https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88622
Andi Kleen <andi-gcc at firstfloor dot org> changed:
What |Removed |Added
----------------------------------------------------------------------------
Target| |x86_64-linux
CC| |jakub at gcc dot gnu.org
--- Comment #1 from Andi Kleen <andi-gcc at firstfloor dot org> ---
Don't have a small reproducer, but on a large LTO build where a few files are
built with
-mpreferred-stack-boundary=4
with the others using the default
I hit the following ICE in ix86_minimum_alignment
29610 /* Don't do dynamic stack realignment for long long objects with
29611 -mpreferred-stack-boundary=2. */
29612 if ((mode == DImode || (type && TYPE_MODE (type) == DImode))
29613 && (!type || !TYPE_USER_ALIGN (type))
29614 && (!decl || !DECL_USER_ALIGN (decl)))
29615 {
29616 gcc_checking_assert (!TARGET_STV);
29617 return 32;
29618 }
I suspect the right fix is to just remove the assert?
Adding Jakub who added it originally in:
commit 1f1475a7e758328a59db17aef5d1ccd81232ea95
Author: jakub <jakub@138bc75d-0d04-0410-961f-82ee72b054a4>
Date: Thu Feb 4 09:02:01 2016 +0000
PR target/69454
* config/i386/i386.c (convert_scalars_to_vector): Remove
stack alignment fixes.
(ix86_option_override_internal): Disable TARGET_STV if stack
might not be aligned enough.
(ix86_minimum_alignment): Assert that TARGET_STV is false.
* gcc.target/i386/pr69454-1.c: New test.
* gcc.target/i386/pr69454-2.c: New test.
