• After 15+ years, we've made a big change: Android Forums is now Early Bird Club. Learn more here.

Root [ROM] MTDEV-CM7 build 2013-03-03

Hi so with the new kernel, by Bsydz has anyone had issues with file corruption?
I know it was a test version no problems there and I think it's great so far.

However I think my data directory got corrupted, currently trying to fix it. This morning Swype dictionary didn't work so I rebooted but it kept force closing on me, I rebooted and did fix permissions and it didn't really help so I updated and was ok. Next I had issues with Whatsapp which force closes as soon as I open it.

Based on alogcat my folder in
/data/data/com.whatsapp is corrupted because it remains even after I uninstall it, reinstall it etc. Actually alogcat has several errors related to /data/data/com.whatsapp/database/msgstore.db

When I go in the folder now and do a LS I get the following error:
ls: ./databases: I/O Error

If I try to delete the folder using rm -f or rm -rf I get this error:
rm: can't stat 'com.whatsapp/databases' : i/o error
rm: can't remove 'com.whatsapp' directory not empty

If we ignore this as a problem as a program but a problem as a folder issue/permission issue then maybe it will make sense.

I searched on google, they said boot to recovery, connect usb to PC run adb shell. Unmount /data unmount /system
then run e2fsck -fv /dev/block/st110 and run e2fsck -fv /dev/block/st10

I did a df -h and I think it should be /dev/block/mmcblk0p7 and 0p8

Can anyone verify my sanity and logic in wanting to run this?
 
(context)
located in /data/data/ still suspect all my errors are due to file system corruption and or permission issues.

Thank you for the suggestion, I have tried -rfv

rm -rfv deleteMe
rm: can't stat 'deleteMe/databases' : I/O error
rm: can't remove 'deleteMe' : Directory not empty


I can rename but not delete. I see app_64 and a lot of other app numbers so I assume the phone knows what the folder is for.
 
awdark said:
Hi so with the new kernel, by Bsydz has anyone had issues with file corruption?
I know it was a test version no problems there and I think it's great so far.

However I think my data directory got corrupted, currently trying to fix it. This morning Swype dictionary didn't work so I rebooted but it kept force closing on me, I rebooted and did fix permissions and it didn't really help so I updated and was ok. Next I had issues with Whatsapp which force closes as soon as I open it.

Based on alogcat my folder in
/data/data/com.whatsapp is corrupted because it remains even after I uninstall it, reinstall it etc. Actually alogcat has several errors related to /data/data/com.whatsapp/database/msgstore.db

When I go in the folder now and do a LS I get the following error:
ls: ./databases: I/O Error

If I try to delete the folder using rm -f or rm -rf I get this error:
rm: can't stat 'com.whatsapp/databases' : i/o error
rm: can't remove 'com.whatsapp' directory not empty

If we ignore this as a problem as a program but a problem as a folder issue/permission issue then maybe it will make sense.

I searched on google, they said boot to recovery, connect usb to PC run adb shell. Unmount /data unmount /system
then run e2fsck -fv /dev/block/st110 and run e2fsck -fv /dev/block/st10

I did a df -h and I think it should be /dev/block/mmcblk0p7 and 0p8

Can anyone verify my sanity and logic in wanting to run this?
Click to expand...
awdark said:
(context)
located in /data/data/ still suspect all my errors are due to file system corruption and or permission issues.

Thank you for the suggestion, I have tried -rfv

rm -rfv deleteMe
rm: can't stat 'deleteMe/databases' : I/O error
rm: can't remove 'deleteMe' : Directory not empty


I can rename but not delete. I see app_64 and a lot of other app numbers so I assume the phone knows what the folder is for.
Click to expand...
Well, I can't say that I know what the issue may be but I can suggest a few things.
First, which kernel are you using?
Which build are you using?

I know I had some issues with CWM version 6 with it corrupting files in my SD card, but that was an early beta build, I suggest CWM 5.0.2.7, it has always worked well for me, for any ROM.

Also, /dev/block/mmcblk0p7 is /system and /dev/block/mmcblk0p8 is data.
 
Thanks for the reply -
CyanogenMod-7 20120926 so September 26 build of CM7.
According to about phone - Kernel version 2.6.32.9-perf g60madman@mtdev #1 sounds like the original one but I did a manual change of the kernel using download mode so I could back up the stock cm7 kernel.

Running the test kernel which you posted. As far as I know your kernel has fixed a lot of the little issues such as the camera crashing when switching to the front and rear camera. It also has the benefit of providing me with a possibly placebo effect for the responsiveness of my screen and touch buttons.

