Its been very quiet in these forums recently, probably because everyone has bought a Galaxy S2 or a HTC Sensation!
I recently came across UPX, this is a compression utility that can shrink exe and dll files quite dramatically, saving space and increasing application launch speed quite dramatically.
I am currently using Mirom8 and even though this Sense 2021 rom is protected have decided to list exe and dll files that can (I think) be UPXed for speed
Here is a list of exe and dlls from the \windows folder and their UPXed size:
AdjustMotionSensor.exe 109k
AdobeReaderLE.exe 1050k
AlarmPopUp.exe 105k
AlbumSearcher.exe 135k
AudioBooster.exe 87k
AudioManager_Eng.exe 76k
BackupRestoreUI.exe 220k
Camera.exe 161k (nice increase in launch speed here)
CommManager.exe 152k
ConnectionSetup.exe 112k
ConnectionSetupAuto.exe 115k
ContactEditor.exe 66k
EmailSetupWizard.exe 97k
Facebook.exe 206k
HTCAlbum.exe 400k
HTCAppointment.exe 105k
HTCBookmark.exe 123k
HTCDRMEngine.exe 949k
HTCMsgEnhance.exe 110k
HTCPhotoPicker.exe 358k
OneNoteMobile.exe 268k
PPCPimBackup.exe 147k
PPT.EXE 687k
PushClient.exe 137k
PWORD.EXE 128k
PXL.EXE 291k
RSSHub.exe 266k
ServiceMode.exe 160k
SettingImprovement.exe 90k
SIM_MGR.EXE 82k
SimLockP.exe 49k
SimMgr.exe 117k
SPMC.EXE 117k
StreamingPlayer.exe 165k
Uploader.exe 97k
WiFiNetwork.exe 67k
wma9prodecoder.dll 236k
wmv9decoder.dll 258k
WMVDMOE.DLL 238k
YouTube.exe 145k
And other useful savings
Opera 10, 2 largest files can be UPXed, flashlite, the large dll can be UPXed
S2U2, all exe's can be UPXed, makes the unlock screen quite smooth to use
Its not worth UPXing files less that 200k although it might help a slow launching app around the 150k size, and normally its not worth UPXing files that stay resident in ram all the time, as it can use quite a lot more ram than normal, but if you launch apps a lot, then it can really make a difference to the speed of opening your favourite applications. Apps that use .net cannot be UPXed either, so that is a limiation otherwise more files could be UPXed.
Very good work!!! Have you send me all the files? I will over weekend swap the files and re cook mirom8.
mirolg said:
Very good work!!! Have you send me all the files? I will over weekend swap the files and re cook mirom8.
Click to expand...
Click to collapse
Hi Miro
I will send you a rar file again via email today at some point.
This can be done to other rom releases too you know, i am unsure of total rom savings space wise, possibly 20MB-30MB, not huge but we are talking a whole 20MB plus of internal storage free for something else.
Also it would be useful to know whereelse in the rom there are exe files i can try to upx, must be other places.
(InsertNameHere) said:
Its been very quiet in these forums recently, probably because everyone has bought a Galaxy S2 or a HTC Sensation!
Click to expand...
Click to collapse
Yes, it seems that only few person are on the forum. Maybe drupad will show up after return from vacation. It is over
Regards
fxdjacentyfxd
(InsertNameHere) said:
Hi Miro
I will send you a rar file again via email today at some point.
This can be done to other rom releases too you know, i am unsure of total rom savings space wise, possibly 20MB-30MB, not huge but we are talking a whole 20MB plus of internal storage free for something else.
Also it would be useful to know whereelse in the rom there are exe files i can try to upx, must be other places.
Click to expand...
Click to collapse
miROM1 and miROM8 are good ROMs and if we can save 20 - 30 MB Storage, then will be much better! i try it and give you a feedback!
good work, tnx
mirolg said:
miROM1 and miROM8 are good ROMs and if we can save 20 - 30 MB Storage, then will be much better! i try it and give you a feedback!
good work, tnx
Click to expand...
Click to collapse
How about coking this with latest manila u had in mind??
thanks for this!!!
nikola92 said:
How about coking this with latest manila u had in mind??
Click to expand...
Click to collapse
Miro is working on it for us
hi, a short statement:
my sense 2021 package i have compressed with cfc.method the result is sense and sense tabs starts quicker and works really faster. ok. miROM1 and mirom8 are in the lastest relases also fast, but all starts 0,5 sec. quicker. UPX method compress the exe + dll files and the result is from app to app different- but i mean the most works better and we save few MB storage. miROM8 has now 270 MB and miROM1 has 256 MB free Storage.
Both ROMs are with all fixes and updates incl. cam-button and toshiba power + wifi settings.
I would test it and will upload it next days
cheers, miRO
mirolg said:
hi, a short statement:
my sense 2021 package i have compressed with cfc.method the result is sense and sense tabs starts quicker and works really faster. ok. miROM1 and mirom8 are in the lastest relases also fast, but all starts 0,5 sec. quicker. UPX method compress the exe + dll files and the result is from app to app different- but i mean the most works better and we save few MB storage. miROM8 has now 270 MB and miROM1 has 256 MB free Storage.
Both ROMs are with all fixes and updates incl. cam-button and toshiba power + wifi settings.
I would test it and will upload it next days
cheers, miRO
Click to expand...
Click to collapse
Good news and sounds quite exciting, i am glad you found a way to get CFC working with 2021, i had read it was somewhat more difficult than previous versions like 2012, and UPX saves some space, thats good too.
270MB user storage is incredible i would say!
Great job!
