bKernel I can guarantee that this is one of the fastest, most stable, and smoothest running kernels available for the Triumph!
Credit goes to shane87(my brother). He is the one who got this build going, and without him I would not be building a kernel. He has taught me everything I know about linux and helped greatly with everything Android I have ever done
Credit also goes to pwnyourace, Napsta, DooMLoRD, and FXP for getting the overclock patch working on these phones, as well as the VDD voltage control, as well as existz and the TalonDEV Team, tickerguy & isaacj87, mantera, Whyzor, Cyanogen and the CyanogenMod team, JoesephMother, Vampirefo, joeykrim, faux123 and many many more along the way.
StockROM bKernel
Kernel Features:
**Bold features indicate new features**
Code:
* Built using linaro's android-toolchain v4.6.3 (Updated to the latest version)
* OC/UC from 61MHz, to 2016MHz, with lots of frequencies in between.
* Undervolt support for all frequencies(Working but me and my brother need to tweak it to get it to work with his Control Freak app we will be using along with this kernel)
* Sio and v(r) I/O schedulers, with v(r) set to default. No-op and ondemand are also included
* Interactive, interactiveX, smartass, smartassv2, smoothass, SavagedZen, performance, ondemand, lagfree, minmax, scary, and conservative govenors. Removed powersave and userspace
* Ext2, ext3, and ext4 support
* Switched back to ext3 mounting
* ISO and UDF cd-rom filesystems supported
* NTFS filesystem read and write support
* Disabled kernel debugging
* Swap support
* Removed init.qcom.post_boot.sh service from the initram so no more worrying about your CPU governor and min frequency not sticking
* CIFS Support
* TUN/TAP support
* JHASH v3
* TINY RCU
* Built with 02 build-flags
* Insecure(so adb shell always has root access)
* I/O schedulers tweaked for NAND memory devices
* Page table optimizations(thanks to spica234)
* SPINLOCK optimization(thanks to spica234)
* Backported I/O fixes from 2.6.33 and 2.6.34 kernel versions(thanks to existz)
* UTF8 support
* Added memcopy(thanks to mantera)
* Undervolt support down to 600mV instead of 750mV
* Switched to Whyzor's latest touchscreen driver
* Added genlock support
* Modified the interactive governor(thanks to mantera)
* Added the two-way calling patch, for more info view this thread: [DEV] Two-way call recording on Desire [ALMOST SOLVED][Sept. 7 update] - xda-developers
* Changed the 1209 MHz state to 1200 MHz state after viewing the Code Aurora msm 3.0 android kernel source
* Increased GPU memory size from 2 to 4
* Enabled force loading of modules with mismatched kernel versions for when I start patching the kernel up to new versions
* Patched kernel up to 2.6.32.59, which is the highest 2.6.32.x kernel version
* Added zRam support, it is initialized by default, and can be enabled by an init.d script
* Disabled lots of debugging from the kernel
* Optimized for Snapdragon SoC's via adding in some CFLAGS to the Makefile
* SFB(Stochastic Fair Blue) Net scheduler enabled
* Fixed the export of scailing_available_frequencies
* Added cleancache
* Added FastFPE
* Various fixes, including memory leak fixes and fixes to TinyRCU, and tweaks
* Fixes to zRam, including adding in a swap spot free callback, fixed sparse warning, made zRam have a default disksize, so now it is setup and ready to go you just have to call mkswap on the device
* Updated lowmemory killer source to latest AOSP source
* Updated binder source to latest AOSP source
* Added fudgeswap, thanks to faux123
* Optimized memcpy & memmove
* More RCU updates, fixes and improvements
* Removed per_cpu based spinlocks
* Added SWP/SWPB emulation
* Optimized arm RWSEM algorithm
* Allowed CPU-support unaligned accesses
* Allow eMMC to be put to sleep before suspend
* Changed kernel version layout
* Fixed a frame size issue in the bunzip2 (decompressor for bzip2) code
* PMEM changes to make gralloc work correctly
* Fixed memory leak in bluetooth thanks to dzo
* Changed the default sleep times to hopefully encourage faster sleeping in kernel
* Enabled KSM for ROM's that support KSM
* Enabled SWPB emulation in the kernel
Feel free to follow me on that repo to see what changes we add to it. And I will post any public test versions here on this thread!
