Skip to contentSkip to conversation

mm/hmm: Prevent infinite loop in hmm_range_fault during EBUSY retries

PATCH In review sooraj 4 participants 4 messages Jan 27, 2025 – Jan 28, 2025

Subsystems (2)
lore
Share & tools
lore

sooraj

author
#0 lore raw

PATCH 1/1 mm/hmm: Prevent infinite loop in hmm_range_fault during EBUSY retries

When hmm_vma_walk_test() skips a VMA (e.g., unsupported VM_IO/PFNMAP range), it must update hmm_vma_walk->last to the end of the skipped VMA. Failing to do so causes hmm_range_fault() to restart from the same address during -EBUSY retries, reprocessing the skipped VMA indefinitely. This results in an infinite loop if the VMA remains non-processable.

Update hmm_vma_walk->last to the VMA's end address in hmm_vma_walk_test() when skipping the range. This ensures subsequent iterations resume correctly after the skipped VMA, preventing infinite retry loops.

---
 mm/hmm.c | 2 ++
 1 file changed, 2 insertions(+)
mm/hmm.c +2 -0
index 7e0229ae4a5a..29e3678fede5 100644
@@ -547,6 +547,8 @@ static int hmm_vma_walk_test(unsigned long start, unsigned long end,
 
 	hmm_pfns_fill(start, end, range, HMM_PFN_ERROR);
 
+	/* Update last to the end of the skipped VMA to prevent reprocessing */
+	hmm_vma_walk->last = end;
 	/* Skip this vma and continue processing the next vma. */
 	return 1;
 }
Signature · 1 line
2.45.2
maintainer
#1 lore raw
Quoted 10 lines · On Tue, 28 Jan 2025 01:34:22 -0500 sooraj <[email protected]> wrote:

When hmm_vma_walk_test() skips a VMA (e.g., unsupported VM_IO/PFNMAP range), it must update hmm_vma_walk->last to the end of the skipped VMA. Failing to do so causes hmm_range_fault() to restart from the same address during -EBUSY retries, reprocessing the skipped VMA indefinitely. This results in an infinite loop if the VMA remains non-processable.

Update hmm_vma_walk->last to the VMA's end address in hmm_vma_walk_test() when skipping the range. This ensures subsequent iterations resume correctly after the skipped VMA, preventing infinite retry loops.

Well that's unfortunate. This code seems quite old - can you tell us what your userspace is doing to trigger this behavior?

mm/hmm.c static int hmm_vma_walk_test(unsigned long start, unsigned long end, · line 554

Quoted 11 lines
mm/hmm.c +2 -0
@@ -547,6 +547,8 @@ static int hmm_vma_walk_test(unsigned long start, unsigned long end,
 
 	hmm_pfns_fill(start, end, range, HMM_PFN_ERROR);
 
+	/* Update last to the end of the skipped VMA to prevent reprocessing */
+	hmm_vma_walk->last = end;
 	/* Skip this vma and continue processing the next vma. */
 	return 1;
 }

This appears to deserve a cc:stable, but I suspect the bug is so old that a Fixes: won't be needed.

#2 lore raw
In reply to sooraj
Quoted 9 lines · sooraj

When hmm_vma_walk_test() skips a VMA (e.g., unsupported VM_IO/PFNMAP range), it must update hmm_vma_walk->last to the end of the skipped VMA. Failing to do so causes hmm_range_fault() to restart from the same address during -EBUSY retries, reprocessing the skipped VMA indefinitely. This results in an infinite loop if the VMA remains non-processable.

Update hmm_vma_walk->last to the VMA's end address in hmm_vma_walk_test() when skipping the range. This ensures subsequent iterations resume correctly after the skipped VMA, preventing infinite retry loops.

I might be missing something here but I don't understand how this causes an infinite loop. If we skip the VMA (ie. hmm_vma_walk_test() returns 1) and hmm_range_fault() subsequently returns -EBUSY it's true that we will reprocess the same non-processable VMA. But a non-processable VMA won't return -EBUSY and therefore won't cause an infinite loop in hmm_range_fault() - it will just fill out the pfns (which is redundant) and continue on to the next VMA.

So it seems this just prevents ueslessly filling out pfns again rather than an infinite loop. What have I missed?

- Alistair

Quoted 22 lines
---
 mm/hmm.c | 2 ++
 1 file changed, 2 insertions(+)
mm/hmm.c +2 -0
index 7e0229ae4a5a..29e3678fede5 100644
@@ -547,6 +547,8 @@ static int hmm_vma_walk_test(unsigned long start, unsigned long end,
 
 	hmm_pfns_fill(start, end, range, HMM_PFN_ERROR);
 
+	/* Update last to the end of the skipped VMA to prevent reprocessing */
+	hmm_vma_walk->last = end;
 	/* Skip this vma and continue processing the next vma. */
 	return 1;
 }
Signature · 1 line
2.45.2
maintainer
#3 lore raw
In reply to sooraj
Quoted 5 lines · sooraj

When hmm_vma_walk_test() skips a VMA (e.g., unsupported VM_IO/PFNMAP range), it must update hmm_vma_walk->last to the end of the skipped VMA. Failing to do so causes hmm_range_fault() to restart from the same address during -EBUSY retries, reprocessing the skipped VMA indefinitely. This results in an infinite loop if the VMA remains non-processable.

This needs alot more explanation.

The only place last is read is here:

		ret = walk_page_range(mm, hmm_vma_walk.last, range->end,
				      &hmm_walk_ops, &hmm_vma_walk);
		/*
		 * When -EBUSY is returned the loop restarts with
		 * hmm_vma_walk.last set to an address that has not been stored
		 * in pfns. All entries < last in the pfn array are set to their
		 * output, and all >= are still at their input values.
		 */
	} while (ret == -EBUSY);

The idea being that when a lower function returns -EBUSY last tells this loop where to restart the iteration.

mm/hmm.c static int hmm_vma_walk_test(unsigned long start, unsigned long end, · line 551

Quoted 6 lines
mm/hmm.c +2 -0
@@ -547,6 +547,8 @@ static int hmm_vma_walk_test(unsigned long start, unsigned long end,
 
 	hmm_pfns_fill(start, end, range, HMM_PFN_ERROR);
 
+	/* Update last to the end of the skipped VMA to prevent reprocessing */
+	hmm_vma_walk->last = end;

This does not return EBUSY, so it does not need to set last.

I just checked against and all the places that return EBUSY, do set last:

static int hmm_vma_fault(unsigned long addr, unsigned long end,
unsigned int required_fault, struct mm_walk *walk)

{
	hmm_vma_walk->last = addr;
..
	return -EBUSY;

static int hmm_vma_handle_pte(struct mm_walk *walk, unsigned long addr,
unsigned long end, pmd_t *pmdp, pte_t *ptep, unsigned long *hmm_pfn)
{
...

			hmm_vma_walk->last = addr;
			migration_entry_wait(walk->mm, pmdp, addr);
			return -EBUSY;

static int hmm_vma_walk_pmd(pmd_t *pmdp,
unsigned long start,
unsigned long end,
struct mm_walk *walk)
{
...

			hmm_vma_walk->last = addr;
			pmd_migration_entry_wait(walk->mm, pmdp);
			return -EBUSY;
		}

Thats it. If an -EBUSY case has been missed it should be fixed at the -EBUSY return point, anything else is wrong.

If you are seeing an infinite loop that this actually fixes please explain in the commit message exactly how it arises, and where the -EBUSY came from to miss the last update.

If there is a loop due to a missed -EBUSY, then this is not the correct way to fix it.

Jason