@@ -2225,7 +2225,8 @@ Cpuset Interface Files
========== =====================================
"member" Non-root member of a partition
"root" Partition root
- "isolated" Partition root without load balancing
+ "isolcpus" Partition root for isolated CPUs pool
+ "isolated" Partition root for isolated CPUs
========== =====================================
The root cgroup is always a partition root and its state
@@ -2237,24 +2238,41 @@ Cpuset Interface Files
its descendants except those that are separate partition roots
themselves and their descendants.
+ When set to "isolcpus", the CPUs in that partition root will
+ be in an isolated state without any load balancing from the
+ scheduler. This partition root is special as there can be at
+ most one instance of it in a system and no task or child cpuset
+ is allowed in this cgroup. It acts as a pool of isolated CPUs to
+ be pulled into other "isolated" partitions. The "cpuset.cpus"
+ of an "isolcpus" partition root contains the list of isolated
+ CPUs it holds, where "cpuset.cpus.effective" contains the list
+ of freely available isolated CPUs that are ready to be pull
+ into other "isolated" partition.
+
When set to "isolated", the CPUs in that partition root will
be in an isolated state without any load balancing from the
scheduler. Tasks placed in such a partition with multiple
CPUs should be carefully distributed and bound to each of the
- individual CPUs for optimal performance.
-
- The value shown in "cpuset.cpus.effective" of a partition root
- is the CPUs that the partition root can dedicate to a potential
- new child partition root. The new child subtracts available
- CPUs from its parent "cpuset.cpus.effective".
-
- A partition root ("root" or "isolated") can be in one of the
- two possible states - valid or invalid. An invalid partition
- root is in a degraded state where some state information may
- be retained, but behaves more like a "member".
-
- All possible state transitions among "member", "root" and
- "isolated" are allowed.
+ individual CPUs for optimal performance. The isolated CPUs can
+ come from either the parent partition root or from an "isolcpus"
+ partition if the parent cannot satisfy its request.
+
+ The value shown in "cpuset.cpus.effective" of a partition root is
+ the CPUs that the partition root can dedicate to a potential new
+ child partition root. The new child partition subtracts available
+ CPUs from its parent "cpuset.cpus.effective". An exception is
+ an "isolated" partition that pulls its isolated CPUs from the
+ "isolcpus" partition root that is not its direct parent.
+
+ A partition root can be in one of the two possible states -
+ valid or invalid. An invalid partition root is in a degraded
+ state where some state information may be retained, but behaves
+ more like a "member".
+
+ All possible state transitions among "member", "root", "isolcpus"
+ and "isolated" are allowed. However, the partition root may
+ not be valid if the corresponding prerequisite conditions are
+ not met.
On read, the "cpuset.cpus.partition" file can show the following
values.
@@ -2262,16 +2280,18 @@ Cpuset Interface Files
============================= =====================================
"member" Non-root member of a partition
"root" Partition root
- "isolated" Partition root without load balancing
+ "isolcpus" Partition root for isolated CPUs pool
+ "isolated" Partition root for isolated CPUs
"root invalid (<reason>)" Invalid partition root
+ "isolcpus invalid (<reason>)" Invalid isolcpus partition root
"isolated invalid (<reason>)" Invalid isolated partition root
============================= =====================================
In the case of an invalid partition root, a descriptive string on
- why the partition is invalid is included within parentheses.
+ why the partition is invalid may be included within parentheses.
- For a partition root to become valid, the following conditions
- must be met.
+ For a "root" partition root to become valid, the following
+ conditions must be met.
1) The "cpuset.cpus" is exclusive with its siblings , i.e. they
are not shared by any of its siblings (exclusivity rule).
@@ -2281,6 +2301,37 @@ Cpuset Interface Files
4) The "cpuset.cpus.effective" cannot be empty unless there is
no task associated with this partition.
+ A valid "isolcpus" partition root requires the following
+ conditions.
+
+ 1) The parent cgroup is a valid partition root.
+ 2) The "cpuset.cpus" must be a subset of parent's "cpuset.cpus"
+ including an empty cpu list.
+ 3) There can be no more than one valid "isolcpus" partition.
+ 4) No task or child cpuset is allowed.
+
+ Note that an "isolcpus" partition is not exclusive and its
+ isolated CPUs can be distributed down sibling cgroups even
+ though they may not appear in their "cpuset.cpus.effective".
+
+ A valid "isolated" partition root can pull isolated CPUs from
+ either its parent partition or from the "isolcpus" partition.
+ It also requires the following conditions to be met.
+
+ 1) The "cpuset.cpus" is exclusive with its siblings , i.e. they
+ are not shared by any of its siblings (exclusivity rule).
+ 2) The "cpuset.cpus" is not empty and must be a subset of
+ parent's "cpuset.cpus".
+ 3) The "cpuset.cpus.effective" cannot be empty unless there is
+ no task associated with this partition.
+
+ If pulling isolated CPUS from "isolcpus" partition,
+ the "cpuset.cpus" must also be a subset of "isolcpus"
+ partition's "cpuset.cpus" and all the requested CPUs must
+ be available for pulling, i.e. in "isolcpus" partition's
+ "cpuset.cpus.effective". In this case, its hierarchical parent
+ does not need to be a valid partition root.
+
External events like hotplug or changes to "cpuset.cpus" can
cause a valid partition root to become invalid and vice versa.
Note that a task cannot be moved to a cgroup with empty
This patch updates the cgroup-v2.rst file to include information about the new "isolcpus" partition type. Signed-off-by: Waiman Long <longman@redhat.com> --- Documentation/admin-guide/cgroup-v2.rst | 89 +++++++++++++++++++------ 1 file changed, 70 insertions(+), 19 deletions(-)