|WARNING| [v2] test: defer timer initialization

[dpdklab at iol.unh.edu]

Test-Label: ai-code-review
Test-Status: WARNING
https://dpdk.org/patch/171627

_AI Code Review_

=== Patch Review: series.patch (via Claude) ===
Review date: 2026-10-19

# DPDK Patch Review

## Errors

None identified.

## Warnings

### Missing release notes update

The patch removes unconditional timer initialization from `app/test/test.c` and moves it into individual timer-dependent test files. This is a behavioral change that affects memory usage (removes a 3MB memzone allocation on ARM) and fixes a test failure.

While the change is internal to the test suite, the memory usage improvement and bug fix (eal_flags_file_prefix_autotest failure) are significant enough to warrant a release note entry in the "Fixed Issues" or "Tested Platforms" section.

**Suggested fix:**
Add a release note entry documenting the memory usage reduction and test fix.

## Info

### Verification of timer initialization pattern

The patch correctly handles the return value of `rte_timer_subsystem_init()` in all added locations:
- Checks for both `ret < 0` and `ret != -EALREADY` (handling the case where subsystem was already initialized)
- Prints an error message and returns `TEST_FAILED` on actual failure
- This pattern is consistent across all modified files

### test_pdcp.c return code handling

The initialization of `ret` was corrected from the implicitly uninitialized state in v1 (where it would only be set on the failure path) to explicit initialization. The v2 change properly sets `ret = TEST_FAILED;` after the timer init success, which is correct since subsequent failures in the function should return `TEST_FAILED`.

### test_timer_secondary.c refactoring

The factorization of the lcore count check is clean - moving the timer init outside the `RTE_PROC_PRIMARY` conditional makes sense since both primary and secondary processes need the timer subsystem initialized.

### Cc: stable at dpdk.org appropriate

The patch fixes a test failure (eal_flags_file_prefix_autotest running out of memory) and reduces unnecessary memory allocation. These are valid reasons for stable backport.

Review-Result: WARNING