Feel free to donate to me and my brother if you like our work!(He needs and deserves it more than me!!!) You can donate to me here on my signature, and to donate to shane87 go to XDA and find a post by him and he has a donate to me button under his name!!
As always check the repo link posted above for more information on changes I have made since the beginning.
Changelog: 2.3: Fixed frame size issue in bunzip2, changes to PMEM for gralloc to work correctly, fixed leak in bt, enabled KSM and SWPB emulation, changed default sleep times
2.2: Updated binder and lowmemory killer souce to latest AOSP source, many fixes to swap and zRam, added fudgeswap, optimized RWSEM algorithm, memcpy and memmove optmizations, various RCU fixes, changed kernel version layout, plus more
2.0: Patched to 2.6.32.59, compiled with linaro's latest android toolchain, added zRam, SFB, FastFPE, cleancache, disabled lots of debugging, many fixes
1.01: Used Whyzor's latest touchscreen driver, added genlock support, enabled UV support down to 600 MHz, chaged from 1209 MHz to 1200 Mhz, enabled forced module loading, increased GPU physical memory size, added two-way calling patch, and updated interactive governor code, removed ext2 version
0.24-DEV: Added lagfree, minmax, scary and conservative governors. Enable UV support down to 650. Added memcopy
0.23-DEV: Added TINY RCU, ext2 mounting in init.rc, added 61 MHZ frequency, added I/O fixes, page table and spinlock optimizations, CIFS, TUN/TAP, and UTF8 support
0.17-DEV: Initial working release
CM7 bKernel
Kernel Features:
The CM7 version has now been caught up and includes all the features of the latest bKernel, and it works with all the latest CM7 builds. It has the updated kgsl drivers, as well all the latest CM7 fixes by Whyzor and mantera!!
**Bold features indicate new features**
Code:
* Built off merged sources from Tickerguy's latest CM7 kernel build, and my bKernel DEV build
* Built using linaro's android-toolchain v4.6(Updated to the latest version)
* OC/UC from 61MHz, to 2016MHz, with lots of frequencies in between.
* Undervolt support for all frequencies(Working but me and my brother need to tweak it to get it to work with his Control Freak app we will be using along with this kernel)
* Sio and v(r) I/O schedulers, with v(r) set to default. No-op and ondemand are also included
* Interactive, interactiveX, smartass, smartassv2, smoothass, SavagedZen, performance, ondemand, lagfree, minmax, scary, and conservative govenors. Removed powersave and userspace
* Ext2, ext3, and ext4 support
* ISO and UDF cd-rom filesystems supported
* NTFS filesystem read and write support
* Disabled kernel debugging
* Swap support
* CIFS Support
* TUN/TAP support
* JHASH v3
* Built with 02 build-flags
* Insecure(so adb shell always has root access)
* I/O schedulers tweaked for NAND memory devices
* Page table optimizations(thanks to spica234)
* SPINLOCK optimization(thanks to spica234)
* Backported I/O fixes from 2.6.33 and 2.6.34 kernel versions(thanks to existz)
* TINY RCU
* UTF8 support
* Fixed WiFi hotspot issue for Tickerguy's final beta release of CM7
* Added memcopy(thanks to mantera)
* Undervolt support down to 600mV instead of 750mV
* Switched to Whyzor's latest touchscreen driver
* Added genlock support
* Modified the interactive governor(thanks to mantera)
* Added the two-way calling patch, for more info view this thread: [DEV] Two-way call recording on Desire [ALMOST SOLVED][Sept. 7 update] - xda-developers
* Changed the 1209 MHz state to 1200 MHz state after viewing the Code Aurora msm 3.0 android kernel source
* Increased GPU memory size from 2 to 4
* Enabled force loading of modules with mismatched kernel versions for when I start patching the kernel up to new versions
* Patched kernel up to 2.6.32.59, which is the highest 2.6.32.x kernel version
* Added zRam support
