[llvm-bugs] [Bug 25016] New: Register Spills Caused by the Machine Reassociation Optimization

[via llvm-bugs]

https://llvm.org/bugs/show_bug.cgi?id=25016

            Bug ID: 25016
           Summary: Register Spills Caused by the Machine Reassociation
                    Optimization
           Product: libraries
           Version: trunk
          Hardware: All
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P
         Component: Scalar Optimizations
          Assignee: unassignedbugs at nondot.org
          Reporter: haicheng at codeaurora.org
                CC: llvm-bugs at lists.llvm.org, mssimpso at codeaurora.org,
                    spatel+llvm at rotateright.com
    Classification: Unclassified

Created attachment 14976
  --> https://llvm.org/bugs/attachment.cgi?id=14976&action=edit
a .ll test case that delays the addition of the first number

I observed some extra register spills when applying the reassociation pass on
some programs.

For example, code like this

…
X=a+b
…
Y=X+c
…
Z=Y+d
…

There are some other instructions between these float adds.  The reassociation
pass first swaps a and c when checking the second add, and then swaps a and d
when checking the third add.  The transformed code looks like

…
X=c+b
…
Y=X+d
…
Z=Y+a

a is pushed all the way down to the bottom and its live range is much larger
now.  

A .ll test case is attached which computes a+b+c+d+e+f+g+h.  The transformation
delays the computation of a to the end and it becomes b+c+d+e+f+g+h+a

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20151001/539eb929/attachment.html>