Problem with Broadcom 5704 B1 on amd64 6.2 release

James Shank james.shank at gmail.com
Mon Jul 16 22:36:45 UTC 2007


Dear John,

Thanks for the suggestion.  I tried the 7.0 livefs snapshot and had
the same results as with the 6.2 release; identical errors on both.

Thanks,

James

On 7/16/07, John Baldwin <jhb at freebsd.org> wrote:
> On Monday 16 July 2007 01:37:40 am James Shank wrote:
> > Greetings,
> >
> > I've run into a problem with the onboard Broadcom 5704 NetXtreme nics
> > on a Tyan S2891 Thunder K8SRE motherboard.
> >
> > Here is the relevant dmesg output to show the error:
> >
> > pcib5: <ACPI PCI-PCI bridge> at device 11.0 on pci8
> > pci10: <ACPI PCI bus> on pcib5
> > pcib5: memory: end (de1fffff) < start (48739b2de100000)
> > pcib5: memory: end (de1fffff) < start (dbca73fdde110000)
> > bge0: <Broadcom BCM5704 B0, ASIC rev. 0x2100> mem
> > 0x48739b2de100000-0x48739b2de10ffff irq 28 at device 9.0 on pci10
> > pcib5: memory: end (de1fffff) < start (48739b2de100000)
> > bge0: couldn't map memory
> > device_attach: bge0 attach returned 6
> > bge1: <Broadcom BCM5704 B0, ASIC rev. 0x2100> mem
> > 0xdbca73fdde110000-0xdbca73fdde11ffff irq 29 at device 9.1 on pci10
> > pcib5: memory: end (de1fffff) < start (dbca73fdde110000)
> > bge1: couldn't map memory
> > device_attach: bge1 attach returned 6
> >
> > It appears to me that the problem might be due to using a 32-bit int
> > for end addresses where it looks like the start address uses 64-bit
> > int.
> >
> > Any input on how to proceed would be greatly appreciated.
> >
> > I've also attached full dmesg output as well as pciconf -l -v output.
> >
> > Thanks!
>
> Can you try a 7.0 snapshot?  It has different handling of 64-bit BARs.  If 7
> works then we can look at backporting those changes to 6.x.
>
> --
> John Baldwin
>


More information about the freebsd-hackers mailing list