Subject: Re: Anyone else seeing fxp0 die?
To: Atsushi Onoe <onoe@sm.sony.co.jp>
From: David Brownlee <abs@netbsd.org>
List: current-users
Date: 10/03/2002 11:26:51
On Thu, 3 Oct 2002, Atsushi Onoe wrote:

> > > > Oct  1 21:31:27 allison /netbsd: fxp0: device timeout
>
> > I just upgraded my fxp box to 1.6 today. When booting the 1.6 CD the
> > kernel fiddled some with the eeprom on my fxp, and it told me I couldn't
> > use the device until my machine was restarted.
>
> It's a completely separete issue, as we are talking about NetBSD-current,
> not 1.6.
>
> > Is this documented somehwre?
>
> The source code may help you to get further information:-)
>
> Captured from sys/dev/ic/i82557.c
> 	/*
> 	 * Systems based on the ICH2/ICH2-M chip from Intel, as well
> 	 * as some i82559 designs, have a defect where the chip can
> 	 * cause a PCI protocol violation if it receives a CU_RESUME
> 	 * command when it is entering the IDLE state.
> 	 *
> 	 * The work-around is to disable Dynamic Standby Mode, so that
> 	 * the chip never deasserts #CLKRUN, and always remains in the
> 	 * active state.
> 	 *
> 	 * Unfortunately, the only way to disable Dynamic Standby is
> 	 * to frob an EEPROM setting and reboot (the EEPROM setting
> 	 * is only consulted when the PCI bus comes out of reset).
> 	 *
> 	 * See Intel 82801BA/82801BAM Specification Update, Errata #30.
> 	 */

	1.6 has this fix (I have a bunch of machines that started working
	after Jason copied it across from FreeBSD :)

-- 
		David/absolute          -- www.netbsd.org: No hype required --