Hi Jakub,
The test needs double, otherwise it wouldn't fail with a buggy cc1 on i?86.
I've been initially contemplating putting the testcase into
gcc.target/i386/, but then thought it would be good to test this on a couple
of other arches too. Guess it can be limited to a couple of targets which
are supposed to handle this, or it can have a list of targets which aren't
supposed to compile this.
Well my tests show that it fails on:
am33-*-*
arc-*-*
avr-*-*
h8300-*-*
frv-*-*
hppa64-*-*
m32c-*-*
m32r-*-*
m68hc1*-*-*
mmix-*-*
mn10300-*-*
powerpc*-*-*
sh-*-*
vax-*-*
I have a patch to add a "dg-skip-if" line to the pr35318.c file for
these targets. Shall I submit it officially ?
Cheers
Nick
