[KERNEL][N5] UBER Kernel

MuggelPu

Zitat

September 12, 2014
Well I decided to update again. While I was searching for call recording I also found a fix for S2W/DT2W firing during phone calls. Now you should be able to make phone calls when using S2W/DT2W without hanging up. So now you have both stereo call recording support and a bug fix for anyone using those screenwake features.
@Williamsss spent a number of hours working on allowing you to use custom gamma profiles with config for those that don't like my profile or stock colors. Here is how it works: Rename a gamma profile of your choice to uber_color_v1.3.cfg (yes name matters) and place them on your sdcard root directory - aka don't place it in a folder!!! then change custom gamma value in /system/etc/uber_v1.3.cfg from 1 (my preset) to 2 for custom) (it reminds you of the correct name and location in cfg too in case you forget) When v1.4 comes out you'll have to update your profile version as well. I'll try to remind you when that day comes.
I've now added uber_v1.3 config to ALL kernels including Release. Keep in mind Release has no GPU or CPU OC so when you see those values marked OC don't try them. Stick with stock frequencies. Everything else should work including UV, UC, colors, CPU governors, GPU Governors and all other kernel features other than OC.
Because this is a config update your old configs will be deleted and new ones will be placed. Make sure to backup your settings if you can't remember them off the top of your head. I would recommend you transfer values to the new configs (rather than just replacing them) just in case you see something new that you didn't know was available because it wasn't in v1.2 or v1.1 or whenever you last updated your configs.
Anyhow,
Enjoy the updates!

Alles anzeigen

Links

Main: http://uber.oceighty.co/HAMMERHEAD/UBER/

Mirror: http://d-h.st/users/cl3kener/HAMMERHEAD/KERNELS/UBER

Cheers!

Cl3Kener

Kernel Changelog

03c842c stereo call recording support
7181fbe Prevent touches during phone calls
ee674aa i2c: Drop missing things to 19.2 MHz
adb3859 Does anyone even read these changelogs anyways?
a4c8ba5 Make sure to like the OP if you haven't already.

Uber Config v1.3 Changelog

- Config now included in Release Kernels as well! (Cl3Kener)
- Custom Gamma Profiles (@Williamsss)
- SDCard cfgs (@Williamsss)
- Min # of CPUs Online (Cl3Kener)
- Minor Clean ups (@Williamsss)

MuggelPu

Update

Zitat

September 25, 2014
It's been a while since the last update. I've tried quite a few things like trying to merge upstream stuff from Google 3.4.y which was a very bad idea. It really bogged down kernel and caused freezes. That's the last time I'll trust Google 3.4.y common branch for updates. I have since reverted and brought more updates from kernel.org and other kernel developers working on this N5.
I've been hearing a lot of complaints recently but I haven't seen any logs with regards to these complaints so here is my thought for the day "If a tree falls in a forest and no one is around to hear it, does it make a sound?"
Seeing as I don't experience any of these so called "bugs" (I don't call them bugs simply because imao bugs are reproduceable and I cannot reproduce them) I can't get a log. If I can't get a log I don't know what is going wrong and therefore the "log" from the tree that fell (hopefully it didn't hit you) made no sound for me....

Anyhow, Here are the topics I'm noticing are trending on this Thread and my thoughts about them:
Sound
Overview: For some very strange reason many users look the to the kernel as the reason for any bug. Keep in mind that both the ROM and APP are also involved in sound. Some apps I've found do poorly with the current default screen off frequency and some play well. My advice is to try something that we know should work without any issues which is Google Play Music. If you can play your songs on that app without any issues it is most likely the APP and not the kernel. Also make sure not to use any sound mods during the process because they may also play a part in distortions.
Not Loud enough/Too loud? Many feel that my Quality settings on Faux Sound version are too quiet. There is an easy fix for that. Change the settings and make them loud. Franco has been said to be too loud on my kernel. There is also and easy fix for that change the settings and lower the volume levels.
Audio Clipping? I never have audio clipping. I almost blew my speakers testing things out to see if I could get audio clipping to happen for me but alas I could not. My guess is your issues are that if you are using Franco sound you have boosted kernel volume levels too high and need to back off. If you are running Faux Sound same thing applies (although default should settings should not have that issue). I had weird audio distortions while running Poweramp a while ago and switched to a simpler app called NexMusic and haven't had any issues with that light little player ever since. Also please make sure to use an AOSP ROM for testing sound. Some hybrids or Caf-based with CAF-AOSP patches may not have sound working properly due to the mixing and matching of CAF and AOSP.
Audio Patches? I currently have all Franco and all Faux audio patches. I watch them like a hawk so rest assured audio issues are not because I'm behind or out-of-date. It may be the drivers can't handle bleeding edge toolchains or optimizations....

Random Reboots
Overview: The only RRs I have ever experienced on this kernel were from changes I made to the test branch which I revert shortly thereafter. I realize it is unfair to many of you because I tailor everything to my phone specifically and not all devices are created equal. I noticed lately Neobuddy bumped up voltages to prevent some RRs for some users. I've also considering doing something like that but I feel like that would be catering to the minority when the majority have no issues with the current UBER setup.
Undervolting: This does have a good purpose which is to reduce heat during video gaming. As far as battery life goes let's just say I never undervolt and can still get PLENTY of hours out of my phone between charges. My test show results are negligeable. If you have RRs please consider bumping up your voltages to +25 like Neobuddy did for his users.
Reboot before unlock: I've never experienced this although it's been reported but never with a log for months now. Please follow Post #2 and go through that to learn how to generate a logcat.
Misc
Older UBER Kernel versions are Better? There have been rumors circulating that older UBER versions (ie 7/29) are better for both battery and performance. I disagree with this idea here is why: 1) With the latest I2C drop my battery increased by 33% and I've never been happier 2) With the introduction of uber.cfg (Anything after 7/29) I decided to make use of some values never used before in the kernel. One of which was the drop cache values. Previously we used the default stock values of 0. Now we have moved to the value of 3. If you read that section in the UBER config you'll start to understand that performance is better with the value of 3 because it causes apps to rebuild their caches based on usage. Unfortunately, this initially drains battery and slows down your phone as you open your apps and it rebuilds everything. This gives some of you the illusion that the new update is "not as fast" or "not as battery friendly" as the previous version. You'll really see the lag on opening apps while using ART because ART takes longer to rebuild caches (as I'm sure you are aware) This is why Facebook takes a minute or so to open after clean install on latest UBER but later opens faster than it ever has before. I'm giving you this explanation so you won't panic when things are slow right after upgrade. After every kernel update I go through and open my apps one by one to ensure that when I need them the caches are rebuilt and ready to go. If you disagree with what I've done here you can change the value from 3 back 0 and it won't hurt my feelings. I'm only trying to help you have your apps open faster and run better.
Color profiles? The idea with using your own color profile in the kernel was so you could use display control by Franco or whatever other display color control app to generate a config like so:
0 12 19 30 39 48 56 72 83 105 118 126 121 114 109 102 93 73 63 54 40 33 20
0 12 19 30 39 48 56 72 83 105 118 130 121 115 114 102 82 76 63 54 40 33 20
0 12 19 30 39 54 61 78 84 107 123 132 117 110 108 97 90 71 61 52 45 37 21
0 12 19 30 39 54 61 78 84 107 121 130 119 112 107 99 76 72 57 50 45 37 21
0 12 19 30 39 48 56 72 83 105 121 126 119 112 107 99 91 71 63 53 44 36 22
0 12 19 30 39 48 56 72 83 105 121 134 119 113 110 98 78 72 59 50 44 36 22
32

The lines repesent colors as follows: (Notice 7 lines above and 7 below in the same order)
Blue Negative
Blue Positive
Green Negative
Green Positive
Red Negative
Red Positive
White Point

After you generate your preferred color profile using an app (because it is easier) you could copy it and rename it to uber_color_v1.3.cfg place it on your sdcard and change the color profile number to 2 for custom color profile.

Logcats: Bottom line is if I don't get a logcat it never happened (like the tree in the forest analogy) Please read Post #2 completely so you are aware of everything you need to know.

Dev Host wouldn't let me upload unfortunately so there is no mirror today. I'll try to rebuild again this weekend so there were be a mirror again.

Anyhow,
Enjoy the updates!

