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

[MassStash]

are the mtd parts susceptible to something like r/w permissions switching with the data not sticking? first off everythign gets written, then when you dont want root and sys to get written, data doesnt anymore and then "doesnt stick" perhaps?

[[ACL]]

Originally Posted by MassStash

are the mtd parts susceptible to something like r/w permissions switching with the data not sticking? first off everythign gets written, then when you dont want root and sys to get written, data doesnt anymore and then "doesnt stick" perhaps?

hmm .. not sure. but i do know i dumped nand after an install and there was stuff there. Then a reboot and it was all blank.

may need to enable debugging on yaffs. or more importantly see if there are checkpoints being made. i'm trying all this now but its taking a while.

[natemcnutty]

Originally Posted by [ACL]

hmm .. not sure. but i do know i dumped nand after an install and there was stuff there. Then a reboot and it was all blank.

may need to enable debugging on yaffs. or more importantly see if there are checkpoints being made. i'm trying all this now but its taking a while.

Just wanted to say again, thank you so much for all of your hard work. Don't know if you saw, but I did manage to get my git push to work. I tested by commenting out the modules part that breaks our boot. I have a couple of other commits that I'll try to make tonight to fix the sbin folder and fix a couple of other things with init.

[[ACL]]

Originally Posted by natemcnutty

Just wanted to say again, thank you so much for all of your hard work. Don't know if you saw, but I did manage to get my git push to work. I tested by commenting out the modules part that breaks our boot. I have a couple of other commits that I'll try to make tonight to fix the sbin folder and fix a couple of other things with init.

thats awesome... as i keep testing this i see more and more weird stuff.

# cat /proc/yaffs |grep check
always_check_erased.. 0
blocks_in_checkpt.... 2
always_check_erased.. 0
blocks_in_checkpt.... 1
# cat /proc/yaffs |grep check
always_check_erased.. 0
blocks_in_checkpt.... 2
always_check_erased.. 0
blocks_in_checkpt.... 0


this is the same session in android. But look at the checkpoint.. it actually went back to 0..

[MassStash]

where are the rootfs and sys getting flashed from as of now?

did you read irc a lil bit ago where wis was, i believe, enabling jit again>?

[natemcnutty]

Originally Posted by MassStash

where are the rootfs and sys getting flashed from as of now?

did you read irc a lil bit ago where wis was, i believe, enabling jit again>?

There is no rootfs currently (unless ACL is still using initramfs/rootfs). system is getting untar'd from /sdcard/andboot/androidinstall.tar (or you can do .tgz). For JIT, it has a nice debugger tool that I'm betting he is looking for something specific. It hasn't been optimized in any way for our device which is why performance is actually better with it turned off at this point.

[MassStash]

oh yea right right, i remember reading the rootfs decomposition... that's cool, one could hypothetically have different builds to flash to if desired that way i suspect

him n argh are claiming it's squashing sys_server roach....

[natemcnutty]

Originally Posted by MassStash

oh yea right right, i remember reading the rootfs decomposition... that's cool, one could hypothetically have different builds to flash to if desired that way i suspect

him n argh are claiming it's squashing sys_server roach....

Well, if something is calling it and doesn't have a fail condition, that could totally do it. Seems really odd that you can call voicemail and it goes away though.

Just fixed the sbin directory, so you guys should now be able to make a manageable initrd.gz from bootenv now. The rest of the changes are going to be to init which I will pass by you guys before I make any commits. I'll add the lcd density back into it and sed it into /system/build.prop tomorrow.

[MassStash]

blast, yet another day til nand play.... thanks bro, ill be waitin to help out like the rest of us tomorrow

[[ACL]]

latest theory is spl might be wiping nand. i cant prove it yet but thats what i gather from convos.