arm64: Enable vmalloc-huge with ptdump

arm64 disables vmalloc-huge when kernel page table dumping is enabled, because an intermediate table may be removed, potentially causing the ptdump code to dereference an invalid address. We want to be able to analyze block vs page mappings for kernel mappings with ptdump, so to enable vmalloc-huge with ptdump, synchronize between page table removal in pmd_free_pte_page()/pud_free_pmd_page() and ptdump pagetable walking. We use mmap_read_lock and not write lock because we don't need to synchronize between two different vm_structs; two vmalloc objects running this same code path will point to different page tables, hence there is no race.

For pud_free_pmd_page(), we isolate the PMD table to avoid taking the lock 512 times again via pmd_free_pte_page().

We implement the locking mechanism using static keys, since the chance of a race is very small. Observe that the synchronization is needed to avoid the following race:

CPU1 CPU2
take reference of PMD table
pud_clear()
pte_free_kernel()
walk freed PMD table

and similar race between pmd_free_pte_page and ptdump_walk_pgd.

Therefore, there are two cases: if ptdump sees the cleared PUD, then we are safe. If not, then the patched-in read and write locks help us avoid the race.

To implement the mechanism, we need the static key access from mmu.c and ptdump.c. Note that in case !CONFIG_PTDUMP_DEBUGFS, ptdump.o won't be a target in the Makefile, therefore we cannot initialize the key there, as is being done, for example, in the static key implementation of hugetlb-vmemmap. Therefore, include asm/cpufeature.h, which includes the jump_label mechanism. Declare the key there and define the key to false in mmu.c.

No issues were observed with mm-selftests. No issues were observed while parallelly running test_vmalloc.sh and dumping the kernel pagetable through sysfs in a loop.

Reviewed-by Ryan Roberts <[email protected]>

Signed-off-by Dev Jain <[email protected]>

---

arm64 disables vmalloc-huge when kernel page table dumping is enabled, because an intermediate table may be removed, potentially causing the ptdump code to dereference an invalid address. We want to be able to

analyze block vs page mappings for kernel mappings with ptdump, so to enable vmalloc-huge with ptdump, synchronize between page table removal in pmd_free_pte_page()/pud_free_pmd_page() and ptdump pagetable walking. We

use mmap_read_lock and not write lock because we don't need to synchronize between two different vm_structs; two vmalloc objects running this same code path will point to different page tables, hence there is no race.

For pud_free_pmd_page(), we isolate the PMD table to avoid taking the lock 512 times again via pmd_free_pte_page().

We implement the locking mechanism using static keys, since the chance of a race is very small. Observe that the synchronization is needed to avoid the following race:

CPU1 CPU2
take reference of PMD table
pud_clear()
pte_free_kernel()
walk freed PMD table

and similar race between pmd_free_pte_page and ptdump_walk_pgd.

Therefore, there are two cases: if ptdump sees the cleared PUD, then we are safe. If not, then the patched-in read and write locks help us avoid the race.

To implement the mechanism, we need the static key access from mmu.c and ptdump.c. Note that in case !CONFIG_PTDUMP_DEBUGFS, ptdump.o won't be a target in the Makefile, therefore we cannot initialize the key there, as is being done, for example, in the static key implementation of hugetlb-vmemmap. Therefore, include asm/cpufeature.h, which includes the jump_label mechanism. Declare the key there and define the key to false in mmu.c.

No issues were observed with mm-selftests. No issues were observed while parallelly running test_vmalloc.sh and dumping the kernel pagetable through sysfs in a loop.

Reviewed-by Ryan Roberts <[email protected]>

Signed-off-by Dev Jain <[email protected]>

---
v3->v4:
- Lock-unlock immediately
- Simplify includes

v2->v3:
- Use static key mechanism

v1->v2:
- Take lock only when CONFIG_PTDUMP_DEBUGFS is on
- In case of pud_free_pmd_page(), isolate the PMD table to avoid taking the lock 512 times again via pmd_free_pte_page()

 arch/arm64/include/asm/ptdump.h |  2 ++
 arch/arm64/mm/mmu.c             | 44 ++++++++++++++++++++++++++++++---
 arch/arm64/mm/ptdump.c          |  5 +++-
 3 files changed, 46 insertions(+), 5 deletions(-)

arch/arm64/include/asm/ptdump.h +2 -0

index fded5358641f..5b331f2a7be1 100644
@@ -7,6 +7,8 @@
 
 #include <linux/ptdump.h>
 
+DECLARE_STATIC_KEY_FALSE(ptdump_lock_key);
+
 #ifdef CONFIG_PTDUMP
 
 #include <linux/mm_types.h>

arch/arm64/mm/mmu.c +0 -0

index 00ab1d648db6..9d3be249047c 100644

arch/arm64/mm/mmu.c +21 -0

+	if (static_branch_unlikely(&ptdump_lock_key) && lock) {
+		mmap_read_lock(&init_mm);
+		mmap_read_unlock(&init_mm);
+	}
+
 	pte_free_kernel(NULL, table);
 	return 1;
 }
 