Which version of UPX did you use ?
can i have a link ?
lesscro said:
Which version of UPX did you use ?
can i have a link ?
Click to expand...
Click to collapse
Any version above 3.03 supplied with this batch utility works fine, but i used the latest version 3.07
Best to do all compression on your pc rather than use a PPC version, it can take quite a long time to compress a large exe file.
thx for all... i will test myself on my TG !
Actually i tried this files to perform a speed test :
Code:
Explorer.exe 268 288 2010-09-07 20:46 -a---
GoogleMaps.exe 2 354 176 2010-06-25 19:47 -a---
iGo.exe 8 613 376 2011-05-08 09:07 -a---
iManager.exe 356 352 2011-02-21 17:00 -a---
InfoTrafic Trains.exe 319 488 2010-03-17 23:28 -a---
Opera10-armv4i.exe 623 400 2010-03-15 20:48 -a---
PhotoManPro.exe 1 108 480 2010-04-26 16:29 -a---
player.exe 1 894 672 2009-09-17 11:14 -a---
RCapture.exe 90 112 2010-04-26 16:28 -a---
s2p.exe 364 032 2010-07-17 05:35 -a---
SKTools.exe 2 657 208 2010-03-02 15:09 -a---
ThemeChanger.exe 113 912 2007-04-18 07:43 -a---
Weather-iGo-PPC-v2.exe 264 192 2010-11-11 17:49 -a---
And only a few Exe seems to be work like my old test with my P535 long time ago...
it appear too somes EXE manage in .NEt/win32 are not supported actually !
But when it work... it s soooooooooooooooooo fastest !
Working file compression :
Code:
Explorer.exe 252 928 2010-09-07 20:46 -a---
GoogleMaps.exe 855 040 2010-06-25 19:47 -a---
iManager.exe 203 264 2011-02-21 17:00 -a---
Opera10-armv4i.exe 246 568 2010-03-15 20:48 -a---
PhotoManPro.exe 499 712 2010-04-26 16:29 -a---
RCapture.exe 53 248 2010-04-26 16:28 -a---
then i have seen on Mirolg ROM thread you manage a complete suite with standard TG exe... can you share this package, i think system files (*.exe) is universal...
And maybe you win when i will finish my new SKIN in test !
other nice idea to still use this tool always updated... i forgot it ! but it s very usefull !
lesscro said:
Actually i tried this files to perform a speed test :
Code:
Explorer.exe 268 288 2010-09-07 20:46 -a---
GoogleMaps.exe 2 354 176 2010-06-25 19:47 -a---
iGo.exe 8 613 376 2011-05-08 09:07 -a---
iManager.exe 356 352 2011-02-21 17:00 -a---
InfoTrafic Trains.exe 319 488 2010-03-17 23:28 -a---
Opera10-armv4i.exe 623 400 2010-03-15 20:48 -a---
PhotoManPro.exe 1 108 480 2010-04-26 16:29 -a---
player.exe 1 894 672 2009-09-17 11:14 -a---
RCapture.exe 90 112 2010-04-26 16:28 -a---
s2p.exe 364 032 2010-07-17 05:35 -a---
SKTools.exe 2 657 208 2010-03-02 15:09 -a---
ThemeChanger.exe 113 912 2007-04-18 07:43 -a---
Weather-iGo-PPC-v2.exe 264 192 2010-11-11 17:49 -a---
And only a few Exe seems to be work like my old test with my P535 long time ago...
it appear too somes EXE manage in .NEt/win32 are not supported actually !
But when it work... it s soooooooooooooooooo fastest !
Working file compression :
Code:
Explorer.exe 252 928 2010-09-07 20:46 -a---
GoogleMaps.exe 855 040 2010-06-25 19:47 -a---
iManager.exe 203 264 2011-02-21 17:00 -a---
Opera10-armv4i.exe 246 568 2010-03-15 20:48 -a---
PhotoManPro.exe 499 712 2010-04-26 16:29 -a---
RCapture.exe 53 248 2010-04-26 16:28 -a---
then i have seen on Mirolg ROM thread you manage a complete suite with standard TG exe... can you share this package, i think system files (*.exe) is universal...
And maybe you win when i will finish my new SKIN in test !
other nice idea to still use this tool always updated... i forgot it ! but it s very usefull !
Click to expand...
Click to collapse
I know .NET files cannot be UPXed, i did say this in my other post about UPX, but this thread was to state what i found could be done on mirom8.
Also bear in mind small files less than maybe 100k are not worth doing, because there is a latency involved with getting file off of storage, and then decompressing it in ram.
Also there is a downside, files UPXed have a higher memory footprint than files not compressed, so for instance, Imanager.exe that would always be resident in ram, is a bad idea, it will use more ram than it would if it wasn't compressed, its more useful for the config program, as that would make it open faster, any program that starts with the phone, don't bother UPXing it, any program thats always running, don't UPX it.
Glad you found the post useful.
then i will try to tes Mirom8 Rom and extract desired system files compresses based on your discussion with Mirolg...
thx for all !
lesscro said:
then i will try to tes Mirom8 Rom and extract desired system files compresses based on your discussion with Mirolg...
thx for all !
Click to expand...
Click to collapse
Mirom8 cfc is a beta and is not protected, so you can either ask Miro for what can and can't be done, or dump it in a kitchen and use it as a guide as to what to UPX and with what settings, not everything can be -Brute compessed, especially dll files, i believe there is a -Best switch, you will have to do a lot of trial and error, Miro has made leaps and bounds in finding what can be done, if you find things he has missed, let him know, anything to get a smaller, faster rom. My favourite example is the camera application, it fles, it opens in about 3 seconds.
