Feature/usblogupload backup - #87

unit_test.sh appends RDK_PROFILE=TV to /etc/device.properties unconditionally. This can create duplicate entries on repeated runs and makes test setup non-idempotent. Follow the pattern used elsewhere in the repo (grep before append, and create the file if missing) to keep the environment stable.

This file uses strncpy in the stubbed get_current_timestamp() but does not include <cstring>/<string.h>. Some toolchains will fail to compile due to a missing declaration. Add the appropriate header (and consider using std::strncpy when including <cstring>).

cov_build.sh is invoked unconditionally. This script clones/builds dependencies and adds network + time cost to every unit test run, even when coverage isn’t requested. Consider gating this behind the coverage flag (or a separate explicit option) and/or caching dependencies in CI.

There is no longer a success-path test for create_temporary_directory() (the previous success test was removed), leaving only error-path coverage below. Add back a positive test that verifies the function returns 0, produces a valid path, and that the directory is created and cleaned up.

DeviceCompatibilityInvalidTest is currently a placeholder (EXPECT_TRUE(true)) and no longer exercises validate_device_compatibility(). This reduces coverage of a key gate in usb_log_main.c. Consider stubbing/mocking getDevicePropertyData("RDK_PROFILE", ...) to return a non-"TV" value and asserting the expected non-zero return code.

MainArgumentValidationTest is still a placeholder, and the file also contains tests that call usb_log_upload_execute() without preparing the required filesystem state (mount point directory must exist, log paths must be set up, etc.). As written, the suite is likely to be flaky/fail on a clean CI runner. Either create/cleanup the required directories in the fixture or mock the filesystem + validation dependencies so assertions are deterministic.

The test-local remove_directory() stub returns false when the directory does not exist, which differs from the production remove_directory() used by cleanup_temporary_files() (it treats “already removed” as success). This makes the unit tests validate behavior that the shipped code does not have. Align the stub semantics with production or explicitly mock remove_directory via the shared mock layer.

CreateUsbLogDirectorySuccessTest no longer verifies that the directory was actually created, and CleanupTemporaryFilesSuccessTest no longer verifies the directory was removed. With the current stubs, these tests can pass even if nothing is created/cleaned up. Add assertions on filesystem state (or use mocks that simulate state changes) so the tests validate observable behavior.

CopyFileAndDelete_Success calls fputs() on f without checking whether fopen() succeeded. If the working directory is not writable, this will crash the test instead of failing cleanly. Add an ASSERT_NE(f, nullptr) (or equivalent) before writing.

The new call to sh cov_build.sh introduces a supply-chain risk because cov_build.sh clones and executes remote GitHub repositories (e.g., rdkcentral/iarmmgrs, rdkcentral/rdk_logger, rdkcentral/telemetry, rdkcentral/common_utilities) using mutable branches without any integrity or version pinning. If any of those upstream repositories or the network path is compromised, an attacker could execute arbitrary code in your CI/test environment and potentially access secrets or tamper with build artifacts. To mitigate this, pin git clone operations to specific commit hashes or verified release tags and/or vendor the required build artifacts instead of executing unpinned remote code during the test run.