(This is a merge of http://go/wvgerrit/97083.)
The switch from LoadKeys to LoadLicense broke entitlement licenses
entirely because the LoadLicense path in CryptoSession didn't include
any affordances for updating the KeySession, unlike the LoadKeys path.
This patch adds code to handle this.
Bug: 152814106
Test: CE CDM Unit Tests
Test: Android Unit Tests
Change-Id: Id0c33a566e17e6be8da04e12be4b0fc87559aa8f
In `Combine Decrypt Calls to OEMCrypto`,
OEMCrypto_ERROR_BUFFER_TOO_LARGE is used as the default error code to
signal fallback to legacy decrypt (sending subsamples separately).
The error code would not be updated if no calls to OEMCrypto were
made (e.g. 0-length buffers).
This change resets error code to OEMCrypto_SUCCESS after fall back to
legacy decrypt. Consequently, buffers that do no require OEMCrypto
calls would return OEMCrypto_SUCCESS.
Merge of http://go/wvgerrit/96743
Bug: 150188155
Test: GtsMediaTestCases
Test: DashTest#testWidevineH264AdaptiveWithRendererDisabling
Change-Id: Ib23803c51f16bc809bda5c2720e628e81f1df1dc
Merge from Widevine repo of http://go/wvgerrit/96783
This CL updates the reference code, unit tests, and adapter to use the
new v16 function OEMCrypto_LoadDRMPrivateKey. This is just an API
change to allow ECC support in the future. The reference code does not
yet support ECC certificates, and the CDM code assumes that all
certificates have an RSA key.
Bug: 152558018
Test: unit tests on taimen and w/v16 mod mock.
Change-Id: I0793b416513b81b3d74849f0b58dbdc91f075ac6
[ Merge of http://go/wvgerrit/95405 ]
There are several OEMCrypto functions that do not require an open
session to be called. This change updates the OEMCrypto functions
related to the Usage Table Header.
Bug: 150888316
Test: Linux unit tests and Android build
Change-Id: Ic879876dd190fb3e058bbe8e0cce37273030b105
[ Merge of http://go/wvgerrit/95404 ]
There are three situtations where error codes from a usage table
operations were not being mapped to unique CDM response types. These
particular errors provide useful information for the CDM during table
defragging.
Also fixed misspelled error code.
Bug: 150890014
Bug: 150891685
Test: Linux unit tests and Android build
Change-Id: I683abdd5fc0871317eede960ea36cfafac7e7f49
(This is a merge of http://go/wvgerrit/95003.)
To reduce the number of OEMCrypto calls on the decrypt path, the maximum
subsample size will now be cached after the first call to retrieve it.
Bug: 150018606
Test: Android Unit Tests
Test: CE CDM Unit Tests
Test: ExoPlayer high-bitrate playback on OEC v15
Change-Id: I0b5d38d8a082c0a127d2a47f112b76c64085bddb
(This is a merge of http://go/wvgerrit/94928.)
In OEMCrypto v16, we dropped support for 'cens' and 'cbc1'. However, we
did not redefine the pattern (0,0) to be a valid pattern for 'cbcs', even
though it was no longer being used to signal 'cbc1'. Instead, we made
the CDM reject CTR with a pattern ('cens') and CBC with a (0,0) pattern
('cbc1') to mirror the behavior of OEMCrypto v16.
However, some apps have been using 'cbc1' mode to decrypt audio in
'cbcs' content. This is normally not possible but is possible for a
subset of content. Furthermore, it is easy to do by accident because of
the way most packagers package 'cbcs' audio and the special significance
Widevine has historically given the (0,0) pattern.
This patch updates the CDM to not reject CBC with a (0,0) pattern but
instead treat it as 'cbcs' content. To decrypt it correctly, the pattern
is treated specially inside the CDM core and converted to the
recommended equivalent pattern — (10,0) — before passing the content to
OEMCrypto.
For more specifics, please see the design doc: http://go/vclfg
Bug: 150219982
Test: ExoPlayer Demo App 'cbcs' Content
Test: GTS 'cbcs' Content
Change-Id: I334ff15db5f7b7d62040a036ba6d17515c3caee4
[ Merge of http://go/wvgerrit/94483 ]
With OEMCrypto V16 comes a new potential error code from calls to
DecryptCENC(). WARNING_MIXED_OUTPUT_PROTECTION may be returned by
supporting devices if one of the output devices does not meet the
required HDCP level for the decryption key/license; however the output
is instead restricted (by OEMCrypto) to devices that are secure. This
warning is informative to the CDM; but no action can/should be taken
by the CDM.
