Reply
Thread Tools
Posts: 646 | Thanked: 1,124 times | Joined on Jul 2010 @ Espoo, Finland
#31
Originally Posted by jalyst View Post
Makes sense, any rationale for the different Fw/eMMC image sizes between different variants that I noted?
It's probably small bits & bobs placed to please this and that opertor. Probably best way would be to get the tools to make an image and use them to open those .bin and compare them
Also, can you point to something that explains exactly why it's absolutely impossible to flash from [e.g.] PR_299 -> PR_235?
Do you remember what kind of error it throws in that case? Is is a 'security error' or a 'downgrade disallowed'?
The flashing verification is like: if (version1 > version2) and (swcert1 > swcert2) and (time_of_day > swcert1) and (other tests) then flash_version1 else exit_with_error
If that verification fails the only way to allow 'downgrade' is to use a rd certificate, but that has been available only in devices used internally in Nokia for testing.
 

The Following 2 Users Say Thank You to minimos For This Useful Post:
Posts: 1,067 | Thanked: 2,383 times | Joined on Jan 2012 @ Finland
#32
Originally Posted by minimos View Post
It's probably small bits & bobs placed to please this and that opertor. Probably best way would be to get the tools to make an image and use them to open those .bin and compare them

Do you remember what kind of error it throws in that case? Is is a 'security error' or a 'downgrade disallowed'?
The flashing verification is like: if (version1 > version2) and (swcert1 > swcert2) and (time_of_day > swcert1) and (other tests) then flash_version1 else exit_with_error
If that verification fails the only way to allow 'downgrade' is to use a rd certificate, but that has been available only in devices used internally in Nokia for testing.
And this is total ********, there is no version or variant name number checks (although variants are usually generated in descending order so those timestamps make it seem that there is variant name check).

The only check that there is is swcert timestamp. (as you can also see in error message that the number in that row is certificate creation time in seconds since 1.1.1970.)

If your flashimages swcert timestamp is older than the swcert's timestamp on device, you get "downgrade disallowed. ([timestamp])"-error message.

And even that check can be bypassed, so if you ever happen to visit Tampere (or see me in some Qt/Sailfish conference) and really want 001 on your device, it can be arranged .

And sorry, NO, before people ask, I'm not sharing how to do the bypass.
__________________
IRC: jonni@freenode
Sailfish: ¤ Qt5 SailfishTouchExample ¤ Qt5 MultiPointTouchArea Example ¤ ipaddress ¤ stoken ¤ Sailbox (Dropbox client) ¤
Harmattan: ¤ Presence VNC for Harmattan ¤ Live-F1 ¤ BTinput-terminal ¤ BabyLock ¤ BabyLock Trial ¤ QML TextTV ¤
Disclaimer: all my posts in this forum are personal trolling and I never post in any official capacity on behalf of any company.
 

The Following 6 Users Say Thank You to rainisto For This Useful Post:
Community Council | Posts: 4,920 | Thanked: 12,867 times | Joined on May 2012 @ Southerrn Finland
#33
Originally Posted by rainisto View Post
And sorry, NO, before people ask, I'm not sharing how to do the bypass.
Of course not

I have a hunch that requires spoofing the timestamp in the certificate, though. Maybe there's a weakness in the verification routines that can be triggered...
 

The Following User Says Thank You to juiceme For This Useful Post:
Moderator | Posts: 5,320 | Thanked: 4,464 times | Joined on Oct 2009
#34
Originally Posted by minimos View Post
It's probably small bits & bobs placed to please this and that opertor. Probably best way would be to get the tools to make an image and use them to open those .bin and compare them
I guess I'm just paranoid that some fw/eMMC images are corrupt*, & hence only a subset of their original size.
Or that they may be missing some critical stuff compared to other variant images...

*@MrCrab, or upon downloading

Originally Posted by rainisto View Post
And this is total ********, there is no version or variant name number checks (although variants are usually generated in descending order so those timestamps make it seem that there is variant name check).
<SNIP>
And even that check can be bypassed, so if you ever happen to visit Tampere (or see me in some Qt/Sailfish conference) and really want 001 on your device, it can be arranged .
And sorry, NO, before people ask, I'm not sharing how to do the bypass.
Easy there tiger, he was just postulating, thanks for the explanation rainisto, tis very much appreciated!

Last edited by jalyst; 2013-06-25 at 15:49. Reason: typo
 

The Following User Says Thank You to jalyst For This Useful Post:
Posts: 646 | Thanked: 1,124 times | Joined on Jul 2010 @ Espoo, Finland
#35
Originally Posted by rainisto View Post
And this is total ********, there is no version or variant name number checks (although variants are usually generated in descending order so those timestamps make it seem that there is variant name check).

The only check that there is is swcert timestamp. (as you can also see in error message that the number in that row is certificate creation time in seconds since 1.1.1970.)

If your flashimages swcert timestamp is older than the swcert's timestamp on device, you get "downgrade disallowed. ([timestamp])"-error message.
Well, if reality is not as complicated as I guessed, I gladly stand corrected
During development of N9, I tested upgrades at a rate of 1-2 per day.
In our team we had some cases of unexpectedly failed upgrade/reflash due to issues with the swcert, but we never got a clear explanation of how the whole verification process works, so yes there are some conditions where I can only offer a guess based on my experience.
 
Moderator | Posts: 5,320 | Thanked: 4,464 times | Joined on Oct 2009
#36
About re-download a whole bunch of apps from the Nokia Store & (this time) capture & store them elsewhere, in-case I need to re-flash several more times in the future.
I wouldn't worry about doing that, if it weren't for the limitation in the no. of times one can download apps from the Nokia Store.
Is N9Qtweak (I also know the associated cmds) still the simplest way to go about that, or is there an even simpler/more_elegant solution nowadays?

Last edited by jalyst; 2013-07-26 at 09:24.
 
F2thaK's Avatar
Posts: 4,365 | Thanked: 2,467 times | Joined on Jan 2010 @ Australia Mate
#37
i d/l with pc, use ssh to copy to n9 then use filebox app
 
Moderator | Posts: 5,320 | Thanked: 4,464 times | Joined on Oct 2009
#38
Originally Posted by minimos View Post
The shortlist of dependencies, nicely sorted is:
Code:
  account-plugin-facebook
  contacts
  contactsd
  evernote-sharing
  facebook
  facebook-meego
  facebookqml
  libqt-facebook
  mp-harmattan-001-pr
  si-helper
  webupload-service-facebook
So mainly you have Contacts tools that depends on FB, the webupload service and then of course the metapackage mp-harmattan to bolt it down.
There seems to have been somewhat of a breakthrough on this:
http://talk.maemo.org/showthread.php...25#post1365525
It's now possible to remove core packages like FB/Twitter without breaking everything else.
Although I imagine things that depend on them would become less functional/stable.

Last edited by jalyst; 2013-08-14 at 04:54.
 
Reply

Tags
nokia n9


 
Forum Jump


All times are GMT. The time now is 09:54.