New (very fast) lag fix using internal memory and playlogo1 hack

[Guest Andrew Luecke]

Hi all, first post here, glad to introduce myself with a quadrant of 2263 :-)

Expect some paypal soon!

thanks

I think when it comes to wear levelling, a lot of you guys are missing the point. What I've been asking is if the phones memory directly, or it uses something like XSR http://www.samsung.com/global/business/sem...R_Brochure.html

Because if it uses XSR, I get the impression that wear leveling is handled by XSR (which it probably is, because FAT32 is certainly a poor choice for flash otherwise).

But if not, we need to ask ourselves where wear leveling takes place. Either way though, writes should be reduced as much as possible. From what I understand, RFS is basically Fat + Journaling anyway, but using filesystems which minimise writing is possibly beneficial (anybody know how many write cycles the NAND is worth, its a SLC, so it might actually be irrelevent).

[Guest edxter]

Another tip, if you check, the fix made backups on my phone of the various folders (/data/data/bak, /data/dalvik-cache.bak, /data/system.bak)), presumably to easily back out of this or as some sort of fail-safe. You could delete those, or better yet back them up to your /sdcard for future emergencies, and that would free up a lot of space (depending on how many apps you'd already had installed when running this.

Could you elaborate more on this and kindly specify the exact files location? I'm using Root explorer and when I have a look at /data I see data within data within data.... not sure which one to delete... :)

[Guest DistortedLoop]

Could you elaborate more on this and kindly specify the exact files location? I'm using Root explorer and when I have a look at /data I see data within data within data.... not sure which one to delete... :)

Don't use root explorer, use a terminal shell or adb shell.

Use the ls -l command:

example, ls -l /data

The output will show you what's a symlink, and what's not. But if you're just interested in saving space removing the backup folders, they should be obvious, if they're there. It will be something like this.

/data/data.bak

/data/dalvik-cache.bak

/data/system.bak

(edit: if you want to use root explorer, these should all be in the /data folder...so just navigate from your root (/) to the first data folder you see, once in there you should see a data folder and a data.bak folder, and the same for the other two. You'd delete the ones that end in .bak --- BE VERY CAREFUL YOU HAVE THE CORRECT FOLDERS)

Those were copies the one-touch script made as backups. They're actually directories, so you have to use the -r flag to remove them from the terminal if you do it.

I removed those files when I manually did the mimocan fix once I was sure my system was stable. I imagine it will be safe to do so with this method ASSUMING that they're not needed for some kind of automatic fall-back/restore method if there's a problem at boot-up mounting the loop drive.

You should ask the method's author or review all the scripts they're running to see if these are ever called for in some scenario I haven't thought of.

I've reviewed the userinit.sh, and the removelagfix.sh that the one-click method uses and they never reference the *.bak files I mentioned above. The removelagfix.sh copies the data back from the ext2 drive - which is the smart way to keep your device current with the apps you've installed since the fix. Based on this, I personally wouldn't be afraid to delete these if space is a concern for me, they're not needed if your system is stable. As usual, make sure you have a good nandroid backup and titanium apps backup before messing around with this stuff.

Are you really that concerned about only having 600mb free space on the paritition that's not going to be used for much stuff now that you've moved all apps and their data to the ext2 partition/file? That 600mb is still more free space than most other Android phones ship with isn't it? And you've got a whole gig still for your apps.

Edited August 10, 2010 by DistortedLoop

[Guest edxter]

Woo, thank you mate! that's a pretty good explanation! i'll have a look this weeekend.

BUG REPORT: Battery indicator: it shows as full in the notifications bar but is at 80% in the settings menu. Not showing charging when plugged to power source neither. I'll wait and see if it notifies when the battery is fully charged, fingers crossed :)

[Guest lucasjohny]

http://forum.xda-developers.com/showthread.php?p=7581558

try this 1 click lag fix, let this script do for you then type yourself.

