Claude review: treewide: Manually convert custom kernel_param_ops .get callbacks

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

Patch Review

These are the callbacks the Coccinelle script couldn't handle — ones that do multi-step string construction, pass the buffer through helper functions, or use `strcpy`/`memcpy`.

**`drivers/gpu/drm/i915/i915_mitigations.c`**: The conversion is well done. The trailing-comma-to-newline replacement now guards against overflow:

```c
+	if (!seq_buf_has_overflowed(buffer) && buffer->len > 0 &&
+	    buffer->buffer[buffer->len - 1] == ',')
+		buffer->buffer[buffer->len - 1] = '\n';
```

This is an improvement over the old code which unconditionally wrote `buffer[count - 1] = '\n'`.

**`drivers/char/ipmi/ipmi_watchdog.c`**: The `get_param_str` rework is notable — the old code called `fn(NULL, buffer)`, checked for errors, then appended a newline. The new code just calls `fn(NULL, buffer)` and returns its result. The newline is now appended inside `action_op`/`preaction_op`/`preop_op` themselves via `seq_buf_printf(outval, "%s\n", ...)`. This is correct — the newline responsibility is pushed into the individual ops.

**`drivers/acpi/sysfs.c`**: The multi-line debug layer/level output is nicely simplified — no more manual offset tracking.

**`drivers/scsi/fcoe/fcoe_transport.c`**: Small behavioral improvement — the old code used `IFNAMSIZ` as the buffer limit per transport name which was arbitrary; the new code relies on seq_buf's natural bounds. The empty-list "none" case was restructured to use an explicit `list_empty()` check instead of checking whether anything was written. Correct.

**`lib/dynamic_debug.c`**: Note that `param_get_dyndbg_classes()` still returns `-1` in the default case rather than a proper `-errno`. This is pre-existing and not introduced by this patch, but it's worth noting.

**Verdict: Well-done manual conversions. No correctness issues.**

---
Generated by Claude Code Patch Reviewer