<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/60942>60942</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            Basic floating-point operations are underspecified
        </td>
    </tr>

    <tr>
      <th>Labels</th>
      <td>
            new issue
      </td>
    </tr>

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

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

<pre>
    Currently, the LangRef does not specify the results of basic floating-point operations (`fadd`, `fsub`, `fmul`, `fdiv`) in any detail. APFloat uses IEEE 754 semantics, but the LangRef does not guarantee it. What guarantees are there about the behavior of floating-point code? If IEEE 754 is the intended model, then x87 codegen is completely broken and probably other targets. If not IEEE 754, then what is the intended model?
</pre>
<img width="1px" height="1px" alt="" src="http://email.email.llvm.org/o/eJyEUs1unDwUfRqzucqIMT8DCxZJviBF-ipV3XRt4wu4NTbyvU46b1-Zps2oitQVOpjzh48isotHHETzIJr_CpV4DXH4lIIvdDDX4THFiJ7dVchH4BXhf-WXLziDCUjgAwPtONn5ehxGpOSYIMygFdkJZhcUW7_c7cF6hrBjVGyDJxCyE205K2NEW2bxjCjpG7Qld4OMfTlQD9aD8lcwyMq6E9x_HrMLJEKC56enJ7g0NRBuyrOdKPN14o_DL0lF5RkRLJ_g66puXhGoiJkWEZQObxIaV_ViQ8wd_2o3BYOiGuF5fo9h6WBZz-gNGtiCQff2Lz386C4Ha0Gfv5zCtjtkdFfQMXzH3NPAHoNW2l0h5CzAKi7IdMo2ucJvqz-ir7nFx77VWJihMn3VqwKHc3u5nOuyOdfFOpRz3bV9N2FVV7JudKfnHpuLwbOe627Gwg6ylFUpZVV28tzUp65Uba_Lumu7WUs1i7rELd-Icy_bKcSlsEQJh7bsa1k4pdHRsTMpPb7CcSikzLOLQ-bc6bSQqEtnieldhS07HB7-sad8V8kbjL_2aNEUKbphZd5JVPdCjkKOi-U16dMUNiHHbPD2uNtj-IYTCzkesUjI8Yj9MwAA__9ZnAne">