+int pmd_free_pte_page(pmd_t *pmdp, unsigned long addr)
+{
+	return __pmd_free_pte_page(pmdp, addr, true);
+}
+
 int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
 {
 	pmd_t *table;

@@ -1301,16 +1314,39 @@ int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
 	}
 
 	table = pmd_offset(pudp, addr);
+	/*
+	 * Isolate the PMD table; in case of race with ptdump, this helps
+	 * us to avoid taking the lock in __pmd_free_pte_page().
+	 *
+	 * Static key logic:
+	 *
+	 * Case 1: If ptdump does static_branch_enable(), and after that we
+	 * execute the if block, then this patches in the read lock, ptdump has
+	 * the write lock patched in, therefore ptdump will never read from
+	 * a potentially freed PMD table.
+	 *

arch/arm64/mm/mmu.c +1 -3

 	pmdp = table;
 	next = addr;
 	end = addr + PUD_SIZE;
 	do {
 		if (pmd_present(pmdp_get(pmdp)))
-			pmd_free_pte_page(pmdp, next);
+			__pmd_free_pte_page(pmdp, next, false);
 	} while (pmdp++, next += PMD_SIZE, next != end);
 
-	pud_clear(pudp);
-	__flush_tlb_kernel_pgtable(addr);
 	pmd_free(NULL, table);
 	return 1;
 }

arch/arm64/mm/ptdump.c +0 -0

index 421a5de806c6..41c9ea61813b 100644

arch/arm64/mm/ptdump.c +2 -0

 #define pt_dump_seq_printf(m, fmt, args...)	\
 ({						\
 	if (m)					\

@@ -311,7 +310,9 @@ void ptdump_walk(struct seq_file *s, struct ptdump_info *info)
 		}
 	};
 
+	static_branch_enable(&ptdump_lock_key);
 	ptdump_walk_pgd(&st.ptdump, info->mm, NULL);
+	static_branch_disable(&ptdump_lock_key);
 }
 
 static void __init ptdump_initialize(void)

Quoted 3 lines · Dev Jain

arm64 disables vmalloc-huge when kernel page table dumping is enabled, because an intermediate table may be removed, potentially causing the ptdump code to dereference an invalid address. We want to be able to

Please keep this paragraph separate explaining the current scenario. It might be worth adding bit more details such CONFIG_PTDUMP_DEBUGFS config dependency in vmalloc()'s arch call backs arch_vmap_pud|pmd_supported().

Quoted 3 lines

analyze block vs page mappings for kernel mappings with ptdump, so to enable vmalloc-huge with ptdump, synchronize between page table removal in pmd_free_pte_page()/pud_free_pmd_page() and ptdump pagetable walking. We

Please keep this paragraph separate explaining what is the benefit of enabling ptdump with huge vmalloc() mappings. This will establish the required motivation to change the current scenario.

Quoted 3 lines

use mmap_read_lock and not write lock because we don't need to synchronize between two different vm_structs; two vmalloc objects running this same code path will point to different page tables, hence there is no race.

Different page tables ? OR different areas in same page table i.e init_mm.

Quoted 3 lines

For pud_free_pmd_page(), we isolate the PMD table to avoid taking the lock 512 times again via pmd_free_pte_page().

This talks about subsequent optimization without establishing the base solution first.

Quoted 17 lines

We implement the locking mechanism using static keys, since the chance of a race is very small. Observe that the synchronization is needed to avoid the following race:

CPU1 CPU2
take reference of PMD table
pud_clear()
pte_free_kernel()
walk freed PMD table

and similar race between pmd_free_pte_page and ptdump_walk_pgd.

Therefore, there are two cases: if ptdump sees the cleared PUD, then we are safe. If not, then the patched-in read and write locks help us avoid the race.

Quoted 7 lines

To implement the mechanism, we need the static key access from mmu.c and ptdump.c. Note that in case !CONFIG_PTDUMP_DEBUGFS, ptdump.o won't be a target in the Makefile, therefore we cannot initialize the key there, as is being done, for example, in the static key implementation of hugetlb-vmemmap. Therefore, include asm/cpufeature.h, which includes the jump_label mechanism. Declare the key there and define the key to false in mmu.c.

