Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site idis.UUCP Path: utzoo!linus!decvax!mcnc!idis!tom From: tom@idis.UUCP Newsgroups: net.bugs,net.news Subject: significant bug in readnews 2.10, fix included Message-ID: <312@idis.UUCP> Date: Thu, 4-Oct-84 13:58:51 EDT Article-I.D.: idis.312 Posted: Thu Oct 4 13:58:51 1984 Date-Received: Sat, 6-Oct-84 06:00:16 EDT Organization: University of Pittsburgh Lines: 23 [bug bait] Having just become the local net-news administrator, my first job was to find why readnews was hanging when it tried to read net.micro. The problem, it turns out, relates to the bitmap[] array and the clear macro. When the article number (stored in the global 'bit') was >= 8193, calling clear(bit) slopped over the end of the bitmap[] array, and fudged the next int, which just happened to be the global 'bit'. 'bit' got reset to 8192 so that when it was incremented to look at the next article, it started looking for 8193 again, and thus entered an infinite loop. The fix is to redefine the size of bitmap[] in rextern.c. I took it from bitmap[1024] to bitmap[2048], and it seems to have solved the problem. I haven't sat down to figure at what point this new sized array will be overrun. If I'm going over old ground, or missing something obvious , please don't turn up the flames too high. As I said, I'm new at this job. The above applies to version 2.10 of readnews. We are running 4.2 BSD. regards, Tom Neuendorffer {mcnc,decvax,pitt} !idis!tom