home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.os2.programmer
- Path: sparky!uunet!usc!sdd.hp.com!elroy.jpl.nasa.gov!nntp-server.caltech.edu!gah
- From: gah@hood.hood.caltech.edu (Glen Herrmannsfeldt)
- Subject: stack size
- Message-ID: <gah.711830283@oldhood>
- Summary: how big a stack can, and should, one have?
- Keywords: stack
- Sender: news@cco.caltech.edu
- Nntp-Posting-Host: oldhood.hood.caltech.edu
- Organization: California Institute of Technology, Pasadena
- Date: Wed, 22 Jul 1992 18:38:03 GMT
- Lines: 26
-
- link386 seems to have a 2k default stack.
-
- I just finished a series of bug reports to bug-gcc that I finally
- decided were due to too small a stack. I call to getchar() would
- write over parts of program memory.
-
-
- Then I tried different stack sizes with link386.
-
- 4k, 256k, 4MB, 64MB, 1GB, 256MB.
-
- Up to 64MB ran fine, fast, etc. Apparently it does not need to
- allocate the whole thing in real or swap space.
-
- At 1GB it refused to load.
-
- At 256MB it got an internal processing error, which, if I figure out
- how, I will report to IBM. (Any one from IBM? How do I do this?)
-
- It seems that there is little penalty for a big stack, and a large
- penalty (clobbered memory) for a small stack. Why not have a big default
- stack?
-
- Has anyone else experienced any of this?
-
- Remember, a call to getchar() and nothing else will overflow a 2K stack!
-