* Disabled lots of debugging from the kernel
* Optimized for Snapdragon SoC's via adding in some CFLAGS to the Makefile
* SFB(Stochastic Fair Blue) Net scheduler enabled
* Fixed the export of scailing_available_frequencies
* Added cleancache
* Added FastFPE
* Various fixes, including memory leak fixes and fixes to TinyRCU, and tweaks
* Whyzor & mantera's fixes to Bluetooth wake-locks
* Updated kgsl driver kernel code
* Disabled GPIO input support to help with proximity issues in CM7
* Fixes to zRam, including adding in a swap spot free callback, fixed sparse warning, made zRam have a default disksize, so now it is setup and ready to go you just have to call mkswap on the device
* Updated lowmemory killer source to latest AOSP source
* Updated binder source to latest AOSP source
* Added fudgeswap, thanks to faux123
* Optimized memcpy & memmove
* More RCU updates, fixes and improvements
* Removed per_cpu based spinlocks
* Added SWP/SWPB emulation
* Optimized arm RWSEM algorithm
* Allowed CPU-support unaligned accesses
* Allow eMMC to be put to sleep before suspend
* Changed kernel version layout
* Fixed a frame size issue in the bunzip2 (decompressor for bzip2) code
* PMEM changes to make gralloc work correctly
* Fixed memory leak in bluetooth thanks to dzo
* Changed the default sleep times to hopefully encourage faster sleeping in kernel
* Enabled KSM for ROM's that support KSM
* Enabled SWPB emulation in the kernel
Changelog: 2.3: Fixed frame size issue in bunzip2, changes to PMEM for gralloc to work correctly, fixed leak in bt, enabled KSM and SWPB emulation, changed default sleep times
1.4: Updated binder and lowmemory killer souce to latest AOSP source, many fixes to swap and zRam, added fudgeswap, optimized RWSEM algorithm, memcpy and memmove optmizations, various RCU fixes, changed kernel version layout, plus more
1.2: Patched to 2.6.32.59, compiled with linaro's latest android toolchain, added zRam, SFB, FastFPE, cleancache, disabled lots of debugging, many fixes, imported mantera and Whyzor's latest CM7 kernel changes, updated kgsl driver code
DEV-0.5: Added lagfree, minmax, scary and conservative governors. Enable UV support down to 650. Added memcopy
DEV-0.4: Fixed WiFI hotspot support
DEV-0.3: Added TINY RCU, added 61 MHZ frequency, added I/O fixes, page table and spinlock optimizations, CIFS, TUN/TAP, and UTF8 support
DEV-0.2: Initial release
One last thing: PLEASE OVERCLOCK AT YOUR OWN RISK!!! THIS PHONE WILL GO TO 2+ GHz BUT RUNNING AT HIGH FREQUENCIES FOR PROLONGED TIMES MAY RESULT IN DAMAGE TO THE CPU CHIPSET, I AM NOT RESPONSIBLE FOR ANY ILL FATED OUTCOMES WHILE USING MY KERNELS!!! YOU HAVE BEEN WARNED!!!!
Some new information I have found puts the max frequency around 1.4 or 1.5 GHz for the 2nd gen Snapdragon SoC's(MSM7x30 and MSM8x55). I would not recommend going above this, but I have left the 1.6 and above frequencies, but you use and of my kernel at your own risk!!!
Location: Boondocks, Florida, where Sprint sucks, along with Verizon!
Posts: 186
Device(s): Motorola Atrix 4G
Blur (modded) 2.3.7
Carrier: Not Provided
Thanks: 111
Thanked 22 Times in 17 Posts
Great
Quote:
Originally Posted by b_randon14
This thread is for bKernel which will be featured in bROM for the Motorol Triumph. This kernel is still in the development stage(hence the DEV tag) and I do not have a release version yet. That being said I will provide you my github address to the repo for this kernel so you can watch and see the changes being made.
Credit goes to shane87(my brother). He is the one who got this build going, and without him I would not be building a kernel. He has taught me everything I know about linux and helped greatly with everything Android I have ever done
Credit also goes to pwnyourace, Napsta, DooMLoRD, and FXP for getting the overclock patch working on these phones, as well as the VDD voltage control.
