sched/cache: Keep nr_pref_llc_running in the runnable domain, to fix LLC mis-scheduling buggrafted

alb_break_llc() decides whether to break LLC preference during active
load balance. It does so by testing that every runnable fair task on the
source rq prefers its LLC:

	env->src_rq->nr_pref_llc_running == env->src_rq->cfs.h_nr_runnable

But the two counters cover different sets. nr_pref_llc_running is updated
in account_llc_enqueue()/account_llc_dequeue(), next to cfs_rq->nr_queued,
so it follows queued tasks. h_nr_runnable is updated in set_delayed()/
clear_delayed() and drops delay-dequeued tasks.

So under DELAY_DEQUEUE, a preferring task that goes to sleep stays counted
in nr_pref_llc_running while h_nr_runnable falls. The equality then breaks,
alb_break_llc() returns false, and active balance is free to pull a task
off its preferred LLC. Active balance only moves runnable tasks, and this
is the only LLC check it consults: once the stopper runs, LBF_ACTIVE_LB
skips the per-task test in can_migrate_task(). The runnable set is the one
we want.

Fix it on the counter side. A task should be counted in
nr_pref_llc_running exactly while it is both queued on its preferred LLC
(pref_llc_queued) and runnable (!sched_delayed). Define that membership
once in task_pref_llc_runnable(), and adjust the counter only through
pref_llc_running_inc()/pref_llc_running_dec() from the four sites that
change either input: account_llc_enqueue(), account_llc_dequeue(),
set_delayed() and clear_delayed(). Gating every update on the same
predicate keeps the delay, wake and dequeue paths from double-counting
or underflowing; see the comments at those sites for the ordering.

nr_llc_running and sd->llc_counts are not touched and stay on queued
semantics.

Fixes: 714059f79ff0 ("sched/cache: Handle moving single tasks to/from their preferred LLC")
Closes: https://lore.kernel.org/lkml/20260827135000.735138-1-zhanxusheng@xiaomi.com/
Reported-by: Zhan Xusheng <zhanxusheng@xiaomi.com>
Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Tim Chen <tim.c.chen@linux.intel.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kayra Cizmeci <kayracizmeci@gmail.com>
Cc: <stable@kernel.org> # v7.2.x
Link: https://patch.msgid.link/06af61afedac32e6477f57feb4d658f6c411c3af.1790035273.git.tim.c.chen@linux.intel.com

