[ Merge of http://go/wvgerrit/189590 ]
[ Cherry-pick of http://ag/26541307 ]
The CDM session shares its CryptoSession instance with a few additional
member objects (CdmLicense and PolicyEngine). When the CDM session's
crypto session is reset, it must also reset the CdmLicense and
PolicyEngine otherwise, a potential stale pointer reference may occur.
Test: request_license_test on Oriole
Test: run_x86_64_tests
Bug: 311239278
Change-Id: Ie175513ae652dcd96e12e5e1def574a8a56d5863
[ Merge of http://go/wvgerrit/173290 ]
* Renew timer offset from when license is loaded verifies that the
rental duration has not expired and begins decryption.
* Renew timer offset from first decrypt bugfix
* Feature is enabled based on oemcrypto v18 presence
* Renewal logic verifies that |can_renew| is enabled
* Unit tests were added to reflect use cases from duration
and renewal documentation
Bug: 278751387
Test: policy unittests, CdmUseCase tests, wvts tests
Change-Id: I3070b3f31b316e150c28ebe38d0440ab1eeb89b9
[ Merge of http://go/wvgerrit/169450 ]
OEMCrypto v17 introduced several new HDCP levels that OEMCrypto may
report; however, the CDM never updated to support them. The enum
values of the additional levels are no longer sequential with their
level of support (v1.1 is 7, and v2.1 is 3), this requires more
considerations when comparing the required HDCP levels (as specified
by the license) and current HDCP level supported by OEMCrypto.
The following rules were used:
1) HDCP_NONE is the absolute lowest level
2) HDCP_NO_DIGITAL_OUTPUT is the absolute highest level
3) HDCP_V1 is treated as equal to all V1.x levels
4) All other versions are based on their major-minor pairs
Bug: 269671291
Test: license_unittest
Test: policy_engine_constraints_unittest
Test: policy_engine_unittest
Test: GtsMediaTestCases
Change-Id: Ibecfcb981d7e019c68cb8e0c7286222253d18369
1. "Change CdmResponseType from enum into a struct"
Merged from http://go/wvgerrit/163199
Bug: 253271674
2. "Log request information when server returns 401"
Bug: 260760387
Bug: 186031735
Merged from http://go/wvgerrit/162798
3. "Specify server version on the command line"
Bug: 251599048
Merged from http://go/wvgerrit/158897
Test: build android.hardware.drm-service.widevine
Test: Netflix and Play Movies & TV
Test: build_and_run_all_unit_tests.sh
Bug: 253271674
Change-Id: I70c950acce070609ee0343920ec68e66b058bc23
[ Merge of http://go/wvgerrit/160000 ]
OEMCrypto v15 licenses made use of several now-obsolete API functions
of OEMCrypto (mainly LoadKeys and RefreshKeys). All license handled
by the CDM must be v16 or newer. The CDM can now rely on all license
requests/responses containing a core message, using v16 policy timers,
and requires loading using LoadLicense() / LoadRenewal().
Bug: 252670759
Test: run_x86_64_tests and policy_engine_unittest
Change-Id: I3f65a6ec0326b4c89d1919b8911e065079cb90d2
[ Merge of http://go/wvgerrit/124063 ]
LicenseDurationRemaining used to indicate the minimum of rental or
license duration till OEMCrypto v16. OEMCrypto v16 onwards it began
reporting rental duration alone.
This is confusing for app developers and content partners. Keeping
LicenseDurationRemaining as apps may depend on it but adding
RentalDurationRemaining for clarity.
Bug: 186838303
Test: WV unit/integration tests, WvCdmRequestLicenseTest.QueryKeyStatus
Change-Id: I6c507150a0945ee36716b4da189f5741b092c0ec
[ Merge of http://go/wvgerrit/108064 ]
The Widevine License Agreement has been renamed to use inclusive
language. This covers files in the core directory.
Bug: 168562298
Test: verified compilation (comment only change)
Change-Id: I8ae5a10cbfdf7faae6a2735e57b33729763f10b8
[ Merge of http://go/wvgerrit/106325 and http://go/ag/12644840 ]
When offline licenses are restored, licenses and any renewals are processed.
License state evaluation occurs and notifications are sent to listeners.
If the license is expired, which is likely if a renewal is present,
the license state will transition to expired. Transitions out of
expired state are not allowed and the renewal has no effect.
If we work around this by allowing transitions out of expired state,
listeners will get notifications that keys have expired and then that are
usable soon after. To avoid delivering erroneous notifications we delay
evaluation of license state while the license and renewal are being processed.
Evaluation occurs at the last stage of license restoration when playback
information from the usage table is being restored.
This only need to occur for when licenses are being restored. In other
cases when a license or renewal is received, license state evaluation
and event listener notification needs to occur immediately.