Above write up is convoluted and not very clear. Please rewrite the solution description while avoiding too much implementation and code file details that can be derived from the patch itself.

Quoted 4 lines

No issues were observed with mm-selftests. No issues were observed while parallelly running test_vmalloc.sh and dumping the kernel pagetable through sysfs in a loop.

Please provide some more details about test_vmallo.sh which seems to be a custom script.

Quoted 42 lines

Reviewed-by Ryan Roberts <[email protected]>

Signed-off-by Dev Jain <[email protected]>

---
v3->v4:
- Lock-unlock immediately
- Simplify includes

v2->v3:
- Use static key mechanism

v1->v2:
- Take lock only when CONFIG_PTDUMP_DEBUGFS is on
- In case of pud_free_pmd_page(), isolate the PMD table to avoid taking the lock 512 times again via pmd_free_pte_page()

  arch/arm64/include/asm/ptdump.h |  2 ++
  arch/arm64/mm/mmu.c             | 44 ++++++++++++++++++++++++++++++---
  arch/arm64/mm/ptdump.c          |  5 +++-
  3 files changed, 46 insertions(+), 5 deletions(-)

arch/arm64/include/asm/ptdump.h +2 -0

index fded5358641f..5b331f2a7be1 100644
@@ -7,6 +7,8 @@
 
  #include <linux/ptdump.h>
 
+DECLARE_STATIC_KEY_FALSE(ptdump_lock_key);
+
  #ifdef CONFIG_PTDUMP
 
  #include <linux/mm_types.h>

arch/arm64/mm/mmu.c +0 -0

index 00ab1d648db6..9d3be249047c 100644
@@ -46,6 +46,8 @@
  #define NO_CONT_MAPPINGS	BIT(1)
  #define NO_EXEC_MAPPINGS	BIT(2)	/* assumes FEAT_HPDS is not used */
 

arch/arm64/mm/mmu.c line 50

Quoted 2 lines

arch/arm64/mm/mmu.c +2 -0

+DEFINE_STATIC_KEY_FALSE(ptdump_lock_key);
+ptdump_lock_key sounds too generic even though this locking

requirement is currently arm64 platform specific. Otherwise this might just appear as a locking from the generic ptdump Could this be renamed as arm64_ptdump_lock_key instead ?

Quoted 8 lines

arch/arm64/mm/mmu.c +0 -0

  enum pgtable_type {
  	TABLE_PTE,
  	TABLE_PMD,

@@ -1267,7 +1269,7 @@ int pmd_clear_huge(pmd_t *pmdp)
  	return 1;
  }
 

arch/arm64/mm/mmu.c int pmd_clear_huge(pmd_t *pmdp) · line 1273

Quoted 3 lines

arch/arm64/mm/mmu.c +1 -1

-int pmd_free_pte_page(pmd_t *pmdp, unsigned long addr)
+static int __pmd_free_pte_page(pmd_t *pmdp, unsigned long addr, bool lock)
  {

'lock' argument needs explanation here in a comment.

Quoted 7 lines

arch/arm64/mm/mmu.c +0 -0

  	pte_t *table;
  	pmd_t pmd;

@@ -1279,13 +1281,24 @@ int pmd_free_pte_page(pmd_t *pmdp, unsigned long addr)
  		return 1;
  	}
 

arch/arm64/mm/mmu.c int pmd_free_pte_page(pmd_t *pmdp, unsigned long addr) · line 1287

Quoted 4 lines

arch/arm64/mm/mmu.c +1 -0

+	/* See comment in pud_free_pmd_page for static key logic */
  	table = pte_offset_kernel(pmdp, addr);
  	pmd_clear(pmdp);
  	__flush_tlb_kernel_pgtable(addr);

Or here ?

Quoted 33 lines

arch/arm64/mm/mmu.c +21 -0

+	if (static_branch_unlikely(&ptdump_lock_key) && lock) {
+		mmap_read_lock(&init_mm);
+		mmap_read_unlock(&init_mm);
+	}
+
  	pte_free_kernel(NULL, table);
  	return 1;
  }
 
+int pmd_free_pte_page(pmd_t *pmdp, unsigned long addr)
+{
+	return __pmd_free_pte_page(pmdp, addr, true);
+}
+
  int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
  {
  	pmd_t *table;

