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: drm: Suppress intentional warning backtraces in scaling unit tests
Date: Sat, 16 May 2026 10:56:14 +1000
Message-ID: <review-patch3-20260514-kunit_add_support-v11-3-b36a530a6d8f@redhat.com>
In-Reply-To: <20260514-kunit_add_support-v11-3-b36a530a6d8f@redhat.com>
References: <20260514-kunit_add_support-v11-0-b36a530a6d8f@redhat.com> <20260514-kunit_add_support-v11-3-b36a530a6d8f@redhat.com>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Patch Review

**Overall: Good practical demonstration.**

**Wrapping non-warning cases in suppression is harmless but unnecessary:**

```c
+	kunit_warning_suppress(test) {
+		scaling_factor =3D drm_rect_calc_hscale(&params->src, &params->dst,
+						      params->min_range,
+						      params->max_range);
+		KUNIT_EXPECT_SUPPRESSED_WARNING_COUNT(test, expected_warnings);
+	}
```

When `expected_warnings =3D=3D 0` (the non-error cases), the entire call is=
 wrapped in a suppression block unnecessarily. This adds trivial overhead (=
kzalloc, list add/remove) for the non-warning parameterized cases. An alter=
native would be to conditionally wrap only the `-EINVAL` cases, but the cur=
rent approach is simpler and more readable, and the overhead is negligible =
for tests. This is a reasonable trade-off.

**`INT_MIN` initialization:**

```c
+	int scaling_factor =3D INT_MIN;
```

The v10 changelog explains this was added to avoid "potentially uninitializ=
ed" warnings. This works but is defensive coding since `scaling_factor` is =
always assigned inside the block before the `KUNIT_EXPECT_EQ`. Still, it's =
harmless and the compiler might not see through the `for`-loop-based macro.

**`#include <linux/limits.h>` for `INT_MIN`:** Correct =E2=80=94 `INT_MIN` =
is defined there.

**CONFIG_BUG skip is correct:**

```c
+	if (expected_warnings && !IS_ENABLED(CONFIG_BUG))
+		kunit_skip(test, "requires CONFIG_BUG");
```

Without `CONFIG_BUG`, WARN_ON becomes a no-op and the count would be 0, fai=
ling the count expectation. Skipping only the warning-expected parameterize=
d cases is the right behavior.

---
Generated by Claude Code Patch Reviewer