[Bug c/17855] New: modification of function return not diagnosed

[Bug c/17855] New: modification of function return not diagnosed

[rth at gcc dot gnu dot org]

struct foo {char x, y, z[2];};
struct foo f();
void bar(int baz)
{
        f().z[baz] = 1;
}

Generates

  z.c:5: internal compiler error: gimplification failed

But I don't see that this is legal according to [6.5.2.2]/5

  # If an attempt is made to modify the result of a function
  # call ... behaviour is undefined.

If we must accept this construct because it might not be
executed, a possible solution is to notice fallback==fb_lvalue
for a CALL_EXPR in c_gimplify_expr and create a temporary for
the call result, and which will also satisfy the lvalue.

This test has crashed in all versions since at least 2.96.

-- 
           Summary: modification of function return not diagnosed
           Product: gcc
           Version: 4.0.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: rth at gcc dot gnu dot org
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] modification of function return not diagnosed

[jsm28 at gcc dot gnu dot org]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2004-10-06 08:40 -------
This does need to be accepted (where warning and compiling into an abort is
one valid form of accepting) in case it is not executed, cf. DR#109:

   A conforming implementation must not fail to translate a strictly
   conforming program simply because *some* possible execution of that
   program would result in undefined behavior.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |jsm28 at gcc dot gnu dot org
OtherBugsDependingO|                            |16989
              nThis|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0 Regression] modification of function return not diagnosed

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-10-06 13:31 -------
Confirmed, the ICE is a regression though.
: Search converges between 2003-06-04-ssa (#3) and 2003-06-05-ssa (#4).

So it is an old regression from the tree-ssa branch.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |pinskia at gcc dot gnu dot
                   |                            |org
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|                            |1
           Keywords|                            |ice-on-valid-code
   Last reconfirmed|0000-00-00 00:00:00         |2004-10-06 13:31:01
               date|                            |
            Summary|modification of function    |[4.0 Regression]
                   |return not diagnosed        |modification of function
                   |                            |return not diagnosed
   Target Milestone|---                         |4.0.0


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0 Regression] modification of function return not diagnosed

[jsm28 at gcc dot gnu dot org]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2004-10-28 16:35 -------
RTH suggested:

> If we must accept this construct because it might not be
> executed, a possible solution is to notice fallback==fb_lvalue
> for a CALL_EXPR in c_gimplify_expr and create a temporary for
> the call result, and which will also satisfy the lvalue.

c_gimplify_expr doesn't have the fallback information available.
gimplify_expr does (and there's already code specific to this
sort of thing in gimplify.c), but the obvious fix

+         if (fallback == fb_lvalue)
+           *expr_p = get_initialized_tmp_var (*expr_p, pre_p, post_p);

just moves the ICE to expand_assignment (while changing the testcase
to insert an explicit temporary yields almost identical tree dumps
but without the ICE).

A related example

struct foo {char x, y, z[2];};
struct foo p, q; int r;
void bar(int baz)
{
  (r ? p : q).z[baz] = 1;
}

ICEs in 3.4 as well as on mainline (in expand_assignment in both cases);
a temporary is already created in gimplifying the COND_EXPR.

While if the non-lvalue structure is the result of an assignment, 3.4 ICEs
and mainline doesn't.

struct foo {char x, y, z[2];};
struct foo p, q;
void bar(int baz)
{
  (p = q).z[baz] = 1;
}


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0/4.1 Regression] modification of function struct return not diagnosed

[jsm28 at gcc dot gnu dot org]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2005-03-30 18:05 -------
Testing a fix.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |jsm28 at gcc dot gnu dot org
                   |dot org                     |
             Status|NEW                         |ASSIGNED
   Last reconfirmed|2004-12-05 04:32:03         |2005-03-30 18:05:58
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0/4.1 Regression] modification of function struct return not diagnosed

