mm/damon: fix misc bugs in DAMON modules

From manual code review, I found below bugs in DAMON modules.

DAMON sample modules crash if those are enabled at boot time, via kernel command line. A similar issue was found and fixed on DAMON non-sample modules in the past, but we didn't check that for sample modules.

DAMON non-sample modules are not setting 'enabled' parameters accordingly when real enabling is failed. Honggyu found and fixed[1] this type of bugs in DAMON sample modules, and my inspection was motivated by the great work. Kudos to Honggyu.

Finally, DAMON_RECLIAM is mistakenly losing scheme internal status due to misuse of damon_commit_ctx(). DAMON_LRU_SORT has a similar misuse, but fortunately it is not causing real status loss.

Fix the bugs. Since these are similar patterns of bugs that were found in the past, it would be better to add tests or refactor the code, in future.

Note that the fix of the second bug for DAMON_STAT is sent separately[2], since it is a fix for a bug in mm-unstable tree at the moment. Also as mentioned above, DAMON_LRU_SORT also has a misuse of damon_commit_ctx(), but it is not causing a real issue, hence the fix is not included in this series. I will post it later.

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

SeongJae Park (6):
samples/damon/wsse: fix boot time enable handling samples/damon/prcl: fix boot time enable crash samples/damon/mtier: support boot time enable setup mm/damon/reclaim: reset enabled when DAMON start failed mm/damon/lru_sort: reset enabled when DAMON start failed mm/damon/reclaim: use parameter context correctly

 mm/damon/lru_sort.c   |  5 ++++-
 mm/damon/reclaim.c    |  9 ++++++---
 samples/damon/mtier.c | 13 +++++++++++++
 samples/damon/prcl.c  | 13 +++++++++++++
 samples/damon/wsse.c  | 15 ++++++++++++++-
 5 files changed, 50 insertions(+), 5 deletions(-)

base-commit: a555ad24c884e9f4ee2f2a0184f5b7b89c8d4a6e

2.39.5

If 'enable' parameter of the 'wsse' DAMON sample module is set at boot time via the kernel command line, memory allocation is tried before the slab is initialized. As a result kernel NULL pointer dereference BUG can happen. Fix it by checking the initialization status.

---
 samples/damon/wsse.c | 15 ++++++++++++++-
 1 file changed, 14 insertions(+), 1 deletion(-)

index e20238a249e7..15e2683fe5f3 100644
@@ -89,6 +89,8 @@ static void damon_sample_wsse_stop(void)
 		put_pid(target_pidp);
 }
 
+static bool init_called;
+
 static int damon_sample_wsse_enable_store(
 		const char *val, const struct kernel_param *kp)
 {

@@ -103,6 +105,9 @@ static int damon_sample_wsse_enable_store(
 		return 0;
 
 	if (enable) {
+		if (!init_called)
+			return 0;
+
 		err = damon_sample_wsse_start();
 		if (err)
 			enable = false;

@@ -114,7 +119,15 @@ static int damon_sample_wsse_enable_store(
 
 static int __init damon_sample_wsse_init(void)
 {
-	return 0;
+	int err = 0;
+
+	init_called = true;
+	if (enable) {
+		err = damon_sample_wsse_start();
+		if (err)
+			enable = false;
+	}
+	return err;
 }
 
 module_init(damon_sample_wsse_init);

2.39.5

If 'enable' parameter of the 'prcl' DAMON sample module is set at boot time via the kernel command line, memory allocation is tried before the slab is initialized. As a result kernel NULL pointer dereference BUG can happen. Fix it by checking the initialization status.

---
 samples/damon/prcl.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

index 5597e6a08ab2..f04a18a8626a 100644
@@ -109,6 +109,8 @@ static void damon_sample_prcl_stop(void)
 		put_pid(target_pidp);
 }
 
+static bool init_called;
+
 static int damon_sample_prcl_enable_store(
 		const char *val, const struct kernel_param *kp)
 {

@@ -122,6 +124,9 @@ static int damon_sample_prcl_enable_store(
 	if (enable == enabled)
 		return 0;
 
+	if (!init_called)
+		return 0;
+
 	if (enable) {
 		err = damon_sample_prcl_start();
 		if (err)

@@ -134,6 +139,14 @@ static int damon_sample_prcl_enable_store(
 
 static int __init damon_sample_prcl_init(void)
 {
+	int err = 0;
+
+	init_called = true;
+	if (enable) {
+		err = damon_sample_prcl_start();
+		if (err)
+			enable = false;
+	}
 	return 0;
 }
 

2.39.5

If 'enable' parameter of the 'mtier' DAMON sample module is set at boot time via the kernel command line, memory allocation is tried before the slab is initialized. As a result kernel NULL pointer dereference BUG can happen. Fix it by checking the initialization status.

