Subject: kern/5139: 'de' driver prints weird media notice on DEC Celebris
To: None <>
From: None <cgd@NetBSD.ORG>
List: netbsd-bugs
Date: 03/09/1998 16:49:33
>Number:         5139
>Category:       kern
>Synopsis:       'de' driver prints weird media notice on DEC Celebris
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    kern-bug-people (Kernel Bug People)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Mon Mar  9 17:05:00 1998
>Originator:     Chris G. Demetriou
Kernel Hackers 'r' Us
>Release:        NetBSD-current as of Mar 9, 1998
System: NetBSD 1.3D NetBSD 1.3D (BRICK) #36: Wed Feb 18 21:00:25 PST 1998 i386
System actually being tested is running a kernel built from today's
(March 9th's) sources.

	[ filed as 'kern' rather than port-i386 because it's a problem
	with an MI driver. ]

	I've got a DEC Celebris GL 6180 (short tower), that appears to
	have a built in 21142 ethernet.

	On boot, the 'de' driver prints:

	de0 at pci0 dev 3 function 0de0: can't find phy 0
	de0: interrupting at irq 9
	de0: DEC 21142 [10-100Mb/s] pass 1.1
	de0: address 00:00:f8:03:93:5a

	At the very least, the printf formatting is losing.  At worst,
	there's some kind of bug in the driver; it seems ... odd (but
	not impossible 8-) to me that a Digital system would trigger
	this warning (which seems to be more of a debugging message
	than anything else) by having a bogus ethernet ROM.

	It's also worth noting that the autoconfiguration output claims
	"10-100Mb/s", but 'ifconfig -m de0' says:

	        media: autoselect (10baseT/UTP) status: active
	        supported media: autoselect 10base5/AUI 10base2/BNC 10baseT/UTP <full-duplex> 10baseT/UTP
	        inet netmask 0xffffff00 broadcast

	If 100Mb/s operation is dependant on a PHY which is missing (?!),
	then should the driver really be claiming 10-100Mb/s at boot?

	Boot a NetBSD/i386 kernel containing the 'de' driver on a Digital
	Celebris GL 6180 (short tower).  (That might have to be a Celebris
	GL 6180 short tower without the 100Mb/s option; it appears that
	there's an add-in card for 100Mb/s operation.)

	Look at the autoconfiguration output, compare it to the 'ifconfig
	-m' output for the interface.
	The autoconfig 'prettyness' problem can probably be fixed by
	some careful manipulation of the driver's printfs.

	I don't know what the deal with the PHY is, but printing
	the correct set of supported speeds at boot time probably wouldn't
	be too hard.