador (2):
x86/vmemmap: Remove !PAGE_ALIGNED case in remove_pte_table x86/vmemmap: Handle unpopulated sub-pmd ranges

 arch/x86/mm/init_64.c | 148 +++++++++++++++++++++++++++++-------------
 1 file changed, 103 insertions(+), 45 deletions(-)
Signature · 1 line
2.26.2
#1 lore raw

PATCH 1/2 x86/vmemmap: Remove !PAGE_ALIGNED case in remove_pte_table

remove_pte_table() is prepared to handle the case where either the start or the end of the range is not PAGE aligned.
This cannot actually happen:

__populate_section_memmap enforces the range to be PMD aligned, so as long as the size of the struct page remains multiple of 8, the vmemmap range will be aligned to PAGE_SIZE.

Drop the unneeded handling of such case and place a VM_BUG_ON in vmemmap_{populate,free} to catch nasty cases.

---
 arch/x86/mm/init_64.c | 61 ++++++++++++++++++-------------------------
 1 file changed, 26 insertions(+), 35 deletions(-)
arch/x86/mm/init_64.c +26 -35
index b5a3fa4033d3..4cfa902ec861 100644
@@ -962,7 +962,6 @@ remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
 {
 	unsigned long next, pages = 0;
 	pte_t *pte;
-	void *page_addr;
 	phys_addr_t phys_addr;
 
 	pte = pte_start + pte_index(addr);
@@ -983,42 +982,19 @@ remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
 		if (phys_addr < (phys_addr_t)0x40000000)
 			return;
 
-		if (PAGE_ALIGNED(addr) && PAGE_ALIGNED(next)) {
-			/*
-			 * Do not free direct mapping pages since they were
-			 * freed when offlining, or simplely not in use.
-			 */
-			if (!direct)
-				free_pagetable(pte_page(*pte), 0);
-
-			spin_lock(&init_mm.page_table_lock);
-			pte_clear(&init_mm, addr, pte);
-			spin_unlock(&init_mm.page_table_lock);
-
-			/* For non-direct mapping, pages means nothing. */
-			pages++;
-		} else {
-			/*
-			 * If we are here, we are freeing vmemmap pages since
-			 * direct mapped memory ranges to be freed are aligned.
-			 *
-			 * If we are not removing the whole page, it means
-			 * other page structs in this page are being used and
-			 * we canot remove them. So fill the unused page_structs
-			 * with 0xFD, and remove the page when it is wholly
-			 * filled with 0xFD.
-			 */
-			memset((void *)addr, PAGE_INUSE, next - addr);
+		/*
+		 * Do not free direct mapping pages since they were
+		 * freed when offlining, or simplely not in use.
+		 */
+		if (!direct)
+			free_pagetable(pte_page(*pte), 0);
 
-			page_addr = page_address(pte_page(*pte));
-			if (!memchr_inv(page_addr, PAGE_INUSE, PAGE_SIZE)) {
-				free_pagetable(pte_page(*pte), 0);
+		spin_lock(&init_mm.page_table_lock);
+		pte_clear(&init_mm, addr, pte);
+		spin_unlock(&init_mm.page_table_lock);
 
-				spin_lock(&init_mm.page_table_lock);
-				pte_clear(&init_mm, addr, pte);
-				spin_unlock(&init_mm.page_table_lock);
-			}
-		}
+		/* For non-direct mapping, pages means nothing. */
+		pages++;
 	}
 
 	/* Call free_pte_table() in remove_pmd_table(). */
@@ -1197,6 +1173,12 @@ remove_pagetable(unsigned long start, unsigned long end, bool direct,
 void __ref vmemmap_free(unsigned long start, unsigned long end,
 		struct vmem_altmap *altmap)
 {
+	/*
+	 * See comment in vmemmap_populate().
+	 */
+	VM_BUG_ON(!IS_ALIGNED(start, PAGE_SIZE));
+	VM_BUG_ON(!IS_ALIGNED(end, PAGE_SIZE));
+
 	remove_pagetable(start, end, false, altmap);
 }
 