Kernel Features(Bold indicated working features in my dev test versions):
OC/UC from 122MHz, to 2016GHz, with lots of frequencies in between.
Undervolt support for all frequencies(Working but me and my brother need to tweak it to get it to work with his Control Freak app we will be using along with this kernel)
Sio and v(r) I/O schedulers, with v(r) set to default. No-op and ondemand are alos included
Interactive, interactiveX(Default for now), smartass, smartassv2, smoothass, minmax(not working at the moment), lagfree(forgot to add on the last test build), SavagedZen, performance, and ondemand govenors. Removed conservative, powersave and userspace
Ext2 and ext4 support
Mounting of ext4 in the init.rc(will require me to format the partitions to ext4 through the updater script in my ROM)
ISO and UDF cd-rom filesystems supported
NTFS filesystem read and write support
Disabled kernel debugging
Swap support
Zram(a better compcache) will hopefully be implemented if we can get the modules to build right.
Applying LMK(Low memory killer) settings in the init.rc through a configurable ram.conf file.
My brothers Memory Freak app configured to work on our phones to control the memory settings(Zram and LMK settings)
My brothers Control Freak app configured to work on our phones to control the OC/UC and Undervolt settings, as well as the governors and I/O schedulers(I think it supports changing the I/O scheduler anyways!)
As many tweaks and optimizations I can cram into this kernel!
Feel free to follow me on that repo to see what changes we add to it. And I will post any public test versions here on this thread!
Feel free to donate to me and my brother if you like our work!(He needs and deserves it more than me!!!) You can donate to me here on my signature, and to donate to shane87 go to XDA and find a post by him and he has a donate to me button under his name!!
Will it work with Cyanogen? this might be a dumb question...
__________________
If I was helpful, the thanks button doesn't bite!
Nah it's for the stock ROM. Im not sure what all would have to be done to make it for CM but I think it's all in the init.rc so it may not be hard to do!
Good work dude, I knew you would make a kernel soon. Keep on taking care of stock!
That's all I know how to do
I should be posting a DEV release build today or tonight. It is still in the dev stage, but it will feature oc/uc from 122MHz to 2016MHz, and it defaults to 1024MHz, so it should boot on any phone, I also modified the ramdisk to remove the service where init.qcom.post_boot.sh is enabled, so that even if you don't modify that file, your frequencies and governor will stick because it never initiates that script in the first place. The ramdisk also contains a minimal busybox so that even without run-parts on your phone memory, init.d scripts will still be executed. There is alot more I did but you can look at that on github lol!
Device(s): HTC One S (CM10),
Kindle Fire (CM10-SGT7)
Carrier: T-Mobile USA
Thanks: 3,844
Thanked 5,034 Times in 2,658 Posts
Quote:
Originally Posted by Austrie
Edit: also some people triumphs can barely handle past 1.5ghz, so I know some random person will be sorry at 2ghz....
And we will see a flood of posts about how people fried their phones and asking why! :P OC'ing is nice, but people need to know what they are doing before messing with that. At least with this phone people can do a lot of things wrong and still get the phone working again, but OC'ing it too much and not being smart about it will screw them over. Make sure the warning on this is HUGE! lol.
And we will see a flood of posts about how people fried their phones and asking why! :P OC'ing is nice, but people need to know what they are doing before messing with that. At least with this phone people can do a lot of things wrong and still get the phone working again, but OC'ing it too much and not being smart about it will screw them over. Make sure the warning on this is HUGE! lol.
yea i dont go past 1.5
The Following User Says Thank You to ziggy46 For This Useful Post:
OVERCLOCK AT YOUR OWN RISK. TO HIGH OF FREQUENCIES FOR PROLONGED PERIODS OF TIME CAN CAUSE DAMAGE TO THE CPU AND OTHER PARTS OF THE INTERNAL CHIPSET. YOU HAVE BEEN WARNED, USE AT YOUR OWN RISK!!
Lol!!!
The Following User Says Thank You to b_randon14 For This Useful Post:
2016 GHz? Thats a lot of GHz! Are you sure it wasn't MHz?
*dreams of ACTUAL 2016 GHz phones, which we'll probably have in a decade*
Lol 2.16 THz and with built in R-134 cooling so our hands don't spontaneously combust under the insane temp of that beast running full bore to display holographic 1080p videogames that we use our minds to control...
