diff options
| author | Robert Killick <robert@killick.dev> | 2026-10-05 14:56:45 +0100 |
|---|---|---|
| committer | Sebastian Reichel <sebastian.reichel@collabora.com> | 2026-10-17 23:30:06 +0200 |
| commit | 4d49915df53ddde71556e507bdc49b81cd4609c8 (patch) | |
| tree | 2c2b030c966a905bfdbd1a427fa3940c106fe779 /rust/kernel/alloc | |
| parent | 1f2a40f954c848dff9d8bb6d81e404f9e47f0e81 (diff) | |
power: supply: sysfs: Ratelimit "unavailable enum value" warning
power_supply_show_enum_with_available() warns every time a driver
reports a value that is not in its declared set of available values.
For usb_type this happens on every sysfs read and on every uevent, so a
driver that reports such a value and calls power_supply_changed()
frequently can flood the kernel log.
This can happen with drivers that pass a firmware-provided value
through unchanged, as qcom_battmgr does for usb_type. On a OnePlus 10
Pro (SM8450), whose firmware reports a vendor-specific usb_type, this
produced ~128 warnings per second and kept systemd-journald at ~30% CPU.
The bug that causes the bad value needs fixing in the driver itself,
but one misbehaving driver should not be able to flood the kernel log.
Use dev_warn_ratelimited(), as power_supply_show_property() already
does for its own errors, so the problem is still reported, just not on
every read.
Signed-off-by: Robert Killick <robert@killick.dev>
Link: https://patch.msgid.link/20261005-ratelimit-enum-warn-v1-1-b77b48de3c47@killick.dev
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Diffstat (limited to 'rust/kernel/alloc')
0 files changed, 0 insertions, 0 deletions