So do you think running e2fsck on the data partition is somewhat "safe"? Current guess is corruption because I/O Errors are usually disk errors and we only have this one flash.
 
awdark said:
Thanks for the reply -
CyanogenMod-7 20120926 so September 26 build of CM7.
According to about phone - Kernel version 2.6.32.9-perf g60madman@mtdev #1 sounds like the original one but I did a manual change of the kernel using download mode so I could back up the stock cm7 kernel.

Running the test kernel which you posted. As far as I know your kernel has fixed a lot of the little issues such as the camera crashing when switching to the front and rear camera. It also has the benefit of providing me with a possibly placebo effect for the responsiveness of my screen and touch buttons.

So do you think running e2fsck on the data partition is somewhat "safe"? Current guess is corruption because I/O Errors are usually disk errors and we only have this one flash.
Click to expand...

You may be right about the kernel, to an extent, If you look at the change log below, the init script was reworked. So it may be a conflict between the two versions as I built the flashable zip with the 1015 build. The actual kernel (zImage) shouldn't be the problem though. If you are comfortable extracting the boot.img and swapping the new zImage into the boot.img from that build it could fix the issue, if it is the issue in the first place. Let me know, if you are comfortable with that or not, if not I can download that release and do it. Or you could check out the newest build 12/28 I believe.

I am not completely convinced it is the kernel but, what I stated above is the only thing I can think of, if it is. Have you put the SD card in a windows computer and scanned it? I really don't know enough about e2fsck to comment on that, so hopefully somebody with more experience can answer that one. The touch screen is not a placebo effect I changed the touch screen drivers. Good luck and let me know what's going on.

10/15/2012
∙ Updated to the Sharp 2.3.5 wlan firmware and init.qcomm etc scripts
∙ Removed system prop line for video suspected of causing wifi issues for some users
∙ Completely reworked the init.triumph.rc.
∙ Removed all the froyo mm modules
∙ Add Sharp 2.3.5 mm modules
∙ Replaced static CM7 wallpaper

09/26/2012
∙ Flicker is back
∙ Camera/camcorder are now working correctly
 
Unfortunately the \data\data is in the internal memory of the phone not on the SD. Unless there is a way I could mount the system\data partitions on Windows through recovery but I don't think so.

Actually the phone was running great for several days, I had updated the kernel manually on 12/27/12 with the one linked on the first page (opened zip manually replaced boot.img [flashing it using the zip right now for curiosity okay it shows a different kernel name now phenom@phenom-desktop #1] )

For now I will just scratch my head, no idea why it flashing it would change other things. I tend to like doing things manually to better understand whats going on. Will check it out and see if I had a half-updated kernel or what. I appreciate your replies though!
 
awdark said:
Unfortunately the \data\data is in the internal memory of the phone not on the SD. Unless there is a way I could mount the system\data partitions on Windows through recovery but I don't think so.

Actually the phone was running great for several days, I had updated the kernel manually on 12/27/12 with the one linked on the first page (opened zip manually replaced boot.img [flashing it using the zip right now for curiosity okay it shows a different kernel name now phenom@phenom-desktop #1] )

For now I will just scratch my head, no idea why it flashing it would change other things. I tend to like doing things manually to better understand whats going on. Will check it out and see if I had a half-updated kernel or what. I appreciate your replies though!
Click to expand...
Not a problem, Any time you see phenom@phenom-desktop in the kernel info that is me, that's my computers name. I was wondering about it showing another name also, that is why I asked which version and such. I got confused with which data you were having issues with, I swore you said something about the SD, Been reading too much lately. Let me know how it goes, if you get it resolved or what ever it may help somebody out in the future.

Also, what command did you use? This is what the updater script uses.

package_extract_file("boot.img", "/dev/block/mmcblk0p5");
 
Unfortunately I just opted to replace the boot.rom file directly from download mode, similar to the way we install recovery.
 
awdark said:
Unfortunately the \data\data is in the internal memory of the phone not on the SD. Unless there is a way I could mount the system\data partitions on Windows through recovery but I don't think so.

Actually the phone was running great for several days, I had updated the kernel manually on 12/27/12 with the one linked on the first page (opened zip manually replaced boot.img [flashing it using the zip right now for curiosity okay it shows a different kernel name now phenom@phenom-desktop #1] )

For now I will just scratch my head, no idea why it flashing it would change other things. I tend to like doing things manually to better understand whats going on. Will check it out and see if I had a half-updated kernel or what. I appreciate your replies though!
Click to expand...

Just for the means of disscution. I wonder if you could mount the other partitions via Cigwin. It might allow it.
 
