Skip to contentSkip to conversation

psi: fix possible missing or delayed pending event

PATCH Superseded Hao Lee 2 participants 7 messages Sep 14, 2022 – Sep 19, 2022

Subsystems (2)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore

Hao Lee

author
#0 lore raw

PATCH 1/1 psi: fix possible missing or delayed pending event

When a pending event exists and growth is less than the threshold, the current logic is to skip this trigger without generating event. However, from e6df4ead85d9 ("psi: fix possible trigger missing in the window"), our purpose is to generate event as long as pending event exists and the rate meets the limit. This patch fixes the possible pending-event missing or delay.

Fixes e6df4ead85d9 ("psi: fix possible trigger missing in the window")
---
 kernel/sched/psi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/psi.c +1 -1
index 9711827e3..0bae4ee2b 100644
@@ -539,7 +539,7 @@ static u64 update_triggers(struct psi_group *group, u64 now)
 
 			/* Calculate growth since last update */
 			growth = window_update(&t->win, now, total[t->state]);
-			if (growth < t->threshold)
+			if (growth < t->threshold && !t->pending_event)
 				continue;
 
 			t->pending_event = true;
Signature · 1 line
2.21.0
#1 lore raw
Quoted 19 lines · Hao Lee

When a pending event exists and growth is less than the threshold, the current logic is to skip this trigger without generating event. However, from e6df4ead85d9 ("psi: fix possible trigger missing in the window"), our purpose is to generate event as long as pending event exists and the rate meets the limit. This patch fixes the possible pending-event missing or delay.

Fixes e6df4ead85d9 ("psi: fix possible trigger missing in the window")
---
 kernel/sched/psi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/psi.c +0 -0
index 9711827e3..0bae4ee2b 100644

kernel/sched/psi.c static u64 update_triggers(struct psi_group *group, u64 now) · line 542

Quoted 5 lines
kernel/sched/psi.c +1 -1
@@ -539,7 +539,7 @@ static u64 update_triggers(struct psi_group *group, u64 now)
 
                        /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);
-                       if (growth < t->threshold)
+                       if (growth < t->threshold && !t->pending_event)

I'm not sure how this additional condition changes things. Current logic is to set t->pending_event=true whenever growth exceeds the t->threshold. This patch will change this logic into setting t->pending_event=true also when t->pending_event=true. But why would you want to set t->pending_event=true if it's already true? What am I missing?

Quoted 6 lines
kernel/sched/psi.c +0 -1
                                continue;
 
                        t->pending_event = true;
--

2.21.0

author
#2 lore raw
Quoted 21 lines · Suren Baghdasaryan
Quoted 20 lines · Hao Lee

When a pending event exists and growth is less than the threshold, the current logic is to skip this trigger without generating event. However, from e6df4ead85d9 ("psi: fix possible trigger missing in the window"), our purpose is to generate event as long as pending event exists and the rate meets the limit. This patch fixes the possible pending-event missing or delay.

Fixes e6df4ead85d9 ("psi: fix possible trigger missing in the window")
---
 kernel/sched/psi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/psi.c +0 -0
index 9711827e3..0bae4ee2b 100644
@@ -539,7 +539,7 @@ static u64 update_triggers(struct psi_group *group, u64 now)
 
Quoted 9 lines

kernel/sched/psi.c static u64 update_triggers(struct psi_group *group, u64 now) · line 542

Quoted 4 lines
kernel/sched/psi.c +1 -1
                        /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);
-                       if (growth < t->threshold)
+                       if (growth < t->threshold && !t->pending_event)

I'm not sure how this additional condition changes things. Current logic is to set t->pending_event=true whenever growth exceeds the t->threshold. This patch will change this logic into setting t->pending_event=true also when t->pending_event=true.

This is right.

Quoted 3 lines

But why would
you want to set t->pending_event=true if it's already true? What am I missing?

If I expand this if-else branch and the pending_event statement to a more detailed snippet, it will be like this:

if (growth < t->threshold && !t->pending_event) // under threshold && no pending event. Skip.

	continue;
else if (growth >= t->threshold) // above threshold. Try to generate event.
	t->pending_event = true;

else // under threshold && have pending events. Try to generate event.
; // pending_event is already true. do nothing

The original code didn't handle the else condition properly. It will skip the trigger when its growth is under the threshold, even though it has a pending event. This patch handles this condition correctly.

