Merge branch 'main' into vector_combine4

Author: RKSimon

lldb/source/Target/Process.cpp

@@ -3256,6 +3256,7 @@ Status Process::ConnectRemote(llvm::StringRef remote_url) {
       if (state == eStateStopped || state == eStateCrashed) {
         // If we attached and actually have a process on the other end, then
         // this ended up being the equivalent of an attach.
+        SetShouldDetach(true);
         CompleteAttach();
 
         // This delays passing the stopped event to listeners till

lldb/test/API/functionalities/gdb_remote_client/TestConnectRemoteDetach.py

@@ -0,0 +1,67 @@
+"""
+Test that ConnectRemote sets ShouldDetach flag correctly.
+
+When connecting to a remote process that stops after connection,
+the process should be marked for detach (not kill) on destruction.
+"""
+
+import lldb
+from lldbsuite.test.lldbtest import *
+from lldbsuite.test.decorators import *
+from lldbsuite.test.gdbclientutils import *
+from lldbsuite.test.lldbgdbclient import GDBRemoteTestBase
+from lldbsuite.test import lldbutil
+
+
+class TestConnectRemoteDetach(GDBRemoteTestBase):
+    """Test that ConnectRemote properly sets ShouldDetach flag."""
+
+    class StoppedResponder(MockGDBServerResponder):
+        """A responder that returns a stopped process."""
+
+        def qfThreadInfo(self):
+            return "m1"
+
+        def qsThreadInfo(self):
+            return "l"
+
+        def qC(self):
+            return "QC1"
+
+        def haltReason(self):
+            # Return that we're stopped
+            return "T05thread:1;"
+
+        def cont(self):
+            # Stay stopped
+            return "T05thread:1;"
+
+        def D(self):
+            # Detach packet: this is what we want to verify gets called.
+            return "OK"
+
+        def k(self):
+            # Kill packet: this is what we want to verify doesn't get called.
+            raise RuntimeError("should not receive k(ill) packet")
+
+    def test_connect_remote_sets_detach(self):
+        """Test that ConnectRemote to a stopped process sets ShouldDetach."""
+        self.server.responder = self.StoppedResponder()
+
+        target = self.createTarget("a.yaml")
+        process = self.connect(target)
+
+        # Wait for the process to be in stopped state after connecting.
+        # When ConnectRemote connects to a remote process that is stopped,
+        # it should call SetShouldDetach(true) before CompleteAttach().
+        lldbutil.expect_state_changes(
+            self, self.dbg.GetListener(), process, [lldb.eStateStopped]
+        )
+
+        # Now destroy the process. Because ShouldDetach was set to true
+        # during ConnectRemote, this should send a 'D' (detach) packet
+        # rather than a 'k' (kill) packet when the process is destroyed.
+        process.Destroy()
+
+        # Verify that the (D)etach packet was sent.
+        self.assertPacketLogReceived(["D"])

llvm/docs/AMDGPUUsage.rst

@@ -1180,6 +1180,51 @@ is conservatively correct for OpenCL.
                              other operations within the same address space.
      ======================= ===================================================
 
