Skip to contentSkip to conversation

mm/damon/core: error damos_commit_quota_goal() for zero target_value

RFC PATCH Applied SJ Park 3 participants 6 messages Oct 26 – Oct 27

Subsystems (3)

v2 of 3 Applied Latest is v3 →

lore
Share & tools
lore

SJ Park

author maintainer
#0 lore raw

RFC PATCH v2 1/1 mm/damon/core: error damos_commit_quota_goal() for zero target_value

Patch series "mm/damon: move zero damos quota target_value handling to the core layer".

Having zero DAMOS quota target value can cause division by zero. DAMON API callers are checking the target value parameters to avoid that. It is easy to make mistakes in some of the multiple API callers. Move that to the core layer.

Patch 1 adds the corner case handling into the core layer DAMON parameters validation logic. Patches 2 and 3 remove no more needed DAMON API callers side handling of the corner case in DAMON_LRU_SORT and DMON_SAMPLIE_MTIER, respectively.

This patch (of 3):

If a DAMOS scheme has a damos_quota_goal of zero target_value, damos_quota_goal() could trigger division-by-zero error. Hence each DAMON API callers should do the zero target_value validation. It is easy to make mistakes. Actually such bugs in DAMON_LRU_SORT and DAMON_SAMPLE_MTIER were found and fixed [1].

It is better to handle the corner case only once in the core layer, instead of multiple places in all DAMON API callers. One straightforward option is using an alternative denominator for the corner case in the damos_quota_goal(). However, the zero target_value is meaningless. In this case, the quota goal is always evaluated as achieved or over-achieved. The quota will only keep being reduced.

Simply avoid using zero target_value by adding a check in the core layer DAMOS quota goal parameters validation/commit path, damos_commit_quota_goal(). Update it to return an error in the case. Also update its caller to propagate the error.

Notes for stable kernel porting. The upstream commit doesn't have
Fixes: and therefore not ported to stable kernels. It turned out it fixes commit 8b549a4fd3c5 ("mm/damon/sysfs-schemes: commit damos quota goals user input to DAMOS") which was merged into v6.8. The user impact is a race condition that can result in divide-by-zero. Hence backport. Reproducing the issue requires sysfs write permission, certain DAMON ABI usages that are supposed to be uncommon, and a rare race timing, though. The issue was discovered [2] by Sashiko as a part [3] of its bug report.

Also, the upstream commit on this stable kernel can fix the issue for parameters commit paths but DAMON start path. That's because DAMON start path does not invoke damon_commit_quota_goals(). To fix the path with minimum change, add a new small validation of quota goal target value before damon_start() call.

[2] https://sashiko.dev/#/bug/linux-7f814c9a-986f-4103-a55a-ef9537dcf62b
[3] https://lore.kernel.org/[email protected]

---
Changes from RFC
- RFC: https://lore.kernel.org/[email protected]
- Fix a typo and macro expansion build error for quota goal iteration.
- Destroy ctx in the before-damon_start() error path.

 mm/damon/core.c  | 18 +++++++++++++-----
 mm/damon/sysfs.c | 18 ++++++++++++++++++
 2 files changed, 31 insertions(+), 5 deletions(-)
mm/damon/core.c +13 -5
index 81d44e41de210..ed32855c16ed8 100644
@@ -698,14 +698,17 @@ static struct damos_quota_goal *damos_nth_quota_goal(
 	return NULL;
 }
 
-static void damos_commit_quota_goal(
+static int damos_commit_quota_goal(
 		struct damos_quota_goal *dst, struct damos_quota_goal *src)
 {
+	if (!src->target_value)
+		return  -EINVAL;
 	dst->metric = src->metric;
 	dst->target_value = src->target_value;
 	if (dst->metric == DAMOS_QUOTA_USER_INPUT)
 		dst->current_value = src->current_value;
 	/* keep last_psi_total as is, since it will be updated in next cycle */
+	return 0;
 }
 