@@ -1556,6 +1538,15 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
 {
 	int err;
 
+	/*
+	 * __populate_section_memmap enforces the range to be added to be
+	 * PMD aligned. Therefore we can be sure that, as long as the
+	 * struct page size is multiple of 8, the vmemmap range will be
+	 * PAGE aligned.
+	 */
+	VM_BUG_ON(!IS_ALIGNED(start, PAGE_SIZE));
+	VM_BUG_ON(!IS_ALIGNED(end, PAGE_SIZE));
+
 	if (end - start < PAGES_PER_SECTION * sizeof(struct page))
 		err = vmemmap_populate_basepages(start, end, node, NULL);
 	else if (boot_cpu_has(X86_FEATURE_PSE))
Signature · 1 line
2.26.2
#2 lore raw

Re: [PATCH 1/2] x86/vmemmap: Remove !PAGE_ALIGNED case in remove_pte_table

Quoted 32 lines
arch/x86/mm/init_64.c +0 -21
@@ -962,7 +962,6 @@ remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
  {
  	unsigned long next, pages = 0;
  	pte_t *pte;
-	void *page_addr;
  	phys_addr_t phys_addr;
 
  	pte = pte_start + pte_index(addr);
@@ -983,42 +982,19 @@ remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end,
  		if (phys_addr < (phys_addr_t)0x40000000)
  			return;
 
-		if (PAGE_ALIGNED(addr) && PAGE_ALIGNED(next)) {
-			/*
-			 * Do not free direct mapping pages since they were
-			 * freed when offlining, or simplely not in use.
-			 */
-			if (!direct)
-				free_pagetable(pte_page(*pte), 0);
-
-			spin_lock(&init_mm.page_table_lock);
-			pte_clear(&init_mm, addr, pte);
-			spin_unlock(&init_mm.page_table_lock);
-
-			/* For non-direct mapping, pages means nothing. */
-			pages++;
-		} else {
-			/*
-			 * If we are here, we are freeing vmemmap pages since
-			 * direct mapped memory ranges to be freed are aligned.
-			 *
-			 * If we are not removing the whole page, it means

arch/x86/mm/init_64.c remove_pte_table(pte_t *pte_start, unsigned long addr, unsigned long end, · line 988

Quoted 10 lines
arch/x86/mm/init_64.c +4 -6
-			 * other page structs in this page are being used and
-			 * we canot remove them. So fill the unused page_structs
-			 * with 0xFD, and remove the page when it is wholly
-			 * filled with 0xFD.
-			 */
-			memset((void *)addr, PAGE_INUSE, next - addr);
+		/*
+		 * Do not free direct mapping pages since they were
+		 * freed when offlining, or simplely not in use.
+		 */

s/simplely/simply/ (I know, not your fault :) )

However, that comment is highly confusing. There is nothing to free in case of the a direct mapping; we never allocated anything. That's how a direct map works.

I'd just get rid of the comment completely - we also don't have one at the other "if (!direct)" places.

(side note: all this freeing before unmapping looks very weird, but at least we should have valid accesses anymore)

Quoted 2 lines
arch/x86/mm/init_64.c +2 -0
+		if (!direct)
+			free_pagetable(pte_page(*pte), 0);

[...]

arch/x86/mm/init_64.c remove_pagetable(unsigned long start, unsigned long end, bool direct, · line 1178

Quoted 4 lines
  {
arch/x86/mm/init_64.c +3 -0
+	/*
+	 * See comment in vmemmap_populate().
+	 */

I'd drop this comment ...

