From mboxrd@z Thu Jan 1 00:00:00 1970 From: Claude Code Review Bot To: dri-devel-reviews@example.com Subject: Claude review: pmdomain: bcm: bcm2835-power: Increase ASB control timeout Date: Sat, 14 Feb 2026 07:21:53 +1000 Message-ID: In-Reply-To: <20260213-v3d-power-management-v5-4-7a8b381eb379@igalia.com> References: <20260213-v3d-power-management-v5-0-7a8b381eb379@igalia.com> <20260213-v3d-power-management-v5-4-7a8b381eb379@igalia.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 > - start =3D ktime_get_ns(); > - > /* Enable the module's async AXI bridges. */ > if (enable) { > val =3D readl(base + reg) & ~ASB_REQ_STOP; > @@ -176,9 +174,10 @@ > } > writel(PM_PASSWORD | val, base + reg); > =20 > + start =3D ktime_get_ns(); > while (!!(readl(base + reg) & ASB_ACK) =3D=3D enable) { > cpu_relax(); > - if (ktime_get_ns() - start >=3D 1000) > + if (ktime_get_ns() - start >=3D 5000) > return -ETIMEDOUT; > } Moving the timestamp after the `writel()` is a good change -- the previous pl= acement counted the write latency against the timeout budget. The increase fr= om 1us to 5us is modest and, as the commit message says, negligible in a powe= r domain transition context. The commit message states this fixes intermittent "Failed to disable ASB mast= er for v3d" errors on BCM2711 during runtime PM suspend. Was this timeout val= ue determined empirically, and has 5us been sufficient in all testing? The co= ncern is whether this is papering over a deeper issue versus being a genuine = timing margin fix. That said, the change itself is correct. No issues found. --- Generated by Claude Code Patch Reviewer