Bug: 166131956
Test: WV unit/integration tests, GtsMediaTestCases tests
Change-Id: Ic8ade25316c5e20cc88de9225c43c24b28f21ac4
[ Merge of http://go/wvgerrit/93865 ]
This allows for handling of timer and clock values as supported when both
the license service and the OEMCrypto on the device support v16.
A flag based on a value in the SignedResponse license indicates
whether this support should be enabled. A new class PolicyTimerV16
performs the duration value evaluation.
Bug: 139372190
Test: Android WV unit/integration tests
Change-Id: Iacbbd51ad26c9f29cb5418ff832f8822982644b7
[ Merge of http://go/wvgerrit/93838 ]
Some more rework of policy engine/policy timers code to support
timer and clock value handling introduced by OEMCrypto v16.
Changes are
* renamed methods to include rental duration since policies for v16 use
rental and playback duration for all licenses. Previously rental and
playback durations enforced timing for persistent licenses and license
duration was used for streaming licenses.
* Moved some common code to the base PolicyTimer class from
PolicyTimerV15.
* Corrected data member naming (policy_timers -> policy_timers_)
* Updated comments
Bug: 139372190
Test: Android WV unit/integration tests
Change-Id: Id925ddcc14608a8500f30c2c68486d91608a9abe
[ Merge of http://go/wvgerrit/93743 ]
Reworks policy engine in preparation for changes to support timer and
clock value handling by OEMCrypto core messages in OEMCrypto v16.
No major functional changes have yet been introduced. Time and duration
evaluation has been devolved to a new policy timer class. Policy
specific to licenses that do not support OEMCrypto core messages
is handled by a Policy Timer V15 class. This ensures backward compatibility.
Backward compatibility may be needed if
* OEMCrypto has not been upgraded to v16
* Licenses were persisted before the device was upgraded to v16
* License service does not yet support core messages
Some minor changes to when the current time was retrieved required
minor modification to test expectations.
Bug: 139372190
Test: Android unit/integration tests
Change-Id: I420fb181f656ed9a6bfe54f09e8b398c130d23da
[ 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
[ Merge of http://go/wvgerrit/77506 ]
Callers of MediaDrm can register to be notified when key status
changes and if they are usable for decryption. A number of factors
are evaluated when making this determination. Key container security
level will now be included in the evaluation.
Bug: 78652608
Test: WV unit/integration test, GtsMediaDrmTest, Play movies playback
testing.
Change-Id: I20243e5cb160f7957e3239e8d05f715ff0ee6dd6
(This is a merge of http://go/wvgerrit/76063)
Now that we have C++11, many places that do string formatting or parsing
can be replaced with std::to_string() or one of the std::sto*() family
of functions. This patch updates places that do simple stringifying or
parsing to use these functions. Some parts of the code are left
untouched because they were using StringStream to do more complex
actions, such as constructing more complex output or checking the status
of the parsing.
Bug: 120599938
Test: CE CDM Unit Tests
Test: Android Unit Tests
Change-Id: I482dc234ecd7c6014fa9b6874387ff51e04b772f
[ Merge of http://go/wvgerrit/74627 ]
Limited duration licenses (where license duration is unlimited and
playback window is set to a small value) did not work correctly when
the initial decrypt call occured before the license was received.
Bug: 109653883
Test: WV unit/integration tests
Change-Id: I0738e97db525b239455e2ac93ab0bedd6611b311
[ 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/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 of http://go/wvgerrit/60240 ]
Since the method is not a general purpose check and only verifies that
the key can be used for a given security level the method
has been renamed PolicyEngine::CanUseKeyForSecurityLevel.
Bug: 115701771
Test: WV unit/integration tests
Change-Id: Icd6789538bb709d2a48c67bbd7bc810f4b000e14
Merge from Widevine repo of http://go/wvgerrit/47860
This CL updates the copyright notice to indicate that files
shared with partners are shared under the Widevine Master
License Agreement.
bug: 77926774
test: comment change only
Change-Id: I0423668111578b80fb39a932d763df2827e2dfc3
[ Merged of http://go/wvgerrit/39766 ]
The security level (software/hardware, decryption/decode)
in the policy that specified how the key was to be used was
not being respected for L3. Playback would either continue or
a vendor specific error would be thrown.
If the device cannot use the key as permitted by the policy
CryptoException#ERROR_INSUFFICIENT_OUTPUT_PROTECTION will be thrown.
Bug: 31913737
Bug: 31913439
Test: WV unit/integration tests
Test: Playback using playmovies and netflix. Cast playback using
playmovies.
Change-Id: If25735ab0f789108431115623cb236687c5ef818
These are a set of CLs merged from the wv cdm repo to the android repo.
