thedead1440 | # 1341 | 2014-01-11, 20:13 | Report

It should be ctrl+z as doing so will suspend the process. I may be wrong but I remember doing so many months ago...

rainisto | # 1342 | 2014-01-12, 16:54 | Report

"flasher -i" doesnt flash anything, but keeps you in charging mode.

peterleinchen | # 1343 | 2014-01-12, 21:40 | Report

That would be too easy to have been implemented by NOkia
Never checked but when Jonni says so
Why did nobody came to find out? Too easy? Up to now everybody fiddled with flasher and the 11% hurdle.

--
But 'flasher - i' terminates itself after a few seconds. Charging goes on?
Or do we need a loop of 'flasher -i' ?

---

Last edited by peterleinchen; 2014-01-12 at 21:43.

juiceme | # 1344 | 2014-01-13, 06:19 | Report

There was one idea I was toying with over the weekend, but have not yet looked into it any deeper.

The way charging-by-flasher works (or actually the whole flashing thing works) is following:

• 1. the flasher loads a small kernel image to the device (called APE algorithm for some obscure reason...?)
  The APE algorithm has an included initramfs that contains the software to flash the device and also a BME implementation that can charge the device if needed
• 2. when the APE is loaded, it either charges the device f under 11% capacity and reports the charging back to the flasher, or it flashes the images transferred down from the flasher

Now, to my knowledge the sources of the APE image have not been published, but as it is a linux implementation it should be gotten from Nokia if somebody just would politely ask about it.

What could be implemented:
If the minimal configuration for charging the device could be determined then it would be no-brainer to make a jumpstarter image that could be started like flasher-charging, but something that does not flash the device, just charges it up to 100% or whatever required.

Additionally I could then implement the charging algorithm in ubiboot, so that would not drain the battery and could be used to jumpstart it if in need.

bitpowers | # 1345 | 2014-01-13, 08:37 | Report

I have a problem with the font in Sailfish. The basic menu and the setting is normal and readable.
In other cases, such as email is pretty much illegible.
Is this normal or do I have something is wrong?
Thank you for answers and solutions.

romu | # 1346 | 2014-01-13, 08:40 | Report

Check the issue tracker, there are some issues about font rendering.

juiceme | # 1347 | 2014-01-13, 09:06 | Report

Started to hack on it, just for fun

It was actually fairly easy to extract the rootfs from APE, now I need just tweak that a bit and try to repack it into a kernel image and see what happens

juiceme | # 1348 | 2014-01-13, 11:00 | Report

mmh... it seems that most intresting things are happening inside a binary "/usr/sbin/softupd" that init calls as the last thing.

What I did is I extracted that and naturally, uploaded it to my N9 and ran it as root
Scared the s**t out of me, that one...

RM696-21-3:~#
RM696-21-3:~#
RM696-21-3:~# ./softupd
[14519.192718] Creating 1 MTD partitions on "omap2-onenand":
[14519.192840] 0x000000000000-0x000001000000 : "tmp"
[14519.230529] Creating 2 MTD partitions on "omap2-onenand":
[14519.230651] 0x000000000000-0x000000100000 : "tmp"
[14519.241851] 0x000000100000-0x000000340000 : "config"
[14519.271820] Creating 6 MTD partitions on "omap2-onenand":
[14519.271942] 0x000000000000-0x000000100000 : "bootloader"
[14519.277404] 0x000000100000-0x0000003c0000 : "config"
[14519.287719] 0x0000003c0000-0x0000013c0000 : "kernel"
[14519.288146] 0x0000013c0000-0x0000015c0000 : "log"
[14519.299133] mtdoops: Attached to MTD device 3
[14519.299224] 0x0000015c0000-0x00001e000000 : "var"
[14519.302978] 0x00001e000000-0x000020000000 : "moslo"
FATAL: Error inserting g_softupd (/lib/modules/2.6.32.54-dfl61-20121301/g_softupd.ko): Device or resource busy
FATAL: Error inserting g_nokia (/lib/modules/2.6.32.54-dfl61-20121301/g_nokia.ko): Device or resource busy
Cannot find device "upnlink0"
ifconfig: SIOCGIFFLAGS: No such device
phonet: interface `upnlink0' not found
BB5 bb5_rdc_cert_read return value. LIB: 0x00000019 (Inappropriate ioctl for device)
BB5 bb5_rdc_cert_read return value. ROM: 0xFFFFFFFF (UNKNOWN)
BB5 bb5_rdc_cert_read return value. PA: 0xFFFFFFFF (UNKNOWN)
BB5 sec_get_locks return value. LIB: 0xFFFFFFFF (Unknown error 4294967295)
BB5 sec_get_locks return value. ROM: 0x0000000A (SEC_ROM_RET_FAIL)
BB5 sec_get_locks return value. PA: 0xFFFFFFFF (UNKNOWN)
Simlock certificate check: 473 SUCCESS ret val: 0
Image lock is open

^C^C

RM696-21-3:~#

Looking at that I really thought now I've blown it big time...
it seems the sneaky bit of software went and repartitioned my MTD just like that... whooooo!
Well, did it also clear it? hopefully not.... no way to check but reboot and see if it comes back alive....

...and it did (phewwww....)

Now I need to experiment a bit with that if I can get anything worthwhile out of it.
Too bad there's probably no source available for softupd.

Hurrian | # 1349 | 2014-01-13, 11:36 | Report

Oh man, that's living on the edge! I actually thought it'd go and flash_erase stuff.

Sounds more like it was preparing the in-memory mappings for MTD partitions for flashing.

IIRC, what it does is open up an interface over USB to accept flash images, then does any tidying up after flashing.

Can't wait to see what you come up with - recovering from a flat battery on every Nxxx device is a pain!

anthonie | # 1350 | 2014-01-13, 12:25 | Report

Running 1.0.2.5 here and I must say, I am pretty impressed about some things. With other things I am underwhelmed. All in all it's a mixed bag with lots of potential.

That being said, there are some issues I encounter that I did not see mentioned in the thread so far.

When I try and enter my simlock pin, often I need to pick up the phone first, before I can enter digits. Almost as if the accelerometer triggers the keypads to function or something. But for that, it should happen consistently, but it doesn't. When I type digits, phone laying down, sometimes it takes 3 of 4 keystrokes before the eventhandler picks it up. Sometimes it doesn't react and I need to pick up the phone first. Odd behaviour, it seems.

In other places keystrokes (the terminal, to name one) also seem a bit sticky, reacting slow at times. But in apps the trouble usually appears after running Sailfish for several hours. A reboot quickly resolves it, until everything connected with typing comes to a grind again

Another thing is my N9 seems to warm up a lot more than it does running Harmattan. Just having the terminal and a browser open, perhaps the store and in half an hour or so the device really starts to warm up. Not to alarming heats but enough to be significant in comparison to Harmattan.

Anyone else experiencing things like this?

Edit: Also, is there a way to get rid of the boot warning (warranty) without flashing the phone through again? I mean, I didn't care at first, but rebooting often makes it quite annoying. However, a clean format/reflash sequence for both Harmattan and Sailfish is not an option.

---

Last edited by anthonie; 2014-01-13 at 12:30.