Related
Hi all.
Not sure whether this was the right place to post, but coming from a development perspective, I thought it was fitting. If not, mods - please feel free to move.
This thread is here to bring the currently worked on 3.0 kernel for the Desire to the HD2 developers' attention. There is minimal difference in the Desire hardware when compared to the HD2, and therefore would probably not be a difficult port to perform.
The main benefit of porting this kernel over to the HD2 is the improvement in ICS, namely through Hardware Acceleration. Although it hasn't been confirmed yet, the OP over in the Desire thread seems to think Hardware Acceleration is working.
I'm reaching out to all the HD2 developers to collaborate their efforts in an attempt to port this over to HD2 - it will bring us an even more well-rounded ICS build, extending the already expanded boundaries of our beloved HD2's.
The link to the original thread is here.
The link to the github source code is here.
All credits go to Chaosz-X for the current work on the kernel. I am simply acting as a messenger.
I wish you guys luck, and hopefully it won't take long to get some work going on this!
MrP.
It is not as easy as you might think. Altough the hardware is the same the radio is completely different because it is from windows mobile. Wich makes most of the work on the ics kernel useless. No and radio's can not be changed easily because they are closed source.
then how did the desire 2.6.35 kernel get ported ?
I thought hardware acceleration was already working in ICS.
jan-willem3 said:
It is not as easy as you might think. Altough the hardware is the same the radio is completely different because it is from windows mobile. Wich makes most of the work on the ics kernel useless. No and radio's can not be changed easily because they are closed source.
Click to expand...
Click to collapse
Forgive me for my lack of extensive knowledge in the kernel department, but, considering that quite a few of the currently developed kernels for the HD2 are based on the Desire sources, I would've assumed that what was done before in porting the 2.6.35 kernel, could be adapted to work with this kernel.
Once again, not saying that's right, because I'm basing that conclusion off logical thought, not experience.
Visentinel said:
then how did the desire 2.6.35 kernel get ported ?
Click to expand...
Click to collapse
My point above exactly. If it's been done before, surely the steps needed to adapt the 3.0 kernel are not far out of reach. The people I think we should be talking to are marc1706, and tytung. I'll pop them a PM when I have a moment.
KyJelly69 said:
I thought hardware acceleration was already working in ICS.
Click to expand...
Click to collapse
It's not true Hardware Acceleration - its a hack of sorts. It gives a boost in performance no doubt, but most likely no where near true Hardware Acceleration.
Thanks for the input guys.
MrP.
It might be possible to port the 3.0 kernel to the HD2 but the thread you linked is currently providing a 2.6.35.14 kernel. They are planning on releasing a 3.0 kernel.
In order to get a 100% supported hardware acceleration we will need a 3.0 kernel, since ICS was build for the 3.0 kernel. It's pretty much the same with the glitches in sense 2.1 and 3.x ROMs we had prior to using the desire .35 kernel.
marc1706 said:
It might be possible to port the 3.0 kernel to the HD2 but the thread you linked is currently providing a 2.6.35.14 kernel. They are planning on releasing a 3.0 kernel.
In order to get a 100% supported hardware acceleration we will need a 3.0 kernel, since ICS was build for the 3.0 kernel. It's pretty much the same with the glitches in sense 2.1 and 3.x ROMs we had prior to using the desire .35 kernel.
Click to expand...
Click to collapse
Thanks for your input marc1076.
I am aware that it's not 3.0 yet, however, he has backported quite a few things for now, mainly concerning hardware acceleration (as written in the OP).
Would it not be a good step to then try this kernel on our current ICS builds in the hopes of any improvement? Or perhaps alternatively incorporate the backports (possibly more difficult) into our current ICS kernel(s) in an attempt to edge closer to true Hardware Acceleration?
A few thoughts, may be possible, or not - feel free to correct me. Thanks again for the advice.
MrP.
MrPadie said:
Thanks for your input marc1076.
I am aware that it's not 3.0 yet, however, he has backported quite a few things for now, mainly concerning hardware acceleration (as written in the OP).
Would it not be a good step to then try this kernel on our current ICS builds in the hopes of any improvement? Or perhaps alternatively incorporate the backports (possibly more difficult) into our current ICS kernel(s) in an attempt to edge closer to true Hardware Acceleration?
A few thoughts, may be possible, or not - feel free to correct me. Thanks again for the advice.
MrP.
Click to expand...
Click to collapse
Once I see some true ics kernels with specific commits for ics you will see them appear in our kernels.
Sent from my HTC HD2 using XDA App
warrenb213 said:
Once I see some true ics kernels with specific commits for ics you will see them appear in our kernels.
