The capacitors (& filter coil) arrived and thus I was able to work on this again.
I replaced the 4.7µF caps, and the 100µF TL494 power supply cap just as a precaution and bumped the output caps to 680µF low ESR (35V still) ones - a random one tested as 0.05ohm in my cheap tester.
The mains caps tested good and i couldnt find any good replacements in the shop I was ordering from so i left them alone - I also tested the 1µF main-side-swingy-cap, that was good too.
I was going to just skip (just remove it and put a big jumper wire under the PCB) the filter coil, but found an 8A 5.6µH filter in the shop, so just so i dont feel so bad about it I installed that instead. I'm sure it has way less filtering effect than the original one, but atleast it is rated high enough amps and is some sort of a filter instead of nothing (also the output caps are higher µF now so maybe that helps a bit too). And also I moved the load resistor further away from the caps so it's not purposefully cooking them.
This is what the output section looks like after these mods:
With these upgrades the power supply (to all the tests i can do right now) seems to operate correctly - can run my load (smallish really, 1 to 2A ish) just fine and doesn't buzz like it used to even on idle.
I also installed a slow fan on it for a longer & cooler life of everything.
(And yeah I know it's wild, this post makes 2020 the first year with more than a single post. OTOH i skipped 2019 so it balances out, don't worry... yet atleast :P)
Ramblings
perjantai 24. heinäkuuta 2020
lauantai 11. heinäkuuta 2020
FDPS-100N-R5 24V 5A power supply schematic
Heyo, this is a schematic I've drawn of this power supply while working to understand it - it might not be all correct, YMMV, no warranty, etc.
It's not a particularly good power supply - the L2 filter/coil is seriously under-rated (and/or insufficiently cooled, however you want to look at it) and this (plus the idle load resistor) cooked the output capacitors in mine - I'm waiting on new caps (& a different filter to experiment with) as i write this, so I don't know if mine has more trouble than that.
That said, it does seem very serviceable - simple one-sided PCB with through-hole components makes for super simple parts replacement, and the PCB even has a nice enough silkscreen with named components so you can find what I'm drawing in the schematic - if i bothered to name that particular component that is ... sorry.
This is the AC input all the way to the rectified mains caps. The most boring bit really. Not drawn: the 220/110V switch: it would in the 110V position connect one of the AC lines past the rectifier to the middle point of the caps.
This is the high-voltage side wiggly bits. I'll be honest and say that I don't (yet) really fully understand how this thing starts up (or some of the details of the controls).
This is how I've described the transformers in the PSU. W is named T2 on the PCB and Y is named T1. I don't really know why i decided to name their terminals Wx and Yx but what is done is done.
This is the first part of the low-voltage side logic - the bottom half of the TL494 chip and what it drives. I know my transistor orientation in the control drive section is funky, but sometimes you make some happy accidents.
This is the top half of the TL494-related circuitry - mostly what it senses.
Finally, the DC output section (and the feedback adjustment stuff).
It's not a particularly good power supply - the L2 filter/coil is seriously under-rated (and/or insufficiently cooled, however you want to look at it) and this (plus the idle load resistor) cooked the output capacitors in mine - I'm waiting on new caps (& a different filter to experiment with) as i write this, so I don't know if mine has more trouble than that.
That said, it does seem very serviceable - simple one-sided PCB with through-hole components makes for super simple parts replacement, and the PCB even has a nice enough silkscreen with named components so you can find what I'm drawing in the schematic - if i bothered to name that particular component that is ... sorry.
This is the AC input all the way to the rectified mains caps. The most boring bit really. Not drawn: the 220/110V switch: it would in the 110V position connect one of the AC lines past the rectifier to the middle point of the caps.
This is the high-voltage side wiggly bits. I'll be honest and say that I don't (yet) really fully understand how this thing starts up (or some of the details of the controls).
This is how I've described the transformers in the PSU. W is named T2 on the PCB and Y is named T1. I don't really know why i decided to name their terminals Wx and Yx but what is done is done.
This is the first part of the low-voltage side logic - the bottom half of the TL494 chip and what it drives. I know my transistor orientation in the control drive section is funky, but sometimes you make some happy accidents.
This is the top half of the TL494-related circuitry - mostly what it senses.
Finally, the DC output section (and the feedback adjustment stuff).
torstai 7. kesäkuuta 2018
Inside of an ebay "EDID Connector VGA DVI HDMI Mini DP Displayport Virtual Display Dummy Adapter"
Specifically, the HDMI pass-through variant, that is useful incase you dont want your OS (eg. M$ crud) to notice you for example using a HDMI switcher or a stupid display that "detaches" when it sleeps.
The thing is just a small 24C02S TSOT-23-5 256-byte EEPROM wired up to the computer side of the thing, with the Hotplug Detect pin pulled up to 5V with a resistor AND connected to the WP pin of the EEPROM.
