Claude review: drm/panel: Add support for Tianma TA066VVHM03 panel

[Claude Code Review Bot <claude-review@example.com>]

Patch Review

**1. Build dependency on unmerged `slice_per_pkt` field (blocking):**

The probe function sets:
```c
dsi->dsc_slice_per_pkt = 2;
```

This field (`dsc_slice_per_pkt`) does not exist in `struct mipi_dsi_device` in the current kernel tree. The cover letter notes this depends on a patch currently under review. This patch cannot be applied or built until that dependency is merged. The series should either explicitly note this with a `Depends-on:` tag or be resubmitted once the dependency lands.

**2. Missing cleanup in `prepare` error paths (bug):**

In `tianma_ta066vvhm03_prepare`, the error paths after `mipi_dsi_picture_parameter_set()` and `mipi_dsi_compression_mode()` just return the error without disabling GPIOs or regulators:

```c
ret = mipi_dsi_picture_parameter_set(ctx->dsi, &pps);
if (ret < 0) {
    dev_err(panel->dev, "failed to transmit PPS: %d\n", ret);
    return ret;  /* leaks: regulators, enable_gpio, reset_gpio */
}

ret = mipi_dsi_compression_mode(ctx->dsi, true);
if (ret < 0) {
    dev_err(dev, "failed to enable compression mode: %d\n", ret);
    return ret;  /* same leak */
}
```

Compare to the earlier error path after `tianma_ta066vvhm03_on()` which correctly cleans up. These later error paths need the same treatment: assert reset, deassert enable, and disable regulators. Consider using a common `goto` label.

Also note the inconsistent `dev` pointer usage: the PPS error message uses `panel->dev` while all other messages use the local `dev` variable. Use `dev` consistently.

**3. LPM mode toggling in `_off` (questionable):**

```c
static int tianma_ta066vvhm03_off(struct tianma_ta066vvhm03 *ctx)
{
    ...
    ctx->dsi->mode_flags &= ~MIPI_DSI_MODE_LPM;

    mipi_dsi_dcs_set_display_off_multi(&dsi_ctx);
    mipi_dsi_dcs_enter_sleep_mode_multi(&dsi_ctx);
```

This clears LPM, meaning display-off and enter-sleep commands are sent in HS mode. Most panels send shutdown commands in LP mode. This may be correct for this specific panel (vendor DT generated), but it's worth double-checking since it's unusual.

Similarly, the backlight callback toggles LPM off for brightness writes then back on:
```c
dsi->mode_flags &= ~MIPI_DSI_MODE_LPM;
ret = mipi_dsi_dcs_set_display_brightness_large(dsi, brightness);
...
dsi->mode_flags |= MIPI_DSI_MODE_LPM;
```

This is a common pattern in generated drivers but is inherently racy if multiple threads could call into the DSI driver simultaneously.

**4. Clock/refresh rate:**

The pixel clock calculation gives:
```c
.clock = (1080 + 24 + 4 + 10) * (2340 + 12 + 1 + 4) * 160 / 1000,
```

This yields ~421 MHz and ~160 Hz refresh rate. For a gaming phone (ROG Phone 3) this seems plausible, but should be verified against hardware documentation. The 160 divisor is the target refresh rate — is 160 Hz confirmed for this panel?

**5. MAINTAINERS entry is fine.** Kconfig/Makefile additions are properly sorted alphabetically.

**6. Minor:** The `_reset` function uses `GPIOD_OUT_HIGH` at probe (assert reset), then in the reset sequence does 0→1→0. With the active-low polarity from the DT (`GPIO_ACTIVE_LOW`), this means physical low→high→low. This should be verified against the panel datasheet reset timing requirements.

---
Generated by Claude Code Patch Reviewer