FDE keys are indexed using the partition GUID, while FBE keys will be
indexed using the filesystem UUID, so pass both of those identifiers
along when forgetting a volume.
Test: cts-tradefed run commandAndExit cts-dev -m CtsAppSecurityHostTestCases -t android.appsecurity.cts.AdoptableHostTest
Bug: 25861755
Change-Id: I6e239d5ba67a01c9a848d705f6167da00f975924
Prevents selinux denials for folders in /proc that do not have the
default /proc label.
Bug: 68146208
Test: no selinux denials for vold attempting to read proc_asound dir.
Merged-In: I7cdd3bbe8e687e078372012773e9a34a5c76e0f8
Change-Id: I7cdd3bbe8e687e078372012773e9a34a5c76e0f8
Prevents selinux denials for folders in /proc that do not have the
default /proc label.
Bug: 68146208
Test: no selinux denials for vold attempting to read proc_asound dir.
Change-Id: I7cdd3bbe8e687e078372012773e9a34a5c76e0f8
I want to use Utils in another executable, so breaking this link.
Bug: 25861755
Test: compiles (and boots, though that doesn't exercise changed code)
Change-Id: I6bb447453bb370fefb7f2f3aceb459428bdee6a7
Moves away from crufty char* operations to std::string utility
methods, including android::base methods for splitting/parsing.
Rewrite of how Process handles scanning procfs for filesystem
references; now uses fts(3) for more sane traversal.
Replace sscanf() with new FindValue() method, also has unit tests.
Remove some unused methods. Switch almost everyone over to using
modern logging library.
Test: cts-tradefed run commandAndExit cts-dev -m CtsOsTestCases -t android.os.storage.cts.StorageManagerTest
Test: cts-tradefed run commandAndExit cts-dev --abi armeabi-v7a -m CtsAppSecurityHostTestCases -t android.appsecurity.cts.AdoptableHostTest
Bug: 67041047
Change-Id: I70dc512f21459d1e25b187f24289002b2c7bc7af
Instead of blindly looping across 4096 possible devices, use
readdir() to only look at valid devices. This speeds up destroyAll()
from 40ms to 0.7ms.
Add tracing information in several places.
Test: external/chromium-trace/systrace.py -b 128768 sched freq am pm ss core_services binder_driver -a system_server,installd,vold
Bug: 65634729, 65737446
Change-Id: If581de47fb55850c0fcd6e25bf33ed246e1b079d
Now that we've moved to Binder, we only have a few lingering atoi()
usages that are cleaned up in this CL.
Rewrite match_multi_entry() entirely, with tests to verify both old
and new implementations.
Test: adb shell /data/nativetest/vold_tests/vold_tests
Bug: 36655947
Change-Id: Ib79dc1ddc2366db4d5b4e1a1e2ed9456a06a983e
Now that we're using Binder, we can have callers provide explicit
listeners for every request instead of trying to squeeze them all
into unsolicited socket events.
Move benchmarking to be async to avoid blocking other commands for
up to several minutes. Remove post-trim benchmarking flag, since
benchmarking now requires a separate callback. Will bring back in
a future CL.
Test: cts-tradefed run commandAndExit cts-dev -m CtsAppSecurityHostTestCases -t android.appsecurity.cts.AdoptableHostTest
Test: adb shell sm fstrim
Bug: 62201209, 13758960
Change-Id: I0f2ebf1ac3b4252ecd6b44303f2887adfdb58e86
This moves fstrim, obb and appfuse commands over to the new Binder
interface. This change also separates creating/destroying and
mounting/unmounting of OBB volumes, which means they finally flow
nicely into the modern VolumeInfo/VolumeBase design.
We now generate unique identifiers for all OBB volumes, instead of
using a shady MD5 hash.
Change all "loop" and "dm" devices to tag the kernel resources with
a vold-specific prefix so that we can clean them up if vold crashes;
there are new destroyAll() methods that handle this cleanup.
