[Kernel/Recovery][P75XX] A1 Kernel V1.9/TWRP2.6.3.0

Stuck After 1.3 on Samsung Screen?

  • No, and hope I never do.

    Votes: 179 82.9%

  • Yes, but I solved it by reflashing

    Votes: 8 3.7%

  • Yes, but I solved it by flashing V1.3 or older

    Votes: 8 3.7%

  • Yes, and had to reflash full stock

    Votes: 21 9.7%
  • Total voters
    216
Search This thread
By:
Advanced…
J

jaswinky

Senior Member
Jun 24, 2009
730
134
AAccount said:
I moved the cpu_speedo_nominal down to 1175. I prefer a decrease, rather than increase in voltage. Then I tried to have all cpu nominals and all core nominals the same so I lowered the core_speedo_nominal down to 1225. Now this way all the cpu nominals are 50mV under the core nominals. Surprise! I got the boot bug with that. A quick reflash solved that and it booted. That's really weird since everything fitted the 50mV rule.

I thought the speedo nominals were "jumpstart" values. I also thought the rail values were the "operating" values. I had the jumpstart values a bit higher because after some trial and error setting the jumpstart values a bit higher than the operating values seemed to work better. I personally call that "jumpstaring it hard". I guess there is no getting around checking every single cpu nominal against every single core nominal value.

Why weren't these inconsistencies a problem in the honeycomb overclock kernel??

PS: to mark something as code put "[ C O D E ]" before and "[ / C O D E ]" after.
Click to expand...
Click to collapse

You get boot freeze after moving cpu_speedo_nominal down to 1175? Then, don't release this change :D

Then, what if you change the order for settings initial values? it is, instead of cpu, core and aon, better core, aon and cpu:

In the void __init tegra_soc_init_dvfs(void) funcition, just move:

Code: 
BUG_ON(speedo_id >= ARRAY_SIZE(cpu_speedo_nominal_millivolts));
	tegra2_dvfs_rail_vdd_cpu.nominal_millivolts =
		cpu_speedo_nominal_millivolts[speedo_id];

after this:

Code: 
tegra2_dvfs_rail_vdd_aon.nominal_millivolts =
		core_speedo_nominal_millivolts[speedo_id];


---------- Post added at 03:53 AM ---------- Previous post was at 03:50 AM ----------
jaswinky said:
You get boot freeze after moving cpu_speedo_nominal down to 1175? Then, don't release this change :D

Then, what if you change the order for settings initial values? it is, instead of cpu, core and aon, better core, aon and cpu:

In the void __init tegra_soc_init_dvfs(void) funcition, just move:

Code: 
BUG_ON(speedo_id >= ARRAY_SIZE(cpu_speedo_nominal_millivolts));
	tegra2_dvfs_rail_vdd_cpu.nominal_millivolts =
		cpu_speedo_nominal_millivolts[speedo_id];

after this:

Code: 
tegra2_dvfs_rail_vdd_aon.nominal_millivolts =
		core_speedo_nominal_millivolts[speedo_id];
Click to expand...
Click to collapse

Anyway, if you got the boot issue after changing that initial value, with only this minor change, it means we are moving exactly around the problem and its solution... isn't it?

---------- Post added at 04:05 AM ---------- Previous post was at 03:53 AM ----------
AAccount said:
I moved the cpu_speedo_nominal down to 1175. I prefer a decrease, rather than increase in voltage. Then I tried to have all cpu nominals and all core nominals the same so I lowered the core_speedo_nominal down to 1225. Now this way all the cpu nominals are 50mV under the core nominals. Surprise! I got the boot bug with that. A quick reflash solved that and it booted. That's really weird since everything fitted the 50mV rule.

I thought the speedo nominals were "jumpstart" values. I also thought the rail values were the "operating" values. I had the jumpstart values a bit higher because after some trial and error setting the jumpstart values a bit higher than the operating values seemed to work better. I personally call that "jumpstaring it hard". I guess there is no getting around checking every single cpu nominal against every single core nominal value.

Why weren't these inconsistencies a problem in the honeycomb overclock kernel??

PS: to mark something as code put "[ C O D E ]" before and "[ / C O D E ]" after.
Click to expand...
Click to collapse

Ey my friend... version 1.5.4XE gives allways me the boot freeze issue!!! But previous 1.5.4 allways works fine for me... It is suposed that 1.5.4XE and 1.5.4 are exactly the same version but... you sure??? I've checked it several times... Maybe you got the freeze after my last suggested change because you used a incorrect version with XE?
 
