> Looks OK to me, AI did find a couple typos that you should fix: > > > Reviewed the whole v6 series applied on top of main. All 15 patches apply > cleanly, net/txgbe builds with no new warnings, and I found no > correctness bugs. Three Info-level nits below, all in 14/15. >
> One design point worth a second opinion rather than a fix: in 10/15, > txgbe_dev_detect_sfp() re-arms itself at the "rearm:" label, which the two > sfp_an_alarm_enabled == 0 early returns deliberately skip. That is > how the poll stops on dev_stop, and it is correct, but it does make that gate > the only thing terminating the 2-second poll that 40G hot-plug > detection now depends on. > Hi Stephen, Thank you for the review. I have fixed the issues you mentioned and sent v7. I also confirmed the design of the AML40 module poll. The persistent 2-second poll is intentional for AML40 non-backplane ports. These devices may not deliver module insertion or removal GPIO events, so polling must continue whether a module is absent, present, or already identified. Therefore the poll is intended to run for the lifetime of a started port and to stop only when the port is stopped. Best regards, Zaiyu