Move appfuse mounting/unmounting into VolumeManager so it can be
shared. Move various model objects into a separate directory to
tidy things up.
Test: cts-tradefed run commandAndExit cts-dev -m CtsOsTestCases -t android.os.storage.cts.StorageManagerTest
Bug: 13758960
Change-Id: I7294e32b3fb6efe07cb3b77bd20166e70b66958f
Keep the old socket-based commands intact for awhile so we can
rapidly disable this change using the ENABLE_BINDER feature flag.
Define constants in AIDL to keep Java and C++ in sync.
Test: cts-tradefed run commandAndExit cts-dev -m CtsAppSecurityHostTestCases -t android.appsecurity.cts.ExternalStorageHostTest
Test: cts-tradefed run commandAndExit cts-dev --abi armeabi-v7a -m CtsAppSecurityHostTestCases -t android.appsecurity.cts.AdoptableHostTest
Bug: 13758960
Change-Id: I0d6f82cbebe67f671b60949fd727409aeb1fdc0d
It fixes the findvolume() / reset() use-after-free issue after
shutdown called to avoid vold crash.
Fixes: a5bbb5e3c1 ("make shutdown safe for double calls.")
Change-Id: I50f216141b20da08549080291091dc5690c00ffe
Signed-off-by: Gao Xiang <gaoxiang25@huawei.com>
- Various sleep(5) for vold shutdown can increase shutdown time a lot.
- If it is shutting down, do not sleep at all. init will take care of
active partitions if not unmounted.
bug: 64143519
Test: reboot and check logs from vold, check if "ShutdownThread: Shutdown wait timed out" happens.
Change-Id: I7cb91427ad2205fe23a054d255caf7ffdfd9f6c3
Start with clang-analyzer-security* and cert-*, but disable two
specific errors:
-- cert-err34-c, which checks for atoi(); heavily triggered by
CommandListener, but will disappear when we move to Binder.
-- cert-err58-cpp, which checks for exceptions before main(); it's
a "Low" severity issue, and filed 36656327 to track cleanup.
Fix all other triggered errors along the way.
Test: builds, boots
Bug: 36655947
Change-Id: I1391693fb521ed39700e25ab6b16bc741293bb79
Also remove some unnecessary SELinux logic when creating image files
for loop devices.
Test: builds, boots, common operations work
Bug: 34903607
Change-Id: I68dfa022ecc39f56c175e786694e0de35b954ca0
It's extremely difficult to test storage related logic on devices
that don't have physical SD card slots. So to support better
debugging and testing, add a new "virtual disk" feature which mounts
a 512MB file through loop device.
It relies on the kernel having the "loop.max_part" value set to
something other than 0 via the boot command line, since that allows
all the existing partition logic to fall into place.
Bug: 34903607
Test: builds, boots, virtual disk works
Change-Id: I04c5b33e37319d867542985a56b7999a9b7cf35d
- In new shutdown sequence, volume shutdown is requested in two places:
system_server and init.
- Make VolumeManager.shutdown safe for double calls.
It crashed before this change.
bug: 36004738
Test: reboot and check last_kmsg for crash / vdc timeout
Change-Id: I296913959b2647b65d66553073f2032545beba57
coldboot is now taken care by ueventd, so removing the duplicate
logic to save boottime.
This CL also fixes a missing lock.
Bug: 33786699
Test: manual
Change-Id: I71270252f3153abc815d142b5da7a9cb85b94dac
* Use const reference type for for-loop index variables
to avoid unnecessary copy.
Bug: 30413223
Change-Id: Id4d980ae8afec1374fc3be0b23f1c6a39bff86e0
Test: build with WITH_TIDY=1
This fixes an eror where resizeAsec would attempt
to read from the superblock struct before initializing
it.
Bug: 28292918
Change-Id: Ic6804e97e7c83bcedfb682a187b8d5e0e1bc51f9
Currently, vold only supports MMC (for SD cards) and SCSI (for USB
drives) devices. It does not recognize any device whose major number is
not one of those used by MMC and SCSI. Unfortunately, virtio-blk is one
such device. It is used by the new Android emulator (a.k.a. qemu2,
featuring the "ranchu" virtual board) for SD card emulation.
