[ROM][GEN2]CyanogenMod 7 (Android 2.3.7)

[Guest 4pda]

I have new problem. Home button not working on "update-cm-7.2.0-RC0-Blade-KANG-n257plus-signed.zip". Before all work (on CM).

OSF with OLED.

[Guest Simono]

New patch for people building ROM blade: add support for new camera sensors

[Guest targetbsp]

New patch for people building ROM blade: add support for new camera sensors

Yeah the weird light sensor behaviour has been reverted. :)

[Guest sej7278]

New patch for people building ROM blade: add support for new camera sensors

built n257+2 with patch4, get it from the usual place

to test it out i restored proximity to default, rebooted and calibrated and it generated "7843,6274,238,238,246,0,0,16,32", much higher than i used to get, but they do seem to work.

autobrightness seems to work again - and light sensor actually returns values like 96/200 now not -1/1 like it did with patch3. i didn't bother with changing the levels.

other than the kernel its pretty much the same as n257+ with the patches listed in my sig below.

[Guest targetbsp]

Cool, will try that. Mine is still repo syncing and iI have an OCD need to reply to the review thingy ASAP lol.

Incidentally, why do yours compile as 7.2.0 and mine 7.1.0? Or are you just renaming them?

Edited November 22, 2011 by targetbsp

[Guest sej7278]

Incidentally, why do yours compile as 7.2.0 and mine 7.1.0? Or are you just renaming them?

that's the version they are now, once you've synced check vendor/cyanogen/products/common_versions.mk

$ git log common_versions.mk

commit 9eb9d741524dd596798821dedad43ee68207b955

Author: Steve Kondik <shadeBLAHchemlabBLAH>

Date:   Sun Nov 13 21:29:10 2011 -0800


	cm: Bump to 7.2.0-RC0


	Change-Id: I19d56107f7684401ea4e91ffda181d762bf8112a

Edited November 22, 2011 by sej7278

[Guest ritterkeks]

Regarding issue "Homekey stopped working"

Mine stopped working too, but not in relation to a flash, so a hardware fault makes more sense.

Since this happened to several people already, use sej's KANG with soft button patch or app ButtonSaviour

HOWEVER I have a much severe problem now: My blade suddenly doesn't accept calls anymore!

This is really bad, will flash now n257p2 now and hope this will change anything, otherwise I will try and restore a backup...

Very weird thing, texts work, but notifaction tones for anyting don't.

ALSO voice input seems to stop now, as I cant use voice actions anymore "No sound input"...

Damn it

[Guest targetbsp]

hi, I recently flashed the latest sej build (n257+) but my proximity sensor is broken. I did 'restore proximity defaults' then reboot and when I'm going to calibrate the proximity sensor it gives an error like 'something went wrong.Aborting'... I read something about manually change the proximity values but I dont really know for what values I need to change... someone help me pls

Do the normal nightlies work for you? How about Sej's new plus 2 rom?

[Guest Amphoras]

I have new problem. Home button not working on "update-cm-7.2.0-RC0-Blade-KANG-n257plus-signed.zip". Before all work (on CM).

OSF with OLED.

Try reflashing, that usually fixes it for me.

[Guest Picazsoo]

built n257+2 with patch4, get it from the usual place

to test it out i restored proximity to default, rebooted and calibrated and it generated "7843,6274,238,238,246,0,0,16,32", much higher than i used to get, but they do seem to work.

autobrightness seems to work again - and light sensor actually returns values like 96/200 now not -1/1 like it did with patch3. i didn't bother with changing the levels.

other than the kernel its pretty much the same as n257+ with the patches listed in my sig below.

I put it to Multiupload if that's ok - to lessen the load on your server. Here is the link: http://www.multiupload.com/EDYUIZP7YM

[Guest sej7278]

I put it to Multiupload if that's ok - to lessen the load on your server. Here is the link: http://www.multiupload.com/EDYUIZP7YM

cheers.

[Guest targetbsp]

that's the version they are now, once you've synced check vendor/cyanogen/products/common_versions.mk

$ git log common_versions.mk

commit 9eb9d741524dd596798821dedad43ee68207b955

Author: Steve Kondik <shadeBLAHchemlabBLAH>

Date:   Sun Nov 13 21:29:10 2011 -0800


	cm: Bump to 7.2.0-RC0


	Change-Id: I19d56107f7684401ea4e91ffda181d762bf8112a

Mine says that when I do a git log but still compiles it as 7.1.0. :S

Package complete: /home/target/android/system/out/target/product/blade/update-cm-7.1.0-Blade-KANG-signed.zip

[Guest sej7278]

Mine says that when I do a git log but still compiles it as 7.1.0. :S

Package complete: /home/target/android/system/out/target/product/blade/update-cm-7.1.0-Blade-KANG-signed.zip

you making sure to delete out/ and maybe ~/.ccache ?

[Guest targetbsp]

I haven't deleted either of those. I'll give them a go in turn. Thanks!

