Claude review: drm/bridge: tc358762: Drop drm_bridge_funcs.mode_set

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

Patch Review

Replaces the deprecated `.mode_set` callback (which stored the mode into `ctx->mode`) with fetching the mode from atomic state in `tc358762_enable()`. Uses `drm_atomic_get_new_connector_for_encoder()` → `drm_atomic_get_new_connector_state()` → `drm_atomic_get_new_crtc_state()` → `crtc_state->mode`.

**Issue: Missing NULL checks.** The return values of `drm_atomic_get_new_connector_for_encoder()`, `drm_atomic_get_new_connector_state()`, and `drm_atomic_get_new_crtc_state()` can all be NULL. While in practice they shouldn't be NULL when called from `atomic_enable` (the CRTC/connector must be active), other bridge drivers (e.g., `dw-dp.c`, `adv7511_drv.c`, `analogix_dp_core.c`) do NULL-check the connector return. The sibling driver `tc358768.c` doesn't check either, but that doesn't make it right.

```c
connector = drm_atomic_get_new_connector_for_encoder(state, bridge->encoder);
conn_state = drm_atomic_get_new_connector_state(state, connector);
crtc_state = drm_atomic_get_new_crtc_state(state, conn_state->crtc);
mode = &crtc_state->mode;
```

A NULL `connector` here would cause a NULL-pointer dereference on the next line. Recommend at minimum a `WARN_ON(!connector)` guard with early return, consistent with the pattern in `adv7511_drv.c`.

Also uses `crtc_state->mode` rather than `crtc_state->adjusted_mode`. This is consistent with the original `.mode_set` which was passed the unadjusted mode, so it preserves behavior. But for a bridge that needs actual hardware timings, `adjusted_mode` is often more appropriate. Worth considering whether `adjusted_mode` would be better, though for this simple bridge the difference is likely negligible.

---
Generated by Claude Code Patch Reviewer