In order to make this virtio-blk based SD card device appear in Android
and appear as an SD card (rather than a USB drive), changes have to be
made to both vold (wherever the device major number is checked) and
ranchu's storage configuration. This CL implements former.
This is a stop-gap solution for emulator in nyc.
A longer term solution in-tune with upstream kernel is in the pipes.
Updated from aosp/master version.
BUG:27431753
Change-Id: I5014edec73be7c5b565d91542464c82cbe58992c
Signed-off-by: Yu Ning <yu.ning@intel.com>
(cherry picked from commit 5b1d1c7dfa13b4dca75213581dc8351b841b76c8)
Changing the num_sectors used in ioctl with BLKGETSIZE because
the kernel expects an unsigned long type and then changes 64 bits
with a 64 bits userspace. This overwrites what's located close to
the parameter location if any.
Change-Id: I78fd61a1084de2741f39b926aa436462518709a0
Signed-off-by: Mateusz Nowak <mateusz.nowak@intel.com>
Signed-off-by: Zhiquan Liu <zhiquan.liu@intel.com>
We have a bunch of magic that mounts the correct view of storage
access based on the runtime permissions of an app, but we forgot to
protect the real underlying data sources; oops.
This series of changes just bumps the directory heirarchy one level
to give us /mnt/runtime which we can mask off as 0700 to prevent
people from jumping to the exposed internals.
Also add CTS tests to verify that we're protecting access to
internal mount points like this.
Bug: 22964288
Change-Id: I83f09f0423f4993e766273c50389dd29b1c50589
When mounting a primary external storage device that is multi-user
aware, ensure that the user-specific directory actually exists before
moving forward.
Bug: 22472026
Change-Id: I33c8eed261a9c0d5acedd5be6133ed9990679d08
Refactor fstrim code to be encapsulated in unique task object, and
give it option of benchmarking when finished. Trimming now includes
both storage from fstab and adopted private volumes. Cleaner timing
stats are logged for each unique volume.
Add wakelock during ongoing async move tasks. Push disk sysfs path
to framework so it can parse any SD card registers as desired.
Bug: 21831325
Change-Id: I76577685f5cae4929c251ad314ffdaeb5eb1c8bf
In order to compare results from readlink() calls, we need to null
terminate the read value, otherwise we can end up doing an infinitely
recursive remount in the root namespace.
When remounting inside a namespace, unmount all existing mounts before
mounting the new storage into place. This also means we need to mount
the user-specific symlinks back into place.
Skip spinning up the FUSE daemon when not visible, otherwise we get
stuck waiting for a daemon that never shows up.
Bug: 22192518, 22204412
Change-Id: Icc7db822354ab7ffc47c39cd0611f65edecc32e5
Now that we're treating storage as a runtime permission, we need to
grant read/write access without killing the app. This is really
tricky, since we had been using GIDs for access control, and they're
set in stone once Zygote drops privileges.
The only thing left that can change dynamically is the filesystem
itself, so let's do that. This means changing the FUSE daemon to
present itself as three different views:
/mnt/runtime_default/foo - view for apps with no access
/mnt/runtime_read/foo - view for apps with read access
/mnt/runtime_write/foo - view for apps with write access
There is still a single location for all the backing files, and
filesystem permissions are derived the same way for each view, but
the file modes are masked off differently for each mountpoint.
During Zygote fork, it wires up the appropriate storage access into
an isolated mount namespace based on the current app permissions. When
the app is granted permissions dynamically at runtime, the system
asks vold to jump into the existing mount namespace and bind mount
the newly granted access model into place.
Bug: 21858077
Change-Id: Iade538e4bc7af979fe20095f74416e8a0f165a4a
Report both the disk and the partition GUID for private volumes to
userspace, and offer to forget the encryption key for a given
partition GUID.
