Claude review: phy: qualcomm: add MSM8974 HDMI PHY support

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

Patch Review

Adds PLL programming and rate calculation for the MSM8974 HDMI PHY, building on the infrastructure from patch 1 and the UNI PLL registers from patches 2-3.

**Comparison bug in `qcom_hdmi_msm8974_phy_pll_set_rate`:**

```c
+	if (pixclk == 297000) {
+		hdmi_phy_write(hdmi_phy, REG_HDMI_8x74_ANA_CFG2, 0x06);
+		hdmi_phy_write(hdmi_phy, REG_HDMI_8x74_ANA_CFG3, 0x03);
+	} else if (pixclk == 268500) {
```

`pixclk` is `hdmi_phy->hdmi_opts.tmds_char_rate` which is in Hz (e.g., 297000000), but the comparisons use kHz values (297000, 268500). These comparisons will never match. This should be either `297000000` / `268500000`, or `pixclk` should be divided by 1000 first. This is a functional bug.

**`qcom_uniphy_recalc` — refclk_div calculation:**

```c
+	rate *= (refclk_cfg >> 2) * 0x3 + 1;
```

The mask `0x3` applied after shifting right by 2 extracts bits [3:2] of refclk_cfg. But looking at the setup code:

```c
+	val = (ref_freq_mult_2 ? BIT(0) : 0) |
+		((refclk_div - 1) << 2);
```

The divider is stored in bits [3:2] as `(refclk_div - 1)`. The recalc multiplies by `(field & 0x3) + 1` which correctly recovers `refclk_div`. However, `0x3` should probably be written as a proper mask. The expression `* 0x3` specifically reads oddly — it looks like it's multiplying by 3 rather than ANDing. It's actually `& 0x3` implicitly due to being the full 2-bit field. Wait, re-reading: `(refclk_cfg >> 2) * 0x3 + 1` — this is `(field_value * 3) + 1`, not `(field_value & 3) + 1`. For `refclk_div=1`, stored as 0, this gives `0*3+1 = 1` (correct). For other values this seems wrong. This looks like it should be `((refclk_cfg >> 2) & 0x3) + 1`. This is another potential bug.

**Missing `Reviewed-by` for patch 4:** Unlike patches 2 and 3 which have Neil Armstrong's Reviewed-by, this patch does not. Being the most complex new patch, it would benefit from additional review.

Overall this is a well-executed series with a few bugs that need fixing before merging, particularly the missing compatibles (8084, 8998), the pixclk unit mismatch in patch 4, and the suspicious recalc formula.

---
Generated by Claude Code Patch Reviewer