Claude review: misc: fastrpc: fix channel ctx ref leak when session alloc fails

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

Patch Review

**Correctness: Good.** The flow is clear:

1. `fastrpc_channel_ctx_get(cctx)` at line 1634 increments the refcount.
2. On success, `fastrpc_device_release()` eventually calls `fastrpc_channel_ctx_put(cctx)` at line 1610.
3. On failure of `fastrpc_session_alloc()`, the early return at line 1652 skips the release path entirely, leaking the reference. The added `fastrpc_channel_ctx_put(cctx)` before the return fixes this.

**One minor style nit:** The patch removes a blank line that existed between `kfree(fl)` and `return -EBUSY`:

```c
-
+		fastrpc_channel_ctx_put(cctx);
 		return -EBUSY;
```

This is fine — the blank line was arguably unnecessary, and inserting the put call there is the logical place (after freeing `fl`, before returning). The ordering is correct: `mutex_destroy` then `kfree` then `put` — the `cctx` doesn't depend on `fl` being alive, so the put can safely go after the free.

**No issues found.** The commit message accurately describes the problem, the Fixes tag references the correct commit that introduced the refcounting, and the stable CC is warranted since this is a resource leak that could prevent channel context cleanup.

---
Generated by Claude Code Patch Reviewer