Link
ONLY HERE TODAY: http://uber.oceighty.co/HAMMERHEAD/UBER/

Cheers!
Cl3Kener

Kernel Changelog

Alles anzeigen

Zitat

f56c6e6 mm: do not sleep in balance_pgdat if there's no i/o congestion
a475053 mm: vmscan: If kswapd has been running too long, allow it to sleep
4a1c0fb qseecom: change the number of bytes to compare app names
4d96bdf Revert "Revert "qseecom: Add support to handle multiple app load & unload""
ad10c70 ARM: fix user unalignment prefetch abort
2a123bf ARM: gic: Bring inline with upstream
8f79e89 cpuidle: check for need_resched() after the idle entry trace event
47562b3 ARM: bitops: prefetch the destination word for write prior to strex
032c069 arm: hammerhead: gpiomux: Pull down the UART suspend gpio
3bff5ea mmc: core: Avoid infinite retries of failed mmc command
4a5eb98 net: wireless: bcmdhd: Disable OOB interrupt when WLAN is off
4451fa4 F2FS: Update Inode Compatibility
27fe796 Fixup Exfat
22a7e38 vfs: Rename end_writeback() to clear_inode()
e257d3e cgroup: fix exit() vs rmdir() race
5825fed msm: Create shadow files under /sys/kernel/ipv4
e1a95bb WLAN subsystem: Sysctl support for key TCP/IP parameters
608f40c SELinux: possible NULL deref in context_struct_to_string
117968d SELinux: use ARRAY_SIZE
14dea71 SELinux: Reduce overhead of mls_level_isvalid() function call
b13e35c i2c-qup: Verify that mem resource exists prior to redirection
a91fe78 Sound Control: add line gain to cached and register lock

Alles anzeigen

TWRP

MuggelPu

Zitat

September 28, 2014
Well 3.4.104 finally came out after 6-7 weeks of waiting but I couldn't upload immediately because Dev Host to come back online. Looks like it is mostly back online but I think some of the uploads failed on for whatever reason so I may be missing a few links there. Oh well you all survived last time right?
So the lockscreen reboot bug was solved thanks to Patrick Perina who determined that changing grace timer in uber.cfg to 0 fixed it for him. We have that feature so that you can control when the config will set after boot. Initially, we thought 10 seconds was good idea but it doesn't hurt you to set it to 0 either. Try that and see if it solves that problem for you. Some of you may not experience this issue because you use something other than slide to unlock.
Check out the changelog for all of the great stuff that was added by kernel.org with this new merge below!

Anyhow,
Enjoy the updates!

Links
Main: http://uber.oceighty.co/HAMMERHEAD/UBER/
Mirror: http://d-h.st/users/cl3kener/HAMMERHEAD/KERNELS/UBER
Cheers!
Cl3Kener

Alles anzeigen

Zitat

Kernel Changelog
Code:
0591827 hid: Insert Missing Bool
86f56d5 cpuidle: Fix finding state with min power_usage
053ee85 cpuidle: Get typical recent sleep interval
884e94b cpuidle: Set residency to 0 if target Cstate not enter
edb9ad8 Defconfig: 3.4.104
bb4a5cf Merge tag 'v3.4.104' into stock
bb4a05a Linux 3.4.104
618dea4 alpha: add io{read,write}{16,32}be functions
3a8f613 ext2: Fix fs corruption in ext2_get_xip_mem()
f79bb94 alpha: Fix fall-out from disintegrating asm/system.h
92a6e26 UniCore32-bugfix: fix mismatch return value of __xchg_bad_pointer
4e36063 UniCore32-bugfix: Remove definitions in asm/bug.h to solve difference between native and cross compiler
62cdcce unicore32: select generic atomic64_t support
82a938ab 8250_pci: fix warnings in backport of Broadcom TruManage support
69db2d4 slab/mempolicy: always use local policy from interrupt context
ab22539 openrisc: add missing header inclusion
c1bc007 MIPS: Fix accessing to per-cpu data when flushing the cache
9a8fa93 MIPS: perf: Fix build error caused by unused counters_per_cpu_to_total()
c804743 USB: serial: fix potential heap buffer overflow
62148f7 USB: serial: fix potential stack buffer overflow
2f3e285 ARM: 8129/1: errata: work around Cortex-A15 erratum 830321 using dummy strex
5624bb3 ARM: 8128/1: abort: don't clear the exclusive monitors
e78c127 HID: picolcd: sanity check report size in raw_event() callback
e115f02 HID: magicmouse: sanity check report size in raw_event() callback
0bfb000 NFSv4: Fix problems with close in the presence of a delegation
818ee41 USB: sisusb: add device id for Magic Control USB video
86d165e HID: logitech-dj: prevent false errors to be shown
2f2c704 USB: whiteheat: Added bounds checking for bulk command response
1d37c3e usb: xhci: amd chipset also needs short TX quirk
42494f0 xhci: Treat not finding the event_seg on COMP_STOP the same as COMP_STOP_INVAL
d232e2c USB: ftdi_sio: Added PID for new ekey device
68e9929 USB: serial: pl2303: add device id for ztek device
262548e USB: ftdi_sio: add Basic Micro ATOM Nano USB2Serial PID
78aea3a USB: option: add VIA Telecom CDS7 chipset device id
c2aa9b7 staging: et131x: Fix errors caused by phydev->addr accesses before initialisation
f40751b xtensa: fix a6 and a7 handling in fast_syscall_xtensa
bcf20fd xtensa: fix TLBTEMP_BASE_2 region handling in fast_second_level_miss
fb2ae73 xtensa: fix address checks in dma_{alloc,free}_coherent
2297927 xtensa: replace IOCTL code definitions with constants
483320c ALSA: hda/realtek - Avoid setting wrong COEF on ALC269 & co
d06e4b0 CIFS: Fix wrong directory attributes after rename
a0e5b9d md/raid6: avoid data corruption during recovery of double-degraded RAID6
77e5657 iommu/amd: Fix cleanup_domain for mass device removal
db83744 pata_scc: propagate return value of scc_wait_after_reset
f0634a3 kvm: iommu: fix the third parameter of kvm_iommu_put_pages (CVE-2014-3601)
33df36f MIPS: OCTEON: make get_system_type() thread-safe
5ccd3e2 isofs: Fix unbounded recursion when processing relocated directories
c945ed6 HID: logitech: perform bounds checking on device_id early enough
dab2f9b HID: fix a couple of off-by-ones
9220628 ibmveth: Fix endian issues with rx_no_buffer statistic
c27d3b5 ASoC: pxa-ssp: drop SNDRV_PCM_FMTBIT_S24_LE
b402b72 stable_kernel_rules: Add pointer to netdev-FAQ for network patches
f56c6e6 mm: do not sleep in balance_pgdat if there's no i/o congestion
a475053 mm: vmscan: If kswapd has been running too long, allow it to sleep
4a1c0fb qseecom: change the number of bytes to compare app names
4d96bdf Revert "Revert "qseecom: Add support to handle multiple app load & unload""
ad10c70 ARM: fix user unalignment prefetch abort
2a123bf ARM: gic: Bring inline with upstream
8f79e89 cpuidle: check for need_resched() after the idle entry trace event
47562b3 ARM: bitops: prefetch the destination word for write prior to strex
032c069 arm: hammerhead: gpiomux: Pull down the UART suspend gpio
3bff5ea mmc: core: Avoid infinite retries of failed mmc command

Alles anzeigen

Gizmodo

EdgA

=Cl3Kener;55874692][size=12][size=14]October 7th, 2014[/SIZE]

Well things are really kicking here at UBER! Let me tell you about some of the new things that are happening!

Basketbuild - I signed up for Basketbuild and I am VERY impressed with them! It has great download speeds and a clean and neat appearance. I believe I'll be dropping Dev Host and going with them as the UBER kernel mirror. Check it out my latest N5 files here: OPPO Forum

Kernel Compression Change - I've decided to shrink the kernel by XZ so that it is now only 2/3 the size which will allow for faster uploads so I don't sit around typing really long messages for you to read. (I'm sure you'll all be grateful.) Although LZ4 is a faster compression than XZ when it came to booting with uber.cfg and everything else differences in boot time was negligeable (like 1 maybe 2 seconds tops on a 50-60 second boot) LZ4 make a noticeable difference on legacy devices but the raw power of the N5 can uncompress boot.img fast regardless of compression. BTW, kernel compression does in no way influence performance it only influences boot time because it has to be uncompressed and loaded to ram.

Nvidia Power savings - I grabbed a series of patches from NVIDIA to help improve battery. They made changes to some critical blocking code paths which will allow the CPU to relax instead of being stuck spinning. This should lower the load average and power consumption and increase you battery a little bit more. I haven't noticed any change in performance so I'd say these patches are awesome!

Better cpuidle - Besides the Nvidia stuff I also grabbed some upstream changes dealing with cpu idle in hopes to improve battery life.