Quoted 7 lines
arch/x86/mm/init_64.c +3 -0
+	VM_BUG_ON(!IS_ALIGNED(start, PAGE_SIZE));
+	VM_BUG_ON(!IS_ALIGNED(end, PAGE_SIZE));
+
  	remove_pagetable(start, end, false, altmap);
  }

arch/x86/mm/init_64.c int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node, · line 1546

Quoted 9 lines
arch/x86/mm/init_64.c +6 -0
@@ -1556,6 +1538,15 @@ int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node,
  {
  	int err;
 
+	/*
+	 * __populate_section_memmap enforces the range to be added to be
+	 * PMD aligned. Therefore we can be sure that, as long as the
+	 * struct page size is multiple of 8, the vmemmap range will be
+	 * PAGE aligned.
+	 */

... and this comment, moving the details into the patch description.

The commit should be easy to detect using git blame in case anybody wonders why.

arch/x86/mm/init_64.c int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node, · line 1552

Quoted 7 lines
arch/x86/mm/init_64.c +3 -0
+	VM_BUG_ON(!IS_ALIGNED(start, PAGE_SIZE));
+	VM_BUG_ON(!IS_ALIGNED(end, PAGE_SIZE));
+
  	if (end - start < PAGES_PER_SECTION * sizeof(struct page))
  		err = vmemmap_populate_basepages(start, end, node, NULL);
  	else if (boot_cpu_has(X86_FEATURE_PSE))

Apart from that looks good, thanks!

Signature · 3 lines
Thanks,

David / dhildenb
#3 lore raw

PATCH 2/2 x86/vmemmap: Handle unpopulated sub-pmd ranges

In reply to Oscar Salvador

When the size of a struct page is not multiple of 2MB, sections do not span a PMD anymore and so when populating them some parts of the PMD will remain unused.
Because of this, PMDs will be left behind when depopulating sections since remove_pmd_table() thinks that those unused parts are still in use.

Fix this by marking the unused parts with PAGE_UNUSED, so memchr_inv() will do the right thing and will let us free the PMD when the last user of it is gone.

This patch is based on a similar patch by David Hildenbrand:

https://lore.kernel.org/linux-mm/[email protected]/
https://lore.kernel.org/linux-mm/[email protected]/

---
 arch/x86/mm/init_64.c | 87 ++++++++++++++++++++++++++++++++++++++-----
 1 file changed, 77 insertions(+), 10 deletions(-)
arch/x86/mm/init_64.c +77 -10
index 4cfa902ec861..b239708e504e 100644
@@ -871,7 +871,72 @@ int arch_add_memory(int nid, u64 start, u64 size,
 	return add_pages(nid, start_pfn, nr_pages, params);
 }
 
-#define PAGE_INUSE 0xFD
+#define PAGE_UNUSED 0xFD
+
+/*
+ * The unused vmemmap range, which was not yet memset(PAGE_UNUSED) ranges
+ * from unused_pmd_start to next PMD_SIZE boundary.
+ */
+static unsigned long unused_pmd_start __meminitdata;
+
+static void __meminit vmemmap_flush_unused_pmd(void)
+{
+	if (!unused_pmd_start)
+		return;
+	/*
+	 * Clears (unused_pmd_start, PMD_END]
+	 */
+	memset((void *)unused_pmd_start, PAGE_UNUSED,
+	       ALIGN(unused_pmd_start, PMD_SIZE) - unused_pmd_start);
+	unused_pmd_start = 0;
+}
+
+/* Returns true if the PMD is completely unused and thus it can be freed */
+static bool __meminit vmemmap_unuse_sub_pmd(unsigned long addr, unsigned long end)
+{
+	unsigned long start = ALIGN_DOWN(addr, PMD_SIZE);
+
+	vmemmap_flush_unused_pmd();
+	memset((void *)addr, PAGE_UNUSED, end - addr);
+
+	return !memchr_inv((void *)start, PAGE_UNUSED, PMD_SIZE);
+}
+
+static void __meminit vmemmap_use_sub_pmd(unsigned long start, unsigned long end)
+{
+	/*
+	 * We only optimize if the new used range directly follows the
+	 * previously unused range (esp., when populating consecutive sections).
+	 */
+	if (unused_pmd_start == start) {
+		if (likely(IS_ALIGNED(end, PMD_SIZE)))
+			unused_pmd_start = 0;
+		else
+			unused_pmd_start = end;
+		return;
+	}
+
+	vmemmap_flush_unused_pmd();
+}
+
+static void __meminit vmemmap_use_new_sub_pmd(unsigned long start, unsigned long end)
+{
+	vmemmap_flush_unused_pmd();
+
+	/*
+	 * Mark the unused parts of the new memmap range
+	 */
+	if (!IS_ALIGNED(start, PMD_SIZE))
+		memset((void *)start, PAGE_UNUSED,
+		       start - ALIGN_DOWN(start, PMD_SIZE));
+	/*
+	 * We want to avoid memset(PAGE_UNUSED) when populating the vmemmap of
+	 * consecutive sections. Remember for the last added PMD the last
+	 * unused range in the populated PMD.
+	 */
+	if (!IS_ALIGNED(end, PMD_SIZE))
+		unused_pmd_start = end;
+}
 
