wayne@ames.arpa (07/09/87)
I have just been given the opportunity to become familiar with
a very large and extremely modularized C application. A major
part of understanding this beast is figuring out who calls whom
and vice versa. Unfortunately, it is way too huge for "cxref"
(don't think we have the disk space!), so I was wondering if
anybody out there knows of any UN*X tools that would help out,
such as by producing a manageable call-graph? I have a Sun 3
running SunOS 3.2 (a 4.2BSD derivative, of course). Be willing
to pay for such a tool if necessary ... Also have access to
Internet FTP.
adTHANXvance!
Wayne Hathaway ultra!wayne@Ames.ARPA
Ultra Corporation
2140 Bering drive with a domain server:
San Jose, CA 95131 wayne@Ultra.COM
408-922-0100guy@gorodish.UUCP (07/09/87)
> Unfortunately, it is way too huge for "cxref" (don't think we have the > disk space!), so I was wondering if anybody out there knows of any UN*X > tools that would help out, such as by producing a manageable call-graph? > I have a Sun 3 running SunOS 3.2 (a 4.2BSD derivative, of course). More correctly, a 4.2BSD and System V Release 2 derivative; note the presence, for example, of "cxref". S5R2 also includes a flow-graph utility called "cflow"; as with "cxref", to get it you have to select the System V Optional Software item when installing SunOS. (It really has nothing to do with System V *per se* - it runs in the 4.2BSD environment, and lives in "/usr/bin".) However, since it, like "cxref", is a derivative of "lint", which produces BIG intermediate-pass files, you may have the same disk space problems with it as you might with "cxref". Guy Harris {ihnp4, decvax, seismo, decwrl, ...}!sun!guy guy@sun.com
franco@MIKEY.BBN.COM (07/09/87)
Wayne: Try 'cflow : generate C flow graph' or 'ctags : create a tags file' or 'ctrace : C program debugger' or 'trace : trace system calls of programs'. If this doesn't help, I would suggest to you to take volume 1 of your systems manual and flip through page by page and learn a few new commands. -franco
barnett@vdsvax.steinmetz.UUCP (Bruce G Barnett) (07/10/87)
Some people here use calls from mod.sources. At least one person here say cflow wasn't as useful as calls. = Subject: v06i003: new calls; shows function call flow (calls.new) = Message-ID: <78@mirror.UUCP> = Date: Mon, 16-Jun-86 20:23:48 EDT = Submitted by: Wombat <cca!caip!pur-ee!pucc-j.Purdue.EDU!rsk> = Archive-name: calls.new = This is a massively revised, bug-fixed, and hacked version of "calls", = which has been bouncing around the net for some time. Comments and = bugs and whatnot to Kevin Braunsdorf at pucc-j!ksb or ksb@j.cc.purdue.edu. = = It's also available via anonymous ftp on j.cc.purdue.edu. -- Bruce G. Barnett (barnett@ge-crd.ARPA) (barnett@steinmetz.UUCP)
skinner@saturn.ucsc.edu (Robert Skinner) (07/23/87)
In article <8230@brl-adm.ARPA>, franco@MIKEY.BBN.COM (Frank A. Lonigro) writes: > Wayne: > Try 'cflow : generate C flow graph' or 'ctags : create a tags file' > or 'ctrace : C program debugger' or 'trace : trace system calls of programs'. > If this doesn't help, I would suggest to you to take volume 1 of your systems > manual and flip through page by page and learn a few new commands. > > -franco great, someone asks about the existence of a tool that I could use also, and we get the above response. OK, Frank, I've 'learned a few new commands' and here's what I know: cflow: not present on either 4.3 or SYSV system I use ctags: sure, it shows you where routines are defined, and I use it for `vi`, but it doesn't show who calls who (it does help narrow down the search by hand) ctrace: not present trace: not present cxref: tried it on our SYSV system. It barfed when I tried to do 'complicated' things like: #define A 5 #define B A and it claims that its errors are cryptic but usually mean that the file wouldn't compile anyway, sure. I also remember that there was a pretty good program also called 'cxref' that we picked up at my last job. I had the impression it was public domain or came from Usenix. It had only minor bugs, and I used it successfully on a 17K line program. back to the original question: what works? Robert Skinner skinner@saturn.ucsc.edu
rsalz@bbn.com (Richard Salz) (07/24/87)
The comp.sources.unix archives has "calls" which does flow-call analysis. Contact your nearest archive site for info on how to get it; if you need a list of archive sites, send a (human-readable :-) request to sources@uunet.uu.net..... /r$ -- Rich $alz Cronus Project, BBN Labs rsalz@bbn.com sources@uunet.uu.net
mlandau@Diamond.BBN.COM (Matt Landau) (07/24/87)
NOTE: I've changed the followup to comp.lang.c, where this discussion
is more appropriate. If you're reading this via Arpa mailing
lists, check info-c (or get news working :-)
In comp.unix.wizards (<566@saturn.ucsc.edu>), skinner@saturn.ucsc.edu
(Robert Skinner) writes:
>
>someone asks about the existence of a tool that I could use also, and
>we get the above response. OK, Frank, I've 'learned a few new commands'
>and here's what I know:
>
> [List of programs that aren't present on stock BSD or certain SysV
> systems...]
>
>back to the original question: what works?
Try "calls", the call-graph generator posted to comp.sources.unix
(formerly mod.sources), and available from you nearest c.s.u archive
site. It's not as nice as cflow in some ways, but it's eminently
serviceable.
--
Matt Landau Oblivion gallops closer,
mlandau@bbn.com favoring the spur, sparing the rein...