[Guest Gustaf Jarnling]

thanks mate! gone from JF3 to Pauls JP3 r2 and this is the only fix I've applied. Pauls ROM doubled the linpack readings and this fix doubled the quadrant ones. but more importantly I can really feel the difference; for example Dolphin used to fc with Flash 10.1 enabled, now it runs like a charm.

just a quick note. the latest version of busybox (1.18.0.git) don't have the mount and remount applets, so as of now it seems you need that 1.17-version linked to above to implement this fix as instructed.

thanks to everybody at MoDaCo (and XDA) for all the good work!

[Guest AndyCr15]

I couldn't resist and just ran through all the install steps by hand. Fairly painless, no problems. Should have done that in the first place. Anyways...

THIS HACK IS CRAZY SICK FAST.

The phone literally flies through everything. It's not placebo like some accused when mimocan first came out. I noticed the difference as soon as the phone booted into the lock screen and all the little apps I have autostarting just seemed to pop open and close very quickly, much more quickly than usual. Things feel snappier too, apps opening, app drawer scrolling etc. Some of that may be placebo, but I don't think so. I should have videod the phone before and after.

My SGS was never a speed demon, even on mimocan I only got low 1400s generally on Quadrant. I just got 2155. Pretty impressive.

I agree. Last night I decided to give FroYo a whirl, so I started from scratch, flashed Paul's r2 (or is it r3 at the mo?) and then applied the one click fix. Wow, that a difference. It's finally performing like it should.

The only question, I'm still only scoring 1750. This is from a totally fresh start. With Mimo's fix I only ever scored 1100. Does my SD card have anything to do with it, as that's only class 2.

[Guest ragflan]

I'm going to try this today. I have the mimocan's fix and I get Quadrant benchmark scores of like 1600-1700. But the phone is way too weird. Sometimes it works like the fastest smoothest phone on the planet. But sometimes, everything lags. For example, yesterday all I did was click on the Phone icon and the whole screen went black and only the notification bar was available which I could not pull down. I press home and nothing happened, back button and nothing. It just stayed there for a LONG time and then appeared. Maybe like 15-20 seconds. It happens at other times too. Its got great scores but the phone still lags and at times I don't want it to lag. That's really hard to bear like when I'm on the job and I need to quickly call someone to check something, it lags.

Since the author of the fix is recommending to flash this over stock firmwares with no other fixes, I wanted to ask. I'm on JG5 with r3 from Paul. What's the best firmware right now?

Edited August 10, 2010 by ragflan

[Guest AndyCr15]

Since the author of the fix is recommending to flash this over stock firmwares with no other fixes, I wanted to ask. I'm on JG5 with r3 from Paul. What's the best firmware right now?

That's what I'm on at the moment, the latest of Paul's. Works great.

Would someone do me a favour? Please post a screenie of a 2200+ quadrant? I only get 1700 :) I want it for my site, if no-one minds, where I'm putting all the info in one place, like I did with Mimo fix.

[Guest DistortedLoop]

I agree. Last night I decided to give FroYo a whirl, so I started from scratch, flashed Paul's r2 (or is it r3 at the mo?) and then applied the one click fix. Wow, that a difference. It's finally performing like it should.

The only question, I'm still only scoring 1750. This is from a totally fresh start. With Mimo's fix I only ever scored 1100. Does my SD card have anything to do with it, as that's only class 2.

My initial answer was going to be that no, for this particular hack the external sdcard is not involved, so it can't be impacting the Quadrant score. I'm pretty sure that's the case because the external card is mounted as /sdcard/sd, so even if Quadrant actually includes "sdcard" testing, it's going to test the standard /sdcard mount point, which for us is the internal card.

It is a bit frustrating to be on the low end of the benchmarks we see reported. I was with mimocan; I routinely got low-1400s/high-1300s, best ever was low 1500s. With this particular hack I seem to be in the upper range for a phone not hacked with anything else (JIT, OC, etc) with a 2177, but I can't get it to punch over 2200.

It's quite possible all phones aren't built the same...different nand batches, cpu runs, etc, but I wonder to what extent other things such as what launcher you're using, widgets, etc impact this. I killed all tasks with ATM after a 2177 run just now, and scores actually went down every run. Go figure.

That's what I'm on at the moment, the latest of Paul's. Works great.