Bug: 21782268
Change-Id: Ie77a3a58e47bf3563cdb3e4b0edfab1de4d0e6b4
vold will eventually use the serial numbers to clean up stale user
directories when mounting private storage devices.
Bug: 20275572
Change-Id: Ia29cb5da23e969f3087bb5caa5dc8f4e88f07613
When formatting volumes, pass along fsType string which can be "auto"
to let the volume select the best choice. For now, private volumes
assume that MMC devices (like SD cards) are best off using f2fs when
both kernel support and tools are present, otherwise fall back to
ext4. Use blkid when mounting to pick the right set of tools.
Move filesystem utility methods into namespaces and place in separate
directory to be more organized.
Bug: 20275581
Change-Id: Id5f82d8672dda2e9f68c35b075f28232b0b55ed4
Now that we're offering to store private app data on adopted storage
devices, the performance of those devices is much more important to
overall user experience.
To help set user expectations, this change offers to execute a
real-world benchmark on a storage device, returning a metric that can
be used to compare internal and external storage. The benchmark is
generated from the strace-instrumented storage access patterns of
typical apps.
A typical device completes the benchmark in under two seconds on
internal storage, a UHS-3 SD card is even faster (!), but a very slow
Class 4 SD card takes about 30 seconds to complete, giving us a clear
signal.
The measured benchmark numbers are logged along with information
about the storage device, such as manufacturer, model, etc. Card
serial numbers are scrubbed from output.
Bug: 21172095
Change-Id: I9b2713dafdfdfcf5d97bf1bc21841f39409a7e54
The VolumeManager::mountObb() creates a mapping between
a loopback device and a dm device. However the device-mapper
carries it out asynchronously, so there is a possibility that
Vold accesses to the dm device which is being built. Added
waiting for completion of the mapping in that function, like
mountAsec().
To verify install FrameworksCoreTests.apk and do:
adb shell am instrument -r -w -e class android.os.storage.\
StorageManagerIntegrationTest#testMountTwoEncryptedObb \
com.android.frameworks.coretests/android.test.\
InstrumentationTestRunner
Change-Id: If42f4b7494bb2f8a8b72d106ad84b3e3bf91fd9b
We eventually should move back to per-disk locks, but use a giant
lock to keep development rolling forward. Also move force adoptable
flag to framework since, since encrypted devices don't have persisted
properties loaded early during boot.
Bug: 19993667
Change-Id: Ifa3016ef41b038f8f71fc30bc81596cfd21dcd2a
This is cleaner and more direct than the reverse of having the disk
publish child volume membership. Rename state constants to match
public API. Add state representing bad removal. Make it clear that
volume flags are related to mounting.
Send new unsupported disk event when we finish scanning an entire
disk and have no meaningful volumes.
Bug: 19993667
Change-Id: I08a91452ff561171a484d1da5745293ec893aec0
Otherwise we get really excited and trip over ourselves while
partitions are still being created.
Bug: 19993667
Change-Id: I034e56b3063a71d73f9311a945c05ea2ae255f7d
When a private volume is mounted, create an emulated volume above it
hosted at the /media path on that device. That emulated volume is
automatically torn down when unmounting the private volume.
Add "removed" state for volume, which signals to framework that
media has left the building, send when the volume is destroyed.
Bug: 19993667
Change-Id: I1f82b51de578ac5cfcc5d7b9a6fb44f6f25c775c
Sadly setexeccon() is process global, so we need to carefully ensure
that all exec() are mutually exclusive to avoid transitioning into
unwanted domains. Also, because we have several threads floating
around, we need to guard all our FDs with O_CLOEXEC.
Format all newly created volumes immediately after partitioning,
but silence all events emitted from those volumes to prevent the
framework from getting all excited. Unify all notify events under a
single codepath to make them easy to silence.
Sent SIGINT before escalating to SIGTERM when unmounting.