 static void __meminit free_pagetable(struct page *page, int order)
 {
@@ -1010,7 +1075,6 @@ remove_pmd_table(pmd_t *pmd_start, unsigned long addr, unsigned long end,
 	unsigned long next, pages = 0;
 	pte_t *pte_base;
 	pmd_t *pmd;
-	void *page_addr;
 
 	pmd = pmd_start + pmd_index(addr);
 	for (; addr < end; addr = next, pmd++) {
@@ -1031,12 +1095,10 @@ remove_pmd_table(pmd_t *pmd_start, unsigned long addr, unsigned long end,
 				spin_unlock(&init_mm.page_table_lock);
 				pages++;
 			} else {
-				/* If here, we are freeing vmemmap pages. */
-				memset((void *)addr, PAGE_INUSE, next - addr);
-
-				page_addr = page_address(pmd_page(*pmd));
-				if (!memchr_inv(page_addr, PAGE_INUSE,
-						PMD_SIZE)) {
+				/*
+				 * Free the PMD if the whole range is unused.
+				 */
+				if (vmemmap_unuse_sub_pmd(addr, next)) {
 					free_hugepage_table(pmd_page(*pmd),
 							    altmap);
 
@@ -1088,10 +1150,10 @@ remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end,
 				pages++;
 			} else {
 				/* If here, we are freeing vmemmap pages. */
-				memset((void *)addr, PAGE_INUSE, next - addr);
+				memset((void *)addr, PAGE_UNUSED, next - addr);
 
 				page_addr = page_address(pud_page(*pud));
-				if (!memchr_inv(page_addr, PAGE_INUSE,
+				if (!memchr_inv(page_addr, PAGE_UNUSED,
 						PUD_SIZE)) {
 					free_pagetable(pud_page(*pud),
 						       get_order(PUD_SIZE));
@@ -1520,11 +1582,16 @@ static int __meminit vmemmap_populate_hugepages(unsigned long start,
 
 				addr_end = addr + PMD_SIZE;
 				p_end = p + PMD_SIZE;
+
+				if (!IS_ALIGNED(addr, PMD_SIZE) ||
+				    !IS_ALIGNED(next, PMD_SIZE))
+					vmemmap_use_new_sub_pmd(addr, next);
 				continue;
 			} else if (altmap)
 				return -ENOMEM; /* no fallback */
 		} else if (pmd_large(*pmd)) {
 			vmemmap_verify((pte_t *)pmd, node, addr, next);
+			vmemmap_use_sub_pmd(addr, next);
 			continue;
 		}
 		if (vmemmap_populate_basepages(addr, next, node, NULL))
Signature · 1 line
2.26.2
#4 lore raw

Re: [PATCH 2/2] x86/vmemmap: Handle unpopulated sub-pmd ranges

arch/x86/mm/init_64.c remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end, · line 1159

Quoted 13 lines
arch/x86/mm/init_64.c +2 -2
@@ -1088,10 +1150,10 @@ remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end,
  				pages++;
  			} else {
  				/* If here, we are freeing vmemmap pages. */
-				memset((void *)addr, PAGE_INUSE, next - addr);
+				memset((void *)addr, PAGE_UNUSED, next - addr);
 
  				page_addr = page_address(pud_page(*pud));
-				if (!memchr_inv(page_addr, PAGE_INUSE,
+				if (!memchr_inv(page_addr, PAGE_UNUSED,
  						PUD_SIZE)) {
  					free_pagetable(pud_page(*pud),
  						       get_order(PUD_SIZE));

I'm sorry to bother you again, but isn't that dead code as well?

