https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=289545
--- Comment #3 from Leo Bicknell <bickn...@ufp.org> --- I think saying "the bug" is a bit ambiguous. There appear to be multiple components to this problem: #1 Some Intel NICs drop link when particular capabilities are changed, like hardware offload. I have never seen clear documentation if this is a problem with the FreeBSD driver, or if this is firmware inside the NIC that is the responsibility of Intel. The attitude in several bug reports is "that's just the way it is with those NICs", but no one has ever documented WHY in any of the posts I have seen. #2 When a bridge is brought up it has to force all component interfaces to have the same set of capabilities. It appears in older code this was not documented, but somewhere along the way was documented and is now spelled out in the man 4 bridge page. While the paragraph is clear, I suspect many people would not understand the implications from just reading the man page. #3 A workaround for the combination of #1 and #2 is to manually disable the capabilities on one or more interfaces so that when the bridge starts up everything matches and the bridge code does not need to change them. This seems like a mostly reasonable workaround to me, and was the one I was trying to implement. It was in the process of implementing that I found this capability "NOMAP" which there appears to be no cli way to turn off, appears to be virtually undocumented, but which bridge must turn off, which means the entire workaround does not work. #4 An unstated workaround is to replace the Intel card with a non-Intel NIC that does not have this problem. In this bug report my main goal is to get an argument to ifconfig to disable "NOMAP" so the workaround is a functional workaround again. That would make me happy. However I can see how other people would prefer that #1 be fixed, either by fixing the FreeBSD driver or by getting Intel to fix their card. Given Intel generally produces high quality NIC cards that people like I can see how getting it fixed could be a priority for some. -- You are receiving this mail because: You are the assignee for the bug.
