Re: wm0 Driver

[buhrow%lothlorien.nfbcal.org@localhost (Brian Buhrow)]

        Hello David.  About a year and a half ago, I did some work on the wm
driver with fiber cards.  Specifically, the driver didn't let you hard code
speed and duplex, prefering to autonegotiate.  I fixed this, and the
changes were pulled into Rev 1.34 of the if_wm.c file.  See PR kern/35797 for
details.
I can't say that I've tried the NetBSD-4 version of the wm(4) driver with
these fixes in place on the fiber cards I'm using, because they're in
production, and we're still running the code I cobled together in NetBSD-3,
but the cards we're using are dual-port half-height PCI cards which probe
as follows:

wm0 at pci3 dev 1 function 0: Intel i82546GB 1000BASE-X Ethernet, rev. 3
wm0: interrupting at ioapic2 pin 0 (irq 10)
wm0: 64-bit 133MHz PCIX bus
wm0: 256 word (8 address bits) MicroWire EEPROM
wm0: Ethernet address xx:xx:xx:xx:xx:xx
wm0: 1000baseSX, 1000baseSX-FDX, auto
wm1 at pci3 dev 1 function 1: Intel i82546GB 1000BASE-X Ethernet, rev. 3
wm1: interrupting at ioapic2 pin 1 (irq 5)
wm1: 64-bit 133MHz PCIX bus
wm1: 256 word (8 address bits) MicroWire EEPROM
wm1: Ethernet address xx:xx:xx:xx:xx:xx
wm1: 1000baseSX, 1000baseSX-FDX, auto

        It sounds like you've got some dangling pointer in the driver
somewhere, leading me to believe that some part of the fiber interface
parameters in the driver aren't getting initialized properly.

-Brian

On Sep 12,  6:20pm, "David Dudley" wrote:
} Subject: wm0 Driver
} This is a MIME message. If you are reading this text, you may want to 
} consider changing to a mail reader or gateway that understands how to 
} properly handle MIME multipart messages.
} 
} --=__Part4D64C25E.0__=
} Content-Type: text/plain; charset=US-ASCII
} Content-Transfer-Encoding: 7bit
} Content-Disposition: inline
} 
} Well, guys, I got a bunch of new DELL1950's(5), 2650's(4), and 2850's(4)
} servers in house to play with today.  Now, all I've got to do is make
} them do something intelligent with NetBSD.
} 
} First, the changes to the NMBCLUSTERS to correct the problems I was
} having when using both bnx0 and bnx1 have corrected the problem  with it
} running out of them.  Everything there is fine.
} 
} Now, each of these machines has at least 1 Intel PRO/1000 multimode
} fiber network card.  Guess what?  When I try to enable one of them, I
} get 'fatal page fault in supervisor mode'. Additional statements are
} 'trap type 6 code 0 eip c05ff9bb cs 8 eflags 10286 cr 2 0 ilevel 6'
} 'kernel: supervisor trap page fault, code=0'
} 'Stopped in pid 1368.1 (ifconfig) at netbsd:wm_tbi_mediachange+0xb 
} movl 0(%eax),%esi'
} 
} Note that I'm copying this off the screen, as I don't have any
} capability of capturing this data.
} 
} Of course, these machines all have USB keyboards, so once they fault,
} there isn't any doing anything except holding down the power button to
} reset them.
} 
} The problem is related specifically to the fiber cards, as two of these
} machines also have Intel PRO/1000 copper connections, and those seem to
} work with no problem.
} 
} David Dudley
} 
} --=__Part4D64C25E.0__=
} Content-Type: text/plain; name="David Dudley.vcf"
} Content-Transfer-Encoding: quoted-printable
} Content-Disposition: attachment; filename="David Dudley.vcf"
} 
} BEGIN:VCARD
} VERSION:2.1
} X-GWTYPE:USER
} FN:David Dudley
} TEL;WORK:826-1265
} ORG:;MIS
} TEL;PREF;FAX:880-3741
} EMAIL;WORK;PREF;NGW:DavidDu%cctexas.com@localhost
} N:Dudley;David
} TITLE:Functional Analyst
} END:VCARD
} 
} 
} --=__Part4D64C25E.0__=--
>-- End of excerpt from "David Dudley"