Re: bge fail on netboot of NetBSD-6 on v215 and v240

[SAITOH Masanobu <msaitoh%execsw.org@localhost>]

(2013/05/16 2:17), Julian Coleman wrote:
> Hi,
> 
>> bge0 at pci0 dev 2 function 0: Broadcom BCM5704C Dual Gigabit Ethernet
>> bge0: interrupting at ivec 37c8
>> data error type 32 sfsr=808008 sfva=e0023040 afsr=10080005000000 
>> afva=7f6000010$
>> data fault: pc=13fe2d0 addr=e0023040 sfsr=0x808008<ASI=0x80,W>
>> kernel trap 32: data access error
>>
>> (gdb) list *(bge_attach+0x784)
>> 0x11a51a4 is in bge_attach (../../../../dev/pci/if_bge.c:3612).
>> 3607             * Read the hardware config word in the first 32k of NIC 
>> internal
>> 3608             * memory, or fall back to the config word in the EEPROM.
>> 3609             * Note: on some BCM5700 cards, this value appears to be 
>> unset.
>> 3610             */
> 
> Is this related to things like:
> 
>   isp0: invalid NVRAM header
> 
> where the built-in sparc64 hardware doesn't include the ROM/NVRAM normally
> associated with some chipsets?  As there is nothing mapped at that location,
> we get the data fault?  I wonder if that is also the case where this value
> appears to be unset, but that hardware is more forgiving.
> 
> Thanks,
> 
> J

 This problem was fixed correctly in if_bge.c rev. 1.128. I tested with v210,
PowerMac G5_11 and i386.

-- 
-----------------------------------------------
                SAITOH Masanobu (msaitoh%execsw.org@localhost
                                 msaitoh%netbsd.org@localhost)