The upgrade routine used to call KeyStore->del which purges the given
key blob from the keystore including all existing grants.
With this patch, upgrade only calls Keymaster::delete on the keyblobs
without purging it from the keystore. Also it only calls
Keymaster::delete once the upgrade key was successfully written to disk.
This patch also calls fsync on the directory containing keyblobs to
narrow the window in which keyblob may be lost due to power loss.
Bug: 110450771
Test: Upgrade path tested by manually creating a key, bumping the
patchlevel, using the key subsequently and inspecting the logs.
Change-Id: I89241b5d4033b270733ff61838ab9244fce28c60
This patch addes verbose logging whenever an error code in the vendor
error code range is returned by keymaster.
Bug: 123562864
Test: atest android.keystore.cts
Merged-In: Ifceece542d6f3536ad87d053145c7aa8dd6d6603
Change-Id: Ifceece542d6f3536ad87d053145c7aa8dd6d6603
Due to lack of thread safety it seems that protobuf values were
occasionally getting deleted during parsing, causing ubsan to register
an error down the line in the protobuf library.
Bug: 128991260
Bug: 128810613
Test: atest cts/tests/tests/keystore/src/android/keystore/cts
Change-Id: Iee7ec6195e1e0aa4b28a7484737f984ed389a75e
Keystore was conceptually single threaded. Even with the introduction of
Keymaster workers it was always assumed that the service dispatcher
thread was single threaded. The wifi keystore service, however, calls
into the keystore service concurrently.
This patch adds a lock around all keystore service entry points to make
sure all dispatcher executions are serialised despite being called from
both the binder and hwbinder service thread.
Bug: 128810613
Bug: 129145334
Bug: 128774635
Test: Regressions tested with Keystore CTS test suite.
Change-Id: I8c5602d2c2cb1dd9423df713037e99b247cee71f
The keystore backend used by racoon interprets the return value of
getKeyCharacteristics such that it thinks that it failed when it didn't.
Test: Initiate VPN connection with racoon.
Bug: 120024003
Change-Id: Ibe936a2d2d81181c10f0dd1075cc5ab3646f736e
The bluetooth user will need to encrypt and decrypt data to ensure
that the integrity of passwords can be verified. These provide
the needed permissions to create/remove keys and sign/verify using them.
Bug: 117993149
Test: Bluetooth operations work from the UI and unit tests pass
Change-Id: I9092c2c282f26b40cd15da84125e6e11354ec48b
This response code needs to be added in the condition where a super
encrypted key blob fails to be read in after a user changes their pin.
Currently, the error code being sent back is VALUE_CORRUPTED, which is
incorrect.
Bug: 118883532
Test: atest cts/hostsidetests/appsecurity/src/android/appsecurity/cts/AuthBoundKeyTest.java
Change-Id: I188948e6e2e66903ee259108db9b8d26d11ca92c
Merged-In: I188948e6e2e66903ee259108db9b8d26d11ca92c
A VTS test for the Wifi Keystore HAL is being created. The test
is run as root and attempts various operations that directly
use the Keystore service. By default that test will not be
able to perform necessary actions to exercise the HAL code,
such as creating keys for tests.
This change will enable the root user to perform all key
operations, but only on userdebug and eng builds. In addition,
the root user will be able to perform actions on behalf of the
wifi user; this is necessary as some operations in the Wifi
Keystore HAL assume the wifi user.
Bug: 120182820
Test: atest system/hardware/interfaces/wifi/keystore/1.0/vts/
functional/VtsHalWifiKeystoreV1_0TargetTest.cpp
Change-Id: Ic6eb5748e0e19b64a44c4bdf88a7074f7367db3d
libc++ has switched this from a class to a struct to match libstdc++.
The standard does not require either specifically, but Clang warns
about the mismatch: https://bugs.llvm.org/show_bug.cgi?id=39871.
Test: m
Bug: None
Exempt-From-Owner-Approval: Janitorial
Change-Id: Iffbd944f60eb363ef7cd8f9f2c2eea77ff967310
This change will reduce the number of files written to dropbox for the
purposes of keystore logging. Previously, if a ton of operations were
being done, a file would be created for each operation which led to
spamming of the dropbox directory.
Now, all equivalent operations are counted and only one copy is sent
along with the corresponding count over the time period of an hour. This
limits the number of files keystore can write to dropbox to 24 a day,
and will reduce the possible size of files since redundant operations
aren't being written.
Bug: 117823210
Bug: 110988360
Test: atest cts/tests/tests/keystore/src/android/keystore/cts
Change-Id: I79367aa7a8eb3679aace2058e128d06e513e25ea
NIAP certification finds that the 128 bit key size is insufficient
and requires a 256 bit key size. This change increases the
size of new master keys to 256 bits. Any existing master keys are
not changed and continue to be supported.
