summaryrefslogtreecommitdiff
path: root/tools/perf/scripts/python/bin/stackcollapse-record
diff options
context:
space:
mode:
authorRong Zhang <i@rong.moe>2026-07-18 07:10:19 +0800
committerRafael J. Wysocki <rafael.j.wysocki@intel.com>2026-07-21 21:55:02 +0200
commit69e81ddfee7603124b7f4e2312dacd988bd82e15 (patch)
tree5383bd2b3bec383cfce94bd77702f40709a56c42 /tools/perf/scripts/python/bin/stackcollapse-record
parent6587dd87595cfae6bf1304cd0cd1df2274424e42 (diff)
ACPI: battery: Merge consecutive battery notifications
It's a very common pattern to emit consecutive battery notifications, for example: Method (_Qxx, 0, NotSerialized) { Notify (BAT0, 0x80) // Status Change Notify (BAT0, 0x81) // Information Change } In this case, the current code path will update battery state twice within a short period, which is not optimal, as the same data are fetched twice. Moreover, both notifications are likely to call power_supply_changed(), causing power_supply_uevent() to read all battery properties in order to assemble uevents. Even worse, after the first uevent reaches userspace, some userspace processes start to read all battery properties in order to refresh their internal states, which competes with the second notification's handling and uevent assembling. This generates significant pressure on _STA, _BST and _BIX/_BIF methods. Not only that, power_supply_ext properties may also rely on some other ACPI methods, so both uevent assembling and userspace processes call them. It becomes a nightmare when all these methods share the same ACPI mutex protecting EC accesses and hence vulnerable to lock starvation. This is exactly the case of some Lenovo devices, where the mentioned EC query pattern eventually leads to a catastrophic situation that a bunch of ACPI methods (including but not limited to the mentioned ones) fail to acquire the same mutex due to timeout. These devices don't handle mutex acquisition failure gracefully and return garbage data, causing even more chaos. Improve battery notification handling by merging at most 16 consecutive battery notifications within 10ms using a delayed work, so that they only refresh and/or update battery state once. ACPI netlink event and notifier call chain are still triggered multiple times in order not to break other components. Finally, call power_supply_changed() once and lead to a single uevent instead of a bunch, preventing userspace programs from causing too much pressure on power supply properties and underlying ACPI methods. If more than 16 battery notifications are queued within 10ms, the firmware/hardware is anyway buggy, and extra notifications will be dropped. Tested-by: Jeffrey Wälti <jeffrey@waelti.dev> Tested-by: Avraham Hollander <anhollander516@gmail.com> Reported-by: Rick <rickk1166@gmail.com> Closes: https://bugzilla.kernel.org/show_bug.cgi?id=221065 Signed-off-by: Rong Zhang <i@rong.moe> Link: https://patch.msgid.link/20260718-b4-acpi-battery-notification-v4-1-599c8ed1072f@rong.moe Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Diffstat (limited to 'tools/perf/scripts/python/bin/stackcollapse-record')
0 files changed, 0 insertions, 0 deletions