[llvm-bugs] [Bug 39462] New: "ran out of registers during register allocation" with MEMCPY on Thumb1

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=39462

            Bug ID: 39462
           Summary: "ran out of registers during register allocation" with
                    MEMCPY on Thumb1
           Product: libraries
           Version: trunk
          Hardware: All
                OS: All
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: Backend: ARM
          Assignee: unassignedbugs at nondot.org
          Reporter: efriedma at codeaurora.org
                CC: llvm-bugs at lists.llvm.org, matze at braunis.de,
                    peter.smith at linaro.org, peter at pcc.me.uk,
                    Ties.Stuij at arm.com

Testcase follows; run with "llc -enable-misched".  Has something to do with the
MEMCPY pseudo (there are 8 total Thumb1 registers, and MEMCPY requires six; if
the register allocator messes up, it's easy to run out of registers).  But I'm
not really familiar with the relevant code.

target datalayout = "e-m:e-p:32:32-i64:64-v128:64:128-a:0:32-n32-S64"
target triple = "thumbv6m-none-unknown-musleabi"

@g = external constant [8 x i32], align 4

declare void @llvm.memcpy.p0i8.p0i8.i32(i8* nocapture writeonly,
                                        i8* nocapture readonly, i32, i1)
declare void @f1(i8* readonly, i32)
declare i8* @f2(i8*)

define i8* @f(i8* readonly %data, i32 %l, i8* %i, i8* %foo) {
entry:
  call void @llvm.memcpy.p0i8.p0i8.i32(
      i8* nonnull align 8 %foo, 
      i8* align 4 bitcast ([8 x i32]* @g to i8*),
      i32 32, i1 false)
  call void @f1(i8* %data, i32 %l)
  %call = call i8* @f2(i8* %i)
  ret i8* %call
}

-- 
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/20181026/6b7d1a8f/attachment.html>