SPW2562@RITVAXC.BITNET (02/20/86)
I've been trying to write a C program on our VAX/VMS system here at
RIT, but have been running into some trouble. The program is run
from my LOGIN.COM to assign symbol names, and delete some symbols
I don't want defined. The assigning part works fine, but when the
program tries deleting symols, the following occurs:
Access violation, reason mask=01, virtual address=7473B3C2, PC=00016831, PSL=03C
00000
Symbolic stack dump follows
module name routine name line rel PC abs PC
00016831 00016831
00016502 00016502
LOGIN main 29 0000017A 0000077A
Being a Unix* hack, I have no idea what this means. (And people
complain about Unix* error messages being cryptic. I understand
Unix* error messages far better than I'll ever understand this
stuff.) The offening line and surrounding code are here:
...
lib$set_symbol("COL*OR", "WRITE SYS$OUTPUT \"\27[2J\27[H\"", tab);
lib$set_symbol("CL*EAR", "WRITE SYS$OUTPUT \"\27[0;41;33m\27[A\"", tab);
=> lib$delete_symbol("ONE", tab);
lib$delete_symbol("LOGOUT", tab);
max=strlen(name)-14;
...
I've run it with 'tab' as an int with value 2, and with 'tab'
#define'd as 2, and it makes no difference. When this function
is called from a basic program it works fine. Can anyone tell me
why this is happening, and how I can correct it?
Please respond directly to either of the below two addresses,
as I do not subscribe to this mailing list.
*UNIX is a trademark of Bell Labs.
==============================================================================
Steve Wall @ Rochester Institute of Technology
Usenet: ..!rochester!ritcv!spw2562 (Fishhook) Unix 4.2 BSD
BITNET: SPW2562@RITVAXC (Snoopy) VAX/VMS 4.2stew%lhasa.UUCP@HARVARD.HARVARD.EDU (02/20/86)
Your problem is with string descriptors. "Quoted Strings" in C are just an address of an array of bytes, terminated by zero. However, the VMS library routines expect strings to be passed by descriptor. See Section 9.4 of "Programming in VAX-11 C" for examples. Stew
tihor@NYU-ACF4.ARPA (Stephen Tihor) (02/21/86)
... the following occurs:
Access violation, reason mask=01, virtual address=7473B3C2, PC=00016831, PSL=03C
00000
Being a Unix* hack, I have no idea what this means.
Being a VMS user I know to look up errors in the manual (try doing `man
bus error` on most unix systems.) Explains says trying to through a bad
location. Ah ha he thinks a parameter error in:#
...
lib$set_symbol("COL*OR", "WRITE SYS$OUTPUT \"\27[2J\27[H\"", tab);
lib$set_symbol("CL*EAR", "WRITE SYS$OUTPUT \"\27[0;41;33m\27[A\"", tab);
=> lib$delete_symbol("ONE", tab);
lib$delete_symbol("LOGOUT", tab);
max=strlen(name)-14;
...
Checking the one line subroutine summaries: HELP LIB$ LIB$DELETE_SYMBOL
I see that the second parameter is the address of a longword containing the
symbol table number. tab might contain that from the code given, but
wait:
I've run it with 'tab' as an int with value 2, and with 'tab'
#define'd as 2, and it makes no difference. When this function
is called from a basic program it works fine. Can anyone tell me
why this is happening, and how I can correct it?
Ah ha, you are passing a (long) the value of tab, the integer 2, which is a
bad address as we all know instead of a (* long) such as &tab which is
what the routine wants.info-vax@ucbvax.UUCP (02/22/86)
In article <8602200928.AA12076@ucbvax.berkeley.edu> SPW2562@RITVAXC.BITNET writes: > > > I've been trying to write a C program on our VAX/VMS system here at > . . . > Being a Unix* hack, I have no idea what this means. (And people > ... > lib$set_symbol("COL*OR", "WRITE SYS$OUTPUT \"\27[2J\27[H\"", tab); > ... > Please respond directly to either of the below two addresses, > as I do not subscribe to this mailing list. > >*UNIX is a trademark of Bell Labs. >============================================================================== > Steve Wall @ Rochester Institute of Technology > Usenet: ..!rochester!ritcv!spw2562 (Fishhook) Unix 4.2 BSD > BITNET: SPW2562@RITVAXC (Snoopy) VAX/VMS 4.2 I have been following the various discussions and postings on mod.computers.vax (and info-vax before it) for a couple of years now and have enjoyed the relatively high technical content to be found here. I find myself increasingly dismayed at the numbers of recent postings that could have been avoided by anything more than a cursory glance at the available documentation. The included posting is to my mind a perfect case in point. Heck, even the HELP command will tell you in no uncertain terms that the strings that are to be passed to LIB$SET_SYMBOL should be: 'access: read only mechanism: by descriptor' in that light one could ask: Why was this posted? Why was this allowed past the moderator? I believe that the moderator should perhaps take a more active role in filtering out inappropriate postings of this sort. How long would it have taken this person to type: $ HELP Rtl LIB$ LIB$SET_SYMBOL as opposed to posting the message that he generated and then caused the rest of the world to pay for its distribution. For all I know, he's still waiting at his terminal with bated breath for a reply. Perhaps the moderator already DOES filter out a lot of garbage like this, and in that case I offer in advance profuse apologies for any aspersions that I've cast on his/her/its character/intellect. BUT...But...but... you know. There... that's a couple of screens full of vitriol.
SPW2562@RITVAXC.BITNET (02/22/86)
Thanks to all who responded to my earlier posting. The consensus
seems to be that C argument passing is quite different than what
VMS expects to get, and that it was just luck that the program didn't
die much earlier. Now I'll see if it works with the changes...
Snoopy
aka Fishhook
Steve Wall @ Rochester Institute of Technology
BITNET: SPW2562@RITVAXC.BITNET
Usenet: ..!rochester!ritcv!spw2562