@@ -1301,16 +1314,39 @@ int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
  	}
 
  	table = pmd_offset(pudp, addr);
+	/*
+	 * Isolate the PMD table; in case of race with ptdump, this helps
+	 * us to avoid taking the lock in __pmd_free_pte_page().
+	 *
+	 * Static key logic:
+	 *
+	 * Case 1: If ptdump does static_branch_enable(), and after that we
+	 * execute the if block, then this patches in the read lock, ptdump has
+	 * the write lock patched in, therefore ptdump will never read from
+	 * a potentially freed PMD table.
+	 *

arch/arm64/mm/mmu.c int pud_free_pmd_page(pud_t *pudp, unsigned long addr) · line 1337

Quoted 10 lines

arch/arm64/mm/mmu.c +10 -0

+	 * Case 2: If the if block starts executing before ptdump's
+	 * static_branch_enable(), then no locking synchronization
+	 * will be done. However, pud_clear() + the dsb() in
+	 * __flush_tlb_kernel_pgtable will ensure that ptdump observes an
+	 * empty PUD. Thus, it will never walk over a potentially freed
+	 * PMD table.
+	 */
+	pud_clear(pudp);
+	__flush_tlb_kernel_pgtable(addr);
+	if (static_branch_unlikely(&ptdump_lock_key)) {

		/* case 1 - ptdump comes first
		 *
		 * mmap_read_lock() here will wait on mmap_write_lock()
	 	 * taken in generic ptdump until it has been released.
		 */

arch/arm64/mm/mmu.c int pmd_free_pte_page(pmd_t *pmdp, unsigned long addr) · line 1291

Quoted 3 lines

arch/arm64/mm/mmu.c +3 -0

+		mmap_read_lock(&init_mm);
+		mmap_read_unlock(&init_mm);
+	}

	} else {
		/* case 2 - pud_free_pmd_page() comes first
		 *
		 * pud_clear() and __flush_tlb_kernel_pgtable() are
		 * sufficient for ptdump to observe an empty PUD.
		 */	
	}> +

Quoted 22 lines

arch/arm64/mm/mmu.c +1 -3

  	pmdp = table;
  	next = addr;
  	end = addr + PUD_SIZE;
  	do {
  		if (pmd_present(pmdp_get(pmdp)))
-			pmd_free_pte_page(pmdp, next);
+			__pmd_free_pte_page(pmdp, next, false);
  	} while (pmdp++, next += PMD_SIZE, next != end);
 
-	pud_clear(pudp);
-	__flush_tlb_kernel_pgtable(addr);
  	pmd_free(NULL, table);
  	return 1;
  }

arch/arm64/mm/ptdump.c +0 -0

index 421a5de806c6..41c9ea61813b 100644
@@ -25,7 +25,6 @@
  #include <asm/pgtable-hwdef.h>
  #include <asm/ptdump.h>
 

arch/arm64/mm/ptdump.c line 28

Quoted 1 line

arch/arm64/mm/ptdump.c +0 -1

-

Stray change. Please drop it.

Quoted 18 lines

arch/arm64/mm/ptdump.c +2 -0

  #define pt_dump_seq_printf(m, fmt, args...)	\
  ({						\
  	if (m)					\

@@ -311,7 +310,9 @@ void ptdump_walk(struct seq_file *s, struct ptdump_info *info)
  		}
  	};
 
+	static_branch_enable(&ptdump_lock_key);
  	ptdump_walk_pgd(&st.ptdump, info->mm, NULL);
+	static_branch_disable(&ptdump_lock_key);
  }
 
  static void __init ptdump_initialize(void)

@@ -353,7 +354,9 @@ bool ptdump_check_wx(void)
  		}
  	};
 

arch/arm64/mm/ptdump.c bool ptdump_check_wx(void)

Quoted 6 lines

arch/arm64/mm/ptdump.c +2 -0

+	static_branch_enable(&ptdump_lock_key);
  	ptdump_walk_pgd(&st.ptdump, &init_mm, NULL);
+	static_branch_disable(&ptdump_lock_key);
 
  	if (st.wx_pages || st.uxn_pages) {
  		pr_warn("Checked W+X mappings: FAILED, %lu W+X pages found, %lu non-UXN pages found\n",Although tempting but I guess encapsulating above ptdump_walk_pgd()

locking sequence inside a new helper arm64_ptdump_walk_pgd() while also explaining this locking might be worth ?

PUD/PMD vmalloc mappings need to be enabled via these arch callback changes via dropping the current CONFIG_PTDUMP_DEBUGFS dependency.

