<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/211007>211007</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            [lldb] DW_OP_fbreg widens its generic result on 32-bit targets
        </td>
    </tr>

    <tr>
      <th>Labels</th>
      <td>
            lldb
      </td>
    </tr>

    <tr>
      <th>Assignees</th>
      <td>
      </td>
    </tr>

    <tr>
      <th>Reporter</th>
      <td>
          firmiana402
      </td>
    </tr>
</table>

<pre>
    
[DWARF v5](https://dwarfstd.org/doc/DWARF5.pdf) Section 2.5.1.2 specifies that `DW_OP_fbreg` produces the frame-base address plus its signed displacement with the generic type. LLVM's [location-description stack extension](https://llvm.org/docs/AMDGPUDwarfExtensionAllowLocationDescriptionOnTheDwarfExpressionStack/AMDGPUDwarfExtensionAllowLocationDescriptionOnTheDwarfExpressionStack.html#a-2-5-4) instead models the result as a default-address-space memory location, but requires it to be converted to a generic value when an arithmetic operation consumes it. Under either model, i386 value arithmetic must use a 32-bit result.

LLDB adds the displacement as an `int64_t` through `Scalar` arithmetic and pushes the result without canonicalizing it to the target address width. A 32-bit frame-base value can therefore become a 64-bit scalar.

## Source Evidence

[`Evaluate_DW_OP_fbreg`](https://github.com/llvm/llvm-project/blob/main/lldb/source/Expression/DWARFExpression.cpp) currently does:

```cpp
Scalar value;
if (llvm::Error err = eval_ctx.frame->GetFrameBaseValue(value))
  return err;
value += fbreg_offset;
eval_ctx.stack.push_back(value);
eval_ctx.stack.back().SetValueType(Value::ValueType::LoadAddress);
```

`fbreg_offset` is an `int64_t`, so the implicit `Scalar` operand is 64 bits. `Scalar::operator+=` promotes to the wider operand before adding, and the handler never truncates or extends the result to `8 * address_size`.

**Related to #209728, but not a duplicate.**

</pre>
<img width="1" height="1" alt="" src="http://email.email.llvm.org/o/eJysVU2P4jgQ_TXmUiIKzgdw4BCWZi49mtH0fByRY1eIdxM7a1fo6fn1KzsBGu1cVlopUnecqlevHq_Kwnt9Nog7VuxZcViIkVrrdo12vRZG5Clf1Fa97VhasWJ_-FF9OcKlYMWB8U1LNHiWVYwfGT-qV-EaTyqx7hxerWT8GBOKZFAN41t4QUnaGuBJkawSDn5AqRuNHqgVBKxMDz9Onz6fmtrhmZUpDM6qUcbvCI0TPS5r4RGEUg69h6EbPWjyEJtQoLQfOiGxR0PwqqmNiWc06LQEehswgefn7x8ZX3tgxb6zUgRGS4VeOj1Edp6E_AvwJ6Hx2prf9dp1l_7ep2f8WH08fPj87RA0eLpmVl1nX5_nEod7hU_ma4tz6BD60Na8hKL_F07SUt8xnoklXxbLPEivjScUCnqrsJv0dOjHjkB4EKCwEWNHy1nYpR-EROixt-4Nriox_gfUI4HDv0ftMCgPZKFGkNZc0BGq8C5ugl9ENyK8tmhAGBBOU9sjaQl2QBchQ6Yf-4iVwDej0AFqatFNTENJnW3KGeodRD96gjF4ATK-rDXN_STBqWn1_HzYB5tMrT74IjRsgtm0oTI_UTAatc6O5zacvkjRCRcO31UTRsEw-hYfpAsOsyOBFMYaLUWnf2lznmUJcSTcGelm11etqE2gujJ-5-ipPylMyHPYWIdQo7R9aLDMY7iPzOYGGc8Yz-DFjk4iPF20QiNx_lbsWZk-BUxBeHocqt_5-aypHetE2n429_xnOTj7J0pi_Fh3tmb82Att4kcV3nyszvjxbsDr0N9PEjkMwYJydA4NdW-gLMbaE9kynZ4QllaT_JMeLNuztNINML6JrLKKZdWTc9YBOgcsOwBeRHeS9DOZxGTZ0wekY_h_Lzx-jzB8M8HxbXjSCsAhjc4EkKnGJD_j-wAZlTrZpvFI0-dbkbgbkuCEUx0H9o78u8A5hvFt8oIUyXx9GwKhiVjs534cX5-tUNVkmBvsTaObZA8cyxT0v00dZsdPPtT90Gmp6dHfcQqNCrllDrUmn7wLiGSmQbVuUmZeyb2lMAcT9KsOM3uFqifnCqW0OQcC4TCEtcKoDh0YvKADcqORIqCEXzLsOfUwWGQDkQ0wXl2H5-T1L2RlerN_eL5gJ-atw3jG0-06iD1tKWMpLLYxdC4IkymDpdVC7TK1zbZigbvVepOX-booNot21zT5tlCpzLdlU-arusjydLXZbvNNmqsyXS30jqe8TNd8tSp4UWwSzLdNndcZV81mzTfI8hR7obvkekEstPcj7vhqlabrRSdq7Hy8ajmfRoiHS9ft4rDV49mzPO20J39HIE1dvJ5jQnGAd-Mc1TfTFXhdurOE1ly3zLSE_GJ03e4_j33kHy64uYXLjv8TAAD__-qFuHM">