[jsm28 at gcc dot gnu dot org]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2005-03-30 23:07 -------
Patch posted <http://gcc.gnu.org/ml/gcc-patches/2005-03/msg02808.html>, awaiting
review.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0/4.1 Regression] modification of function struct return not diagnosed

[jsm28 at gcc dot gnu dot org]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2005-03-31 10:09 -------
[resend of comment#6 which did not get emailed out]

The reason I don't consider compiling into an abort to be suitable here is that
the problem could in principle arise with more complicated code where the
pointer in the array reference isn't immediately syntactically from a non-lvalue
array but has been through other assignments, casts, etc., the optimizers then
removing these so that at some point an assignment to such an array needs to be
expanded.  Ultimately I think these cases may most reliably be handled by having
the front end create explicit addressable temporaries for every non-lvalue
structure containing an array (mapping non-lvalue-struct to *(tmp =
non-lvalue-struct, &tmp)), but I think putting the special cases in the
gimplifier makes more sense in the context of fixing now a regression we have now.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0/4.1 Regression] modification of function struct return not diagnosed

[rth at gcc dot gnu dot org]

------- Additional Comments From rth at gcc dot gnu dot org  2005-03-31 22:16 -------
Patch is ok.  I would prefer that this (eventually) be handled by the front end,
but certainly this approach is appropriate for fixing the regression in 4.0.

-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0/4.1 Regression] modification of function struct return not diagnosed

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-03-31 23:34 -------
Subject: Bug 17855

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	jsm28@gcc.gnu.org	2005-03-31 23:34:45

Modified files:
	gcc            : ChangeLog gimplify.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/compile: struct-non-lval-1.c 
	                                     struct-non-lval-2.c 
	                                     struct-non-lval-3.c 

Log message:
	PR c/17855
	* gimplify.c (gimplify_expr): Create a temporary for lvalue
	COND_EXPR and CALL_EXPR.
	
	testsuite:
	* gcc.c-torture/compile/struct-non-lval-1.c,
	gcc.c-torture/compile/struct-non-lval-2.c,
	gcc.c-torture/compile/struct-non-lval-3.c: New tests.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&r1=2.8061&r2=2.8062
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/gimplify.c.diff?cvsroot=gcc&r1=2.120&r2=2.121
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.5254&r2=1.5255
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/struct-non-lval-1.c.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/struct-non-lval-2.c.diff?cvsroot=gcc&r1=NONE&r2=1.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/struct-non-lval-3.c.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0/4.1 Regression] modification of function struct return not diagnosed

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2005-03-31 23:37 -------
Subject: Bug 17855

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-4_0-branch
Changes by:	jsm28@gcc.gnu.org	2005-03-31 23:37:11

Modified files:
	gcc            : ChangeLog gimplify.c 
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.c-torture/compile: struct-non-lval-1.c 
	                                     struct-non-lval-2.c 
	                                     struct-non-lval-3.c 

Log message:
	PR c/17855
	* gimplify.c (gimplify_expr): Create a temporary for lvalue
	COND_EXPR and CALL_EXPR.
	
	testsuite:
	* gcc.c-torture/compile/struct-non-lval-1.c,
	gcc.c-torture/compile/struct-non-lval-2.c,
	gcc.c-torture/compile/struct-non-lval-3.c: New tests.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.7592.2.104&r2=2.7592.2.105
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/gimplify.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=2.113.2.2&r2=2.113.2.3
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=1.5084.2.83&r2=1.5084.2.84
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/struct-non-lval-1.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/struct-non-lval-2.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.2.1
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.c-torture/compile/struct-non-lval-3.c.diff?cvsroot=gcc&only_with_tag=gcc-4_0-branch&r1=NONE&r2=1.1.2.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855

[Bug c/17855] [4.0/4.1 Regression] modification of function struct return not diagnosed

[jsm28 at gcc dot gnu dot org]

------- Additional Comments From jsm28 at gcc dot gnu dot org  2005-03-31 23:42 -------
Fixed on 4.0 branch and mainline.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=17855