Bug report: pointer to array of incomplete type causes segfault in mcpcom

[wulf4]

Hello,

I found no better place for a bug report, so I'll try here.

I wondered if this program is correct C; gcc does not compile it, Sun C does. So I tried icc to get a third opinion :-)

int main(void)
{
struct foo { struct foo (*ptr)[1]; };
}

$ icc blah.c

icc: error #10106: Fatal error in /home/wulf4/intel/Compiler/11.1/064/bin/ia32/mcpcom, terminated by segmentation violation
compilation aborted for blah.c (code 1)

valgrind is very noisy, first message shown:

==13563== Conditional jump or move depends on uninitialised value(s)
==13563== at 0x85EAE25: __intel_sse2_strlen (in /home/wulf4/intel/Compiler/11.1/064/bin/ia32/mcpcom)

system information:

debian unstable, linux 2.6.30-1-686, eglibc 2.10.2

icc Version 11.1

Celeron M CPU 520

Cheers

Wulf4

[TimP]

icc aims for gcc compatibility, so it's not entirely surprising if it doesn't go beyond what your active gcc does. An internal error, at least when running on a supported system, would be a bug reportable on your premier.intel.com account.

$ icl wulf.c
Intel C++ Compiler Professional for applications running on IA-32, Version 11
.1 Build 20091130 Package ID: w_cproc_p_11.1.054
Copyright (C) 1985-2009 Intel Corporation. All rights reserved.

wulf.c
compilation aborted for wulf.c (code -1073741571)

This does seem excessively cryptic, given that gcc gives an indication what it doesn't like:

$ gcc wulf.c
wulf.c: In function `main':
wulf.c:3:27: error: array type has incomplete element type

[Om_S_Intel]

The statement is ill formed.

The compiler must not crash. The icc crashing is bug. I have submitted a report to Intel compiler development team.

[wulf4]

Thank you; please let me know if you need additional information!