In addition, if DecryptCENC() returns an error/warning, it is likely
that the same status code will be returned on subsequent calls to
decrypt within the same crypto session. To reduce the number of logs
the CDM produces within the same crypto session only changes in error
codes are logged.
Bug: 140825538
Change-Id: Iaf9da3f0c88889525f73f3153a5977c8416286bb
(cherry picked from commit d9c703ef9e)
Merged-In: Iaf9da3f0c88889525f73f3153a5977c8416286bb
Merge from Widevine repo of http://go/wvgerrit/94243
When generating a second provisioning request, a new session should be
opened because a session can only have one nonce for v16.
For Provisioning 3.0 devices, the OEM Cert's private key must be
loaded in the new session.
Test: WvCdmRequestLicenseTest.ProvisioningInterposedRetryTest
Bug: 135288420 Nonce reuse
Bug: 141655126 Cert reload
Change-Id: I8a96566142c4d4380e2bdd571e8d363a7a1f74d4
(This is a merge of http://go/wvgerrit/93829,
http://go/wvgerrit/93830, http://go/wvgerrit/93832,
http://go/wvgerrit/93833, and http://go/wvgerrit/93834 from the
Widevine repo.)
This implements the CDM code changes necessary to take advantage of
Combined Decrypt Calls on OEMCrypto v16. The result of this is that
WVCryptoPlugin is much lighter now because it can pass the full sample
down to the core in one call, but CryptoSession is heavier, as it now
has to handle more complex fallback logic when devices can't handle
multiple subsamples at once.
This patch also removes support for the 'cens' and 'cbc1' schema, which
are being dropped in OEMCrypto v16. This fixes an overflow in the code
for handling those schemas by removing it entirely.
This patch also fixes the "in chunks" legacy decrypt path to use larger
chunk sizes on devices with higher resource rating tiers.
Bug: 135285640
Bug: 123435824
Bug: 138584971
Bug: 139257871
Bug: 78289910
Bug: 149361893
Test: no new CE CDM Unit Test failures
Test: Google Play plays
Test: Netflix plays
Test: no new GTS failures
Change-Id: Ic4952c9fa3bc7fd5ed08698e88254380a7a18514
[ Merge of http://go/wvgerrit/89848 ]
Apps query a number of properties at initialization. The mediaDrm
API getProperty allows the query of a single property at a time.
This causes a series of requests. If no crypto
sessions are concurrently open, a series of expensive OEMCrypto
Initialization and Termination calls will occur.
In this change OEMCrypto termination is delayed. If an OEMCrypto
Terminate is followed in close succession by an Initialize, neither
will occur avoiding the overhead. A timer enables a countdown process.
If no session activity occurs, the timer will eventually terminate
OEMCrypto and exit.
Bug: 136282358
Test: Android unit/integration tests
Change-Id: I442b7919b4e7835c52583516c8bc64d0c150241d
This CL merges some missed pieces of the merge from Widevine repo of
http://go/wvgerrit/93405
Test: unit tests
Bug: 147879734, 147396294, 141247171
Change-Id: I3b92b983f68f6a529f109bc5dec91b313e3fd7fb
Merge from Widevine repo of http://go/wvgerrit/93404
This is the unit tests, reference code, and documentation for
OEMCrypto v16.2. Backwards compatibility should work for a v15
OEMCrypto.
Some review comments will be addressed in future CLs.
Bug: 141247171
Test: Unit tests
Test: Media GTS tests on bonito
Change-Id: I9d427c07580e180c0a4cfdc4a68f538d351c0ddd
Merge of http://go/wvgerrit/93404
This CL updates the Widevine CDM to support OEMCrypto v16.1
Test: Tested in 16.2 CL
Bug: 141247171
Change-Id: I69bd993500f6fb63bf6010c8b0250dc7acc3d71b
[ Merge of http://go/wvgerrit/89906 ]
The change allows the GetDecryptHashSupport method to return
an error.
Bug: 144851430
Test: WV android unit/integration tests
Change-Id: Ib3b95788adb21b5ed0daee51ad338f9674b04c3c
[ Merge of http://go/wvgerrit/89888 ]
A macro validates parameters in CryptoSession and return a
specified error code. Some error codes have been retired and replaced with
the error code PARAMETER_NULL.
