Skip to contentSkip to conversation

New: local struct uinput_event inconsistency with kernel y2038 safe struct input_event: AVRCP events vanish on 32-bit

Bug 215089 [email protected] 1 participant 1 message Nov 21, 2021 Bluetooth drivers

lore
Share & tools
lore
#0 lore raw

https://bugzilla.kernel.org/show_bug.cgi?id=215089

Bug ID: 215089
Summary: local struct uinput_event inconsistency with kernel
y2038 safe struct input_event: AVRCP events vanish on 32-bit
Product: Drivers
Version: 2.5
Kernel Version: 5.10
Hardware: All
OS: Linux
Tree: Mainline
Status: NEW
Severity: high
Priority: P1
Component: Bluetooth
Assignee: [email protected]
Reporter: [email protected]
Regression: No

Under bluez (up to 5.62) compiled with time64 support (like musl1.2 or possibly newer glibc with __USE_TIME_BITS64) on 32-bit systems (x86 or armhf for instance), AVRCP events just never show-up on input device.

Tools like evtest will not report any event upon bluetooth headset media button keypress, whereas btmon sees them flowing through.

This bug does not hit x86-64 releases, only 32-bit are affected.

Issue has been discussed here [1], with a simple & replicable Alpinelinux 3.14 test case (based on musl 1.2).
Legacy local uinput_event struct might not play well with newer time64-aware input_event kernel structure.

[1] https://github.com/bluez/bluez/issues/84#issuecomment-942155841

Signature · 4 lines
You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.