But I think assigning true to pending_event when it's already true doesn't have other side effects, so I eliminate the else if branch. Maybe we'd better make it explicit, like the above snippet? Thanks.

Quoted 7 lines
Quoted 7 lines
kernel/sched/psi.c +0 -1
                                continue;
 
                        t->pending_event = true;
--

2.21.0

#3 lore raw
Quoted 43 lines · Hao Lee
Quoted 21 lines · Suren Baghdasaryan
Quoted 20 lines · Hao Lee

When a pending event exists and growth is less than the threshold, the current logic is to skip this trigger without generating event. However, from e6df4ead85d9 ("psi: fix possible trigger missing in the window"), our purpose is to generate event as long as pending event exists and the rate meets the limit. This patch fixes the possible pending-event missing or delay.

Fixes e6df4ead85d9 ("psi: fix possible trigger missing in the window")
---
 kernel/sched/psi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/psi.c +0 -0
index 9711827e3..0bae4ee2b 100644
@@ -539,7 +539,7 @@ static u64 update_triggers(struct psi_group *group, u64 now)
 
Quoted 9 lines

kernel/sched/psi.c static u64 update_triggers(struct psi_group *group, u64 now) · line 542

Quoted 4 lines
kernel/sched/psi.c +1 -1
                        /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);
-                       if (growth < t->threshold)
+                       if (growth < t->threshold && !t->pending_event)

I'm not sure how this additional condition changes things. Current logic is to set t->pending_event=true whenever growth exceeds the t->threshold. This patch will change this logic into setting t->pending_event=true also when t->pending_event=true.

This is right.

Quoted 3 lines

But why would
you want to set t->pending_event=true if it's already true? What am I missing?

If I expand this if-else branch and the pending_event statement to a more detailed snippet, it will be like this:

if (growth < t->threshold && !t->pending_event) // under threshold && no pending event. Skip.

Quoted 8 lines
        continue;
else if (growth >= t->threshold) // above threshold. Try to generate event.
        t->pending_event = true;

else // under threshold && have pending events. Try to generate event.
; // pending_event is already true. do nothing

The original code didn't handle the else condition properly.

The else condition in your code does nothing, and that's why the original code does not implement a handler for that case.

Quoted 7 lines

It will
skip the trigger when its growth is under the threshold, even though it has a pending event. This patch handles this condition correctly.

But I think assigning true to pending_event when it's already true doesn't have other side effects, so I eliminate the else if branch. Maybe we'd better make it explicit, like the above snippet? Thanks.

The new code you posted is functionally the same as the original one while being more verbose and IMO less readable. Unless you can explain the problem with the original code, I don't see any reason to change it.

Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
kernel/sched/psi.c +0 -1
                                continue;
 
                        t->pending_event = true;
--

2.21.0

author
#4 lore raw
Quoted 60 lines · Suren Baghdasaryan
Quoted 43 lines · Hao Lee
Quoted 21 lines · Suren Baghdasaryan
Quoted 20 lines · Hao Lee

When a pending event exists and growth is less than the threshold, the current logic is to skip this trigger without generating event. However, from e6df4ead85d9 ("psi: fix possible trigger missing in the window"), our purpose is to generate event as long as pending event exists and the rate meets the limit. This patch fixes the possible pending-event missing or delay.

Fixes e6df4ead85d9 ("psi: fix possible trigger missing in the window")
---
 kernel/sched/psi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/psi.c +0 -0
index 9711827e3..0bae4ee2b 100644
@@ -539,7 +539,7 @@ static u64 update_triggers(struct psi_group *group, u64 now)
 
Quoted 9 lines

kernel/sched/psi.c static u64 update_triggers(struct psi_group *group, u64 now) · line 542

Quoted 4 lines
kernel/sched/psi.c +1 -1
                        /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);
-                       if (growth < t->threshold)
+                       if (growth < t->threshold && !t->pending_event)

I'm not sure how this additional condition changes things. Current logic is to set t->pending_event=true whenever growth exceeds the t->threshold. This patch will change this logic into setting t->pending_event=true also when t->pending_event=true.

