[Bug tree-optimization/122079] [16 regression] ICE on valid code at -O{s,2,3} on x86_64-linux-gnu: in compute_antic, at tree-ssa-pre.cc:2449

[rguenth at gcc dot gnu.org via Gcc-bugs]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122079

--- Comment #5 from Richard Biener <rguenth at gcc dot gnu.org> ---
ANTIC_IN is MAX on 4->15
ANTIC_OUT[4] := { f_12(D) (0008), {mem_ref<0B>,f_12(D)}@.MEM_13(D) (0007),
{mem_ref<0B>,addr_expr<&a>}@.MEM_13(D) (0003) }
[changed] ANTIC_IN[4] := { f_12(D) (0008),
{mem_ref<0B>,addr_expr<&a>}@.MEM_13(D) (0003), c.0_1 (0001) }

->

ANTIC_OUT[4] := { f_12(D) (0008), _11 (0007), {mem_ref<0B>,f_12(D)}@.MEM_13(D)
(0007), {mem_ref<0B>,addr_expr<&a>}@.MEM_13(D) (0003) }
[changed] ANTIC_IN[4] := { f_12(D) (0008), _11 (0007),
{mem_ref<0B>,f_12(D)}@.MEM_13(D) (0007), {mem_ref<0B>,addr_expr<&a>}@.MEM_13(D)
(0003), c.0_1 (0001) }

->

ANTIC_OUT[4] := { f_12(D) (0008), a.2_3 (0003),
{mem_ref<0B>,addr_expr<&a>}@.MEM_13(D) (0003) }
[changed] ANTIC_IN[4] := { f_12(D) (0008), a.2_3 (0003),
{mem_ref<0B>,addr_expr<&a>}@.MEM_13(D) (0003), c.0_1 (0001) }


so it's I think again the issue that we prune values after removing a subset
of expressions but the MAX set is involved.  This time through
prune_clobbered_mems.