+Target Types
+------------
+
+The AMDGPU backend implements some target extension types.
+
+.. _amdgpu-types-named-barriers:
+
+Named Barriers
+~~~~~~~~~~~~~~
+
+Named barriers are fixed function hardware barrier objects that are available
+in gfx12.5+ in addition to the traditional default barriers.
+
+In LLVM IR, named barriers are represented by global variables of type
+``target("amdgcn.named.barrier", 0)`` in the LDS address space. Named barrier
+global variables do not occupy actual LDS memory, but their lifetime and
+allocation scope matches that of global variables in LDS. Programs in LLVM IR
+refer to named barriers using pointers.
+
+The following named barrier types are supported in global variables, defined
+recursively:
+
+* a single, standalone ``target("amdgcn.named.barrier", 0)``
+* an array of supported types
+* a struct containing a single element of supported type
+
+.. code-block:: llvm
+
+      @bar = addrspace(3) global target("amdgcn.named.barrier", 0) undef
+      @foo = addrspace(3) global [2 x target("amdgcn.named.barrier", 0)] undef
+      @baz = addrspace(3) global { target("amdgcn.named.barrier", 0) } undef
+
+      ...
+
+      %foo.i = getelementptr [2 x target("amdgcn.named.barrier", 0)], ptr addrspace(3) @foo, i32 0, i32 %i
+      call void @llvm.amdgcn.s.barrier.signal.var(ptr addrspace(3) %foo.i, i32 0)
+
+Named barrier types may not be used in ``alloca``.
+
+Named barriers do not have an underlying byte representation.
+It is undefined behavior to use a pointer to any part of a named barrier object
+as the pointer operand of a regular memory access instruction or intrinsic.
+Pointers to named barrier objects are intended to be used with dedicated
+intrinsics. Reading from or writing to such pointers is undefined behavior.
+
 LLVM IR Intrinsics
 ------------------
 

llvm/lib/Analysis/DependenceAnalysis.cpp

@@ -407,9 +407,10 @@ static void dumpExampleDependence(raw_ostream &OS, DependenceInfo *DA,
         continue;
       Value *Ptr = getLoadStorePointerOperand(&Inst);
       const Loop *L = LI.getLoopFor(Inst.getParent());
+      const Loop *OutermostLoop = L ? L->getOutermostLoop() : nullptr;
       const SCEV *PtrSCEV = SE.getSCEVAtScope(Ptr, L);
       const SCEV *AccessFn = SE.removePointerBase(PtrSCEV);
-      SCEVMonotonicity Mon = Checker.checkMonotonicity(AccessFn, L);
+      SCEVMonotonicity Mon = Checker.checkMonotonicity(AccessFn, OutermostLoop);
       OS.indent(2) << "Inst: " << Inst << "\n";
       OS.indent(4) << "Expr: " << *AccessFn << "\n";
       Mon.print(OS, 4);
@@ -945,6 +946,8 @@ SCEVMonotonicity SCEVMonotonicityChecker::invariantOrUnknown(const SCEV *Expr) {
 SCEVMonotonicity
 SCEVMonotonicityChecker::checkMonotonicity(const SCEV *Expr,
                                            const Loop *OutermostLoop) {
+  assert((!OutermostLoop || OutermostLoop->isOutermost()) &&
+         "OutermostLoop must be outermost");
   assert(Expr->getType()->isIntegerTy() && "Expr must be integer type");
   this->OutermostLoop = OutermostLoop;
   return visit(Expr);

llvm/test/Analysis/DependenceAnalysis/monotonicity-no-wrap-flags.ll

@@ -298,7 +298,8 @@ exit:
 }
 
 ; The value of step reccurence is not invariant with respect to the outer most
-; loop (the i-loop).
+; loop (the i-loop). It is theoretically multivariate monotonic by definition,
+; but we cannot handle non-affine addrec for now.
 ;
 ; offset_i = 0;
 ; for (int i = 0; i < 100; i++) {
@@ -312,7 +313,8 @@ define void @step_is_variant(ptr %a) {
 ; CHECK-NEXT:  Monotonicity check:
 ; CHECK-NEXT:    Inst: store i8 0, ptr %idx, align 1
 ; CHECK-NEXT:      Expr: {%offset.i,+,1}<nuw><nsw><%loop.j>
-; CHECK-NEXT:      Monotonicity: MultivariateSignedMonotonic
+; CHECK-NEXT:      Monotonicity: Unknown
+; CHECK-NEXT:      Reason: %offset.i
 ; CHECK-EMPTY:
 ; CHECK-NEXT:  Src: store i8 0, ptr %idx, align 1 --> Dst: store i8 0, ptr %idx, align 1
 ; CHECK-NEXT:    da analyze - confused!
@@ -346,6 +348,56 @@ exit:
   ret void
 }
 