Last edited:
AAccount

AAccount

Senior Member
Sep 8, 2010
1,005
1,909
jaswinky said:
Ey my friend... version 1.5.4XE gives allways me the boot freeze issue!!! But previous 1.5.4 allways works fine for me... It is suposed that 1.5.4XE and 1.5.4 are exactly the same version but... you sure??? I've checked it several times... Maybe you got the freeze after my last suggested change because you used a incorrect version with XE?
Click to expand...
Click to collapse

The only thing that has been changed between 1.5.4XE and 1.5.5XE is that cpu_speedo voltage has been moved down to 1175. I meant another change that I made caused it to get the boot freeze when it shouldn't have theoretically because the 50mV rule was followed. That other change did boot after a reflash but that defies all logic. Specifically in the other change I tried to make all cpu nominals (rail and speedo) 1175 and all core nominals (rail and sppedo) 1225. This way I won't have to do a 4 way check to make sure cpu speedo, cpu rail, core speedo, core rail all follow the 50mV rule.

To be honest, my tablet seems to be pretty flexible with voltages.so it's hard for me to fix a problem I don't see. Perhaps by moving the cpu speedo from 1225 to 1175, your tablet isn't jumpstarted hard enough? See a few posts back for what I mean by jumpstarting.

Oh yeah, any guesses as to why copying and pasting the exact voltages from the honeycomb kernel isn't working? I mean, it worked like a dream in honeycomb.
 
Last edited:
  • Like
Reactions: mamemo1 and jaswinky
C

cpope-11

Member
Apr 21, 2012
47
2
vancouver
On the newest one ex boots up then freezes and es works for about 5 minutes the freeze that only build that really works with my tab 1.1
 
J

jaswinky

Senior Member
Jun 24, 2009
730
134
AAccount said:
The only thing that has been changed between 1.5.4XE and 1.5.5XE is that cpu_speedo voltage has been moved down to 1175. I meant another change that I made caused it to get the boot freeze when it shouldn't have theoretically because the 50mV rule was followed. That other change did boot after a reflash but that defies all logic. Specifically in the other change I tried to make all cpu nominals (rail and speedo) 1175 and all core nominals (rail and sppedo) 1225. This way I won't have to do a 4 way check to make sure cpu speedo, cpu rail, core speedo, core rail all follow the 50mV rule.

To be honest, my tablet seems to be pretty flexible with voltages.so it's hard for me to fix a problem I don't see. Perhaps by moving the cpu speedo from 1225 to 1175, your tablet isn't jumpstarted hard enough? See a few posts back for what I mean by jumpstarting.

Oh yeah, any guesses as to why copying and pasting the exact voltages from the honeycomb kernel isn't working? I mean, it worked like a dream in honeycomb.
Click to expand...
Click to collapse

1.5.5 worked for me perfect! This was my results:

1.5.4: never freeze at boot
1.5.4XE: allways freeze at boot
1.5.5XE: never freeze at boot

Thanks!

Enviado desde mi GT-P7510 usando Tapatalk 2
 
AAccount

AAccount

Senior Member
Sep 8, 2010
1,005
1,909
cpope-11 said:
On the newest one ex boots up then freezes and es works for about 5 minutes the freeze that only build that really works with my tab 1.1
Click to expand...
Click to collapse

Do you still have V1.1? I'm really at quite a loss as to why seeing as my tablet has never given me attitude when flashing a kernel that did 1.2GHz OC. 1.4GHz and 1.6GHz OC yes, my tablet was not always happy but it's been pretty cool with 1.2GHz.

Sorry about that.
 
  • Like
Reactions: mamemo1
P

Philx64

Member
Aug 14, 2008
48
1
Thanks, Xe means more powerfull - more performance ?

Sent from my GT-P7500 using XDA Premium HD app
 
You must log in or register to reply here.

Similar threads

P
[Kernel] - 3.1.10 - JB - 1.4GHZ - Galaxy Tab 10.1 (11/19/12)
Replies
4K
Views
853K
P
pershoot
nakedninja42
[ROM][P75xx][JB][4.2.2] CyanogenMod 10.1 Nightly Discussions
Replies
2K
Views
731K
E
Eaglebiker
Toldo
  • Poll
[ROM][ICS][AROMA] JellyBeanRom ICS v8 [P75xx][End of development&support]
Replies
2K
Views
643K
Toldo
Toldo
alterbridge86
[7.March.2012][ROM][P75xx] ♫♫ Overcome 10.1 Series v2.3.1 ♫♫
Replies
2K
Views
550K
jacky03
jacky03
alterbridge86
  • Locked
[22.Oct.2011][ROM][P75xx] ◄► Overcome 10.1 Series v1.2.3 ◄► Recovery v5.1.2.6 ◄►
Replies
747
Views
188K
alterbridge86
alterbridge86

Top Liked Posts

24 Hours All time
  • There are no posts matching your filters.
  • 187
    AAccount
    AAccount
    :good: First of all :good: Thank you for your patronage!!! :D:cowboy::laugh::victory:

    Before you do anything: This is purely at your own risk and I will not be responsible for anything that happens to your tablet. I flashed it on mine at my own risk.