Would someone do me a favour? Please post a screenie of a 2200+ quadrant? I only get 1700 :) I want it for my site, if no-one minds, where I'm putting all the info in one place, like I did with Mimo fix.

Here's a 2177 image if that's close enough. Feel free to use it until someone else posts something faster.

[Guest Compuse]

Talking about overkill? :)

btw. this is no solution against the lag!

Edited August 10, 2010 by Compuse

[Guest DistortedLoop]

I'm sure most reading here are also reading the xda threads, but in case you're not, there's a linux guru on the loop mount hack thread insisting that this ext2 is not stable and it's not a matter of you might lose data, it's a matter of you will lose data running ext2, even within RFS.

His best argument/example is to set up an ext2 partition within a VMWare virtual machine and then kill the power to it while it's writing data and see that when you restart the VM that the fsck is ran because of the dirty shutdown. The logic there apparently being that your VMWare machine is running on top of a journaled OS (like HPFS on a Mac).

It's a good argument on the surface, but I can see the counter-argument in my mind of it's not the same, since VMWare is a virtual OS whereas the loop mount is just a virtual partition.

Still think if you have anything critical in your apps you should keep current and multiple version backups with Titanium or something similar.

[Guest DistortedLoop]

Talking about overkill? :)

btw. this is no solution against the lag!

I've seen a few people say that, but what specifically makes you say it?

I did notice a stutter once or twice yesterday, but not sure I'm ready to say I have lag on the phone. Even with the occasional stutter/lag, the phone otherwise runs so wicked fast at everything else it may be worth the trade-off.

I think morfic's anticipatory scheduler hack isn't getting the attention it probably deserves because of its release so coincidental in time with this one. Morfic's doesn't seem to affect benchmarks, but he's reporting no lag/stutters with it. Worth checking out since it's so easy to implement and so easy to undo.

[Guest JexWang]

If i reset to factory setting my phone, will it automatically erase this trick?

[Guest DistortedLoop]

If i reset to factory setting my phone, will it automatically erase this trick?

I'm not sure...Reset only wipes certain parts of the phone from user installed stuff/settings. I don't think reset will erase the symlinks that this made or re-create the deleted directories.

That said, when I managed to screw up my phone to the point it wouldn't even boot past the boot logo experimenting with the one-click lag-fix script, a factory reset from Recovery did get my phone back to a bootable state so that I could at least put the Clockwork ROM update.zip back in the proper location to restore a nandroid backup.

[Guest Compuse]

I've seen a few people say that, but what specifically makes you say it?

I did notice a stutter once or twice yesterday, but not sure I'm ready to say I have lag on the phone. Even with the occasional stutter/lag, the phone otherwise runs so wicked fast at everything else it may be worth the trade-off.

I think morfic's anticipatory scheduler hack isn't getting the attention it probably deserves because of its release so coincidental in time with this one. Morfic's doesn't seem to affect benchmarks, but he's reporting no lag/stutters with it. Worth checking out since it's so easy to implement and so easy to undo.

to my believes, we're tricked with a "better"(other) memory check in the quadra bench .. the lag still occurs; occationally, why would i lie? :)

[Guest Compuse]

didn't noticed some smoother fade to apps and back extra, and it feels slightly more direct upon my keystrokes i've made ..

[Guest DistortedLoop]

to my believes, we're tricked with a "better"(other) memory check in the quadra bench .. the lag still occurs; occationally, why would i lie? :)

I didn't say you were lying... I just asked what made you think this didn't address lag? I was curious if you have experienced lag, and under what situations. If we all share our experiences, we will all learn better.

[Guest Compuse]

At first(before the fix) the lag seems to occur when hitting the message button or switching between programs; several hours of playing with the phone gave around 10/15 small black lockups..

Yesterday night i tried freaking around a bit more with it, .. only 2 times it lagged on me, definitely an improvement!!

[Guest DeepBass909]

Guys,

I've tried the hack yesterday evening, feeling kinda confident because I know linux quite good and so I'm allready familiar with most of the nessecery commands needed. But... Something went wrong and the fix didn't work and caused my phone to start-up badly (slow, no start-up animation).

