Page MenuHomeFreeBSD

bxe: Add warpcore 2500baseX support
Needs ReviewPublic

Authored by adam.jaremko_freebsd_gmail.com on May 31 2023, 6:18 PM.
Tags
None
Referenced Files
Unknown Object (File)
May 10 2024, 2:02 PM
Unknown Object (File)
Feb 10 2024, 9:39 PM
Unknown Object (File)
Feb 3 2024, 8:18 AM
Unknown Object (File)
Dec 20 2023, 4:59 AM
Unknown Object (File)
Dec 18 2023, 5:10 AM
Unknown Object (File)
Dec 12 2023, 5:42 AM
Unknown Object (File)
Oct 15 2023, 7:13 AM
Unknown Object (File)
Oct 15 2023, 7:13 AM

Details

Reviewers
None
Group Reviewers
Contributor Reviews (src)
Summary

Support 2500Base-X on NetXtreme II BCM5710 adapters according to NVRAM configuration

Test Plan

The hardware and nvram configuration have been thoroughly documented by myself at https://www.dslreports.com/forum/r32230041-Internet-Bypassing-the-HH3K-up-to-2-5Gbps-using-a-BCM57810S-NIC

Diff Detail

Repository
rG FreeBSD src repository
Lint
Lint Skipped
Unit
Tests Skipped
Build Status
Buildable 51813
Build 48704: arc lint + arc unit

Event Timeline

luca.piccirillo_gmail.com added inline comments.
sys/dev/bxe/bxe_elink.c
5050

I cannot understand this bitmask conditions. It looks to me that you are enforcing 1G capability to be enabled in cap_mask even for 2_5G. Why?

sys/dev/bxe/bxe_elink.c
5050

SFP+ transceivers must support 1G, be it if they're 2.5G, 10G, etc. and it should be present in the capability mask when fiddling with ediag and was used as a catch guard against misconfiguration.

This also aligns with some previous code handling 2.5G and how the MII control register is set for 1G. But it's not strictly required to include 1G capability in the mask for this condition.