Strangely it seems like installing the old version, uninstall, rename folder, restart a few times, then put in the newest one seemed to have gotten the app working again. Maybe it decided to finally ignore not sure. But ill be monitoring this.

Touchscreen and button behavior feels the same, working more often than not but still using button savior so I won't bash my phone's screen in.

@Eduardo121842
Have you tried installing a different keyboard to see if that is also unresponsive? Mine does have a tendency to skip randomly as I swype producing really random words.
 
MikeRL said:
I will upload a screenshot (with the streets blurred out for privacy reasons) that shows what I mean.
Click to expand...

From the screenshot, it looks like there is no GPS lock (the GPS icon circle is not white inside), which would indicate it is using network location services instead of GPS location services, which would explain the significantly reduced accuracy.

Can you get a GPS lock in the app "GPS Status"?

And/or, if you turn off Network Location services (deselect "Use wireless networks" in Location & Security settings), does maps find any location?
 
dsmryder said:
One of the other things I wanted to know about is video performance. I convert my DVDs into m4p files and I get a slight skip while playing them on stock and CM7. On CM9 it runs smooth. If possible I would like to get that smooth on CM7.
Click to expand...

I had always had problems trying to play back downloaded torrents using both MX & Mobo going all the way back to TG's builds. They would either skip, freeze or go out of sync. Probably around the time Reloaded came out I tried VitalPlayer and have not had any issues since.

One of things noticed why playing with the code is if the overlay renderer code is removed in stagefrighthw or if I build using msm7k instead of msm7x30 which does not use the overlay encoder in stagefrighthw video playback is vastly improved but the Camera will break. In short removing the overlay renderer code in stagefrighthw when building msm7x30.

https://github.com/MTDEV-CM7/hardware/tree/cm7/qcom/media/libstagefrighthw

so it looks like the code used when building msm7k

https://github.com/MTDEV-CM7/hardware/tree/cm7/msm7k/libstagefrighthw

seems to vastly improve video playback & fix flicker but breaks the Camera. I have been able to git the Camera up and running building as msm7k by stealing the overlay & gralloc libs from BSydz GB Backups posted in his Triumph experiment thread but it has the same issue with the Camcorder that I have if removing the Overlay renderer code from stagefrighthw when building as msm7x30.

Another thing that I have had a hard time with is that if you look in,

https://github.com/MTDEV-CM7/hardware/tree/cm7/qcom/media/mm-core/omxcore/src

will notice the qc registry tables for 7630
https://github.com/MTDEV-CM7/hardware/tree/cm7/qcom/media/mm-core/omxcore/src/7630

lists pretty much the same omx libs as all but the Sharp 2.3.5 GB variants but if you look at the 7627 tables that are used when building as msm7k as every variant is msm7k

https://github.com/MTDEV-CM7/hardware/tree/cm7/qcom/media/mm-core/omxcore/src/7627

will not see the omx libs used in any of the msm7k variants. This is not just CM7, I have the Source code for all the different GB code Aurora branches and some other lge sources and its always the same in that the omx libs from our msm7k variants are all listed in the 7630 tables and not the tables used for msm7k.

Also looking in the msm7k stagefrighthw code

https://github.com/MTDEV-CM7/hardwa...ighthw/stagefright_surface_output_msm72xx.cpp

will see line number 37 lists “OMX.qcom.video.decoder.", 23” compared to what is used in msm7x30

https://github.com/MTDEV-CM7/hardwa...ighthw/stagefright_surface_output_msm7x30.cpp

will see line 35 lists to use "OMX.qcom.7x30.video.decoder.", 28”. So have tried swapping it so the MSM7x30 line uses “OMX.qcom.video.decoder.", 23” in place of "OMX.qcom.7x30.video.decoder.", 28” and it behaves exactly like I built it using the msm7k BoardConfig but again messes up the Camcorder.

In the previous builds that I used the omx libs from Sharp 2.3.5 which at least by name are the same as the other variants in additionn to the below 3 not found in the other 3 gb variantss like Cherry, Spice & WellcoM If counting the open core libs.

libOmxH264Dec.so
libOmxQcelpHwDec.so
libopencorehw.so

Was not able to use the Sharp 2.3.5 “libomxcore.so” without adding code for DRM Encoding to stagefrighthw I picked up from Code Aurora in addition to adding in

libmmparser.so
libmmparser_divxdrmlib.so
libdivxdrmdecrypt.so
libmm-omxcore.so
libqcomm_omx.so
libplayready.so
libgemini.so

not to mention the MM-Modules & other bin files like
sw-adec-omxaac-test
sw-adec-omxamr-test
sw-adec-omxamrwb-test
sw-adec-omxmp3-test

and because I was not able to successfully work in the drm divx code from Aurora ware I found the drm encoder code for stagefrighthw also needed to add

libOmxVdec.so
libOmxVenc.so

we previously built from Source because “libOmxVdec.so” from Sharp 2.3.5 already knows about the divx drm libs listed above.

http://androidforums.com/triumph-all-things-root/653173-media-library-dependency-list.html

Anyhow in the last build ditched anything media related used from Sharp 2.3.5 and instead used the Cherry files as they don't use any of the DRM stuff and could cut the media files used in half and also thought maybe would help a bit with the Camera as we use the Cherry Camera libs.