Optimized Memcpy - I grabbed some patches meant to improve memcpy or copy of information into memory on Krait-based processor.

EXT4 Speedup - You can see I grabbed a commit to help speed up some EXT4 functions.

Fauxsound config explanations - update faux sound explanations so they make more sense. Hopefully that will help some of you.

Upstream fix for longtime Kernel regression - This commit is cool because I saw it upstream before any other N5 developer. It appears vfs was getting slower since 3.2.y and this fixes that and was made by Linux Torvalds himself. https://github.com/Cl3Kener/UBER-…8b22ba83361691b

Okay well I could keep going since there were some pretty cool little things here and there. You'll see this is quite a long changelog but my time is almost up. Kernels are almost uploaded.

Anyhow,

Enjoy the updates!

[size=14]
Links [/SIZE]

Main: http://uber.oceighty.co/HAMMERHEAD/UBER/

Mirror: OPPO Forum

Cheers!

Cl3Kener

[size=14]Kernel Changelog[/SIZE]

Code

1018cad Makefile: Use Pipe Option
b511363 msm: tspp: use an actual device structure for DMA allocation
5248947 tspp: fix bam irq request preventing sleep
f72d076 msm: tspp: Request IRQs on first use
19df51f msm: tspp: Fix memory leak
16b2831 tspp: Protect against buffer overflow
dc6b128 msm: tspp: Fix usage of physical addresses
48f10cf arch:arm: Use generic rwsem
5544299 ARM: 7494/1: use generic termios.h
c52da3c HID: input: generic hidinput_input_event handler
cdfa739 arm:lib: Correct Neon Flags
4ac2023 USB: storage: ene_ub6250: fix warning for the frame size
58f0495 ARM: 7493/1: use generic unaligned.h
2ed21ef Disable -fconserve-stack on ARM v2
818027c mm: Add NULL check before de-referencing vma
18197cb mm: failslab: Add support to force slab alloc failures based on size.
3b5ff1f kernel: power: Add PM_USERSPACE_FROZEN workqueue
292cf91 freezer: set PF_SUSPEND_TASK flag on tasks that call freeze_processes
cd59062 alarmtimer: Do not check for wake capability in registration
f801c43 ARM: Honor udelay convention for timer-based delay
3c8973f ARM: kgdb: ignore breakpoint instructions from user mode
5d7db2c ARM: Export LoUIS flush
b0a20b3 ARM: mm: adding a check for slab objects in flush_dcache_page
1d895fb ARM: mm: Enable NCSE feature for A15 only
3e9cb13 ARM: mm: enable non-cacheable streaming enhancement
813253e ARM: 7650/1: mm: replace direct access to mm->context.id with new macro
c5885ef ARM: cache-l2x0: get size of outer cache
e61f5f4 ARM: dma: Drop GFP_COMP for DMA-IOMMU memory allocations
ba12177 ARM: mm: cache-l2x0: Implement outer_clean_all()
1e8e9eb ARM: LPAE: Invalidate the TLB for module addresses during translation fault
cbad24af ARM: remove unnecessary dcache_clean_area
95f958f ARM: mm: Skip I-cache invalidate for Cortex-A15 boot
f92ae37 ARM: mm: Remove unnecessary CMO in Cortex A15 startup
0d21e70 ARM: cleanup undefined instruction entry code
978fce4 arch: arm: lib: optimize memcpy for cortex-A15
9680309 kernel/smp.c: use '|=' for csd_lock
f364ad8 smp: enhance power efficiency
a9b1956 hrtimer: enhance power efficiency
cf7063c sched: Trying to reduce power usage
c7c5e3d seqlock: enhance power efficiency
480fe53 cpu: enhance power efficiency
9c26fc5 ARM: including asm-generic/processor.h
075c66f asm-generic: processor.h: adding new header file
cbc6885 llist: defining relaxed version of llist_empty
de6868c asm-generic: relaxed.h: defined relaxed.h
2cf971c sched: defining relaxed version of idle_cpu
dc1dbb3 arm: asm: relaxed.h: defined relaxed.h
bca5d32 arm: atomic.h: defined relaxed atomic_read
a0cc10b lib/lzo: huge LZO decompression speedup on ARM by using unaligned access
bd3083d ARM: 7483/1: vfp: only advertise VFPv4 in hwcaps if CONFIG_VFPv3 is enabled
83f88d7 msm: msm_bus: Suspected fix for deadlock
b82df88 kthread_work: wake up worker only when the worker is idle
535713f ARM: make kernel oops easier to read
2a9876b Defconfig: Set XZ as Kernel Compression
0dca45f vfs: fix bad hashing of dentries
812e1fb persistent_ram: check PERSISTENT_RAM_SIG before writing
7a7ed1b staging: android: persistent_ram: Introduce  persistent_ram_post_init()
76693cc persistent_ram: Remove prz->node
424bf61 msm: peripheral_loader: Use the new request_firmware_direct API
b7eb839 firmware_class: Introduce the request_firmware_direct API
f59e78b input: SynaImage_ds5: Firmware E079 (ORISE)
d42ccd7 kernel/smp.c: quit unconditionally enabling irqs in on_each_cpu_mask().
ae14e08 genirq: Introduce irq_do_set_affinity() to reduce duplicated code
4e49265 arch: arm: TLB invalidate if power collapse gets aborted
0e42e77 Defconfig: Add Memcpy Optimizations
410ea15 arch: arm: lib: optimize memcpy for cortex-A15
fcaa5fe mm: vmscan: when kswapd is awoken due to reclaim by a running task
370f6c8 ext4: Speedup WB_SYNC_ALL pass called from sync(2)
03770e2 ACPI idle, CPU hotplug: Fix NULL pointer dereference during hotplug
b0906ba cpuidle: move field disable from per-driver to per-cpu
1f4225f cpuidle: add checks to avoid NULL pointer dereference
69b01d5 cpuidle: remove unused hrtimer_peek_ahead_timers() call
9092439 cpuidle: remove cross-cpu IPI by new latency request.

Alles anzeigen

[/SIZE]

MuggelPu

Zitat

October 20th, 2014
Well things are still happening here at UBER (even if nobody has posted on this thread in few days)! Let me tell you about some of the new things that are going on!
New Server on bbqdroid.org - Special thanks to @chadouming BBQDroid.org for hosting me on their server now! These kernel builds were so much easier to make this time around! If you haven't heard of the BBQ Team check out their apps here:https://play.google.com/store/apps/d...er?id=Team+BBQ or visit their website (bbqdroid.org).
New SaberMod Toolchains - @sparksco has been working tirelessly on bringing up SaberMod toolchains to the latest version of everything available! (For example binutils from 2.24 to 2.25, isl from 0.12.1 to 0.13.0, cloog from 18.1 to 18.2, gmp from 5.1.2 to 6.0.0 as well as upstream merges from gnu to gcc and gdb). I felt accomplished this time around because I was able to help fix a weird automake related errorhttps://github.com/SaberMod/cloog-cu...de9fb36956cc34 in the cloog repo.
Note: Linaro toolchain components I only use release versions so I'm still waiting on them to update again before rebuild. Nothing changed since the last versions I made for the last batch of UBER kernels.
Kernel Changes - As far as the kernel changes there wasn't anything amazing. Flar2 updated his wake timeout driver so I merged those changes. You'll also see several arm topology commits, some error message fixes/changes, and some upstream merges on lz4/lzo compressions.
I also decided to simplify ondemand governor by removing trivial features (like input boost because we already have this feature in hotplug for example) and adding down_differential_multicore from CM kernel. Hopefully this will yield better battery results for those of you staying with kernel defaults (although I'm still sticking with optimax)
Anyhow,
Enjoy the updates!