* Correct error logging
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/40000 ]
In tests, we set the cipher list to avoid using insecure
ciphers when connecting to the provisioning/license service.
The result of setting the cipher list was being incorrectly
validated.
Bug: 64847919
* Move mips cache headers to clear_cache_function.h
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/39700 ]
Since the clear_cache function has been moved away from the dynamic
adapter, we need these conditional includes to be migrated as well for
MIPS.
* Comment out Level 3 debug call until merge
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/39761 ]
This call was introduced in go/wvgerrit/34260/. Since the haystack tool
in google3 still needs this merge, this should be commented out so the
tool can still build until the merge has finished.
* Add logging for MAC keys to mock
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/39740 ]
Bug: 70637842
* Move external interfaces into level3.h + refactor
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/39673 ]
As part of b/70523618, this CL moves interfaces that partners are
responsible for in Level 3 to level3.h so they can be visible as
part of the CDM release process. It also cleans up some of the
names of the files and adds documentation.
* Corrected close session logging level
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/39676 ]
Bug: 69460963
* Remove Security Level Path Backward Compatibility Support
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/39505 ]
From the android K release onwards certificates were stored in
security level specific directories. If upgrading from
previous releases persistent information needed to be moved
to those directories.
Since no device is likely to upgrade from J to Pi, comptibility
support can be removed.
Bug: 70160032
* Rename privacy_crypto_openssl To privacy_crypto_boringssl
Author: John W. Bruce <juce@google.com>
[ Merge of http://go/wvgerrit/37122 ]
Now that we no longer support OpenSSL in the Shared Source CDM, the name
of this file can be updated.
Bug: 67907873
Test: build.py x86-64
Test: wv_ce_cdm_unittest
Test: jenkins/linux_unit_tests
* Remove Conditional Compilation from OpenSSL/BoringSSL
Author: John W. Bruce <juce@google.com>
[ Merge of http://go/wvgerrit/39460 ]
This change removes the usages of conditional compilation to support
both BoringSSL and OpenSSL, as well as to support multiple versions of
the OpenSSL API. All code is now compiled against one of the two
versions of BoringSSL in third_party/.
Note that in some cases, the kit/ and legacy_kit/ versions of BoringSSL
had different APIs, so when removing the OpenSSL version compatibility
conditional compilation, sometimes the older branch was kept and
sometimes the newer branch was kept.
Bug: 67907873
Test: build.py x86-64
Test: wv_ce_cdm_unittest
Test: jenkins/linux_unit_tests
* Build CE & Jenkins CDMs With BoringSSL from third_party/
Author: John W. Bruce <juce@google.com>
[ Merge of http://go/wvgerrit/37120 ]
Up until now, integrators have been responsible for providing a
compatible crypto library for use by the CE CDM. (either OpenSSL or
BoringSSL) After this change, this decision will no longer be in their
hands. The CE CDM build will always use the copy of BoringSSL in
third_party/, which will be statically linked with our library with
hidden visibility. This allows us to better control what crypto library
we use and will prevent continuing problems with trying to support both
OpenSSL and BoringSSL.
Unfortunately, BoringSSL began using C++11 in mid-2017, and we can't
support C++11 right now. Until we can, we need to use a C++11-free
version of BoringSSL for libssl. The CDM itself will continue to use a
recent BoringSSL, as it only needs libcrypto. But the unit tests that
need libssl have to use the legacy version.
Bug: 67907873
Test: build.py x86-64
Test: wv_ce_cdm_unittest
Test: jenkins/linux_unit_tests
* Modified RNG for Level3 to use more entropy
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/39220 ]
Bug: 65165076
Modified seed generation to use an xor of clock_gettime and
client-implemented code to supply random seeds to the RNG. Modified the RNG
as well to use xoroshiro128+ instead of xorshift, since it uses more
than one seed/state (which are 64-bit) and has higher "statistical quality".
The default implementations for the seed generation use /dev/urandom.
* Configure base path for Level3FileSystem
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/39506 ]
This is in response to b/70354006. This change makes the
Android Level3FileSystem use the existing properties method
GetDevicesFilesBasePath for binderization. The same is done for the
Linux implementation.
* Add legacy_kit/ to BoringSSL Directory
Author: John W. Bruce <juce@google.com>
[ Merge of http://go/wvgerrit/38861 ]
This adds a second copy of BoringSSL to the third_party/boringssl/
directory. This second copy is pinned to the last revision of BoringSSL
not to require C++11 and is not updated by the UPDATE_BORINGSSL.sh
script. This second copy will be used to provide libssl to the tests on
devices that do not support C++11.
Once we support C++11 in the CDM again, this weight should be removed
and all targets should use the copy of BoringSSL in the kit/ directory.
