Claude review: drm/amd/display: Increase DCN314 SR latency by 1us

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

Patch Review

**Correctness:** The patch consistently updates all relevant locations:

1. **`dcn314_clk_mgr.c`** — The `lpddr5_wm_table` has all four watermark entries (WM_A through WM_D) updated identically:
   - `sr_exit_time_us`: 30.0 → 31.0
   - `sr_enter_plus_exit_time_us`: 32.0 → 33.0

   This is correct — the difference between `sr_enter_plus_exit_time_us` and `sr_exit_time_us` remains 2.0μs (i.e., the enter time alone is still 2μs), which preserves the original relationship.

2. **`dcn314_fpu.c`** — The `dcn3_14_soc` bounding box is updated:
   - `sr_exit_time_us`: 16.5 → 17.5
   - `sr_enter_plus_exit_time_us`: 18.5 → 19.5

   Again, the 2μs delta between enter+exit and exit-only is preserved.

**Observations:**

- The DDR5 table (`ddr5_wm_table` at line 410) is **not** modified — only the LPDDR5 table is changed. This is consistent with the commit message stating the issue is specific to LPDDR5 modules. However, the soc bounding box values in `dcn314_fpu.c` are global (not memory-type-specific), so both DDR5 and LPDDR5 paths will see the increased latency from the FPU side. This seems intentional as a conservative safety margin, but the reviewer may want to confirm the author considered whether the FPU change could affect DDR5 configurations.

- The `Fixes:` tag references a freedesktop.org GitLab issue rather than a commit SHA. This is non-standard for the kernel — typically a `Fixes:` tag references a commit that introduced the bug (format: `Fixes: <sha1> ("commit subject")`). If this is intended for stable backports, a proper `Fixes:` tag with a commit hash would be needed. If the issue just tracks the bug report, consider using `Link:` or `Closes:` instead.

- The patch lacks a `Cc: stable@vger.kernel.org` tag, which would be appropriate if this is meant to fix a user-visible regression.

**Nit:** The commit message `[Why & How]` marker is an AMD-internal convention — no issue, just noting it.

**Verdict:** The change itself looks correct and safe. The main feedback items are:
1. Fix the `Fixes:` tag to reference a commit hash (or change to `Link:`/`Closes:`)
2. Confirm the FPU bounding box change (which affects all memory types) is intentional
3. Consider adding `Cc: stable@vger.kernel.org` if this should be backported

---
Generated by Claude Code Patch Reviewer