    Compatibility Information: There are 2 different versions. 1 for stock and 1 for Cyanogenmod. Even though the kernels are intended for the 3G tablet it will work just fine on a WiFi only tablet (like mine). This makes the build process less messy.

    **Sources Fully documented v1.9 beginner friendly sources now available including the build script. If more explanations are required, please say so. However, remember that the explanations are not intended to be a blind step by step instructions of going from samsung source to A1.

    TWRP V2.6.0.0: Please read: do not worry about any kind of "E:A: Render Error". This is a side effect of removing some image files to save place on the recovery image so it will fit on the 5MB. The error simply means it failed to show (render) and image file that it is looking for. This is not harmful at all. If you can't flash the recovery please try the dd method.

    TWRP V2.6.3.0: There are no more E:A Render errors. All graphics that were removed are there. Also, the recovery will install supersu if no root is detected. That feature has also been restored. The kernel has been stripped of anything not recovery related to shrink it down. It also has selinux support which allows the "set_metadata" zip installation to work. The download is in the attachments below.

    Changelog:
    Changelog:
    • Voodoo sound <-- new to version 1.0
    • USB charging enabled as well as charging from other chargers *** fixed V1.5.4 thanks to pershoot
    • Overclock to 1.2GHz <-- new to version 1.2
    • NTFS modules enabled, ondemand default frequency scaler, noop default disk io scheduler <-- new to version 1.2.1
    • GPU overclock from 333MHz to 400MHz. <-- new to version 1.3
    • Overclock to 1.4GHz <--new to version 1.4
    • Support for Current Widget <-- new to version 1.5
    • CIFS enabled in kernel config, 1.4GHz attempted stability fix (thanks to jaswinky) <--new to version 1.5.2
    • Wifi driver compiled as module <--new to V1.5.3
    • More voltage anomalies fixed, updated battery driver to pershoot's 3.1 <-- new to V1.5.4 and V1.5.5XE
    • Fixed charging boot bug <--new to V1.5.6SE
    • (In case you were wondering 1.5.7 had an issue in it.)
    • Disabled frequency lock, added 1GHz default max (you can change it to 1.2GHz or 1.4GHz after boot), 216MHz default min CazeW's suggestion <--new to 1.5.8SE+XE
    • Enabled pershoot build/vfp optimization commit, added 150MHz minimum frequency <-- new to 1.5.9SE+XE
    • Boot script support (put yours as /data/local/a1init.sh) <-- new to V2.0
    • Smarter ram disk that only initializes 3G on 3G tablets<-- new to V2.1
    • 1.5GHz && 1.6GHz OC available, cifs utf8 kernel module available <-- new to V2.2
    • Support for CM9 <-- new to V2.2CM
    • Enabled xbox controller module as xpad.ko <--new to V2.2.1
    • NTFS moved to kernel module instead of built in, CIFS moved to kernel module instead of built in, enabled option.ko and usb_wwan.ko for connection of 3G dongle, all nls charsets/codepages built as modules <--new to V2.2.2
    • V1.6.3 is pretty much V2.2.2 but without my custom ram disk. I've also tried a boot script method proposed by Toldo but haven't had time to test it yet. Please flash the CM9 stock kernel or Samsung stock kernel before trying V1.6.3.
    • Applied git patch to disable mmc_cap_erase <-- new to V1.6.4
    • Voltage control, cifs built in again <-- new to V1.7
    • Galaxy Tab 8.9 specific fix <--new to V1.7.1
    • Cyanogenmod charging mode fix <--new to V1.7.2
    • Galaxy Tab 8.9 specific feature <--new to V1.7.3
    • CM10 compatibility fix, enabled Samsung's powersaving governor <--new to V1.7.4
    • followed pershoot's "-defconfig: jb: unset TEGRA_CONVSERVATIVE_GOV_ON_EARLYSUPSEND" change to fix odd 450MHz capping issue <--new to V1.7.4b
    • Copied in pershoot's keyboard dock driver <--new to V1.7.5
    • 801MB RAM enabled <-- new to V1.8
    • Rebased on pershoot's November 17 source <--new to V1.8.1
    • Back to samsung base, follow pershoot's suggestion of Nexus7 mtp drivers to fix mtp, enabled zRam <--new to V1.8.2
    • Reduced RAM to 785MB to prevent intense 3D games issue, enabled KSM (check in settings --> performance --> memory management)
    • (CM Version only) works on CM10.1 <--new to V1.8.4
    • Incorporated touchscreen death fix. Thank the user who reported it. <-- new to V1.8.5
    • Used part of pershoot's kernel config to get data usage working in CM <--new to V1.8.6
    • Added CM's version of the ROW (read over write) disk IO scheduler for GT-I9300<-- new to V1.8.7
    • Added XDA's grzwolf's solution to the hsic wakeclock problem <--new to V1.8.8
    • Added frandom kernel module <-- new to V1.9
    Credits:
    Supercurio: voodoo sound
    Pershoot: voodoo port, usb charging, flashable zip, OC template, current widget setup
    XDA forum's CazeW: used your replace suggestions to patch up wm8994_voodoo.c and wm8994_voodoo.h so it will compile properly, and finding the missing link for 1.4GHz OC, and disabling frequency locks
    Task650: providing a nice, clean deodexed rom for me to try this on
    XDA forum's RaymanFx: OC instructions
    Me: GPU OC
    XDA forum's jaswinky for valuable hints and checking my work over.
    Xoom ICS kernel: for getting an approximate idea of voltages
    (If I missed any credits, please forgive me as this is my first attempt at any kind of development)
    XDA forum's motley for voltage control
    XDA forum's Unknownzd for mmc_cap_erase patch
    XDA's grzwolf for the hsic wakelock fix
    View
    11
    AAccount
    AAccount
    Very Good News

