diff options
| author | Tejun Heo <tj@kernel.org> | 2026-09-14 22:12:34 -1000 |
|---|---|---|
| committer | Tejun Heo <tj@kernel.org> | 2026-09-15 06:57:20 -1000 |
| commit | a9e3760b0838299649c0d57cca44daaf40ba3c33 (patch) | |
| tree | 95807ff0e511c05602ef34808cfc660303309057 /Documentation/ABI/testing/debugfs-tpmi | |
| download | linux-stable-a9e3760b0838299649c0d57cca44daaf40ba3c33.tar.gz linux-stable-a9e3760b0838299649c0d57cca44daaf40ba3c33.zip | |
sched_ext: Maintain an online cid mask in the scheduler arenagrafted
Schedulers on the default cid mapping treat [0, nr_online_cids) as the
online set and restart on hotplug. Schedulers that install their own mapping
with scx_bpf_cid_override() have no way to learn which cids are online: the
count no longer identifies members and the CPU-form cpumask is unusable from
cid programs. This is an obvious hole in the cid API.
Add scx_bpf_online_cmask(), a kernel-maintained cmask in the scheduler's
arena, allocated alongside the per-CPU scratch masks and populated after the
cid mapping is finalized and before ops.init(), for child schedulers too.
The pointer stays valid through ops.exit() with no reference to take. It is
the arena offset as a void pointer, the same form struct_ops arena arguments
arrive in. The verifier types the void return as a scalar for the program's
arena cast.
The mask follows the SCX hotplug notifications: seeded from cpu_active_mask
and updated before ops.cid_online/offline() runs, so it lags cpu_online_mask
only inside a hotplug transition. Updates walk the scheduler list under the
lock that also serializes unlinking. Reads are live, not atomic snapshots.
Root initialization excludes hotplug.
v2: Reworded the getter kerneldoc (Andrea Righi).
Signed-off-by: Tejun Heo <tj@kernel.org>
Reviewed-by: Andrea Righi <arighi@nvidia.com>
Diffstat (limited to 'Documentation/ABI/testing/debugfs-tpmi')
| -rw-r--r-- | Documentation/ABI/testing/debugfs-tpmi | 40 |
1 files changed, 40 insertions, 0 deletions
diff --git a/Documentation/ABI/testing/debugfs-tpmi b/Documentation/ABI/testing/debugfs-tpmi new file mode 100644 index 000000000..c493a1403 --- /dev/null +++ b/Documentation/ABI/testing/debugfs-tpmi @@ -0,0 +1,40 @@ +What: /sys/kernel/debug/tpmi-<n>/pfs_dump +Date: November 2023 +KernelVersion: 6.6 +Contact: srinivas.pandruvada@linux.intel.com +Description: +The PFS (PM Feature Structure) table, shows details of each power +management feature. This includes: +tpmi_id, number of entries, entry size, offset, vsec offset, lock status +and disabled status. +Users: Debugging, any user space test suite + +What: /sys/kernel/debug/tpmi-<n>/tpmi-id-<n>/mem_dump +Date: November 2023 +KernelVersion: 6.6 +Contact: srinivas.pandruvada@linux.intel.com +Description: +Shows the memory dump of the MMIO region for a TPMI ID. +Users: Debugging, any user space test suite + +What: /sys/kernel/debug/tpmi-<n>/tpmi-id-<n>/mem_write +Date: November 2023 +KernelVersion: 6.6 +Contact: srinivas.pandruvada@linux.intel.com +Description: +Allows to write at any offset. It doesn't check for Read/Write access +as hardware will not allow to write at read-only memory. This write is +at offset multiples of 4. The format is instance,offset,contents. +Example: +echo 0,0x20,0xff > mem_write +echo 1,64,64 > mem_write +Users: Debugging, any user space test suite + +What: /sys/kernel/debug/tpmi-<n>/plr/domain<n>/status +Date: Aug 2024 +KernelVersion: 6.11 +Contact: Tero Kristo <tero.kristo@linux.intel.com> +Description: +Shows the currently active Performance Limit Reasons for die level and the +individual CPUs under the die. The contents of this file are sticky, and +clearing all the statuses can be done by writing "0\n" to this file. |