 /**
@@ -723,18 +726,23 @@ static void damos_commit_quota_goal(
 int damos_commit_quota_goals(struct damos_quota *dst, struct damos_quota *src)
 {
 	struct damos_quota_goal *dst_goal, *next, *src_goal, *new_goal;
-	int i = 0, j = 0;
+	int i = 0, j = 0, err;
 
 	damos_for_each_quota_goal_safe(dst_goal, next, dst) {
 		src_goal = damos_nth_quota_goal(i++, src);
-		if (src_goal)
-			damos_commit_quota_goal(dst_goal, src_goal);
-		else
+		if (src_goal) {
+			err = damos_commit_quota_goal(dst_goal, src_goal);
+			if (err)
+				return err;
+		} else {
 			damos_destroy_quota_goal(dst_goal);
+		}
 	}
 	damos_for_each_quota_goal_safe(src_goal, next, src) {
 		if (j++ < i)
 			continue;
+		if (!src_goal->target_value)
+			return -EINVAL;
 		new_goal = damos_new_quota_goal(
 				src_goal->metric, src_goal->target_value);
 		if (!new_goal)
mm/damon/sysfs.c +18 -0
index 5aa23c6545cb8..2c8749b62c621 100644
@@ -1497,6 +1497,20 @@ static struct damon_ctx *damon_sysfs_build_ctx(
 	return ctx;
 }
 
+static bool damon_sysfs_valid_start_ctx(struct damon_ctx *ctx)
+{
+	struct damos *s;
+	struct damos_quota_goal *g;
+
+	damon_for_each_scheme(s, ctx) {
+		damos_for_each_quota_goal(g, (&s->quota)) {
+			if (!g->target_value)
+				return false;
+		}
+	}
+	return true;
+}
+
 static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond)
 {
 	struct damon_ctx *ctx;
@@ -1517,6 +1531,10 @@ static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond)
 	ctx = damon_sysfs_build_ctx(kdamond->contexts->contexts_arr[0]);
 	if (IS_ERR(ctx))
 		return PTR_ERR(ctx);
+	if (!damon_sysfs_valid_start_ctx(ctx)) {
+		damon_destroy_ctx(ctx);
+		return -EINVAL;
+	}
 	err = damon_start(&ctx, 1, false);
 	if (err) {
 		damon_destroy_ctx(ctx);

base-commit: 85d82e0d7c2dfd3e75d34f35b7ebbf41f45c9c4e

Signature · 1 line
2.47.3
#1 lore raw
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider: - [Medium] mm/damon/sysfs: pid leak in damon_sysfs_turn_damon_on() on invalid start ctx --

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] mm/damon/sysfs: pid leak in damon_sysfs_turn_damon_on() on invalid start ctx --

commit dc23c827b1a01b5ed768b0a218bb6a67def40106
Author: SJ Park <[email protected]>

mm/damon/core: error damos_commit_quota_goal() for zero target_value

This commit validates that DAMOS quota goal target_value is non-zero in damos_commit_quota_goal() and damos_commit_quota_goals() to prevent a division-by-zero error. For the stable backport, it also adds damon_sysfs_valid_start_ctx() to reject zero target_value before starting DAMON in damon_sysfs_turn_damon_on().

Quoted 26 lines
mm/damon/sysfs.c +14 -0
index 5aa23c6545cb8..2c8749b62c621 100644
@@ -1497,6 +1497,20 @@ static struct damon_ctx *damon_sysfs_build_ctx(
 	return ctx;
 }
 
+static bool damon_sysfs_valid_start_ctx(struct damon_ctx *ctx)
+{
+	struct damos *s;
+	struct damos_quota_goal *g;
+
+	damon_for_each_scheme(s, ctx) {
+		damos_for_each_quota_goal(g, (&s->quota)) {
+			if (!g->target_value)
+				return false;
+		}
+	}
+	return true;
+}
+
 static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond)
 {
 	struct damon_ctx *ctx;

mm/damon/sysfs.c static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond) · line 1537

Quoted 7 lines
mm/damon/sysfs.c +4 -0
@@ -1517,6 +1531,10 @@ static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond)
 	ctx = damon_sysfs_build_ctx(kdamond->contexts->contexts_arr[0]);
 	if (IS_ERR(ctx))
 		return PTR_ERR(ctx);
+	if (!damon_sysfs_valid_start_ctx(ctx)) {
+		damon_destroy_ctx(ctx);
+		return -EINVAL;
+	}

[Severity: Medium]
Does this error path leak the target struct pid references (t->pid) when damon_target_has_pid(ctx) is true?

