If we created a BROWSE PendingIntent early during boot (while the
device was still locked), we would resolve the MIME type as "null"
meaning the intent would fail to launch.
Since we always know what the MIME type will be, set it explicitly.
Test: boot device while USB device attached
Bug: 30742481
Change-Id: I23c523cbe2660ca63a3b7df1f41300ab803b314c
We call uncrypt services to setup / clear bootloader control block (BCB)
for scheduling tasks under recovery (applying OTAs, performing FDR).
However, we cannot start multiple requests simultaneously. Because they
all use the same socket (/dev/socket/uncrypt) for communication and init
deletes the socket on service exits.
This CL fixes the issue by a) adding synchronized blocks for the service
requests; b) checking the availability of the socket before initiating a
new one.
Note that adding synchronized blocks to RecoverySystem doesn't help,
because the calls could be made from different processes (e.g. priv-app,
system_server).
Bug: 31526152
Test: FDR works while a priv-app keeps calling clear BCB.
Change-Id: I95308989e849a9c98a9503ac509f2bc51ed3de19
(cherry picked from commit 794c8b0b3f)
Log the error code to uncrypt_status if uncrypt gets killed because
of timeout.
Test: We log the error code correctly in uncrypt_status when uncrypt timeouts.
Bug: 31603820
Change-Id: Ia623c333714295e68f4269257fbb4297a867e42b
(cherry picked from commit 036d08638e)
When we create a file that already exists, we try attaching a suffix
like "(1)" to the filename to avoid the conflict. The newly added
rename method should do the same, since developers may not have
access to delete the conflicting file.
Test: boots, rename via UI, new unit tests
Bug: 31545404
Change-Id: Ie397eebb0fbf98cf079eee3bbbb6c6b7ca627d91
Read and report time cost of uncrypt (in seconds) from last_install.
Bug: 31383361
Change-Id: I283970c33b2a0324f2f51cde328e3e527eff7c58
(cherry picked from commit 3477efce30)
This change adds a new pathMatch attribute to the intent-filter tag
that supports a subset of regular expression syntax. Supported
tokens include dot (.) and set ([]) and supported modifiers include
zero-or-more (*), one-or-more(+) and ranges ({}).
Change-Id: I2ec3bc8a9eb3e94d21caf34a1ed0f3fd3fb33a35
Bug: 30805203
Fixes a bug where DozeService would not properly initialize
because the CPU went to sleep before onDreamingStarted completed,
causing the pickup gesture to not work.
Change-Id: I85955a2b7d6bad5171accbc336117a9660b1b198
Fixes: 31044352
Don't write partial requests, and don't return (or throw) early after
partially reading a response.
bug: 30143607
Change-Id: I5881fdd5e81023cd21fb4d23a471a5031987a1f1
When cancelling, the exception will be eaten by FutureTask, but
onCancelled would end up never being called due to the exception
causing the code to skip calling postResult.
b/30304893
Change-Id: I47d859d9ef77245889816b2b6e10bb380bc4979b
Also fix b/30230996.
TODO: Support the sourceBounds argument to LauncherApps.
startShortcut().
Bug 30218829
Bug 30230996
Change-Id: Ibb6c59d5b7c0a320c21d09436d016d0eac94644f
These appear as a new event in the battery stats history,
"longwake" in the long version and "Elw" in the checkin.
The power manager keeps track of which wake locks are held
for a long time and reports them to battery stats. Long
is currently considered 1 minute or more. Once it is long,
the start event will appear, and once if is released the
event will end.
In the case of a wake lock changing (typically its work
source changing), for purposes of this accounting this is
considering a pure release of the old state and start of
the new state... so the timer will reset back to one
minute until the wake lock is considered long. This is done
to prevent things that make lots of changes to wake lock
work sources from spamming the log.
Bug: 28753137
Change-Id: I33b6168c57a7ea6ea558273dec731704123124a5
The total storage size was calculating by reading and multiplying the
contents of /sys/block/mmcblk0/size and
/sys/block/mmcblk0/queue/hw_sector_size.
On some devices, such calculation doesn't work because:
1.The primary block is not /sys/block/mmcblk0 .
2.The sector size is not the right value to use.
These 2 issues are temporarily addressed by providing alternative
primary blocks and hardcoding the size (512 bytes). In the long term,
the size should be calculated by vold, so each device could provide its
own calculation if necessary.
BUG: 30216622
Change-Id: I8f9a9f4f753d3c92bab9257062b61ed2b9d665c5
Fixes: 24128505
If an application thread disables disk reads on the main thread,
isThreadInProcess will trigger a strict mode violation. This is bad if
the RenderThread is created after disk reads have been disallowed.
bug 30170339
Change-Id: I6918605d11f0c586c156f3168051376c3209b3f1
If the OTA updates a system app which was previously preopted to a non-
preopted state the app might end up being verified at runtime. That's
because by default the apps are verify-profile but for preopted apps
there's no profile.
Do a hacky check to ensure that if we have no profiles (a reasonable
indication that before the OTA the app was preopted) system apps get
compiled with a non-profile filter (by default interpret-only).
Bug: 30032273
Test: Andreas "has verified that Calin's change to A/B works as expected
and promotes things like SystemUI to speed. From my side, that's
ready to be merged"
Change-Id: I7a052a8ea76cab7f649dc993237ea05534d6c4b9