It presents the EDID of an oddball 4k TV, so in case that doesnt fit your use case (eg. not a 4k device, or not audio capable, etc), you can write that EEPROM with
your EEPROM writer of choice just by getting a female (i think, like in a GPU) HDMI connector (or splice an extension cable) and wiring HPD to ground and the rest 4 (GND,5V,SDA,SCL) to your programmer.
I suppose if you have a GPU with working I2C drivers in linux, you can also just get an extension cable and ground the HPD inside it and flash with that.
Other generic notes: this will break HDCP since nothing is connected to the display side DDC I2C pins, and yeah you plug the thing into your GPU (ignore the source and sink texts, they just confused me).
The thing is just a small 24C02S TSOT-23-5 256-byte EEPROM wired up to the computer side of the thing, with the Hotplug Detect pin pulled up to 5V with a resistor AND connected to the WP pin of the EEPROM.
It presents the EDID of an oddball 4k TV, so in case that doesnt fit your use case (eg. not a 4k device, or not audio capable, etc), you can write that EEPROM with
your EEPROM writer of choice just by getting a female (i think, like in a GPU) HDMI connector (or splice an extension cable) and wiring HPD to ground and the rest 4 (GND,5V,SDA,SCL) to your programmer.
I suppose if you have a GPU with working I2C drivers in linux, you can also just get an extension cable and ground the HPD inside it and flash with that.
Other generic notes: this will break HDCP since nothing is connected to the display side DDC I2C pins, and yeah you plug the thing into your GPU (ignore the source and sink texts, they just confused me).
keskiviikko 11. tammikuuta 2017
Performance of unxz and xz (on a 25Mhz 486SL) built with gcc 3.4, 4.1, 4.7 and 6 (* -Os/-O2/-O3)
The decompression test was decoding the .tar.xz (742.5K) that contained these binaries to /dev/null. The tar.xz was compressed with xz -2.
Times:
3.4-O2
real 0m 48.50s
user 0m 38.13s
sys 0m 0.94s
3.4-O3
real 0m 42.01s
user 0m 39.27s
sys 0m 0.88s
3.4-Os
real 0m 31.21s
user 0m 29.70s
sys 0m 0.73s
4.1-O2
real 0m 31.93s
user 0m 29.10s
sys 0m 0.77s
4.1-O3
real 0m 30.72s
user 0m 29.30s
sys 0m 0.63s
4.1-Os
real 0m 31.93s
user 0m 30.37s
sys 0m 0.76s
4.7-O2
real 0m 33.08s
user 0m 30.07s
sys 0m 0.69s
4.7-O3
real 0m 31.74s
user 0m 30.03s
sys 0m 0.74s
4.7-Os
real 0m 30.24s
user 0m 28.58s
sys 0m 0.69s
6.3-O2
real 0m 31.77s
user 0m 30.17s
sys 0m 0.69s
6.3-O3
real 0m 31.53s
user 0m 30.14s
sys 0m 0.70s
6.3-Os
real 0m 28.51s
user 0m 27.05s
sys 0m 0.73s
I've highlighted worst and best user time. It seems that -Os is pretty good for decompression (in this odd case)...
The compression test was an xz -0 compression of the uncompressed .tar (2.7M) to /dev/null (the output would be 931.7K).
Times:
3.4-O2
real 2m 14.93s
user 1m 55.39s
sys 0m 3.15s
3.4-O3
real 1m 55.66s
user 1m 46.14s
sys 0m 2.83s
3.4-Os
real 2m 9.30s
user 1m 58.94s
sys 0m 2.45s
4.1-O2
real 1m 55.53s
user 1m 41.88s
sys 0m 2.40s
4.1-O3
real 1m 49.50s
user 1m 37.86s
sys 0m 2.93s
4.1-Os
real 2m 10.07s
user 1m 49.07s
sys 0m 4.22s
4.7-O2
real 1m 54.09s
user 1m 38.77s
sys 0m 3.10s
4.7-O3
real 1m 52.92s
user 1m 42.22s
sys 0m 2.83s
4.7-Os
real 2m 8.43s
user 1m 56.91s
sys 0m 2.82s
6.3-O2
real 1m 48.35s
user 1m 37.14s
sys 0m 2.71s
6.3-O3
real 1m 48.53s
user 1m 37.91s
sys 0m 3.06s
6.3-Os
real 2m 3.00s
user 1m 55.42s
sys 0m 2.94s
Again highlighted worst and best. It seems -Os is pretty bad for compression.
Also of some interest might be the stripped binary size for these builds of xz+liblzma:
232K xz-3.4-O2
244K xz-3.4-O3
204K xz-3.4-Os
220K xz-4.1-O2
240K xz-4.1-O3
196K xz-4.1-Os
256K xz-4.7-O2
288K xz-4.7-O3
212K xz-4.7-Os
264K xz-6.3-O2
304K xz-6.3-O3
212K xz-6.3-Os
Why all of this? Because.
PS. xz -0 compresses faster than gzip -9 and makes a smaller file. OTOH gunzip is about 3x faster (1/3 of the time) than unxz on this hw.