When damon_sysfs_build_ctx() constructs ctx via damon_sysfs_apply_inputs() -> damon_sysfs_add_targets() -> damon_sysfs_add_target(), it acquires a struct pid reference for each target:

mm/damon/sysfs.c:damon_sysfs_add_target() {
	...
	if (damon_target_has_pid(ctx)) {
		t->pid = find_get_pid(sys_target->pid);
		if (!t->pid)
			goto destroy_targets_out;
	}
	...
}

Because damon_sysfs_valid_start_ctx(ctx) fails before damon_start() is called, kdamond never runs and ctx->callback.before_terminate (damon_sysfs_before_terminate()) is never invoked to release those references. Calling damon_destroy_ctx(ctx) invokes damon_destroy_targets(ctx), which frees each target without calling put_pid(t->pid):

mm/damon/core.c:damon_destroy_targets() {
	struct damon_target *t, *next_t;

	if (ctx->ops.cleanup) {
		ctx->ops.cleanup(ctx);
		return;
	}

	damon_for_each_target_safe(t, next_t, ctx)
		damon_destroy_target(t);
}

In comparison, damon_sysfs_commit_input() cleans up an unstarted context from damon_sysfs_build_ctx() by calling damon_sysfs_destroy_targets() before damon_destroy_ctx():

mm/damon/sysfs.c:damon_sysfs_commit_input() {
	...
	err = damon_commit_ctx(kdamond->damon_ctx, param_ctx);
	damon_sysfs_destroy_targets(param_ctx);
	damon_destroy_ctx(param_ctx);
	return err;
}

Should damon_sysfs_turn_damon_on() also call damon_sysfs_destroy_targets(ctx) before damon_destroy_ctx(ctx) here?

Signature · 1 line
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1
author maintainer
#2 lore raw
Quoted 101 lines · On Mon, 26 Oct 2026 14:11:30 +0000 [email protected] wrote:

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] mm/damon/sysfs: pid leak in damon_sysfs_turn_damon_on() on invalid start ctx --

commit dc23c827b1a01b5ed768b0a218bb6a67def40106
Author: SJ Park <[email protected]>

mm/damon/core: error damos_commit_quota_goal() for zero target_value

This commit validates that DAMOS quota goal target_value is non-zero in damos_commit_quota_goal() and damos_commit_quota_goals() to prevent a division-by-zero error. For the stable backport, it also adds damon_sysfs_valid_start_ctx() to reject zero target_value before starting DAMON in damon_sysfs_turn_damon_on().

Quoted 26 lines
mm/damon/sysfs.c +14 -0
index 5aa23c6545cb8..2c8749b62c621 100644
@@ -1497,6 +1497,20 @@ static struct damon_ctx *damon_sysfs_build_ctx(
 	return ctx;
 }
 
+static bool damon_sysfs_valid_start_ctx(struct damon_ctx *ctx)
+{
+	struct damos *s;
+	struct damos_quota_goal *g;
+
+	damon_for_each_scheme(s, ctx) {
+		damos_for_each_quota_goal(g, (&s->quota)) {
+			if (!g->target_value)
+				return false;
+		}
+	}
+	return true;
+}
+
 static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond)
 {
 	struct damon_ctx *ctx;

mm/damon/sysfs.c static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond) · line 1537

Quoted 7 lines
mm/damon/sysfs.c +4 -0
@@ -1517,6 +1531,10 @@ static int damon_sysfs_turn_damon_on(struct damon_sysfs_kdamond *kdamond)
 	ctx = damon_sysfs_build_ctx(kdamond->contexts->contexts_arr[0]);
 	if (IS_ERR(ctx))
 		return PTR_ERR(ctx);
+	if (!damon_sysfs_valid_start_ctx(ctx)) {
+		damon_destroy_ctx(ctx);
+		return -EINVAL;
+	}

[Severity: Medium]
Does this error path leak the target struct pid references (t->pid) when damon_target_has_pid(ctx) is true?