Adding to your list of what you found could be done, i can tell you both of the largest files in the Opera application can be UPXed, so thats the exe and the largest dll file, in flashlite the largest dll can also be UPXed.
i will tru one by one Windows Exe files...
but i repeat i m pretty sure system base system is same on all rom especially system configuration files...
and All TG tool file, like camera and more other indispensable application...
You know what i m looking for... that's why i ask you to send me a package with already identified files system 'UPX transformed'...
I will won lot of time...
indeed, i will report your technic 'UPX' to a French Rom Cooker...
Actually my ROM, maybe it will try it himself too... and more tester more report !
have fun !
lesscro said:
i will tru one by one Windows Exe files...
but i repeat i m pretty sure system base system is same on all rom especially system configuration files...
and All TG tool file, like camera and more other indispensable application...
You know what i m looking for... that's why i ask you to send me a package with already identified files system 'UPX transformed'...
I will won lot of time...
indeed, i will report your technic 'UPX' to a French Rom Cooker...
Actually my ROM, maybe it will try it himself too... and more tester more report !
have fun !
Click to expand...
Click to collapse
Some components cannot be UPXed, cprog.exe, unless you first compress the png files, and gwes.exe, will eat more memory than ever if upxed, however if you found a technique to make it work postively on memory usage, please share how you did it.
I read some of this thread, maybe useful to you.
Only files i have are the ones listed in the first post, better off looking at miros roms as he has a lot of files upxed there.
Damn ! Beautifull... all your idea is awesome !
All your link is so cool...
i will try to test some new files... another tester is in place now... we will report all files we can test here !
Thx...
Related
What is this file ???
Is it aku 3.3 in english or russian????
-----------------------
MODERATOR EDIT:
Changed the subject, use this as the main thread to discuss HTC 2.05 WWE AKU 3.3 ROM issues.
Download and ROM details in the wiki:
http://wiki.xda-developers.com/index.php?pagename=Hermes_Upgrades
BE SURE TO READ THE WIKI BEFORE ASKING QUESTIONS PLEASE
English.
If you need Russian one look for it in Upload/Temp directory
It is in English, I just flashed it using 1.10mfg method
An still no MUI I presume ?
What Where Who Its New The Wwe Version?
Plz Give The Link And Why Dont You Say Anything?
bubu23 said:
What Where Who Its New The Wwe Version?
Plz Give The Link And Why Dont You Say Anything?
Click to expand...
Click to collapse
ftp://ftp.xda-developers.com/Uploads/Hermes/
Here you have.
BuT WhY TyPinG LikE ThAt ! Are YoU PaRkInsOn ?
Is it from HTC or cingular? I've heard both are about to release an AKU 3.3 rom
loll no im not parkinston maybe my keboard is
but why the file finish with test?
its not good its just test???
plz tell me im horny on that fileee(rom)
English confirmed.
Just flushed OS and Ext rom.
ROM Version:2.05.255.1
ROM Date: 01/17/07
ExtROM Version: 2.05.255.101
OS 5.1.461 (build 15671.3.3.0)
I can't get to FTP :/ what user and password ?
reg
Sar
Username = xda
Password = xda
You get to the ftp by this link no pass needed...
ftp://xda:[email protected]/Hermes/
Cheers
Seems very fast...
cant see where it says TEST though. My of been a test for the carriers, so likely the RTM for the carriers to then implement.
So far, so good... at least this one is english, unlike the Swedish update i did yesterday
in the ftp thers only the swedish rom i cant find the wwe rom??
RUU_Hermes_2.05.263.1_6275_1.34.00.10_108_SVE_HTCEUR.exe is now up... Guessing it is an official release from htc?
Cheers
Well i flashed it also.
the Test is included in the splash.bmp. =)
really... But imho the swedish one is a bit faster, but that is just subjectiv i didnt tried to test it with a benchmark program.
btw. any update from cingular?
Somethings fishy about this.
This is supposed to be AKU 3.3, but theres no WPA2 support???
not sure this is a complete ROM?
Does this mean MS Voice Command over bluetooth is fixed without the hacks?
I have detect problems with this rom! Is better wait for another!
Problems detect:
- No wpa2
- Phone connection always breaking down
- Can't go to my contacts!
- ActiveSync always says pocket is out of memory (yes I reset several times)
Flash Using Lnb
This is my first time flashing using lnb and i followed the wiki about using mtty..i just wanted an experimented persons to examine the output of mtty and tell me if the update was perfectly done and if there was no problems in nand please...
Thanks
Iverson....
USB>
USB>task 32
Level = 0
USB>task 28
Storage format start
Write Nand Success
dwBlockToWrite = 13
Storage start block: 461
Storage Total block: 475
Total Bad Block in CE: 0
NeedToEraseBlockStart: 474
Storage format success
USB>lnb 06_OS.nb
:F=06_OS.nb
:A=501A0000
:O=00000000
:L=FFFFFFFF
start NB image downloadSH
Load ADDR: 501A0000 Length: 3C0C800
*********************************************
*********************************************
*********************************************
*********************************************
*********************************************
*********LAST BLOCK, dwBytesCollected=0x13800
Code entry point at 0x53DA0000
Write Nand Success
USB>task 28
Storage format start
Write Nand Success
dwBlockToWrite = 13
Storage start block: 473
Storage Total block: 463
Total Bad Block in CE: 0
NeedToEraseBlockStart: 486
Storage format success
USB>
What is the Page Pool?
Quoting mamaich:
The large the paging pool is, the more code pages can present in RAM simultaneously -> the less loads from slow ROM -> the faster the device is.