PPS. The 486 has 8M of RAM and runs linux 2.4.36 with uClibc 0.9.30.3 :P
PPPS. The binaries and little helper scripts for running them and the results (and the test files used) are at https://urjaman.ddns.net/xz-486-tests.tar.xz
Times:
3.4-O2
real 0m 48.50s
user 0m 38.13s
sys 0m 0.94s
3.4-O3
real 0m 42.01s
user 0m 39.27s
sys 0m 0.88s
3.4-Os
real 0m 31.21s
user 0m 29.70s
sys 0m 0.73s
4.1-O2
real 0m 31.93s
user 0m 29.10s
sys 0m 0.77s
4.1-O3
real 0m 30.72s
user 0m 29.30s
sys 0m 0.63s
4.1-Os
real 0m 31.93s
user 0m 30.37s
sys 0m 0.76s
4.7-O2
real 0m 33.08s
user 0m 30.07s
sys 0m 0.69s
4.7-O3
real 0m 31.74s
user 0m 30.03s
sys 0m 0.74s
4.7-Os
real 0m 30.24s
user 0m 28.58s
sys 0m 0.69s
6.3-O2
real 0m 31.77s
user 0m 30.17s
sys 0m 0.69s
6.3-O3
real 0m 31.53s
user 0m 30.14s
sys 0m 0.70s
6.3-Os
real 0m 28.51s
user 0m 27.05s
sys 0m 0.73s
I've highlighted worst and best user time. It seems that -Os is pretty good for decompression (in this odd case)...
The compression test was an xz -0 compression of the uncompressed .tar (2.7M) to /dev/null (the output would be 931.7K).
Times:
3.4-O2
real 2m 14.93s
user 1m 55.39s
sys 0m 3.15s
3.4-O3
real 1m 55.66s
user 1m 46.14s
sys 0m 2.83s
3.4-Os
real 2m 9.30s
user 1m 58.94s
sys 0m 2.45s
4.1-O2
real 1m 55.53s
user 1m 41.88s
sys 0m 2.40s
4.1-O3
real 1m 49.50s
user 1m 37.86s
sys 0m 2.93s
4.1-Os
real 2m 10.07s
user 1m 49.07s
sys 0m 4.22s
4.7-O2
real 1m 54.09s
user 1m 38.77s
sys 0m 3.10s
4.7-O3
real 1m 52.92s
user 1m 42.22s
sys 0m 2.83s
4.7-Os
real 2m 8.43s
user 1m 56.91s
sys 0m 2.82s
6.3-O2
real 1m 48.35s
user 1m 37.14s
sys 0m 2.71s
6.3-O3
real 1m 48.53s
user 1m 37.91s
sys 0m 3.06s
6.3-Os
real 2m 3.00s
user 1m 55.42s
sys 0m 2.94s
Again highlighted worst and best. It seems -Os is pretty bad for compression.
Also of some interest might be the stripped binary size for these builds of xz+liblzma:
232K xz-3.4-O2
244K xz-3.4-O3
204K xz-3.4-Os
220K xz-4.1-O2
240K xz-4.1-O3
196K xz-4.1-Os
256K xz-4.7-O2
288K xz-4.7-O3
212K xz-4.7-Os
264K xz-6.3-O2
304K xz-6.3-O3
212K xz-6.3-Os
Why all of this? Because.
PS. xz -0 compresses faster than gzip -9 and makes a smaller file. OTOH gunzip is about 3x faster (1/3 of the time) than unxz on this hw.
PPS. The 486 has 8M of RAM and runs linux 2.4.36 with uClibc 0.9.30.3 :P
PPPS. The binaries and little helper scripts for running them and the results (and the test files used) are at https://urjaman.ddns.net/xz-486-tests.tar.xz
perjantai 11. marraskuuta 2016
JD0171R (R for red) small 3-digit 7-segment voltmeter schematic
So yeah this one can be had for less than a dollar from aliexpress or similar in various display colors, mine was red. It was a 2-wire model but as can be seen from the schematic it is trivial to change it to 3-wire operation if needed (remove the SMD resistor with 0R on it and solder a wire).
This reverse engineering was done thinking i could maybe reuse the assembly and swap in a different MCU (this chip has no markings, and no obvious way to access it, thus i guess it is mask rom or similar), but it is being a bit of a challenge to find an MCU with a suitable pinout & package.
The other side of the PCB has the text SJ-DC028-AB and the 7-segment display says YU2?81BS (? = 3 or 8, not sure).
The ???-resistor was likely a resistor (hypotethically could have been a cap...), but i managed to destroy it when disassembling this, so no value known.
That resistor and the 180R resistors (made of a 4x pack and 4 individual ones, what) are between the PCB and the 7-segment. Rest of the components are accessible for fun and mods :P
Oh yeah and Vcc is 3.3V, the 0uF cap is a place for a cap, nothing installed.
This reverse engineering was done thinking i could maybe reuse the assembly and swap in a different MCU (this chip has no markings, and no obvious way to access it, thus i guess it is mask rom or similar), but it is being a bit of a challenge to find an MCU with a suitable pinout & package.