A new BlobType, TYPE_MASTER_KEY_AES256, is used to signal when a
key is the new larger size.
Bug: 121272336
Test: (1) Ran Keystore CTS tests against Walleye.
(2) Created keys in build without change, moved to build
with change and verified old key could be loaded and
used. Also, a new key could be created with the
increased size and could be reloaded after a reboot.
Change-Id: If00331c303e6cc7bc95a2ab624d0e19bec4e587e
In the future the key size for new master keys will increase.
To maintain backwards compatibility the size of the key
can no longer be assumed. To help communicate the actual
size of the key, it will be passed around in a vector.
Bug: 121272336
Test: Ran Keystore CTS tests against Walleye
Change-Id: I4c05acb15b77959f2bf89abbdc325904fffb497a
For NIAP certification keys need to be generated using SHA256 or
higher. Presently SHA1 is used. To satisfy this requirement,
SHA256 will be used for new keys. As the master key has recently
increased in size, the key size is used to determine if SHA1 is used
(for older keys) or SHA256.
Bug: 121272336
Test: Ran Keystore CTS tests against Walleye
Change-Id: I6099156173e04b22c6edafd9fb0e072f7201c5ee
* Add explicit to conversion constructors/operators
* Use NOLINT or NOLINTNEXTLINE to suppress warnings on intended converters
Bug: 28341362
Test: make with WITH_TIDY=1 DEFAULT_GLOBAL_TIDY_CHECKS=-*,google-explicit-constructor
Change-Id: I4ed5aea36fcdcd8dbda9a4be9607c5af606f2a08
/dev/urandom is not an approved random number generator
for NIAP certification. Changing to use BoringSSL's
RAND_bytes(), which is approved.
Bug: 121272336
Test: Ran Keystore CTS tests against Walleye
Change-Id: I579d140ef56c90b477b0d8989e3b02375681aee8
Why?: 1) Returning an int array is unsafe because it must be allocated in Java and C++ must not change the size. 2) List<Integer> is not supported by AIDL, but List<String> is. I decided it was simpler to pass back integers encoded as strings than to create yet another parcelable.
Bug: b/119616956
Test: ./list_auth_bound_keys_test.sh
Test: Temporarily modified settings app to call listUidsOfAuthBoundKeys
Change-Id: Ibf86864a5df1608a39f438745dde6f2f8c296b66
* changes:
Replace cast operator with getValue() for key store return codes
Check key store result with isOk() instead of casted value
Use stream operator to report result code
The key name is encoded into the filename containing the (encrypted) key
material.
Since the key name might contain characters that are not valid in a
filename, the name is encoded using a multi-character custom encoding
scheme.
However, the decoding function did not decode the key name correctly -
in particular, spaces were decoded to 'P', causing CtsVerifier tests
that install a key with a space in the name to fail (due to internal
inconsistency between the key names in KeyChain's DB and key names
obtained from Keystore).
Fix by correctly compensating for the "carrier" character.
Test: atest keystore_unit_tests
Bug: 116716944
Change-Id: I0326a9d9e6912b04bb13b3b350ead8ddcfcc12f8
Since getKeyCharacteristins uses a cache file it is it is no longer
guarantied to call upgrade key any more. So we have to put the upgrade
key logic back into begin. Also we need to get extract the key blob data
from the key blob object every time the key blob could have changed.
Test: enroll a password, bump the patch level, rebuild and flash.
Then attempt to unlock the device with the password.
Bug: 120063166
Change-Id: If91c30d3f0599452b43923255bb88fee490beb21
The KeyStoreService returns NO_ERROR if the key was found,
and another response code otherwise. All of these are
mapped to non-zero values. As a result, if a key's
existence was queried it would always respond "true",
regardless if it exists or not or if there was a permissions
error.
Test: Key existence can be successfully checked with the
keystore_cli_v2 tool.
Change-Id: Iffc2e155a61354f1fbffbece093b19e5cbc537fd
The value of the error code is not intended to be used
in checks, and instead isOk() should be used. A few places
were found which used the error codes directly via the
cast operator. To make it less likely this will happen
in the future unintentionally, the cast operator is being
removed. Some code still wants to access the error code
directly, such as when logging, so getValue() is added
for these cases.
Bug: 119771891
Test: Built for walleye successfully, basic operations with
keystore_cli_v2 tool work correctly.
Change-Id: I46e82d66dc4932472d8a5b2749ece08e398e7c88
The underlying value of the return code is not meaningful
in a check, instead isOk() should be used.
Bug: 119771891
Test: Built for walleye successfully, basic operations with
keystore_cli_v2 tool work correctly, new cts test
no longer finds a crash in keystore.
Change-Id: Id20612824677619cbbd0b2ba4a11b15fe5258ecb