How do we ever end up using 1GB pages for the vmemmap? At least not via vmemmap_populate() - so I guess never? There are not many occurrences of "PUD_SIZE" in the file after all ...

I think we can simplify that code.

Signature · 3 lines
Thanks,

David / dhildenb
#5 lore raw

Re: [PATCH 2/2] x86/vmemmap: Handle unpopulated sub-pmd ranges

Quoted 6 lines · David Hildenbrand
Quoted 6 lines
arch/x86/mm/init_64.c +1 -1
@@ -1088,10 +1150,10 @@ remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end,
  				pages++;
  			} else {
  				/* If here, we are freeing vmemmap pages. */
-				memset((void *)addr, PAGE_INUSE, next - addr);
+				memset((void *)addr, PAGE_UNUSED, next - addr);
Quoted 8 lines

arch/x86/mm/init_64.c remove_pud_table(pud_t *pud_start, unsigned long addr, unsigned long end, · line 1159

Quoted 6 lines
arch/x86/mm/init_64.c +1 -1
  				page_addr = page_address(pud_page(*pud));
-				if (!memchr_inv(page_addr, PAGE_INUSE,
+				if (!memchr_inv(page_addr, PAGE_UNUSED,
  						PUD_SIZE)) {
  					free_pagetable(pud_page(*pud),
  						       get_order(PUD_SIZE));

I'm sorry to bother you again, but isn't that dead code as well?

Heh, I spotted that earlier, but I did not think much of it honestly.

All this was introduced by:

commit ae9aae9eda2db71bf4b592f15618b0160eb07731
Author: Wen Congyang <[email protected]>
Date: Fri Feb 22 16:33:04 2013 -0800

memory-hotplug: common APIs to support page tables hot-remove

Quoted 3 lines

How do we ever end up using 1GB pages for the vmemmap? At least not via vmemmap_populate() - so I guess never? There are not many occurrences of "PUD_SIZE" in the file after all ...

AFAICT, we don't. The largest we populate for vmemmap is 2MB. I see init_memory_mapping can use 1G, but that should not affect us.

I guess that the vmemmap handling for 1GB can go as well. I will update the patchset.

Signature · 2 lines
Oscar Salvador
SUSE L3
#6 lore raw

Re: [PATCH 2/2] x86/vmemmap: Handle unpopulated sub-pmd ranges

In reply to Oscar Salvador
Thank you for the patch! Yet something to improve:

Hi Oscar,

Thank you for the patch! Yet something to improve:

[auto build test ERROR on tip/x86/mm]
[also build test ERROR on hnaz-linux-mm/master v5.11-rc6 next-20210125] [cannot apply to luto/next]
[If your patch is applied to the wrong git tree, kindly drop us a note. And when submitting patch, we suggest to use '--base' as documented in
https://git-scm.com/docs/git-format-patch]

url:    https://github.com/0day-ci/linux/commits/Oscar-Salvador/Cleanup-and-fixups-for-vmemmap-handling/20210202-192636
base:   https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git 167dcfc08b0b1f964ea95d410aa496fd78adf475
config: x86_64-randconfig-r004-20210202 (attached as .config)
compiler: clang version 13.0.0 (https://github.com/llvm/llvm-project 275c6af7d7f1ed63a03d05b4484413e447133269)

reproduce (this is a W=1 build):
wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross
# install x86_64 cross compiling tool for clang build
# apt-get install binutils-x86-64-linux-gnu
# https://github.com/0day-ci/linux/commit/2995155f4651bbb8c0d5f2e58e6e77321c5a889a
git remote add linux-review https://github.com/0day-ci/linux
git fetch --no-tags linux-review Oscar-Salvador/Cleanup-and-fixups-for-vmemmap-handling/20210202-192636 git checkout 2995155f4651bbb8c0d5f2e58e6e77321c5a889a
# save the attached .config to linux build tree
COMPILER_INSTALL_PATH=$HOME/0day COMPILER=clang make.cross ARCH=x86_64

