[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference turris::fortran

Title:	Digital Fortran
Notice:	Read notes 1.* for important information
Moderator:	QUARK::LIONEL

Created:	Thu Jun 01 1995
Last Modified:	Fri Jun 06 1997
Last Successful Update:	Fri Jun 06 1997
Number of topics:	1333
Total number of notes:	6734

1302.0. "Initialize stack variables" by HYDRA::NEWMAN (Chuck Newman, 508/467-5499 (DTN 297), MRO1-3/F26) Mon May 19 1997 13:31

I'm trying to track down a problem that I think is caused by using a stack
variable before it is set (I caught one on Friday that the compiler was
unable to catch, and I suspect there are others).

Is there a switch (probably undocumented) that will generate code to initialize
allocated stack to some bizarre number (e.g. something that is an IEEE NaN and
perhaps a huge negative integer?)

Thanks!

								-- Chuck Newman

T.R	Title	User	Personal Name	Date	Lines
1302.1		QUARK::LIONEL	Free advice is worth every cent	`Mon May 19 1997 14:02`	3
	No. Steve
1302.2	third degree?	MSBCS::SCHNEIDER	Say it with ASCII	`Mon May 19 1997 16:03`	3
	I think third degree does that. Chuck
1302.3	Third degree is helpful, but has limitations	HYDRA::NEWMAN	Chuck Newman, 508/467-5499 (DTN 297), MRO1-3/F26	`Thu May 29 1997 10:28`	21
	Re: .2 Yes, third degree does that, but I have two problems with it: 1) In turning my 16+ MB file into a 52+ MB file I get the following warning: atom: Warning: Branch instruction overflowed. Executable may not work. I guess I'll be okay as long as I don't try to execute that particular code. 2) It doesn't yet fix up the symbol table information, so finding the code that causes my FPE in the .third executable is well nigh impossible. Is there any likelyhood that a "-poison stack" qualifier would be implemented, or would it be too difficult and too low on the priority list to ever see the light of day? -- Chuck Newman
1302.4	We'll think hard about "-poison stack"	GEMGRP::GROVE		`Thu May 29 1997 11:09`	14
	re: .3 < Is there any likelyhood that a "-poison stack" qualifier would be implemented, < or would it be too difficult and too low on the priority list to ever see the < light of day? Chuck - it's clear that a "-poison stack" option would be useful. I'd like to see the GEM team implement that for all our Alpha compilers. If we do it as a somewhat inefficent debugging option, I don't think it would be too hard. If we try to make it extremely efficient, we raise the bar so high that we then might not do it at all. /Rich Grove, GEM