There are two ways for the kernel to load iris5.fw. The first way is
via fw_get_filesystem_firmware(), which read iris5.fw directly from the
filesystem using filp_read(). The second way gets FW via ueventd, which
calls request_firmware().
Because the kernel unconditionally calls fw_get_filesystem_firmware()
for iris5, the first way always gets executed. However, the current
sepolicy denies read access to iris5.fw for the kernel, causing this way
to never work. The second way is used only on ioctl(), which would
likely be sent from HWC. Unlike the first way, the second way *does*
work as-is, and since HWC is always started when booting into Android,
we don't have to worry about the first way being broken.
However, HWC doesn't run offline charging mode, so we have to rely on
the first way for that case. Unfortunately, since the first way is
broken, offline charging fails to work properly as a result. As such, we
need to fix that denial to make the first way work again.
Change-Id: Iba43afc5dba1c64d8ced9f374bbf9f0457db3c92
For aw8697_haptic, OOS 12 never really properly implemented any effect
other than CLICK and HEAVY_CLICK. Let's implement our own TEXTURE_TICK,
TICK, CLICK, and HEAVY_CLICK so that the user can actually feel a
noticeable difference between each haptic effect.
As a quick summary:
- CLICK: same as oplus CLICK, but with vmax set to 2500.
- DOUBLE_CLICK: replaced with our CLICK played twice in a row, w/ a
100ms delay
- TICK: same as oplus CLICK, but with vmax set to 1400.
- HEAVY_CLICK: same as oplus CLICK, but with vmax set to 2500 and
waveform_index set to 4.
- TEXTURE_TICK: same as oplus CLICK, but with vmax set to 60 and
waveform_index set to 2.
In addition, modify LedVibratorDevice's on() to better reflect the
timeout duration actually being passed in to the function. The original
tried to change waveforms instead for various durations at 102 ms or
lower, which didn't really work out in practice for our vibrator.
Change-Id: Ifb769bc2292774afa4977026787a7882430c09d0
It seems that the DOUBLE_CLICK and HEAVY_CLICK effects assume that
`waveform_index` is always set to 1 and `vmax` is set to 1600, because
if the last played effect is TICK, then all the other effects (aside
from CLICK) end up feeling different, presumably because
`waveform_index` is left as set to 2 and `vmax` is left as set to 1600.
Fix this by having DOUBLE_CLICK and HEAVY_CLICK set `waveform_index`
back to 1 and `vmax` back to 1600.
Change-Id: Ic116f45894cf4af25bcd25b66764eeb472631af5
The ndk_platform backend will soon be deprecated because the ndk backend
can serve the same purpose. This is to eliminate the confusion about
having two variants (ndk and ndk_platform) for the same ndk backend.
Bug: 161456198
Test: m
Merged-In: I14a1c57bd06f1f2aa52491f779c7030d4de03547
Change-Id: I7b9055909b901843631314b9577a9e578a739d00
The AOSP TICK effect generated from the following test app:
- https://github.com/luk1337/VibeTest
was used as the starting point. It is more or less the same as oplus
CLICK, except waveform_index is set to 2.
The vibration effect from this TICK was stronger than oplus CLICK, which
is not really desirable, so adjustments were made to vmax in an
unscientific manner until it felt and sounded just right.
Change-Id: I472db022c2b0182e9a0d544c9d439bc3e47e4d82