Ha lol yeah its definitely not capable of 2.16 THz lol! Proofreading has never been my strong suit lol! I'm going to run a 2ghz quad lol and hopefully my phone doesn't meltdown in my hand!!
R-134 cooling now there is an awesome idea hive it 5 years lol!
Device(s): HTC EVO V 4G, Running: AOKP 4.1.2 v3.5.2,
Skill Level(s): App, ROM, & Kernel Dev
Carrier: Virgin Mobile
Thanks: 0
Thanked 790 Times in 292 Posts
This kernel should be fast, GUARANTEED faster than stock CM7, even at same voltage, because of this kernel using ext2.
~~~
B_randon I was thinking of using ext2 too, but to many "new" people are using the triumph, so I decided to stick with ext4. To many people like to do battery pulls...
>* Mounting of ext2 in the init.rc(This will mount the ext3 partitions as non-journaled ext2 partitions)
I'm confused. I assume this init.rc file that you mention is in the ramdisk, correct? Can you explain how this would affect the final filesystem once the actual file system is loaded? And did you figure out where the file system gets mounted as ext4?
I thought that the actual file system that is formatted during the actual ROM flashing? For example, looking in the updater script for the cm7 zip package, I see:
I assume this means that during the flashing of CM7, /system is formatted as ext4 and so it will remain as ext4 until you reformat it as something else, correct?
Device(s): HTC EVO V 4G, Running: AOKP 4.1.2 v3.5.2,
Skill Level(s): App, ROM, & Kernel Dev
Carrier: Virgin Mobile
Thanks: 0
Thanked 790 Times in 292 Posts
Quote:
Originally Posted by mantera
>* Mounting of ext2 in the init.rc(This will mount the ext3 partitions as non-journaled ext2 partitions)
I'm confused. I assume this init.rc file that you mention is in the ramdisk, correct? Can you explain how this would affect the final filesystem once the actual file system is loaded? And did you figure out where the file system gets mounted as ext4?
I thought that the actual file system that is formatted during the actual ROM flashing? For example, looking in the updater script for the cm7 zip package, I see:
I assume this means that during the flashing of CM7, /system is formatted as ext4 and so it will remain as ext4 until you reformat it as something else, correct?
Only bkernel is ext2, cm7 is still ext4. Stock triumph is ext3, when you it becomes ext4, you have to change "("ext3", "EMMC"," to "("ext4", "EMMC",". It have 3 different sections that can be formated, data, cache, and system. In all you can have a mix combination of 3 filesytems on one phone. bkernel mount everything as ext2, so its a ext2 kernel, while cm7 and MIUI is ext4.
I'm confused(or I find it unnecessary) about, using the update script to change to ext4 if the kernel is already ext4, its not like the kernel is a different filesystem.
Edit: Its not the Init.rc in ramdisk...
Last edited by Austrie; November 16th, 2011 at 07:22 PM.
>* Mounting of ext2 in the init.rc(This will mount the ext3 partitions as non-journaled ext2 partitions)
I'm confused. I assume this init.rc file that you mention is in the ramdisk, correct? Can you explain how this would affect the final filesystem once the actual file system is loaded? And did you figure out where the file system gets mounted as ext4?
I thought that the actual file system that is formatted during the actual ROM flashing? For example, looking in the updater script for the cm7 zip package, I see:
I assume this means that during the flashing of CM7, /system is formatted as ext4 and so it will remain as ext4 until you reformat it as something else, correct?
Well that is on the stock based kernel. Yeah the init.RC is in the ramdisk and it mounts everything and sets everything up to boot. So if the partition is formatted as ext3 like it is from the factory, you can have the init.rc mount that ext3 partition as an ext2 because it is backwards compatible with ext2. On cm7 for some reason, the init.qcom.RC is what mounts the partitions. It mounts them as ext4. The ext3 partitions can be mounted as ext4 as well cause if you notice in tickerguys cm7 updater script he only formats system and maybe cache but it doesn't format data as ext4 but through the initram data gets mounted as ext4. But like on the intercept you could mount an rfs partition as ext2 or ext4 because they ain't compatible(rfs is based off fat) so you had to convert to one or the other. And an ext2 formatted partition can't be mounted as ext4 I believe. The updater script actually formats the partition, but with the initrc youcan choose what to mount so long as whatever the partition is formatted as can be mounted as that(I.e ext3 mounted as ext4).
