This is the mail archive of the gcc-bugs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

[Bug c/33725] New: Could eliminate argument push for the second const function for same arguments


This testcase:

--cut here--
extern const int foo (int a);
extern const int bar (int a);

int test (int a)
{
  return foo (a) + bar (a);
}
--cut here--

compiles to:

test:
        subl    $12, %esp
        movl    %ebx, 4(%esp)
        movl    16(%esp), %ebx
        movl    %esi, 8(%esp)
        movl    %ebx, (%esp)    (*)
        call    foo
        movl    %ebx, (%esp)    (**)
        movl    %eax, %esi
        call    bar
        movl    4(%esp), %ebx
        addl    %esi, %eax
        movl    8(%esp), %esi
        addl    $12, %esp
        ret

However, since argument is already pushed to the stack for foo(), there is no
need to push it again for the bar() function [foo() is const function]. If gcc
eliminates the second push, a call-clobbered reg could be used to move argument
to (esp), resulting in something like:

test:
        subl    $12, %esp
        movl    16(%esp), %eax
        movl    %esi, 8(%esp)
        movl    %eax, (%esp)
        call    foo
        movl    %eax, %esi
        call    bar
        addl    %esi, %eax
        movl    8(%esp), %esi
        addl    $12, %esp
        ret


-- 
           Summary: Could eliminate argument push for the second const
                    function for same arguments
           Product: gcc
           Version: 4.3.0
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: c
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: ubizjak at gmail dot com
GCC target triplet: i686-pc-linux-gnu


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=33725


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]