So I've tried to revert the fix, but did something wrong... And now the original "playlogo1" got overwritten with the content of the replacement "playlogo1".

I did some damage investigation, and found out that "playlogo1" (the one from the fix) is nothing more than a script, used to call "userinit.sh" and "playlogonow". Because "playlogosnow" also got corrupted, I can't get the startup animation anymore, and I can imagin that on start-up the system generates some errors about files not being found...

Can someone please confirm that the original "playlogo1" is also nothing more than a shell-script, and if so, what is the original content of this script? And if not, can someone please provide a copy of the original "playlogo1"-file?

Many, many, many thanks!

[Guest DistortedLoop]

Guys,

I've tried the hack yesterday evening, feeling kinda confident because I know linux quite good and so I'm allready familiar with most of the nessecery commands needed. But... Something went wrong and the fix didn't work and caused my phone to start-up badly (slow, no start-up animation).

So I've tried to revert the fix, but did something wrong... And now the original "playlogo1" got overwritten with the content of the replacement "playlogo1".

I did some damage investigation, and found out that "playlogo1" (the one from the fix) is nothing more than a script, used to call "userinit.sh" and "playlogonow". Because "playlogosnow" also got corrupted, I can't get the startup animation anymore, and I can imagin that on start-up the system generates some errors about files not being found...

Can someone please confirm that the original "playlogo1" is also nothing more than a shell-script, and if so, what is the original content of this script? And if not, can someone please provide a copy of the original "playlogo1"-file?

Many, many, many thanks!

If you re-read/walk through the scripts to create this "fix" and to undo this "fix", you see the following commands:

create:

echo "Moving boot animation"

busybox mv /system/bin/playlogos1 /system/bin/playlogosnow

echo "Copying over init scripts"

busybox cp /sdcard/playlogos1 /system/bin/playlogos1

undo:

echo "Removing auto boot support"

busybox mv /system/bin/playlogosnow /system/bin/playlogos1

So, the original playlogos1 was moved/renamed to playlogosnow

then the new script was moved/named playlogos1

To undo, all we do is move/rename the playlogosnow file back to playlogos1

I just checked on my phone and the playlogosnow is not a script, it is some kind of binary.

Pretty clear that what we did was rename the bootlogo binary with a script that then calls the bootlogo binary to play. The original playlogos1 was not just a script as far as I can tell.

Are you sure your playlogosnow is actually corrupted. You should be able to just move/rename it back to playlogos1 to get the original playlogos1 back on your phone.

Anyways, in case you really did corrupt your playlogosnow file, here's a copy off my phone you can use.

So remember, if your after the original playlogos1 file, just rename this upload to playlogos1, since this is the playlogos1 from my phone that was renamed playlogosnow as part of the fix.

Hope this helps!

playlogosnow.zip

Edited August 11, 2010 by DistortedLoop

[Guest Paul]

Ran this fix on Froyo and it does seem pretty good, will see how it pans out over time! :D

P

[Guest Astridax]

Could any one imagine doing this with the overclocked 1.25 ghz kernel. People are getting 2500 with that and possibly some sort of lag fix, so would be interesting to see this with the OC'd kernel.

[Guest DistortedLoop]

Bluetooth was never a problem on my SGS with my car. After the fix here, bluetooth has had issues connecting to the car unless I reboot, and one attempt to connect to bluetooth caused the phone to shut itself off on its own. That's the first time the phone's ever crashed itself in just normal use.

Again, never a problem on the phone, even with the mimocan fix applied.

Not ready to point the finger 100% at the fix, because I think I had been running the "anticipatory scheduler" hack at the same time. I'm not going to enable the anticipatory scheduler and see if the bluetooth settles down.

Anyone else noticed bluetooth issues with this.

[Guest Xoneso]

Ran this fix on Froyo and it does seem pretty good, will see how it pans out over time! :D

P

Hey Paul.

Have you had a chance to look at a fix so we can use update.zip's on JP3 again? Kinda hate that i can't run my own update.zip that i have made for eclair and jp1+jp2 on xda.

Also running this with the anticipatory hack from morfic and i have to say it's running pretty nicely together.