As for battery life I haven't put much hours on the cm7 build but on the stock based roms I get pretty good battery life
>* Mounting of ext2 in the init.rc(This will mount the ext3 partitions as non-journaled ext2 partitions)
I'm confused. I assume this init.rc file that you mention is in the ramdisk, correct? Can you explain how this would affect the final filesystem once the actual file system is loaded? And did you figure out where the file system gets mounted as ext4?
I thought that the actual file system that is formatted during the actual ROM flashing? For example, looking in the updater script for the cm7 zip package, I see:
I assume this means that during the flashing of CM7, /system is formatted as ext4 and so it will remain as ext4 until you reformat it as something else, correct?
Well that is on the stock based kernel. Yeah the init.RC is in the ramdisk and it mounts everything and sets everything up to boot. So if the partition is formatted as ext3 like it is from the factory, you can have the init.rc mount that ext3 partition as an ext2 because it is backwards compatible with ext2. On cm7 for some reason, the init.qcom.RC is what mounts the partitions. It mounts them as ext4. The ext3 partitions can be mounted as ext4 as well cause if you notice in tickerguys cm7 updater script he only formats system and maybe cache but it doesn't format data as ext4 but through the initram data gets mounted as ext4. But like on the intercept you could mount an rfs partition as ext2 or ext4 because they ain't compatible(rfs is based off fat) so you had to convert to one or the other. And an ext2 formatted partition can't be mounted as ext4 I believe. The updater script actually formats the partition, but with the initrc youcan choose what to mount so long as whatever the partition is formatted as can be mounted as that(I.e ext3 mounted as ext4).
As for battery life I haven't put much hours on the cm7 build but on the stock based roms I get pretty good battery life
Device(s): HTC EVO V 4G, Running: AOKP 4.1.2 v3.5.2,
Skill Level(s): App, ROM, & Kernel Dev
Carrier: Virgin Mobile
Thanks: 0
Thanked 790 Times in 292 Posts
Quote:
Originally Posted by b_randon14
Well that is on the stock based kernel. Yeah the init.RC is in the ramdisk and it mounts everything and sets everything up to boot. So if the partition is formatted as ext3 like it is from the factory, you can have the init.rc mount that ext3 partition as an ext2 because it is backwards compatible with ext2. On cm7 for some reason, the init.qcom.RC is what mounts the partitions. It mounts them as ext4. The ext3 partitions can be mounted as ext4 as well cause if you notice in tickerguys cm7 updater script he only formats system and maybe cache but it doesn't format data as ext4 but through the initram data gets mounted as ext4. But like on the intercept you could mount an rfs partition as ext2 or ext4 because they ain't compatible(rfs is based off fat) so you had to convert to one or the other. And an ext2 formatted partition can't be mounted as ext4 I believe. The updater script actually formats the partition, but with the initrc youcan choose what to mount so long as whatever the partition is formatted as can be mounted as that(I.e ext3 mounted as ext4).
As for battery life I haven't put much hours on the cm7 build but on the stock based roms I get pretty good battery life
Filesystems being mounted as a another, confusing like hell....
Well that is on the stock based kernel. Yeah the init.RC is in the ramdisk and it mounts everything and sets everything up to boot. So if the partition is formatted as ext3 like it is from the factory, you can have the init.rc mount that ext3 partition as an ext2 because it is backwards compatible with ext2. On cm7 for some reason, the init.qcom.RC is what mounts the partitions. It mounts them as ext4. The ext3 partitions can be mounted as ext4 as well cause if you notice in tickerguys cm7 updater script he only formats system and maybe cache but it doesn't format data as ext4 but through the initram data gets mounted as ext4. But like on the intercept you could mount an rfs partition as ext2 or ext4 because they ain't compatible(rfs is based off fat) so you had to convert to one or the other. And an ext2 formatted partition can't be mounted as ext4 I believe. The updater script actually formats the partition, but with the initrc youcan choose what to mount so long as whatever the partition is formatted as can be mounted as that(I.e ext3 mounted as ext4).