Links
Main:http://uber.oceighty.co/HAMMERHEAD/UBER/
Mirror:Chainfire

Cheers!
Cl3Kener

Alles anzeigen

Zitat

Kernel Changelog
Code:
a4a6618 DT2W: Add Missing Include
1917f74 s2w/dt2w: make sure irq_wake gets disabled if s2w/dt2w are disabled while screen is off
ab87082 wake_timeout: do not start wake timeout if s2w and dt2w are disabled
d697b77 wake_timeout: fully disable wake functions if s2w and dt2w are disabled while screen is off
452c909 wake_timeout: increase delay to ensure panel is shut down properly
72f049e power: Avoids bogus error messages for the suspend aborts.
44472d3 Simplify Ondemand
869e75c lzo: check for length overrun in variable length encoding.
724c600 Revert "lzo: properly check for overruns"
e5c68ee crypto: lz4,lz4hc - fix decompression
d334bc7 ARM: topology: Add efficiency coefficient for Krait
b6e5452 ARM: dt: Only print warning, not WARN() on bad cpu map in device tree
5deae3e ARM: 7763/1: kernel: fix __cpu_logical_map default initialization
21ead1e ARM: kernel: add logical mappings look-up
d5ab2f5 ARM: 7762/1: kernel: fix arm_dt_init_cpu_maps() to skip non-cpu nodes
ff31736 ARM: 7585/1: kernel: fix nr_cpu_ids check in DT logical map init
e969fb5 ARM: kernel: add device tree init map function
4548fdf of: Fix iteration bug over CPU reg properties
5e66eda of: Make cpu node handling more portable.
d38ec26 of: fix unnecessary warning on missing /cpus node
e897cd3 driver/core: cpu: initialize of_node in cpu's device struture
cf7c97c numa, cpu hotplug: change links of CPU and node when changing node number by onlining CPU
d81a818 sysfs: fix crash_notes_size build warning
651e70d sysfs: Add crash_notes_size to export percpu note size
b510d69 drivers/base/cpu.c: Fix typo in comment
d75796f cpupower: Remove mc and smt power aware scheduler info/settings
10042cc of: move of_get_cpu_node implementation to DT core library
6be1e88 ARM: 8008/1: topology: Coding style fixes
382163a ARM: 7920/1: topology: Staticise non-exported symbols
80041e4 ARM: topology: remove hwid/MPIDR dependency from cpu_capacity
0e0f949 arm: remove cast for kzalloc return value
385fef9 ARM: kernel: update topology to use new MPIDR macros
a9b4d6b ARM: kernel: enhance MPIDR macro definitions
1d3c7fb ARM: 7482/1: topology: fix section mismatch warning for init_cpu_topology
ae91a83 ARM: 7463/1: topology: Update cpu_power according to DT information
5636087 ARM: 7462/1: topology: factorize the update of sibling masks
9e2ff02 ARM: 7461/1: topology: Add arch_scale_freq_power function
fc1f264 ARM: 8168/1: extend __init_end to a page align address
e35e29c ARM: 8088/1: vmlinux.lds.S: drop redundant .comment
6a520cc ARM: 7605/1: vmlinux.lds: Move .notes section next to the rodata
8db58e6 ARM: 8165/1: alignment: don't break misaligned NEON load/store

Alles anzeigen

sparksco

MuggelPu

Zitat

[size=14]Chet Kener[/SIZE]
Experimental/Test Builds - 22.10.2014
[COLOR=#ff0000]New UBER Lollipop Preview Kernel - October 21, 2014[/COLOR]
Well I decided to finally build an Lollipop kernel since this preview seemed nice enough to me to be worth running.
The changelog is huge since I had to bring over 1111 commits!
It seems stable enough to me but I'll leave it to you to find any unforeseen bugs.
Go here to see more more details::
Amazon hid a full app store in its Android app for six weeks and no one noticed
theverge.com

Alles anzeigen

Zitat

[COLOR=#ff0000]Android Lollipop Preview Kernel[/COLOR]
October 21th, 2014
I've spent almost all day merging in everything from kitkat into the new Lollipop Preview (LPX13D). Here is what I managed to accomplish today:

Features Added:
Code:
* MultiROM Compatible (has Kexec patch)
* CPU Governors added: Intellidemand, Intelliactive, Adaptive, Lionheart, AbyssplugV2, Ondemandplus, Badass, Wheatley, Hyper, Lazy, PegasusQ, Nightmare, DanceDance, Darkness, Slim, Optimax. and Uberdemand
* GPU Governors: Performance, Ondemand, and Interactive (Franco)
* IO Schedulers: Noop, Deadline, VR, SIO, ROW, CFQ, BFQ, FIOPS, FIFO, ZEN and Tripndroid
* Snapdragon & CortexA15 optimizations
* Non-Stock Hotplug and Thermal regulator (from showp1984)
* Touch Boost - Showp1984 (comes with hotplug)
* GPU OverClock up to 650 MHz (27 MHz is still default idle with lowest frequency being underclocked to 100 MHz) (Cl3Kener)
* Allow CPU OverClock up to 3.0 Ghz (Cl3Kener)
* Allow CPU UnderClock to 96Mhz (showp1984 implementation)
* L2 Cache OverClock (flar2)
* Ramdisk Edits on the fly for greater compatibility. (Strongly based on Showp1984 implement)
* Undervolting Control (Franco)
* LZ4 support implemented kernel wide - fastest compression/decompression available (Cl3Kener)
* XZ Compression to latest 3.17.y (Cl3Kener)
* KSM (Kernel Samepage Merging)
* Faux Sound (Faux123)
* Sweep2Wake, Double Tap 2 Wake (Showp1984 with Faux123 verticle s2w support and flar2 dt2w enhancements)
* Powerkey suspend for S2W/DT2W (Flar2 version)
* Wake timeout for S2W/DT2W (Flar2)
* Dynamic and Asynchronous Fsync (Faux123)
* USB Fast Charge with many voltage options (Faux123)
* Gamma Control (Various Contributers)
* Color Control (savoca)
* Fstrim (reduces lag see more: http://man7.org/linux/man-pages/man8/fstrim.8.html)
* ExFat Support (you'll need to load it via this init.d https://www.dropbox.com/s/tfm8k9vmskuhs4v/loadexfat) Updated to latest version (Neobuddy)
* Low Memory Killer - Many patches from Neobuddy
* Zram support upstreamed to 3.17.y branch (Cl3Kener)
* Zram supports lz4 compression now (kernel.org)
* Zcache, and Cleancache support updated to 3.14.y (Cl3Kener)
* BLX - Battery Life Extender (See here Amazon hid a full app store in its Android app for six weeks and no one noticed
theverge.com) Ported by Ayysir to msm8974 devices
* F2FS file system support from Samsung
* Mount internal storage as CDROM (Frozen Cow)
* Advanced TCP Congestion Options (Veno as default, all 12 from Amazon hid a full app store in its Android app for six weeks and no one noticed
theverge.com)
* Stereo Call Recording Support
* Uber.cfg for tuning kernel thanks to @janres155 & @Williamsss
To Do:* Merge to latest 3.4.104 kernel version from Kernel.org
* Make Francosound and Release kernels for L (Only FauxSound so far)

Not Happening:* Frandom Support- It's broken in Android L - don't have time to worry about fixing it now.
* Backlight Dimmer (Flar2) - L has adaptive brightness already

I'm too tired to go and figure out all of the scripts to build this kernel with every different toolchain so I'm just uploading a single SM Faux 4.9.2 Kernel to the XDA OP under DevDB section. That should be easily accessible to all of you and it will keep it separate from the kitkat kernels until I figure out how I want to manage 2 different kernel projects.
I hope you enjoy the changelog today. Grand total of commits was 1111!!!
Enjoy the new UBER Lollipop Kernel!!!

[B]Link
Amazon hid a full app store in its Android app for six weeks and no one noticed
theverge.com

Installation
* Reboot to Recovery
* I recommend flashing a root compatible boot.img from here for example Amazon hid a full app store in its Android app for six weeks and no one noticed
theverge.com
* Then flash uber and reboot.
* You will have a working root this method.
Note: If you have flashed N5_speedpatch.zip or other mods that are causing your install to fail you will have to replace the build.prop with stock build.prop or you will have to remove these lines from the build.prop for it to flash.
dalvik.vm.dex2oat-flags "--compiler-filter=interpret-only"
dalvik.vm.image-dex2oat-flags ""
After kernel install you can reflash your MODs

Cheers!
Cl3Kener

Kernel Changelog
It was too long to post on XDA (limit of 30,000 characters) so you'll have to go here to see all of the changes:
http://hastebin.com/ugofezarak.vhdl
[/B]

Alles anzeigen

https://plus.google.com/10438486723571…sts/Syy1wzgUKBH

Amazon hid a full app store in its Android app for six weeks and no one noticed
theverge.com

MuggelPu

[COLOR=#ff0000]INFO

[/COLOR]
[COLOR=#000000]

Zitat

How to get uber.cfg working on Lollipop
Many are complaining about not having uber.cfg work for them. There is an easy fix for that. Go into SuperSu menu and enable su during boot by checking the box.
That fixed the issue for me!

