pan/bi: Gate late DCE/CSE on "optimize"
Otherwise we can end up with unlowered ATOM.i32 on Valhall. Signed-off-by: Alyssa Rosenzweig <alyssa@collabora.com> Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/15223>
This commit is contained in:
committed by
Marge Bot
parent
3485b8dc78
commit
b70a7c97bb
@@ -4125,8 +4125,10 @@ bi_compile_variant_nir(nir_shader *nir,
|
||||
bi_lower_fau(ctx);
|
||||
|
||||
/* Lowering FAU can create redundant moves. Run CSE+DCE to clean up. */
|
||||
bi_opt_cse(ctx);
|
||||
bi_opt_dead_code_eliminate(ctx);
|
||||
if (likely(optimize)) {
|
||||
bi_opt_cse(ctx);
|
||||
bi_opt_dead_code_eliminate(ctx);
|
||||
}
|
||||
|
||||
/* Analyze before register allocation to avoid false dependencies. The
|
||||
* skip bit is a function of only the data flow graph and is invariant
|
||||
|
||||
Reference in New Issue
Block a user