[clang] [llvm] Fix Windows EH IP2State tables (remove +1 bias) (PR #144745)

[Eli Friedman via llvm-commits]

================
@@ -37,9 +37,11 @@ cond.end:                                         ; preds = %entry, %cond.true
 ; CHECK: testq
 ; CHECK: je
 ; CHECK: callq alloc
+; CHECK-NEXT: nop
----------------
efriedma-quic wrote:

> No IP2State table mean no need to insert NOPs at all, even for SEH, right?

I think the SEH tables have essentially the same issue: if we don't insert a nop, the table entry overlaps with the epilogue or another table entry, and things break.  But if we don't need to handle the exception, we don't need a nop, I think.

In terms of modifying win64_call_epi.ll it probably makes sense to have tests for both?

https://github.com/llvm/llvm-project/pull/144745