+; The value of step reccurence is not invariant with respect to the outer most
+; loop (the i-loop). Actually, `offset_i` is not monotonic.
+;
+; offset_i = 0;
+; for (int i = 0; i < 100; i++) {
+;   for (int j = 0; j < 100; j++)
+;     a[offset_i + j] = 0;
+;   offset_i += (i % 2 == 0) ? -1 : 3;
+; }
+;
+define void @step_is_variant2(ptr %a) {
+; CHECK-LABEL: 'step_is_variant2'
+; CHECK-NEXT:  Monotonicity check:
+; CHECK-NEXT:    Inst: store i8 0, ptr %idx, align 1
+; CHECK-NEXT:      Expr: {%offset.i,+,1}<nsw><%loop.j>
+; CHECK-NEXT:      Monotonicity: Unknown
+; CHECK-NEXT:      Reason: %offset.i
+; CHECK-EMPTY:
+; CHECK-NEXT:  Src: store i8 0, ptr %idx, align 1 --> Dst: store i8 0, ptr %idx, align 1
+; CHECK-NEXT:    da analyze - confused!
+;
+entry:
+  br label %loop.i.header
+
+loop.i.header:
+  %i = phi i64 [ 0, %entry ], [ %i.inc, %loop.i.latch ]
+  %offset.i = phi i64 [ 0, %entry ], [ %offset.i.next, %loop.i.latch ]
+  %step.i.0 = phi i64 [ -1, %entry ], [ %step.i.1, %loop.i.latch ]
+  %step.i.1 = phi i64 [ 3, %entry ], [ %step.i.0, %loop.i.latch ]
+  br label %loop.j
+
+loop.j:
+  %j = phi i64 [ 0, %loop.i.header ], [ %j.inc, %loop.j ]
+  %offset = add nsw i64 %offset.i, %j
+  %idx = getelementptr inbounds i8, ptr %a, i64 %offset
+  store i8 0, ptr %idx
+  %j.inc = add nsw i64 %j, 1
+  %exitcond.j = icmp eq i64 %j.inc, 100
+  br i1 %exitcond.j, label %loop.i.latch, label %loop.j
+
+loop.i.latch:
+  %i.inc = add nsw i64 %i, 1
+  %offset.i.next = add nsw i64 %offset.i, %step.i.0
+  %exitcond.i = icmp eq i64 %i.inc, 100
+  br i1 %exitcond.i, label %exit, label %loop.i.header
+
+exit:
+  ret void
+}
+
 ; The AddRec doesn't have nsw flag for the j-loop, since the store may not be
 ; executed.
 ;

