Mount OBB containers using shared app GID, so that an app can read
the mount point across users.
Bug: 7212801
Change-Id: Ia1be52df9854c259b20728111f3a2c9facf4beaa
Augment the Ext4::format function to take
a mountpoint parameter. This will then
be passed to make_ext4fs through the
-a option to allow proper security labeling.
Change-Id: Ic26703406a2c463c12e32c8103a0c75c727b7d29
Any ASEC or OBB files were unmounted when USB storage was set to UMS
mode. This changes it so only ASEC files on external storage and OBB
files mounted from external storage are unmounted.
(Cherry-pick of 93ecb38dad)
Bug: 6948035
Change-Id: Ib60727bd360caa32173797ff5b4e1e21fcf20054
Any ASEC or OBB files were unmounted when USB storage was set to UMS
mode. This changes it so only ASEC files on external storage and OBB
files mounted from external storage are unmounted.
Bug: 6948035
Change-Id: I91bc09ee5b792970b0eef895f6886f3ffad00e8f
Now that forward locked apps are stored on /data as asec image files
that are mounted, they need to be unmounted before /data can be unmounted
so it can be encrypted.
Change-Id: I7c87deb52aaed21c8ad8ce8aceb7c15c2338620a
There appears to be a race condition from when the device mapper is
asked to create a device and when it actually appears. When we moved
ASECs to use Ext4, mount started winning the race more often.
Just insert a sleep-retry loop here to counter-act this race. We should
ideally look at the uevent replies, but it takes a bit more effort to
separate them out.
Change-Id: Ie8a5b36b1c9a26f2320a178d37312059d03a1281
When calling System.loadLibrary(), it needs to be able to read the
directory to load the file. We could probably fix that, but changing
permissions here is faster.
Bug: 6478606
Change-Id: I296b0805839da5a19950157f9a16755a4d258ca8
Traversal would mark directories with the correct permissions, but
they're visited again in post-order which is a different fts_info flag.
Then it would set that to regular file permissions.
Explicitly check to make sure we're looking at a file instead.
Bug: 6478606
Change-Id: I13cab3e69f451da6a994fa974d575ef366f82025
Now forward locked applications will be in ASEC containers both internal
to the system and externally.
This change adds support for putting applications in ext4-based ASECs.
Change-Id: I8d6765b72dd2606e429c067b47a2dbcaa8bef37d
Add the force_and_revert option to the unmount command which will force
the unmount, and revert a crypto mapping. This is used during factory
reset so that when the internal sdcard volume is formatted, it formats
the raw device, not the encrypted mapping.
Change-Id: I36b6ff9bb54863b121de635472a303bf4a2334a9
Mounting was already not allowed, but also unshare before starting
encryption, and don't allow sharing or formatting to be initiated
during encrytion.
Change-Id: Ida188d81f025739ba4dd90492b3e66088735991e
Add support for keeping the keys in a separate file on another partition,
for devices with no space reserved for a footer after the userdata filesystem.
Add support for encrypting the volumes managed by vold, if they meet certain
criteria, namely being marked as nonremovable and encryptable in vold.fstab.
A bit of trickiness is required to keep vold happy.
Change-Id: Idf0611f74b56c1026c45742ca82e0c26e58828fe
Improves UI responsiveness when copying large amount of data to the device.
BUG: 3131847
Change-Id: I4aa5ade7e2cd7e5110c8f0f7ee43bdc57577e11d
Signed-off-by: Mike Lockwood <lockwood@google.com>
ENOENT is more descriptive than the generic error that EINVAL gets
turned into. Then we can actually treat it how we want on the other
side.
Change-Id: I9b4f3be6308e13f680eae368d2167ab9ee6aae5f
OBB and ASEC are tracked in the same active container list, but when it
comes time to unmount everything, it was trying to unmount the OBBs
according to ASEC rules. This led to the OBB not being unmounted and the
volume unmount failing.
Change-Id: I12c1d4d387b8022185d552b63edd61a50b9c0fc3
Only share a single partition via UMS if a specific partition
is specified in vold.fstab (rather than "auto")
Do not fail to reformat if MBR cannot be found.
Change-Id: I544ca2ee325c308af656ab9f4732c788a964c156
Signed-off-by: Mike Lockwood <lockwood@android.com>
This allows us to place shared libraries in these containers which may
only be loaded if they are executable.
Change-Id: I78fa9ab6d5c58ec8b98c40004da72aebc0aade2a
* Rename all functions dealing with OBB files to mention Obb
* Add 'path' and 'list' functionality to OBB commands
* Store hashed filename in loop's lo_crypt_name and keep lo_file_name
for the real source filename. That way we can recover it later with an
ioctl call.
Change-Id: I29e468265988bfb931d981532d86d7be7b3adfc8
Allow the mounting of OBB filesystem images if they're encrypted with
twofish and in FAT filesystem format.
Change-Id: I54804e598f46b1f3a784ffe517ebd9d7626de7aa
The usb_mass_storage switch no longer exists in our 2.6.35 kernel.
Instead we will consider mass storage to be available if both USB is connected
and the USB mass storage function is enable.
Change-Id: I730d1b3cb3cac664fc2abcdc36cd39856a08404a
Signed-off-by: Mike Lockwood <lockwood@android.com>
Fixes bug: http://b/issue?id=2567572
Note: The framework will still likely restart since the system_server
is holding references to assets on the card which are mmaped, but
at least now storage will be available when a new card is re-inserted.
Change-Id: I4e195c0c666426b93da47198fa826a6f58d855a9
Signed-off-by: San Mehat <san@google.com>
- Fix issue where container-names > 64 bytes were getting truncated in the
kernel. lo_name is only 64 bytes in length, so we now hash the container
id via md5
- Add 'dump' command to dump loop and devicemapper status
- Add 'debug' command to enable more detailed logging at runtime
- Log vold IPC arguments (minus encryption keys)
- Fix premature return from Loop::lookupActive() and friends
Change-Id: I0e833261a445ce9dc1a8187e5501d27daba1ca76
Signed-off-by: San Mehat <san@google.com>
Add an un-encrypted superblock to the end of a secure container to
help identify the crypto cipher used. This is required in order to
provide some semblence of versioning, in case we have cause to
change the crypto type on an update - say for example switching
from software to hardware encryption, or from 128 -> a zillion
bit keys.
NOTE: This format is incompatible with previous secure containers.
After this change they will no longer be mountable.
Signed-off-by: San Mehat <san@google.com>
In order to protect the '/android_secure' directory on VFAT removable media
from being mucked with by 3rd party applications on the device, we hide the
directory with a read-only, zero-sized tmpfs mounted on-top. A reference to the
hidden directory is kept by a bind-mount which is mounted at a location which
only root can access.
Staging consists of:
1. Mount checked media at a secure location (/mnt/secure/staging)
2. Ensure /android_secure exists on the media, (creating if it doesnt)
3. Bind-mount /mnt/secure/staging/android_secure -> /mnt/secure/asec
(where only root can access it)
4. Mount an RDONLY zero-sized tmpfs over /mnt/secure/staging/android_secure
5. Atomically move /mnt/secure/staging to the publicly accessable storage
directory (/mnt/sdcard)
Signed-off-by: San Mehat <san@google.com>
