diff options
| author | Jiri Kosina <jkosina@suse.com> | 2025-03-26 13:42:07 +0100 |
|---|---|---|
| committer | Jiri Kosina <jkosina@suse.com> | 2025-03-26 13:42:07 +0100 |
| commit | b3cc7428a32202936904b5b07cf9f135025bafd6 (patch) | |
| tree | d4a1a6180ac5939fccd92acd6f8d7d1388575c4a /kernel/time/timer_migration.c | |
| parent | db52926fb0be40e1d588a346df73f5ea3a34a4c6 (diff) | |
| parent | 01601fdd40ecf4467c8ae4d215dbb7d2a0599a2c (diff) | |
Merge branch 'for-6.15/amd_sfh' into for-linus
From: Mario Limonciello <mario.limonciello@amd.com>
Some platforms include a human presence detection (HPD) sensor. When
enabled and a user is detected a wake event will be emitted from the
sensor fusion hub that software can react to.
Example use cases are "wake from suspend on approach" or to "lock
when leaving".
This is currently enabled by default on supported systems, but users
can't control it. This essentially means that wake on approach is
enabled which is a really surprising behavior to users that don't
expect it.
Instead of defaulting to enabled add a sysfs knob that users can
use to enable the feature if desirable and set it to disabled by
default.
Diffstat (limited to 'kernel/time/timer_migration.c')
| -rw-r--r-- | kernel/time/timer_migration.c | 10 |
1 files changed, 9 insertions, 1 deletions
diff --git a/kernel/time/timer_migration.c b/kernel/time/timer_migration.c index 9cb9b6584ea18..2f6330831f084 100644 --- a/kernel/time/timer_migration.c +++ b/kernel/time/timer_migration.c @@ -1675,6 +1675,9 @@ static int tmigr_setup_groups(unsigned int cpu, unsigned int node) } while (i < tmigr_hierarchy_levels); + /* Assert single root */ + WARN_ON_ONCE(!err && !group->parent && !list_is_singular(&tmigr_level_list[top])); + while (i > 0) { group = stack[--i]; @@ -1716,7 +1719,12 @@ static int tmigr_setup_groups(unsigned int cpu, unsigned int node) WARN_ON_ONCE(top == 0); lvllist = &tmigr_level_list[top]; - if (group->num_children == 1 && list_is_singular(lvllist)) { + + /* + * Newly created root level should have accounted the upcoming + * CPU's child group and pre-accounted the old root. + */ + if (group->num_children == 2 && list_is_singular(lvllist)) { /* * The target CPU must never do the prepare work, except * on early boot when the boot CPU is the target. Otherwise |