Bug: 67907873
* Use Shared Libraries for Unit Tests
Author: John W. Bruce <juce@google.com>
[ Merge of http://go/wvgerrit/38860 ]
Some unit tests were using a statically-linked CDM instead of a
dynamically-linked one. (Or, in one case, trying to link both ways into
the same binary.) For now, we need to only link dynamically, so that the
unit tests and the CDM can use different versions of BoringSSL.
Long-term, we would like to test both kinds of linkage. (See b/69548115
for that.)
Some unit tests were also using a dynamicaly-linked CDM that was named
such that it appeared to be statically-linked. This patch renames some
targets to make the linkage clearer.
Bug: 67907873
* Change CDM_Backwards_Compatiblity_Tests to dedicated brances
Author: Fred Gylys-Colwell <fredgc@google.com>
[ Merge of http://go/wvgerrit/39003 ]
The build scripts used by CDM_Backwards_Compatiblity_Tests now pull
old versions of oemcrypto from the dedicated branches oemcrypto-v*,
which [will eventually] contain old oemcrypto versions, that build
with the current build system with a current boringssl version.
bug: 67907873
* Fix spacing on level3 header
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/38760 ]
* Correct Query status calls
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/38640 ]
Bug: 70160032
* Refactoring to allow encryption of client ID
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/37460 ]
The code has been restructured to allow encryption of client
identification in provisioning requests. This will be enabled
when server side changes have been made (b/69427217).
* Additional information is included in the Client Identification
portion of the provisioning request.
* Client identification will be encrypted with a service
certificate provided by the app/client. Platform changes
to enable passing this to core are needed. If a service certificate
is not provided, a default one associated with the production Keysmith
will be used.
* Switched APIs in CdmEngine to take a service certificate for
provisioning rather than licensing. Service certificates for
licensing are session based and passed as properties from platform
code.
Bug: 30737060
* Allow some CDM errors to be reported from multiple locations
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/38360 ]
This creates some CdmResponseType errors which may be reused
PARAMETER_NULL, NOT_INITIALIZED_ERROR, REINIT_ERROR.
I have made changes to a few classes to report these errors.
Will work on additional classes in a separate CL.
Bug: 69864404
BUG: 71650075
Test: WV Unit/integration tests
Change-Id: Icc048770d424ac537d11ff327cda2cb142da802d
These are a set of CLs merged from the wv cdm repo to the android
repo.
* Android build fixes
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/36322 ]
* Address android compilation errors and warnings
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/36300 ]
* Gyp cleanup and OpenSSL v10.1 support.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/36001 ]
OpenSSL 10.1 has a small number of incompatible changes.
A desktop system upgrade exposed some issue in the build scripts.
Specifically, the linux build was using both third_party/protobufs (2.6.1)
and the version installed on the system (3.0 in this case). The linux
cdm.gyp depended on cdm/cdm.gyp which caused that plus some
additional issues.
These changes are necessary to support g++ version:
g++ (Debian 6.3.0-18) 6.3.0 20170516
Also did some cosmetic rework on run_current_tests to make it easier
to figure out what is going on when something fails.
Also tweaked some of the compiler settings for g++ support (revisit
this later).
* Refactored Service Certificate encryption to allow encryption of arbitrary data.
Author: Thomas Inskip <tinskip@google.com>
[ Merge of http://go/wvgerrit/36141 ]
* Send cdm test requests to UAT.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/36221 ]
This change resolves the all of the
CdmDecryptTest/CdmTestWithDecryptParam.DecryptToClearBuffer
tests.
The license servers will return different keys and keyids.
Sending the request to staging returned key ids and keys that were
not matching what was expected in the unit tests.
* Fix for building L3 OEMCrypto with clang and libc++
Author: yucliu <yucliu@google.com>
[ Merge of http://go/wvgerrit/35740 ]
1. Include <time.h> for time(time_t*).
2. Create endian check union on stack. Clang may create const union
somewhere else, which may cause crash.
* Remove error result when a sublicense session does
not exist. This is not considered an error.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/36080 ]
* Set default mock handler for GetSupportedCertificateTypes
for all unit tests and removed the use of StrictMock from
MockCryptoSession.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/35922 ]
The handler for this was only set for one test and resulted
in a number of failures.
* Set default handler for GetHdcpCapabilities. For
now the default action is to call the real
GetHdcpCapabilities of crypto_session.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/36140 ]
I also changed the mock to a NiceMock to silence
responses to unexpected calls to GetHdcpCapabilities.
The default handler can be overridden as needed in
the individual tests.
This resolves the policy engine test failures.
* Finalize merge of cdm_partner_3.4 to master.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/35360 ]
This is the final set of updates to merge all v3.4.1
changes into master.
* Embedded license: Sublicense rotation.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/35360 ]
Handle sublicense rotation event.
* Embedded license: Initial license phase.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/34280 ]
Initial license phase - key loading subsession.