Ahh ok. After a lot more looking through the files, I see the mount commands now. thanks.
Anyone know how battery life is with the CM7 version of this kernel?
I've been using mantera's OC kernel for the last few days and battery life has tanked compared to Tickerguy's...
Are you overclocking? If so, I would expect the battery life to be worse since you're using more power... Try setting back to 1024 and does that put your battery usage back to what you were expecting?
Filesystems being mounted as a another, confusing like hell....
Isn't it the init.rc in kernel?
Ha yeah lol. It's in the init.rc of the stock ROM kernel, now on CM7 its in init.qcom.rc(tickerguys) or I believe in isaac's builds it's init.triumph.rc, and I assume its somthing like that for the MIUI kernel as well. Yeah I believe that ext3 will mount as ext2 or ext4, and ext4 will mount as ext3 or ext2, but I believe ext2 will only mount as ext2 and ext3. Not 100% on that, but you could find out on google or somthing. Now you an't do something crazy like mount a ext3 partition as ntfs or fat or somthing lol!!! But the ext(extended filesystem) are somewhat backwards and even forwards compatible.
@ziggy, possibly if the kernel is requesting a write to the nand at the moment you pull the battery and all the writes hadn't been synced before. But i never had had a problem. And it's not so much loosing data, just it becoming corrupted. It may still appear to be there, but not work quite right. Like it still shows you have a text file there, but it don't open right. I have never had a problem that I have seen surface itself using ext2
i've been doning some research on the MSM8655 which is the chipset used in the triumph, and most of the OCing info i'm finding is based off the HTC thunderbolt because it uses the same GPU and chipset as this phone. now from what i could find is that the chipset itself is stable at 1.9ghz, but not at 2ghz. and i went through multiple articles to confirm this. from everything i read though is that the thunderbolt runs stable at 1.996GHz, but our phones when you use the 1.9GHz freq is only running at 1.901GHz, which in theory SHOULD mean our chipset should be stable at this speed, but as we all know this isn't the case at all with a lot of phones not being able to work above certain freq's most likely because of the imperfection of processor making, but if you google around a bit you will find out that the thunderbolt runs stable at 1.996GHz.
Yeah software can affect how stable a cpu runs. For example on the samsung based ROMs for the captivate(this is what phone shane87, my brother, has) they can only go to about 1400 stable, more like 1200 and be completely stable. But on the official CM7 builds, 1.6 is really stable. Their SoC is the hummingbird though and it's a little bit older, and not as robust.
I was wondering for a 3 days, why can the evolution shift handle 2ghz and we have a better processor. Next thing bkernel comes out with 2ghz...
Edit: also some people triumphs can barely handle past 1.5ghz, so I know some random person will be sorry at 2ghz....
I think voltages are coming into play as well. My phone was not very stable past 1.4 until we got voltage control in the kernel. I now have been at 61 min 1.7 max for the last few days with a gentle undervolt at low steps and a a slight overvolt at the 1.6 and 1.7 Ghz frequencies... It has been rock solid no reboots, cold to the touch, and running around 15-18hrs per full charge of here and there use. I`m guessing there are variations in the powering to chip from phone to phone... thus contributing to the stability or lack thereof at the upper rung frequencies. I`m rocking 60 MFlops in linpack and 4000 ish on antutu @ 1.7GHz. Currently sitting at 15h 42min with 32% left on battery
Last edited by streetpounder; November 18th, 2011 at 09:17 AM.
LinkBack to this Thread: http://androidforums.com/triumph-all-things-root/435668-kernel-update-05-24-2012-bkernel-v2-3-bkernel-cm7-v2-3-2-6-32-59-a.html
[KERNEL][UPDATE: 05/24/2012] bKernel-v2.3 & bKernel-CM7-v2.3 - 2.6.32.59
~Faster Internet~