Bug: 136123217
Test: Android unit/integration tests
Change-Id: I6ecbad53e87cce04dfd9ea27861400e83044cf5e
(This is a merge of http://go/wvgerrit/86383)
When Key Sessions were added to CryptoSession, the initialization of the
initial Key Session was placed at the end of the initialization of the
owning CryptoSession. That's all well and good except the block right
before that assumed that it was safe to abort initialization early in
order to swallow errors when setting up usage tables. As a result, if
anything caused usage table initialization to fail, it would leave the
CryptoSession without a Key Session, resulting in an inevitable segfault
further down the line.
There's no reason the Key Session can't be initialized first. This
change moves initialziation order around to avoid the bug.
Bug: 141021960
Test: CE CDM Unit Tests
Test: Android Unit Tests
Change-Id: Ic78005c831d2a24d7d6de22df54462b2bd7085f0
[ Merge of http://go/wvgerrit/84647 ]
[ Merge of http://go/wvgerrit/84648 ]
Replacing most instances of C's NULL with C++'s nullptr. Also changed
how a NULL check is performed on smart pointers. They provided an
implicit boolean operator for null checks, meaning the underlying
pointer does not need to be compared directly (as it was in some places
before).
Note that clang-format has performed additional changes to some of the
test files that have not yet been formatted.
Bug: 120602075
Test: Linux and Android unittests
Change-Id: I06ddebe34b0ea6dfecedb5527e7e808e32f5269a
[ Merge of http://go/wvgerrit/83423 ]
[ Merge of http://go/wvgerrit/83424 ]
[ Merge of http://go/wvgerrit/83425 ]
[ Merge of http://go/wvgerrit/83426 ]
[ Merge of http://go/wvgerrit/83427 ]
Types of cleanup:
- Removed function / class prefixes from the logs.
- Fixed log string format options to match the types passed
- Corrected small spelling mistakes / typos
- _Tried_ to make the log format more consistent
- Added static_cast<int> conversion on enumerations when logged
- Changed several LOGE to LOGW and vice versa
- Used LOGE if the triggering condition stops the method/function
from completing its task
- Used LOGW if the triggering condition changes the expected
outcome but does not stop the rest of the method/function's
task
- Changed several instances of `NULL` to `nullptr`
- Ran clang-format on files after cleanup
This is part of a larger code quality effort in Widevine DRM.
Test: WV linux unittests and WV Android unit tests
Bug: 134460638
Bug: 134365840
Bug: 136123217
Change-Id: I958ec70ef99eef95c38dbebd7a1acd62ef304145
[ Merge of http://go/wvgerrit/80484 ]
Clang-format has been run on files in core/src. clang-format has been turned
off for some blocks but otherwise no other changes have been made.
Bug: 134365840
Test: WV unit/integration tests
Change-Id: I6e509f25136f84d37de3d920084302f0f2c23dc4
(This is a merge of http://go/wvgerrit/77803)
In OEMCrypto 15.2, OEMCrypto_GenerateNonce() is a Session Initialization
Function, which means the global OEMCrypto lock must be held before
calling it. There's no harm updating our code to behave this way now,
even before OEMCrypto 15.2 implementations are ready, since it's a
stricter guarantee than the current behavior.
Bug: 131327826
Test: CE CDM Unit Tests
Test: Android Unit Tests
Change-Id: Icb6a86dea7298c99596fa9e3d90cec721ebcb8a6
Merge from Widevine repo of http://go/wvgerrit/78623
This updates the license request client identification to include
OEMCrypto build information.
Bug: 129070445
Test: ExoPlayer on crosshatch with mod mock
Change-Id: I0dbce0cca4e9810e14f60561e4e434f1dbcadfb6
Merge from Widevine repo of http://go/wvgerrit/75123
Merge from Widevine repo of http://go/wvgerrit/75114
This changes the encoding for the hash to be hex instead of base64.
Also, the bad frame number is initialized to 0 to make it easier to
debug. And the FDPT test app now uses the correct byte order.
Bug: 129100318
Test: unit tests, FDPT test app.
Change-Id: I296bab990125a4e18bec92f3316e8289a3b25a6b
[ Merge from http://go/wvgerrit/74924 ]
These were not previously being collected or were collected
insufficiently.
BUG: http://b/121090396http://b/112919252
Test: Unit tests, Gplay, Nflix, GTS
Change-Id: I32b7206cbe6071519b4a483fbcd0920dc1a26961
(This is a merge of http://go/wvgerrit/74628)
There is some old legacy behavior in CryptoSession for supporting
Chromecast. When a platform tries to use a combination of L3 and opaque
handles, it silently substitutes the type "clear buffers" instead.
