[clang] [llvm] [X86][AVX10.2] Support YMM rounding new instructions (PR #101825)

[Shengchen Kan via cfe-commits]

================
@@ -785,9 +785,9 @@ void DisassemblerTables::emitModRMDecision(raw_ostream &o1, raw_ostream &o2,
     break;
   }
 
-  // We assume that the index can fit into uint16_t.
-  assert(sEntryNumber < 65536U &&
-         "Index into ModRMDecision is too large for uint16_t!");
+  // We assume that the index can fit into uint32_t.
+  assert(sEntryNumber < -1U &&
+         "Index into ModRMDecision is too large for uint32_t!");
----------------
KanRobert wrote:

I suggest not using the tricky `-1`, it's not robust b/c the suffix `U` can represent unsigned long long (https://en.cppreference.com/w/cpp/language/integer_literal)

https://github.com/llvm/llvm-project/pull/101825