<table border="1" cellspacing="0" cellpadding="8">
    <tr>
        <th>Issue</th>
        <td>
            <a href=https://github.com/llvm/llvm-project/issues/60969>60969</a>
        </td>
    </tr>

    <tr>
        <th>Summary</th>
        <td>
            [MLIR] Invalid IRs generated by irrelevant optimization pass
        </td>
    </tr>

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

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

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

<pre>
    This is a very interesting test case. I found that there are some passes on it that output invalid IRs. There is no `scf` dialect in this test case, so `-scf-for-to-while` should not take effect, but actually produces invalid IRs with no content in funcOp. 

I also found several similar bugs on this test case. Maybe the reason behind this problem may be similar, I will try to find the root cause of the bugs.

Reproduce cmd:
```
mlir-opt -scf-for-to-while core-ops.mlir
```
IR:
```
func.func @affine_apply1() {
  %i = "arith.constant"() {value = 0: index} : () -> index
  %j = "arith.constant"() {value = 1: index} : () -> index
  %a = "affine.apply" (%i) { map = affine_map<(d0) -> (d0 + 1)> } :
    (index) -> (index)
  %b = affine.apply affine_map<()[x] -> (x+1)>()[%i]
 return
}
```
Outputs:
```
module {
  func.func @affine_apply1() {
    return
  }
}
```

</pre>
<img width="1px" height="1px" alt="" src="http://email.email.llvm.org/o/eJyUVF-PmzgQ_zTmZRRETCDhgYemuUiRrqq06vtpgCG4Z2xkD8nmPv3JDptN1e7DShGRze_feMyg9-psiGpR7EVxSHDmwbraX4mwSxrb3eofg_KgPCBcyN1AGSZHnpU5A5NnaNFTCifo7Ww64AEZeCBHgI7A25FgQu_JgzWg-A6wM08zgzIX1KqD04tP4UckKQ_Ggigz3_aizKBTqKkNUOAQ5GEp5FfwEbnybb_qrVuxXV0HpSnw_GBn3YGxDIz_ElDfU8uB1MwM2PKMWt9gcrabW_LPUeCqeAgpWmuYTPTuZ9N-n1IQ2UFkX-7PE6D2dqnb04UcavBqVBodNPM5Vvxr6BS-4a2hcEDgCL010NCg4rkpH9I0mkYY8QYNvWmF0Ce4Kq2B3Q3YQn9nEDhrg_DsCWwfd4Jv-pzyhZYaoR07kS-7osyWX1yOWrmVnRh-O0toraOVnXwaMH8kn14-kg2nloYHiE2Gfa8M_YPTpG9rIXdCViC2-zsSQMhCgcgPIKREp3hIW2s8o2Eh5Tv8gnqmiMtE_gWU6ehVbA8QFgtqJfK_lhfv2j8_pb3-lDY-tGOJaSxRSHknFWqRhxGniFxOYsRJ5F-F3HXZQzquQMg9rIWs4s49wJtdMNzdAzxx3jaeQjVPVvdEv_kGQrF_FcXhIfQq5H5xfgBiCcVh0XbEszNLu7eHP_b9e_y8_Ye3zXazpufmf-aiwC8RAN5DfJAm6eq8q_IKE6rX5XYr5baqimSoO1msd7LEdUU7aspivc0z2TflJts066LcJKqWmcwzKTfZLq-KIt0UuKNd2VVV1fX9eic2GY2odKr1ZUytOyfK-5nqMqvKKtHYkPZxtEpp6ArxZbhyxSFxdeCswvcqNplWnv27CivWcSZ_-_v0EvpzehpPZzLkkKmD5gbKOdJ0QcNgJ1aj-g9ZWRNnbjI7XQ_MU-yEPAp5PCse5iZt7SjkMdgtf6vJ2Z9xQB5jSC_kMRbxfwAAAP__WNvSFg">