* Embedded license: generate session data.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/33722 ]
Generate session data and add it to the license request for
any embedded license material.
* Resolve missing symbol when building cd-cdm
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/35840 ]
* C++11: Replace OVERRIDE def with override keyword
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/35400 ]
BUG: 71650075
Test: Not currently passing. Will be addressed in a subsequent
commit in the chain.
Change-Id: I37d0cb17f255ac6389030047d616ad69f895748c
These are a set of CLs merged from the wv cdm repo to the android repo.
* Resolve intermittent decrypt error.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/35720 ]
The CdmSession's closed state was not properly
initialized resulting in intermittent
SESSION_NOT_FOUND_FOR_DECRYPT errors.
In CdmEngine::Decrypt the session is looked up by
the key id. A list of open sessions is acquired
by calling CdmSessionMap::GetSessionList and each
session in the list is queried to see if it has
the key.
In building the list in CdmSessionMap::GetSessionList,
sessions are only added to the query list *if* the session
is not closed.
The closed status was not initialized and during testing
the query list would not contain the session causing
CdmEngine::Decrypt to return SESSION_NOT_FOUND_FOR_DECRYPT
resulting in the ce cdm api returning widevine::Cdm::kNoKey.
* No support for pre- C++11 compilation.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/35381 ]
* Handle unaligned nonce pointer in RewrapDeviceRSAKey calls.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/35340 ]
The pointer points into a message and it may not be aligned.
Always copy the nonce into aligned memory before checking it.
BUG: 38140370
Add note to CHANGELOG for this.
* Compiler strictness: more checks and code cleanup.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/35300 ]
Use the switches proposed in b/38033653 (as much as possible - some
conflicts with protobufs and gtest prevent fully accepting them).
Switch to clang for x32 build; ensure that both x86-64 and x86-32 builds
compile and link cleanly.
BUG: 38032429
BUG: 38033653
This partially resolves b/38458986
* Android build fixes
Author: Rahul Frias <rfrias@google.com>
[ Merge of http://go/wvgerrit/35102 ]
These corrections address compile warnings and errors for android
and unit tests.
* Embedded License: Add sub license key sessions.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/33680 ]
NOTE: this adds the AddSubSession() method, but it is not yet being
used. Use and proper cleanup is in an upcoming CL.
* Embedded license: Add track label field.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/33660 ]
A new track label field (a string) is added to the key container and the
sub session data objects.
This field will be used in handling sub license requests.
* Embedded license: extract keys from init_data.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/33621 ]
* Embedded license: add protobuf messages.
Author: Jeff Fore <jfore@google.com>
[ Merge of http://go/wvgerrit/33620 ]
also sync the widevine header definition with recent naming changes.
* Improve handling of provisioning response errors.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/33600 ]
Separate out the case of no response and the case
where the message is believed to be a JSON+base64
message but it doesn't parse properly.
BUG: 71650075
Test: Not currently passing. Will be addressed in a subsequent
commit in the chain.
Change-Id: I3c86f1c54980b071aec7461ac58541836551f896
These are a set of CLs merged from the wv cdm repo to the android repo.
* Add CDM status return for decrypt blocked by HDCP.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/28062 ]
New status code is kKeyUsageBlockedByPolicy. It is returned by the decrypt()
call instead of kDecryptError or kNoKey.
Also shuffled the CDM status returns to define the EME-aligned codes
first, and added comments to highlight the differences in handling.
BUG: 37540672
* Change division and mod ops to relocatables
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/28600 ]
This is similar to I2dad1028acf295288cd10817a2bcff2513c053c9.
We should be using the relocatable functions instead of the
native division and mod operations.
* Cleanup Encrypted ClientID in provisioning request
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/28083 ]
b/36897239
Staging server does not support it (or the client is not constructing
it properly). Leave it disabled pending investigation.
* Certificate Provisioning fixes.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/28066 ]
Partial fix for BUG: 37482676
Partial fix for BUG: 37481392
Update service certificates, get rid of DEV/QA root certificate.
Provisioning request and response are base64 (web-safe) encoded.
Response is optionally JSON-wrapped.
Change ConfigTestEnv; clearer comments and a closer match to reality.
BUG: 71650075
Test: Not currently passing. Will be addressed in a subsequent
commit in the chain.
Change-Id: I79d3c4bf1124e5e0d3e4d40baead65a8266ea874
Below are a set of CLs being merged from the wv cdm repo to the android repo.
* Fix handling of OEM Cert public key.
Author: Srujan Gaddam <srujzs@google.com>
[ Merge of http://go/wvgerrit/27921 ]
This is a potential fix for b/36656190. Set aside public
key on first call to get the public key, and use it afterwards.
