Claude review: [PATCH] drm/amd/display: Fix uninitialized variable which breaks full LTO

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

**Bug analysis is correct.** Looking at the current source (`amdgpu_dm.c`):

- Line 2570: `memset(&fw_meta_info_params, 0, ...)` — zeroed here
- Line 2574: `fw_meta_info_params.bss_data_size = le32_to_cpu(hdr->bss_data_bytes);` — populated here  
- Line 2578: `fw_meta_info_params.fw_bss_data = region_params.bss_data_size ? ...` — **BUG**: reads uninitialized `region_params`
- Line 2591: `memset(&region_params, 0, ...)` — `region_params` only zeroed *after* the buggy use

**The fix is correct:**
```diff
-	fw_meta_info_params.fw_bss_data = region_params.bss_data_size ? adev->dm.dmub_fw->data +
+	fw_meta_info_params.fw_bss_data = fw_meta_info_params.bss_data_size ? adev->dm.dmub_fw->data +
```

This changes the condition to read from `fw_meta_info_params.bss_data_size` which holds the same value (`le32_to_cpu(hdr->bss_data_bytes)`) and is already initialized. After the fix, `region_params.bss_data_size` is later set from `fw_meta_info_params.bss_data_size` at line 2594, so the data flow is consistent.

**Minor observations:**
- The v2 improvement (using `fw_meta_info_params.bss_data_size` rather than re-reading from the header) is the cleaner approach, as Nathan suggested.
- It's noteworthy that without LTO, the uninitialized stack variable apparently happened to be zero or non-zero in a way that matched the correct behavior — classic undefined behavior that "works" by accident until optimization settings change.
- The Fixes tag correctly identifies the commit that introduced the bug.

**No issues found.** This is a clean, correct one-line fix.

---
Generated by Claude Code Patch Reviewer