* There is zero reason to ever build without blobs. We've even seen
this situation with official builds from our servers. It's always
better for a build to fail than it is for it to produce something
that has no chance at working.
Change-Id: I4968795670c91f691e9ecdc0e4af62e16ba3a93a
media_profiles_V1_0.xml is created from media_profiles.xml by replacing
the dtd schema inside media_profiles.xml with an external reference to
/system/etc/media_profiles_V1_0.dtd.
Change-Id: I0fbb03603a72d0108a96bfa0f0a4ceb12ef3d430
We have issues with camera blobs which doesn't play
well with scudo allocation/deallocation tracking.
Scudo reports corrupted header when stopping a
video recording.
Because we aren't alone with this (also upstream seems
to have trouble with camera modules) use the 32 libc
which is used only by a few components, like camera
and audio.
Because docs states, that libc with the svelte take less RSS,
but is a little bit slower.
https://android.googlesource.com/platform/bionic/+/master/docs/native_allocator.md
This is the reason to not enable this target wide.
Change-Id: Ic0d950817238a227218f59b95e5fd2ffd66e786a
Also remove the widevine plugin we don't have
Reference:
LineageOS/android_hardware_interfaces@lineage-17.0/drm/1.1/README.md
Change-Id: Ie4ea8d79496ba6181ef6c45b0ce0a993a3e442c4
* So, this one from bullhead can cause everything to get janky AF
every time RenderScript is actually used, and finding one that
actually loads (no missing symbols) AND actually works well is,
well, challenging. The device works fine with the normal
non-adreno driver in AOSP, so let's use that until/unless we find
an actually working adreno driver.
Change-Id: I4ab19269eb49f5f2aed375d9055d68c1b9835f69
…nifest
* The cmd_list node exists, and it contains hover_mode, so
isSupported() returns true. Not having a supported interface
declared leads to boot-breaks starting with R.
Change-Id: If2081dc796ff1a33043d9894aa556028d4aa2d68
What this do is only write to core_ctl nodes every time the
screen turns on and off, as the new Lineage power HAL decides
to just change the migration value rather than messing with
cores, we may drop this as well.
Change-Id: I318268842aa0a4680883990bb8076e224ca9234d
This addresses an underlying problem since /dev/hw_random is used. The
issue is that EntropyMixer, which runs in system_server, needs to read
from /dev/hw_random, but thanks to our qualcomm ueventd, EntropyMixer
can never accress this node, resulting in warnings like this:
W EntropyMixer: Failed to add HW RNG output to entropy pool
W EntropyMixer: java.io.FileNotFoundException: /dev/hw_random (Permission denied)
AOSP ueventd handles this well, so just remove this line to get things
back to stage. This issue was never exposed because Google derps the
logging here, which is already fixed on Pie.
c78a463e87
Change-Id: I42c9aaa8f57306536e51c0a17febad57a5a77581
* Apparently this is now required on all non full treble devices since
google added a sleep(1) in libhidl...
Change-Id: I6d8492dd7c56bfdb2ca00f3477634a46fa55b1c0
In Q, this looks like it's making the system wait for a
specific widevine service, which we don't have.
Removing this appears to make L3 start working again.
Change-Id: Ib410029519e1d3cb0f0f3a5aa553976b36bca8c3
This CL removes the definition for the hostapd hidl interface from
manifest.xml and rely on vint-fragements entry.
This is to avoid the need to modify the revision in future
Bug: 145780363
Test: builds and wifi softap works
Change-Id: I2a9cb25209efc94f6306f38d89cb5e1ff6f72bed
This CL removes the definition for the IWifi hidl interface from
manifest.xml and rely on vint-fragements entry.
This is to avoid the need to modify the revision in future
Bug: 138243400
Test: builds and wifi works
Change-Id: Icaff93193e10d4953d9947d2739b05959ece79f1
Shipping API level less than 26 corresponds to legacy FCM version.
This addresses the following build warning:
Warning: Shipping FCM Version is inferred from Shipping API level. Declare Shipping FCM Version in device manifest directly.
Change-Id: Ib230c345ff7deb552597824838b8809ceefbbe8a
* AOSP removed this as a platform-wide rule since it is "...not
present on Pixel devices". As it turns out, our RIL gets grumpy
and won't latch a data connection if it cannot open the device.
* own/mode aligned to AOSP values from before the change
* E RILD : ElapsedRealTime() Cannot open file /dev/alarm
* ref: LineageOS/android_system_core@16cdffe
Change-Id: Ib369e4b76653a412d75a3c8c37ee283f8c8e9239