[/COLOR]
[COLOR=#000000]

Lukas Becker[/COLOR]

MuggelPu

[COLOR=#ff0000]UBER-L Test builds (All Toolchains and Versions)[/COLOR]
[COLOR=#ff0000]
[/COLOR]
[COLOR=#000000]

Zitat

Well I've been dying for official L release and while I've been waiting I decided to make all version of UBER-L Preview and test the new scripts and locations. Seems everything was a success. If you are on L Preview or one of the AOSP L builds please give one (or more) of these a try and let me know how it goes!
Also to avoid confusion I've made new folders deemed UBER-L instead of just UBER. All of the ZIPs also say "L" in the name in case you forget that there is a new directory.
Downloads
ungültiger Link
http://uber.oceighty.co/HAMMERHEAD/UBER-L/
Changelog
* Fix Slim Governor so it now compiles and is present in the kernel
* Added back ksgl commits from previous UBER to hopefully help GPU.
* Enabled F2FS again in the kernel so it will be ready when F2FS ROMs become available.

Alles anzeigen

[/COLOR]
[COLOR=#000000]
https://plus.google.com/10438486723571…sts/ikpanzMyuMM[/COLOR]

MuggelPu

[COLOR=#80ff00][size=24]UBER-L Official Release Build - November 12, 2014[/SIZE][/COLOR]
[size=24]
[/SIZE]
[size=14]As we welcome android Lollipop I have decided to build all UBER-L kernels almost immediately. There weren't many changes in the official kernel source release and it merged right in without a single conflict so I'm feeling confident that it will be smooth sailing.

For those of you that are not aware, in order to have working ROOT in L you need a ramdisk with a modified selinux policy (see herehttp://www.androidpolice.com/2014/10/20/cha…-a-requirement/)

What that means is that you'll need to flash a stock kernel with ROOT selinux policy fixes in order to have working root with UBER because UBER use the ramdisk that comes with whatever ROM you flash. This allows for greater ROM compatibility but now with more security in L it adds one more step for those of you wanting to run rooted stock. Once all of the custom ROMs start putting up builds you'll find uber won't break any of the different rooting methods thanks to the current ramdisk snatching install.

I don't currently have a ROOTED stock kernel but I'm sure one will appear here rather quicklyAFH Mirror #2. I guess Chainfire release cf-autoroot so you can probably use that too. Whatever the method you just need to make sure you have root obtain in the stock kernel you flash before flashing UBER. (There are some links to things in comments below)

Also, if you want to have uber.cfg work on L you need to install busybox after rooting. Search play store for busybox and use an app to install it. That should allow cfg to work.

Anyhow
Enjoy!

Downloads
http://uber.oceighty.co/HAMMERHEAD/UBER-L/
or
Basketbuild Mirror #1

Changelog
https://github.com/Cl3Kener/UBER-L/commits/master[/SIZE]

https://plus.google.com/10438486723571…sts/355Sd8tKU3F

MuggelPu

[COLOR=#ff0000][size=18]Test UBER-L Kernel - November 16, 2014[/SIZE][/COLOR]

Okay so I did a complete kernel rebase last night and this morning. I updated the kernel to 3.4.104. There seems to be some sort of negative stigma regarding 3.4.y kernel updates. I've noticed kernel developers and user alike spread false information about kernels that use kernel.org updates as "not as stable" or "bleeding edge." Both of these notions are FALSE. Every release from kernel.org is considered stable. Here is an example of a post from Greg KH on the subject of 3.4.100 saying this is a "stable review cycle"kernel.org. Allkernel.org updates undergo a review process and are safe to merge and use. If you see anyone posting anything about these updates being "not stable" please kindly correct them and help combat the false information circulating forums and G+.

As part of the test kernel, I've also removed a few things that I think may be causing Screen Freezes. Hopefully that will fix problems for the people that are having them.

I've updated uber.cfg files to be a bit more stable (although you'll notice a slight reduction in performance due to the nature of these changes you'll need to reenable some of the memory optimizations yourself under section 9 of the cfg)

Let me know if you still are having screen freezes!

Download Link

[DLMURL]http://uber.oceighty.co/HAMMERHEAD/UBER-L/TEST/UBER-L-N5-SM-4.9.3-11-16-2014-Test.zip[/DLMURL]

https://plus.google.com/10438486723571…sts/VWd9Mayb7Dn

MuggelPu

Zitat

November 17th, 2014
Well I've finally figured out a fix for those of you experiencing screen freezes. Kernels from today should work for you.
I've also spent several hours merging 5000+ commits from kernel.org bringing this kernel back to version 3.4.104. I also did a full kernel rebase on the 3.4.104 base. You can find my stock fused with kernel.org branch here:https://github.com/Cl3Kener/UBER-L/commits/merged. I went commit by commit to ensure proper merging of all updates. There were many fixes/improvements relevant to this device (although there were also many for other devices that were not relevant to us at all). Kernel.org really is a mixed bag but it's worth ensuring we have the latest improvements, security patches, etc.
Speaking of kernel.org: There seems to be some sort of negative stigma regarding 3.4.y kernel updates. I've noticed kernel developers and users alike spread false information about kernels that use kernel.org updates as "not as stable" or "bleeding edge - beware!." Both of these notions are FALSE. Every release from kernel.org is considered stable. Here is an example of a post from Greg KH on the subject of 3.4.100 saying this is a "stable review cycle" supported builds. All kernel.org updates undergo a review process and are safe to merge and use. If you see anyone posting anything about these updates being "not stable" please kindly correct them and help combat the false information circulating forums.
In this update I've also updated uber.cfg files to be a bit more stable (although you'll notice a slight reduction in performance due to the nature of these changes you'll need to reenable some of the memory optimizations yourself under section 9 of the cfg) I've removed functions no longer supported and added a couple more. Some of the most notable changes were 1) I removed frandom support (doesn't work for Lollipop), I removed backlight dimmer (use adaptive brightness instead), I added arch power toggle, I added a gentle fair sleepers toggle, I changed Drop_Caches values back to 0 (since the value of 3 agressively dropped all caches which take longer to rebuild in ART so it causes lag initially when you open apps although performance is great after you get past the lag) I changed FS_DELAY value to 0 just to ensure there would be no data loss caused by this feature while we are still in the testing phase of Lollipop (I may revert this change later).

Just so we are on the same page these are all of the features that made the 3.4.104 rebase:

Zitat

Features In Release Today:
* 3.4.104 from Kernel.org
* MultiROM Compatible (has Kexec patch)
* CPU Governors added: Intellidemand, Intelliactive, Adaptive, Lionheart, AbyssplugV2, Ondemandplus, Badass, Wheatley, Hyper, Lazy, PegasusQ, Nightmare, DanceDance, Darkness, Slim, Optimax. and Uberdemand
* GPU Governors: Performance, Ondemand, and Interactive (Franco)
* IO Schedulers: Noop, Deadline, VR, SIO, ROW, CFQ, BFQ, FIOPS, FIFO, ZEN and Tripndroid
* Snapdragon & CortexA15 optimizations
* Non-Stock Hotplug and Thermal regulator (from showp1984)
* Touch Boost - Showp1984 (comes with hotplug)
* GPU OverClock up to 650 MHz (27 MHz is still default idle with lowest frequency being underclocked to 100 MHz) (Cl3Kener)
* Allow CPU OverClock up to 3.0 Ghz (Cl3Kener)
* Allow CPU UnderClock to 96Mhz (showp1984 implementation)
* L2 Cache OverClock (flar2)
* Ramdisk Edits on the fly for greater compatibility. (Strongly based on Showp1984 implement)
* Undervolting Control (Franco)
* LZ4 support implemented kernel wide - fastest compression/decompression available (Cl3Kener)
* XZ Compression to latest 3.17.y (Cl3Kener)
* KSM (Kernel Samepage Merging)
* Faux Sound (Faux123)
* Sweep2Wake, Double Tap 2 Wake (Showp1984 with Faux123 verticle s2w support and flar2 dt2w enhancements)
* Powerkey suspend for S2W/DT2W (Flar2 version)
* Wake timeout for S2W/DT2W (Flar2)
* Dynamic and Asynchronous Fsync (Faux123)
* USB Fast Charge with many voltage options (Faux123)
* Gamma Control (Various Contributers)
* Color Control (savoca)
* Fstrim (reduces lag see more: http://man7.org/linux/man-pages/man8/fstrim.8.html)
* ExFat Support (you'll need to load it via this init.d https://www.dropbox.com/s/tfm8k9vmskuhs4v/loadexfat) Updated to latest version (Neobuddy)
* Low Memory Killer - Many patches from Neobuddy
* Zram support upstreamed to 3.17.y branch (Cl3Kener)
* Zram supports lz4 compression now (kernel.org)
* Zcache, and Cleancache support updated to 3.14.y (Cl3Kener)
* BLX - Battery Life Extender (See here supported builds) Ported by Ayysir to msm8974 devices
* F2FS file system support from Samsung
* Mount internal storage as CDROM (Frozen Cow)
* Advanced TCP Congestion Options (Veno as default, all 12 from supported builds)
* Stereo Call Recording Support
* Uber.cfg for tuning kernel thanks to @janres155 & @Williamsss
I hope you enjoy the changelog today. Grand total of commits handpicked by me in the last 24 hours was 5442!!!
Enjoy the new UBER Lollipop Kernel Rebase!!!
Cheers!
Cl3Kener

