[RABasic] Properly update the LiveRegMatrix when LR splitting occur

Prior to this patch we used to not touch the LiveRegMatrix while doing
live-range splitting. In other words, when live-range splitting was
occurring, the LiveRegMatrix was not reflecting the changes.
This is generally fine because it means the query to the LiveRegMatrix
will be conservately correct. However, when decisions are taken based on
what is going to happen on the interferences (e.g., when we spill a
register and know that it is going to be available for another one), we
might hit an assertion that the color used for the assignment is still
in use.

This patch makes sure the changes on the live-ranges are properly
reflected in the LiveRegMatrix, so the assertions don't break.
An alternative could have been to remove the assertion, but it would
make the invariants of the code and the general reasoning more
complicated in my opnion.

http://llvm.org/PR33057

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@304603 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/CodeGen/RegAllocBasic.cpp b/lib/CodeGen/RegAllocBasic.cpp
index e8ecea98f5672ff9f92a84b4f9ee3065226a5b2a..24be7ea98d82434ea4fa958259d68c216de9a889 100644 (file)
--- a/lib/CodeGen/RegAllocBasic.cpp
+++ b/lib/CodeGen/RegAllocBasic.cpp
@@ -58,8 +58,9 @@ namespace {
 /// whenever a register is unavailable. This is not practical in production but
 /// provides a useful baseline both for measuring other allocators and comparing
 /// the speed of the basic algorithm against other styles of allocators.
-class RABasic : public MachineFunctionPass, public RegAllocBase
-{
+class RABasic : public MachineFunctionPass,
+                public RegAllocBase,
+                private LiveRangeEdit::Delegate {
   // context
   MachineFunction *MF;
 
@@ -72,6 +73,9 @@ class RABasic : public MachineFunctionPass, public RegAllocBase
   // selectOrSplit().
   BitVector UsableRegs;
 
+  bool LRE_CanEraseVirtReg(unsigned) override;
+  void LRE_WillShrinkVirtReg(unsigned) override;
+
 public:
   RABasic();
 
@@ -138,6 +142,28 @@ INITIALIZE_PASS_DEPENDENCY(LiveRegMatrix)
 INITIALIZE_PASS_END(RABasic, "regallocbasic", "Basic Register Allocator", false,
                     false)
 
+bool RABasic::LRE_CanEraseVirtReg(unsigned VirtReg) {
+  if (VRM->hasPhys(VirtReg)) {
+    LiveInterval &LI = LIS->getInterval(VirtReg);
+    Matrix->unassign(LI);
+    aboutToRemoveInterval(LI);
+    return true;
+  }
+  // Unassigned virtreg is probably in the priority queue.
+  // RegAllocBase will erase it after dequeueing.
+  return false;
+}
+
+void RABasic::LRE_WillShrinkVirtReg(unsigned VirtReg) {
+  if (!VRM->hasPhys(VirtReg))
+    return;
+
+  // Register is assigned, put it back on the queue for reassignment.
+  LiveInterval &LI = LIS->getInterval(VirtReg);
+  Matrix->unassign(LI);
+  enqueue(&LI);
+}
+
 RABasic::RABasic(): MachineFunctionPass(ID) {
 }
 
@@ -207,7 +233,7 @@ bool RABasic::spillInterferences(LiveInterval &VirtReg, unsigned PhysReg,
     Matrix->unassign(Spill);
 
     // Spill the extracted interval.
-    LiveRangeEdit LRE(&Spill, SplitVRegs, *MF, *LIS, VRM, nullptr, &DeadRemats);
+    LiveRangeEdit LRE(&Spill, SplitVRegs, *MF, *LIS, VRM, this, &DeadRemats);
     spiller().spill(LRE);
   }
   return true;
@@ -266,7 +292,7 @@ unsigned RABasic::selectOrSplit(LiveInterval &VirtReg,
   DEBUG(dbgs() << "spilling: " << VirtReg << '\n');
   if (!VirtReg.isSpillable())
     return ~0u;
-  LiveRangeEdit LRE(&VirtReg, SplitVRegs, *MF, *LIS, VRM, nullptr, &DeadRemats);
+  LiveRangeEdit LRE(&VirtReg, SplitVRegs, *MF, *LIS, VRM, this, &DeadRemats);
   spiller().spill(LRE);
 
