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

[Starfox]

Well, as long as there's some headway in the PM dept and as long as wifi works, I'm game. Right now my TP2 is usually left on airplane mode and accessing free wifi when I need to do stuff. pm.sleepmode=1 really shuts down everything, afaik it's like sleep on PCs and causes the kernel to stop ticking (as seen via dmesg).

-- Starfox

[natemcnutty]

Originally Posted by [ACL]

I guess i'll make PM my next priority for this kernel. I'm no longer gonna commit anything to haret unless its a hand me down. So i'll save time there.

Lot of work to do so i'll need the rest of the team to handle GB while i handle the kernel.

Edit: Started looking into the power manangement. I'm unsure of the pmsleep modes we pass down on the cmdline. Anyone know the specifics of what these pm sleep modes do? I want to replace all this cmdline pmsleep modes since i dont see why the option is even there.

Swear I'm not dead. Just going through the insane process of buying my first home! Anyway, here's the details on power management (according to a post I can no longer access at XDA):


Sleep Modes: (pm.sleep_mode=) in the startup.txt
This is supposedly for topaz/rhodium that I found on the xda wiki:
* '4' will do "wait for interrupt", no change in arm11's clock or voltage
* '3' will do "wait for interrupt and ramp clock", the arm11's clock is lowered to 20MHz instead of 300-500, and voltage is lowered too.
* '2' will do "app sleep", arm11 is still on, but put into low power mode (registers are still saved)
* '1' and '0' will totally power off the arm11 (so we have to restore registers and things ourself), don't know the differences between them.

I believe we want to use 1, and if I remember correctly, emwe had a patch that fixed the issue of sleep on OC when using pm.sleep_mode=1.

[arrrghhh]

Originally Posted by natemcnutty

Swear I'm not dead. Just going through the insane process of buying my first home!

I just did that a year ago dude - maddening experience at times, but when it's all done it's totally worth it... Congrats!

Originally Posted by natemcnutty

I believe we want to use 1, and if I remember correctly, emwe had a patch that fixed the issue of sleep on OC when using pm.sleep_mode=1.

Yes, 1 is recommended. bzo fixed a PLL2 issue a while ago on sleep_mode=1 and OC. This has been mainline for a while now... .

[[ACL]]

Originally Posted by natemcnutty

Swear I'm not dead. Just going through the insane process of buying my first home! Anyway, here's the details on power management (according to a post I can no longer access at XDA):


Sleep Modes: (pm.sleep_mode=) in the startup.txt
This is supposedly for topaz/rhodium that I found on the xda wiki:
* '4' will do "wait for interrupt", no change in arm11's clock or voltage
* '3' will do "wait for interrupt and ramp clock", the arm11's clock is lowered to 20MHz instead of 300-500, and voltage is lowered too.
* '2' will do "app sleep", arm11 is still on, but put into low power mode (registers are still saved)
* '1' and '0' will totally power off the arm11 (so we have to restore registers and things ourself), don't know the differences between them.

I believe we want to use 1, and if I remember correctly, emwe had a patch that fixed the issue of sleep on OC when using pm.sleep_mode=1.

I was afraid of that.. our cmdline always had pm.sleep_more=1 since the early days. I doubt its worth it to study 0. Now comes the painful task of studying winmo.

congrats on the new home

[[ACL]]

Originally Posted by arrrghhh

I just did that a year ago dude - maddening experience at times, but when it's all done it's totally worth it... Congrats!



Yes, 1 is recommended. bzo fixed a PLL2 issue a while ago on sleep_mode=1 and OC. This has been mainline for a while now... .

I was afraid of this too.. means the code doesnt work well on nand since PM is almost identical (few exceptions for a proper shutdown).

I may have to release multiple test kernels for pm. .. looking at the dlls now to find something juicy.

[arrrghhh]

Originally Posted by [ACL]

I was afraid of this too.. means the code doesnt work well on nand since PM is almost identical (few exceptions for a proper shutdown).

