Re: optimization/7714: inlining of function with inline assembler causes duplicate assembler lables

Re: optimization/7714: inlining of function with inline assembler causes duplicate assembler lables

[bangerth]

Synopsis: inlining of function with inline assembler causes duplicate assembler lables

State-Changed-From-To: open->feedback
State-Changed-By: bangerth
State-Changed-When: Fri Dec  6 13:09:07 2002
State-Changed-Why:
    Maybe someone else can chime in here:
    
    I think it is not possible to use labels in asm statements
    if they will appear in different places (i.e. in functions
    that are inlined, or in templates). Unfortunately, the
    manual doesn't say anything about this; rather, it even
    uses a label in asm code in at least one place.
    
    What do the asm specialists say?
    
    Regards
      Wolfgang

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=7714

optimization/7714: inlining of function with inline assembler causes duplicate assembler lables

[linux_dr]

>Number:         7714
>Category:       optimization
>Synopsis:       inlining of function with inline assembler causes duplicate assembler lables
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          rejects-legal
>Submitter-Id:   net
>Arrival-Date:   Sat Aug 24 18:06:03 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Loren Osborn <linux_dr@yahoo.com>
>Release:        gcc 3.1.1
>Organization:
>Environment:
Linux Mandrake 8.1
>Description:
I found a similar bug to PR233, and was disappointed to
see that PR233 was closed, appearantly just because the
URL to the test case became a broken link.

I think I have found a manifistaion of the same bug, 
although it doesn't involve unrolling loops.  It involves 
inlining of functions with inline-assembler lables in it...

I have included a minimal test case 

As far as I am aware, the code itself (while not 
technically vailid ANSI C, because ANSI C doesn't 
explicitly allow inline assembler) should be valid in
gcc. 
>How-To-Repeat:
gcc -O3 -c -o InlineLableBug.o InlineLableBug.c
>Fix:
gcc should probably prepend (or append) a string to each assembler label that is unique to each instantiation of that function.
>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="InlineLableBug.c"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="InlineLableBug.c"