This gets rid of extra calls to OEMCrypto_GetOEMPublicCertificate(),
which has side-effect of staging the OEM private key.
This also fixes a problem where the public cert string was
not being trimmed to match the size returned by
OEMCrypto_GetOEMPublicCertificate().
* Complete provisioning request/response for Provisioning 3.0
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/27780 ]
Fix bug on provisioning request path where GenerateDerivedKeys()
was being called when preparing to generate the signature.
Add message signature verification, and call correct OEMCrypto
routine to rewrap the private key (OEMCrypto_RewrapDeviceRSAKey30).
* Implement Cdm::deleteAllUsageRecords()
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/27780 ]
Delete all usage records for current origin. Removes usage
records from file system and retains the PSTs. The deletes
any usage entries matching those PSTs held by OEMCrypto.
BUG: 35319024
* Remove stringencoders library from third_party.
Author: Jacob Trimble <modmaker@google.com>
[ Merge of http://go/wvgerrit/27585 ]
We have a fork of the stringencoders library that we use for base64
encoding. This reimplements base64 encoding to remove the extra
dependency and to reduce the amount of code.
* Add Cdm::deleteUsageRecord() based on key_set_id.
Author: Gene Morgan <gmorgan@google.com>
[ Merge of http://go/wvgerrit/27605 ]
Delete specified usage record from file system usage info and
from OEMCrypto.
BUG: 35319024
* Modifiable OEMCrypto
Author: Fred Gylys-Colwell <fredgc@google.com>
[ Merge of http://go/wvgerrit/24729 ]
This CL adds a new variant of the OEMCrypto mock code that adjusts its
behavior based on a configuration file. This is intended for
testing.
For example, a tester can set current_hdcp to 2 in the options.txt
file, push it to the device, and verify that a license is granted for
HDCP 2.0. Then the tester can edit the value of current_hdcp to 1 and
push the file to the device. Playback should stop because the license
is no longer valid.
This variant uses a real level 1 liboemcrypto.so to push data to a
secure buffer. That means we can test playback for a license that
requires secure buffers on an Android device with real secure buffers.
BUG: 35141278
BUG: 37353534
BUG: 71650075
Test: Not currently passing. Will be addressed in a subsequent
commit in the chain.
Change-Id: I58443c510919e992bb455192e70373490a00e2b6
[ Merge of http://go/wvgerrit/34900 ]
Querying information about licenses with uncommon duration values
did not, under all conditions, return expected values. This corrects
for licenses where
* playback and license durations were set to unlimited
* rental and license durations were set to unlimited
Bug: 65669869
Test: 9 new policy_engine_unittests
Test: WV unit/integration tests
Test: GTSMediaTestCases
Test: Playback using netflix and play movies
Change-Id: I6e48f82f1f194fb7e04491d33054dfe74d9465eb
[ Merge from http://go/wvgerrit/33440 ]
Introduce checks to guard against clock tampering.
Bug: 62037413
Test: WV unit, integration tests on angler
Test: Manual clock rollback test
Test: Playback testing (Netflix, Play movies) on sailfish
Test: GTS test on angler
Change-Id: I47938109adb5c0f5e9aefc58eb5dac156b9f16ef
(This is a merge of http://go/wvgerrit/31040)
Because the Policy Engine was only consulting the result of the Max-Res
Decode check when it was in kLicenseStateCanPlay and not in other states
that imply kKeyStatusUsable, like kLicenseStateWaitingLicenseUpdate, the
Max-Res Decode results would not be honored during the interval between
requesting a renewal and receiving the result. (Or until the key
expired.) This was particularly problematic for keys with renewal delays
less than ten seconds long, which would freeze the Max-Res state before
it had a chance to update for the first time, effectively disabling
Max-Res Decode until renewal was received.
Fixing this required changing how the Policy Engine and the
LicenseKeyStatus objects communicate about the changing usability state
of the LicenseKeyStatus objects. Before, a call to ApplyConstraints()
might calculate a Max-Res failure, but this failure would be pending
until the Policy Engine deigned to call ApplyStatusChange() again.
Without a call to ApplyStatusChange(), it could pend forever. This put a
burden on the PolicyEngine to poll the LicenseKeys with redundant
ApplyStatusChange() calls using the same CdmKeyStatus that the keys were
already in, just in case Max-Res had changed since the last necessary
call to ApplyStatusChange().
If the Policy Engine got the timing of these calls wrong, it would
result in Max-Res results being ignored. (as in the linked bug) If it
ever polled with the wrong CdmKeyStatus, it would update the
LicenseKeys' status when it did not mean to. It would be preferable if
this polling were not needed, so that the Policy Engine couldn't get it
wrong.
This patch changes the API between these classes so that when Max-Res
fails, the state change can be reported immediately instead of pending
until ApplyStatusChange() is called, eliminating the need for polling.
