Claude review: drm/amdgpu/mes12: fix queue init wptr reset

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

Patch Review

**The change:**
```c
-		*(ring->wptr_cpu_addr) = 0;
+		atomic64_set((atomic64_t *)ring->wptr_cpu_addr, 0);
```

At `mes_v12_0.c:1479`, in the reset/suspend path of `mes_v12_0_queue_init()`.

**Correctness:** Same analysis as patch 1. The writer `mes_v12_0_ring_set_wptr()` (line 58) uses `atomic64_set()` and the reader `mes_v12_0_ring_get_wptr()` (line 76) uses `atomic64_read()`. The fix aligns the reset path with these.

**Fixes tag:** `Fixes: 785f0f9fe742 ("drm/amdgpu: Add mes v12_0 ip block support (v4)")` — reasonable, identifies when the v12 queue_init was introduced.

**Same observation about rptr:** `mes_v12_0_ring_get_rptr()` (line 67) uses `return *ring->rptr_cpu_addr` — 32-bit read — so leaving the rptr reset as a 32-bit store is consistent.

**No issues with this patch.**

---

**Summary:** Both patches are correct, minimal, and well-scoped. The commit messages are clear and honest about testing limitations (build-tested only, no hardware). The series makes the reset path consistent with the normal read/write accessors. The only thing worth noting to the author is that the cover letter claims this was audited on "Linux v6.18.21" which doesn't exist — v6.18 hasn't had 21 point releases. This is likely a version numbering mistake and doesn't affect the patches themselves.

---
Generated by Claude Code Patch Reviewer