I may have to release multiple test kernels for pm. .. looking at the dlls now to find something juicy.

Nice dude. I'm always here to test... you know where to reach me.

It seems to be a panel issue to be honest. Haret still has panel issues, even after WisTilt2's fine tuning - don't get me wrong, it's a heckuva lot better now because of his careful tuning, but still has issues with waking at times. If it's in the middle of collapsing, it seems if you try to wake it there's some sort of race condition or something getting klobbered (panel? ) and the whole wake process just gets messed up.

It wasn't an issue until the panel was officially collapsed fully - again, thanks to WisTilt2. Before that, the panel always woke quickly because it was never really off/asleep... Oh the irony...

[[ACL]]

Originally Posted by arrrghhh

Nice dude. I'm always here to test... you know where to reach me.

It seems to be a panel issue to be honest. Haret still has panel issues, even after WisTilt2's fine tuning - don't get me wrong, it's a heckuva lot better now because of his careful tuning, but still has issues with waking at times. If it's in the middle of collapsing, it seems if you try to wake it there's some sort of race condition or something getting klobbered (panel? ) and the whole wake process just gets messed up.

It wasn't an issue until the panel was officially collapsed fully - again, thanks to WisTilt2. Before that, the panel always woke quickly because it was never really off/asleep... Oh the irony...

I'm not gonna concentrate on the SODs yet. I want to get it to the point we can go 8hrs with no issues. So improved battlife is my priority. Once we have that i can attack the SODs and all that nonsense.

Funny thing is i'm pretty sure Lmiller had this bad boy running for a while with no charge, i guess something recent broke it. I'll have to check with him again.

[Lmiller1708]

Originally Posted by [ACL]

I'm not gonna concentrate on the SODs yet. I want to get it to the point we can go 8hrs with no issues. So improved battlife is my priority. Once we have that i can attack the SODs and all that nonsense.

Funny thing is i'm pretty sure Lmiller had this bad boy running for a while with no charge, i guess something recent broke it. I'll have to check with him again.

Yes I have had my phone on with NO data and a few calls here and there for ~6 days.
I'm running an old Kernal though... 2.6.27.46-01268-ga175a7c-dirty (built by me) This was right after a bunch of patches from NAND to get the camera working...
With data I can get about ~3 days, again just a few calls here and there.
But I do not use my TP2 that much at all anymore other then phone calls... So I would expect better battery life then most.

As for the SOD's... I get none! But I do have to hit the top button and then the menu key every time to get it to turn on. (Just something you get used too)

[[ACL]]

Originally Posted by Lmiller1708

Yes I have had my phone on with NO data and a few calls here and there for ~6 days.
I'm running an old Kernal though... 2.6.27.46-01268-ga175a7c-dirty (built by me) This was right after a bunch of patches from NAND to get the camera working...
With data I can get about ~3 days, again just a few calls here and there.
But I do not use my TP2 that much at all anymore other then phone calls... So I would expect better battery life then most.

As for the SOD's... I get none! But I do have to hit the top button and then the menu key every time to get it to turn on. (Just something you get used too)

What would i do without you ? only the heavens know. =D>

So this is a while back? any ideas on the date of that test kernel you have ?

There was one nasty patch i made that didnt allow some clocks to be turned off to troubleshoot gsm. Those also may need to be re-enabled.

Its gonna be a pain to find the patch that killed the life time.

[Lmiller1708]

Originally Posted by [ACL]

What would i do without you ? only the heavens know. =D>

So this is a while back? any ideas on the date of that test kernel you have ?

There was one nasty patch i made that didnt allow some clocks to be turned off to troubleshoot gsm. Those also may need to be re-enabled.

Its gonna be a pain to find the patch that killed the life time.

Haha! Same here Bro!

According to our chats... It looks like it was around May 10th -12th.

Hope that helps!!