Quoting Mike Calligaro from Windows Mobile Team Blog:
In general terms, there are a number RAM uses that aren't reported in the Memory CPL.
For instance, an OEM might set aside a fair amount of RAM to allow the camera to write directly into it without using the CPU (a DMA buffer). Typically the way they do that is to tell the system there is less than the available RAM and then use the RAM the system doesn't know about.
Another example is something called the "page pool." This is similar to the "page file" on a desktop system, and is heavilly used on NAND systems. The page pool isn't something a user can configure, so there's not much point in showing it in the Memory CPL.
Code that can't be shut down or paged out, especially drivers, isn't shown on that Memory CPL either. There's no reason to show you the RAM the display driver is using, because you can't do anything about it.
Applications use RAM in two ways. There is code that runs, and there is data that is created while it is running. On a NOR device, the code can run directly from the ROM and not be loaded into RAM first. This process is called XIP (eXecute In Place). NAND devices can't XIP, so their code is loaded into RAM and executed from there. If you don't have a Page Pool, this code is loaded into normal RAM. The Page Pool is a mechanism to limit how much code is loaded into normal RAM. With a Page Pool, we can unload code that hasn't been used in a while and reload it later if we need to. We can't do that without a Page Pool.
More information:
http://forum.xda-developers.com/showthread.php?t=276630
http://blogs.msdn.com/windowsmobile/archive/2005/11/17/494177.aspx
http://blogs.msdn.com/windowsmobile/archive/2005/08/19/453784.aspx
How to find the page pool offset on Hermes ROMs?
Take the Hex editor of your choice (I use radare) and open OS.nb.
Search the ASCI string "NKKD8", you'll find it two times, take the one more near to the end of the file.
Scroll down a bit until you find the HEXA string "64 00 00 00", page pool should be around there.
If anyone knows a better method please explain
Example using OS.nb from WM6 ROM for hermes
We find the second "NKKD8" at offset 0x3B6C90:
Code:
offset 0 1 2 3 4 5 6 7 8 9 A B C D E F 0123456789ABCDEF
----------------+----------------------------------------+----------------.
00000000003B6C90 4E4B 4B44 3800 0000 0000 0005 0000 0000 |NKKD8...........|
00000000003B6CA0 00C8 FFFF 8039 2D80 4060 2880 B069 2880 |[email protected]`(..i(.|
00000000003B6CB0 B869 2880 AC69 2880 144A 0180 805D 2D80 |.i(..i(..J...]-.|
00000000003B6CC0 602A 2D80 1829 0080 0000 0000 0000 0000 |`*-..)..........|
Further down we can find this:
Code:
offset 0 1 2 3 4 5 6 7 8 9 A B C D E F 0123456789ABCDEF
----------------+----------------------------------------+----------------.
00000000003B6F40 581D 0480 C41C 0480 BC1C 0480 00B0 0400 |X...............|
00000000003B6F50 FFFF FFFF FFFF FFFF 9B4F FFFF 64B0 0000 |.........O..d...|
[COLOR="Sienna"]00000000003B6F60 6400 0000 A85B 0480 0000 [b]6000[/b] 0000 0000 |d....[....`.....|[/COLOR]
00000000003B6F70 38E8 0180 C43C 0280 DC6B 0180 00B0 0380 |8....<...k......|
00000000003B6F80 882D 0280 FC2C 0280 D488 0480 5489 0480 |.-...,......T...|
So, we can see the page pool is at offset 0x3B6F6B and is set to 6 Mb.
The pagepool value is specified in HEXA, so for example:
Change "6000" to "4000" for a 4Mb paging pool
Change "6000" to "8000" for a 8Mb paging pool
Change "6000" to "C000" for a 12Mb paging pool
Change "6000" to "F000" for a 16Mb paging pool
Finding the best page pool size
The page pool size is a compromise between memory available to programs and speed. On WM6 ROM:
With 4Mb page pool we get a Total Program Memory available of 50.68 Mb, and SPB Benchmark index 347.88.
With 6Mb page pool we get a Total Program Memory available of 48.68 Mb, and SPB Benchmark index 346.03.
With 8Mb page pool we get a Total Program Memory available of 46.67 Mb, and SPB Benchmark index 345.95.
With 12Mb page pool we get a Total Program Memory available of 42.66 Mb, and SPB Benchmark index 348.53.
With 16Mb page pool we get a Total Program Memory available of 39.64 Mb, and SPB Benchmark index 351.26.
I attach the SPB benchmark results, so anyone can take its conclusions
Does it works for WM5 too ?
yes, it's exactly the same in wm5
Suprising results. Smallest and largest pool size were the fastest. It was obviously an increase as the size increased.
No point IMO changing it to anything other than 16mb or 4mb. Its stange how the least amount of memory allows for the second fastest speed.
When I run some programs then after I have killed the running program some other programs fail to start till I do a soft reset. One example is Opera 8.60. after running and killing Opera 8.60 I cannot start Netfront 3.3 and Softmaker's Planmaker.
I don't know if this is even relevant but Just wanted to know if I experiment with different page pool sizes do you think my problem can be solved.?
Thanks and Regards
this is great n all but just curious on how you would know that that is the exact thing that you need to change. theres books on this kinda stuff.. ?!?!
-ps thanx for da info
pof said:
yes, it's exactly the same in wm5
Click to expand...
Click to collapse
on AKU 3.3 I had lot's of out of driver memory errors with the widcomm BT drivers. So I guess it may have a smaller page pool then the WM6 os has.
Too bad I can't bench the BT performance. But maybe I will try a bigger pagepool when the new WM6 rom comes out, see if it feels more responsive and faster then.
