[dimerr]

The internal-PHY model reported an invented ID (0x0007c0f1) and set the BMSR link bit. A real GK7101, with a 100 Mb/s full-duplex link up, reads:

BMCR 3100  BMSR 78e9  ID1 0000  ID2 0000  ANAR 01e1  ANLPAR cde1

There is no ID, and BMSR shows autoneg complete but never the link bit. U-Boot's phylib therefore found no PHY on the camera, while it worked in qemu. The model now reads like the silicon, so that failure shows here too.

Checked:

• the vendor-based kernel logs mii id = 0 and boots to login;
• U-Boot without an ID workaround reports Could not get PHY, and with one (OpenIPC/u-boot hi3516ev200 / hi3518ev300: u-boot "hardware decompress overtime" — HW decompressor unimplemented #60) it transfers a file over TFTP;
• test-gk710x.py and test-gk710xs.py pass.

[qodo-free-for-open-source-projects]

PR Summary by Qodo

Match GK7101 internal PHY register reads

🐞 Bug fix 🕐 10-20 Minutes

AI Description

• Match measured GK7101 BMSR, PHY ID, and link-partner register values instead of reporting invented
 values.
• Expose the missing PHY ID and link bit so emulation reproduces U-Boot’s hardware failure.

Diagram

graph TD
  Guest["Guest PHY driver"] --> Read["EPHY MMIO read"] --> Select{"Register index"} -->|BMSR| Status["BMSR 0x78E9"]
  Select -->|PHY IDs| IDs["Zero PHY IDs"]
  Select -->|ANLPAR| Partner["ANLPAR 0xCDE1"]
  Select -->|Other| Stored["Stored registers"]

Loading

High-Level Assessment

Use the measured register values directly. Retaining or making the invented PHY ID configurable would preserve guest compatibility at the cost of hiding the U-Boot failure this hardware model needs to reproduce.

Files changed (1) +12 / -7

Bug fix (1) +12 / -7

gk-ephy.cReturn silicon-matched internal PHY register values +12/-7

Return silicon-matched internal PHY register values

• Changes BMSR to 0x78E9, clears both PHY ID responses, and sets ANLPAR to 0xCDE1 to match observed GK7101 reads. The comments explain why a guest driver cannot rely on a nonzero ID or the BMSR link bit.

qemu/hw/misc/gk-ephy.c

[qodo-free-for-open-source-projects]

Code Review by Qodo

🐞 Bugs (0) 📘 Rule violations (1) 📎 Requirement gaps (0) 🎨 UX issues (0) 🔗 Cross-repo conflicts (0) 📜 Skill insights (0)

1. New PHY register values have no regression test 📘 Rule violation ⚙ Maintainability

Description

gk_ephy_read now returns BMSR 0x78E9, zero for both ID registers and ANLPAR 0xCDE1, but the PR
adds or changes no register-level test for gk-ephy. Neither test-gk710x.py nor test-gk710xs.py
reads the PHY window, so a later change that restores an ID or the link bit would pass both tests.

Code

qemu/hw/misc/gk-ephy.c[R49-55]

+    case 1:     /* BMSR: 100/10 FD/HD, preamble suppression, AN complete */
+        return 0x78E9;
+    case 2:     /* PHYID1 / PHYID2 */
   case 3:
-        return 0xC0F1;
-    case 5:     /* ANLPAR: the partner advertises the same */
-        return 0x45E1;
+        return 0;
+    case 5:     /* ANLPAR, as a switch advertised it to that GK7101 */
+        return 0xCDE1;

Evidence

Rule 9 requires regression coverage for changed device register behavior. A grep of the tests for
the PHY finds no check of the gk-ephy window in the GK710x test scripts, and the PR touches no
test file.

CLAUDE.md: Verify Device Wiring and Register Behavior with Tests: CLAUDE.md: Verify Device Wiring and Register Behavior with Tests
qemu/tests/test-gk710x.py[270-282]

Agent prompt

The issue below was found during a code review. Follow the provided context and guidance below and implement a solution

## Issue description
The `gk-ephy` read values (BMSR `0x78E9`, ID1/ID2 0, ANLPAR `0xCDE1`) are not covered by any register test.
## Fix Focus Areas
- qemu/tests/test-gk710x.py[121-290]
- qemu/hw/misc/gk-ephy.c[42-58]
## Recommended Fix
Add a `test_ephy` to `test-gk710x.py` that reads the PHY MMIO window (`ephy_base` from `qemu/hw/arm/goke.c`) at +0x4 (BMSR), +0x8 and +0xC (IDs), +0x10 (ANAR) and +0x14 (ANLPAR). It should assert `0x78E9`, 0, 0, `0x01E1` and `0xCDE1`, and that the link bit (`0x4`) is clear. Call it from `main`.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools

---

Tip of the day

💡 Did you know, you can enable the Remediation agent and Qodo fixes findings in a dedicated fix PR

More tips ↗ | Customize Qodo ↗ | Qodo docs ↗

[qodo-free-for-open-source-projects]

1. New phy register values have no regression test 📘 Rule violation ⚙ Maintainability

gk_ephy_read now returns BMSR 0x78E9, zero for both ID registers and ANLPAR 0xCDE1, but the PR
adds or changes no register-level test for gk-ephy. Neither test-gk710x.py nor test-gk710xs.py
reads the PHY window, so a later change that restores an ID or the link bit would pass both tests.

Agent Prompt

## Issue description
The `gk-ephy` read values (BMSR `0x78E9`, ID1/ID2 0, ANLPAR `0xCDE1`) are not covered by any register test.

## Fix Focus Areas
- qemu/tests/test-gk710x.py[121-290]
- qemu/hw/misc/gk-ephy.c[42-58]

## Recommended Fix
Add a `test_ephy` to `test-gk710x.py` that reads the PHY MMIO window (`ephy_base` from `qemu/hw/arm/goke.c`) at +0x4 (BMSR), +0x8 and +0xC (IDs), +0x10 (ANAR) and +0x14 (ANLPAR). It should assert `0x78E9`, 0, 0, `0x01E1` and `0xCDE1`, and that the link bit (`0x4`) is clear. Call it from `main`.

ⓘ Copy this prompt and use it to remediate the issue with your preferred AI generation tools