[MinGW-notify] [ mingw-Bugs-3291311 ] Incorrect jump address in assembly code

[SF/projects/mingw n. l. <min...@li...>]

Bugs item #3291311, was opened at 2011-04-22 14:52
Message generated for change (Tracker Item Submitted) made by sam9721
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102435&aid=3291311&group_id=2435

Please note that this message will contain a full copy of the comment thread,
including the initial issue submission, for this request,
not just the latest update.
Category: binutils
Group: None
Status: Open
Resolution: None
Priority: 5
Private: No
Submitted By: Peter (sam9721)
Assigned to: Nobody/Anonymous (nobody)
Summary: Incorrect jump address in assembly code

Initial Comment:
Incorrect jump address for backward local labels with Intel syntax in as 2.20.

File test.asm:

    .intel_syntax
    mov eax, 5
1:
    dec eax
    jnz 1b

    ret


Command line (under msys):
   as test.asm -o test.o
   ld test.o -o test.exe


Actual result: jump two bytes forward.

00401000      B8 05000000   mov eax,5
00401005      48            dec eax
00401006 +--- 75 02         jne short test.0040100A ; error
00401008 |    C3            ret
00401009 |    90            nop
0040100A +->  90            nop


Expected result: jump three bytes back:

00401000      B8 05000000   mov eax,5
00401005 +->  48            dec eax
00401006 +--- 75 FD         jne short test.00401005
00401008      C3            ret


Additional information

If you specify -a switch (as test.asm -o test.o -a), the bug disappears. If you
rewrite the assembly code using AT&T syntax, it also disappears:
    mov $5, %eax
1:
    dec %eax
    jnz 1b

    ret

I checked the object file (test.o) generated by the GNU Assembler and found the
erroneous jump address in it (so it's not a linker bug). I also reported the bug to as maintainers (http://sourceware.org/bugzilla/show_bug.cgi?id=12687), but they were unable to reproduce it under Linux, so it should be MinGW bug.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102435&aid=3291311&group_id=2435