Hello @jianghao ,
I think there is a small terminology mix-up here regarding Auto‑Negotiation vs. RGMII interface clocking.
Auto‑Negotiation is not a function of RGMII. RGMII is just the parallel MAC↔MAC / MAC↔PHY interface (data + source‑synchronous clocks). The link partner “Auto‑Negotiation” (speed/duplex capability exchange) happens on the PHY line side (e.g., 100BASE‑TX / 1000BASE‑T / 100BASE‑T1 etc.), not on the RGMII pins. In other words, S32K344 EMAC and SJA1105 do not “negotiate” speed over RGMII in the way SGMII/PCS-based links can.
The SJA1105 is an SPI-managed switch; you must load a static configuration after reset/power-up, and this configuration defines per-port interface mode and operational parameters. The typical software model is “configure and keep,” with only limited dynamic updates at runtime.
If your S32K344 EMAC ↔ SJA1105 connection is a MAC‑to‑MAC (PHY-less) RGMII link, the usual approach is to treat it as a fixed link and keep it at a chosen speed (S32K344 EMAC doesn't support 1Gbps).
Best regards,
Pavel