#define arch_vmap_pud_supported arch_vmap_pud_supported static inline bool arch_vmap_pud_supported(pgprot_t prot) {
return pud_sect_supported()
}

#define arch_vmap_pmd_supported arch_vmap_pmd_supported static inline bool arch_vmap_pmd_supported(pgprot_t prot) {
return return.
}

arch/arm64/mm/mmu.c +5 -0

@@ -1301,16 +1314,39 @@ int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
 	}
 
 	table = pmd_offset(pudp, addr);
+	/*
+	 * Isolate the PMD table; in case of race with ptdump, this helps
+	 * us to avoid taking the lock in __pmd_free_pte_page().
+	 *
+	 * Static key logic:

Quoted 9 lines · Dev Jain

arch/arm64/mm/mmu.c +5 -0

@@ -1301,16 +1314,39 @@ int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
  	}
 
  	table = pmd_offset(pudp, addr);
+	/*
+	 * Isolate the PMD table; in case of race with ptdump, this helps
+	 * us to avoid taking the lock in __pmd_free_pte_page().
+	 *
+	 * Static key logic:

arch/arm64/mm/mmu.c int pud_free_pmd_page(pud_t *pudp, unsigned long addr) · line 1331

Quoted 10 lines

arch/arm64/mm/mmu.c +10 -0

+	 *
+	 * Case 1: If ptdump does static_branch_enable(), and after that we
+	 * execute the if block, then this patches in the read lock, ptdump has
+	 * the write lock patched in, therefore ptdump will never read from
+	 * a potentially freed PMD table.
+	 *
+	 * Case 2: If the if block starts executing before ptdump's
+	 * static_branch_enable(), then no locking synchronization
+	 * will be done. However, pud_clear() + the dsb() in
+	 * __flush_tlb_kernel_pgtable will ensure that ptdump observes an

Statements like "observes" really need to be relative, not absolute. Like in "observes an empty PUD before/after ...".

arch/arm64/mm/mmu.c int pud_free_pmd_page(pud_t *pudp, unsigned long addr) · line 1340

Quoted 9 lines

arch/arm64/mm/mmu.c +9 -0

+	 * empty PUD. Thus, it will never walk over a potentially freed
+	 * PMD table.
+	 */
+	pud_clear(pudp);
+	__flush_tlb_kernel_pgtable(addr);
+	if (static_branch_unlikely(&ptdump_lock_key)) {
+		mmap_read_lock(&init_mm);
+		mmap_read_unlock(&init_mm);
+	}

This needs a formal model ;).

static_branch_enable() is called before the mmap_write_lock(), so even if the above observes the new branch, it may do the read_unlock() before the ptdump_walk_pgd() attempted the write lock. So your case 1 description is not entirely correct.

I don't get case 2. You want to ensure pud_clear() is observed by the ptdump code before the pmd_free(). The DSB in the TLB flushing code ensures some ordering between the pud_clear() and presumably something that the ptdump code can observe as well. Would that be the mmap semaphore? However, the read_lock would only be attempted if this code is seeing the static branch update, which is not guaranteed. I don't think it even matters since the lock may be released anyway before the write_lock in ptdump.

For example, you do a pud_clear() above, skip the whole static branch. The ptdump comes along on another CPU but does not observe the pud_clear() since there's no synchronisation. It goes ahead and dereferences it while this CPU does a pmd_free().

And I can't get my head around memory ordering, it doesn't look sound. static_branch_enable() I don't think has acquire semantics, at least not in relation to the actual branch update. The static_branch_unlikely() test, again, not sure what guarantees it has (I don't think it has any in relation to memory loads). Maybe you have worked it all out and is fine but it needs a better explanation and ideally some simple formal model. Show it's correct with a global variable first and then we can optimise with static branches.

Quoted 19 lines · Catalin Marinas

Quoted 18 lines · Dev Jain

arch/arm64/mm/mmu.c +15 -0

@@ -1301,16 +1314,39 @@ int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
  	}
  	table = pmd_offset(pudp, addr);
+	/*
+	 * Isolate the PMD table; in case of race with ptdump, this helps
+	 * us to avoid taking the lock in __pmd_free_pte_page().
+	 *
+	 * Static key logic:
+	 *
+	 * Case 1: If ptdump does static_branch_enable(), and after that we
+	 * execute the if block, then this patches in the read lock, ptdump has
+	 * the write lock patched in, therefore ptdump will never read from
+	 * a potentially freed PMD table.
+	 *
+	 * Case 2: If the if block starts executing before ptdump's
+	 * static_branch_enable(), then no locking synchronization
+	 * will be done. However, pud_clear() + the dsb() in
+	 * __flush_tlb_kernel_pgtable will ensure that ptdump observes an

Quoted 9 lines

I don't get case 2. You want to ensure pud_clear() is observed by the ptdump code before the pmd_free(). The DSB in the TLB flushing code ensures some ordering between the pud_clear() and presumably something that the ptdump code can observe as well. Would that be the mmap semaphore? However, the read_lock would only be attempted if this code is seeing the static branch update, which is not guaranteed. I don't think it even matters since the lock may be released anyway before the write_lock in ptdump.

