From mboxrd@z Thu Jan  1 00:00:00 1970
From: sms@2BSD.COM (Steven M. Schultz)
Date: Wed, 30 Apr 2003 08:56:58 -0700 (PDT)
Subject: [pups] 2.11BSD device config trouble
Message-ID: <200304301556.h3UFuwb20269@moe.2bsd.com>

Hi -

> From: Jochen Kunz <jkunz at unixag-kl.fh-kl.de>
> > > cn 1 csr 176540 vector 344 no address found for kl/dl-11
> > 	Is 'cn 1 ...' line 38 of the /etc/dtab file?
> Yes: 
> cn      1 176540 344    5       cnrint  cnxint

	Interesting.   I do not recall any particular problem getting 
	additional DL devices recognized (the 11/93 had 7 of them).

	Seeing the 'no address found' error is *strange* though - that
	would indicate that 'autoconfig' could not find 'cnrint' or 'cnxint'
	in the /unix symbol table.   Look at /sys/autoconfig and you can
	see where that message is coming from.

	If you do 

		nm /unix | egrep 'cnxint|cnrint' 

	what do you see?

> NKL             4               # KL11, DL11
> The card has four ports, one of them is the console. (The M8192 CPU card
> has no SLU / ROM / ...) 
	
	Ah, ok.   My 11/73 has a SLU/ROM card and the console is on that.  I
	also have a DHV installed (alas, the system is powered down now
	so I can not check for more information).

> But I would prefere to get the DHV11 working. It seams that this device
> is more suitable for multi user operation. 

	Yes, it's a little better.  Not as nice as a DHQ-11 though (which can
	run in DHU or DHV modes - with DHU mode having much better silo 
	handling).

	I forget the exact error you were getting on the DHV but if it was
	'no interrupt' then it might be that the DHV clone is not behaving
	exactly like a DEC DHV

	In /sys/autoconfig/dhvauto.c here is how the probing attempts to
	force an interrupt:

dhvprobe(addr,vector)
        struct dhvdevice *addr;
        int vector;
{
    if ( grab ( &(addr->dhvcsr) ) & DHV_CS_MCLR )
        DELAY(35000L);
    if ( grab ( &(addr->dhvcsr) ) & (DHV_CS_MCLR|DHV_CS_DFAIL) )
        return ( 0 );
    stuff ( DHV_CS_RI | DHV_CS_RIE, &(addr->dhvcsr) );
    DELAY(3500L);
    stuff ( 0, &(addr->dhvcsr) );
    return(ACP_IFINTR);
}

	Either 3500 microseconds (very approximately of course) is too short 
	of a wait _or_ the method of trying to generate an interrupt is 
	not correct.    You can try changing 'ACP_IFINTR' to 'ACP_EXISTS'
	which tells autoconf to not care if the device interrupted or not.

> ** Last Mounted on /
> ** Root file system
> ** Phase 1 - Check Blocks and Sizes
> ** Phase 2 - Check Pathnames
> ** Phase 3 - Check Connectivity
> ** Phase 4 - Check Reference Counts
> ** Phase 5 - Check Free List
> BLK(S) MISSING
> SALVAGE? y
> 
> ** Phase 6 - Salvage Free List
> 1364 files, 11625 used, 2430 free
> 
> ***** FILE SYSTEM WAS MODIFIED *****
> 
> But on the next reboot I get the same when running fsck. Any hints? 

	How are you rebooting?  With the "reboot" command or by using the
	'halt' button?   You do not want to use the 'reboot' command because
	that does a "sync" which flushes the disc cache (and superblock) back
	out to disc - that overwrites the work that 'fsck' did.

	A few 'missing' blocks is not a serious problem though and can be
	ignored.

	Cheers,
	Steven Schultz