On Tue, Nov 03, 2026 at 08:42:28AM -0000, Stuart Henderson wrote: > On 2026-11-03, Samuel Thibault <[email protected]> wrote: > > Hello, > > > > David Gwynne, le mar. 03 nov. 2026 14:10:29 +1000, a ecrit: > >> what physical interfaces are you using as ports in aggr(4) (and > >> trunk(4))? > > > > They are > > > > bnx0 at pci1 dev 0 function 0 "Broadcom BCM5709" rev 0x20: apic 1 int 4 > > bnx1 at pci1 dev 0 function 1 "Broadcom BCM5709" rev 0x20: apic 1 int 16 > > bnx2 at pci2 dev 0 function 0 "Broadcom BCM5709" rev 0x20: apic 1 int 0 > > bnx3 at pci2 dev 0 function 1 "Broadcom BCM5709" rev 0x20: apic 1 int 10 > > > >> generally, if an interface supports multiple rx rings, it > >> will set a flowid on mbufs for you. > > > > It looks like it's not happening. > > The bnx(4) driver doesn't support MSI or multiple rings. From what I > read, BCM5709 hardware would do if the driver supported it. > > >> the diff below tries to fix the M_FLOWID setting in aggr_port_input(), > >> and adds a link0 flag that enables packet parsing/hashing in a similar > >> style to what trunk(4) did. you'll have to "ifconfig aggr0 link0" to > >> enable it. > > Is there a benefit to hiding behind link0 rather than juat falling back > if M_FLOWID is not set?
it makes it easy to test either behaviour at the moment. if there's no performance difference doing the more complicated packet parsing then it makes sense to do it as a matter of course.