The other side of the PCB has the text SJ-DC028-AB and the 7-segment display says YU2?81BS (? = 3 or 8, not sure).
The ???-resistor was likely a resistor (hypotethically could have been a cap...), but i managed to destroy it when disassembling this, so no value known.
That resistor and the 180R resistors (made of a 4x pack and 4 individual ones, what) are between the PCB and the 7-segment. Rest of the components are accessible for fun and mods :P
Oh yeah and Vcc is 3.3V, the 0uF cap is a place for a cap, nothing installed.
sunnuntai 13. syyskuuta 2015
Creating a new btrfs for the jolla phone
Okay so this is just a mind dump in case i need to do this later. I wiped my factory setup when doing this. Be warned, No warranty.
I assume you 1. have a backup of atleast the files you want your jolla btrfs to have (i had a clone of the emmc) and 2. the btrfs on the jolla is borked.
Making a blank btrfs image (because no mkfs.btrfs in the recovery mode):
1. sparse file: (if you plan on loop mounting this dont do sparse)
dd if=/dev/zero bs=1k count=0 seek=14415855 of=blank-btrfs.bin
2. make a btrfs on it with no extensions (jolla: old kernel)
mkfs.btrfs -L sailfish -O ^extref,^skinny-metadata blank-btrfs.bin
2.5. note the UUID in the output
3. make a tiny file out of it (2x gzip because gzip isnt _that_ good at compressing zeroes... result:40k)
gzip -9 blank-btrfs.bin
gzip -9 < blank-btrfs.bin.gz > blank-btrfs.bin.gz.gz
Write the blank btrfs image to the jolla:
1. boot in recovery mode and telnet there and open the shell (4)
on the jolla:
nc -l -p 1234 | gunzip | gunzip | dd of=/dev/mmcblk0p28 bs=16k
on the pc:
nc -c 10.42.66.66 1234 < blank-btrfs.bin.gz.gz
Mount the blank btrfs image on the jolla
mount -o subvolid=0 /dev/mmcblk0p28 /mnt
cd /mnt
Create the @ and @home (and @swap but i think thats unused) subvolumes on the jolla
btrfs subvolume create @
btrfs subvolume create @home
btrfs subvolume create @swap # just to copy what i had on the jolla
Mount the backup btrfs on the PC with -o subvolid=0 or have identical file tree somewhere, example:
losetup -P -f mmc.bin
mount -o subvolid=0,ro /dev/loop?p28 mount-point
Copy the files from the PC to the new jolla fs - on the jolla:
nc -l -p 1234 | tar xz
On the pc:
cd mount-point
tar -cz @ @home | nc -c 10.42.66.66 1234
This will take a while - you can take a second telnet connection to the jolla to look at df -h to see how much has been written etc
When this has been done, on the jolla, we need to fix the fstab since the new btrfs has a new UUID (you noted that above, did you. worry not, btrfs filesystem show -d or -m does show the fs with uuid)
On the jolla:
vi @/etc/fstab
and modify the two lines with UUID= to have the new UUID
Then we need to finish up the btrfs setup and make the phone behave as if the new filesystem is the factory filesystem (kernel etc stuff i dont know that much of yet.) - on the jolla:
btrfs subvolume snapshot @ factory-@
btrfs subvolume snapshot @home factory-@home
btrfs subvolume list # note ID for the @ subvolume
btrfs subvolume set-default <ID-OF-@> .
cd ..
umount mnt
Then return to the menu (reboot to recovery or exit, i cant remember) and use the factory reset option and the recovery shell will do some magic to make sure you have the correct stuff elsewhere.
Then the phone will reboot and it should work. If it doesnt you can try to debug
by booting back to the shell, mounting the fs, and setting
@/etc/systemd/journald.conf Storage=persistent
and trying to boot again (preferably try to turn it off with power button after giving it a moment)
and returning to recovery shell and chroot:ing to the @ subvolume and using journalctl
to read the logs recorded.
I assume you 1. have a backup of atleast the files you want your jolla btrfs to have (i had a clone of the emmc) and 2. the btrfs on the jolla is borked.