Thanx for this great info!
but hermes have 64M rams, where are other rams
pof said:
yes, it's exactly the same in wm5
Click to expand...
Click to collapse
Tried it on the new 2.11 (AKU3.3) rom.
And it works mutch better now when using the widcomm stack and browsing the net.
Although I'm not sure what speeds it up, could be the pagepool or this rom, but it works great, mutch faster then the WM6 roms.
we can use pocket machanic to change the 3 cache settings
dont forget you have auto mode as well, i wonder how will that check out
qooqoo said:
but hermes have 64M rams, where are other rams
Click to expand...
Click to collapse
other rams maybe to Extended_ROM??
how to change Extended_ROM size????
philpwong said:
other rams maybe to Extended_ROM??
how to change Extended_ROM size????
Click to expand...
Click to collapse
no, Extended_ROM is part of NAND flash, it's different from RAM.
qooqoo said:
but hermes have 64M rams, where are other rams
Click to expand...
Click to collapse
Just think about the PC's windows OS, something like that
clfk said:
Just think about the PC's windows OS, something like that
Click to expand...
Click to collapse
but i don't think so.
hard disk and flash memory are use 1k=1000b instead of 1024b, but ram don't like that. 64M of RAM do = 64 * 1024 *1024 b.
My x01ht is
qooqoo said:
but i don't think so.
hard disk and flash memory are use 1k=1000b instead of 1024b, but ram don't like that. 64M of RAM do = 64 * 1024 *1024 b.
Click to expand...
Click to collapse
My x01ht is
ExtROM=10M+Pagepool=6M+Program Memory=48M
page pool
Hello i have try a benchmark with VsBenchmark 2006 on my Hermes tytn herm 200 and wm6 LVSW 3.24.0.4 radio 1.40 + 4mo page pool.
(original is 6mo)
the result score bench is 982
test 1:950
test 2:900
test 3:1263
test 4:654
test 5:1146
dluisao said:
Hello i have try a benchmark with VsBenchmark 2006 on my Hermes tytn herm 200 and wm6 LVSW 3.24.0.4 radio 1.40 + 4mo page pool.
(original is 6mo)
the result score bench is 982
test 1:950
test 2:900
test 3:1263
test 4:654
test 5:1146
Click to expand...
Click to collapse
I am writing because I have been trying to change the pagepool of LVSW 08-04 to 4mb but I can´t locate the offset location on this ROM when searching with my HEX editor. I followed the instructions posted in this forum but.... Any help would be fantastic. Thanks!
update.... sorted!!!
christonge said:
I am writing because I have been trying to change the pagepool of LVSW 08-04 to 4mb but I can´t locate the offset location on this ROM when searching with my HEX editor. I followed the instructions posted in this forum but.... Any help would be fantastic. Thanks!
update.... sorted!!!
Click to expand...
Click to collapse
you must find the string in ascy NKKD8 at the rom in NB format and don't in NBH format.
convert your rom NBH in NB format and use ultraedit for find this string NKKD8 or in ultraedit(editor hexa) go to
at 00000000003B6C90 line the NKKD8 is there and at the line 00000000003B6F60 you must change 6000 to 4000.
(00000000003B6F60 6400 0000 A85B 0480 0000 6000 0000 0000)
4000
and at finish reconvert the NB rom to NBH with the 4000 modified.
it seems that new wm6 rom for hermes,
Build: 17741.0.2.1
AKU: 0.2.1
CE: 5.2.1236
offset is different from other roms.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
how can i change new rom's pagepool,plz?
thanks in advance
@qooqoo: read the first post again, follow the three easy steps under "How to find the page pool offset on Hermes ROMs?".
wen i try 2 acc4ss d windows folder either frm default explorer or resco explorer....it takes more dan 2-3 mins 2 open......plz help mods......it feels very irritating........ma current ROM is Sumit's Blackmetal.......
Try using Advanced Config tool from Sumit's Manilla Vanilla post and modify:
File System Cache : 256 KB
and Glyph Cache: 32 KB Under Performance.
This should give you good speed.
hw 2 modify dis cache........???
Use this: http://forum.xda-developers.com/attachment.php?attachmentid=151013&d=1233099165
After installing the cab, you will find an option called Performance, Click on the plus sign, there you will find the cache.
android read and writes SDcard really slow
even if you got a class ten card this really sucks
why is it so slow?
the cache size to read and write the sdcard is set to 128kB on our devices
can we change this?
yes we can
this speeds are with my sdcard they are different for each sdcard
128 kB: <- original
Write: 4.1 MB/s
Read: 11.2 MB/s
1024 kB: <- sucks
Write: 4.5 MB/s
Read: 18.1 MB/s
2048 kB: <- write spees is best
Write: 6.7 MB/s
Read: 35.5 MB/s
3072 kB: <- overal speed is best
Write: 6.4 MB/s
Read: 45.8 MB/s
4096 kB: <- also sucks
Write: 3.9 MB/s
Read: 35.3 MB/s
now you got proof you can make this better
Code:
open root explorer
go to "/system/etc" and open "hw_config.sh" with Text Editor
paste this line at the end
"# sd speed fix
echo 3072 > /sys/devices/virtual/bdi/179:0/read_ahead_kb"
without quotes ofc
quick note i will include this in my rom updates
Enjoy!