Sent from my HTC HD2 using XDA App
Click to expand...
Click to collapse
Thanks for your reply warrenb213.
However, I'm kinda lost as to what you are saying - are you saying that our current kernel (namely Tytung's kernel) for ICS is not a proper ICS kernel (which it isn't really, it's adapted from GB kernel from what I can tell), and until we have a kernel dedicated to ICS, there won't be any of the above mentioned additions?
If that's the case, what is your reasoning for saying that if I may ask? And also, where can these 'true' ICS kernels be sourced from? (As far as I know, the only 'true' ICS kernels around are based on 3.0, which in essence, defeats the object of us porting these additions back in the first place).
MrP.
MrPadie said:
Thanks for your reply warrenb213.
However, I'm kinda lost as to what you are saying - are you saying that our current kernel (namely Tytung's kernel) for ICS is not a proper ICS kernel (which it isn't really, it's adapted from GB kernel from what I can tell), and until we have a kernel dedicated to ICS, there won't be any of the above mentioned additions?
If that's the case, what is your reasoning for saying that if I may ask? And also, where can these 'true' ICS kernels be sourced from? (As far as I know, the only 'true' ICS kernels around are based on 3.0, which in essence, defeats the object of us porting these additions back in the first place).
MrP.
Click to expand...
Click to collapse
First of all, tytung's kernel which is based on hastarin's sd kernel is a base of .32 kernel from froyo source of the eVO 4g. It was not supposed to be used in gb but since it works, no worries.
Even though AOSP is opensource, the driver code for the devices is not, which makes porting difficult and moreso when the device actually never had any drivers for that specific os (read Android on HD2).
There is already an initiative (thread to be precise) in the qualcomm developer forums pleading them to opensource the qsd8250 drivers so that devs who are not so rich can backport ics to n1, desire (or HD2 for that matter) since it is clear that HTC certainly won't and google too has given up n1.
What tytung did to make his kernel work with ics is add upon the old base, it works but is dirty and should lead to unexpected issues.
The true ics kernels are based on the 3.0.1-tag of linus's repo.
You can sync into linus's 3.0.1 and compare it to the kernel source code of Galaxy Nexus, see the android changes, port them to the vanilla kernel, copy htcleo board files while rebasing them on 3.0.1 and adding the radio code from current kernels. It is far easy to say this than to actually do this when looking at the staggering size of the linux codebase.
The question is how many of the HD2-specific changes were broken by
a) changes in the upstream Linux codebase in between 2.6.32 and 2.6.40 (aka 3.0)?
b) android-specific changes between Froyo, GB and ICS
EDIT: There seems to be a working (except for USB mass storage) 2.6.35 kernel here. This at least reduces the number of upstream code changes to deal with.
Dr_Grip said:
EDIT: There seems to be a working (except for USB mass storage) 2.6.35 kernel here.
Click to expand...
Click to collapse
USB mass storage works, confirmed it once again. USB Tethering doesn't work. Still working on it.
Dr_Grip said:
The question is how many of the HD2-specific changes were broken by
a) changes in the upstream Linux codebase in between 2.6.32 and 2.6.40 (aka 3.0)?
b) android-specific changes between Froyo, GB and ICS
EDIT: There seems to be a working (except for USB mass storage) 2.6.35 kernel here. This at least reduces the number of upstream code changes to deal with.
Click to expand...
Click to collapse
And a fully working one here
Visentinel said:
then how did the desire 2.6.35 kernel get ported ?
Click to expand...
Click to collapse
Because it was an htc kernel just like the prevous evo 4g kernel so porting it was more easy. If you look at the 2.6.35 cm kernel it is highly unstable and the MSM kernel is also an AOSP kernel like the cm kernel.
jan-willem3 said:
Because it was an htc kernel just like the prevous evo 4g kernel so porting it was more easy. If you look at the 2.6.35 cm kernel it is highly unstable and the MSM kernel is also an AOSP kernel like the cm kernel.
Click to expand...
Click to collapse
That does not make sense, I think. If the Desire devs get a working kernel for the Desire, the changes needed to make it working on the HD2 should be the same no matter if they are starting from a HTC or an AOSP codebase.
That it will be more complicated to get a working 3.0 Desire kernel without HTC's codebase is a challenge the Desire devs will have to face unrelated of the porting effort to the HD2.
uzi2 said:
And a fully working one here
Click to expand...
Click to collapse
Can you confirm that everything works?
My kernel is forked from ACA 2.6.35 non-sense and developed further...
fhasovic said:
Can you confirm that everything works?
