Was ist blu_plug?
[KERNEL] [4.4.4] Code_Blue [r450] [Linaro 2014.09 Cortex-A15 GCC 4.9.2]
-
-
Zitat
Originally Posted by dhalham
May I ask what is bluplug? Or hot plug even and how does it benefit us?Zitat von AerowinderPlease call it blu_plug. When I read "bluplug" I thought it said something very different! :laugh:
It is a better implementation of Google's logic that tells your 4 cores when they should be online, and when they should be offline.
-
Danke. Na dann bin ich mal auf die Auswirkungen gespannt - und darauf was ihr zu berichten habt.
-
-
Code_Blue r316
Zitat
Wicked!Finally pushing out blu_plug: Dynamic hotplug for hammerhead
Lot's and lot's of hours coding and testing, very proud of the outcome.
Special thanks to my testers and apb_axel for making UKM tottaly compatible with blu_plug and Code_Blue.blu_plug is still WIP, stay tune for more nice stuff in upcoming weeks
Both my cpu governor blu_active and hotplug blu_plug were forked to other devices like Nexus 4 and LG pad and so on.
Damn proud to be BLUE! ANd just the other day, a naive guy called me insecure...Keep it real, keep it true, keep it #Code_Blue
Changelog
Zitat
[size=14]Code_Blue r316[/SIZE]
- blu_plug: Dynamic hotplug driver for hammerhead (enabled by default min 2 core online, max 1 core screenoff) , Driver recode and rework, Tune initial values, Improve suspend/resume routines, down_timer_cnt can't be lower than up__timer_cnt, missing bits, Do not offline a cpu if it's boosted by cpu-boost, Set up_threshold to 70, let's try without jiffies
- blu_active: Since we got blu_plug, let's rename and keep it linear, let's add the file, shall we?
- cpufreq: create a helper function to get CPU utilization. Picked from @stratosk
- Revert "[backport] sched_rt: Upstream fixes"
- Revert "bluactive: let's try and live without the idle notifier and rely on the timer instead."
- zlib: clean up some dead code
- netlink: add signed types
- net/l2tp: don't fall back on UDP [get|set]sockopt
- Tuned some cpufreq tunables, set sched_mc_power_savings to 2
- Use UKM v3.3.1 or higher to control all tunables -
Danke
ich hab den Opener auch mal wieder aktualisiert
-
Da kommt was feines
Wie hat sich blu_plug jetzt bemerkbar gemacht?
-
Code_Blue r350
- makefile: use arm-cortex_a15-linux-gnueabihf-linaro_4.9.2-2014.08 toolchain by default
- blu_plug: driver is always enabled, add max_freq_screenoff (1190 MHz by default), add powersaver mode (1 core online @ 1728 MHz switch), reworked suspend/resume, fix powersaver_mode when screenoff to save more juice, misc fixes
- dt2w: add doubletap2wake (disabled by default), Prevent touches during phone calls, Power button also disable irq_wake in touch driver, fix unbalanced irq_wake, Option to disable with power button, add wake_timeout to disable wake functions after a determined period, prevent end call when wake_timeout and end call with power button are enabled
- CHROMIUM: clocksource: allow clocksources to be selected earlier during boot
- Revert "net: wireless: bcmdhd: Set KSO enable sleep to 5ms"
- Tuned blu_active & blu_plug tunablesNexus 5 - Purity - Hells
-
Thread ist mal wieder aktualisiert
-
Code_Blue r376
- blu_active: rework driver and merge upstream, Linux 3.5 improvements, use chose_freq
- blu_plug: remove deprecated cpu methods, faster cancel work
- intelli-thermal v2: initial adaptation
- msm: thermal: Update thermal subsystem
- cpufreq: Manage only online cpus, Remove unnecessary initialization of a local variable, Fix sparse warning by making local function static, return early from __cpufreq_driver_getavg()
- cpufreq / core: Fix typo in comment describing show_bios_limit()
- Sound Control: add line gain to cached and register lock
- ARM: dts: hammerhead: Change i2c clock rate to 19.2 MHz
- exFAT support, enable missing filesystems
- futex: Make lookup_pi_state more robust, Always cleanup owner tid in unlock_pi, Validate atomic acquisition in futex_lock_pi_atomic(), futex-prevent-requeue-pi-on-same-futex.patch futex, Add another early deadlock detection check, Prevent attaching to kernel threads
- stop_machine: Mark per cpu stopper enabled early, Use smpboot threads, Store task reference in a separate per cpu variable
- smp: Add generic smpboot facility, Provide generic idle thread allocation
- smpboot: Allow selfparking per cpu threads, Provide infrastructure for percpu hotplug threads
- smpboot, idle: Fix comment mismatch over idle_threads_init(), Optimize calls to smp_processor_id() in idle_threads_init(), Allocate idle thread for each possible cpu during boot
- hotplug: Fix UP bug in smpboot hotplug code
- sched: Fix memory leakage in build_sched_groups(), Fix clear NOHZ_BALANCE_KICK, Micro-optimize the smart wake-affine logic, Implement smarter wake-affine logic, Fix select_idle_sibling() bouncing cow syndrome, add wait_for_completion_io[_timeout]
- sched/debug: Fix sd->*_idx limit range avoiding overflow, Limit sd->*_idx range on sysctl
- sched/cpuidle: reduce IPI store, Backport upstream 3.16 scheduler updates
- kthread: Implement park/unpark facility, Prevent unpark race which puts threads on the wrong cpu
- kthread_worker: reimplement flush_kthread_work() to allow freeing the work item being executed, reorganize to prepare for flush_kthread_work() reimplementation
- writeback: Fix occasional slow sync(1), fix race that cause writeback hung, fix writeback cache thrashing
- tick: Dynamically set broadcast irq affinity
- AIO: Don't plug the I/O queue in do_io_submit()
- block: Reserve only one queue tag for sync IO if only 3 tags are available, account iowait time when waiting for completion of IO request
- switch do_fsync() to fget_light()
- nohz: Reduce overhead under high-freq idling patterns
- init: debug: bye bye
- Bluetooth: fix possible info leak in bt_sock_recvmsg()
- sync: don't block the flusher thread waiting on IO
- arm/dts: hammerhead: Remove erroneous qcom,msm-bus,active-only property
- wcnss: add null check in pm_ops unregister.