Nice find owain :thumb: A bit off topic, if i'm not mistaken, there's some pieces inside there about LED keyboard with light sensor automation. Can it be implemented directly into you ROM? CMIIW but If i'm outdated about your CM progress then i'd say sorry i'll try 3.6 now..
hansip87 said:
Nice find owain :thumb: A bit off topic, if i'm not mistaken, there's some pieces inside there about LED keyboard with light sensor automation. Can it be implemented directly into you ROM? CMIIW but If i'm outdated about your CM progress then i'd say sorry i'll try 3.6 now..
Click to expand...
Click to collapse
nah i've tried that
Thank you!, brainmasters methods were no good for me.
The improvements are fantastic, absurd this isnt this stocked.
Whats the default permissions for "hw_config.sh"?
owain94 said:
android read and writes SDcard really slow
even if you got a class ten card this really sucks
why is it so slow?
the cache size to read and write the sdcard is set to 128kB on our devices
can we change this?
yes we can
this speeds are with my sdcard they are different for each sdcard
128 kB: <- original
Write: 4.1 MB/s
Read: 11.2 MB/s
1024 kB: <- sucks
Write: 4.5 MB/s
Read: 18.1 MB/s
2048 kB: <- write spees is best
Write: 6.7 MB/s
Read: 35.5 MB/s
3072 kB: <- overal speed is best
Write: 6.4 MB/s
Read: 45.8 MB/s
4096 kB: <- also sucks
Write: 3.9 MB/s
Read: 35.3 MB/s
now you got proof you can make this better
Code:
open root explorer
go to "/system/etc" and open "hw_config.sh" with Text Editor
paste this line at the end
"# sd speed fix
echo 3072 > /sys/devices/virtual/bdi/179:0/read_ahead_kb"
without quotes ofc
quick note i will include this in my rom updates
Enjoy!
Click to expand...
Click to collapse
Nice Owain!
Waiting for updates in your ROM!
dj_deez said:
Whats the default permissions for "hw_config.sh"?
Click to expand...
Click to collapse
#ls -l /system/etc/hw_config.sh
-rw-r--r-- 1 root root 1443 Aug 1 2008 /system/etc/hw_config.sh
miniCM RC1??
death_syndrome said:
miniCM RC1??
Click to expand...
Click to collapse
Maby you can make a sentence then we might understand you -.-''
Sent from my U20i using XDA Premium App
owain94 said:
Maby you can make a sentence then we might understand you -.-''
Sent from my U20i using XDA Premium App
Click to expand...
Click to collapse
.
.
lol . . does t work with miniCM RC1 rom?
5.0 Write
43.4 Read!!!!
I SD is on steroids!!!
Thanx!!!
death_syndrome said:
.
.
lol . . does t work with miniCM RC1 rom?
Click to expand...
Click to collapse
it works on every rom stock cyanogen sense
do u know same app to test it?
I used SD tools
Sent from my U20i using XDA Premium App
@Owain----on my class 4,8GB Transcend card-----I get write/read speeds of 11 and 16 by default.Upon tweaking as suggested by you,the reading speed goes up to around 21 whereas the writing speed actually drops(just a wee bit though)!
sathkartha said:
@Owain----on my class 4,8GB Transcend card-----I get write/read speeds of 11 and 16 by default.Upon tweaking as suggested by you,the reading speed goes up to around 21 whereas the writing speed actually drops(just a wee bit though)!
Click to expand...
Click to collapse
I posted 3 different cach sizes in first post by my tests try the other sizes by just replacing the 3072 with another it may work better with your sd card
Sent from my U20i using XDA Premium App
@Owain--thanks.
for the records--the best results for me were at 4096 when the speeds went to 10.9/38.4 compared to 12.4/15.9 for default.
thanks owain, at last found the simple configuration.
tried on my x10 mini, got improvement.
thx owain
huge improvement on stock 2gb memory card
Hey, I tried all the different values 1024, 2048, 3072 etc.
I get worse results in write and marginally better in read.
I am not sure, it is technically possible...
But at the moment, all ROMs need SD card for installation and usage...
But I think, we can make it also possible to use our Waves without SD...
Maybe we can modify waves, that Odin or Flash tool can hanlde our Waves...
Let's collect some thoughts...!
For Odin you need change Bootloader...
If you don't know what Bootloader is... check this out:
http://forum.xda-developers.com/showthread.php?t=897468
SBL from I9000 for instance... (but then other problems or more problems)
I have tested few SBL with Odin and S8500... long time ago...
But you need JTAG or this:
http://forum.xda-developers.com/showthread.php?t=1250270
NOT support nor finished solution... only "concept"...
WARNING! You can Hardbrick your S8500...
Same warning for S8530...
About ""Flash tool""...
If FOTA would support read from 512 MB OneNAND... you could copy zImage + ROM to this memory instead 2 GB moviNAND...
With Multiloader... and WaveRemaker created files...
And don't forget since bada 2... Samsung kastrierte moviNAND to same size...
but S8500 have more memory as S8530...
If I remember correct... 90 MB unused in S8500...
Unsere Boo.L Experten sind alle... gelangweilt... haben "wichtigeres" zu tun...
Somit wird es auch extrem schwer... FOTA neue Kunststücke beizubringen...
Weil dazu mußt Du den Bootloader BL3 halbwegs begriffen haben...
Siehe das:
http://forum.xda-developers.com/samsung-tizen/bada-android/fboot-fota-noob-edition-t2821156
Best Regards
Edit 1.
Here ""evidence""/example for Odin with S8500:
http://forum.xda-developers.com/showpost.php?p=44993772&postcount=49
Somewhere else I have tested more SBLs from different Android handsets with similar CPU...
adfree said:
About ""Flash tool""...
If FOTA would support read from 512 MB OneNAND... you could copy zImage + ROM to this memory instead 2 GB moviNAND...