Alles anzeigen

Links

Main: http://uber.oceighty.co/HAMMERHEAD/UBER-L/

Mirror: #HT

Kernel Changelog

http://hastebin.com/uwiqogasis.dos

(Much too long for XDA character limits)

MuggelPu

Zitat

November 27th, 2014 - UBER Kernel Updates
Before anyone asks me if I'm going to add faux123's latest intellimm governor to UBER I thought I'd rebuild with it.
I've also include a few reverts to solve some lag for a few certain people (those heavy RAM users). Apparently some modifications to the out of memory killer were causing it to not work properly causing lag in some cases.
This Update is also for Kitkat as well. You too can run intellimm

Enjoy!
Chet

Lollipop Links
Main: http://uber.oceighty.co/HAMMERHEAD/UBER-L/
Mirror:@Google+

Kitkat Links
Main: http://uber.oceighty.co/HAMMERHEAD/UBER/
Mirror:@Google+

Cheers!
And Happy Thanksgiving!
Cl3Kener

Kernel Changelog
- Intellimm
- Fixed oom killer

Note: I forgot to add intellimm as an option to uber.cfg. I've been with family today and have been trying to be discrete with my developing which is why I forgot to do it. Don't worry though, if you put "intellimm" in for governor it will work fine. No need to panic when you don't see it in the list in the cfg it was just a small oversight on my part.

Alles anzeigen

https://plus.google.com/10438486723571…sts/jTuGWwVC9HD

MuggelPu

[size=14]Update auf 3.4.105 & mehr ...[/SIZE][size=24] [/SIZE][size=8](Ist der neu? Nein mit Perwoll gewaschen!) :moustach:[/SIZE]

Zitat

December 8th, 2014
Well after lots of testing kernels/toolchains it seems 3.4.105 is stable and I've updated both Lollipop and Kitkat Kernels to this new version. This was no small update. See the changelog below.
Also I apologize for the delay. I didn't feel any rush to push out 3.4.105 since I bet it will be another month or two until we see 3.4.106. I also wanted to get the latest Linaro toolchain components into the toolchain builds before the update.
For people wondering about F2FS support, MultiROM and all of that jazz check out this post here:https://plus.google.com/104384867235...ts/jgGGnRsuQiy just in case you missed it.
Enjoy the Updates!
ChetInstallation for Lollipop
1. Flash UBER after clean install or
2. Wipe system, dirty flash ROM, gapps, F2FS/EXT4 Kernel of you choice (link above) UBER then SuperSU. (Note failure to flash SuperSU last will result in loss of root)

Cheers!
Cl3Kener

Alles anzeigen

Lollipop Links

Main: http://uber.oceighty.co/HAMMERHEAD/UBER-L/

Mirror: Mirror #1

Kitkat Links

Main: http://uber.oceighty.co/HAMMERHEAD/UBER/

Mirror: Mirror #1

Zitat

Kernel Changelog

888ded4 Defconfig: 3.4.105
64b4302 Linux 3.4.105
a61fc2d l2tp: fix race while getting PMTU on PPP pseudo-wire
351629d ixgbevf: Prevent RX/TX statistics getting reset to zero
c151025 net: Do not enable tx-nocache-copy by default
8bdda44 ipv6: reuse ip6_frag_id from ip6_ufo_append_data
6dd11b5 rtl8192ce: Fix null dereference in watchdog
f71f9e4 ipv4: disable bh while doing route gc
0f310f8 ipv4: avoid parallel route cache gc executions
cba1056 ipv4: move route garbage collector to work queue
2fd7dda Fix spurious request sense in error handling
12e77f3 dm crypt: fix access beyond the end of allocated space
c9d49b3 mnt: Only change user settable mount flags in remount
ac7c0d4 perf: Handle compat ioctl
15261a9 NFS: fix stable regression
6f0805a ext4: avoid trying to kfree an ERR_PTR pointer
9856cc6 ext4: propagate errors up to ext4_find_entry()'s callers
af62f7d nl80211: clear skb cb before passing to netlink
4b99e3f genhd: fix leftover might_sleep() in blk_free_devt()
aa18b20 init/Kconfig: Hide printk log config if CONFIG_PRINTK=n
cf9076d perf: fix perf bug in fork()
26dbe48 mm: migrate: Close race between migration completion and mprotect
6918a56 shmem: fix nlink for rename overwrite directory
a4e2250 ocfs2/dlm: do not get resource spinlock if lockres is new
6ead374 nilfs2: fix data loss with mmap()
de66acd MIPS: mcount: Adjust stack pointer for static trace in MIPS32
6e08dce cpuset: PF_SPREAD_PAGE and PF_SPREAD_SLAB should be atomic flags
eb64d72 sched: add macros to define bitops for task atomic flags
90aaff9 sched: Fix unreleased llc_shared_mask bit during CPU hotplug
7313bfe parisc: Only use -mfast-indirect-calls option for 32-bit kernel builds
731a965 Fix nasty 32-bit overflow bug in buffer i/o code.
7ec50d1 ALSA: pcm: fix fifo_size frame calculation
e9a4eb4 can: at91_can: add missing prepare and unprepare of the clock
ea09f11 can: flexcan: put TX mailbox into TX_INACTIVE mode after tx-complete
77de70d can: flexcan: implement workaround for errata ERR005829
ffd705b can: flexcan: correctly initialize mailboxes
ac355c8 can: flexcan: mark TX mailbox as TX_INACTIVE
e877038 USB: storage: Add quirks for Entrega/Xircom USB to SCSI converters
fb8d324 USB: storage: Add quirk for Ariston Technologies iConnect USB to SCSI adapter
f0e0823 USB: storage: Add quirk for Adaptec USBConnect 2000 USB-to-SCSI Adapter
9523cbb libiscsi: fix potential buffer overrun in __iscsi_conn_send_pdu
af3004e NFSv4: Fix another bug in the close/open_downgrade code
4f2bf5ec iscsi-target: avoid NULL pointer in iscsi_copy_param_list failure
7bdb166 iscsi-target: Fix memory corruption in iscsit_logout_post_handler_diffcid
d4964b2 Input: i8042 - add nomux quirk for Avatar AVIU-145A6
d3bb852 Input: i8042 - add Fujitsu U574 to no_timeout dmi table
f2ae0d2 ASoC: samsung-i2s: Check secondary DAI exists before referencing
3ca58fd perf: Fix a race condition in perf_remove_from_context()
9feee3f MIPS: ZBOOT: add missing <linux/string.h> include
8116207 jiffies: Fix timeval conversion to jiffies
dc2f6c8 don't bugger nd->seq on set_root_rcu() from follow_dotdot_rcu()
a15244e alarmtimer: Lock k_itimer during timer callback
64e1991 alarmtimer: Do not signal SIGEV_NONE timers
2b70c7e alarmtimer: Return relative times in timer_gettime
00c896c futex: Unlock hb->lock in futex_wait_requeue_pi() error path
24e0338 xhci: Fix null pointer dereference if xhci initialization fails
fcb8eac storage: Add single-LUN quirk for Jaz USB Adapter
a6049f1 usb: hub: take hub->hdev reference when processing from eventlist
edf1ccf Input: serport - add compat handling for SPIOCSTYPE ioctl
2198d6f libceph: do not hard code max auth ticket len
4a74c78 libceph: add process_one_ticket() helper
4f9c65a libceph: gracefully handle large reply messages from the mon
6d500b0 Input: synaptics - add support for ForcePads
60849e7 uwb: init beacon cache entry before registering uwb device
b560b4c USB: ftdi_sio: Add support for GE Healthcare Nemo Tracker device
11d4532f Input: elantech - fix detection of touchpad on ASUS s301l
0cba95e usb: host: xhci: fix compliance mode workaround
de6677e drm/radeon: add connector quirk for fujitsu board
1ad43cd ahci: add pcid for Marvel 0x9182 controller
a6cd173 block: Fix dev_t minor allocation lifetime
bde8959 xen/manage: Always freeze/thaw processes when suspend/resuming
6363b1f USB: sierra: add 1199:68AA device ID
69e7ca4 USB: sierra: avoid CDC class functions on "68A3" devices
453a915 USB: ftdi_sio: add support for NOVITUS Bono E thermal printer
8d62523 ata_piix: Add Device IDs for Intel 9 Series PCH
3658160 ahci: Add Device IDs for Intel 9 Series PCH
d27cdc3 rtlwifi: rtl8192cu: Add new ID
fa44e3b cgroup: reject cgroup names with ' '
c5b8f75 percpu: free percpu allocation info for uniprocessor system
2e63c15 percpu: perform tlb flush after pcpu_map_pages() failure
5ceda6a percpu: fix pcpu_alloc_pages() failure path
34e59d0 regulatory: add NUL to alpha2
992bb58 ACPI / cpuidle: fix deadlock between cpuidle_lock and cpu_hotplug.lock
f1f8f0e ALSA: hda - Fix COEF setups for ALC1150 codec
5877a41 drm/vmwgfx: Fix a potential infinite spin waiting for fifo idle
b4dd701 get rid of propagate_umount() mistakenly treating slaves as busy.
aa94baa drm/i915: Remove bogus __init annotation from DMI callbacks
1f38a75 regmap: Fix handling of volatile registers for format_write() chips
d35c6be KVM: s390: Fix user triggerable bug in dead code