- Revert "dts: change back to sensor 0, sensor 5 specifically tracks cpu0.", "cpufreq: Introduce new relation for freq selection", "epoll: use freezable blocking call"Nexus 5 - Purity - hellscore
-
Code_Blue_r398
- Milestone 6 release
- Build with Linaro 2014.09 Cortex-A15 GCC 4.9.2
- Compressed with LZ4-r122
- blu_active: merge Linux 3.5 upstream and CAF updates, remove variable frequency calculations, use cal_freq, remove target_loads, add down load threashold to keep cal_freq better on low loads, use get_cpu_idle_time_cb
- cpufreq: conservative: add load calc
- cgroup: fix exit() vs rmdir() race
- msm: Create shadow files under /sys/kernel/ipv4
- WLAN subsystem: Sysctl support for key TCP/IP parameters
- net: wireless: bcmdhd: Disable OOB interrupt when WLAN is off, Add WIPHY_WOWLAN_ANY support
- SELinux: possible NULL deref in context_struct_to_string, use ARRAY_SIZE, Reduce overhead of mls_level_isvalid() function call
- Ramdisk updates, refresh tunables
- Revert "qup_i2c: Initialize I2C resource before registering.", "block: Disable IO Stats"Nexus 5 - Purity - hellscore
-
Habe den Opener mit Erlaubnis von @Eisbaer-Dan aktualisiert.
-
Zitat
Code_Blue r407
All right crew!Just pushed r407, go get it. Now.
Add the ability of touch boost to blu_plug (now with 1 core min, 2 cores on touch), you can use older setting of 2 always online if you prefer.
Also I tweaked thermal levels to match more conservative values.CM branch has all the goodies merged now, but I am not happy with dt2w in either branches: I still notice some leakages in some particular situations, will test more during next week. If I am not happy with the results, I am going to remove touchy touch stuff from my kernels.
Some fun facts: r398 just hit the amazing 5,2k hits. Damn
Get involved, test, play, donate, help, contribute.
You keep the spirit alive!Keep it real, keep it true, keep it #Code_Blue
ungültiger Link
-
Vom 407 habe ich bis jetzt nichts gutes gehört (Ruckler etc.)
-
Er und seine Tester,
haben keine lags .............
sagt er;) -
Auf dem AH Forum haben viele (alle)
lags.
Wollte es nur mal weiter geben -
Über die Qualität dieser "Bugreports" lässt sich auch auch trefflich streiten...
Kein einziger nennt die Rom, sonstige installierte mods und/oder Tweaks (Xposed, UV etc). -
Zitat
Information about the lag with Build 407
The issues with 407 (some of you report lags) are located, a updated build with the lag fix is coming soon, no ETA, It will work as quickly as possible.
Be patient.
ungültiger Link
Zitat
All right crew!input boost bug on blu_plug is fixed. Big shouts to my blue beta posse :highfive:
Both AOSP and CM builds were updated (although the bug in CM didn't show). Never settle for nothing more than perfection.
I also made an improvement on the process, let's see if anybody notices it...Get involved, test, play, donate, help, contribute.
You keep the spirit alive!
ungültiger Link -
LOL...
-
Dafür funktioniert bei Version 412 jetzt der DeepSleep leider nicht mehr, ... ich warte auf eine neue Version, solange kommt wieder die 350 / 376 / 398 drauf.
-
Jetzt mitmachen!
Sie haben noch kein Benutzerkonto auf unserer Seite? Registrieren Sie sich kostenlos und nehmen Sie an unserer Community teil!