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/msm/hdmi: switch to generic PHY subsystem
Date: Sun, 22 Mar 2026 04:56:09 +1000
Message-ID: <review-patch1-20260319-fd-hdmi-phy-v6-1-cefc08a55470@oss.qualcomm.com>
In-Reply-To: <20260319-fd-hdmi-phy-v6-1-cefc08a55470@oss.qualcomm.com>
References: <20260319-fd-hdmi-phy-v6-0-cefc08a55470@oss.qualcomm.com> <20260319-fd-hdmi-phy-v6-1-cefc08a55470@oss.qualcomm.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

This is the core patch (~5500 lines of diff) that removes the old PHY code =
from `drivers/gpu/drm/msm/hdmi/` and creates new standalone PHY drivers und=
er `drivers/phy/qualcomm/`.

**Critical: Missing compatible strings**

The old code in `hdmi_phy.c` handled these compatibles:
```
{ .compatible =3D "qcom,hdmi-phy-8084", .data =3D &msm_hdmi_phy_8x74_cfg },
{ .compatible =3D "qcom,hdmi-phy-8998", .data =3D &msm_hdmi_phy_8998_cfg },
```

The new `phy-qcom-hdmi-28hpm.c` only has:
```c
static const struct of_device_id qcom_hdmi_phy_28hpm_of_match[] =3D {
	{ .compatible =3D "qcom,hdmi-phy-8974", .data =3D &msm8974_hdmi_phy_cfg },
	{ },
};
```

And the new `phy-qcom-qmp-hdmi-msm8998.c` only has MSM8996 in its match tab=
le (if any). The `qcom,hdmi-phy-8084` compatible (which shared the 8x74 PHY=
 config) and `qcom,hdmi-phy-8998` are dropped, breaking those platforms.

**Use `device_get_match_data()` instead of `of_device_get_match_data()`**

In `phy-qcom-hdmi-preqmp.c`:
```c
const struct qcom_hdmi_preqmp_cfg *cfg =3D of_device_get_match_data(dev);
```

And in `phy-qcom-qmp-hdmi-base.c`:
```c
const struct qmp_hdmi_phy_cfg *cfg =3D of_device_get_match_data(dev);
```

The kernel community prefers `device_get_match_data()` which works for both=
 OF and ACPI matching, and avoids the `__iomem` casting warnings that `of_d=
evice_get_match_data()` can produce.

**Typo in Kconfig help text**

```
config PHY_QCOM_HDMI
	...
	help
	  Enable this to support HDMI PHY presend on pre-QMP Qualcomm platforms
```

"presend" should be "present".

**Typo in MODULE_DESCRIPTION**

In `phy-qcom-hdmi-preqmp.c`:
```c
MODULE_DESCRIPTION("Qualcomm MSMpreqmp HDMI PHY driver");
```

Should have a space: "Qualcomm MSM pre-QMP HDMI PHY driver" or similar.

**Duplicated helper functions**

`write16()`, `write24()`, and `read24()` are defined identically in both `p=
hy-qcom-hdmi-28lpm.c` and `phy-qcom-hdmi-28hpm.c`. These could be placed in=
 the shared `phy-qcom-hdmi-preqmp.h` header or a common source file to avoi=
d duplication.

**`pm_runtime_get_sync` vs `pm_runtime_resume_and_get`**

In `phy-qcom-hdmi-preqmp.c`:
```c
ret =3D pm_runtime_get_sync(dev);
if (ret < 0) {
	pm_runtime_put_sync(dev);
	return ret;
}
```

`pm_runtime_resume_and_get()` is the preferred pattern =E2=80=94 it handles=
 the error path automatically, avoiding the need for the manual `pm_runtime=
_put_sync()` on failure.

**Bridge code conversion looks correct**

The `hdmi_bridge.c` changes properly sequence the generic PHY calls:
```c
phy_init(hdmi->phy);
phy_configure(hdmi->phy, &opts);
phy_power_on(hdmi->phy);
...
phy_power_off(hdmi->phy);
phy_exit(hdmi->phy);
```

This correctly maps to the old `hdmi->phy->funcs->powerup()` / `powerdown()=
` flow.

---

---
Generated by Claude Code Patch Reviewer