With Multiloader... and WaveRemaker created files...
Click to expand...
Click to collapse
Could you please explain step by step how to prepare this file for multiloader with zImage(boot.img) inside?
Rebellos, many time ago said that it possible to read RAW(binary) file from OneNAND via FOTA
Could you please explain step by step how to prepare this file for multiloader with zImage(boot.img) inside?
Click to expand...
Click to collapse
Also long time ago.
Will try to find my old posts.
For S8500/S8530 it should be easy to use .PFS file... made via WaveRemaker...
But PFS writes into 2 GB moviNAND not 512 MB OneNAND...
For OneNAND we could add zImage to RC1 or RC2 or replace instead...
Need to check reserved space for partitions...
http://forum.xda-developers.com/showpost.php?p=37698018&postcount=10
Hmmm.
If bada not needed... then we could create template RC1...
with 10 MB or something in this direction... and place zImage easily with WinHex into same address...
So in FOTA it could read from correct address in OneNAND...
Best Regards
Edit 1.
For zImage on 2 GB moviNAND with PFS inside...
http://forum.xda-developers.com/showpost.php?p=50254876&postcount=408
This Folder for instance:
Code:
Media/zImage
And for RC1 template I need little bit time... 1 or 2 days... maybe faster...
in last ROMs MoviNAND is fully "busy" by android.
Correct way is OneNand
Also it will be good to save bada offline charger, i don't remember correctly where is located images for it in RC1 or RC2, like bada splash screen and pictured with for offline charger
Correct way is OneNand
Click to expand...
Click to collapse
Okidoki, then we should take RC1 for tests...
S8500 flash and load from:
Code:
0x1980 0000
I need few minutes, because I must charge my test S8500... before I flash something...
Also it will be good to save bada offline charger, i don't remember correctly where is located images for it in RC1 or RC2, like bada splash screen and pictured with for offline charger
Click to expand...
Click to collapse
RC2 have 1 Picture as Placeholder for Charger... before apps_compressed.bin is fully loaded and executed...
If someone is able to create FOTA with loading from address 0x1980 0000 zImage... then we could use every other address...
But before for faster tests should 10 MB empty RC1 be enough...
Ehm... 16 MB... easier length...
QMD Header is not mandatory... if I remember correct...
MD5 Check for Multiloader disabled...
File without zImage yet, but easy replace 0x0 with content of zImage...
Result should be... zImage at address 0x1980 0000 in OneNAND...
File not tested yet, because charging battery...
Later I will try to flash self... then report if my S8500 explode...
Best Regards
@volk204 can you tell me what we'll gain changing location of boot.img(zImage)? Only place or something about performance or etc.?
hero355 said:
@volk204 can you tell me what we'll gain changing location of boot.img(zImage)? Only place or something about performance or etc.?
Click to expand...
Click to collapse
Nothing for performance.
We just remove dependency from SD card
Code:
Download Start Ch[0]
Rsrc1 16777.2KB OK[6.7s]
All files complete[7.4s]
7 seconds for 16 MB RC1...
...
Testresult via PM...
Best Regards
Edit 1.
Short logged via UART...
Code:
Hangs at:
[PAM: ] OneNAND physical base address : 0xb0000000
[PAM: ] OneNAND virtual base address : 0xb0000000
[PAM: ] FSR_PAM_InitNANDController Success!!
[PAM: ] --FSR_PAM_Init
bl3_info_block 1 age = 1
bl3_info_block 2 not found, BL3_1 Loading
+-------------------------------+
| Bootloader Shadowing FINISHED |
+-------------------------------+
Launch Image at 0x42080000
[BOOT_V1.0 (Jan 5 2012, 19:08:14)]
SelectBootingMode: H/W...0xe.
[BOOT] ARMCLK: 400000 KHz, MSYSHCLK 200000 KHz,MSYSPCLK: 100000 KHz, [BOOT] DSYSHCLK 166750 KHz,DSYSPCLK: 83375 KHz,PSYSHCLK: 133400 KHz, PSYSPCLK: 66700 KHz,SYSCON_A2M: 200000 KHz
+++FIMD_Drv_INITIALIZE
FIMD_Drv_ChangeMode: MDNIE_MODE
Frame Rate:62 SCLK_FIMD:133400 kHz ClkDiv:4
S6E63M0 : LDI_Pentile_Set_Change Pentile_Value =6c
---FIMD_Drv_INITIALIZE
---FIMD_Drv_SetWinOnOff(WIN4:1)
LCD initialize Finished
Flash_Unlock failed
Poweron status - 20
FSA9480 0x03 Register = 1
FSA9480 0x0A Register = 0
FSA9480 0x0B Register = 8
FSA9480 0x07 Register = 1c
[B]SelectBootingMode: Boot Mode = 1...[/B]
Info from Rebellos:
<Rebellos> To read from oneNAND you need Flash_Read_Data
<Rebellos> protype of it would be...
<Rebellos> uint32_t Flash_Read_Data(void* buf, uint32_t addr, uint32_t size);
<Rebellos> It should be easily detectable by mijoma's tool
<Rebellos> (you can change void* to char* or any pointer type that suits you)
<Volk204> Rebellos, do i need something like OneNand_Init before?
<Rebellos> No. Onenand is already initialized by BL3 (since FOTA is being read from there)
I will try to add it in FOTA later, or maybe someone else
edit1:
8500:
unsigned long c_Flash_Read_Data[] = { 0xf5bd478f, 0 };
8530:
unsigned long c_Flash_Read_Data[] = { 0x7dc705fa, 0 };
I can test tomorrow morning
I am not sure that it is correct but
Modified Fota.c http://pastebin.com/S62SMwnH
Test FOTA: https://yadi.sk/d/OMvi6FwMa2HgL i tried to test, but kernel don't start, output is ok, probably i incorrect flashed rc1 with boot.img.