Making a blank btrfs image (because no mkfs.btrfs in the recovery mode):
1. sparse file: (if you plan on loop mounting this dont do sparse)
dd if=/dev/zero bs=1k count=0 seek=14415855 of=blank-btrfs.bin
2. make a btrfs on it with no extensions (jolla: old kernel)
mkfs.btrfs -L sailfish -O ^extref,^skinny-metadata blank-btrfs.bin
2.5. note the UUID in the output
3. make a tiny file out of it (2x gzip because gzip isnt _that_ good at compressing zeroes... result:40k)
gzip -9 blank-btrfs.bin
gzip -9 < blank-btrfs.bin.gz > blank-btrfs.bin.gz.gz
Write the blank btrfs image to the jolla:
1. boot in recovery mode and telnet there and open the shell (4)
on the jolla:
nc -l -p 1234 | gunzip | gunzip | dd of=/dev/mmcblk0p28 bs=16k
on the pc:
nc -c 10.42.66.66 1234 < blank-btrfs.bin.gz.gz
Mount the blank btrfs image on the jolla
mount -o subvolid=0 /dev/mmcblk0p28 /mnt
cd /mnt
Create the @ and @home (and @swap but i think thats unused) subvolumes on the jolla
btrfs subvolume create @
btrfs subvolume create @home
btrfs subvolume create @swap # just to copy what i had on the jolla
Mount the backup btrfs on the PC with -o subvolid=0 or have identical file tree somewhere, example:
losetup -P -f mmc.bin
mount -o subvolid=0,ro /dev/loop?p28 mount-point
Copy the files from the PC to the new jolla fs - on the jolla:
nc -l -p 1234 | tar xz
On the pc:
cd mount-point
tar -cz @ @home | nc -c 10.42.66.66 1234
This will take a while - you can take a second telnet connection to the jolla to look at df -h to see how much has been written etc
When this has been done, on the jolla, we need to fix the fstab since the new btrfs has a new UUID (you noted that above, did you. worry not, btrfs filesystem show -d or -m does show the fs with uuid)
On the jolla:
vi @/etc/fstab
and modify the two lines with UUID= to have the new UUID
Then we need to finish up the btrfs setup and make the phone behave as if the new filesystem is the factory filesystem (kernel etc stuff i dont know that much of yet.) - on the jolla:
btrfs subvolume snapshot @ factory-@
btrfs subvolume snapshot @home factory-@home
btrfs subvolume list # note ID for the @ subvolume
btrfs subvolume set-default <ID-OF-@> .
cd ..
umount mnt
Then return to the menu (reboot to recovery or exit, i cant remember) and use the factory reset option and the recovery shell will do some magic to make sure you have the correct stuff elsewhere.
Then the phone will reboot and it should work. If it doesnt you can try to debug
by booting back to the shell, mounting the fs, and setting
@/etc/systemd/journald.conf Storage=persistent
and trying to boot again (preferably try to turn it off with power button after giving it a moment)
and returning to recovery shell and chroot:ing to the @ subvolume and using journalctl
to read the logs recorded.
sunnuntai 24. elokuuta 2014
Nokia 700 059G2D6 log making of a simple CFW (not a simple log...)
This is mostly just a log for myself to keep up with what i did. The first try was a weird shotgun effort and resulted in a not clean enough and slightly broken FW, so I'm doing it methodically this time.
There's small bits of "howto-style" stuff put in there to make this maybe slightly less intimidating.
Lots of the tools can be had from there. Huge thanks to freaxs_r_us for this.
You're gonna need the base software, for that eg get it via mrcab.net
Next thing is make a copy of the base software, and open the uda with NokiaCooker.
Then use the UDA apps cleaner to get rid of all the crap. you want to. I googled lots of stuff. The badly named thing that has nokiapt files is the public transportation thing. Got rid of that. Mostly didnt remove small stuff that i dunno what it was rewally and stuff that i want (like maps). Got rid of weather because that doesnt work anymore. And all M$ crap :P (well now that nokia=m$ that would be hard, but the office etc i mean...) The App cleaner "crashes" on Cpix but you can say continue, look at the file list for cpix and delete the ones the app cleaner failed to remove (seems like the problem is that the sisregistry listed a non-existent folder) and then restart the app cleaner.
Also get rid of the languages you dont need, .r01 is english, r09 is finnish, .rsc is generic. I destroyed all other .r<number>'s. Check the Rofs2 lang tool for a text list of languages if you desire to see what the numbers mean.
Copy the UDA files folder to a safe place when ready. Save the firmware in nokiacooker if you wish to. Maybe should to have proper info in partitioning later. I did.
Switch to ROFS3.
https://sites.google.com/site/i8910wizard/calendar/the-vmp-of-the-game/101f8763-txt
ROFS3 will end up almost empty except for that (there was something random). I only changed the default suggested date to 2014.
Save and switch to ROFS2.
Run the ROFS2 Lang Tool. I ended up with only 01 and 09,d.
1. The lang tool isnt perfect, theres still lots of crap there relating not to finnish or english.
2. There's other crap in the ROFS2 i wanna get rid of (facebook, flickr, ...).
So i go into system/install and _move_ the .sis's for things i dont want out to another folder.
Then one by one open them with SISWare, delete everything in the file list they have, then delete them (if you're not sure you got all, you can leave the sis for later re-parsing.) Some of them are just empty markers, you can just note that, delete it, go on (i think).
I think there should be a way to speed this up, but i didnt find anything pretty.
There are some things like "Facebookplugins" that says to contain just that. I skipped them for now (didnt delete from my temp folder). I think they have files associated but just not listed or something... Will think about later.
This way you can also still easily undo the "think i'm gonna remove this" thought by just moving the sis back to system/install. I just undoed my thought of removing harfbuzz. Dunno if anybody needs it but seems to be a legit library.