Alles anzeigen

Happy Flashing und so

MuggelPu

INFO zu Bye Bye GCC 4.10

"UBER Kernel Update - GCC 4.10 Kernels to be removed"

Zitat

[COLOR=#ff0000]UBER Kernel Update - GCC 4.10 Kernels to be removed[/COLOR]
Let me start with a little history lesson before you get upset for my decision to remove GCC 4.10. I promise there are a number of very good reasons.
GCC 4.10 was actually originally a mistake by gnu. Based on how they name things it should have just been GCC 5.0 but it wasn't which is why it came and went in something like 6 short months. A few weeks before it died it was already giving me grief and stopped booting. GCC 5.0 has now been out for a while and even though the kernel builds with GCC 5.0 it doesn't boot either which leads me to believe 5.0 is never going to boot for us on N5 unfortunately.

So why have I decided to remove GCC 4.10?
This is not a spur of the moment sort of thing. I've contemplated this for a long time and these are my reasons for removing GCC 4.10:
1) GCC 4.10 is no longer compatible with the latest Toolchain components As I mentioned it was a mistake from the beginning and has already died off months ago. With recent major version updates to toolchain components GCC 4.10 no longer builds which means I actually haven't been able to build it and you're getting the same old toolchain while the other toolchains are constantly getting fixed/improved through the gnu review process.
2) GCC 4.10 drains your battery more rapidly than 4.9.3 and I really can't tell much of a difference in performance, UI fluidity or anything really.
3) GCC 4.10 has been known to cause audio glitchesThis point is pretty self-explanatory. Every time I receive a message with regards to audio glitches it is on the 4.10 kernel. Is it really worth trading a very slight extra bit of performance for audio and battery?
4) Benchmark Gap between 4.9.3 and 4.10 has closed significantly Originally I kept trying to milk 4.10 for a long as I could because of benchmarks were higher in 4.10 but in light of recent toolchain upgrades that gap has closed and the difference is no longer significant between the two.
5) Many people complain about UBER and switch to other kernels because of GCC 4.10 This reason is obviously pretty silly... especially since I don't solicit donations or make a livelihood out of doing this (Quick note to clarify my stance on donations so you don't misunderstand me. Of course I enjoy it when people show their thanks for the many hours I spend doing this for the benefit of us all by sending a little donation but no I don't expect it or advertise it). It really doesn't hurt my feelings if you leave but if your reason for leaving is because you only tried one of my experimental 4.10 kernels..... yeah won't even comment on this one. Actually I'm more about reducing the number of complaints I receive with regards to UBER more than anything. Seeing as most of them are with regards to 4.10 (maybe 75%) they just waste my time (and yours) so what's the point of keeping 4.10 around anyways?

So as you can see I've really put a lot of thought into this and I really do think it is time to retire GCC 4.10.

At this point I've noticed that this cuts the kernel options down from 10 down to only 6. I'm open to maybe trying something new that would suit some or your needs better like a Franco Release kernel (ie NO OC/UC of CPU or GPU for stability sake and not just a fauxsound only release option in that category), a kernel without any sound modification that is "Google Factory," Linaro 4.8.4 2014.11 kernels, or maybe I should just be happy with 6 kernels now since I have to build for KK and LP which ends up being 12.
Anyhow, your votes below would be appreciated. Pick the one you would prefer to see most. If you can think of another option that wouldn't require me to completely rebuild the kernel from scratch (like using another hotplug/thermal setup for example) you can comment with your idea below. (Or you can express how terrible you think it is that I'm removing 4.10 in the comments too. I didn't give you that option in this survey because I'm interested in reading your arguments for why I should keep GCC 4.10)
Thanks for you input! It's very much appreciated!

Alles anzeigen

https://plus.google.com/10438486723571…sts/1ibkANEBpoN

MuggelPu

Zitat

December 18th, 2014
Kernel Update Info (See Also Changelog at bottom)
I've updated the kernel default IO (BFQ) now to BFQv7r6 and so I thought it was worth an update just for that. (Plus it has been 10 days) I also grabbed a couple more commits from google 3.4.y common branch and a couple kernel.org updates as well.
Toolchain Related Info
For those of you that haven't been following my activity in G+ [COLOR=#ff0000]I've announced that 4.10 is now discontinued[/COLOR]. You can read more about my reason here:https://plus.google.com/104384867235...ts/1ibkANEBpoN. The vote was a 2/3rds majority so even if I tried to veto I would be overridden by the house and senate. I really feel this decision is for the best. If you are upset by this don't worry I've made it quite easy for you to download my 4.10 toolchains and unofficially build UBER for yourself. http://uber.oceighty.co/TOOLCHAIN/ Find a tutorial about how to make the zImage and use this toolchain. Once you finish kernel compile copy zImage-dtb into uber zip as zImage and flash and you'll be good to go.
Speaking of toolchains I decided to make my linaro toolchains easier to build now than ever before. As long as you are running 64-bit ubuntu (I would say do 14.04 LTS but it may work on others) you should just be able to follow this Manifest ReadMe ([DLMURL]https://github.com/UBERUTILS/linaro-manifest[/DLMURL]), run the scripts, and make yourself a linaro arm-eabi toolchain (unofficial obviously). I've been asked before why I don't thank Christopher83 (see his toolchain work here: Mobiflip) in my thread like everyone else in this forum seems to do for the use of Linaro toolchains. The reason why is he builds gnueabi toolchains and I build arm-eabi. We both use the same linaro snapshots for main toolchain components but his toolchains include some additional components (Namely Eglibc and an entire 3.4.105 kernel source) which allow his toolchains to compile the kernel faster but takes literally like 20-30 minutes for him to build (unless he's got a beastly computer which I would image he does based on how many toolchains he builds). My arm-eabi toolchains take less than 10 minutes to build but do not build the kernel as fast however I find the difference in kernel compile is really more or less a minute and so I decided on arm-eabi. This makes sense for me because I'm always updating and testing new toolchain updates it made more sense to go the faster toolchain compile route than the faster kernel compile. I realize this is probably more than you wanted to hear on this subject but I just don't want to have to explain to everyone via PM. It's much easier to inform you all here publicly that is the difference between my linaro toolchains and his gnueabi.
Oh and one more thing, before anyone asks which one is faster/has better battery/benchmarks higher and all of that jazz I'm going to inform you that I really don't know but I am willing to find out if you are willing to test. I'm going to see if I can build uber with my arm-eabi and the very popular linaro gnueabi so you can realistically compare them side by side. I'm interested to know if it really is worth it to spend half an hour building a gnueabi toolchain or not. How many of you are willing to test? Keep an eye out this weekend for the Linaro Toolchain comparison test!

