Compound Literals patch breaks kernel and glibc! (Was Re: Testcase for a new ICE)

[Andreas Jaeger]

"Joseph S. Myers" <jsm28@cam.ac.uk> writes:

> On 16 Dec 2001, Alexandre Oliva wrote:
>
>> On Dec 16, 2001, Andreas Jaeger <aj@suse.de> wrote:
>> 
>> > Footnotes: [1] Not in the testsuite but in compilation of Linux kernel
>> > and glibc.
>> 
>> Andreas, please post your testcase extracted from glibc as a patch to
>> introduce a new test in the testsuite.
>
> I'm testing a patch to fix the problem; of course it adds the testcase to
> the testsuite.

Thanks a lot! I'm just bootstrapping the compiler - and will close the
PR myself after doing a number of tests.

The example from glibc is slightly different from the kernel example
(that's the testcase you committed):

inline int
finite (double __x)
{
  return (__extension__
	  (((((union { double __d; int __i[2]; }) {__d: __x}).__i[1]
	     | 0x800fffffu) + 1) >> 31));
}

int
main (void)
{
  double x = 1.0;
  
  return finite (x);
}

Should I commit this testcase?  It seems also to be fixed now,

Andreas
-- 
 Andreas Jaeger
  SuSE Labs aj@suse.de
   private aj@arthur.inka.de
    http://www.suse.de/~aj