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

[coolsilver]

Too bad there isn't a simple way to take the haret kernel pull in the nand kernel changes like a patch and go from there.

Sent from my HTC Touch Pro 2 on Android with Tapatalk Pro.

[arrrghhh]

Originally Posted by coolsilver

Too bad there isn't a simple way to take the haret kernel pull in the nand kernel changes like a patch and go from there.

What? He didn't lose the changes to the NAND tree, it's his phone that's not working...

[natemcnutty]

Originally Posted by arrrghhh

What? He didn't lose the changes to the NAND tree, it's his phone that's not working...

I think he just meant to merge the new changes from the linux-msm project into the linux-msm-rhod-nand project.

@Coolsilver: Unfortunately, there are some major differences between booting with haret vs. nand. Some of the changes don't translate over, and nand is a branch that only supports the rhodium devices whereas the "haret" tree supports many devices.

[natemcnutty]

WOOHOO!!! I found the hw3d problem Just an over-provisioning of memory

Line 647 in /arch/arm/mach-msm/board-htcrhodium.c

Change from

mi->bank[0].size = (107 * 1024 * 1024);

To

mi->bank[0].size = (99 * 1024 * 1024);

I also made the following change (but I don't think it affects anything).

Line 280 in /arch/arm/mach-msm/pmem.c

Change from

pmem_setting.ram_console_size=0x2000;

To

pmem_setting.ram_console_size=0x20000;

[pts69666]

Originally Posted by natemcnutty

WOOHOO!!! I found the hw3d problem Just an over-provisioning of memory

Line 647 in /arch/arm/mach-msm/board-htcrhodium.c

Change from

mi->bank[0].size = (107 * 1024 * 1024);

To

mi->bank[0].size = (99 * 1024 * 1024);

I also made the following change (but I don't think it affects anything).

Line 280 in /arch/arm/mach-msm/pmem.c

Change from

pmem_setting.ram_console_size=0x2000;

To

pmem_setting.ram_console_size=0x20000;

AWESOME!!! Congratz! =D

[arrrghhh]

Good news! You the man Hopefully now we can push forward with NAND! C'mon acl, fix that brick!!

[[ACL]]

Originally Posted by natemcnutty

WOOHOO!!! I found the hw3d problem Just an over-provisioning of memory

Line 647 in /arch/arm/mach-msm/board-htcrhodium.c

Change from

mi->bank[0].size = (107 * 1024 * 1024);

To

mi->bank[0].size = (99 * 1024 * 1024);

I also made the following change (but I don't think it affects anything).

Line 280 in /arch/arm/mach-msm/pmem.c

Change from

pmem_setting.ram_console_size=0x2000;

To

pmem_setting.ram_console_size=0x20000;

ahh man .. thats odd.. in reality that should also kill haret version even more.

ANyways if it fixes it then ill run it by bzo who was the original author of the 3d mem patch. If its kosher, we can commit this all around.

Thanks for looking into it while i deal with my nand issues. I tried everything to get it to work and Jonpry has a theory there is a bad block somewhere. Hopefully i can get it resolved soon.

In the mean time can you compile from the latest source now with no issues?

[natemcnutty]

Originally Posted by [ACL]

ahh man .. thats odd.. in reality that should also kill haret version even more.

ANyways if it fixes it then ill run it by bzo who was the original author of the 3d mem patch. If its kosher, we can commit this all around.

Thanks for looking into it while i deal with my nand issues. I tried everything to get it to work and Jonpry has a theory there is a bad block somewhere. Hopefully i can get it resolved soon.

In the mean time can you compile from the latest source now with no issues?

Actually, I'm pretty sure someone already fixed it in linux-msm. I had done a full file comparison when the problem first started happening, and that wasn't there back then. What's odd is the change is listed here:

Commit in linux-msm in Linux on Qualcomm's MSM - Gitorious

Bummer to hear about your nand. I wonder if you've worn out some of the flash on the internal nand. When using a WM ROM, does it have a way of getting wear leveling information that maybe the tinboot isn't doing?

As for compiling, with those changes I have no issues, and boot is crazy fast! I have uploaded it to the first post and cleaned up the post as well.

Edit: Also, I noticed there was some new commits to the RootFS, so I've updated the modified rootfs to include those changes and uploaded that to the first page as well

[[ACL]]

Originally Posted by natemcnutty

Actually, I'm pretty sure someone already fixed it in linux-msm. I had done a full file comparison when the problem first started happening, and that wasn't there back then. What's odd is the change is listed here:

Commit in linux-msm in Linux on Qualcomm's MSM - Gitorious

Bummer to hear about your nand. I wonder if you've worn out some of the flash on the internal nand. When using a WM ROM, does it have a way of getting wear leveling information that maybe the tinboot isn't doing?

As for compiling, with those changes I have no issues, and boot is crazy fast! I have uploaded it to the first post and cleaned up the post as well.

Edit: Also, I noticed there was some new commits to the RootFS, so I've updated the modified rootfs to include those changes and uploaded that to the first page as well

yeah i spoke to bzo. He pointed out my patch was broken on nand. I dont know how i missed it for so long. I wonder now dirty my working tree at home is since i have to commit to both all the time.. lol..

[damage_them]

Originally Posted by natemcnutty

WOOHOO!!! I found the hw3d problem Just an over-provisioning of memory

Line 647 in /arch/arm/mach-msm/board-htcrhodium.c

Change from

mi->bank[0].size = (107 * 1024 * 1024);

To

mi->bank[0].size = (99 * 1024 * 1024);

I also made the following change (but I don't think it affects anything).

Line 280 in /arch/arm/mach-msm/pmem.c

Change from

pmem_setting.ram_console_size=0x2000;

To

pmem_setting.ram_console_size=0x20000;

OHHH YEAAA!! this is the kinda of news i love to see when i come on here great work mehn ..going to test right now!!!