   // The live virtual register requesting allocation was spilled, so tell

diff --git a/test/CodeGen/SystemZ/RAbasic-invalid-LR-update.mir b/test/CodeGen/SystemZ/RAbasic-invalid-LR-update.mir
new file mode 100644 (file)
index 0000000..eadad68
--- /dev/null
+++ b/test/CodeGen/SystemZ/RAbasic-invalid-LR-update.mir
@@ -0,0 +1,279 @@
+# RUN: llc -mtriple=s390x-linux-gnu -mcpu=z13 -run-pass=regallocbasic %s -o - | FileCheck %s
+# This test used to assert in RABasic. The problem was when we split live-ranges,
+# we were not updating the LiveRegMatrix properly and the interference calculation
+# wouldn't match what the assignment thought it could do.
+# In other words, this test case needs to trigger live-range splitting to exercise
+# the problem.
+#
+# PR33057
+--- |
+  target datalayout = "E-m:e-i1:8:16-i8:8:16-i64:64-f128:64-v128:64-a:8:16-n32:64"
+  target triple = "s390x--linux-gnu"
+  
+  define void @autogen_SD21418() #0 {
+    ret void
+  }
+  
+  attributes #0 = { "target-cpu"="z13" }
+
+...
+
+# CHECK: name: autogen_SD21418
+# Check that at least one live-range has been split
+# CHECK: id: 114, class
+---
+name:            autogen_SD21418
+alignment:       2
+tracksRegLiveness: true
+registers:       
+  - { id: 0, class: vr128bit }
+  - { id: 1, class: vr128bit }
+  - { id: 2, class: vr128bit }
+  - { id: 3, class: vr64bit }
+  - { id: 4, class: gr64bit }
+  - { id: 5, class: vr128bit }
+  - { id: 6, class: grx32bit }
+  - { id: 7, class: vr128bit }
+  - { id: 8, class: vr128bit }
+  - { id: 9, class: gr32bit }
+  - { id: 10, class: gr64bit }
+  - { id: 11, class: vr128bit }
+  - { id: 12, class: fp64bit }
+  - { id: 13, class: vr64bit }
+  - { id: 14, class: vr64bit }
+  - { id: 15, class: gr64bit }
+  - { id: 16, class: gr128bit }
+  - { id: 17, class: gr64bit }
+  - { id: 18, class: gr32bit }
+  - { id: 19, class: gr32bit }
+  - { id: 20, class: gr128bit }
+  - { id: 21, class: gr32bit }
+  - { id: 22, class: gr64bit }
+  - { id: 23, class: gr32bit }
+  - { id: 24, class: gr32bit }
+  - { id: 25, class: gr128bit }
+  - { id: 26, class: grx32bit }
+  - { id: 27, class: gr64bit }
+  - { id: 28, class: gr64bit }
+  - { id: 29, class: vr128bit }
+  - { id: 30, class: vr128bit }
+  - { id: 31, class: gr64bit }
+  - { id: 32, class: gr32bit }
+  - { id: 33, class: gr32bit }
+  - { id: 34, class: gr128bit }
+  - { id: 35, class: gr32bit }
+  - { id: 36, class: vr128bit }
+  - { id: 37, class: gr64bit }
+  - { id: 38, class: gr32bit }
+  - { id: 39, class: gr32bit }
+  - { id: 40, class: gr128bit }
+  - { id: 41, class: gr32bit }
+  - { id: 42, class: addr64bit }
+  - { id: 43, class: grx32bit }
+  - { id: 44, class: addr64bit }
+  - { id: 45, class: vr64bit }
+  - { id: 46, class: vr64bit }
+  - { id: 47, class: gr32bit }
+  - { id: 48, class: gr32bit }
+  - { id: 49, class: grx32bit }
+  - { id: 50, class: vr64bit }
+  - { id: 51, class: gr64bit }
+  - { id: 52, class: grx32bit }
