>Number:         3902
>Category:       bin
>Synopsis:       gcc doesn't promote correctly to (off_t)
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    bin-bug-people (Utility Bug People)
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Jul 22 05:35:01 1997
>Last-Modified:
>Originator:     Ignatios Souvatzis
>Organization:
computer science department, university of Bonn, Germany
>Release:        last thursday
>Environment:
	
System: NetBSD cosinus.cs.uni-bonn.de 1.2G NetBSD 1.2G (COSINUS) #64: Fri Jul 18 12:32:54 MEST 1997 ignatios@cosinus.cs.uni-bonn.de:/usr/obj.i386/kernel/COSINUS i386

>Description:
gcc promotes -sizeof(cbuf) (where cbuf is a small structure) in a way that

reed(fd, &cbuf, sizeof(cbuf));
lseek(fd, -sizeof(cbuf), SEEK_CUR);

seeks to 148+2^32-148 instead of back to 0.
Casting to (off_t) doesn't help.

lseek(fd, (off_t)-sizeof(cbuf), SEEK_CUR);

Casting to (long) helps:

lseek(fd, (long)-sizeof(cbuf), SEEK_CUR);

First casting to (off_t), then taking the negative value works:

lseek(fd, -(off_t)sizeof(cbuf), SEEK_CUR);

ktrace excerpt for non-working (all same):
  2358 rc5client CALL  lseek(0x3,0,0xffffff6c,0,0x1)

ktrace excerpt for working:
   261 rc5client CALL  lseek(0x3,0,0xffffff6c,0xffffffff,0x1)


>How-To-Repeat:
Try to compile & run the Finnish rc5 team's rc5client.c.
Wonder why it computes the same keyblock over and over again.
Wonder why the cache file is 4294967444 bytes long, instead of 8880.

>Fix:
If I knew how, I'd fix it.
>Audit-Trail:
>Unformatted: