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]

middle-end/8502: wrong code generated for switch statement


>Number:         8502
>Category:       middle-end
>Synopsis:       wrong code generated for switch statement
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Fri Nov 08 03:16:00 PST 2002
>Closed-Date:
>Last-Modified:
>Originator:     Volker Reichelt
>Release:        gcc 3.1, gcc 3.2, 3.2-branch
>Organization:
>Environment:
i686-pc-linux-gnu
>Description:
Consider the following testcase:

--------------------------snip here-------------------------
bool foo (int i)
{
    switch (i)
    {
        case 0: return false;
        case 1: return false;
        case 2: return false;
        case 3: return false;
        case 4: return false;
    }

    return false;
}

int main() { return 0; }
--------------------------snip here-------------------------

Compiling this with gcc 3.1, 3.2 or the 3.2-branch on i686-pc-linux-gnu I get
the following error message:

/tmp/ccii3YKY.o: In function `foo(int)':
/tmp/ccii3YKY.o(.text+0x11): undefined reference to `.L8'
collect2: ld returned 1 exit status

The assembler code for the function "foo" is:

	.align 2
.globl _Z3fooi
	.type	_Z3fooi,@function
_Z3fooi:
.LFB1:
	pushl	%ebp
.LCFI0:
	movl	%esp, %ebp
.LCFI1:
	cmpl	$4, 8(%ebp)
	ja	.L1
	movl	8(%ebp), %eax
	sall	$2, %eax
	movl	.L8(%eax), %eax     <--------- !!!!!
.L1:
	movl	$0, %eax
	popl	%ebp
	ret
.LFE1:
.Lfe1:
	.size	_Z3fooi,.Lfe1-_Z3fooi

Note that the problem disappears with only four cases in the switch statement.

The problem doesn't arise on the main trunk, though.
Because this is a regression from 3.0.x, I rate the PR as "high priority".

BTW, the PR is actually a distilled version of PR 6679 of
R.S. Veldema, which is quite large and and slightly corrupted.
Therefore, I closed PR 6679 and created this one.
>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted:


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