1 files changed, 220 insertions, 0 deletions

diff --git a/Documentation/gpu/drm-usage-stats.rst b/Documentation/gpu/drm-usage-stats.rst
new file mode 100644
index 000000000..70b7cfcc1
--- /dev/null
+++ b/Documentation/gpu/drm-usage-stats.rst
@@ -0,0 +1,220 @@
+.. _drm-client-usage-stats:
+
+======================
+DRM client usage stats
+======================
+
+DRM drivers can choose to export partly standardised text output via the
+`fops->show_fdinfo()` as part of the driver specific file operations registered
+in the `struct drm_driver` object registered with the DRM core.
+
+One purpose of this output is to enable writing as generic as practically
+feasible `top(1)` like userspace monitoring tools.
+
+Given the differences between various DRM drivers the specification of the
+output is split between common and driver specific parts. Having said that,
+wherever possible effort should still be made to standardise as much as
+possible.
+
+.. contents::
+
+File format specification
+=========================
+
+- File shall contain one key value pair per one line of text.
+- Colon character (`:`) must be used to delimit keys and values.
+- All standardised keys shall be prefixed with `drm-`.
+- Driver-specific keys shall be prefixed with `driver_name-`, where
+  driver_name should ideally be the same as the `name` field in
+  `struct drm_driver`, although this is not mandatory.
+- Whitespace between the delimiter and first non-whitespace character shall be
+  ignored when parsing.
+- Keys are not allowed to contain whitespace characters.
+- Numerical key value pairs can end with optional unit string.
+- Data type of the value is fixed as defined in the specification.
+
+Key types
+---------
+
+1. Mandatory, fully standardised.
+2. Optional, fully standardised.
+3. Driver specific.
+
+Data types
+----------
+
+- <uint> - Unsigned integer without defining the maximum value.
+- <keystr> - String excluding any above defined reserved characters or whitespace.
+- <valstr> - String.
+
+Mandatory fully standardised keys
+---------------------------------
+
+- drm-driver: <valstr>
+
+String shall contain the name this driver registered as via the respective
+`struct drm_driver` data structure.
+
+Optional fully standardised keys
+--------------------------------
+
+Identification
+^^^^^^^^^^^^^^
+
+- drm-pdev: <aaaa:bb.cc.d>
+
+For PCI devices this should contain the PCI slot address of the device in
+question.
+
+- drm-client-id: <uint>
+
+Unique value relating to the open DRM file descriptor used to distinguish
+duplicated and shared file descriptors. Conceptually the value should map 1:1
+to the in kernel representation of `struct drm_file` instances.
+
+Uniqueness of the value shall be either globally unique, or unique within the
+scope of each device, in which case `drm-pdev` shall be present as well.
+
+Userspace should make sure to not double account any usage statistics by using
+the above described criteria in order to associate data to individual clients.
+
+- drm-client-name: <valstr>
+
+String optionally set by userspace using DRM_IOCTL_SET_CLIENT_NAME.
+
+
+Utilization
+^^^^^^^^^^^
+
+- drm-engine-<keystr>: <uint> ns
+
+GPUs usually contain multiple execution engines. Each shall be given a stable
+and unique name (keystr), with possible values documented in the driver specific
+documentation.
+
+Value shall be in specified time units which the respective GPU engine spent
+busy executing workloads belonging to this client.
+
+Values are not required to be constantly monotonic if it makes the driver
+implementation easier, but are required to catch up with the previously reported
+larger value within a reasonable period. Upon observing a value lower than what
+was previously read, userspace is expected to stay with that larger previous
+value until a monotonic update is seen.
+
+- drm-engine-capacity-<keystr>: <uint>
+
+Engine identifier string must be the same as the one specified in the
+drm-engine-<keystr> tag and shall contain a greater than zero number in case the
+exported engine corresponds to a group of identical hardware engines.
+
+In the absence of this tag parser shall assume capacity of one. Zero capacity
+is not allowed.
+
+- drm-cycles-<keystr>: <uint>
+
+Engine identifier string must be the same as the one specified in the
+drm-engine-<keystr> tag and shall contain the number of busy cycles for the given
+engine.
+
+Values are not required to be constantly monotonic if it makes the driver
+implementation easier, but are required to catch up with the previously reported
+larger value within a reasonable period. Upon observing a value lower than what
+was previously read, userspace is expected to stay with that larger previous
+value until a monotonic update is seen.
+
+- drm-total-cycles-<keystr>: <uint>
+
+Engine identifier string must be the same as the one specified in the
+drm-cycles-<keystr> tag and shall contain the total number cycles for the given
+engine.
+
+This is a timestamp in GPU unspecified unit that matches the update rate
+of drm-cycles-<keystr>. For drivers that implement this interface, the engine
+utilization can be calculated entirely on the GPU clock domain, without
+considering the CPU sleep time between 2 samples.
+
+A driver may implement either this key or drm-maxfreq-<keystr>, but not both.
+
+- drm-maxfreq-<keystr>: <uint> [Hz|MHz|KHz]
+
+Engine identifier string must be the same as the one specified in the
+drm-engine-<keystr> tag and shall contain the maximum frequency for the given
+engine.  Taken together with drm-cycles-<keystr>, this can be used to calculate
+percentage utilization of the engine, whereas drm-engine-<keystr> only reflects
+time active without considering what frequency the engine is operating as a
+percentage of its maximum frequency.
+
+A driver may implement either this key or drm-total-cycles-<keystr>, but not
+both.
+
+Memory
+^^^^^^
+
+Each possible memory type which can be used to store buffer objects by the GPU
+in question shall be given a stable and unique name to be used as the "<region>"
+string.
+
+The region name "memory" is reserved to refer to normal system memory.
+
+The value shall reflect the amount of storage currently consumed by the buffer
+objects belong to this client, in the respective memory region.
+
+Default unit shall be bytes with optional unit specifiers of 'KiB' or 'MiB'
+indicating kibi- or mebi-bytes.
+
+- drm-total-<region>: <uint> [KiB|MiB]
+
+The total size of all requested buffers, including both shared and private
+memory. The backing store for the buffers does not need to be currently
+instantiated to count under this category. To avoid double-counting, if a buffer
+has multiple regions where it can be allocated to, the implementation should
+consistently select a single region for accounting purposes.
+
+- drm-shared-<region>: <uint> [KiB|MiB]
+
+The total size of buffers that are shared with another file (i.e., have more
+than one handle). The same requirement to avoid double-counting that applies to
+drm-total-<region> also applies here.
+
+- drm-resident-<region>: <uint> [KiB|MiB]
+
+The total size of buffers that are resident (i.e., have their backing store
+present or instantiated) in the specified region.
+
+- drm-memory-<region>: <uint> [KiB|MiB]
+
+This key is deprecated and is only printed by amdgpu; it is an alias for
+drm-resident-<region>.
+
+- drm-purgeable-<region>: <uint> [KiB|MiB]
+
+The total size of buffers that are resident and purgeable.
+
+For example, drivers that implement functionality similar to 'madvise' can count
+buffers that have instantiated backing stores but have been marked with an
+equivalent of MADV_DONTNEED.
+
+- drm-active-<region>: <uint> [KiB|MiB]
+
+The total size of buffers that are active on one or more engines.
+
+One practical example of this could be the presence of unsignaled fences in a
+GEM buffer reservation object. Therefore, the active category is a subset of the
+resident category.
+
+Implementation Details
+======================
+
+Drivers should use drm_show_fdinfo() in their `struct file_operations`, and
+implement &drm_driver.show_fdinfo if they wish to provide any stats which
+are not provided by drm_show_fdinfo().  But even driver specific stats should
+be documented above and where possible, aligned with other drivers.
+
+Driver specific implementations
+-------------------------------
+
+* :ref:`i915-usage-stats`
+* :ref:`panfrost-usage-stats`
+* :ref:`panthor-usage-stats`
+* :ref:`xe-usage-stats`
+* :ref:`amdxdna-usage-stats`