NAND Testing - 05-25 Update: New LK, Recovery.img, Kernel Updates through Recovery

[natemcnutty]

Originally Posted by [ACL]

Nate. must have broken something. Cant seem to boot off the autobuild version of the initrd.

Missing commits to tinboot? I got the recovery even smaller but hangs on main logo. no double vib so im assuming it may be the same issue as before. Right now the nbh is about 3.4mb

My tinboot2.S is the same as git. All I did was name it recovery.tgz, put it in /bootenv/initrd, and then built initrd.gz. Then I copied that into /tinboot-linux-msm/kernel, and built the rest. Here is my build script:

#!/bin/sh
#-------------------------------------------------------------------------
# Build Script for linux-msm-rhod-nand (Android on HTC) kernel, modules, XIP, and NBH
#-------------------------------------------------------------------------
#
# Set the following variables
# ANDROID -- Root director for Android files
ANDROID=~/Desktop/android
# TINBOOT -- tinboot directory
TINBOOT=$ANDROID/tinboot-linux-msm
# TOOLCHAIN_PATH -- directory containing the arm toolchain
TOOLCHAIN_PATH=$ANDROID/toolchain/arm-2010.09/bin/arm-none-eabi
#-------------------------------------------------------------------------

#build initrd and copy to Tinboot
cd $ANDROID/bootenv
sh buildInitrd
cp -f initrd.gz $TINBOOT/kernel

#-----start build XIP-----
# remove old XIP if it exists
cd $TINBOOT
if [  -f xip/rhod ]; then
    rm xip/rhod
    echo Deleting old XIP
fi

# make XIP folder if it does not exist
if [ ! -d "$TINBOOT/xip" ] ; then
mkdir $TINBOOT/xip
fi

# compile tinboot XIP

echo "Start compiling tinboot/xip for rhod....."
$TOOLCHAIN_PATH-as  tinboot/tinboot2.S -o tinboot.o --defsym rhod=1 --defsym MTYPE=2292
$TOOLCHAIN_PATH-objcopy tinboot.o -O binary tinbootxip
mv tinbootxip xip/rhod

# cleanup tinboot
rm tinboot.o

# make sure XIP build was successful
if [ ! -f xip/rhod ] ; then
    fail "XIP build failed. XIP file was not found."
fi


#-----start build NBH-----
# clean up previous log file
if [ -e "tools/log" ]; then
rm tools/log
fi

# compile tinboot
cp tools/rhod_payload2 os.nb.payload

echo "Inserting tinboot into payload"
wine tools/osnbtool -c os.nb.payload 1 xip/rhod >> tools/log
mv os.nb.payload.NEW os.nb.payload >> tools/log

echo "Inserting blank imgfs into payload"
wine tools/ImgfsToNb.exe  tools/imgfs.bin os.nb.payload os-new.nb.payload >> tools/log

echo "Creating os.nb portion of nbh"
tools/nbmerge < os-new.nb.payload > os-new.nb

echo "Creating NBH"
wine tools/yang.exe -F RHODIMG.NBH -f os-new.nb -t 0x400 -s 64 -d RHOD****0 -c 11111111 -v Tinboot -l WWE >> tools/log

# clean up payloads
rm os.nb.payload
rm os-new.nb.payload
rm os-new.nb

# clean up tinboot directory
rm -rf xip
mv -f RHODIMG.NBH ~/Desktop

I know this works because I had arrrghhh and others flash the 5MB one that had full busybox in recovery without any issues. Do you think it could be a toolchain difference, or is it possible I'm compiling the NBH differently?

[[ACL]]

Originally Posted by natemcnutty

My tinboot2.S is the same as git. All I did was name it recovery.tgz, put it in /bootenv/initrd, and then built initrd.gz. Then I copied that into /tinboot-linux-msm/kernel, and built the rest. Here is my build script:

#!/bin/sh
#-------------------------------------------------------------------------
# Build Script for linux-msm-rhod-nand (Android on HTC) kernel, modules, XIP, and NBH
#-------------------------------------------------------------------------
#
# Set the following variables
# ANDROID -- Root director for Android files
ANDROID=~/Desktop/android
# TINBOOT -- tinboot directory
TINBOOT=$ANDROID/tinboot-linux-msm
# TOOLCHAIN_PATH -- directory containing the arm toolchain
TOOLCHAIN_PATH=$ANDROID/toolchain/arm-2010.09/bin/arm-none-eabi
#-------------------------------------------------------------------------

#build initrd and copy to Tinboot
cd $ANDROID/bootenv
sh buildInitrd
cp -f initrd.gz $TINBOOT/kernel

#-----start build XIP-----
# remove old XIP if it exists
cd $TINBOOT
if [  -f xip/rhod ]; then
    rm xip/rhod
    echo Deleting old XIP
fi

# make XIP folder if it does not exist
if [ ! -d "$TINBOOT/xip" ] ; then
mkdir $TINBOOT/xip
fi

# compile tinboot XIP

echo "Start compiling tinboot/xip for rhod....."
$TOOLCHAIN_PATH-as  tinboot/tinboot2.S -o tinboot.o --defsym rhod=1 --defsym MTYPE=2292
$TOOLCHAIN_PATH-objcopy tinboot.o -O binary tinbootxip
mv tinbootxip xip/rhod