My kernel is forked from ACA 2.6.35 non-sense and developed further...
Click to expand...
Click to collapse
There are voice recorder issues in the current version, but these are fixed in the latest release 0103
Swyped from my aHD2oid
SPEAKING FOR THE NOOBS
I LOVE MY PHONE TO ME IT'S SUPER FAST,
BUT WHAT YOU GUYS ARE SAYING IS THAT IF WE HAD A NATIVE ANDROID WITH THE SAME SPECS AS OUR HD2 IT WOULD BE EVEN FASTER AND BETTER?
THIS IS MY FIRST SMART PHONE SO I'M JUST CURIOUS ABOUT THAT.
Sent from my HD2 using xda premium
turn your caps lock off.
Short answer: You can expect the same performance from the HD2 as from the Nexus One and Desire, given it runs the same software.
Slightly longer answer: This is a development thread. What we are talking about is getting the linux kernel (think: core of the system) version used by ICS running on our phone.
Getting this to work -as opposed to trying to get a Froyo/GB kernel working with ICS- would benefit both performance and stability.
There are some issues with that. Most of them are related to the fact that Google and HTC have dropped official support for the Nexus One and Desire. The only issue related to the HD2 not being an native Android phone is the need for a special radio driver.
EDIT: If you tried to be funny: You failed.
uzi2 said:
There are voice recorder issues in the current version, but these are fixed in the latest release 0103
Swyped from my aHD2oid
Click to expand...
Click to collapse
Uzi is correct. Ill be doing a fs thing soon, a Bluetooth thing soon, and more cam related stuff. But overall 99% is fully working.
Sent from my HTC HD2 using XDA App
Now that bytecode has given us CWM, the question becomes what current ROM or branch of cyanogenmod would give us the biggest jump on getting a working image for our phones?
I think CM9 would be a good candidate. It would probably be the most popular choice to get working. I know I'm itching for ICS on our phone. I think it would make it the best phone on the market for now.
I'm no programmer so I can only say my opinion... I think whatever can be done first to get away from LG / AT&T software. If that's CM7, cool, if it's CM9, cool.
I'd think CM7 is the way to go to jump into a custom ROM that you could get working the fastest for this phone. CM7 has been out for a LONG time and it's been well tweaked and it's SOLID!
BUT... CM9 is what's coming (and for more and more phones/tabs, it's what's already here!).
I say even if it takes more effort now (vs. an older and more stable CM7), go right with CM9 and let's get this party started!
Namuna said:
I say even if it takes more effort now (vs. an older and more stable CM7), go right with CM9 and let's get this party started!
Click to expand...
Click to collapse
Why not ICS? Maybe it would be easy to port ICS for Galaxy Nexus to Nitro?
I'd say CM7 since by the time you finish getting CM9 figured out and working with the current kernal, it will have already been released by LG.
I'm going to start working on my own custom rom, at least as long as i can figure out how to get my Ubuntu setup... ugg..
Does anyone know any good guides for making your own AOSP ROMS? Or is it all, learned knowledge and experience?
I sure wish I knew what you guys are talking about. Lol
Sent from my LG-P930 using XDA App
I think CM9 is the way to go.
Question to any who know, if we were to put CM9 or any ICS on the Nitro, will it use hardware acceleration? That is the main feature I have been waiting for.
red_solo_cup said:
I think CM9 is the way to go.
Question to any who know, if we were to put CM9 or any ICS on the Nitro, will it use hardware acceleration? That is the main feature I have been waiting for.
Click to expand...
Click to collapse
Hardware acceleration for the UI? Or 3D / 2D graphics?
I think I read the other day that you can force the UI to use hardware acceleration in the build.prop file. With CWM feel free to tinker.
red_solo_cup said:
I think CM9 is the way to go.
Question to any who know, if we were to put CM9 or any ICS on the Nitro, will it use hardware acceleration? That is the main feature I have been waiting for.
Click to expand...
Click to collapse
unless someone on the CM team gets a nitro, there will be no ICS/CM9 until LG puts out a rom.
We need to just start w/ the basics. Optimize (battery life and performance) what we currently have and go from there.
I find your lack of faith...disturbing
Seriously, I think we can get a pretty damn good rom going if we put some work into it.
AllstarE4 said:
I sure wish I knew what you guys are talking about. Lol
Sent from my LG-P930 using XDA App
Click to expand...
Click to collapse
Allstar, here u go
CyanogenMod: http://www.cyanogenmod.com/
CM7 is based on Gingerbread and is a very optimized and customizable ROM (OS) for the phone, which is generally close to perfect and has a myriad of features that you don't normally find in regular/stock ROMs.