This is right.

Quoted 3 lines

But why would
you want to set t->pending_event=true if it's already true? What am I missing?

If I expand this if-else branch and the pending_event statement to a more detailed snippet, it will be like this:

if (growth < t->threshold && !t->pending_event) // under threshold && no pending event. Skip.

Quoted 8 lines
        continue;
else if (growth >= t->threshold) // above threshold. Try to generate event.
        t->pending_event = true;

else // under threshold && have pending events. Try to generate event.
; // pending_event is already true. do nothing

The original code didn't handle the else condition properly.

The else condition in your code does nothing, and that's why the original code does not implement a handler for that case.

Quoted 4 lines

It will
skip the trigger when its growth is under the threshold, even though it has a pending event. This patch handles this condition correctly.

Quoted 8 lines
Quoted 3 lines

But I think assigning true to pending_event when it's already true doesn't have other side effects, so I eliminate the else if branch. Maybe we'd better make it explicit, like the above snippet? Thanks.

The new code you posted is functionally the same as the original one while being more verbose and IMO less readable. Unless you can explain the problem with the original code, I don't see any reason to change it.

Hi, for the original code, let's assume t->pending_event is true:
* if new_stall is false, we will try to check event ratelimit and generate an event for this psi_trigger. This case is right.
* but if new_stall is true, we will skip this psi_trigger if growth growth < t->threshold. I think we shouldn't skip this psi_trigger in this case because it has a pending event.

Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
Quoted 9 lines
kernel/sched/psi.c +0 -1
                                continue;
 
                        t->pending_event = true;
--

2.21.0

#5 lore raw
Quoted 71 lines · Hao Lee
Quoted 60 lines · Suren Baghdasaryan
Quoted 43 lines · Hao Lee
Quoted 21 lines · Suren Baghdasaryan
Quoted 20 lines · Hao Lee

When a pending event exists and growth is less than the threshold, the current logic is to skip this trigger without generating event. However, from e6df4ead85d9 ("psi: fix possible trigger missing in the window"), our purpose is to generate event as long as pending event exists and the rate meets the limit. This patch fixes the possible pending-event missing or delay.

Fixes e6df4ead85d9 ("psi: fix possible trigger missing in the window")
---
 kernel/sched/psi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/psi.c +0 -0
index 9711827e3..0bae4ee2b 100644
@@ -539,7 +539,7 @@ static u64 update_triggers(struct psi_group *group, u64 now)
 
Quoted 9 lines

kernel/sched/psi.c static u64 update_triggers(struct psi_group *group, u64 now) · line 542

Quoted 4 lines
kernel/sched/psi.c +1 -1
                        /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);
-                       if (growth < t->threshold)
+                       if (growth < t->threshold && !t->pending_event)

I'm not sure how this additional condition changes things. Current logic is to set t->pending_event=true whenever growth exceeds the t->threshold. This patch will change this logic into setting t->pending_event=true also when t->pending_event=true.

This is right.

Quoted 3 lines

But why would
you want to set t->pending_event=true if it's already true? What am I missing?

If I expand this if-else branch and the pending_event statement to a more detailed snippet, it will be like this:

if (growth < t->threshold && !t->pending_event) // under threshold && no pending event. Skip.

Quoted 8 lines
        continue;
else if (growth >= t->threshold) // above threshold. Try to generate event.
        t->pending_event = true;

else // under threshold && have pending events. Try to generate event.
; // pending_event is already true. do nothing

The original code didn't handle the else condition properly.

The else condition in your code does nothing, and that's why the original code does not implement a handler for that case.

Quoted 4 lines

It will
skip the trigger when its growth is under the threshold, even though it has a pending event. This patch handles this condition correctly.

Quoted 8 lines
Quoted 3 lines

But I think assigning true to pending_event when it's already true doesn't have other side effects, so I eliminate the else if branch. Maybe we'd better make it explicit, like the above snippet? Thanks.

The new code you posted is functionally the same as the original one while being more verbose and IMO less readable. Unless you can explain the problem with the original code, I don't see any reason to change it.

