From eed1f754151965c4f9f7eef877801043101b83e1 Mon Sep 17 00:00:00 2001
From: bjorn3 <17426603+bjorn3@users.noreply.github.com>
Date: Sat, 25 Mar 2023 14:30:12 +0000
Subject: Don't store vector types in ssa variables

They have been causing a lot of trouble. For example current MIR
building thinks that it is fine to dynamically index into them. And
there are different paths depending on if the repr(simd) struct uses
fields or a single array as interior. There is also trouble with moving
the inner array of a repr(simd) type that is an array wrapper.

If performance becomes a concern, I will implement this in a more
principled way.
---
 scripts/test_rustc_tests.sh | 1 -
 1 file changed, 1 deletion(-)

(limited to 'scripts')

diff --git a/scripts/test_rustc_tests.sh b/scripts/test_rustc_tests.sh
index 20dcb4cf34d..527a505ccc4 100755
--- a/scripts/test_rustc_tests.sh
+++ b/scripts/test_rustc_tests.sh
@@ -112,7 +112,6 @@ rm tests/incremental/spike-neg2.rs # same
 
 rm tests/ui/simd/intrinsic/generic-reduction-pass.rs # simd_reduce_add_unordered doesn't accept an accumulator for integer vectors
 
-rm tests/ui/simd/intrinsic/generic-as.rs # crash when accessing vector type field (#1318)
 rm tests/ui/simd/simd-bitmask.rs # crash
 
 # bugs in the test suite
-- 
cgit 1.4.1-3-g733a5