The following reply was made to PR kern/32968; it has been noted by GNATS.

From: Petar Bogdanovic <p+netbsd@2005.smokva.net>
To: Hubert Feyrer <hubert@feyrer.de>
Cc: port-i386@NetBSD.org,
	"Bayer, David" <david.bayer@Vanderbilt.Edu>, gnats-bugs@NetBSD.org
Subject: Re: kern/32968
Date: Sat, 10 Jun 2006 01:18:10 +0200

 A copy for gnats. :)
 
 Hubert Feyrer wrote:
 > 
 > There's some more fun with NetBSD-current (3.99.17?) and an 
 > almost-GENERIC kernel on new HP dc7600 machines. I've never seen that 
 > NMI message before... any ideas?
 > 
 > 
 >  - Hubert
 > 
 > ---------- Forwarded message ----------
 > Date: Tue, 23 May 2006 13:25:50 -0500
 > From: "Bayer, David" <david.bayer@Vanderbilt.Edu>
 > To: Hubert Feyrer <hubert@feyrer.de>, g4u-help@feyrer.de
 > Subject: RE: [g4u-help] FYI:  new HP dc7600 problem
 > 
 > Sorry about that - I didn't have the info available when I wrote the
 > email.  Here are the on-screen messages:
 > 
 > ppb1 at pci0 dev 28 function 1: Intel 82801GB/GR PCI Express Port #2
 > (rev. 0x01)
 > pci2 at ppb1 bus 2
 > pci2: i/o space, memory space enabled, rd/line, wr/inv ok
 > bge0 at pci2 dev 0 function 0: Broadcom BCM5752 Gigabit Ethernet
 > bge0: interrupting at irq 10
 > NMI ... going to debugger
 > Stopped in pid 0.1 (swapper) at 0xc044ab22:     xorl     %eax,%eax
 > db>
 > 
 > If I type the "s" command, I get the following:
 > Stopped in pid 0.1 (swapper) at 0xc044ab24:     mov1     %ecx,%edx
 > 
 > I can type "continue" and move on.  Then it stops again:
 > audio0 at azalia0: full duplex, independent
 > NMI ... going to debugger
 > Stopped in pid 0.1 (swapper) at 0xc056e720:     inb     $0x71,%al
 > 
 > The "s" command returns this:
 > Stopped in pid 0.1 (swapper) at 0xc056e722:     movzbl  %eax,%eax
 > 
 > I can again "continue", and I get a bunch of bge0 errors and finally
 > make it to the menu.  The first few bge0 errors are as follows:
 > bge0: firmware handshake timed out, val = 0
 > bge0: RX CPU self-diagnostics failed!
 > bge0: flow-through queue init failed
 > bge0: initialization error 6
 > bge0: block failed to stop: reg 0x468, bit 0x00000002
 > bge0: block failed to stop: reg 0x2c00, bit 0x00000002
 > bge0: block failed to stop: reg 0x2000, bit 0x00000002
 > bge0: block failed to stop: reg 0x2400, bit 0x00000002
 > bge0: block failed to stop: reg 0x2800, bit 0x00000002
 > 
 > Etc...
 > 
 > Once I get to the menu, if I run ifconfig -a I get:
 > bge0: flags=8803<UP,BROADCAST,SIMPLEX,MULTICAST> mtu 1500
 > 
 > capabilities=3f00<IP4CSUM_Rx,IP4CSUM_Tx,TCP4CSUM_Rx,TCP4CSUM_Tx,UDP4CSUM
 > _Rx,UDP4CSUM_Tx>
 >         enabled=0
 >         address: 00:16:35:67:31:b4
 > bge0: PHY read timed out
 > bge0: PHY read timed out
 > bge0: PHY read timed out
 > bge0: link state changed to DOWN
 > bge0: PHY read timed out
 > bge0: PHY read timed out
 > bge0: PHY read timed out
 >         media: Ethernet autoselect
 >         status: no carrier
 > lo0: flags=8008<LOOPBACK,MULTICAST> mtu 33192
 > 
 > I hope that's the information you need - if you need something different
 > I'll be happy to provide it.
 > 
 > Thanks
 > 
 > David R. Bayer
 > david.bayer@vanderbilt.edu
 > 
 > 
 > -----Original Message-----
 > From: Hubert Feyrer [mailto:hubert@feyrer.de]
 > Sent: Tuesday, May 23, 2006 11:32 AM
 > To: Bayer, David
 > Cc: g4u-help@feyrer.de; Hubert Feyrer
 > Subject: Re: [g4u-help] FYI: new HP dc7600 problem
 > 
 > On Tue, 23 May 2006, Bayer, David wrote:
 >> g4u 2.2 beta 2 has a problem with the Broadcom network card in the
 >> latest HP dc7600 computers running Pentium 4 HT processors.  g4u sees
 >> the network card and tries to load the driver and immediately dies.  I
 > 
 >> verified that it is the network card by disabling it in the bios and
 >> using a slightly older pci broadcom card, at which point g4u works
 > fine.
 > 
 > OK, need more data:
 >   * how does it "see the network card", what does it print?
 >   * how does it "try to load the driver" (NetBSD uses a monolithic
 > kernel
 >     with no modules, thus has all drivers compiled in and doesn't load
 >     anything), what does it print?
 >   * how does it "immediately die", what does it print?
 > 
 > Unfortunately, exact data is needed to have a remote chance to debug
 > this.
 > Thanks!
 > 
 > 
 >   - Hubert