    So it looks like CM10.2 is actually alive and in testing on my tablet. Big thank you to user epinter for helping me with the graphics issue. I will play with CM10.2 to see how it is. First 10 seconds impression: buttery smooth, wifi works, bluetooth works, sound works, A1 kernel with SELinux firing on all cylinders! Unfortunately it has the 2 bar UI... time to search for that post on how to make tabletui.

    Please donate to epinter for CM10.2. He gets the credit for making it work. I just did some minor tweaking to the fstab and audio flinger.

    Gotta go to bed. Just check out the screenshot zips to see it working. I have no time adjust the sizes for xda.

    THE ZIP IS THE SCREENSHOTS, NOT THE ROM. I will post the rom later this week after testing.
    View
    10
    AAccount
    AAccount
    Honeycomb Tablet Mode Partial Fix

    Hey everyone,

    I managed to figure out how to get the classic honeycomb tablet mode to work in CM10.1 (mostly). You know, the one where you get 1 status bar for the soft buttons and tray. Simply move the jars to /system/framework and the apk to /system/app. Do this with any root explorer or the terminal.

    I tried to make a flashable zip, but I managed to screw the permissions in the process and accidentally pull the plug on my computer, corrupting its data partition. I did setup my Linux install to have a system (/) and data partition (/home) like android.

    The only down side is that you can't customize the power widget but, on the upside, the UI seems noticeably less laggy than the stock CM10.1 UI with the 2 bars. Why??? Who knows. I think this benefit more than justifies the small annoyance of unchangeable toggles.

    Here are the instructions I followed: http://xdaforums.com/showpost.php?p=28957307&postcount=1, http://xdaforums.com/showpost.php?p=36319776&postcount=974.

    Last gotcha, it's for wifi only since that's what I have and I don't really want to wait for my computer to compile a whole other rom for the 3G version. (Have mercy on the Core2 mobile).
    View
    8
    AAccount
    AAccount
    RavenY2K3 said:
    Great to see someone else kernel developing, could be the start of something here... Lol, I wanna give this a try but I think It'll stop 3G working if I flash the 7510 kernel.

    Sent from my GT-N7000 using Tapatalk 2
    Click to expand...
    Click to collapse

    Care to try the 3G version I just uploaded? I don't own the 3G version but I followed the same idea for compiling the 3G kernel.
    View
    7
    AAccount
    AAccount
    Please do not advertise your roms in my kernel thread. This thread is mainly for bug reports and feature requests. Thank you for your understanding. (I did not advertise my A1 in pershoot's kernel thread saying "hey you want some of these features on stock rom, try my kernel". ) I will fix V2.0 to use a 3G ramdisk so 3G tablets will be able to use boot scripts later today.
    View

New posts