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/bridge: ite-it66121: Set TX mode in the .atomic_enable callback
Date: Sat, 16 May 2026 13:19:45 +1000
Message-ID: <review-patch1-20260512132232.333654-2-javierm@redhat.com>
In-Reply-To: <20260512132232.333654-2-javierm@redhat.com>
References: <20260512132232.333654-1-javierm@redhat.com> <20260512132232.333654-2-javierm@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

**Issue: Duplicate register writes not removed from `.mode_set`**

The patch adds a new `it66121_set_tx_mode()` called from `it66121_bridge_en=
able()`, but the existing code in `it66121_bridge_mode_set()` still writes =
the same registers:

```c
/* Enable AVI infoframe */
if (regmap_write(ctx->regmap, IT66121_AVI_INFO_PKT_REG,
		 IT66121_AVI_INFO_PKT_ON | IT66121_AVI_INFO_PKT_RPT))
	goto unlock;

/* Set TX mode to HDMI */
if (regmap_write(ctx->regmap, IT66121_HDMI_MODE_REG, IT66121_HDMI_MODE_HDMI=
))
	goto unlock;
```

These lines at the current tree's lines 796-803 are not removed. The diffst=
at shows `1 file changed, 18 insertions(+)` with zero deletions, confirming=
 the old code remains. The commit message says "it is more correct to do th=
is in .atomic_enable" but doesn't explain why the old writes are kept. Thes=
e should be removed from `.mode_set` to avoid writing the same registers tw=
ice per mode change, and to avoid confusion about which callback is the aut=
horitative one.

**Minor: Error handling style**

```c
if (regmap_write(ctx->regmap, IT66121_AVI_INFO_PKT_REG,
		 IT66121_AVI_INFO_PKT_ON | IT66121_AVI_INFO_PKT_RPT))
	goto unlock;

regmap_write(ctx->regmap, IT66121_HDMI_MODE_REG, IT66121_HDMI_MODE_HDMI);
```

The first `regmap_write` checks for errors and jumps to `unlock`, but the s=
econd `regmap_write` result is silently discarded. This is inconsistent =E2=
=80=94 either both should check for errors or neither should (the existing =
`.mode_set` code checks both). This is a minor nit since the function retur=
ns void and there's nothing to do on error except release the lock, but con=
sistency is preferred.

**Ordering observation**: The call to `it66121_set_tx_mode()` is placed *af=
ter* `it66121_set_mute(ctx, false)` in `.atomic_enable`. This means the dis=
play is unmuted before the TX mode is configured. On a DVI monitor receivin=
g HDMI-mode infoframes during this window, there could be a brief glitch. C=
onsider placing `it66121_set_tx_mode()` before the unmute. This ordering al=
so matters because in the existing `.mode_set`, the TX mode/infoframe setup=
 is done *before* the AFE configuration and *before* unmute =E2=80=94 movin=
g it after unmute changes the sequencing.

---
Generated by Claude Code Patch Reviewer