[Guest ritterkeks]

whoah... now i flashed it after formatting /system.

had to install gapps then, but

now my home key is working again, so you were right with it being a software issue

AND I can now finally be called again, as the blade know rings again instead of simply remaining uninvolved, while the other side hears it ringing

:-) blade, I can now keep you a few more months

[Guest sej7278]

looks like the "8 cores, 24gb ram" specs are over-the-top for building ics.

i've just built aosp master full-eng for the emulator on debian7 in 45mins. same box builds cm7 in 30mins.

ram has barely been touched - maximum 48% of my 12gb has been used, most of the time it was around 25% (its at 9% at boot).

repo is about 10gb download, ~/.ccache is 900mb, out/ comes to 14gb. linux and ccache is on an ssd but the repo itself is on a hard disk.

cpu usage has been mostly 100% on all of my 4 cores on the 3.2ghz core-i5/750 using make -j8 (4 threads over 4 cores).

looking at the times on the google group cpu speed is the key, not ram or disk. an 8gb core-i7 is plenty.

oh and JBQ confirmed that sun-jdk-6 is pretty much the only way to go, openjdk is just not going to cut it and java7 of any sort is not supported. that's going to be a problem moving forward as oracle end-of-life java6 and a lot of the linux distro's remove sun-jdk in favour of openjdk.....

[Guest KonstaT]

looks like the "8 cores, 24gb ram" specs are over-the-top for building ics.

i've just built aosp master full-eng for the emulator on debian7 in 45mins. same box builds cm7 in 30mins.

ram has barely been touched - maximum 48% of my 12gb has been used, most of the time it was around 25% (its at 9% at boot).

repo is about 10gb download, ~/.ccache is 900mb, out/ comes to 14gb. linux and ccache is on an ssd but the repo itself is on a hard disk.

cpu usage has been mostly 100% on all of my 4 cores on the 3.2ghz core-i5/750 using make -j8 (4 threads over 4 cores).

looking at the times on the google group cpu speed is the key, not ram or disk. an 8gb core-i7 is plenty.

oh and JBQ confirmed that sun-jdk-6 is pretty much the only way to go, openjdk is just not going to cut it and java7 of any sort is not supported. that's going to be a problem moving forward as oracle end-of-life java6 and a lot of the linux distro's remove sun-jdk in favour of openjdk.....

Yeah, those initial build specs were highly exaggerated.

I built my first ICS for emulator (full-eng) in 1h22mins. My computer builds CM7 in about 40mins.

I have dual-core i5-2410M with 6gb of RAM on my laptop. Building ICS is certainly more CPU bound than other things. SSD certainly can speed things up a bit. Memory usage peaked couple of times to ~4,5gb.

I've been compiling it with 'make -j4', but apparently I could use -j8? I have two cores with four threads. That could speed it up, if my hard disk doesn't become the bottle neck?

[Guest sej7278]

Yeah, those initial build specs were highly exaggerated.

I built my first ICS for emulator (full-eng) in 1h22mins. My computer builds CM7 in about 40mins.

I have dual-core i5-2410M with 6gb of RAM on my laptop. Building ICS is certainly more CPU bound than other things. SSD certainly can speed things up a bit. Memory usage peaked couple of times to ~4,5gb.

I've been compiling it with 'make -j4', but apparently I could use -j8? I have two cores with four threads. That could speed it up, if my hard disk doesn't become the bottle neck?

laptop users seem to be suffering - even those with 7200rpm drives instead of the regular 5400rpm are slow. it could be turboboost/speedstep and less cache on the cpu too i guess.

the make -j value should be between 1 and 2 times the number of threads, so a core-i5 should be 4-8, but i usually find the higher the better, especially if you see cpu usage drop below 100% across all four cores.

my 8gb 2.66ghz core2quad with ssd builds cm7 about 10mins slower than my core-i5, so it is largely down to cpu.

[Guest targetbsp]

you making sure to delete out/ and maybe ~/.ccache ?

Deleting out sorted it. Thanks again!

[Guest hugobosslives]

Yeah, those initial build specs were highly exaggerated.

I built my first ICS for emulator (full-eng) in 1h22mins. My computer builds CM7 in about 40mins.

I have dual-core i5-2410M with 6gb of RAM on my laptop. Building ICS is certainly more CPU bound than other things. SSD certainly can speed things up a bit. Memory usage peaked couple of times to ~4,5gb.

I've been compiling it with 'make -j4', but apparently I could use -j8? I have two cores with four threads. That could speed it up, if my hard disk doesn't become the bottle neck?

so now i'm confused.

i have the same i5 as you with 8GB of ram. I like you have always used j4. (even on my dad's i7-2600k desktop) because, if im being honest, i have only a vague idea what that is. could some one give a brief explanation on what it is and which to use for given cores/ number of cpus please? :)

Generally I agree with sej above that the whole compiling thing is more cpu related than ram. turning up the clock speed on a i7-2600k makes a hell of a difference, whilst letting it have more ram has made no noticeable difference. hmm confused, as everyone always goes on about 24GB system monsters for compiling.

[Guest targetbsp]

hmm confused, as everyone always goes on about 24GB system monsters for compiling.