+  - { id: 53, class: gr32bit }
+  - { id: 54, class: gr64bit }
+  - { id: 55, class: grx32bit }
+  - { id: 56, class: gr32bit }
+  - { id: 57, class: gr128bit }
+  - { id: 58, class: gr128bit }
+  - { id: 59, class: gr32bit }
+  - { id: 60, class: gr64bit }
+  - { id: 61, class: grx32bit }
+  - { id: 62, class: gr32bit }
+  - { id: 63, class: gr64bit }
+  - { id: 64, class: grx32bit }
+  - { id: 65, class: gr32bit }
+  - { id: 66, class: gr128bit }
+  - { id: 67, class: gr128bit }
+  - { id: 68, class: grx32bit }
+  - { id: 69, class: gr64bit }
+  - { id: 70, class: gr64bit }
+  - { id: 71, class: vr128bit }
+  - { id: 72, class: vr128bit }
+  - { id: 73, class: gr64bit }
+  - { id: 74, class: grx32bit }
+  - { id: 75, class: gr32bit }
+  - { id: 76, class: gr64bit }
+  - { id: 77, class: grx32bit }
+  - { id: 78, class: gr32bit }
+  - { id: 79, class: gr128bit }
+  - { id: 80, class: gr128bit }
+  - { id: 81, class: gr32bit }
+  - { id: 82, class: vr128bit }
+  - { id: 83, class: gr64bit }
+  - { id: 84, class: grx32bit }
+  - { id: 85, class: gr32bit }
+  - { id: 86, class: gr64bit }
+  - { id: 87, class: grx32bit }
+  - { id: 88, class: gr32bit }
+  - { id: 89, class: gr128bit }
+  - { id: 90, class: gr128bit }
+  - { id: 91, class: gr32bit }
+  - { id: 92, class: grx32bit }
+  - { id: 93, class: gr64bit }
+  - { id: 94, class: gr32bit }
+  - { id: 95, class: gr32bit }
+  - { id: 96, class: gr32bit }
+  - { id: 97, class: gr64bit }
+  - { id: 98, class: gr64bit }
+  - { id: 99, class: grx32bit }
+  - { id: 100, class: grx32bit }
+  - { id: 101, class: gr128bit }
+  - { id: 102, class: gr128bit }
+  - { id: 103, class: gr128bit }
+  - { id: 104, class: gr64bit }
+  - { id: 105, class: gr128bit }
+  - { id: 106, class: gr128bit }
+  - { id: 107, class: gr64bit }
+  - { id: 108, class: gr128bit }
+  - { id: 109, class: gr128bit }
+  - { id: 110, class: gr64bit }
+  - { id: 111, class: gr128bit }
+  - { id: 112, class: gr128bit }
+  - { id: 113, class: gr64bit }
+constants:       
+  - id:              0
+    value:           double 0xD55960F86F577076
+    alignment:       8
+body:             |
+  bb.0:
+    %11 = VGBM 0
+    %43 = LHIMux 0
+    %44 = LARL %const.0
+    %45 = VL64 %44, 0, _ :: (load 8 from constant-pool)
+  
+  bb.1:
+    ADJCALLSTACKDOWN 0, 0
+    %12 = LZDR
+    %f0d = COPY %12
+    CallBRASL $fmod, killed %f0d, undef %f2d, csr_systemz, implicit-def dead %r14d, implicit-def dead %cc, implicit-def %f0d
+    ADJCALLSTACKUP 0, 0
+    KILL killed %f0d
+  
+  bb.2:
+    successors: %bb.2(0x7c000000), %bb.3(0x04000000)
+  
+    %17 = VLGVH %11, _, 0
+    %19 = LHR %17.subreg_l32
+    undef %20.subreg_l64 = LGHI 0
+    %20 = DSGFR %20, %19
+    %22 = VLGVH %11, _, 3
+    %24 = LHR %22.subreg_l32
+    undef %25.subreg_l64 = LGHI 0
+    %25 = DSGFR %25, %24
+    %31 = VLGVH %11, _, 1
+    %33 = LHR %31.subreg_l32
+    undef %34.subreg_l64 = LGHI 0
+    %34 = DSGFR %34, %33