CM9 is the same basic concept as CM7, except it is based on ICS (Ice Cream Sandwich, Android's latest OS).
ICS: http://www.android.com/about/ice-cream-sandwich/
---------- Post added at 08:27 AM ---------- Previous post was at 08:24 AM ----------
In my opinion, our phone is built to support ICS or any iteration of it (clear stated by the 3 softkeys instead of 4 in gingerbread). So I suggest we proceed with either ICS or CM9.
I don't know if it matters, but I think someone should reach out to CM team and inform them of the development. I am sure they will be very interested in developing CM9 for a phone that is clearly intended to be a rival to the Galaxy Nexus.
The main thing to consider is the kernel , ICS runs on 3.xx kernel and gingerbread on 2.6 , porting 2.6 kernel drivers to 3.xx is alot of work. codeaurora kernel is a good start
https://www.codeaurora.org/contribute/projects/qkernel
bytecode64 said:
The main thing to consider is the kernel , ICS runs on 3.xx kernel and gingerbread on 2.6 , porting 2.6 kernel drivers to 3.xx is alot of work. codeaurora kernel is a good start
https://www.codeaurora.org/contribute/projects/qkernel
Click to expand...
Click to collapse
+1
Sent from my LG-P930 using XDA App
I think a simple debloated init.d optimized stock ROM should be the starting point. Starting a long project like CM9 will only slow down the adoption rate of the phone because the progress will be slow. I think most people will want to see a couple of ROM choices before they jump aboard.
Once you use CM9 it will automatically start hardware acceleration for smoothness and my upcoming base ROM that i'm including is going to have hardware acceleration and much more, please stay tune.
mod ammend
moved to general - this isn't a development query
I'm going to bet it will be cm9!
isnt riccardo from G+ already getting close to a ics release for the nitro...
Sent from my LG-P930 using XDA
People what is the best and most stable ROM available for the kindle fire? I want a stable, reliable version for daily use.
Thanks in advance
ZiGi888 said:
People what is the best and most stable ROM available for the kindle fire? I want a stable, reliable version for daily use.
Thank in advance
Click to expand...
Click to collapse
That depends on what you're looking for. If you want to keep the amazon UI then Modaco is the way to go. If you're wanting more of an AOSP feel then GedeROM, AOKP, or Jandycane are all great options. GedeROM is based off of CM9, I believe. AOKP is it's own animal but it is based off of Ice Cream Sandwich source I believe and Jandycane is based off of Jellybean Source. Modaco probably has the best battery life out of them all but obviously is pretty much standard amazon. It is deodexed and optimized and rooted though without the worry of OTA updates messing with your root permissions. The biggest selling point for AOKP in my opinion is customization options. You can even change the soft keys to something you are more comfortable with on AOKP. GedeROM is a very stable CM9 and also has plenty of customization options that come with cyanogenmod. Jandycane as I said is based on Jellybean and it is buttery as they come. Right now my kindle is running Jandycane 1.4, soon to be 1.5 when I find time to update it. Battery life is a little sorry though but I am sure in time it will improve as lithium comes up with more fixes.
Yeah, it does depend on what you want. These are the ROMs I recommend. I have used them all and they all will work fine.
Jellybean/CM10 ROMs: CM10 SGT7 ROM, Hashcode's ROM(AOSP)
ICS/CM9 ROMs: gedeROM, SGT7 ROM, Glazed ICS
CM7: Barebones CM7
I was using GedeROM ... very stable IceCreamSandwich rom.
I just switched to JandyCane .. its Jellybean. I'm not sure what's missing .. it's also nice.
Stick with gedeRom if this is your first forray into Kindle Android. Then when you're comfortable .. look around.
I am very excited seeing kernel 3.0 being developed for Desire. But we will have vsync on our device? Butter is just a new thing that is very eye-catchy and it could bring a lot of speed on JB for Desire
AFAIK the hardware is incapable of vsync, but i've seen some sort of software vsync related change in some ICS or JB (dont remember) commits. Guess it will end like hw accel for ICS, as software based hack... I trust on our fellow devs
if you look at the htc hd2 forums, you will see everyone is using tytung kernel for their 4.0 to 4.2 roms as it has fully working 720p video playback, camera and camcorder... why hasn't any htc desire developer ported his work? i would think it would be easier/more flexible than trying to port kernel 3.0 which seems to never be able to work completely because of lack of drivers.
yeahman45 said:
if you look at the htc hd2 forums, you will see everyone is using tytung kernel for their 4.0 to 4.2 roms as it has fully working 720p video playback, camera and camcorder... why hasn't any htc desire developer ported his work? i would think it would be easier/more flexible than trying to port kernel 3.0 which seems to never be able to work completely because of lack of drivers.
