small bug in init

raj at UCI-750B.ARPA raj at UCI-750B.ARPA
Wed Aug 15 15:46:44 AEST 1984


From:  Richard Johnson <raj at UCI-750B.ARPA>

Description:
	The init program uses the assumption that if after running getty
	that fork returns in less than 1 minute, then getty failed.  This
	is not always true.
Repeat-By:
	Pick a terminal and keep logging in and back out again making sure
	you never stay logged in for more than 1 minute.  After about 5 times
	of logging in and out again, the console will produce the bogus
	message "init: getty failing; sleeping...".
Fix:
	Reset the time of the last try at running getty on a line to zero
	when you know for sure a person has logged out on that line.
	The only fix needed is at the end of the routine "rmut".  After this
	fix has been installed, the only way to produce the message is by
	Control-C'ing getty over and over.  This is not worth worrying about.

	*** bk_init.c	Sun Jun 12 01:39:21 1983
	--- init.c	Mon Aug  6 21:58:51 1984
	
	The line numbers for "bk_init.c" are correct for the original berkeley
	source but the ones for "init.c" can't be trusted.
	
	***************
	*** 479,484
	  			write(f, (char *)&wtmp, sizeof(wtmp));
	  			close(f);
	  		}
	  	}
	  }
	  
	
	--- 508,517 -----
	  			write(f, (char *)&wtmp, sizeof(wtmp));
	  			close(f);
	  		}
	+ #ifdef	UCI			/* raj - change 1 */
	+ 		p->gettytime = 0;/* force time since last getty to be
	+ 				    a large number to reset p->gettycnt */
	+ #endif	UCI
	  	}
	  }



More information about the Comp.unix.wizards mailing list