+    %37 = VLGVH %11, _, 2
+    %39 = LHR %37.subreg_l32
+    undef %40.subreg_l64 = LGHI 0
+    %40 = DSGFR %40, %39
+    CHIMux %43, 0, implicit-def %cc
+    BRC 14, 6, %bb.2, implicit killed %cc
+    J %bb.3
+  
+  bb.3:
+    successors: %bb.1(0x7c000000), %bb.4(0x04000000)
+  
+    WFCDB undef %46, %45, implicit-def %cc
+    %48 = IPM implicit killed %cc
+    %48 = AFIMux %48, 268435456, implicit-def dead %cc
+    %6 = RISBMux undef %6, %48, 31, 159, 35
+    WFCDB undef %50, %45, implicit-def %cc
+    BRC 15, 6, %bb.1, implicit killed %cc
+    J %bb.4
+  
+  bb.4:
+    %36 = VLVGP %25.subreg_l64, %25.subreg_l64
+    %36 = VLVGH %36, %20.subreg_l32, _, 0
+    %36 = VLVGH %36, %34.subreg_l32, _, 1
+    dead %36 = VLVGH %36, %40.subreg_l32, _, 2
+    %4 = LG undef %42, 0, _ :: (load 8 from `i64* undef`)
+    undef %57.subreg_h64 = LLILL 0
+    undef %66.subreg_h64 = LLILL 0
+    undef %79.subreg_h64 = LLILL 0
+    undef %89.subreg_h64 = LLILL 0
+    %92 = LHIMux 0
+  
+  bb.5:
+  
+  bb.6:
+    successors: %bb.6, %bb.7
+  
+    %51 = VLGVH undef %7, _, 0
+    %53 = LLHRMux %51.subreg_l32
+    %54 = VLGVH undef %1, _, 0
+    %57.subreg_l32 = LLHRMux %54.subreg_l32
+    %58 = COPY %57
+    %58 = DLR %58, %53
+    %60 = VLGVH undef %7, _, 3
+    %62 = LLHRMux %60.subreg_l32
+    %63 = VLGVH undef %1, _, 3
+    %66.subreg_l32 = LLHRMux %63.subreg_l32
+    %67 = COPY %66
+    %67 = DLR %67, %62
+    %73 = VLGVH undef %7, _, 1
+    %75 = LLHRMux %73.subreg_l32
+    %76 = VLGVH undef %1, _, 1
+    %79.subreg_l32 = LLHRMux %76.subreg_l32
+    %80 = COPY %79
+    %80 = DLR %80, %75
+    %83 = VLGVH undef %7, _, 2
+    %85 = LLHRMux %83.subreg_l32
+    %86 = VLGVH undef %1, _, 2
+    %89.subreg_l32 = LLHRMux %86.subreg_l32
+    %90 = COPY %89
+    %90 = DLR %90, %85
+    CHIMux %92, 0, implicit-def %cc
+    BRC 14, 6, %bb.7, implicit killed %cc
+    J %bb.6
+  
+  bb.7:
+    successors: %bb.6(0x7c000000), %bb.8(0x04000000)
+  
+    CGHI undef %93, 0, implicit-def %cc
+    %96 = IPM implicit killed %cc
+    CGHI undef %97, 0, implicit-def %cc
+    BRC 14, 6, %bb.6, implicit killed %cc
+  
+  bb.8:
+    successors: %bb.5(0x7c000000), %bb.9(0x04000000)
+  
+    CHIMux %6, 0, implicit-def %cc
+    %10 = LLILL 41639
+    dead %10 = LOCGR %10, %4, 14, 6, implicit killed %cc
+    CHIMux %92, 0, implicit-def %cc
+    BRC 14, 6, %bb.5, implicit killed %cc
+    J %bb.9
+  
+  bb.9:
+    %82 = VLVGP %67.subreg_h64, %67.subreg_h64
+    %82 = VLVGH %82, %58.subreg_hl32, _, 0
+    %82 = VLVGH %82, %80.subreg_hl32, _, 1
+    dead %82 = VLVGH %82, %90.subreg_hl32, _, 2
+    %96 = AFIMux %96, 1879048192, implicit-def dead %cc
+    %96 = SRL %96, _, 31
+    dead %11 = VLVGF %11, %96, _, 1
+    %100 = LHIMux 0
+  
+  bb.10:
+    successors: %bb.10(0x7c000000), %bb.11(0x04000000)
+  
+    CHIMux %100, 0, implicit-def %cc
+    BRC 14, 6, %bb.10, implicit killed %cc
+    J %bb.11
+  
+  bb.11:
+    Return
+
+...