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: power: supply: sgm41542: Add SG Micro sgm41542 charger
Date: Tue, 28 Apr 2026 14:19:02 +1000
Message-ID: <review-patch2-20260427170914.5062-3-macroalpha82@gmail.com>
In-Reply-To: <20260427170914.5062-3-macroalpha82@gmail.com>
References: <20260427170914.5062-1-macroalpha82@gmail.com> <20260427170914.5062-3-macroalpha82@gmail.com>
X-Mailer: Claude Code Patch Reviewer
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
MIME-Version: 1.0

Patch Review

This is a large BSP-derived driver. Several issues:

**Bug: `power_supply_put_battery_info` called on error path.**
When `power_supply_get_battery_info()` fails, `bat_info` is uninitialized/invalid, but the code falls through to `power_supply_put_battery_info()`:

```c
	ret = power_supply_get_battery_info(sgm->charger, &bat_info);
	if (ret)
		dev_warn(sgm->dev, "sgm4154x: cannot read battery info\n");
	else {
		...
	}

	power_supply_put_battery_info(sgm->charger, bat_info);
```

This should be inside the `else` block, or `bat_info` should be initialized to NULL and the put function checked for NULL safety.

**Bug: `sgm4154x_hw_chipid_detect` failure is not fatal.**
The probe function logs the error but continues anyway:

```c
	ret = sgm4154x_hw_chipid_detect(sgm);
	if (ret)
		dev_err_probe(dev, ret, "Unable to read HW ID\n");
```

If the chip ID can't be read at all (I2C failure), probe should fail. The `dev_err_probe` return value is discarded.

**Missing error check for `sgm4154x_vbus_regulator_register`:**

```c
	sgm4154x_vbus_regulator_register(sgm);
```

The return value is ignored. If OTG regulator registration fails, it should be propagated.

**`sgm4154x_set_watchdog_timer` uses raw values instead of defines for the argument.**
In `sgm4154x_charger_set_property`:

```c
		sgm4154x_set_watchdog_timer(sgm, SGM4154X_WDT_TIMER_40S);
```

This passes the register bit value (`BIT(4)` = 16) but the function expects seconds (0, 40, 80, 160). This is a semantic mismatch. The call should pass `40`, not `SGM4154X_WDT_TIMER_40S`.

**Typo in file header:**

```c
 * Chrager driver for Sgm4154x
```

Should be "Charger".

**Style issues from BSP:**
- `enum SGM4154X_VINDPM_OS` uses uppercase for the enum tag, which is not kernel coding style.
- Workqueue naming uses `%s` format specifier but passes a string literal:
```c
	sgm->sgm_monitor_wq = alloc_ordered_workqueue("%s",
			WQ_MEM_RECLAIM | WQ_FREEZABLE, "sgm-monitor-wq");
```
This works but is odd; just use `"sgm-monitor-wq"` directly as the format string.

- The `REGCACHE_NONE` in regmap config is the default and doesn't need to be specified.

- The `i2c_client_get_device_id()` result (`id`) is used for `id->name` but there's no NULL check. If probed via DT only, `id` could be NULL. The model_name copy via `strscpy(sgm->model_name, id->name, ...)` would then crash.

---
Generated by Claude Code Patch Reviewer