Click to expand...
Click to collapse
Maybe because making a working kernel 3 is more important than porting other kernels.
abaaaabbbb63 said:
Maybe because making a working kernel 3 is more important than porting other kernels.
Click to expand...
Click to collapse
i think we will never get a complete kernel 3 tbh as there's a lot of drivers not available..
It looks much easier/feasible to port Tytung work on camera/camcorder as the 2.x kernel of HD2 is very similar to htc desire's 2.x kernel..that would require less effort than trying to port kernel 3 (maybe i am wrong..) Qiss Rom and Tytung have been able to get a fully working camera/camcorder using kernel 2.x; so it is possible
at least i have been able to enjoy ICS with qiss rom and marius.a's rom (not a single issue to the major components : camera/camcorder is perfect, BT, GPS, Fm radio and all apps and games work great) ... i would have liked enjoying Jelly bean on my desire before changing phone next year
yeahman45 said:
i think we will never get a complete kernel 3 tbh as there's a lot of drivers not available..
It looks much easier/feasible to port Tytung work on camera/camcorder as the 2.x kernel of HD2 is very similar to htc desire's 2.x kernel..that would require less effort than trying to port kernel 3 (maybe i am wrong..) Qiss Rom and Tytung have been able to get a fully working camera/camcorder using kernel 2.x; so it is possible
at least i have been able to enjoy ICS with qiss rom and marius.a's rom (not a single issue to the major components : camera/camcorder is perfect, BT, GPS, Fm radio and all apps and games work great) ... i would have liked enjoying Jelly bean on my desire before changing phone next year
Click to expand...
Click to collapse
man you are wrong i have a second device (samsung galaxy5) and it has kernel 3.x working great and it's the only ARMV6 phone whit this kernel and if someone could do it for such a low end phone we can do it so don't worry we'll have our 3.x kernel
tcris45 said:
man you are wrong i have a second device (samsung galaxy5) and it has kernel 3.x working great and it's the only ARMV6 phone whit this kernel and if someone could do it for such a low end phone we can do it so don't worry we'll have our 3.x kernel
Click to expand...
Click to collapse
i mean apart from qiss rom team(unfortunately they did not released their source), no one has been able to get a stable camera/camcorder on kernel 2.x ... what makes you think with kernel 3.x it will be better and easier? and unfortunately there are very very few developers on the htc desire nowadays.. look at the android dev forum, it is quite quiet nowadays
yeahman45 said:
i mean apart from qiss rom team(unfortunately they did not released their source), no one has been able to get a stable camera/camcorder on kernel 2.x ... what makes you think with kernel 3.x it will be better and easier? and unfortunately there are very very few developers on the htc desire nowadays.. look at the android dev forum, it is quite quiet nowadays
Click to expand...
Click to collapse
out there are millions of devices whitout any dev and you complain about your phone whit A LOT of devs? if you want camera to full work stay on sense roms
yeahman45 said:
i mean apart from qiss rom team(unfortunately they did not released their source), no one has been able to get a stable camera/camcorder on kernel 2.x ... what makes you think with kernel 3.x it will be better and easier? and unfortunately there are very very few developers on the htc desire nowadays.. look at the android dev forum, it is quite quiet nowadays
Click to expand...
Click to collapse
Trust me, nikez and team are working on it. The xda thread may seem dead because of a bit of a mishap with a forum mod, but hes is still actively working and reporting his progress on his forum and on twitter.
chromium96 said:
Trust me, nikez and team are working on it. The xda thread may seem dead because of a bit of a mishap with a forum mod, but hes is still actively working and reporting his progress on his forum and on twitter.
Click to expand...
Click to collapse
i know he's developing and i am glad he is still continuing working on k3 ... i am just raising the question why no one ported tytung's work instead of working from "scratch"
tcris45 said:
out there are millions of devices whitout any dev and you complain about your phone whit A LOT of devs? if you want camera to full work stay on sense roms
Click to expand...
Click to collapse
i already have a working camera/camcorder with qiss/marius.a rom and it's ICS!
yeahman45 said:
i know he's developing and i am glad he is still continuing working on k3 ... i am just raising the question why no one ported tytung's work instead of working from "scratch"
Click to expand...
Click to collapse
It wouldn't work, because nikez uses ion now in his rom instead of pmem! With pmem and tytung's camera fix, which tytung already dicussed with nikez in nikez's thread, it would work, but nikez dont want to revert back to pmem
Sent from my HTC Desire using xda app-developers app