# cleanup tinboot
rm tinboot.o

# make sure XIP build was successful
if [ ! -f xip/rhod ] ; then
    fail "XIP build failed. XIP file was not found."
fi


#-----start build NBH-----
# clean up previous log file
if [ -e "tools/log" ]; then
rm tools/log
fi

# compile tinboot
cp tools/rhod_payload2 os.nb.payload

echo "Inserting tinboot into payload"
wine tools/osnbtool -c os.nb.payload 1 xip/rhod >> tools/log
mv os.nb.payload.NEW os.nb.payload >> tools/log

echo "Inserting blank imgfs into payload"
wine tools/ImgfsToNb.exe  tools/imgfs.bin os.nb.payload os-new.nb.payload >> tools/log

echo "Creating os.nb portion of nbh"
tools/nbmerge < os-new.nb.payload > os-new.nb

echo "Creating NBH"
wine tools/yang.exe -F RHODIMG.NBH -f os-new.nb -t 0x400 -s 64 -d RHOD****0 -c 11111111 -v Tinboot -l WWE >> tools/log

# clean up payloads
rm os.nb.payload
rm os-new.nb.payload
rm os-new.nb

# clean up tinboot directory
rm -rf xip
mv -f RHODIMG.NBH ~/Desktop

I know this works because I had arrrghhh and others flash the 5MB one that had full busybox in recovery without any issues. Do you think it could be a toolchain difference, or is it possible I'm compiling the NBH differently?

ahh i see the differences. The payload is diff plus you are using osnbtool.

I lost track of all the payloads we had. Is your rhod_payload2 different that the rhod_payload we use for git? remember autobuild uses exactly wahts on git.

[natemcnutty]

Wirelessly posted (Opera/9.80 (Android; Opera Mini/6.0.24212/24.743; U; en) Presto/2.5.25 Version/10.54)

Originally Posted by [ACL

]

ahh i see the differences. The payload is diff plus you are using osnbtool.

I lost track of all the payloads we had. Is your rhod_payload2 different that the rhod_payload we use for git? remember autobuild uses exactly wahts on git.

Ahh yes. The payload 2 is really important. That's the one I fixed the partition table! In the other payload, fatfs partition overwrites tinboot area on reboot.

Also, using nbtools fixes the partition tables based on size of tinboot's XIP.

[[ACL]]

Originally Posted by natemcnutty

Wirelessly posted (Opera/9.80 (Android; Opera Mini/6.0.24212/24.743; U; en) Presto/2.5.25 Version/10.54)



Ahh yes. The payload 2 is really important. That's the one I fixed the partition table! In the other payload, fatfs partition overwrites tinboot area on reboot.

Also, using nbtools fixes the partition tables based on size of tinboot's XIP.

ok so this is a new creation? nothing we have on git rite. Upload it so i can put it on git and have autobuild make one of these nbh with the recovery on it.

[natemcnutty]

Wirelessly posted (Opera/9.80 (Android; Opera Mini/6.0.24212/24.743; U; en) Presto/2.5.25 Version/10.54)

Originally Posted by [ACL

]

ok so this is a new creation? nothing we have on git rite. Upload it so i can put it on git and have autobuild make one of these nbh with the recovery on it.

You have the rhod_payload2 already on git. It's the one I made a few weeks back. Maybe autobuild was just never updated. Try using my script

[[ACL]]

Originally Posted by natemcnutty

Wirelessly posted (Opera/9.80 (Android; Opera Mini/6.0.24212/24.743; U; en) Presto/2.5.25 Version/10.54)



You have the rhod_payload2 already on git. It's the one I made a few weeks back. Maybe autobuild was just never updated. Try using my script

ahh ok.. i forgot u made that one. I though it was the original original we used. Ok never mind ill try it out.

[pts69666]

You know what I find a bit odd? Data works for me in Android; however, when I flash back to WinMo, I have no data. When I try to update data profile, I get an error 1012. I flash back to Android, and I have data again. Is it a sign? =P

[natemcnutty]

Originally Posted by pts69666

You know what I find a bit odd? Data works for me in Android; however, when I flash back to WinMo, I have no data. When I try to update data profile, I get an error 1012. I flash back to Android, and I have data again. Is it a sign? =P

Yikes, that doesn't sound good. I hope nothing we are doing in Android is affecting Windows Mobile. A good thing to test would be flashing stock and see if that fixes data in WinMo. I don't think we can affect the radio firmware from within Android, but I guess anything is possible.

[MassStash]

[acl], double check the auto build script is useing payload2, that was my prob being stuck there... hopefully that easy

[arrrghhh]

Originally Posted by pts69666

You know what I find a bit odd? Data works for me in Android; however, when I flash back to WinMo, I have no data. When I try to update data profile, I get an error 1012. I flash back to Android, and I have data again. Is it a sign? =P

lol, it is a sign.

I don't have this experience tho. Have you tried different WinMo ROM's, or OEM like nate suggested?

Stock should include radio firmware.