openssh bugs - May 2006 - [Bug 1192] warning: comparison between signed and unsigned

http://bugzilla.mindrot.org/show_bug.cgi?id=1192

           Summary: warning: comparison between signed and unsigned
           Product: Portable OpenSSH
           Version: 4.3p2
          Platform: amd64
               URL: http://www.openbsd.org/cgi-
                   
bin/cvsweb/src/usr.bin/ssh/scp.c.diff?r1=1.121&r2=1.122&
                    f=h
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: P2
         Component: scp
        AssignedTo: bitbucket at mindrot.org
        ReportedBy: carenas at sajinet.com.pe


introduced as part of scp version 1.122 (OpenBSD 3.8/OpenSSH 4.2), and
which results in the following warning when compiled with
-Wsign-compare :

scp.c: In function `source':
scp.c:632: warning: comparison between signed and unsigned
scp.c:639: warning: comparison between signed and unsigned

the problem is that off_t is defined signed (long long for OpenBSD and
long int for linux/amd64), while size_t is unsigned (unsigned long for
OpenBSD and long unsigned int for Linux).

for LPI32 architectures (like i386), the expansion from unsigned long
(4 bytes) to long long (8 bytes) is safe, but no so for LP64
architectures (like amd64) where "long long" and "unsigned
long" are 8
bytes wide.




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.mindrot.org/show_bug.cgi?id=1192





------- Comment #1 from carenas at sajinet.com.pe  2006-05-22 13:36 -------
Created an attachment (id=1142)
 --> (http://bugzilla.mindrot.org/attachment.cgi?id=1142&action=view)
reverting first snippet for scp.c version 1.122

compiling without warnings and validated for OpenBSD/amd64,
OpenBSD/i386 and Gentoo Linux 2006.0/x86_64




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.mindrot.org/show_bug.cgi?id=1192





------- Comment #2 from dtucker at zip.com.au  2006-05-22 13:55 -------
(From update of attachment 1142)
>-	size_t result;
>+	off_t i, amt, result, statbytes;
I don't think that's the right way forward since later there is:

        result = atomicio(read, fd, bp->buf, amt)

and atomicio now returns a size_t.

Instead, I think we should be changing the progressmeter interface to
u_int64_t from off_t (since ultimately, the size it operates on is
fixed at 64 bits by the packet format of the sftp spec).  This would
change "amt" and also fix the warning.

There's an old bug for this (#842) but I suspect the diff will need
some tweaking.




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.

http://bugzilla.mindrot.org/show_bug.cgi?id=1192





------- Comment #3 from dtucker at zip.com.au  2006-05-22 14:01 -------
Err, ignore comment #2.  It's a similar but unrelated issue (but in the
same chunk of code).




------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.