After those... more cleanup:
resource/qt/translations contains stuff _da, _is, _nb, _sv ... got rid of those (everything not _en or _fi). There was/is some stuff there that i think i should/could get rid of entirely, but not yet (will investigate):
-- investigation log --
namely accounts_facebook and weatherwidget.
qt/plugins also has facebookplugin.qtplugin and flickroauth...
qt/imports/com/nokia also has facebook and flickr
qt/crml has fbshareplugins.qcrml
resource/ has facebookproviderdata and flickrproviderdata
resource/apps has weatherpublisher.r*
-- found this text on this somewhere on the net --
* How to remove the non-working Nokia weather widget *
Files and Folders to remove:
- Rofs2/private/2003A94C
- Rofs2/private/10003a3f/import/apps/weatherpublisher_reg
- Rofs2/private/10202D56/import/packages/2003a94e also go to rofs2/private/200159c0/install and delete anything with the name containin "weather". If you are using a CFW, therre might be weathersmall or weathermedium as well.
- Rofsx/resource/apps/weatherwidget/ (contains so many .png pictures)
- Rofsx/resource/apps/weatherpublisher.rxx (xx= your languages. eg. r01, r50)
- Rofsx/resource/apps/weatherwidget.mif
- Rofsx/resource/apps/weatherwidget_ah.mif
- Rofsx/resource/qt/translations/weatherwidget_xx.qm (xx is your language abbreviation like en, fa)
- Rofsx/sys/bin/weatherpublisher.exe
- Rofsx/sys/bin/weatherwidgetchsplugin.dll
For a clean wipe out, you need also to remove the ecom plugin of weatherwidgetchsplugin.rsc
-- i dunno what that last line means .. yet --
200159c0/install/*/*/ has folders for languages, keep 00 (generic), 01 and 09 (or whatever you're keeping).
200113dd/content also has folders per languages (matrixmenudata.dtd)
2000f8aa seems some nokia offer whatever crap, need to investigate what all it entails.
2003a945/import/languagedb also seems to have .pdct per language (named fi_FI etc)
-- investigation end --
deleted:
private/10009d8f/ ...
remove .sNN not 01 or 09, rename remaining to -2-1, -2-2...
not sure of this, hopefully did right.
sys/bin/xt9<wronglang>.dll
ptixt9kdb_itut_NN.dll
ptikeymappings_itut_NN.dll
share_youtube_plugin.exe (2003e208)
google_plugin.dll 2003e1c2
googlesignin.dll 2003e1c3
flickrsignin.dll 2003e1bc had resource FlickrConnectingTbItems.qml
flickroauth_plugin.dll
serviceframework_flickr_share.dll 2003e1be
serviceframework_facebook_share.dll 2003deb9
serviceframework_youtube_share.dll 2003deba YtSharingView.qml
flickr_share_plugin.exe 2003e1bf
fb_share_plugin.exe 2003de5d
facebooksignin.dll 2003e1b8 FbSigninView.qml
facebookplugin.dll 2003e1b6
system/accounts/providers/
provider_facebook.xml
- here was data to watch out for facebookloginapp.exe
provider_flickr.xml
- and here data to watch out for flickrloginapp.exe
provider_google.xml
- googleloginapp.exe
resource/facebookproviderdata
resource/flickrproviderdata
resource/googleproviderdata
private/2002AC7F/import/*
qt/imports/com/nokia/*
resource/qt/crml/fbshareplugins.qcrml
qt/plugins/*.qtplugin
qt/plugins/accountauthplugins/*
qt/translations/{fbstuff,google}
private/20031dc2
private/200113dd/NN
private/200159c0/install/*/*/NN
NN there meaning the numbers not 00,01 or 09.
That took a moment to go through, used NokiaCooker UI.
private/2003A945/import/languagedb/{not fin,en}
private/2002A867/translations/{*_not_en_or_fi}.qm
private/200159c0/install/weather*
resource/qt/translations/weatherwidget*
resource/apps/weatherpublisher.r*
Then closed all the explorer windows and tools (except NokiaCooker) and saved the ROFS2 (now 42.22Mb)
Then opened the core / ROFS1.
Here I'll mostly want to clean up,
embed RP+ and an AllFiles X-Plore (and unlock ROFS1).
1. Did as the weather widget removal above suggests.
2. cleaned up exes stuff from sys\bin that nokia has
put there to make old hacks no longer work, as suggested by the RP+ Readme.
3. Put in RP+
3.1. I'm not really sure what the RP+ author means when he says to mod noncriticalcmdlist_hw.rsc (eg. exactly what to remove/change-to-rp+ or if theres an editor that can add to these lists), but i ended up replacing sipprofilesrv.exe (with RPPAuto2.exe) that i believe does something related to SIP (that i dont use), so should be fine. YMMV, ... did that using the RCS editor.
4. Put in X-plore with all capabilities.
I'm sure i have a tool to edit the capabilites but since
the delight fw already has a fancy icon and capabilitized .exe i used their stuff. Just put in in Z (ROFS1, what we're editing now) though, seemed to work the last time.
I just looked up the places from the x-plore sis with SiSWare and took the files from delight fw with nokiacooker.
5. Click Unlock ROFS in the NokiaCooker, change all three values from 2B to 2C. You only need to do this once*
I'll find a better explanation of how this works later, but basically that skips some checksum check in the flashing/booting/somewhere. Critical step.
6. Save the core.... and NokiaCooker crashed...
Umm.. copied the Files folder of core into safety.
Closed nokiacooker (wiped files).
Run partition manager. Made everything like minimum +1 or 2Mb (ROFS3 i left 4Mb) except UDA. Will get more space for C: on the phone.
Click save. UDA will be erased.
I have it saved, np.
Open core again. Replace it with my Copy of files.
Unlock again. Try to save again.
Successful.
Open the UDA. Remember that it got emptied when partitioning. So now copy back in your backup of the files in UDA. Save it.
Then follow the flashing tutorial as per delight people.
*Yeah due to the crash i needed to do it twice.
Initial reports from the phone: seems fine.I might put the SW up some day, as well as mirror of all the files, but thats pending a really good file host.
Oh, NokiaCooker wants me to give a shoutout to them, great sw, but i cant remember what the exact banner was :P.
There's small bits of "howto-style" stuff put in there to make this maybe slightly less intimidating.
Lots of the tools can be had from there. Huge thanks to freaxs_r_us for this.
You're gonna need the base software, for that eg get it via mrcab.net
Next thing is make a copy of the base software, and open the uda with NokiaCooker.
Then use the UDA apps cleaner to get rid of all the crap. you want to. I googled lots of stuff. The badly named thing that has nokiapt files is the public transportation thing. Got rid of that. Mostly didnt remove small stuff that i dunno what it was rewally and stuff that i want (like maps). Got rid of weather because that doesnt work anymore. And all M$ crap :P (well now that nokia=m$ that would be hard, but the office etc i mean...) The App cleaner "crashes" on Cpix but you can say continue, look at the file list for cpix and delete the ones the app cleaner failed to remove (seems like the problem is that the sisregistry listed a non-existent folder) and then restart the app cleaner.
Also get rid of the languages you dont need, .r01 is english, r09 is finnish, .rsc is generic. I destroyed all other .r<number>'s. Check the Rofs2 lang tool for a text list of languages if you desire to see what the numbers mean.
Copy the UDA files folder to a safe place when ready. Save the firmware in nokiacooker if you wish to. Maybe should to have proper info in partitioning later. I did.
Switch to ROFS3.
https://sites.google.com/site/i8910wizard/calendar/the-vmp-of-the-game/101f8763-txt
ROFS3 will end up almost empty except for that (there was something random). I only changed the default suggested date to 2014.
Save and switch to ROFS2.
Run the ROFS2 Lang Tool. I ended up with only 01 and 09,d.
1. The lang tool isnt perfect, theres still lots of crap there relating not to finnish or english.
2. There's other crap in the ROFS2 i wanna get rid of (facebook, flickr, ...).
So i go into system/install and _move_ the .sis's for things i dont want out to another folder.
Then one by one open them with SISWare, delete everything in the file list they have, then delete them (if you're not sure you got all, you can leave the sis for later re-parsing.) Some of them are just empty markers, you can just note that, delete it, go on (i think).
I think there should be a way to speed this up, but i didnt find anything pretty.
There are some things like "Facebookplugins" that says to contain just that. I skipped them for now (didnt delete from my temp folder). I think they have files associated but just not listed or something... Will think about later.
This way you can also still easily undo the "think i'm gonna remove this" thought by just moving the sis back to system/install. I just undoed my thought of removing harfbuzz. Dunno if anybody needs it but seems to be a legit library.
After those... more cleanup:
resource/qt/translations contains stuff _da, _is, _nb, _sv ... got rid of those (everything not _en or _fi). There was/is some stuff there that i think i should/could get rid of entirely, but not yet (will investigate):
-- investigation log --
namely accounts_facebook and weatherwidget.
qt/plugins also has facebookplugin.qtplugin and flickroauth...
qt/imports/com/nokia also has facebook and flickr
qt/crml has fbshareplugins.qcrml
resource/ has facebookproviderdata and flickrproviderdata
resource/apps has weatherpublisher.r*
-- found this text on this somewhere on the net --
* How to remove the non-working Nokia weather widget *
Files and Folders to remove:
- Rofs2/private/2003A94C
- Rofs2/private/10003a3f/import/apps/weatherpublisher_reg
- Rofs2/private/10202D56/import/packages/2003a94e also go to rofs2/private/200159c0/install and delete anything with the name containin "weather". If you are using a CFW, therre might be weathersmall or weathermedium as well.
