Skip to contentSkip to conversation

Documentation: index.rst: add entry of other sub-directory

PATCH In review Manuel Ebner 3 participants 7 messages Jun 3 – Jun 15

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Documentation: index.rst: add entry of other sub-directory

add reference to scheduler/sched-rt-group.rst

---
 Documentation/core-api/real-time/index.rst | 1 +
 1 file changed, 1 insertion(+)
Documentation/core-api/real-time/index.rst +1 -0
index f08d2395a22c..661b419e7f8f 100644
@@ -15,3 +15,4 @@ the required changes compared to a non-PREEMPT_RT configuration.
    differences
    hardware
    architecture-porting
+   Real-Time group scheduling <../../scheduler/sched-rt-group>
Signature · 1 line
2.54.0
#1 lore raw

PATCH 1/1 Documentation: real-time/index: add entry

I figured out the subject and changelog aren't good. It should be:

[PATCH] Documentation: real-time/index: add entry

Add reference to realtime/index.rst pointing to scheduler/sched-rt-group.rst

---

I'm waiting for feedback before making [v2].

Thanks
Manuel

#2 lore raw
In reply to Manuel Ebner
Quoted 12 lines · Manuel Ebner

add reference to scheduler/sched-rt-group.rst

---
 Documentation/core-api/real-time/index.rst | 1 +
 1 file changed, 1 insertion(+)
Documentation/core-api/real-time/index.rst +0 -0
index f08d2395a22c..661b419e7f8f 100644

Documentation/core-api/real-time/index.rst the required changes compared to a non-PREEMPT_RT configuration. · line 18

Quoted 4 lines
Documentation/core-api/real-time/index.rst +1 -0
@@ -15,3 +15,4 @@ the required changes compared to a non-PREEMPT_RT configuration.
    differences
    hardware
    architecture-porting
+   Real-Time group scheduling <../../scheduler/sched-rt-group>

Have you been using this and are referencing this because it is useful or did you just happen to find it?
People have been complaining about this and I *think* it fails to work in multi-CPU scenarios (it might work if the task is restricted to single CPU usage). Not entirely certain. But complains, that I am certain of.
There is replacement underway to utilize SCHED_DL for this as in "Hierarchical Constant Bandwidth Server (HCBS)"

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

Sebastian

#3 lore raw
Quoted 8 lines · Sebastian Andrzej Siewior
Quoted 7 lines · Manuel Ebner

add reference to scheduler/sched-rt-group.rst

---
 Documentation/core-api/real-time/index.rst | 1 +

 1 file changed, 1 insertion(+)
Quoted 13 lines
Quoted 10 lines
Documentation/core-api/real- renamed from Documentation/core-api/real-time/index.rst +0 -0

time/index.rst
index f08d2395a22c..661b419e7f8f 100644

Documentation/core-api/real-time/index.rst +0 -0
@@ -15,3 +15,4 @@ the required changes compared to a non-PREEMPT_RT configuration.

    differences
    hardware
    architecture-porting
+   Real-Time group scheduling <../../scheduler/sched-rt-group>

Have you been using this and are referencing this because it is useful or did you just happen to find it?

I happened to find it.

Quoted 1 line

People have been complaining about this

Maybe with the help of them we can figure out what's happening. adding [email protected]

Quoted 7 lines

and I *think* it fails to work
in multi-CPU scenarios (it might work if the task is restricted to single CPU usage). Not entirely certain. But complains, that I am certain of.
There is replacement underway to utilize SCHED_DL for this as in "Hierarchical Constant Bandwidth Server (HCBS)"

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

will look into this.

Quoted 2 lines

Sebastian

Thanks for the feedback
Manuel

#4 lore raw
Quoted 4 lines · Manuel Ebner
Quoted 1 line

People have been complaining about this

Maybe with the help of them we can figure out what's happening. adding [email protected]

good luck. What I managed to find out is that those, that did not copy statements but run into trouble themself, were using in a multi-CPU scenario. Based on other details I *think* it should not happen single CPU usage but nobody confirmed.

But this HCBS thingy aims at replacing/ removing it entirely…

Sebastian

maintainer
#5 lore raw
In reply to Manuel Ebner
Quoted 12 lines · Manuel Ebner

add reference to scheduler/sched-rt-group.rst

---
 Documentation/core-api/real-time/index.rst | 1 +
 1 file changed, 1 insertion(+)
Documentation/core-api/real-time/index.rst +0 -0
index f08d2395a22c..661b419e7f8f 100644

Documentation/core-api/real-time/index.rst the required changes compared to a non-PREEMPT_RT configuration. · line 18

Quoted 4 lines
Documentation/core-api/real-time/index.rst +1 -0
@@ -15,3 +15,4 @@ the required changes compared to a non-PREEMPT_RT configuration.
    differences
    hardware
    architecture-porting
+   Real-Time group scheduling <../../scheduler/sched-rt-group>

This file already appears in the toctree in Documentation/scheduler/index.rst; adding it elsewhere can create build problems. If you really want a link here, the way to do it is to add a cross reference instead.

jon

#6 lore raw
Quoted 9 lines · Jonathan Corbet
Quoted 7 lines · Manuel Ebner

add reference to scheduler/sched-rt-group.rst

---
 Documentation/core-api/real-time/index.rst | 1 +
 1 file changed, 1 insertion(+)
Quoted 14 lines

Documentation/core-api/real-time/index.rst the required changes compared to a non-PREEMPT_RT configuration. · line 18

Quoted 9 lines
Documentation/core-api/real-time/index.rst +1 -0
index f08d2395a22c..661b419e7f8f 100644
@@ -15,3 +15,4 @@ the required changes compared to a non-PREEMPT_RT configuration.
    differences
    hardware
    architecture-porting
+   Real-Time group scheduling <../../scheduler/sched-rt-group>

This file already appears in the toctree in Documentation/scheduler/index.rst; adding it elsewhere can create build problems. If you really want a link here, the way to do it is to add a cross reference instead.

The long term goal is to remove sched-rt-group and replace it with something else. Manuel stated that he just found it and was not/ is not using it. Maybe just drop it entirely.

Quoted 1 line

jon

Sebastian