Quoted 6 lines

Hi, for the original code, let's assume t->pending_event is true:
* if new_stall is false, we will try to check event ratelimit and generate an event for this psi_trigger. This case is right.
* but if new_stall is true, we will skip this psi_trigger if growth growth < t->threshold. I think we shouldn't skip this psi_trigger in this case because it has a pending event.

Ok, I see the issue now. I think the following fix would be the simplest:

                       /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);

- if (growth < t->threshold)
- continue;
+ if (!t->pending_event) {
+ if (growth < t->threshold)
+ continue;

- t->pending_event = true;
+ t->pending_event = true;
+ }

Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
Quoted 10 lines
kernel/sched/psi.c +0 -1
                                continue;
 
                        t->pending_event = true;
--

2.21.0

author
#6 lore raw
Quoted 83 lines · Suren Baghdasaryan
Quoted 71 lines · Hao Lee
Quoted 60 lines · Suren Baghdasaryan
Quoted 43 lines · Hao Lee
Quoted 21 lines · Suren Baghdasaryan
Quoted 20 lines · Hao Lee

When a pending event exists and growth is less than the threshold, the current logic is to skip this trigger without generating event. However, from e6df4ead85d9 ("psi: fix possible trigger missing in the window"), our purpose is to generate event as long as pending event exists and the rate meets the limit. This patch fixes the possible pending-event missing or delay.

Fixes e6df4ead85d9 ("psi: fix possible trigger missing in the window")
---
 kernel/sched/psi.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
kernel/sched/psi.c +0 -0
index 9711827e3..0bae4ee2b 100644
@@ -539,7 +539,7 @@ static u64 update_triggers(struct psi_group *group, u64 now)
 
Quoted 9 lines

kernel/sched/psi.c static u64 update_triggers(struct psi_group *group, u64 now) · line 542

Quoted 4 lines
kernel/sched/psi.c +1 -1
                        /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);
-                       if (growth < t->threshold)
+                       if (growth < t->threshold && !t->pending_event)

I'm not sure how this additional condition changes things. Current logic is to set t->pending_event=true whenever growth exceeds the t->threshold. This patch will change this logic into setting t->pending_event=true also when t->pending_event=true.

This is right.

Quoted 3 lines

But why would
you want to set t->pending_event=true if it's already true? What am I missing?

If I expand this if-else branch and the pending_event statement to a more detailed snippet, it will be like this:

if (growth < t->threshold && !t->pending_event) // under threshold && no pending event. Skip.

Quoted 8 lines
        continue;
else if (growth >= t->threshold) // above threshold. Try to generate event.
        t->pending_event = true;

else // under threshold && have pending events. Try to generate event.
; // pending_event is already true. do nothing

The original code didn't handle the else condition properly.

The else condition in your code does nothing, and that's why the original code does not implement a handler for that case.

Quoted 4 lines

It will
skip the trigger when its growth is under the threshold, even though it has a pending event. This patch handles this condition correctly.

Quoted 8 lines
Quoted 3 lines

But I think assigning true to pending_event when it's already true doesn't have other side effects, so I eliminate the else if branch. Maybe we'd better make it explicit, like the above snippet? Thanks.

The new code you posted is functionally the same as the original one while being more verbose and IMO less readable. Unless you can explain the problem with the original code, I don't see any reason to change it.

Quoted 6 lines

Hi, for the original code, let's assume t->pending_event is true:
* if new_stall is false, we will try to check event ratelimit and generate an event for this psi_trigger. This case is right.
* but if new_stall is true, we will skip this psi_trigger if growth growth < t->threshold. I think we shouldn't skip this psi_trigger in this case because it has a pending event.

Ok, I see the issue now. I think the following fix would be the simplest:

                       /* Calculate growth since last update */
                        growth = window_update(&t->win, now, total[t->state]);

Review

Quoted 9 lines
(quoted diff) +5 -3
-                       if (growth < t->threshold)
-                                continue;
+                       if (!t->pending_event) {
+                              if (growth < t->threshold)
+                                        continue;
 
-                        t->pending_event = true;
+                               t->pending_event = true;
+                       }

Great! I will update in v2.

Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
Quoted 12 lines
kernel/sched/psi.c +0 -1
                                continue;
 
                        t->pending_event = true;
--

2.21.0