Quoted 4 lines

For example, you do a pud_clear() above, skip the whole static branch. The ptdump comes along on another CPU but does not observe the pud_clear() since there's no synchronisation. It goes ahead and dereferences it while this CPU does a pmd_free().

The objective is: ptdump must not dereference a freed pagetable.
So for your example, if the static branch is not observed by pud_free_pmd_page, this means that ptdump will take the write lock after the execution of flush_tlb_kernel_pagetable completes (for if ptdump takes the write lock before execution of flush_tlb_kernel_pagetable completes, we have executed static_branch_enable(), contradiction).

Quoted 8 lines

And I can't get my head around memory ordering, it doesn't look sound. static_branch_enable() I don't think has acquire semantics, at least not in relation to the actual branch update. The static_branch_unlikely() test, again, not sure what guarantees it has (I don't think it has any in relation to memory loads). Maybe you have worked it all out and is fine but it needs a better explanation and ideally some simple formal model. Show it's correct with a global variable first and then we can optimise with static branches.

What do you suggest? As in what do you mean by showing its correct with a global variable first...and, for the formal model thingy, do you want mathematical rigor similar to what you explain in [1] :P, because unfortunately (and quite embarrassingly) I didn't have formal methods in college : )

Quoted 20 lines · Dev Jain

Quoted 19 lines · Catalin Marinas

Quoted 18 lines · Dev Jain

arch/arm64/mm/mmu.c +15 -0

@@ -1301,16 +1314,39 @@ int pud_free_pmd_page(pud_t *pudp, unsigned long addr)
   	}
   	table = pmd_offset(pudp, addr);
+	/*
+	 * Isolate the PMD table; in case of race with ptdump, this helps
+	 * us to avoid taking the lock in __pmd_free_pte_page().
+	 *
+	 * Static key logic:
+	 *
+	 * Case 1: If ptdump does static_branch_enable(), and after that we
+	 * execute the if block, then this patches in the read lock, ptdump has
+	 * the write lock patched in, therefore ptdump will never read from
+	 * a potentially freed PMD table.
+	 *
+	 * Case 2: If the if block starts executing before ptdump's
+	 * static_branch_enable(), then no locking synchronization
+	 * will be done. However, pud_clear() + the dsb() in
+	 * __flush_tlb_kernel_pgtable will ensure that ptdump observes an

[...]

Quoted 13 lines

Quoted 13 lines

I don't get case 2. You want to ensure pud_clear() is observed by the ptdump code before the pmd_free(). The DSB in the TLB flushing code ensures some ordering between the pud_clear() and presumably something that the ptdump code can observe as well. Would that be the mmap semaphore? However, the read_lock would only be attempted if this code is seeing the static branch update, which is not guaranteed. I don't think it even matters since the lock may be released anyway before the write_lock in ptdump.

For example, you do a pud_clear() above, skip the whole static branch. The ptdump comes along on another CPU but does not observe the pud_clear() since there's no synchronisation. It goes ahead and dereferences it while this CPU does a pmd_free().

Quoted 6 lines

The objective is: ptdump must not dereference a freed pagetable.
So for your example, if the static branch is not observed by pud_free_pmd_page, this means that ptdump will take the write lock after the execution of flush_tlb_kernel_pagetable completes (for if ptdump takes the write lock before execution of flush_tlb_kernel_pagetable completes, we have executed static_branch_enable(), contradiction).

I don't see why the write lock matters since pud_free_pmd_page() doesn't

take the read lock in the second scenario. What we need is acquire semantics after the static branch update on the ptdump path but we get it before we even attempt the write lock.

For simplicity, ignoring the observability of instruction writes and considering the static branch a variable, if pud_free_pmd_page() did not observe the static branch update, is the ptdump guaranteed to see the cleared pud subsequently?

With initial state pud=1 (non-zero), stb=0 (static branch):

P0 (pud_free_pmd_page) P1 (ptdump)

     W_pud=0			   W_stb=1
     DSB				   barrier/acq
     R_stb=0			   R_pud=?

The write to the static branch on P1 will be ordered after the read of the branch on P0, so the pud will be seen as 0. It's not even worth mentioning the semaphore here as the static branch update has enough barriers for cache flushing and kick_all_cpus_sync().

