Skip to contentSkip to conversation

sched_ext: Fix stale errno in scx_sub_enable_workfn()

PATCH Superseded Cui Jian 3 participants 5 messages Jul 18 – Jul 22

Subsystems (2)
lore
Share & tools
lore

Cui Jian

author
#0 lore raw

PATCH 1/1 sched_ext: Fix stale errno in scx_sub_enable_workfn()

scx_sub_enable_workfn() drops the return value of validate_ops(). When validate_ops() fails, the function jumps to err_disable with ret still holding 0 from the previous call, so the fallback error added by commit db4e9defd2e8 ("sched_ext: Record an error on errno-only sub-enable failure") reports "scx_sub_enable() failed (0)".

This is currently harmless because validate_ops() records its own scx_error() first and the first error wins, but it leaves the fallback broken for this path. Save the return value into ret, like scx_root_enable_workfn() already does.

The nesting depth check and the cgroup online check also reach err_disable without setting ret. Set -EINVAL and -ENODEV there so the fallback always reports a real errno.

---
 kernel/sched/ext/ext.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
kernel/sched/ext/ext.c +4 -1
index e3fa7b2fac9d..e623d6375f66 100644
@@ -7559,6 +7559,7 @@ static void scx_sub_enable_workfn(struct kthread_work *work)
 	if (sch->level >= SCX_SUB_MAX_DEPTH) {
 		scx_error(sch, "max nesting depth %d violated",
 			  SCX_SUB_MAX_DEPTH);
+		ret = -EINVAL;
 		goto err_disable;
 	}
 
@@ -7580,7 +7581,8 @@ static void scx_sub_enable_workfn(struct kthread_work *work)
 	if (ret)
 		goto err_disable;
 
-	if (validate_ops(sch, ops))
+	ret = validate_ops(sch, ops);
+	if (ret)
 		goto err_disable;
 
 	struct scx_sub_attach_args sub_attach_args = {
@@ -7613,6 +7615,7 @@ static void scx_sub_enable_workfn(struct kthread_work *work)
 	set_cgroup_sched(sch_cgroup(sch), sch);
 	if (!(cgrp->self.flags & CSS_ONLINE)) {
 		scx_error(sch, "cgroup is not online");
+		ret = -ENODEV;
 		goto err_unlock_and_disable;
 	}
 

base-commit: 1229e2e57a5c2980ccd457b9b53ea0eed5a22ab3

Signature · 1 line
2.34.1
reviewer
#1 lore raw

Hi Cui,

Quoted 12 lines · Cui Jian

scx_sub_enable_workfn() drops the return value of validate_ops(). When validate_ops() fails, the function jumps to err_disable with ret still holding 0 from the previous call, so the fallback error added by commit db4e9defd2e8 ("sched_ext: Record an error on errno-only sub-enable failure") reports "scx_sub_enable() failed (0)".

This is currently harmless because validate_ops() records its own scx_error() first and the first error wins, but it leaves the fallback broken for this path. Save the return value into ret, like scx_root_enable_workfn() already does.

Quoted 5 lines

The nesting depth check and the cgroup online check also reach err_disable without setting ret. Set -EINVAL and -ENODEV there so the fallback always reports a real errno.

This doesn't apply anymore, but the logic should still be valid. Can you rebase it to the latest branch?

git://git.kernel.org/pub/scm/linux/kernel/git/tj/sched_ext.git for-7.3

Thanks,
-Andrea

Quoted 39 lines
---
 kernel/sched/ext/ext.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)
kernel/sched/ext/ext.c +4 -1
index e3fa7b2fac9d..e623d6375f66 100644
@@ -7559,6 +7559,7 @@ static void scx_sub_enable_workfn(struct kthread_work *work)
 	if (sch->level >= SCX_SUB_MAX_DEPTH) {
 		scx_error(sch, "max nesting depth %d violated",
 			  SCX_SUB_MAX_DEPTH);
+		ret = -EINVAL;
 		goto err_disable;
 	}
 
@@ -7580,7 +7581,8 @@ static void scx_sub_enable_workfn(struct kthread_work *work)
 	if (ret)
 		goto err_disable;
 
