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

Root TWRP - Spectrum

That's actually a lot of good info. I'm going to be free most of this weekend so I'll fire up the build box and update all my source code tomorrow. I may be able to edit something to get the button to show up all the time.
 
That's actually a lot of good info. I'm going to be free most of this weekend so I'll fire up the build box and update all my source code tomorrow. I may be able to edit something to get the button to show up all the time.

That's cool. If im around, I'll likely be on IRC (freenode) as yoinx, if you use it.

What I've been trying to do is recompile the iproj kernel by hand to change some of the video/display settings and recompile. The bad part is that I can't get any the kernels that I build by hand to boot with twrp. Though, I've noticed that they seem to be building larger than the original that built from within the cm repo.

I suspect that one of the settings for the the video are causing the issues. There are a couple in there to fix overlay issues and things like that. It could even be as simple as disabling the hardcoded resolution, since TWRP can be set for different resolutions and sets values based on that (which are hardcoded in the kernel)... maybe this is causing issues.
 
Silly question, but I'll ask anyways. Have you tried pulling a zImage out of an official CM10.1 build and inserting it into the ROM? Also, what about building from TDMs repo instead of the CM repo?
 
Silly question, but I'll ask anyways. Have you tried pulling a zImage out of an official CM10.1 build and inserting it into the ROM? Also, what about building from TDMs repo instead of the CM repo?

Fair enough question.

The kernel out of a 10.1 build should be the same one that the make was generating (if you don't have a pre-built kernel in the device tree and have the kernel source in the build directory). I imagine that what'll end up happening will be the same thing as before with the UI lag.

I imagine TDM's repo is pretty much the CM repo, since his moved into official support.

Though I could be wrong.


I think one of the issues may be the larger size for the 10.1 kernel... but I'm not positive. My recovery images have varied from 6-8.5 megs. But, the partition 13 is like 16 megs, so I dont see why that should be an issue.
 
TDM's kernel is the same in most respects, but not all. Case in point, TDM supports overclocking and stock does not. As for pulling a zImage, this would rule out a possible bad environment setup. That doesn't seem like it should be an issue, but I've also seen the difference between using Linaro and the Google GCC.
 
And 'BOOM' goes the dynamite.

http://d-h.st/uU5

TWRP 2.4.4.0 with working USB storage.

I've got a script in there to set the time like I have in my modded cwm as well, but it's not executing right now.


On a side not to Neph81:
It's odd that I got this compile with the 10.1 kernel (3.0.31-cyanogenmod+) and work correctly. The 2.5.0.0 gives me crazy UI lag. I'm thinking that I may try yanking a graphics.c file out of the 2.4.4.0 source and use it as a custom graphics file in 2.5.0.0

I take that back... I think the problem all along was that the iproj common board config was calling a custom graphics file which was screwing me up. I have 2.5.0.0 working but it's mounting the internal storage when I mount usb instead of the external. I should hopefully have that fixed soon then upload a 2.5.0.0 and upload to github.
 
So the custom graphic.c file from 2.4 fixed your issue in 2.5, but still kept the mount as USB button working?
 
So the custom graphic.c file from 2.4 fixed your issue in 2.5, but still kept the mount as USB button working?

No, I think the custom graphic file that's specified in the iproj common folder is what was causing me issues all along. The new kernel started working fine after I commented it out.

Both these builds present the mount as usb option. However, when you 'unmount usb' it doesn't correctly unmount. You still need to eject from the computer.

The same build tree works in both 2.4 and 2.5. well should anyway. I had to modify board config for the internal memory's mount line from /data to /data/data to get it stop mounting the internal memory as usb on 2.5, in 2.4 it was mounting correctly with just /data
 
Is this still at an "experimental" stage or safe for regular use now?

Thanks for putting in so much work on this, I personally prefer TWRP.
 
Is this still at an "experimental" stage or safe for regular use now?

Thanks for putting in so much work on this, I personally prefer TWRP.

Since I haven't heard anything back about either of them yet I'd call them experimental still.

If you're brave enough to test them... make a cwm backup an and copy it to your computer. Then flash this and test away

I prefer twrp to because it's open source for the touch...unlike cwm

Sent from my VS920 4G using Tapatalk 2
 
Yoinx:

I installed 2.5.0.0 that you posted, but I'm not getting ADB access even though I'm showing a ADB device in Windows device manager. I'm also not getting it to mount the USB drive. Which OS are you on that you were able to get these to work? I'm using Win7 X64.
 
Yoinx:

I installed 2.5.0.0 that you posted, but I'm not getting ADB access even though I'm showing a ADB device in Windows device manager. I'm also not getting it to mount the USB drive. Which OS are you on that you were able to get these to work? I'm using Win7 X64.

Interesting.

I'm running Ubuntu 13.04. Both ADB and mass storage worked flawlessly for me across multiple tests.

I'll try a couple times under windows 7x64 as that my other os and see if it manages to pick either up. Just gotta get some tweaks uploaded again for a different thread first.

I was going to work on getting some other things working right like the gpio for the power button which I suspect is supposed to turn the screen off, since everything turns it back on.


Actually... I wonder if windows needs the custom ums path to be set >.>
 
Very possible. Linux see android extremely easy while Windows likes to whine and complain about drivers and missing paths. Case in point, on CM9, you could mount the USB drive without turning on the mass storage option if you were under linux, but not under Windows.
 