If you fix the issue, kindly add following tag as appropriate

All errors (new ones prefixed by >>):

>> arch/x86/mm/init_64.c:1588:6: error: implicit declaration of function 'vmemmap_use_new_sub_pmd' [-Werror,-Wimplicit-function-declaration]
                                           vmemmap_use_new_sub_pmd(addr, next);
                                           ^
>> arch/x86/mm/init_64.c:1594:4: error: implicit declaration of function 'vmemmap_use_sub_pmd' [-Werror,-Wimplicit-function-declaration]
                           vmemmap_use_sub_pmd(addr, next);
                           ^

2 errors generated.

vim +/vmemmap_use_new_sub_pmd +1588 arch/x86/mm/init_64.c

1535
1536 static int __meminit vmemmap_populate_hugepages(unsigned long start,
1537 unsigned long end, int node, struct vmem_altmap *altmap)
1538 {
1539 unsigned long addr;
1540 unsigned long next;
1541 pgd_t *pgd;
1542 p4d_t *p4d;
1543 pud_t *pud;
1544 pmd_t *pmd;
1545
1546 for (addr = start; addr < end; addr = next) {
1547 next = pmd_addr_end(addr, end);
1548
1549 pgd = vmemmap_pgd_populate(addr, node);
1550 if (!pgd)
1551 return -ENOMEM;
1552
1553 p4d = vmemmap_p4d_populate(pgd, addr, node);
1554 if (!p4d)
1555 return -ENOMEM;
1556
1557 pud = vmemmap_pud_populate(p4d, addr, node);
1558 if (!pud)
1559 return -ENOMEM;
1560
1561 pmd = pmd_offset(pud, addr);
1562 if (pmd_none(*pmd)) {
1563 void *p;
1564
1565 p = vmemmap_alloc_block_buf(PMD_SIZE, node, altmap);
1566 if (p) {
1567 pte_t entry;
1568
1569 entry = pfn_pte(__pa(p) >> PAGE_SHIFT,
1570 PAGE_KERNEL_LARGE);
1571 set_pmd(pmd, __pmd(pte_val(entry)));
1572
1573 /* check to see if we have contiguous blocks */
1574 if (p_end != p || node_start != node) {
1575 if (p_start)
1576 pr_debug(" [%lx-%lx] PMD -> [%p-%p] on node %d\n",
1577 addr_start, addr_end-1, p_start, p_end-1, node_start);
1578 addr_start = addr;
1579 node_start = node;
1580 p_start = p;
1581 }
1582
1583 addr_end = addr + PMD_SIZE;
1584 p_end = p + PMD_SIZE;
1585
1586 if (!IS_ALIGNED(addr, PMD_SIZE) ||
1587 !IS_ALIGNED(next, PMD_SIZE))

> 1588						vmemmap_use_new_sub_pmd(addr, next);
  1589					continue;
  1590				} else if (altmap)
  1591					return -ENOMEM; /* no fallback */
  1592			} else if (pmd_large(*pmd)) {
  1593				vmemmap_verify((pte_t *)pmd, node, addr, next);
> 1594				vmemmap_use_sub_pmd(addr, next);

1595 continue;
1596 }
1597 if (vmemmap_populate_basepages(addr, next, node, NULL))
1598 return -ENOMEM;
1599 }
1600 return 0;
1601 }
1602

---
0-DAY CI Kernel Test Service, Intel Corporation
https://lists.01.org/hyperkitty/list/kbuild-all(a)lists.01.org