<html>
    <head>
      <base href="https://bugs.llvm.org/">
    </head>
    <body><table border="1" cellspacing="0" cellpadding="8">
        <tr>
          <th>Bug ID</th>
          <td><a class="bz_bug_link 
          bz_status_NEW "
   title="NEW - [IPRA] wrong code"
   href="https://bugs.llvm.org/show_bug.cgi?id=36587">36587</a>
          </td>
        </tr>

        <tr>
          <th>Summary</th>
          <td>[IPRA] wrong code
          </td>
        </tr>

        <tr>
          <th>Product</th>
          <td>libraries
          </td>
        </tr>

        <tr>
          <th>Version</th>
          <td>trunk
          </td>
        </tr>

        <tr>
          <th>Hardware</th>
          <td>PC
          </td>
        </tr>

        <tr>
          <th>OS</th>
          <td>Linux
          </td>
        </tr>

        <tr>
          <th>Status</th>
          <td>NEW
          </td>
        </tr>

        <tr>
          <th>Severity</th>
          <td>enhancement
          </td>
        </tr>

        <tr>
          <th>Priority</th>
          <td>P
          </td>
        </tr>

        <tr>
          <th>Component</th>
          <td>Register Allocator
          </td>
        </tr>

        <tr>
          <th>Assignee</th>
          <td>unassignedbugs@nondot.org
          </td>
        </tr>

        <tr>
          <th>Reporter</th>
          <td>paulsson@linux.vnet.ibm.com
          </td>
        </tr>

        <tr>
          <th>CC</th>
          <td>llvm-bugs@lists.llvm.org
          </td>
        </tr></table>
      <p>
        <div>
        <pre>Created <span class=""><a href="attachment.cgi?id=19992" name="attach_19992" title="reduced testcase">attachment 19992</a> <a href="attachment.cgi?id=19992&action=edit" title="reduced testcase">[details]</a></span>
reduced testcase

It seems that enabling interprocedural register allocation with (reduced)
csmith program gives wrong code:

bin/llc -mtriple=s390x-linux-gnu -mcpu=z13  tc_ipra_reduced.ll -o tc.s
-enable-ipra                                                                    
bin/clang -O3 -march=z13 tc.s -o a.out -w; ./a.out                              
checksum = 0                                   

bin/llc -mtriple=s390x-linux-gnu -mcpu=z13  tc_ipra_reduced.ll -o tc.s
bin/clang -O3 -march=z13 tc.s -o a.out -w; ./a.out                              
checksum = 5 
(checksum = 5 also with gcc -O0 and more)

At first glance, it seems that while fn2() actually uses %r0 (which is caller
saved), ipra (right) seems to result in %r0 being live accross the call to
fn2():

                                        IPRA

        lgrl    %r0, q       |          llihl   %r0, 0
        lgrl    %r1, o       |          stgrl   %r1, n
        larl    %r13, a      |          stg     %r1, 0(%r2)
        lhi     %r12, 0      <
        stgrl   %r0, n       <
        stg     %r0, 0(%r1)  <
        brasl   %r14, fn2@PL            brasl   %r14, fn2@PL
        larl    %r1, f                  larl    %r1, f
        chsi    0(%r1), 7               chsi    0(%r1), 7
        lgrl    %r1, m                  lgrl    %r1, m
        lochie  %r12, 1      |          lochhie %r0, 1
        st      %r12, 0(%r1) |          stfh    %r0, 0(%r1)


To the left, %r12 is used by first setting it to 0, then after fn2()
conditionally set it to 1, and then store it to @m.

To the right, %r0 is used the exact same way, which is incorrect since fn2()
changes its value.</pre>
        </div>
      </p>


      <hr>
      <span>You are receiving this mail because:</span>

      <ul>
          <li>You are on the CC list for the bug.</li>
      </ul>
    </body>
</html>