---
 samples/damon/mtier.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

index 97892ade7f31..20c3102242ec 100644
@@ -157,6 +157,8 @@ static void damon_sample_mtier_stop(void)
 	damon_destroy_ctx(ctxs[1]);
 }
 
+static bool init_called;
+
 static int damon_sample_mtier_enable_store(
 		const char *val, const struct kernel_param *kp)
 {

@@ -170,6 +172,9 @@ static int damon_sample_mtier_enable_store(
 	if (enable == enabled)
 		return 0;
 
+	if (!init_called)
+		return 0;
+
 	if (enable) {
 		err = damon_sample_mtier_start();
 		if (err)

@@ -182,6 +187,14 @@ static int damon_sample_mtier_enable_store(
 
 static int __init damon_sample_mtier_init(void)
 {
+	int err = 0;
+
+	init_called = true;
+	if (enable) {
+		err = damon_sample_mtier_start();
+		if (err)
+			enable = false;
+	}
 	return 0;
 }
 

2.39.5

When the startup fails, 'enabled' parameter is not reset. As a result, users show the parameter 'Y' while it is not really working. Fix it by resetting 'enabled' to 'false' when the work is failed.

---
 mm/damon/reclaim.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

index a675150965e0..c91098d8aa51 100644
@@ -329,7 +329,7 @@ static int __init damon_reclaim_init(void)
 	int err = damon_modules_new_paddr_ctx_target(&ctx, &target);
 
 	if (err)
-		return err;
+		goto out;
 
 	ctx->callback.after_wmarks_check = damon_reclaim_after_wmarks_check;
 	ctx->callback.after_aggregation = damon_reclaim_after_aggregation;

@@ -338,6 +338,9 @@ static int __init damon_reclaim_init(void)
 	if (enabled)
 		err = damon_reclaim_turn(true);
 
+out:
+	if (err && enabled)
+		enabled = false;
 	return err;
 }
 

2.39.5

When the startup fails, 'enabled' parameter is not reset. As a result, users show the parameter 'Y' while it is not really working. Fix it by resetting 'enabled' to 'false' when the work is failed.

---
 mm/damon/lru_sort.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

index 4af8fd4a390b..9bd8a1a115e0 100644
@@ -325,7 +325,7 @@ static int __init damon_lru_sort_init(void)
 	int err = damon_modules_new_paddr_ctx_target(&ctx, &target);
 
 	if (err)
-		return err;
+		goto out;
 
 	ctx->callback.after_wmarks_check = damon_lru_sort_after_wmarks_check;
 	ctx->callback.after_aggregation = damon_lru_sort_after_aggregation;

@@ -334,6 +334,9 @@ static int __init damon_lru_sort_init(void)
 	if (enabled)
 		err = damon_lru_sort_turn(true);
 
+out:
+	if (err && enabled)
+		enabled = false;
 	return err;
 }
 

2.39.5

damon_reclaim_apply_parameters() allocates a new DAMON context, stages user-specified DAMON parameters on it, and commits to running DAMON context at once, using damon_commit_ctx(). The code is mistakenly over-writing the monitoring attributes and the reclaim scheme on the running context. It is not causing a real problem for monitoring attributes, but the scheme overwriting can remove scheme's internal status such as charged quota. Fix the wrong use of the parameter context.

---
 mm/damon/reclaim.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

index c91098d8aa51..0fe8996328b8 100644
@@ -194,7 +194,7 @@ static int damon_reclaim_apply_parameters(void)
 	if (err)
 		return err;
 
-	err = damon_set_attrs(ctx, &damon_reclaim_mon_attrs);
+	err = damon_set_attrs(param_ctx, &damon_reclaim_mon_attrs);
 	if (err)
 		goto out;
 

@@ -202,7 +202,7 @@ static int damon_reclaim_apply_parameters(void)
 	scheme = damon_reclaim_new_scheme();
 	if (!scheme)
 		goto out;
-	damon_set_schemes(ctx, &scheme, 1);
+	damon_set_schemes(param_ctx, &scheme, 1);
 
 	if (quota_mem_pressure_us) {
 		goal = damos_new_quota_goal(DAMOS_QUOTA_SOME_MEM_PSI_US,

2.39.5