Claude review: drm/bridge: ti-sn65dsi83: add test pattern generation support

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

Patch Review

**Verdict: Has a bug in the hdisplay halving logic.**

**Bug — hdisplay halved unconditionally instead of only for dual LVDS:**

The commit message says:
> "Active line length needs to be halved for test pattern generation **in dual LVDS output**."

But the code halves `hdisplay` whenever the test pattern is enabled, regardless of whether dual LVDS mode is active:

```c
le16val = cpu_to_le16(mode->hdisplay / (sn65dsi83_test_pattern ? 2 : 1));
```

On an SN65DSI83 (single LVDS only), enabling the test pattern would incorrectly halve the active line length. This should use the `dual_factor` introduced in patch 2:

```c
le16val = cpu_to_le16(mode->hdisplay / (sn65dsi83_test_pattern ? dual_factor : 1));
```

This way single LVDS (`dual_factor == 1`) gets no division, and dual LVDS (`dual_factor == 2`) gets the correct halving.

**Missing MODULE_PARM_DESC:**

The module parameter has no description:

```c
static bool sn65dsi83_test_pattern;
module_param_named(test_pattern, sn65dsi83_test_pattern, bool, 0644);
```

Adding a `MODULE_PARM_DESC` would help users discover and understand the parameter (e.g., `MODULE_PARM_DESC(test_pattern, "Enable LVDS test pattern output");`).

**Minor — global state for per-device feature:**

The module parameter is a file-scope `static bool`, meaning it applies to all SN65DSI83/84 instances in the system. For a debug/bringup feature this is probably acceptable, but worth noting. A per-device debugfs entry would be cleaner if there are multi-bridge use cases.

**Minor — runtime writability vs. effect:**

The parameter is 0644 (writable at runtime), but changes only take effect on the next `atomic_pre_enable` (i.e., next modeset). This is expected behavior but could surprise users. The `MODULE_PARM_DESC` could note this.

---
Generated by Claude Code Patch Reviewer