No platform uses this behavior anymore. In fact, the CE CDM contains
code that explicitly prevents this case from being triggered. That code
has broken in every one of the last 3 releases. To allow the CE CDM to
end that bug farm, this patch removes the behavior that no one wants.
Bug: 112703532
Test: CE CDM Unit Tests
Test: Android Unit Tests
Change-Id: I7f70483fac46c75637da5378c5a8b1bf7a2c2860
(This is a merge of http://go/wvgerrit/71883)
This moves all the SSL code to privacy_crypto so we can use the
iOS-specific versions and not use any BoringSSL. The iOS version
doesn't support OEM certificates.
Note that the tests still use BoringSSL.
Bug: 126559819
Test: build_and_run_all_unit_tests.sh
Change-Id: Ib0fad5d95b283b6cd6e02d8a08bcf248c5900bc4
(This is a merge of http://go/wvgerrit/72867)
This patch replaces the previous static std::mutexes in CryptoSession
with shared_mutexes, allowing multiple readers to access the resources
they protect. For the shared fields, this means only Initialize(),
Terminate(), and the code that sets up the usage table headers needs
exclusive access. All other CryptoSession code is able to read these
fields in parallel.
For OEMCrypto, the static OEMCrypto lock is joined by a per-session
std::mutex, which are used in concert to enforce the OEMCrypto v15
threading guarantees.
On my machine this results in a noticeable increase in performance for
the parallel unit tests.
Bug: 70889998
Bug: 118584039
Test: CE CDM Unit Tests
Test: Android Unit Tests
Test: Jenkins Tests
Change-Id: Ie6332ae4926ed4f14af897685d37bfe63831b14f
[ Merge of http://go/wvgerrit/72703 ]
SRM is an optional feature and whether it is implemented is upto the
discretion of OEMs. If it is not, avoid logging this information.
Bug: 124391178
Test: WV unit/integration tests
Change-Id: If8d2b1e0b59fb11825f832a5d4259b03c482fd6b
(This is a merge of http://go/wvgerrit/71324)
This patch increases the granularity of the locking in CryptoSession
without substantially changing its locking semantics. Where before
there was a single |crypto_lock_| performing multiple duties, now
there are three locks:
1) |static_field_lock_|, which is used when needing to access the
non-atomic static member fields of CryptoSession.
2) |oem_crypto_lock_|, which is used when needing to call into
OEMCrypto.
3) |factory_lock_|, used only by the functions that interact with the
CryptoSession factory.
All the code in CryptoSession has been updated to use these locks. It
has also been updated to only hold them for the minimal amount of time
necessary, as opposed to holding them for a whole function. This should
help some with the ability of CryptoSession calls to happen
concurrently. To assist in taking locks in a consistent manner, two
helper functions, |WithStaticFieldLock()| and |WithOecLock()| have been
added. Also, for the very common case of reading |initialized_|, the
accessor |IsInitialized()| will read the value safely.
While changing all the code to lock differently, I found that some
places in CryptoSession were *not* locking before accessing static state
or calling into OEMCrypto. I have made these callsites consistent with
the rest of CryptoSession.
As a result of taking locks for only the minimum time necessary, it is
no longer necessary for functions to make assumptions about whether the
lock will already be held before they are called. Locks should not be
held while calling helper functions, and code should always take a lock
for the brief time it is necessary to do so.
In tests, including the concurrent unit tests coming in the following
patch, this code did not perform substantially better or worse than the
code that preceded it, but the hope is that it will experience less
contention on devices that are more resource-constrained than my
desktop, such as older game consoles.
This patch appears to address some real threading issues. Hopefully, it
will also make it easier to maintain soundness in the future and to
reason about when code in CryptoSession needs to take a lock.
This is the first step to implementing the "Finer-Grained Locking in
CryptoSession" specification. A future patch will make some of these
locks reader-writer locks, to allow even greater parallelism.
Bug: 70889998
Bug: 118584039
Bug: 123319961
Test: CE CDM Unit Tests
Test: Android Unit Tests
Test: GTS
Test: Play Movies
Test: Netflix
Change-Id: I346c04a5d9875723db54af33ee91772bf49ca12f
[ Merge of http://go/wvgerrit/71326 ]
Nonce flood, frame size, session and system invalidation errors
will now bubble up to the app. OEMCrypto v15 returns
OEMCrypto_ERROR_BUFFER_TOO_LARGE, OEMCrypto_ERROR_SESSION_LOST_STATE,
OEMCrypto_ERROR_SYSTEM_INVALIDATED and a variety of nonce errors.