- Rofsx/resource/apps/weatherwidget/ (contains so many .png pictures)
- Rofsx/resource/apps/weatherpublisher.rxx (xx= your languages. eg. r01, r50)
- Rofsx/resource/apps/weatherwidget.mif
- Rofsx/resource/apps/weatherwidget_ah.mif
- Rofsx/resource/qt/translations/weatherwidget_xx.qm (xx is your language abbreviation like en, fa)
- Rofsx/sys/bin/weatherpublisher.exe
- Rofsx/sys/bin/weatherwidgetchsplugin.dll
For a clean wipe out, you need also to remove the ecom plugin of weatherwidgetchsplugin.rsc
-- i dunno what that last line means .. yet --
200159c0/install/*/*/ has folders for languages, keep 00 (generic), 01 and 09 (or whatever you're keeping).
200113dd/content also has folders per languages (matrixmenudata.dtd)
2000f8aa seems some nokia offer whatever crap, need to investigate what all it entails.
2003a945/import/languagedb also seems to have .pdct per language (named fi_FI etc)
-- investigation end --
deleted:
private/10009d8f/ ...
remove .sNN not 01 or 09, rename remaining to -2-1, -2-2...
not sure of this, hopefully did right.
sys/bin/xt9<wronglang>.dll
ptixt9kdb_itut_NN.dll
ptikeymappings_itut_NN.dll
share_youtube_plugin.exe (2003e208)
google_plugin.dll 2003e1c2
googlesignin.dll 2003e1c3
flickrsignin.dll 2003e1bc had resource FlickrConnectingTbItems.qml
flickroauth_plugin.dll
serviceframework_flickr_share.dll 2003e1be
serviceframework_facebook_share.dll 2003deb9
serviceframework_youtube_share.dll 2003deba YtSharingView.qml
flickr_share_plugin.exe 2003e1bf
fb_share_plugin.exe 2003de5d
facebooksignin.dll 2003e1b8 FbSigninView.qml
facebookplugin.dll 2003e1b6
system/accounts/providers/
provider_facebook.xml
- here was data to watch out for facebookloginapp.exe
provider_flickr.xml
- and here data to watch out for flickrloginapp.exe
provider_google.xml
- googleloginapp.exe
resource/facebookproviderdata
resource/flickrproviderdata
resource/googleproviderdata
private/2002AC7F/import/*
qt/imports/com/nokia/*
resource/qt/crml/fbshareplugins.qcrml
qt/plugins/*.qtplugin
qt/plugins/accountauthplugins/*
qt/translations/{fbstuff,google}
private/20031dc2
private/200113dd/NN
private/200159c0/install/*/*/NN
NN there meaning the numbers not 00,01 or 09.
That took a moment to go through, used NokiaCooker UI.
private/2003A945/import/languagedb/{not fin,en}
private/2002A867/translations/{*_not_en_or_fi}.qm
private/200159c0/install/weather*
resource/qt/translations/weatherwidget*
resource/apps/weatherpublisher.r*
Then closed all the explorer windows and tools (except NokiaCooker) and saved the ROFS2 (now 42.22Mb)
Then opened the core / ROFS1.
Here I'll mostly want to clean up,
embed RP+ and an AllFiles X-Plore (and unlock ROFS1).
1. Did as the weather widget removal above suggests.
2. cleaned up exes stuff from sys\bin that nokia has
put there to make old hacks no longer work, as suggested by the RP+ Readme.
3. Put in RP+
3.1. I'm not really sure what the RP+ author means when he says to mod noncriticalcmdlist_hw.rsc (eg. exactly what to remove/change-to-rp+ or if theres an editor that can add to these lists), but i ended up replacing sipprofilesrv.exe (with RPPAuto2.exe) that i believe does something related to SIP (that i dont use), so should be fine. YMMV, ... did that using the RCS editor.
4. Put in X-plore with all capabilities.
I'm sure i have a tool to edit the capabilites but since
the delight fw already has a fancy icon and capabilitized .exe i used their stuff. Just put in in Z (ROFS1, what we're editing now) though, seemed to work the last time.
I just looked up the places from the x-plore sis with SiSWare and took the files from delight fw with nokiacooker.
5. Click Unlock ROFS in the NokiaCooker, change all three values from 2B to 2C. You only need to do this once*
I'll find a better explanation of how this works later, but basically that skips some checksum check in the flashing/booting/somewhere. Critical step.
6. Save the core.... and NokiaCooker crashed...
Umm.. copied the Files folder of core into safety.
Closed nokiacooker (wiped files).
Run partition manager. Made everything like minimum +1 or 2Mb (ROFS3 i left 4Mb) except UDA. Will get more space for C: on the phone.
Click save. UDA will be erased.
I have it saved, np.
Open core again. Replace it with my Copy of files.
Unlock again. Try to save again.
Successful.
Open the UDA. Remember that it got emptied when partitioning. So now copy back in your backup of the files in UDA. Save it.
Then follow the flashing tutorial as per delight people.
*Yeah due to the crash i needed to do it twice.
Initial reports from the phone: seems fine.I might put the SW up some day, as well as mirror of all the files, but thats pending a really good file host.
Oh, NokiaCooker wants me to give a shoutout to them, great sw, but i cant remember what the exact banner was :P.
Tilaa:
Kommentit (Atom)







