From mboxrd@z Thu Jan  1 00:00:00 1970
From: Claude Code Review Bot <claude-review@example.com>
To: dri-devel-reviews@example.com
Subject: Claude review: dma/contiguous: Fix broken build
Date: Tue, 31 Mar 2026 17:07:44 +1000
Message-ID: <review-patch1-20260330-dma-build-fix-v2-1-26d844575815@kernel.org>
In-Reply-To: <20260330-dma-build-fix-v2-1-26d844575815@kernel.org>
References: <20260330-dma-build-fix-v2-1-26d844575815@kernel.org> <20260330-dma-build-fix-v2-1-26d844575815@kernel.org>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0

Patch Review

**Commit message**: The description is clear about the problem and rationale. One sentence is grammatically broken:

> In the case where the default CMA region is not defined in the DT though used to be the case covered by the now removed dma_heap_cma_register_heap() in dma_contiguous_reserve().

This should read something like: "In the case where the default CMA region is not defined in the DT, this used to be the case covered by..."

**Forward declaration and stub** (`kernel/dma/contiguous.c`):

```c
+#ifdef CONFIG_OF_RESERVED_MEM
+static int rmem_cma_insert_area(struct cma *cma);
+#else
+static inline int rmem_cma_insert_area(struct cma *cma)
+{
+	return 0;
+}
+#endif
```

This is the correct pattern for conditionally-compiled functions in the kernel. The `#else` stub returning 0 means that when `CONFIG_OF_RESERVED_MEM` is disabled, the default CMA area silently won't be queued for legacy-name heap creation. This seems acceptable since:
- The "default_cma_region" name is still created via `dev_get_cma_area()` in the CMA heap driver.
- The legacy name is primarily relevant for DT-based systems, which would have `CONFIG_OF_RESERVED_MEM` enabled.

One minor concern: the forward declaration lacks `__init`. If the actual definition of `rmem_cma_insert_area()` (from commit 3a236f6a5cf2) is marked `__init`, there would be a section mismatch warning. I can't verify this since the function definition isn't in the drm-next tree I have access to, but it's worth checking.

**Call site replacement** (`kernel/dma/contiguous.c`):

```c
-		ret = dma_heap_cma_register_heap(dma_contiguous_default_area);
+		ret = rmem_cma_insert_area(dma_contiguous_default_area);
 		if (ret)
-			pr_warn("Couldn't register default CMA heap.");
+			pr_warn("Couldn't queue default CMA region for heap creation.");
```

Clean replacement. The updated warning message is more accurate for the new semantics.

**Header removal** (`include/linux/dma-buf/heaps/cma.h`): The header only contained the `dma_heap_cma_register_heap()` declaration. With the function gone, removing the entire header and its includes from `cma_heap.c` and `contiguous.c` is correct cleanup.

**Verdict**: The patch looks correct as a build fix. The only actionable item is the grammatical issue in the commit message and verifying the `__init` annotation on the actual `rmem_cma_insert_area()` definition.

---
Generated by Claude Code Patch Reviewer