llvm/test/CodeGen/AMDGPU/call-args-inreg-bfloat.ll

@@ -0,0 +1,130 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 4
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx900 < %s | FileCheck -enable-var-scope -check-prefix=GFX9 %s
+; RUN: llc -mtriple=amdgcn-amd-amdhsa -mcpu=gfx1100 < %s | FileCheck -enable-var-scope -check-prefix=GFX11 %s
+
+; We've separated this file from call-args-inreg.ll since GlobalISel does not support the bfloat type.
+; Ideally, we should merge the two files once that support lands.
+
+declare hidden void @external_void_func_bf16_inreg(bfloat inreg) #0
+declare hidden void @external_void_func_v2bf16_inreg(<2 x bfloat> inreg) #0
+
+define void @test_call_external_void_func_bf16_inreg(bfloat inreg %arg) #0 {
+; GFX9-LABEL: test_call_external_void_func_bf16_inreg:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX9-NEXT:    s_mov_b32 s17, s33
+; GFX9-NEXT:    s_mov_b32 s33, s32
+; GFX9-NEXT:    s_or_saveexec_b64 s[18:19], -1
+; GFX9-NEXT:    buffer_store_dword v40, off, s[0:3], s33 ; 4-byte Folded Spill
+; GFX9-NEXT:    s_mov_b64 exec, s[18:19]
+; GFX9-NEXT:    v_writelane_b32 v40, s17, 2
+; GFX9-NEXT:    s_addk_i32 s32, 0x400
+; GFX9-NEXT:    v_writelane_b32 v40, s30, 0
+; GFX9-NEXT:    s_getpc_b64 s[18:19]
+; GFX9-NEXT:    s_add_u32 s18, s18, external_void_func_bf16_inreg@rel32@lo+4
+; GFX9-NEXT:    s_addc_u32 s19, s19, external_void_func_bf16_inreg@rel32@hi+12
+; GFX9-NEXT:    s_mov_b32 s0, s16
+; GFX9-NEXT:    v_writelane_b32 v40, s31, 1
+; GFX9-NEXT:    s_swappc_b64 s[30:31], s[18:19]
+; GFX9-NEXT:    v_readlane_b32 s31, v40, 1
+; GFX9-NEXT:    v_readlane_b32 s30, v40, 0
+; GFX9-NEXT:    s_mov_b32 s32, s33
+; GFX9-NEXT:    v_readlane_b32 s4, v40, 2
+; GFX9-NEXT:    s_or_saveexec_b64 s[6:7], -1
+; GFX9-NEXT:    buffer_load_dword v40, off, s[0:3], s33 ; 4-byte Folded Reload
+; GFX9-NEXT:    s_mov_b64 exec, s[6:7]
+; GFX9-NEXT:    s_mov_b32 s33, s4
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-LABEL: test_call_external_void_func_bf16_inreg:
+; GFX11:       ; %bb.0:
+; GFX11-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s1, s33
+; GFX11-NEXT:    s_mov_b32 s33, s32
+; GFX11-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX11-NEXT:    scratch_store_b32 off, v40, s33 ; 4-byte Folded Spill
+; GFX11-NEXT:    s_mov_b32 exec_lo, s2
+; GFX11-NEXT:    v_writelane_b32 v40, s1, 2
+; GFX11-NEXT:    s_add_i32 s32, s32, 16
+; GFX11-NEXT:    s_getpc_b64 s[2:3]
+; GFX11-NEXT:    s_add_u32 s2, s2, external_void_func_bf16_inreg@rel32@lo+4
+; GFX11-NEXT:    s_addc_u32 s3, s3, external_void_func_bf16_inreg@rel32@hi+12
+; GFX11-NEXT:    v_writelane_b32 v40, s30, 0
+; GFX11-NEXT:    v_writelane_b32 v40, s31, 1
+; GFX11-NEXT:    s_swappc_b64 s[30:31], s[2:3]
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX11-NEXT:    v_readlane_b32 s31, v40, 1
+; GFX11-NEXT:    v_readlane_b32 s30, v40, 0
+; GFX11-NEXT:    s_mov_b32 s32, s33
+; GFX11-NEXT:    v_readlane_b32 s0, v40, 2