The other scenario is P0 (pud_free_pmd_page) observing the write to the static branch (that's case 1 in your comment). This doesn't say anything about whether P1 (ptdump) sees a clear or valid pud. What we do know is that P0 will try to acquire (and release) the lock. If P1 already acquired the write lock, P0 will wait and the state of the pud is irrelevant (no freeing). Similarly if P1 already completed by the time P0 takes the lock.

If P0 takes the lock first, the lock release guarantees that the pud_clear() is seen by the ptdump code _after_ it acquired the lock.

W.r.t. the visibility of the branch update vs pud access, the combinations of DSB+ISB (part of the TLB flushing) on P0 and cache maintenance to PoU together with kick_all_cpus_sync() on P1 should suffice.

I think the logic works (though I'm heavily jetlagged and writing from a plane) but the comments need to be improved. As described above, case 1 has two sub-cases depending on when P0 runs in relation to the write lock (before or during/after). And the write lock doesn't matter for case 2.

Quoted 8 lines

Quoted 8 lines

And I can't get my head around memory ordering, it doesn't look sound. static_branch_enable() I don't think has acquire semantics, at least not in relation to the actual branch update. The static_branch_unlikely() test, again, not sure what guarantees it has (I don't think it has any in relation to memory loads). Maybe you have worked it all out and is fine but it needs a better explanation and ideally some simple formal model. Show it's correct with a global variable first and then we can optimise with static branches.

Quoted 4 lines

What do you suggest? As in what do you mean by showing its correct with a global variable first...and, for the formal model thingy, do you want mathematical rigor similar to what you explain in [1] :P, because unfortunately (and quite embarrassingly) I didn't have formal methods in college : )

Neither did I ;) (mostly analogue electronics). I was thinking something like our cat/herd tools where you can write some simple assembly. It's a good investment if you want to give it a try.

Quoted 41 lines · Catalin Marinas

Quoted 26 lines · Dev Jain

Quoted 25 lines · Catalin Marinas

Quoted 24 lines · Dev Jain

arch/arm64/mm/mmu.c +2 -0

@@ -1301,16 +1314,39 @@ int pud_free_pmd_page(pud_t *pudp, 
 unsigned long addr)
        }
        table = pmd_offset(pudp, addr);
+    /*
+     * Isolate the PMD table; in case of race with ptdump, this 

helps
+     * us to avoid taking the lock in __pmd_free_pte_page(). +     *
+     * Static key logic:
+     *
+     * Case 1: If ptdump does static_branch_enable(), and after that we
+     * execute the if block, then this patches in the read lock, ptdump has
+     * the write lock patched in, therefore ptdump will never read from
+     * a potentially freed PMD table.
+     *
+     * Case 2: If the if block starts executing before ptdump's +     * static_branch_enable(), then no locking synchronization +     * will be done. However, pud_clear() + the dsb() in +     * __flush_tlb_kernel_pgtable will ensure that ptdump observes an

[...]

Quoted 13 lines

Quoted 13 lines

I don't get case 2. You want to ensure pud_clear() is observed by the ptdump code before the pmd_free(). The DSB in the TLB flushing code ensures some ordering between the pud_clear() and presumably something that the ptdump code can observe as well. Would that be the mmap semaphore? However, the read_lock would only be attempted if this code is seeing the static branch update, which is not guaranteed. I don't think it even matters since the lock may be released anyway before the write_lock in ptdump.

For example, you do a pud_clear() above, skip the whole static branch. The ptdump comes along on another CPU but does not observe the pud_clear() since there's no synchronisation. It goes ahead and dereferences it while this CPU does a pmd_free().

Quoted 7 lines

Quoted 6 lines

The objective is: ptdump must not dereference a freed pagetable.
So for your example, if the static branch is not observed by pud_free_pmd_page, this means that ptdump will take the write lock after the execution of flush_tlb_kernel_pagetable completes (for if ptdump takes the write lock before execution of flush_tlb_kernel_pagetable completes, we have executed static_branch_enable(), contradiction).

I don't see why the write lock matters since pud_free_pmd_page() doesn't

True.

Quoted 54 lines

take the read lock in the second scenario. What we need is acquire semantics after the static branch update on the ptdump path but we get it before we even attempt the write lock.

For simplicity, ignoring the observability of instruction writes and considering the static branch a variable, if pud_free_pmd_page() did not observe the static branch update, is the ptdump guaranteed to see the cleared pud subsequently?

With initial state pud=1 (non-zero), stb=0 (static branch):

P0 (pud_free_pmd_page)        P1 (ptdump)

     W_pud=0               W_stb=1
     DSB                   barrier/acq
     R_stb=0               R_pud=?

