10_100_1000 Mbps tri-mode ethernet MAC

Issue List
Potentially a Bug in Phy_int #21
Open jefflieu opened this issue about 9 years ago
jefflieu commented about 9 years ago

Not sure whether this is a bug or not. (I'm using RGMII interface, when I convert RGMII to GMII following the standard before feeding GMII signals into phy_int) In phy_int, rx path, the error signal (MRxErr) goes through 1 stage of FF, whilst the Data valid signal (MCrs_dv) going through 3 stages of FFs. The symptom would be receiving a lot of Error for good packet if you plug the module into a normal network switch.

jefflieu commented about 9 years ago

Type your text here

snetoc commented over 5 years ago

I can confirm that it is a bug, at least I tried for 1000Mbps. Users also reported that RX packets with an odd number of bytes get lost. I also noticed this behavior. When fixing this bug, also RX packets with odd number of bytes are received fine.

The right behavior can be easily checked with a testbench, where the RxErr = !RxDv. So the error signal is always '1' between frames. According to the 802.3 standard, this behavior is valid and some PHYs (at least my one as it seems) behave like this.

No one