tequilaOS/platform_system_vold
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity

Stop using trigger_reset_main.

Browse source 
This trigger was used on FDE devices to bring down the minimal
framework, and worked by shutting down the 'main' service class.

With APEX being introduced, we want to restart all services that were
started after the tmpfs /data was mounted, as those are the services
that haven't been able to use updated APEXes in the (real) /data.

In order to do this, we need to reset more classes; that in turn
made the 'shutdown_main' trigger pretty much similar to the
previously existing 'trigger_shutdown_framework' trigger; so instead
of keeping two duplicate triggers, use only the
'trigger_shutdown_framework' one.

Bug: 118485723
Test: Taimen configured as FDE boots, Taimen configured as FBE boots
Change-Id: I0d80ef2528bd70870b063a2c580cd00a03de9961
This commit is contained in:
Martijn Coenen 2019-04-24 10:41:11 +02:00
parent 8d4164f92f
commit aec7a0a165
1 changed files with 19 additions and 8 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

27
cryptfs.cpp
View file

@ -1569,12 +1569,23 @@ static int cryptfs_restart_internal(int restart_main) {
if (restart_main) {
/* Here is where we shut down the framework. The init scripts
* start all services in one of three classes: core, main or late_start.
* On boot, we start core and main. Now, we stop main, but not core,
* as core includes vold and a few other really important things that
* we need to keep running. Once main has stopped, we should be able
* start all services in one of these classes: core, early_hal, hal,
* main and late_start. To get to the minimal UI for PIN entry, we
* need to start core, early_hal, hal and main. When we want to
* shutdown the framework again, we need to stop most of the services in
* these classes, but only those services that were started after
* /data was mounted. This excludes critical services like vold and
* ueventd, which need to keep running. We could possible stop
* even fewer services, but because we want services to pick up APEX
* libraries from the real /data, restarting is better, as it makes
* these devices consistent with FBE devices and lets them use the
* most recent code.
*
* Once these services have stopped, we should be able
* to umount the tmpfs /data, then mount the encrypted /data.
* We then restart the class main, and also the class late_start.
* We then restart the class core, hal, main, and also the class
* late_start.
*
* At the moment, I've only put a few things in late_start that I know
* are not needed to bring up the framework, and that also cause problems
* with unmounting the tmpfs /data, but I hope to add add more services
@ -1582,10 +1593,10 @@ static int cryptfs_restart_internal(int restart_main) {
* till the user is asked for the password to the filesystem.
*/
/* The init files are setup to stop the class main when vold.decrypt is
* set to trigger_reset_main.
/* The init files are setup to stop the right set of services when
* vold.decrypt is set to trigger_shutdown_framework.
*/
property_set("vold.decrypt", "trigger_reset_main");
property_set("vold.decrypt", "trigger_shutdown_framework");
SLOGD("Just asked init to shut down class main\n");
/* Ugh, shutting down the framework is not synchronous, so until it