Maybe someone will play with it
P.S. modified rc1 kill bada offline charger
volk204 said:
I am not sure that it is correct but
Modified Fota.c http://pastebin.com/S62SMwnH
Test FOTA: https://yadi.sk/d/OMvi6FwMa2HgL i tried to test, but kernel don't start, output is ok, probably i incorrect flashed rc1 with boot.img.
Maybe someone will play with it
P.S. modified rc1 kill bada offline charger
Click to expand...
Click to collapse
Maybe some parts of charger located there, maybe we don't need to fully empty rc1 with boot.img, just deleting unneeded ui parts can work.
In RC1 can only be few Pics...
But they are for apps_compressed.bin... So only animation is not visible...
Charging at all should work...
Later we can add Pics and copy zImage to other position...
But priority is to start zImage from address 0x1980 0000...
Need few minutes for test... then report...
Best Regards
Edit 1.
First attempt...
I see on Display...
Mounted partitions
Copied kernel from onennand
Init Modem
Boot in Normal Mode
Copied kernel to boot
Wait!
I have taken older zImage... Now I am waiting...
Will test with empty RC1 and RC1 with zImage inside... to check if textoutput is static or really detected zImage...
Later more...
Edit 2.
Textoutput is static but work. :good:
Same result with my attached empty RC1.
@volk204
Your FOTA work. :good:
Congratulation!
Zimage content should be at 0x0...
Header QMD must be overwritten...
Now we could change location of zImage... for instance... take full RC1 and add zImage at end of content, before last 1024 Byte
Depend on size of RC1...
But then 0x1980 0000 is wrong/obsolete...
Also we could create minor/small/slim RC1... with only few MB for charger Animation... if somebody need...
If someone need, I could upload my RC1 template... but for now I am tooo lazy to attach 6 MB... upload...
Edit 3.
If I remember correct... smallest RC1 was 7 MB or less 10 MB... to start bada...
My tests long time ago...
With FOTA from volk204 this RC1 boot zImage from 512 MB OneNAND.
:good:
Inside older kernel, not latest... only for test...
You can compare textoutput between this RC1 and my empty template:
http://forum.xda-developers.com/showpost.php?p=54892518&postcount=6
Same text, but this one really load and execute zImage from OneNAND.
Thanx.
Best Regards
adfree said:
Textoutput is static but work. :good:
Same result with my attached empty RC1.
@volk204
Your FOTA work. :good:
Congratulation!
Zimage content should be at 0x0...
Header QMD must be overwritten...
Now we could change location of zImage... for instance... take full RC1 and add zImage at end of content, before last 1024 Byte
Depend on size of RC1...
But then 0x1980 0000 is wrong/obsolete...
Also we could create minor/small/slim RC1... with only few MB for charger Animation... if somebody need...
Click to expand...
Click to collapse
I will replace textoutput by bada splash (Sasung Wave picture) from rc2
so if FOTA stuck on this picture, need to reflash kernel, we can't do anything more for debug
0x1980 0000 is easy to change, no problem, and probably it is different for wave and wave 2
I will push sources to github later, maybe today
and about charger, i see only picture from rc2 and then reboot in 5-10 sec, so it not work at all
Okidoki.
Later I could do some tests with zImage at end of RC1...
So RC1 is original + zImage...
Need some time...
Best Regards
i pushed sources on github:
https://github.com/Badadroid/android_bootable_wave-fboot/tree/android_onenand
TextOuput replaced by Bada splash screen,
If FOTA stuck on this picture more than 15-20 sec - something wrong with kernel
OneNand address for 8500: https://github.com/Badadroid/androi.../blob/android_onenand/FBOOT_S8500_b2x.lds#L28
OneNand address for 8530: https://github.com/Badadroid/androi.../blob/android_onenand/FBOOT_S8530_b2x.lds#L28
looks like rsrc1 address for 8530 is 0x36000000
Compiled FOTA for wave and wave 2 is in attachments
Edit:
for offline charger we need PshRsrcBmpIdleAni.rbm from rc1
rc1 with only PshRsrcBmpIdleAni.rbm attached, flashed and charger work
now need to create template with this file inside and boot.img inside and count memory address for boot.img, but i don't know how to do it
First look at charger.rc1...
Content ends at 0x0014 0000
We could make this area bigger... 0x0020 0000
And add zImage at this address...
So for FOTA new entry point is 0x1980 0000 + 0x0020 0000 =
19A0 0000
Theory... I can make tests only later...
Need to do some other things now...
Best Regards
adfree said:
First look at charger.rc1...
Content ends at 0x0014 0000
We could make this area bigger... 0x0020 0000
And add zImage at this address...
So for FOTA new entry point is 0x1980 0000 + 0x0020 0000 =
19A0 0000
Theory... I can make tests only later...
Need to do some other things now...
Best Regards
Click to expand...
Click to collapse
tested and working
fota with start in 0x19A00000 is in attachments
offline charger also work
Edit
Attached rsrc template for 8500 with charger inside, just add boot.img in 0x00200000
@adfree, could you please create empty rsrc1 template ~16mb for wave 2 (8530)?
And could you please re-check onenand adress for wave2?
in multiloader i see 0x36000000, but Rebellos marked it as 0x03600000 in kernel sources https://github.com/Badadroid/androi...i_nand/drivers/mtd/onenand/samsung_wave.h#L47