The write to the static branch on P1 will be ordered after the read of the branch on P0, so the pud will be seen as 0. It's not even worth mentioning the semaphore here as the static branch update has enough barriers for cache flushing and kick_all_cpus_sync().

The other scenario is P0 (pud_free_pmd_page) observing the write to the static branch (that's case 1 in your comment). This doesn't say anything about whether P1 (ptdump) sees a clear or valid pud. What we do know is that P0 will try to acquire (and release) the lock. If P1 already acquired the write lock, P0 will wait and the state of the pud is irrelevant (no freeing). Similarly if P1 already completed by the time P0 takes the lock.

If P0 takes the lock first, the lock release guarantees that the pud_clear() is seen by the ptdump code _after_ it acquired the lock.

W.r.t. the visibility of the branch update vs pud access, the combinations of DSB+ISB (part of the TLB flushing) on P0 and cache maintenance to PoU together with kick_all_cpus_sync() on P1 should suffice.

I think the logic works (though I'm heavily jetlagged and writing from a plane) but the comments need to be improved. As described above, case 1 has two sub-cases depending on when P0 runs in relation to the write lock (before or during/after). And the write lock doesn't matter for case 2.

Quoted 8 lines

Quoted 8 lines

And I can't get my head around memory ordering, it doesn't look sound. static_branch_enable() I don't think has acquire semantics, at least not in relation to the actual branch update. The static_branch_unlikely() test, again, not sure what guarantees it has (I don't think it has any in relation to memory loads). Maybe you have worked it all out and is fine but it needs a better explanation and ideally some simple formal model. Show it's correct with a global variable first and then we can optimise with static branches.

Quoted 7 lines

Quoted 4 lines

What do you suggest? As in what do you mean by showing its correct with a global variable first...and, for the formal model thingy, do you want mathematical rigor similar to what you explain in [1] :P, because unfortunately (and quite embarrassingly) I didn't have formal methods in college : )

Neither did I ;) (mostly analogue electronics). I was thinking something like our cat/herd tools where you can write some simple assembly. It's a good investment if you want to give it a try.

Will the following proof work -

Proof of correctness: The below diagram represents pud_free_pmd_page executing on P0 and ptdump executing on P1. Note that, we can ignore the situation when processes migrate to another CPU, since we will have extra barriers because of switch_to(), and all of the embedded barriers that are used in the reasoning of the proof below apply to the inner shareable domain, and therefore will be observed by all CPUs, therefore it suffices to prove for the case where pud_free_pmd_page executes completely on P0 and ptdump executes completely on P1.

Let t_i, 0 <= i <= 8 denote the *global* timestamp taken for the corresponding instruction to complete. Therefore from here on we do not need to use the term "observe" in a relative context. Let t_i' (t_i dash) denote the global timestamp for the corresponding instruction to start. That is, an instruction labelled with t_i implies that it started at t_i' and finished at t_i.

P0                P1:

W_PUD = 0: t0            x = 1: t2

if (x == 1) {: t7            write lock: t3
    read lock: t6        R_PUD = 1: t4
    read unlock: t8        write unlock: t5
}

Free PUD: t1

We need to prove that ptdump completely finishes before we free the PUD. Since write unlock has release semantics, if the write unlock finishes, it is guaranteed that ptdump has finished => it suffices to prove that t5 < t1'.

R_PUD = 1 => t4 < t0 .... (i)

Because of acquire semantics of down_write(rw_semaphore lock), t3 < t4' .... (ii)

(i) and (ii) (and t4' < t4) => t3 < t0 ... (iii)

ptdump is executed on a single kernel thread, which implies that the transition x = 1 -> x = 1 will never happen; that is, when static_branch_enable() is executed, then x was 0, which means that the call chain static_key_enable -> static_key_enable_cpuslocked -> jump_label_update -> jump_label_can_update/ arch_jump_label_transform_apply
-> kick_all_cpus_sync -> smp_mb -> __smp_mb -> dmb(ish) will always be followed.
The emission of dmb(ish) => t2 < t3 ... (iv)

(iii) and (iv) => t2 < t0, also, flush_tlb_kernel_pgtable -> dsb(ish) ensures that t0 < t7' => t2 < t7' => the static branch is observed by P0 always => t6 and t8 exist.

Now, t0 < t6' because of flush_tlb_kernel_pgtable; combining with (iii), this gives us t3 < t6' => the write lock is observed first => t5 < t6 (the read lock cannot be taken before the write unlock finishes) ...(v)

Release semantics of read unlock => t8 < t1' ...(vi) Also, trivially t6 < t8...(vii)

Combining v, vi and vii, t5 < t1'. Q.E.D

Quoted 1 line