Possible. I managed to get usb mounted in windows, but it wanted me to format the drive, I'm guessing that it was mounted a different path... perhaps my internal data partition. But I couldn't get ADB to work on either linux or windows. ... Very odd, since everything was working fine last night.

*edit*

Actually, looking in the android.mk for twrp it looks like the custom ums file is somewhat useless unless the device is actually mapping somewhere else for UMS.

This bit of code sets them to the same value if there's not a custom path set.

[high]
ifneq ($(TARGET_USE_CUSTOM_LUN_FILE_PATH),)
LOCAL_CFLAGS += -DCUSTOM_LUN_FILE=\"$(TARGET_USE_CUSTOM_LUN_FILE_PATH)\"
endif
ifneq ($(BOARD_UMS_LUNFILE),)
LOCAL_CFLAGS += -DCUSTOM_LUN_FILE=\"$(BOARD_UMS_LUNFILE)\"
endif
[/high]
 
Ok. So a little more investigation...

Under linux, everything still works as expected. Though, it's a little finnicky and sometimes I'm needing to reboot from recovery back into recovery to get adb to see the phone. Mass storage works as expected as well.

Under windows, adb works for me, though sometimes I have to reboot from recovery back into recovery to get it to work. Mass storage mounts, but prompts to format the drive. Obviously it's mounting a partition other than what I'm expecting. When I shell in and check to see what should be being mounted by
cat /sys/devices/platform/msm_hsusb/gadget/lun0/file
I get
/dev/block/mmcblk1p1

As I would expect.

Really, I'm starting to be at a loss as to what's going on.

I did end up having to reinstall my adb drivers for windows though... it was a little odd.
 
What happens if you set your recovery.fstab to the same one is in the normal CM file?

Yours:
/sdcard vfat /dev/block/mmcblk1p1

CM:
/sdcard vfat /dev/block/mmcblk1p1 /dev/block/mmcblk1
 
What happens if you set your recovery.fstab to the same one is in the normal CM file?

Yours:
/sdcardvfat/dev/block/mmcblk1p1

CM:
/sdcard vfat /dev/block/mmcblk1p1 /dev/block/mmcblk1

Either should be okay. At least in theory. But you know about theory.. :)
 
Either should be okay. At least in theory. But you know about theory.. :)
TDM:

Sadly yes. Theory says I should be able to find my notes for working on UMS the first time. Reality says otherwise and I'm really hoping I didn't toss them. I did find part of what I was looking for though.

Yoinx:
The old CWM kernel does in fact support UMS. The path is this:

/sys/devices/platform/usb_mass_storage/lun0/file

or

/sys/devices/platform/msm_hsusb/gadget/lun2/file

I'm still not having any luck getting that custom line to work when using your GIT info. I'm hoping you may have better luck on your setup.
 
TDM:

Sadly yes. Theory says I should be able to find my notes for working on UMS the first time. Reality says otherwise and I'm really hoping I didn't toss them. I did find part of what I was looking for though.

Yoinx:
The old CWM kernel does in fact support UMS. The path is this:

/sys/devices/platform/usb_mass_storage/lun0/file

or

/sys/devices/platform/msm_hsusb/gadget/lun2/file

I'm still not having any luck getting that custom line to work when using your GIT info. I'm hoping you may have better luck on your setup.

Yeah I've tried both those lines with the old kernel as well. Never got anything that worked horribly reliable.

The kernel I synced to hit should be the 5.7mb 3.0.31 kernel which as configured works pretty well under Linux. Windows I'm getting somewhat decent results except that it wants to format Thayer it's presenting as a mounted drive.

I can't figure out what that is though... cuz while it's doing that everything is pointing to the sd card partition. In the paths through adb

I was trying to include a stock init.usb.rc to attempt to fix it. Not sure if it's really helping our nit. But I had duty yesterday so I couldn't work on it much

Sent from my VS920 4G using Tapatalk 2
 
Has anyone actually done any operation testing with either of these builds?

IE: Backup/restore/install roms/fix permissions/force boot recovery with the button combo and ensuring you dont get bootloops/ etc?

I tried a few of them, and they all worked fine for me which is why I'm focusing on the mass storage/adb/usb issues. Just wondering what kinda of mileage everyone else is getting. I've been splitting my time between this and my cwm time mod hack... which *should* apply to this too once I get everything working right.
 
I'd rather wait until I know for sure that it's working decently for more than just myself (other than the usb issues that I'm aware of).

I'm not sure if you're asking for convenience or inability... but... If you can't flash it manually, you would have a hell of time restoring CWM. Since you won't be able to use a flashable zip to restore it if something goes wrong.

If you understand that and really want me to make one for convenience, i will. But I won't be able to help much reverting back.
 
Ahh okay. I'm very inexperienced with recovery-type stuff, and with adb. So the only way I'd be able to test it out would be the only way I know how, flashing through cwm. BUT I figured out how to do it, so I'll start messing with it tonight! :D thank you for all your work Yoinx!!
 
Ahh okay. I'm very inexperienced with recovery-type stuff, and with adb. So the only way I'd be able to test it out would be the only way I know how, flashing through cwm. BUT I figured out how to do it, so I'll start messing with it tonight! :D thank you for all your work Yoinx!!

No worries, Glad you managed to get it worked out.

The only reason I'm trying to avoid the flashable zips right now is if you can flash this recovery manually through ADB you can flash a CWM image with adb from within your rom.

if you can't do that... and you mess up recovery... you're pretty much stuck with cab flashing. I guess you could *probably* run the pwnage tool too to restore cwm.
 
Back
Top Bottom