[prev in list] [next in list] [prev in thread] [next in thread] 

List:       freebsd-bugs
Subject:    bin/23912: underflow of cnt in vs_paint() by O_NUMBER_LENGTH when both leftright and number options
From:       gladiatr () boneyard ! lawrence ! ks ! us
Date:       2000-12-28 21:17:20
[Download RAW message or body]


> Number:         23912
> Category:       bin
> Synopsis:       underflow of cnt in vs_paint() by O_NUMBER_LENGTH when both \
>                 leftright and number options in use
> Confidential:   no
> Severity:       non-critical
> Priority:       low
> Responsible:    freebsd-bugs
> State:          open
> Quarter:        
> Keywords:       
> Date-Required:
> Class:          sw-bug
> Submitter-Id:   current-users
> Arrival-Date:   Thu Dec 28 13:20:01 PST 2000
> Closed-Date:
> Last-Modified:
> Originator:     Stephen D. Spencer
> Release:        4.2
> Organization:
> Environment:
FreeBSD gandhi.eecs.ukans.edu 4.2-STABLE FreeBSD 4.2-STABLE #0: Mon Dec 18 09:41:03 \
CST 2000     sdspence@gandhi.eecs.ukans.edu:/usr/src/sys/compile/gandhi  i386

> Description:
in versions of nvi including the newer development versions, if the
leftright and number options are both active, the editor will dive into
a not-quite-inifinite loop.  in several seconds, it will return.  if 
the ruler option is active, it shows (1,4294967288).  nvi remains
fairly useless from this point forward and must be sigkill'd
> How-To-Repeat:
vi /your/favorite/file that is long enough to require a page redraw
> set leftright
> set number

<Ctrl>F a few times... (croak)

> Fix:
this problem was traced back to vs_columns() returning a 0 value 
regardless of whether or not the number option is being used.

according to vi/vi.h, #define O_NUMBER_LENGTH 8

which should be the offset returned when 'set number' is active so 
that when the O_NUMBER_LENGH margin used for displaying the line number
is subtracted from the literal screen column, you will end up with 
column '0' rather than 0xFFFFFFF8.

Please note, I have read the comments towards the last nvi bug report
that was submitted here and understand that nothing will be 
done until the vendor commits the patch to their distribution.  I
thought it would be nice in case someone else has tried these two 
options in tandem and not had time to poke around to at least have a 
manual recourse.

I have submitted a bug report to Keith Bostic.  They are in the middle
of getting ready for a DB rollout, so it will be a while before he is
able to review the report.

--- vs_relative.c.orig  Mon May 13 09:20:04 1996
+++ vs_relative.c       Thu Dec 28 00:54:32 2000
@@ -122,7 +122,7 @@
        if (lp == NULL) {
 done:          if (diffp != NULL)              /* XXX */
                        *diffp = 0;
-               return (0);
+               return (O_ISSET(sp, O_NUMBER) ? O_NUMBER_LENGTH : 0);
        }
 
        /* Store away the values of the list and leftright edit options. */


> Release-Note:
> Audit-Trail:
> Unformatted:


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-bugs" in the body of the message


[prev in list] [next in list] [prev in thread] [next in thread]