Silly IRQ allocation on Dell 1950

John Baldwin jhb at freebsd.org
Thu Nov 2 22:07:32 UTC 2006


On Thursday 02 November 2006 11:03, John Polstra wrote:
> I've been testing a Dell 1950 box with -current from late July.  The
> box has two integrated BCM5708 gigabit ethernet interfaces that work
> with the bce driver.  I noticed that both bce0 and bce1 are assigned
> to IRQ 16.  With two IOAPICs on board, it's hard to believe that Dell
> would be so silly as to force the two gigabit interfaces to share an
> IRQ.  Is this sharing really necessary, or is it caused by a bug or
> limitation in the OS?  The verbose dmesg output follows below.

It's in the hardware.  Both your bce devices are on pci4 and pci8 both
of which route their interrupts up through their bridges onto pci3,
and Dell mapped both of those to IRQ 16.

> pci8: <PCI bus> on pcib4
> pcib3: matched entry for 7.0.INTA
> pcib3: slot 0 INTA hardwired to IRQ 16
> pcib4: slot 0 INTA is routed to irq 16
> bce0: <Broadcom NetXtreme II BCM5708 1000Base-T (B1), v0.9.6> mem 0xf4000000-0xf5ffffff irq 16 at
> device 0.0 on pci8
> pci4: <PCI bus> on pcib14
> pcib13: matched entry for 3.0.INTA
> pcib13: slot 0 INTA hardwired to IRQ 16
> pcib14: slot 0 INTA is routed to irq 16
> bce1: <Broadcom NetXtreme II BCM5708 1000Base-T (B1), v0.9.6> mem 0xf8000000-0xf9ffffff irq 16 at
> device 0.0 on pci4

-- 
John Baldwin


More information about the freebsd-current mailing list