These will be reported to HIDL as OUTPUT_TOO_LARGE_ERROR,
ERROR_DRM_SESSION_LOST_STATE, ERROR_DRM_INVALID_STATE and
ERROR_DRM_RESOURCE_CONTENTION.
Bug: 120572706
Test: Unit/Integration tests
Change-Id: Ida177300046327ce81592a273028ef6c3a0d9fd9
(This is a merge of http://go/wvgerrit/70667)
Request ID Index generation has historically worked by incrementing a
shared variable in one place and reading it in another place and
trusting the fact that CdmLicense calls these operations in a certain
order and only once per session to give each session a unique value.
This patch cleans this up a bit, having each session store the current
Request ID Index at the same time as it stores its Request ID Base. This
guarantees that each CryptoSession will receive a unique but stable
combination of Base and ID rather than relying on the calling pattern.
Since all this generation happens during the same function, the full
Request ID can be generated up-front and stored, making
GenerateRequestId() no longer necessary.
This patch also simplifies the threading story around this shared state
by using a std::atomic<uint64_t>. Bringing the code that interacts with
the shared state together into one place and replacing it with atomic
operations will simplify locking around this code when CryptoSession
locking is revamped in a future patch.
Bug: 70889998
Bug: 118584039
Test: CE CDM Unit Tests
Test: Android Unit Tests
Change-Id: I12d2f6501f872f1973e5a9af5125ca03f23e5a56
(This is a merge of http://go/wvgerrit/70303)
This adds a platform.h file to abstract some of the differences
between Windows and POSIX platforms. This includes ntohl, setenv,
and ssize_t.
Bug: 122953649
Test: Android Unit Tests
Change-Id: I3235f3f284b53d24d7365ff3f4a06dcd9b403697
[ Merge of http://go/wvgerrit/69724 ]
Some queries no longer require a session to be opened before they
can be answered - security level, current HDCP level, max HDCP level,
usage support, number of open sessions, max sessions,
OEMCrypto API version, current SRM version, SRM update support,
resource rating tier and OEMCrypto build information.
b/117104043
Test: WV unit/integration tests
Change-Id: I92f8249e5599860da8cbf42d3b16f25515a46c55
[ Merge of http://go/wvgerrit/68083 ]
Add ability to query decrypt hash support, set a hash computed over a frame
and retrieve the last error at a later point.
Bug: 34080802
Test: WV unit/integration tests. New tests added to cdm_engine_test,
libwvdrmdrmplugin_hidl_test and request_license_test.
Change-Id: I7548c8798c873a6af3e1cfc0df57c117e1e474a6
[ Merge of http://go/wvgerrit/67884 ]
Now that we can use C++11, we should use the cross-platform std::mutex
type, not the custom pthread version.
Bug: 111850982
Test: WV unit/integration tests
Change-Id: If2fde2836826c5184609e6b1f3a6511206bd4594
[ Merge of http://go/wvgerrit/67324 ]
This introduces the ability to query resource rating tier information
through the plugin and CDM. Resource rating tiers are also
sent in the client identification portion of the license request.
Bug: 117112392
Test: WV unit/integration tests
Change-Id: I68ac6dfc4362f61150af822bd526e346b5cc4bf7
(This is a merge of http://go/wvgerrit/66643)
The sub-license feature has been removed from the server and packager.
So that we do not have to continue maintaining the code that supports
this feature that never shipped, I am removing it from the CDM as well.
Bug: 113165466
Test: CE CDM Unit Tests
Test: Android Unit Tests
Change-Id: I5d25844b161e74aa19adf19a29c56e4881aa7304
Merge from master branch of Widevine repo of http://go/wvgerrit/66078
Merge from oemcrypto-v15 branch of Widevine repo of http://go/wvgerrit/64022
This CL updates OEMCrypto ref code, unit tests, and core code for
setting the sandbox id before initializing OEMCrypto.
Test: unit tests only
Test: tested as part of http://go/ag/5501993
Bug: 115834255
Change-Id: Id9831680fe4db1c69413815931cae4bc80df0c01
Merge from master branch of Widevine repo of http://go/wvgerrit/66073
Merge from oemcrypto-v15 branch of Widevine repo of http://go/wvgerrit/64083
As part of the update to v15, LoadKeys, RefreshKeys, and
LoadEntitledContentKeys should all use offsets and lengths into the
message rather than a pointer for its parameters. The CDM, tests,
adapters, and OEMCrypto implementations are changed to reflect this.
Test: tested as part of http://go/ag/5501993
Bug: 115874964
Change-Id: I981fa322dec7c565066fd163ca5775dbff71fccf