F2FS Support Info
Many users are still very confused about F2FS and how it all works. If in doubt flash the permissive F2FS/EXT4 kernel. That kernel should boot no matter what. Here are links to information regarding F2FS.
Explanation of Different F2FS/EXT4 Kernels found on BasketBuild (also talk of MultiROM): https://plus.google.com/104384867235...ts/jgGGnRsuQiy
Tutorial on Converting to F2FS: https://plus.google.com/104384867235...ts/7No16fUQaFz
Due to some people having difficulty in the F2FS conversion process here is my current simplified recommendation for you should you choose to accept it. Rather than convert all partitions to F2FS convert your cache and data (which will erase you internal sdcard and all files there so MAKE A BACKUP BEFORE YOU BEGIN!!!) and keep you system EXT4 without wiping it and leave the ROM there so you can easily boot up. Don't forget to install either the F2FS/EXT4 permissive or enforcing kernel before you proceed to wipe you system!!! Once you are booted you can restore you settings, copy files over to your phone again, etc. This will give you the full benefits of F2FS while enabling you to flash any Lollipop ROM (since they all flash on EXT4 system and they all support F2FS thanks to Google). You'll need to continue to flash the F2FS/EXT4 Kernel after every ROM install but it will be much easier to switch ROMs. As I have said before system is Read Only and EXT4 and F2FS read at roughtly the same speed. It is the F2FS write speed which is significantly faster than EXT4 which is why you'll want cache and sdcard formated to F2FS (because they do lots of writes all of the time). Before someone ask well what about MultiROM? Here is what I have to say on the subject: If you are using MultiROM... good luck... I'm not even going to go there....
Broken Charging Animation when Phone Powered Off?
You may also notice that the charging animation is broken on some ROMs when the phone is powered off. You can also flash those F2FS/EXT4 kernels to fix that animation if it bothers you that it doesn't work whichever ROM you are on.
Is UBER not working for you on your favorite ROM?
I've recently become aware that UBER is not working on certain ROMs even after flashing stock kernel, removing init.d scripts, etc. One such ROM is slim apparently. I haven't had time to investigate but while I'm at it I might as well investigate any other ROMs that don't seem to like UBER to see if I can do some fancy install script voodoo to broaden UBER compatibility with all AOSP ROMs. If you have discovered any other ROMs that aren't slim or slim-based that don't work let me know and I'll download and investigate them too this weekend.
I think that just about covers all of the questions being asked... if not, until next time!
Enjoy the Updates!
Cheers!
Cl3Kener

Alles anzeigen

Zitat

Kernel Changelog
914b8c4 mm: vmscan: If kswapd has been running too long, allow it to sleep
a5a09fe CPU hotplug, debug: detect imbalance between get_online_cpus() and put_online_cpus()
a355533 mm, vmalloc: remove useless variable in vmap_block
d136491 mm, vmalloc: use well-defined find_last_bit() func
cc985a6 writeback: fix race that cause writeback hung
dcdfde6 binfmt_elf.c: use get_random_int() to fix entropy depleting
1fc0cb1 block: Update to BFQv7r6
4d400c3 net: ipv6: allow choosing optimistic addresses with use_optimistic
43203ce net/ping: handle protocol mismatching scenario
de0e56e net: ipv6: Add a sysctl to make optimistic addresses useful candidates

Alles anzeigen

Lollipop Links
Main: http://uber.oceighty.co/HAMMERHEAD/UBER-L/

Mirror: Business Insider

Kitkat Links

Main: http://uber.oceighty.co/HAMMERHEAD/UBER/

Mirror: Business Insider

MuggelPu

Bissl INFO

[COLOR=#ff0000]Linaro Toolchain Wars - arm-eabi vs. gnueabi[/COLOR]

Zitat

Okay so it's time for a toolchain show down as I mentioned I would do yesterday. Let's lay down some ground rules before we get too carried away.
Both Toolchains are tuned to these parameters to be equally tuned: cortex-a15, neon-vfpv4, and hard float so this should be a fair fight.
As mentioned previously the build methods are completely different. gnueabi uses a tool called crosstool-ng and includes and uses a specific kernel snapshot (3.4.105 for example) and eglibc to assemble. These additional components allow it compile kernels faster but also add more time to the toolchain build itself (20-30 minutes per toolchain). arm-eabi uses a partial android build environment and makes use of android bionic to optimize itself. In this case I used optipop bionic to make it device specific. I build arm-eabi via script which is similar to how linaro builds their snapshots and it takes less than 10 minutes for a complete toolchain build. I've always used arm-eabi because I update the kernel toolchain all of the time and I liked being able to rebuild quickly but I'm interested to know which one is better suited for UBER (or if they are more or less equal) so that is why I'm enlisting your help!

What I'm looking for from anyone willing to participate:
1. You must flash both kernels on the SAME ROM and should spend at least a day on each (but 2 or 3 days would be best if you can, this is not a quick fight this is all out toolchain war!)
2. You must keep track of battery life as best as possible. Take screenshots of you SOT, Stand by, voice calls, other apps you think are significant. I will be posting a short questionnaire/survey where I expect you to have some details on battery life. I don't need to see all of your screenshots but you are welcome to post them all here in this community for us to all see.
3. Try and to keep track of performance the best you can Obviously I don't want you to shy away from benchmarks. By all means run as many of them as you like. I also understand that benchmarks don't always give the whole picture on performance so I'm also looking for you to describe any differences you see between these two linaro kernels. Comparison and contrast are the preferred method. (ie. This one opens camera app faster than that one. This one has smoother scrolling than that one. This one has a more fluid UI than that one. etc.)
4. Look for anything else that is measurable that you can test Test boot times, test app opening times on large apps like Facebook. Look for operations that cause lag on the kernel and try to reproduce the lag doing the same thing on the other etc.
5 Most importantly have fun and keep track of your observations! If this isn't going to help make your weekend more enjoyable don't participate. I'm not trying to ruin your weekend.

How I plan on determining the winner
I will be posting a survey for you to take either Sunday or Monday with some questions relating to these things I've mentioned above for you to write you experiences/feedback. Right now I want you to focus on the kernels and not be distracted by extra links. I'll be testing right with you so I'll make a summary post of what the results of the test show.
Good luck and thank you for taking the time to help out! The future of which Linaro toolchain I use on UBER depends on you!
Cheers!
Chet

Download Links
ARM EABI:http://uber.oceighty.co/HAMMERHEAD/UBE…0-2014-Test.zip
GNUEABI:http://uber.oceighty.co/HAMMERHEAD/UBE…0-2014-Test.zip

Alles anzeigen

https://plus.google.com/10438486723571…sts/MVYvonN2mf2

MuggelPu

Zitat

January 1st, 2015
Happy New Year Everyone!
To kick off the new year let's celebrate with new Kernels to play with!
So apparently there was some idle over estimation in the scheduler which was causing CPUs to jump towards max more than they should have been. I grabbed the patch for this and I'm hoping it will help us get a little bit more battery and won't slow us down any.
I also found a patch that removed some dead armv6 code from spinlock and replaced it with newer/better multicore code for our processor.
Toolchains were updated as always can be found http://uber.oceighty.co/TOOLCHAIN/. If you haven't had a chance to try the new Linaro ARM-EABI or Linaro Gnueabi UBER kernels you really ought to consider mixing things up this new year. Just for clarification sake kernels labeled Linaro are arm eabi and ones label Linaro-gnueabi are made with linaro gnueabi. Gnueabi is the more popular type of Linaro toolchain that you'll find many kernels in this forum use. Linaro arm-eabi is manufactured by yours truly. Give them both a spin and see what you think. Also, in case you missed it, here are some pretty indepth results on how these two compare: https://plus.google.com/102836405889...ts/AHoavmKHrKf
If you aren't up for an adventure there's always SaberMod which is fast, reliable, and battery friendly as well. Don't change to Linaro if you are satisfied with SaberMod and happy with how things are going for you.
Happy New Year!!!
Enjoy the Updates!
Cl3Kener

Alles anzeigen

Zitat

Kernel Changelog
adea4a2 Revamp spinlock ASM code
be115df sched: Reduce overestimating rq->avg_idle

Lollipop Links

Main:http://uber.oceighty.co/HAMMERHEAD/UBER-L/

Mirror:Hier

Kitkat Links

Main:http://uber.oceighty.co/HAMMERHEAD/UBER/

Mirror:Hier

MuggelPu

Kernel Update und gaaanz viel INFO auf g+ ...

Mobilegeeks.de

EdgA

https://plus.google.com/10438486723571…sts/TFitmk1J1ZV
Update

MuggelPu

Kernel Update ... UBERTC

Zitat

2/28/15
- Merged latest Color Drivers for N5 (neobuddy)
- Changed Toolchains from SaberMod (Now partially closed source so that you cannot build for yourself anymore) to an AOSP-based open source toolchain called UBERTC. Many patches from Linaro have been included in source and also some GCC build fix patches from SM to allow for the latest toolchain components. All sources are GPL and can be found https://github.com/UBERTC/. New toolchains have less of a tendency to lag in scrolling and other screen transitions in comparison to SaberMod and I think are better suited for UBER.

[KERNEL][N5] UBER Kernel

Jetzt mitmachen!

Benutzer online in diesem Thema