[Lldb-commits] [lldb] [lldb] Support arbitrary precision integer registers (PR #166363)

[Matej Košík via lldb-commits]

================
@@ -207,6 +207,18 @@ Status RegisterValue::SetValueFromData(const RegisterInfo &reg_info,
         int128.x[1] = data1;
       }
       SetUInt128(llvm::APInt(128, 2, int128.x));
+    } else {
+      std::vector<uint8_t> bytes(src_len, 0);
+      for (size_t i = 0; i < src_len; i++)
+        bytes[i] = src.GetU8(&src_offset);
+      if (src.GetByteOrder() == eByteOrderBig)
+        std::reverse(bytes.begin(), bytes.end());
+      // The number of 64-bit wide words that are stored in "src".
----------------
sedymrak wrote:

I have pushed commit that (hopefully) explains the situation.

We have to be more careful when LLDB is running on big-endian machine. Below we call `llvm::LoadIntFromMemory` function. On big-endian machine that function reads from the provided region of memory in 64-bit (i.e. 8 byte) chunks. We must ensure that the whole value is stored in a memory region whose size a (smallest) multiple of 64-bit (i.e. 8 bytes).

https://github.com/llvm/llvm-project/pull/166363