Otherwise just went on a rant but thought maybe some if this info could help out anyone else who may want to look at the Video for CM7. Also if you look at CM9 for comparisons will find the code looks very different then it does in CM7 so was no help at least in my case.

Edit: 01.05.2013

One last note, well probably not but anyway I noticed after removing all the MM Modules CM7 is now building 4 Modules on its own we previously added from Stock Froyo or more recently Sharp 2.3.5.

mm-vdec-omx-test
mm-venc-omx-test720p
mm-video-driver-test
mm-video-encdrv-test

in the past if I put in the full set of Modules from either Stock or Sharp 2.3.5 also needed to add "libgemini.so" from the Rom I took the Modules from. Just as an experiment I copied in just these 4 Modules only from Sharp 2.3.5 expecting I would not boot past the M logo like I would expect when using the full set of modules from either Stock or Sharp 2.3.5 using the Cherry "libgemini.so" but had booted up just fine so it looks like its one or more of the other Modules that are tied to "libgemini.so" and not the Modules listed above. I did another build adding back in the full set of Modules again without changing to the Sharp 2.3.5 "libgemini.so" and it went back to not booting past the M logo again. Dont know yet how this all ties together but why playing youtube in portrait mode see libstagfrighthw errors referencing venc-omx-test720p light the logcat up red like a Christmas tree. This has been the case building msm7x30 going back all the way to the first TG builds. Swapping out "OMX.qcom.7x30.video.decoder.", 28” with “OMX.qcom.video.decoder.", 23” clears up the errors but like mentioned above causes Camcorder issues.
 
I'm pretty happy with the latest 12-28 release... Bluetooth MOGA controller works. The battery doesn't seam to drain as bad. I had my wifi on all night last night when I was sleeping and it was still above 80%. I was using bluetooth speakers today and the wifi didn't function correctly.. But that's minor.

Thanks, I was getting ready to get a new phone.. I think this one should last me another year.
 
Chairshot215 said:
I had always had problems trying to play back downloaded torrents using both MX & Mobo going all the way back to TG's builds. They would either skip, freeze or go out of sync. Probably around the time Reloaded came out I tried VitalPlayer and have not had any issues since.
Click to expand...
Thanks for that tip, it solved my problem. I was a mx player diehard
 
Installed the 12/28 ROM 3 days ago.

Pros: GPS is superfast, much like the Sharp ROM! Initial lock took about 1 minute indoor (where I wasn't able to get a lock with any of the many CM7 ROMs I used before) but any subsequent locks take under 5 seconds. This is almost as fast as my Nexus 4! GPS Test showed 3 ft accuracy while driving at 50 mph! Absolutely amazing!

Cons: Phone randomly freezes and wouldn't take commands or takes 30+ seconds to process them. Had to pull the battery once as it was unresponsive for more than 3 minutes. Got 2 random reboots in 2 days, something I haven't experienced with the past MTDEV CM7 releases.

Keep up the fantastic job you guys are doing! I'll keep rooting for you even though I'm moving to a Nexus 4 on Straight Talk next week. I would have trashed this phone long time ago if it wasn't for many devs spending countless hours to make this phone shine.

THANK YOU!
 
First of all, thank you MT-DEV for all your awesome work!

I'm using the 12/28 rom and I don't understand why the font is now changed to open-sans. I prefer the roboto font because it seems cleaner and easier to read in smaller font sizes.
 
You gonna be kidding me.... home button quit working after flashing the rom....
flashed again, home button still not working.... contact, talk syncs fine, but no play/gmail..... it runs a lot smoother, but I need the phone is pretty much useless without play/gmail....
 
realdreams said:
You gonna be kidding me.... home button quit working after flashing the rom....
flashed again, home button still not working.... contact, talk syncs fine, but no play/gmail..... it runs a lot smoother, but I need the phone is pretty much useless without play/gmail....
Click to expand...

Did you flash gapps? If you did, did you sign in with your account?
 
You must log in or register to reply here.
Back
Top Bottom