------- =_aaaaaaaaaa0
Content-Type: text/plain; charset="us-ascii"

Hi,

I tried sending in the following bug report with send-pr, but received no
automatic ack and no personal reaction either.  Are you guys all swamped
with other higher-priority tasks, or did the bug report disappear into the
bit-bucket?

- Havard

------- =_aaaaaaaaaa0
Content-Type: text/plain; charset="us-ascii"
Content-Description: Copy of bug report

SEND-PR: -*- send-pr -*-
SEND-PR: Lines starting with `SEND-PR' will be removed automatically, as
SEND-PR: will all comments (text enclosed in `<' and `>').
SEND-PR: 
SEND-PR: Please consult the send-pr man page `send-pr(1)' or the Texinfo
SEND-PR: manual if you are not sure how to fill out a problem report.
SEND-PR:
SEND-PR: Choose from the following categories:
SEND-PR:
SEND-PR: bin         kern        lib         misc        
SEND-PR: port-amiga  port-hp300  port-i386   port-m68k   port-mac    
SEND-PR: port-sparc  port-sun3   
SEND-PR:
To: gnats-bugs@sun-lamp.cs.berkeley.edu
Subject: if_ed, multicast & promiscuous mode bugs
From: he@rype.runit.sintef.no
Reply-To: he@rype.runit.sintef.no

>Submitter-Id:  net
>Originator: 	Havard Eidnes
>Organization:	SINTEF RUNIT
>Confidential:  no
>Synopsis:	promiscuous mode doesn't work with (relatively) new if_ed
>Severity:	serious
>Priority:	medium
>Category: 	port-i386
>Class:		sw-bug
>Release:	NetBSD-current as of one or two weeks ago
>Environment:	Configured for MULTICAST
System: NetBSD rype.runit.sintef.no 0.9a RYPE#32 i386

>Description:
	Due to relatively recent changes in the if_ed driver to support 
	multicast, it has become impossible to use BPF and set promiscuous 
	mode. The reason is that the BPF code fiddles the interface flags 
	and then calls the interface SIOCSIFFLAGS ioctl, but the code there will
	only do something if the interface is either being taken up or down.

	The attached patch is an attempt at circumventing that problem,
	but it appears to have other unintended side effects.  In effect,
	it makes other programs stop receiving network I/O (?) and they
	will be unable to resume until the interface is taken out of
	promiscuous mode.  This most visibly affects tcpdump, where it will
	hang trying to tranlslate IP addresses to names via the DNS (run
	it with -n and it produces output just fine, confirming that 
	promiscuous mode has been turned on).

>How-To-Repeat:
	See above.
>Fix:
	Here is the patch which tries to make promiscuous mode work again.
	As said above, it appears to have undesireable side effects which
	I have been unable to find a solution to.

*** if_ed.c.old	Thu Jan 13 11:42:29 1994
--- if_ed.c	Fri Jan 21 20:48:30 1994
***************
*** 171,174 ****
--- 171,175 ----
  	u_char	rec_page_stop;	/* last page of RX ring-buffer */
  	u_char	next_packet;	/* pointer to next unread RX packet */
+ 	short	old_if_flags;	/* old value of corresponding if_flags */
  } ed_softc[NED];
  
***************
*** 1300,1303 ****
--- 1301,1307 ----
  	ifp->if_flags &= ~IFF_OACTIVE;
  
+ 	/* Remember for next SIOCSIFFLAGS ioctl */
+ 	sc->old_if_flags = ifp->if_flags;
+ 
  	/*
  	 * ...and attempt to start output
***************
*** 1930,1938 ****
  		} else {
  		/*
! 		 * If interface is marked up and it is stopped, then start it
  		 */
  			if ((ifp->if_flags & IFF_UP) &&
! 		    	    ((ifp->if_flags & IFF_RUNNING) == 0))
! 				ed_init(ifp->if_unit);
  		}
  		/*
--- 1934,1948 ----
  		} else {
  		/*
! 		 * If interface is marked up and it is either stopped or some
! 		 * some of the changeable flags changed, then (re)initialize 
! 		 * and (re)start it
  		 */
  			if ((ifp->if_flags & IFF_UP) &&
! 		    	    ((ifp->if_flags & IFF_RUNNING) == 0) ||
! 		    	     ((ifp->if_flags & ~IFF_CANTCHANGE) !=
! 		    	      (sc->old_if_flags & ~IFF_CANTCHANGE)))
! 			{
! 			 	ed_reset(ifp->if_unit);
! 			}
  		}
  		/*

------- =_aaaaaaaaaa0--

------------------------------------------------------------------------------