Re: optimization/7515: invalid inlining

Re: optimization/7515: invalid inlining

[Geoff Keating]

The following reply was made to PR optimization/7515; it has been noted by GNATS.

From: Geoff Keating <geoffk@geoffk.org>
To: drepper@redhat.com
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: optimization/7515: invalid inlining
Date: 06 Aug 2002 21:11:41 -0700

 This is not a bug unless -fpic is used, since executables themselves
 will not have any symbols overriden, correct?
 -- 
 - Geoffrey Keating <geoffk@geoffk.org> <geoffk@redhat.com>

Re: optimization/7515: invalid inlining

[rth]

Synopsis: invalid inlining

Responsible-Changed-From-To: geoffk->rth
Responsible-Changed-By: rth
Responsible-Changed-When: Fri Aug 30 19:43:04 2002
Responsible-Changed-Why:
    .
State-Changed-From-To: analyzed->closed
State-Changed-By: rth
State-Changed-When: Fri Aug 30 19:43:04 2002
State-Changed-Why:
    http://gcc.gnu.org/ml/gcc-patches/2002-08/msg01829.html
    http://gcc.gnu.org/ml/gcc-patches/2002-08/msg01830.html

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=7515

Re: optimization/7515: invalid inlining

[Ulrich Drepper]

The following reply was made to PR optimization/7515; it has been noted by GNATS.

From: Ulrich Drepper <drepper@redhat.com>
To: Geoff Keating <geoffk@geoffk.org>
Cc: gcc-gnats@gcc.gnu.org
Subject: Re: optimization/7515: invalid inlining
Date: Tue, 06 Aug 2002 21:17:05 -0700

 -----BEGIN PGP SIGNED MESSAGE-----
 Hash: SHA1
 
 Geoff Keating wrote:
 > This is not a bug unless -fpic is used, since executables themselves
 > will not have any symbols overriden, correct?
 
 This assumes clueful programmers.  Often code for DSO are compiled 
 without -fpic.  Sometimes it gets forgotten.  Other times people believe 
 the "PIC register prevents optimization" myth.  The text relocations 
 DSOs and up with having often go completely unnoticed.
 
 So, no, -fpic is no prerequisite for this being a bug.
 
 - -- 
 - ---------------.                          ,-.   1325 Chesapeake Terrace
 Ulrich Drepper  \    ,-------------------'   \  Sunnyvale, CA 94089 USA
 Red Hat          `--' drepper at redhat.com   `------------------------
 -----BEGIN PGP SIGNATURE-----
 Version: GnuPG v1.0.6 (GNU/Linux)
 Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
 
 iD8DBQE9UJ9H2ijCOnn/RHQRAiJwAKCgut+4fC4Zoq3VUC2a/knlBz/pRACggv8B
 gg0/a/wOdwawFNE4IOxUseg=
 =Mdrw
 -----END PGP SIGNATURE-----
 

Re: optimization/7515: invalid inlining

[geoffk]

Synopsis: invalid inlining

Responsible-Changed-From-To: unassigned->geoffk
Responsible-Changed-By: geoffk
Responsible-Changed-When: Tue Aug  6 21:10:33 2002
Responsible-Changed-Why:
    Working on fix.
State-Changed-From-To: open->analyzed
State-Changed-By: geoffk
State-Changed-When: Tue Aug  6 21:10:33 2002
State-Changed-Why:
    Reproduced with GCC 3.1 and -O3 -fpic.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=7515

optimization/7515: invalid inlining

[drepper]

>Number:         7515
>Category:       optimization
>Synopsis:       invalid inlining
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Aug 06 20:26:00 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Ulrich Drepper
>Release:        at least 2.96 and later
>Organization:
>Environment:
x86 Linux, but probably all others
>Description:
When -O3 is used gcc performs invalid inlining optimizations.  Functions, which are not declared static or have a visibility != default must not be inlined.  At least when generating code for ELF binaries.  The semantic changes since 'foo' will not be looked up using the ELF symbol resolution rules which would allow interposition.
>How-To-Repeat:
Compile the attached code with -O3.  The result is that 'foo' gets inlined.
>Fix:
Only inline if the function is static, has visibility != default, or possibly if a new function attribute or command line option is used.
>Release-Note:
>Audit-Trail:
>Unformatted:
----gnatsweb-attachment----
Content-Type: application/octet-stream; name="o3.c"
Content-Transfer-Encoding: base64
Content-Disposition: attachment; filename="o3.c"

aW50CmZvbyAoaW50IGEpCnsKICAgIHJldHVybiBhOwp9CgppbnQKYmFyIChpbnQgYSkKewogICAg
cmV0dXJuIGZvbyAoYSkgPT0gMTsKfQo=