Bug: 19993667
Change-Id: Idc6c806afc7919a004a93e2240b42884f6b52d6b
This adds support for private volumes which is just a filesystem
wrapped in a dm-crypt layer. For now we're using the exact same
configuration as internal encryption (aes-cbc-essiv:sha256), but we
don't store any key material on the removable media. Instead, we
store the key on internal storage, and use the GPT partition GUID
to identify which key should be used.
This means that private external storage is effectively as secure as
the internal storage of the device. That is, if the internal storage
is encrypted, then our external storage key is also encrypted.
When partitioning disks, we now support a "private" mode which has
a PrivateVolume partition, and a currently unused 16MB metadata
partition reserved for future use. It also supports a "mixed" mode
which creates both a PublicVolume and PrivateVolume on the same
disk. Mixed mode is currently experimental.
For now, just add ext4 support to PrivateVolume; we'll look at f2fs
in a future change. Add VolumeBase lifecycle for setting up crypto
mappings, and extract blkid logic into shared method. Sprinkle some
more "static" around the cryptfs code to improve invariants.
Bug: 19993667
Change-Id: Ibd1df6250735b706959a1eb9d9f7219ea85912a0
Wire up new Disk and VolumeBase objects and events to start replacing
older DirectVolume code. Use filesystem UUID as visible PublicVolume
name to be more deterministic.
When starting, create DiskSource instances based on fstab, and watch
for kernel devices to appear. Turn matching devices into Disk
objects, scan for partitions, and create any relevant VolumeBase
objects. Broadcast all of these events towards userspace so the
framework can decide what to mount.
Keep track of the primary VolumeBase, and update the new per-user
/storage/self/primary symlink for all started users.
Provide a reset command that framework uses to start from a known
state when runtime is restarted. When vold is unexpectedly killed,
try recovering by unmounting everything under /mnt and /storage
before moving forward.
Remove UMS sharing support for now, since no current devices support
it; MTP is the recommended solution going forward because it offers
better multi-user support.
Switch killProcessesWithOpenFiles() to directly take signal. Fix
one SOCK_CLOEXEC bug, but SELinux says there are more lurking.
Bug: 19993667
Change-Id: I2dad1303aa4667ec14c52f774e2a28b3c1c1ff6d
ioctl(BLKGETSIZE) expects unsigned long
(8 bytes on 64 bit environment).
This is fixing fails in android.os.storage.StorageManagerIntegrationTest
(in FrameworkCoreTests).
To verify, install FrameworksCoreTests.apk and do:
adb shell am instrument -r -w -e class android.os.storage.\
StorageManagerIntegrationTest#testMountSingleEncryptedObb \
com.android.frameworks.coretests/android.test.InstrumentationTestRunner
Change-Id: Ib6d5c7490c02521c93f107c35ad0aac49f6a3f1a
Don't use faccessat(AT_SYMLINK_NOFOLLOW). In Android, AT_SYMLINK_NOFOLLOW
is ignored. In glibc, it returns counter intuitive results when a
symbolic link is encountered, returning true all the time even though
an open(O_NOFOLLOW) will eventually fail.
Instead, stat the file and check to see if it's a regular file,
not a directory or symlink or some other weirdness.
In addition, fix a bug where isAsecInDirectory would return
true ("-1") if the asec directory didn't exist. It should return
false.
Bug: 18867827
Change-Id: I33d90e9095fad36ce0f83fde105b70f72e4eaef4
-Wno-missing-field-initializers is used as well, but that is an
overzealous warning from initializing structs with {0} and not a
real warning.
bug 18736778 and 16868177
Change-Id: Iffde89cd7200d9a11193e1614f1819f9fcace30a
Previously this would fail if the framework wasn't stopped. The failure
would then stop full disk encryption. The fact that the unmount worked,
however, would then stop the second attempt from achieving anything.
Fix in line with current retry philosophy
We still need to figure out why Devmapper::destroy() fails at first.
Bug: 17301843
Change-Id: I405a36c832ccdebf2d904bef77f15eea174a6bfb
Resize is no-op when sector count is unchanged; the caller can't
anticipate how vold does its sector calculations.