-	if (validate_ops(sch, ops))
+	ret = validate_ops(sch, ops);
+	if (ret)
 		goto err_disable;
 
 	struct scx_sub_attach_args sub_attach_args = {
@@ -7613,6 +7615,7 @@ static void scx_sub_enable_workfn(struct kthread_work *work)
 	set_cgroup_sched(sch_cgroup(sch), sch);
 	if (!(cgrp->self.flags & CSS_ONLINE)) {
 		scx_error(sch, "cgroup is not online");
+		ret = -ENODEV;
 		goto err_unlock_and_disable;
 	}
 

base-commit: 1229e2e57a5c2980ccd457b9b53ea0eed5a22ab3

Signature · 1 line
2.34.1
author
#2 lore raw

PATCH v2 1/1 sched_ext: Fix stale errno in scx_sub_enable_workfn()

The nesting depth check and the cgroup online check in scx_sub_enable_workfn() reach err_disable without setting ret, so the fallback error added by commit db4e9defd2e8 ("sched_ext: Record an error on errno-only sub-enable failure") reports "scx_sub_enable() failed (0)".

This is currently harmless because both paths record their own scx_error() first and the first error wins, but it leaves the fallback broken for these paths. Set -EINVAL and -ENODEV there so the fallback always reports a real errno.

v2: The validate_ops() path from v1 is already fixed in for-7.3
(sub.c already has ret = scx_validate_ops()), so only the two remaining paths are addressed.

---
 kernel/sched/ext/sub.c | 2 ++
 1 file changed, 2 insertions(+)
kernel/sched/ext/sub.c +2 -0
index d7842a609d96..6da6c91e4287 100644
@@ -1331,6 +1331,7 @@ void scx_sub_enable_workfn(struct kthread_work *work)
 	if (sch->level >= SCX_SUB_MAX_DEPTH) {
 		scx_error(sch, "max nesting depth %d violated",
 			  SCX_SUB_MAX_DEPTH);
+		ret = -EINVAL;
 		goto err_disable;
 	}
 
@@ -1378,6 +1379,7 @@ void scx_sub_enable_workfn(struct kthread_work *work)
 	set_cgroup_sched(sch_cgroup(sch), sch);
 	if (!(cgrp->self.flags & CSS_ONLINE)) {
 		scx_error(sch, "cgroup is not online");
+		ret = -ENODEV;
 		goto err_unlock_and_disable;
 	}
 
Signature · 1 line
2.34.1
reviewer Reviewed-by on v2 1/1
#3 lore raw

Hi Cui,

Quoted 11 lines · Cui Jian

The nesting depth check and the cgroup online check in scx_sub_enable_workfn() reach err_disable without setting ret, so the fallback error added by commit db4e9defd2e8 ("sched_ext: Record an error on errno-only sub-enable failure") reports "scx_sub_enable() failed (0)".

This is currently harmless because both paths record their own scx_error() first and the first error wins, but it leaves the fallback broken for these paths. Set -EINVAL and -ENODEV there so the fallback always reports a real errno.

Quoted 5 lines

v2: The validate_ops() path from v1 is already fixed in for-7.3
(sub.c already has ret = scx_validate_ops()), so only the two remaining paths are addressed.

Looks good to me now.

Thanks,
-Andrea

Quoted 27 lines
---
 kernel/sched/ext/sub.c | 2 ++
 1 file changed, 2 insertions(+)
kernel/sched/ext/sub.c +2 -0
index d7842a609d96..6da6c91e4287 100644
@@ -1331,6 +1331,7 @@ void scx_sub_enable_workfn(struct kthread_work *work)
 	if (sch->level >= SCX_SUB_MAX_DEPTH) {
 		scx_error(sch, "max nesting depth %d violated",
 			  SCX_SUB_MAX_DEPTH);
+		ret = -EINVAL;
 		goto err_disable;
 	}
 
@@ -1378,6 +1379,7 @@ void scx_sub_enable_workfn(struct kthread_work *work)
 	set_cgroup_sched(sch_cgroup(sch), sch);
 	if (!(cgrp->self.flags & CSS_ONLINE)) {
 		scx_error(sch, "cgroup is not online");
+		ret = -ENODEV;
 		goto err_unlock_and_disable;
 	}
 
Signature · 1 line
2.34.1
reviewer Applied
#4 lore raw
In reply to Cui Jian

Hello,

Quoted 5 lines · Cui Jian

The nesting depth check and the cgroup online check in scx_sub_enable_workfn() reach err_disable without setting ret, so the fallback error added by commit db4e9defd2e8 ("sched_ext: Record an error on errno-only sub-enable failure") reports "scx_sub_enable() failed (0)".

Applied to sched_ext/for-7.3.

Thanks.

Signature · 1 line
tejun