All state changes to the LicenseKeyStatus objects go through a unified
ApplyStatusChange() method that can update the CdmKeyStatus, resolution,
and/or HDCP level and report any resulting usability changes
immediately. This patch updates the unit tests to exercise this new API
instead of the old API.
Previously, the linked bug slipped past our unit tests because we only
test unrenewable, streaming licenses against Max-Res. This patch adds
several more variants to
policy_engine_constraints_unittest so that it tests six kinds of
license to provide better coverage.
Bug: 62393949
Test: build_and_run_all_unit_tests
Change-Id: I0dfdbf6b8ea39abb446089aef5f6ea0502e9b4c6
[ Merge of http://go/wvgerrit/25781 ]
The security level (software/hardware, decryption/decode)
in the policy that specified how the key was to be used was
not being respected for L3. Playback would either continue or
a vendor specific error would be thrown.
If the device cannot use the key as permitted by the policy
CryptoException#ERROR_INSUFFICIENT_OUTPUT_PROTECTION will be thrown.
Test: Verified by WV unit+integration tests.
Verified by WidevineDashPolicyTests
Verified by WidevineDashPolicyTests#testL3SoftwareSecureDecoderRequired,
testL3HardwareSecureCryptoRequired, testL3HardwareSecureDecodeRequired,
testL3SecureVideoPathRequired.
b/31913737
b/31913439
Change-Id: Ibfc7f3dd6fc7264e8cf9b0d33f6f8d619eed6c00
(This is a merge of go/wvgerrit/23686)
This patch removes the makefile lines that previously prevented the
Widevine DRM Plugin from being built as 64-bit on Android. 64-bit
builds are now fully supported.
Only one piece of CDM code has had to change. Due to a bug in
libprotobuf before v3.0, int64 values from protobufs are technically a
different type from int64_t values in code on some 64-bit
architectures. Both have the same in-memory representation at runtime
but are seen as distinct types by the compiler. The compiler will
automatically convert in most places, but template instantiation is
not one of them, so a few places that passed a Protobuf int64 directly
into a template had to be modified.
Please note that tweaks to the mediadrmserver (not covered by this
patch) are needed in order for it to run as 64-bit and load 64-bit DRM
Plugins. Please also note that, as we have no 64-bit L1 OEMCrypto on
any devices, using the 64-bit mediadrmserver and Widevine library will
make your device fall back to L3 for the time being.
Bug: 18949752
Test: OEMCrypto unit tests
Test: Widevine unit tests
Test: Google Play (on Marlin)
Test: Widevine GTS Tests (on Marlin)
Change-Id: Ib6cdf2dd1ff75a1c473cacdc5e22397caa0a656c
[ Merge of http://go/wvgerrit/22565 ]
When using the grace period, the CDM will need to override the values
given to use by the TEE (through OEMCrypto). Normally the first (and
last) decrypt times are stored securely by the TEE. To avoid extra
complexity in OEMCrypto, we will simply ignore the values given to us
by the TEE when using this feature.
However, the TEE will still enforce the (hard) license duration. So
only the rental/playback durations will be affected by malicious
editing of files.
b/34211676
Test: Reran unittests including newly added tests. All tests other than
some oemcrypto, request_license_test passed. Those tests failed with
or without this CL.
Change-Id: I6d7b5bfb669fd8603b474b68c2f7175b0c30901d
* CDM license protocol updates
[ Merge of http://go/wvgerrit/22789 ]
No functional changes (yet) - all tests in widevine_ce_cdm_unittest
run successfully.
* Address android test build failures
[ Merge of http://go/wvgerrit/22983 ]
Updates to the license_protocol.proto in go/wvgerrit/22789
did not include the integration tests for android.
b/34202048
Test: Reran unittests. All tests other than some oemcrypto,
request_license_test passed. Those tests failed with or without this CL.
Change-Id: Ib9041d397187859b8fcbc1b1f7d275f8c4ef6aba
[ Merge of http://go/wvgerrit/22564 ]
b/34211676
Test: All unittests other than some oemcrypto, request_license_test
passed. Those tests failed with or without this CL.
Change-Id: I20474339aa1777da2db3677c10f186726505ecc8
[ Merge of http://go/wvgerrit/22517 ]
b/34211676
Test: All unittests other than some oemcrypto, request_license_test
passed. Those tests failed with or without this CL.
Change-Id: I86a2ff041aae57ac46e9f9f7bac38ec4599a0fa7
[ Merge of http://go/wvgerrit/22516 ]
b/34211676
Test: All unittests other than some oemcrypto, request_license_test
passed. Those tests failed with or without this CL.
Change-Id: Ie973f468f9efd05bdafcf90164dae185a6ce11dc
[ Merge of http://go/wvgerrit/22237 ]
This only changes the existing fields of the policy.