When damon_sysfs_build_ctx() constructs ctx via damon_sysfs_apply_inputs() -> damon_sysfs_add_targets() -> damon_sysfs_add_target(), it acquires a struct pid reference for each target:

mm/damon/sysfs.c:damon_sysfs_add_target() {
	...
	if (damon_target_has_pid(ctx)) {
		t->pid = find_get_pid(sys_target->pid);
		if (!t->pid)
			goto destroy_targets_out;
	}
	...
}

Because damon_sysfs_valid_start_ctx(ctx) fails before damon_start() is called, kdamond never runs and ctx->callback.before_terminate (damon_sysfs_before_terminate()) is never invoked to release those references. Calling damon_destroy_ctx(ctx) invokes damon_destroy_targets(ctx), which frees each target without calling put_pid(t->pid):

mm/damon/core.c:damon_destroy_targets() {
	struct damon_target *t, *next_t;

	if (ctx->ops.cleanup) {
		ctx->ops.cleanup(ctx);
		return;
	}

	damon_for_each_target_safe(t, next_t, ctx)
		damon_destroy_target(t);
}

In comparison, damon_sysfs_commit_input() cleans up an unstarted context from damon_sysfs_build_ctx() by calling damon_sysfs_destroy_targets() before damon_destroy_ctx():

mm/damon/sysfs.c:damon_sysfs_commit_input() {
	...
	err = damon_commit_ctx(kdamond->damon_ctx, param_ctx);
	damon_sysfs_destroy_targets(param_ctx);
	damon_destroy_ctx(param_ctx);
	return err;
}
Quoted 2 lines

Should damon_sysfs_turn_damon_on() also call damon_sysfs_destroy_targets(ctx) before damon_destroy_ctx(ctx) here?

Nice catch. I will fix this in the next revision.

FYI, this patch for 6.18 and later kernels doesn't have this problem, since they have commit 7114bc5e01cf ("mm/damon/core: add cleanup_target() ops callback"). The commit made damon_destroy_ctx() to do the pid reference drop by itself, and merged into 6.17.

Quoted 3 lines
Signature · 2 lines
-- 
Sashiko AI review · https://sashiko.dev/#/patchset/[email protected]?part=1

Thanks,
SJ

maintainer Applied
#3 lore raw
In reply to SJ Park
Quoted 5 lines

Also, the upstream commit on this stable kernel can fix the issue for parameters commit paths but DAMON start path. That's because DAMON start path does not invoke damon_commit_quota_goals(). To fix the path with minimum change, add a new small validation of quota goal target value before damon_start() call.

Queued for 6.12, thanks.

Signature · 2 lines
Thanks,
Sasha
author maintainer
#4 lore raw

Hello Sasha,

Quoted 7 lines · On Mon, 26 Oct 2026 18:55:58 -0400 Sasha Levin <[email protected]> wrote:
Quoted 5 lines

Also, the upstream commit on this stable kernel can fix the issue for parameters commit paths but DAMON start path. That's because DAMON start path does not invoke damon_commit_quota_goals(). To fix the path with minimum change, add a new small validation of quota goal target value before damon_start() call.

Queued for 6.12, thanks.

I'm sorry but could you please unqueue? Sashiko found [1] a bug that can leak pids.

Also, could we avoid queueing DAMON patches that marked as RFC in general?

[1] https://lore.kernel.org/[email protected]

Thanks,
SJ

[...]

maintainer
#5 lore raw
Quoted 10 lines · SJ Park

Hello Sasha,

Quoted 6 lines · On Mon, 26 Oct 2026 18:55:58 -0400 Sasha Levin <[email protected]> wrote:
Quoted 5 lines

Also, the upstream commit on this stable kernel can fix the issue for parameters commit paths but DAMON start path. That's because DAMON start path does not invoke damon_commit_quota_goals(). To fix the path with minimum change, add a new small validation of quota goal target value before damon_start() call.

Quoted 4 lines
Quoted 1 line

Queued for 6.12, thanks.

I'm sorry but could you please unqueue? Sashiko found [1] a bug that can leak pids.

No problem, now dropped.

Quoted 1 line

Also, could we avoid queueing DAMON patches that marked as RFC in general?

Sorry about this. We don't normally get RFC tags for backports so my brain didn't register this at all :)

Signature · 2 lines
Thanks,
Sasha