To compile GB uses a max of 1.7GB (inc what the OS needs) under Ubuntu v11 64bit for me. So I have no idea where the 24gb came from. Are people keeping their source caode tree in ram??? :D

[Guest sej7278]

so now i'm confused.

i have the same i5 as you with 8GB of ram. I like you have always used j4. (even on my dad's i7-2600k desktop) because, if im being honest, i have only a vague idea what that is. could some one give a brief explanation on what it is and which to use for given cores/ number of cpus please? :)

Generally I agree with sej above that the whole compiling thing is more cpu related than ram. turning up the clock speed on a i7-2600k makes a hell of a difference, whilst letting it have more ram has made no noticeable difference. hmm confused, as everyone always goes on about 24GB system monsters for compiling.

the -j flag refers to the number of jobs (you can use --jobs=4) that make spawns in parallel, so -j4 fires off 4 threads, and starts a new one when one of those is finished. i believe that with higher values (not too high) the kernel scheduler can prioritise the load better, so you're not waiting a long time for one big job to finish (e.g. compiling the kernel) when you could have executed 3 others in that time (e.g. odex'ing some apk's)

your dad's core-i7 is 4 cores with 8 threads (hyperthreading) so anything less than -j8 is doing it an injustice, you'd probably not see the 4 cores under more than 50% load, you could go up to -j16

jbq uses -j32 on his dual-socket quad-core with hyperthreading, and builds in 20mins or something crazy.

you certainly can't assume that by doubling the jobs you're halving the build time (like hyperthreading or multiple cores don't actually double your pc speed) but it will be faster the larger the number, subject to the law of diminishing returns.

holy moly that was not a brief explanation! ok think of it as typing with one hand - its faster with two hands (but not twice as fast) and as you have two hands, you might as well use them both. better?

[Guest hugobosslives]

To compile GB uses a max of 1.7GB (inc what the OS needs) under Ubuntu v11 64bit for me. So I have no idea where the 24gb came from. Are people keeping their source caode tree in ram??? :D

yer i guess they are either doing that or outputting the compile straight to ram. no idea how to try that.....

the -j flag refers to the number of jobs (you can use --jobs=4) that make spawns in parallel, so -j4 fires off 4 threads, and starts a new one when one of those is finished. i believe that with higher values (not too high) the kernel scheduler can prioritise the load better, so you're not waiting a long time for one big job to finish (e.g. compiling the kernel) when you could have executed 3 others in that time (e.g. odex'ing some apk's)

your dad's core-i7 is 4 cores with 8 threads (hyperthreading) so anything less than -j8 is doing it an injustice, you'd probably not see the 4 cores under more than 50% load, you could go up to -j16

jbq uses -j32 on his dual-socket quad-core with hyperthreading, and builds in 20mins or something crazy.

you certainly can't assume that by doubling the jobs you're halving the build time (like hyperthreading or multiple cores don't actually double your pc speed) but it will be faster the larger the number, subject to the law of diminishing returns.

holy moly that was not a brief explanation! ok think of it as typing with one hand - its faster with two hands (but not twice as fast) and as you have two hands, you might as well use them both. better?

cheers man. maybe that's why the 2600k was not as fast as i was expecting it to be when i first did it on there?

I will try j8 and see what kind of difference it makes next time I'm at home (crimbo).

so basically you pick by how many threads your system can simultaneously handle? So in dummy terms, on the new (single cpu) intel architecture, its double the cores? makes sense I guess. But you say above the i7-2600k could handle j16?

Makes alot more sense now, thanks. And yer I get the thing about not doubling the speed. Anyone that's into gaming will know this with SLI graphic card combo's.

[Guest sej7278]

cheers man. maybe that's why the 2600k was not as fast as i was expecting it to be when i first did it on there?

I will try j8 and see what kind of difference it makes next time I'm at home (crimbo).

so basically you pick by how many threads your system can simultaneously handle? So in dummy terms, on the new (single cpu) intel architecture, its double the cores? makes sense I guess. But you say above the i7-2600k could handle j16?

Makes alot more sense now, thanks. And yer I get the thing about not doubling the speed. Anyone that's into gaming will know this with SLI graphic card combo's.

the i7 is quad core but has hyperthreading too so that means double the threads of the i5, so you go go up to double that with make, so 4 cores x2 threads x2 = 16 jobs.

i've just noticed that "brunch blade" is calling make -j4, like "mka" does and takes 28mins to build cm7.

i tried "lunch cyanogen_blade && make -j8" and that took 22mins, but it doesn't make a full flashable zip, just the img files, brunch doesn't take "-j<whatever>" as an argument anymore, so i've tried editing the mka command in envsetup.sh, which is generating "make -j8 bacon" in the process list. oddly enough its made not a minute's difference it still takes 28mins, i guess as mka already uses schedtool

its looking like your average 8gb quad core can handle ICS just fine, no need to buy shares in newegg or dixons :D

i just fired my aosp master image up in the emulator, ICS is pretty sweet - settings are much cleaner and more organised than cm7, ringlock is the default lockscreen, also better than the cm7 one.

Edited November 22, 2011 by sej7278

[Guest samjam]

I finally worked out that ICS is Ice Cream Sandwich and not Internet Connection Sharing (aka tethering) - I was wondering what the big deal with ICS was.