Made a bunch of progress today.

I got a kernel running with support for both USA and INTL devices in the same kernel binary. This is pretty huge.

Also got a pure MTD kernel compiled and created scripts to convert BML/STL to MTD. This is not really tested yet but it's getting closer.

The next big things on my list are redoing the initramfs in a standard way and fixing the deep sleep bug. Maybe after that, we'll have something that can almost pass for a daily driver.

Originally Posted by tdmcode

Made a bunch of progress today.

I got a kernel running with support for both USA and INTL devices in the same kernel binary. This is pretty huge.

Also got a pure MTD kernel compiled and created scripts to convert BML/STL to MTD. This is not really tested yet but it's getting closer.

The next big things on my list are redoing the initramfs in a standard way and fixing the deep sleep bug. Maybe after that, we'll have something that can almost pass for a daily driver.

Thanks for all your hard work! If it weren't for you I wonder where all the 4.0 development would come from...

Sent using Tapatalk

Seriously I don't know how to say you thanks (I mean you did a heck of a job here). Where is your donations link?

Originally Posted by zaclimon

Seriously I don't know how to say you thanks (I mean you did a heck of a job here). Where is your donations link?

I don't accept donations. Keeps things fair and ensures I don't feel guilty if I pause development for a while. But thanks for the thought.

Still working on the initramfs but it's getting close. I'm having fits trying to read the initramfs directly from the kernel. It will probably involve some low level code, possibly assembly, to pass the offset+size from the decompressor to the main kernel. So I'm using the teamhacksung method for now so that I can make progress on the other stuff.

Also decided to integrate the buyer code into the kernel. It's going to work like this:

There will be /proc/buyer_code file exposed by the kernel. The init.rc will copy /efs/buyer_code.dat to it so that the touchscreen driver and sound driver can access it. This will simplify the touchscreen driver loading, which is currently a simple script that passes a module parameter. It will also eliminate the code in the sound code that reads it from /efs. I'll probably do the same with the bt mac addr, because reading files from the kernel is strongly frowned upon.

Had an epiphany on the way home today. I think I can steal some bytes from the end of the cmdline block to pass the initramfs addr+size from the decompresser to the init code. I'll investigate this on Monday.

Still making progress.

I have the efs stuff incorporated into the kernel so that the boot scripts can just do this:

copy /efs/buyer_code.dat /sys/firmware/efs/buyer_code

Then the touchscreen driver and whatever else can use the data.

I put my device back to stock (that boot sound drives me nuts!!) and tested an upgrade. The MTD convert script seems to be working, which is nice.

I think the last piece is getting my boot and recovery initramfs in order. Switching over to the teamhacksung stuff means that the boot and recovery initramfs is taken from the main build like it was meant to be. But that also means that I need to ensure my files in the main build are functional, where I didn't before.

I was able to see the nifty new blue color scheme in CWM 5.x a couple times, so I know it's getting close.

Originally Posted by tdmcode

Still making progress.

I have the efs stuff incorporated into the kernel so that the boot scripts can just do this:

copy /efs/buyer_code.dat /sys/firmware/efs/buyer_code

Then the touchscreen driver and whatever else can use the data.

I put my device back to stock (that boot sound drives me nuts!!) and tested an upgrade. The MTD convert script seems to be working, which is nice.

I think the last piece is getting my boot and recovery initramfs in order. Switching over to the teamhacksung stuff means that the boot and recovery initramfs is taken from the main build like it was meant to be. But that also means that I need to ensure my files in the main build are functional, where I didn't before.

I was able to see the nifty new blue color scheme in CWM 5.x a couple times, so I know it's getting close.

Keep up the awesome work! Nice to see that you are making lots of progress.

Sent using Tapatalk

Originally Posted by tdmcode

Still making progress.

I have the efs stuff incorporated into the kernel so that the boot scripts can just do this:

copy /efs/buyer_code.dat /sys/firmware/efs/buyer_code

Then the touchscreen driver and whatever else can use the data.

I put my device back to stock (that boot sound drives me nuts!!) and tested an upgrade. The MTD convert script seems to be working, which is nice.

I think the last piece is getting my boot and recovery initramfs in order. Switching over to the teamhacksung stuff means that the boot and recovery initramfs is taken from the main build like it was meant to be. But that also means that I need to ensure my files in the main build are functional, where I didn't before.

I was able to see the nifty new blue color scheme in CWM 5.x a couple times, so I know it's getting close.

Great work！ THX！

Well, I thought that I had an error-free kernel partition, so I have been avoiding running bml_over_mtd in the boot process to load the initramfs. But I spent several hours trying to figure out why my recovery would not boot most of the time, but occasionally it would. Turns out my kernel partition has mysteriously developed a bad block that is high enough to only be needed sometimes.

So the bad part is that I wasted a bunch of time trying to figure out my problem. The good part is that bml_over_mtd is now functional.

I swear this device hates me. It's giving me problems at every turn. But I guess that's expected... I hate how Samsung made it do all this weird and proprietary stuff.