Skip to contentSkip to conversation

fs/proc/task_mmu: avoid spurious seq_file warning on interrupted VMA lookup

PATCH In review Adrian Huang (Lenovo) 2 participants 2 messages Nov 18

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH v2 1/1 fs/proc/task_mmu: avoid spurious seq_file warning on interrupted VMA lookup

Commit 5631da56c9a8 ("fs/proc/task_mmu: read proc/pid/maps under per-vma lock") introduced the possibility of returning -EINTR from proc_get_vma() when a fatal signal interrupts mmap_read_lock_killable().

The issue can be reproduced using stress-ng's VMA stressor, which reads /proc/self/maps while concurrently modifying VMAs and periodically terminates child processes with SIGKILL. The reproduced command is:

  # stress-ng --vma 0 -t 60

When the VMA lookup is interrupted, proc_get_vma() returns an error pointer without advancing the position index. Consequently, seq_read_iter() emits the following rate-limited informational message:

seq_file: buggy .next function m_next did not update position index

Save the current position in priv->last_pos before checking the VMA lookup result, allowing it to be reused by both successful and error paths. Advance the position index when returning an error to avoid the spurious warning, while preserving the original error handling.

---
Changes since v1:
- Simplify the error handling and comments as suggested by Lorenzo.
- Clarify the fatal-signal trigger and the cosmetic nature of the warning.

 fs/proc/task_mmu.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
fs/proc/task_mmu.c +4 -3
index 22361607020d..564e52f7ddc6 100644
@@ -216,16 +216,17 @@ static struct vm_area_struct *proc_get_vma(struct seq_file *m, loff_t *ppos)
 
 retry:
 	vma = get_next_vma(priv, *ppos);
-	/* EINTR of EAGAIN is possible */
+	/* Store previous position to be able to restart if needed */
+	priv->last_pos = *ppos;
 	if (IS_ERR(vma)) {
 		if (PTR_ERR(vma) == -EAGAIN && fallback_to_mmap_lock(priv, *ppos))
 			goto retry;
 
+		/* Advance to avoid spurious buggy seq warning in dmesg */
+		(*ppos)++;
 		return vma;
 	}
 
-	/* Store previous position to be able to restart if needed */
-	priv->last_pos = *ppos;
 	if (vma) {
 		/*
 		 * Track the end of the reported vma to ensure position changes
Signature · 1 line
2.43.0
maintainer Reviewed-by on v2 1/1
#1 lore raw

(micro-nit - no need to say 1/1 for a single patch :)

Quoted 16 lines · Adrian Huang (Lenovo)

Commit 5631da56c9a8 ("fs/proc/task_mmu: read proc/pid/maps under per-vma lock") introduced the possibility of returning -EINTR from proc_get_vma() when a fatal signal interrupts mmap_read_lock_killable().

The issue can be reproduced using stress-ng's VMA stressor, which reads /proc/self/maps while concurrently modifying VMAs and periodically terminates child processes with SIGKILL. The reproduced command is:

  # stress-ng --vma 0 -t 60

When the VMA lookup is interrupted, proc_get_vma() returns an error pointer without advancing the position index. Consequently, seq_read_iter() emits the following rate-limited informational message:

seq_file: buggy .next function m_next did not update position index

Quoted 7 lines

Save the current position in priv->last_pos before checking the VMA lookup result, allowing it to be reused by both successful and error paths. Advance the position index when returning an error to avoid the spurious warning, while preserving the original error handling.

LGTM and:

Quoted 34 lines

---
Changes since v1:
- Simplify the error handling and comments as suggested by Lorenzo.
- Clarify the fatal-signal trigger and the cosmetic nature of the warning.

 fs/proc/task_mmu.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
fs/proc/task_mmu.c +4 -4
index 22361607020d..564e52f7ddc6 100644
@@ -216,16 +216,17 @@ static struct vm_area_struct *proc_get_vma(struct seq_file *m, loff_t *ppos)
 
 retry:
 	vma = get_next_vma(priv, *ppos);
-	/* EINTR of EAGAIN is possible */
+	/* Store previous position to be able to restart if needed */
+	priv->last_pos = *ppos;
 	if (IS_ERR(vma)) {
 		if (PTR_ERR(vma) == -EAGAIN && fallback_to_mmap_lock(priv, *ppos))
 			goto retry;
 
+		/* Advance to avoid spurious buggy seq warning in dmesg */
+		(*ppos)++;
 		return vma;
 	}
 
-	/* Store previous position to be able to restart if needed */
-	priv->last_pos = *ppos;
 	if (vma) {
 		/*
 		 * Track the end of the reported vma to ensure position changes
--
Quoted 2 lines

2.43.0

--
Cheers, Lorenzo