After resizing, we need to mount the container read-write, so allow
the caller to request "ro" or "rw" mode.
Handle ENOTSUP when trying to fallocate() on some filesystems
Bug: 16514385
Change-Id: I0d3a378280d4c36d14f8108ff428102283d583fa
Sometimes when an sdcard is already mounted,
some info like uuid and label are not re-broadcast to new listeners.
The extra argument to list allows late listeners to catch up by asking
volume list to broadcast that info again.
Bug: 16253597
Bug: 16306775
Change-Id: Ie7d0c1132c22d307a5b2a0e50075a3716138d00b
Signed-off-by: Benson Huang <benson.huang@mediatek.com>
(cherry picked from commit 85f4700f44170b772697e627b3075dcb9137e1b7)
From Shawn Heo's patch:
Ext4 introduced reserved clusters to prevent costly zeroout, or
unexpected ENOSPC. The size is 2% or 4096 clusters, whichever
is smaller (http://lwn.net/Articles/546473/).
So, we need to allocate additionally this amount of free space
to asecs when vold create asec images. This is required when
Android runs on Linux kernel 3.10 or later.
see: https://android-review.git.corp.google.com/#/c/96160
Change-Id: Iacff16b8cf0314493c355fa741bcfa519f744d6c
Signed-off-by: Daniel Rosenberg <drosen@google.com>
ASECs formatted as ext4 can now be resized using vdc asec resize.
Refactored some common code.
Requires resize2fs.
Change-Id: Ie78bb6015114a7bc4af42b16d1f299322ffc1e2a
Signed-off-by: Daniel Rosenberg <drosen@google.com>
Changed ext4 to be 4kb aligned, and fat to be 32kb aligned.
Fixed issue that could potentially cause unencrypted ext4
ASECS to overwrite the ASEC super block when filled.
Change-Id: I890426c82ac9cbc65add85a8e3f5063504193c31
Signed-off-by: Daniel Rosenberg <drosen@google.com>
Must limit vold calls to fs_mkdirs() only when the volume is mounted.
If NOT, it will trigger selinux warning as follows.
audit(1398835637.785:8): avc: denied { write } for pid=137 comm="vold" name="sdcard0" dev="rootfs" ino=3191 scontext=u:r:vold:s0 tcontext=u:object_r:rootfs:s0 tclass=dir
Change-Id: I1113fc961cbdd8bbd2fcbf740c2f504628c8399d
Signed-off-by: Cylen Yao <cylen.yao@mediatek.com>
libselinux selinux_android_restorecon API is changing to the more
general interface with flags and dropping the older variants.
Also get rid of the old, no longer used selinux_android_setfilecon API
and rename selinux_android_setfilecon2 to it as it is the only API in use.
Change-Id: I1e71ec398ccdc24cac4ec76f1b858d0f680f4925
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
This will allow fine-grained labeling of the
contents of ASEC containers. Some of the contents
need to be world readable and thus should be
distinguishable in policy.
Change-Id: Iefee74214d664acd262edecbb4f981d633ff96ce
Signed-off-by: rpcraig <rpcraig@tycho.ncsc.mil>
Replace MINOR(dev_t) and MAJOR(dev_t) with minor and major,
which cast to int.
Cast int to uintptr_t before casting to pointer
Change-Id: I59375518f15d27f400fcd4f8a8dfe5ebdd8350e6
Paul Crowley
Jeff Sharkey
Jeff Vander Stoep
TreeHugger Robot
Jeff Sharkey
Keun-young Park
Gao Xiang
Elliott Hughes
Wei Wang
Chih-hung Hsieh
Daniel Rosenberg
Hidehiko Abe
Yu Ning
Daichi Hirono
Mateusz Nowak
yoshiyuki hama
Hiroaki Miyazawa
Nick Kralevich
Yabin Cui
Tim Murray
Paul Lawrence
JP Abgrall
Cylen Yao
Stephen Smalley
Colin Cross
Robert Craig