License::Policy::license_duration_seconds represents the end time
(relative to the license start time) that the license can be used.
This overriding other times if this is earlier.
License::Policy::rental_duration_seconds represents the end time
(relative to the license start time) that the license can be used
before playback starts. Once playback starts, this no longer applies.
License::Policy::playback_duration_seconds represents the end time
(relative to the playback start time) that the license can be used after
playback has started.
b/34211676
Test: Ran new unittests and reran old tests. All tests other than some
oemcrypto, request_license_test passed. Those tests failed with or
without this CL.
Change-Id: I34e7e39a7ab864300806c557b480f093aec8e545
This CL merges several CLs from the widevine repo:
http://go/wvgerrit/18012 Add support for querying allowed usage for key.
http://go/wvgerrit/17971 Add per-origin storage.
http://go/wvgerrit/18152 Add OEMCrypto's generic crypto operations to CDM.
http://go/wvgerrit/17911 QueryKeyControlInfo => QueryOemCryptoSessionId
Note: numbering in wv_cdm_types.h was added in this CL and will be
back ported to wvgerrit in a future CL.
Change-Id: Idb9e9a67e94f62f25dc16c5307f75a08b3430b64
[ Merge of https://go/wvgerrit/16940 ]
An alternate scenario to renewing keys is to load the same keys in
a separate session and make use of them by using the session sharing
feature.
Session sharing involves iterating through a map of sessions and
returning the first session that contains the Key ID. In certain cases
(license about to expire) we might prefer an alternate session
be chosen.
Licenses may expire in two ways. Policy engine, driven by a 1 second
timer may detect expiry and send an asynchronous event. OEMCrypto may
also detect expiry based on information in the key control block
and return an error during decryption. It is possible that these
may differ by upto a second. This can lead to issues where decryption
fails but EVENT_KEY_EXPIRED is not generated till later.
It is possible to address this by using information from both timers
to notify the app about expiry. To implement this correctly will
add complexity and require synchronization between threads. To avoid
this an alternate solution is, if session sharing is used, to pick
the session that has a license with the longest remaining validity.
b/27041140
Change-Id: I398cc4c10ee3a2f192d4a0befe7c8a469dd5bf86
* Add CE test for incomplete remove()
[ Merge of http://go/wvgerrit/14658 ]
This depends on I064c053dd986a432865163aed5c9c3493f14340b to get
PolicyEngine to implement the EME semantics expressed in this test.
This also excludes another error code from causing an error log in
CdmEngine::AddKey, because this is actually an expected, handled
error in the CE CDM and it causes some confusing noise during testing
and development.
* Drop CdmEngine test main
[ Merge of http://go/wvgerrit/14693 ]
The command-line arguments are no longer in use anywhere, and
dropping the CdmEngine test's main allows me to add those tests to
the CE test suite.
* Add PolicyEngine::SetLicenseForRelease()
[ Merge of http://go/wvgerrit/14651 ]
In order to implement the EME-specified behaviors for load() &
remove(), some small changes are required in PolicyEngine.
According to EME, you should be able to remove() an active session.
This means that releasing a persistent session is not a separate load
operation. EME also states that the keys should be expired when this
is done.
Remove() is implemented using GenerateKeyRequest(type=release). This
leads to CdmLicense::RestoreLicenseForRelease, which in turn calls
PolicyEngine::SetLicense. When removing an active session, the policy
engine will have keys already loaded. The old behavior would cause
these keys to be reloaded. We need them to be expired, instead.
Once a remove() has been started, the keys should never be loadable
again. If a release confirmation is not received by the CDM, the
session should still be loadable. EME states that once a session has
had remove() called, then is loaded again later, there should be no
keys. Not that they should be expired, but not present. The old
behavior would cause these keys to be reloaded as usable.
This new method allows EME remove() and load() behaviors to be
faithfully implemented in the CE CDM. When removing an active
session, the old keys become expired. When removing a partially-
removed, newly-loaded session, no keys will be loaded at all.
This change does not affect any existing tests in core/.
New tests have been added in PolicyEngineTest to cover the behavior
of the new method.
Change-Id: Idd61487c277c9eadb4a044cb2a563e151442a548
Also fix a missing change for
"playback duration should override license duration".
Merged from Widevine CDM repo:
https://widevine-internal-review.googlesource.com/#/c/14435/
Bug: 21393975
Change-Id: Ibfcf3ae4c13db8944ea285bcc79b6312ea621e1b
[ Merge of go/wvgerrit/14360 ]
If within playback window, do not expire license on expiry of rental or
license duration. In this case playback duration will extend the license.
b/17791094
Change-Id: I26d255aa8f0287bd583ebdeec991c613d49d8f22