+; GFX11-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX11-NEXT:    scratch_load_b32 v40, off, s33 ; 4-byte Folded Reload
+; GFX11-NEXT:    s_mov_b32 exec_lo, s1
+; GFX11-NEXT:    s_mov_b32 s33, s0
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    s_setpc_b64 s[30:31]
+  call void @external_void_func_bf16_inreg(bfloat inreg %arg)
+  ret void
+}
+
+define void @test_call_external_void_func_v2bf16_inreg(<2 x bfloat> inreg %arg) #0 {
+; GFX9-LABEL: test_call_external_void_func_v2bf16_inreg:
+; GFX9:       ; %bb.0:
+; GFX9-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX9-NEXT:    s_mov_b32 s17, s33
+; GFX9-NEXT:    s_mov_b32 s33, s32
+; GFX9-NEXT:    s_or_saveexec_b64 s[18:19], -1
+; GFX9-NEXT:    buffer_store_dword v40, off, s[0:3], s33 ; 4-byte Folded Spill
+; GFX9-NEXT:    s_mov_b64 exec, s[18:19]
+; GFX9-NEXT:    v_writelane_b32 v40, s17, 2
+; GFX9-NEXT:    s_addk_i32 s32, 0x400
+; GFX9-NEXT:    v_writelane_b32 v40, s30, 0
+; GFX9-NEXT:    s_getpc_b64 s[18:19]
+; GFX9-NEXT:    s_add_u32 s18, s18, external_void_func_v2bf16_inreg@rel32@lo+4
+; GFX9-NEXT:    s_addc_u32 s19, s19, external_void_func_v2bf16_inreg@rel32@hi+12
+; GFX9-NEXT:    s_mov_b32 s0, s16
+; GFX9-NEXT:    v_writelane_b32 v40, s31, 1
+; GFX9-NEXT:    s_swappc_b64 s[30:31], s[18:19]
+; GFX9-NEXT:    v_readlane_b32 s31, v40, 1
+; GFX9-NEXT:    v_readlane_b32 s30, v40, 0
+; GFX9-NEXT:    s_mov_b32 s32, s33
+; GFX9-NEXT:    v_readlane_b32 s4, v40, 2
+; GFX9-NEXT:    s_or_saveexec_b64 s[6:7], -1
+; GFX9-NEXT:    buffer_load_dword v40, off, s[0:3], s33 ; 4-byte Folded Reload
+; GFX9-NEXT:    s_mov_b64 exec, s[6:7]
+; GFX9-NEXT:    s_mov_b32 s33, s4
+; GFX9-NEXT:    s_waitcnt vmcnt(0)
+; GFX9-NEXT:    s_setpc_b64 s[30:31]
+;
+; GFX11-LABEL: test_call_external_void_func_v2bf16_inreg:
+; GFX11:       ; %bb.0:
+; GFX11-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s1, s33
+; GFX11-NEXT:    s_mov_b32 s33, s32
+; GFX11-NEXT:    s_or_saveexec_b32 s2, -1
+; GFX11-NEXT:    scratch_store_b32 off, v40, s33 ; 4-byte Folded Spill
+; GFX11-NEXT:    s_mov_b32 exec_lo, s2
+; GFX11-NEXT:    v_writelane_b32 v40, s1, 2
+; GFX11-NEXT:    s_add_i32 s32, s32, 16
+; GFX11-NEXT:    s_getpc_b64 s[2:3]
+; GFX11-NEXT:    s_add_u32 s2, s2, external_void_func_v2bf16_inreg@rel32@lo+4
+; GFX11-NEXT:    s_addc_u32 s3, s3, external_void_func_v2bf16_inreg@rel32@hi+12
+; GFX11-NEXT:    v_writelane_b32 v40, s30, 0
+; GFX11-NEXT:    v_writelane_b32 v40, s31, 1
+; GFX11-NEXT:    s_swappc_b64 s[30:31], s[2:3]
+; GFX11-NEXT:    s_delay_alu instid0(VALU_DEP_1)
+; GFX11-NEXT:    v_readlane_b32 s31, v40, 1
+; GFX11-NEXT:    v_readlane_b32 s30, v40, 0
+; GFX11-NEXT:    s_mov_b32 s32, s33
+; GFX11-NEXT:    v_readlane_b32 s0, v40, 2
+; GFX11-NEXT:    s_or_saveexec_b32 s1, -1
+; GFX11-NEXT:    scratch_load_b32 v40, off, s33 ; 4-byte Folded Reload
+; GFX11-NEXT:    s_mov_b32 exec_lo, s1
+; GFX11-NEXT:    s_mov_b32 s33, s0
+; GFX11-NEXT:    s_waitcnt vmcnt(0)
+; GFX11-NEXT:    s_setpc_b64 s[30:31]
+  call void @external_void_func_v2bf16_inreg(<2 x bfloat> inreg %arg)
+  ret void
+}
+