Claude review: drm/bridge: Fix refcount shown via debugfs for encoder_bridges_show()

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

Patch Review

**Correctness: The fix is correct.** The `drm_for_each_bridge_in_chain_scoped()` macro (defined at `include/drm/drm_bridge.h:1490`) calls `drm_bridge_chain_get_first_bridge()` and `drm_bridge_get_next_bridge_and_put()`, holding an extra reference on the current bridge during each iteration. This inflates the `kref_read()` value by 1 in the debugfs output.

**Style nit — prefer `refcount - 1` over `--refcount`:**

```c
drm_printf(p, "\trefcount: %u%s\n",
           scoped ? --refcount : refcount,
           lingering ? " [lingering]" : "");
```

Using pre-decrement (`--refcount`) on a local variable is unnecessarily side-effectful. Since `refcount` is never used again, it works, but `refcount - 1` would be clearer and more idiomatic for a display-only adjustment.

**Over-engineering concern:** The patch adds 20+ lines (a `__` prefixed inner function, two wrappers, and a new `bool scoped` parameter) to fix a one-line display issue. A simpler approach would avoid the extra abstraction entirely. For example, `encoder_bridges_show()` could just call the existing function and the refcount adjustment could be handled more locally, or the original function could simply take the offset as a parameter:

```c
static void drm_bridge_debugfs_show_bridge(struct drm_printer *p,
                                           struct drm_bridge *bridge,
                                           unsigned int idx,
                                           bool lingering,
                                           unsigned int refcount_adj)
{
    unsigned int refcount = kref_read(&bridge->refcount) - refcount_adj;
    ...
```

This avoids needing both a `__` function and two thin wrappers, and it's more transparent about what's happening (subtracting an adjustment rather than a boolean that means "subtract 1").

Alternatively, the simplest possible fix would be to just adjust the refcount inline in `encoder_bridges_show()` without changing the helper at all — though that would mean duplicating the format string.

**Conceptual question:** Debugfs is a debugging interface. Arguably, showing the *actual* refcount (including the iterator's reference) is the most truthful thing to do, and the "expected" value of 3 is a convention, not a contract. A comment in the output or in the code explaining the +1 from the scoped iterator might be more appropriate than silently adjusting the value. That said, consistency between `allbridges_show` and `encoder_bridges_show` for the same bridge is a reasonable goal, so the fix is defensible.

**Summary:** The fix is correct and addresses a real inconsistency. Recommend simplifying the implementation (drop the double-wrapper pattern, use `refcount - 1` instead of `--refcount`) before merging.

---
Generated by Claude Code Patch Reviewer