[Bug tree-optimization/31727] New: Functions with non-local gotos are considered not to have side effects

[Bug tree-optimization/31727] New: Functions with non-local gotos are considered not to have side effects

[rakdver at gcc dot gnu dot org]

In the following testcase with -O1, dce removes the bla call, thus
making the otherwise infinite loop terminate:

static void bla(void *label)
{
  goto *label;
}

int main (void)
{
l1:
  bla (&&l1);

  return 0;
}


-- 
           Summary: Functions with non-local gotos are considered not to
                    have side effects
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Keywords: wrong-code
          Severity: normal
          Priority: P3
         Component: tree-optimization
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: rakdver at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31727

[Bug tree-optimization/31727] Functions with non-local gotos are considered not to have side effects

[rguenth at gcc dot gnu dot org]

------- Comment #1 from rguenth at gcc dot gnu dot org  2007-04-27 12:04 -------
I think this is ipa-pure-const again:

Function found to be const: bla
Function found to be const: main

and DCE removes pure/const function calls.  (We should also not CSE such
calls)


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31727

[Bug tree-optimization/31727] Functions with non-local gotos are considered not to have side effects

[pinskia at gcc dot gnu dot org]

------- Comment #2 from pinskia at gcc dot gnu dot org  2007-04-28 00:36 -------
This code is undefined
Take a look at the manual:
http://gcc.gnu.org/onlinedocs/gcc-4.1.2/gcc/Labels-as-Values.html

You may not use this mechanism to jump to code in a different function. If you
do that, totally unpredictable things will happen. The best way to avoid this
is to store the label address only in automatic variables and never pass it as
an argument. 

There is nothing special about this code, except it is undefined :).


-- 

pinskia at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31727

[Bug tree-optimization/31727] Functions with non-local gotos are considered not to have side effects

[pinskia at gcc dot gnu dot org]

------- Comment #3 from pinskia at gcc dot gnu dot org  2007-04-28 00:37 -------
Oh and this goto is considered not a non-local goto, it is just considered a
computed goto.  A non-local goto is only with nested functions.  Like:
int main (void)
{
__label__ l1;
void bla(void)
{
  goto l1;
}
l1:
 bla();

  return 0;
}


Which we handle correctly by the way :).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31727

[Bug tree-optimization/31727] Functions with non-local gotos are considered not to have side effects

[pinskia at gcc dot gnu dot org]

------- Comment #4 from pinskia at gcc dot gnu dot org  2007-04-28 00:40 -------
By the way non local gotos are documented at:
http://gcc.gnu.org/onlinedocs/gcc-4.1.2/gcc/Nested-Functions.html


I really guess nobody actually reads the documentation (yes we can improve it
but even then people still won't read it :) ).

I don't know how to fix that issue except keep on pointing out in our
documentation about where stuff is documented.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=31727