wdc@ATHENA.MIT.EDU (Bill Cattey) (10/02/90)
I am getting random core dumps of ez on the PS/2. Sometimes it happens when I ^X2 Most times when I us compose character. It always seems to choke and die inside malloc. Since PS/2 dbx core dumps when you trace malloc, (I hope that the PS/2 and dbx people are properly and publicly embarrassed for their inability to come together and get a good debugger up for the PS/2...) I have been trying to find other ways to solve this problem. I got nasty and did ar d /lib/libc.a calloc.o malloc.o Then when I linked with plumber I found the following undefined symbols: %_Djyj.Q %_Sjyj.Q %_DJCp2R %_SJCp2R The first two are defined in calloc.o, and the second two in malloc.o Since these important symbols are not defined anywhere else, they force load malloc.o and calloc.o from libc.a These 4 symbols are asked for in globals.o. Can someone help me determine if: a. these are important symbols that have important value that must be duplicated. (Much like the realloc_srchlen dummy variable is that appears in malloc.ci.) b. these can be removed from the list of globals. c. there is some correctable problem with the routine that generates the list of globals so that these 4 symbols can be automagically eradicated. -wdc
jis@mtgzx.att.com (J Mukerji) (10/02/90)
Excerpts from info-andrew: 2-Oct-90 One step closer to plumber .. Bill Cattey@athena.mit.e (1234) > I am getting random core dumps of ez on the PS/2. > Sometimes it happens when I ^X2 > Most times when I us compose character. > It always seems to choke and die inside malloc. Hmmm.... Very interesting. That is exactly one of the various places that ez dies on the SunOS 4.1 Sparcstation! Now that you mention it, many (but not all) of the other places where it dies could possibly be in malloc too. Thanks for the info.