On devices with cellular data available, I've been experiencing Wi-Fi
dropouts on 5 GHz networks where it disconnects and falls back to
cellular data around a RSSI of -77 dBm. While the Wi-Fi quality may not
be ideal at this signal level, it is still better to stay on it than
switch to cellular data because switching networks can be very
disruptive to the user.
To make matters worse, the signal tends to oscillate around -77 dBm in
my case, which causes it to oscillate between Wi-Fi and cellular data
every few seconds. This causes far more disruptions than staying on weak
Wi-Fi would.
These signal levels were measured empirically on a Pixel 5, but they
should apply to most devices. 2.4 GHz values were found to be more or
less accurate, but 5 GHz networks continued to work past the AOSP
thresholds. The iPhone 6s was also content with these signal levels and
still displayed 2 of 3 signal levels at -77 dBm.
Change-Id: I377be8374955530a5f6c084620460cac87e6a126
Signed-off-by: pix106 <sbordenave@gmail.com>
* In settings, it called "extends capatibility"
* packages/apps/Settings/src/com/android/settings/wifi/tether/WifiTetherMaximizeCompatibilityPreferenceController.java#L76 private boolean is5GhzBandSupported();
Signed-off-by: pix106 <sbordenave@gmail.com>
* gSoftApMaxPeers is not set in WCNSS_qcom-cfg.ini
* As kernel define it default 32 in drivers/staging/qcacld-3.0/components/mlme/dispatcher/inc/cfg_mlme_sap.h, make system settings same
Signed-off-by: pix106 <sbordenave@gmail.com>
* Yet another disablement we need, to have bluetooth working properly on A13 (QPR2)
Change-Id: Id3889d6310bac6d417ff2871d5e4a8c513e77354
Signed-off-by: pix106 <sbordenave@gmail.com>
* BOARD_HAVE_BLUETOOTH is uneeded.
* BOARD_HAVE_BLUETOOTH_QCOM is only important if
you build libbt-vendor which we don't.
Change-Id: Ib0465b3c0d5138a70cee6a3c3d5f08dd7ce9aa57
Signed-off-by: pix106 <sbordenave@gmail.com>
[basamaryan: This is needed for Android U to fix RIL]
Signed-off-by: basamaryan <basam.aryan@gmail.com>
Change-Id: Ie3fa610f71077b4ee2af1b4d57bd0c30b34f30fa
Signed-off-by: pix106 <sbordenave@gmail.com>
Mount of tracefs in /sys/kernel/debug/tracing will fail when DEBUG_FS
is disabled. So, mount tracefs in /sys/kernel/tracing to still use all the
tracing abilities when CONFIG_DEBUG_FS is disabled
Change-Id: Ib37332c3f1108d7c4798717f0f009c891db72850
Signed-off-by: pix106 <sbordenave@gmail.com>
* That makes clang17 happy
* Fix:
device/xiaomi/violet/gps/android/utils/battery_listener.cpp:191:13: error: misleading indentation; statement is not part of the previous 'if' [-Werror,-Wmisleading-indentation]
auto r = mHealth->unlinkToDeath(this);
^
device/xiaomi/violet/gps/android/utils/battery_listener.cpp:189:9: note: previous statement is here
if (mHealth != NULL)
^
1 warning and 1 error generated.
Change-Id: I3be1e3de45ce184c9d3dbcede57a312375a91498
Signed-off-by: Nanhumly <admin@warbler.cc>
Signed-off-by: pix106 <sbordenave@gmail.com>
* As https://source.android.com/docs/security/features/encryption/file-based said emmc_optimized is a inlinecrypt flavor and can only be applied on emmc 5.2 device which has not been released yet
* Apparently Xiaomi 660 is launched with emmc 5.1 5 years ago and never met the 5.2 spec
* Now really enables inlinecrypt hardware by adding it in mnt_flags
* Use AOSP recommended inlinecrypt_optimized
* Use lazytime for reducing meaningless io
* Label metadata partition
(pix106) i use logdump partition for metadata, not cust
Signed-off-by: pix106 <sbordenave@gmail.com>
Sdcardfs in deprecated
Enable Project ID Quota support on userdata.
* These features enable wide inodes on userdata, which
is required for project ID quotas. Wide inodes can
only be set at MKFS time.
* This also enables casefolding, which is required for the
removal of SDCardFS.
Bug: 138321217
Test: Examine filesystem features for project quota and wide inodes.
Change-Id: I9c17cda2af34c6fbd73c2d90eb6b465f4939df67
Signed-off-by: Omkar Chandorkar <gotenksIN@aosip.dev>
Signed-off-by: pix106 <sbordenave@gmail.com>
* as c999a1a2b9d975610ffa359ba34cb27d016165cc has unlocked setupwizard rotation, auto rotation default as "on" is not meaningless now
Signed-off-by: pix106 <sbordenave@gmail.com>
* as rom with vendor_gms repo will set this as true and locks the orientation for setup
* that is not so good for tablet type device
* override this prop in vendor_init to make it always landscape available
Signed-off-by: pix106 <sbordenave@gmail.com>
* allowing any extra permission for "untrustred_app" domain is DANGER
* the "untrustred_app" domain rule should ONLY be defined by aosp
* kill all don't audit except getopt for untrusted_app. it's a tool to show which app are evil, let it show in audit logs
Signed-off-by: pix106 <sbordenave@gmail.com>
