[protector] Bump Preferences backup version to 3 to mitigate the startup prefs migration.

[protector] Bump Preferences backup version to 3 to mitigate the startup prefs migration.

Since r134147 (https://chromiumcodereview.appspot.com/10213009), startup prefs migration is more complex and may trigger the protector for users on M19.


BUG=122936
TEST=ProtectedPrefsWatcherTest.*


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=134698

[Ivan Korotkov, 2012-04-28 12:30:44]

Hi Mattias and Tyler,

recent change to startup prefs migration
(https://chromiumcodereview.appspot.com/10213009) causes a Protector's false
positive for users who:
 (1) already have startup prefs protection (i.e., Dev and Beta users if merged
to M19)
 (2) have homepage set to non-NTP
 (3) have kRestoreOnStartup not set.

My fault to not catch it in the review in the first place :(

Initially, startup prefs protection was added *after* the first migration.
Therefore, the migration step was forced to happen before the preferences are
checked for any changes -- this worked fine with the first migration and would
trigger protector as expected if old malware would change user's settings to
pre-migrated state.

But now there is a more complicated migration step that happens for users with
the setup described above (who *already* have prefs protection initialized and
working), and from the Protector's point of view it means sudden change from
starting with NTP to starting with homepage.

Hence there is a need to do a one-time migration step of backup as well (should
have been done in the same CL, actually...)

[Tyler Breisacher (Chromium), 2012-04-30 17:59:38]

Good point. I didn't think of this because I was only thinking about users on
the stable channel. lgtm

[commit-bot: I haz the power, 2012-05-01 12:39:05]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/ivankr@chromium.org/10170046/1

[commit-bot: I haz the power, 2012-05-01 14:10:36]

Change committed as 134698

[Mattias Nissler (ping if slow), 2012-05-02 09:28:00]

On 2012/05/01 14:10:36, I haz the power (commit-bot) wrote:
> Change committed as 134698

Just a question: Do we have to add special-case code for all protected prefs
even if they get written from within Chrome? I may be lacking context, but that
seems pretty unsustainable to me.

[Ivan Korotkov, 2012-05-02 10:26:15]

This one is different because the migration step happens explicitly before
ProtectorService starts observing pref changes. It could be done after, but then
one could externally change Preferences to a pre-migrated state, migration would
happen and mask the fact of a change.
The same way Web Data migrations happen before checking for search engine
changes.