[llvm] b5cf030 - [AArch64][SME] Fix accessing the emergency spill slot with hazard padding (#142190)

via llvm-commits llvm-commits at lists.llvm.org
Mon Jun 2 02:52:13 PDT 2025


Author: Benjamin Maxwell
Date: 2025-06-02T10:52:10+01:00
New Revision: b5cf03033251a642b184b2e0ea6bdac171c17702

URL: https://github.com/llvm/llvm-project/commit/b5cf03033251a642b184b2e0ea6bdac171c17702
DIFF: https://github.com/llvm/llvm-project/commit/b5cf03033251a642b184b2e0ea6bdac171c17702.diff

LOG: [AArch64][SME] Fix accessing the emergency spill slot with hazard padding (#142190)

This patch fixes an issue where when hazard padding was enabled locals,
including the emergency spill slot, could not be directly addressed.

Generally, this is fine, we can materialize the constant offset in a
scratch register, but if there's no register free we need to spill, and
if we can't even reach the emergency spill slot then we fail to compile.

This patch fixes this by ensuring that if a function has variable-sized
objects and is likely to have hazard padding we enable the base pointer.
Then if we know a function has hazard padding, place the emergency spill
slot next to the BP/SP, to ensure it can be directly accessed without
stepping over any hazard padding.

Added: 
    llvm/test/CodeGen/AArch64/framelayout-scavengingslot-stack-hazard.mir

Modified: 
    llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp
    llvm/test/CodeGen/AArch64/stack-hazard.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp b/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp
index 8caa49de0af43..1afe23e637e8d 100644
--- a/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp
+++ b/llvm/lib/Target/AArch64/AArch64RegisterInfo.cpp
@@ -18,6 +18,7 @@
 #include "AArch64Subtarget.h"
 #include "MCTargetDesc/AArch64AddressingModes.h"
 #include "MCTargetDesc/AArch64InstPrinter.h"
+#include "Utils/AArch64SMEAttributes.h"
 #include "llvm/ADT/BitVector.h"
 #include "llvm/BinaryFormat/Dwarf.h"
 #include "llvm/CodeGen/LiveRegMatrix.h"
@@ -632,14 +633,27 @@ bool AArch64RegisterInfo::hasBasePointer(const MachineFunction &MF) const {
       return true;
 
     auto &ST = MF.getSubtarget<AArch64Subtarget>();
+    const AArch64FunctionInfo *AFI = MF.getInfo<AArch64FunctionInfo>();
     if (ST.hasSVE() || ST.isStreaming()) {
-      const AArch64FunctionInfo *AFI = MF.getInfo<AArch64FunctionInfo>();
       // Frames that have variable sized objects and scalable SVE objects,
       // should always use a basepointer.
       if (!AFI->hasCalculatedStackSizeSVE() || AFI->getStackSizeSVE())
         return true;
     }
 
+    // Frames with hazard padding can have a large offset between the frame
+    // pointer and GPR locals, which includes the emergency spill slot. If the
+    // emergency spill slot is not within range of the load/store instructions
+    // (which have a signed 9-bit range), we will fail to compile if it is used.
+    // Since hasBasePointer() is called before we know if we have hazard padding
+    // or an emergency spill slot we need to enable the basepointer
+    // conservatively.
+    if (AFI->hasStackHazardSlotIndex() ||
+        (ST.getStreamingHazardSize() &&
+         !SMEAttrs(MF.getFunction()).hasNonStreamingInterfaceAndBody())) {
+      return true;
+    }
+
     // Conservatively estimate whether the negative offset from the frame
     // pointer will be sufficient to reach. If a function has a smallish
     // frame, it's less likely to have lots of spills and callee saved
@@ -764,7 +778,8 @@ AArch64RegisterInfo::useFPForScavengingIndex(const MachineFunction &MF) const {
   assert((!MF.getSubtarget<AArch64Subtarget>().hasSVE() ||
           AFI->hasCalculatedStackSizeSVE()) &&
          "Expected SVE area to be calculated by this point");
-  return TFI.hasFP(MF) && !hasStackRealignment(MF) && !AFI->getStackSizeSVE();
+  return TFI.hasFP(MF) && !hasStackRealignment(MF) && !AFI->getStackSizeSVE() &&
+         !AFI->hasStackHazardSlotIndex();
 }
 
 bool AArch64RegisterInfo::requiresFrameIndexScavenging(

diff  --git a/llvm/test/CodeGen/AArch64/framelayout-scavengingslot-stack-hazard.mir b/llvm/test/CodeGen/AArch64/framelayout-scavengingslot-stack-hazard.mir
new file mode 100644
index 0000000000000..52ac36f801854
--- /dev/null
+++ b/llvm/test/CodeGen/AArch64/framelayout-scavengingslot-stack-hazard.mir
@@ -0,0 +1,99 @@
+# RUN: llc -mtriple=aarch64-none-linux-gnu -aarch64-stack-hazard-size=1024 -run-pass=prologepilog %s -o - | FileCheck %s
+--- |
+
+  define void @stack_hazard_streaming_compat() "aarch64_pstate_sm_compatible" { entry: unreachable }
+  define void @stack_hazard_streaming_compat_emergency_spill_slot() "aarch64_pstate_sm_compatible" { entry: unreachable }
+
+...
+
+# +------------------+
+# | GPR callee-saves |
+# +------------------+ <- FP
+# | <hazard padding> |
+# +------------------+
+# | FPR locals       |
+# | %stack.1         |
+# +------------------+
+# | <hazard padding> |
+# +------------------+
+# | GPR locals       |
+# | %stack.2         |
+# | <emergency spill>|
+# +------------------+ <- BP
+# | <VLA>            |
+# +------------------+ <- SP (can't be used due to VLA)
+
+# In this case without the base pointer we'd need the emergency spill slot to
+# access both %stack.1 and %stack.2. With the base pointer we can reach both
+# without spilling.
+
+name: stack_hazard_streaming_compat
+# CHECK-LABEL: name: stack_hazard_streaming_compat
+# CHECK: bb.0:
+# CHECK:      STRDui $d0, $x19, 131
+# CHECK-NEXT: STRXui $x0, $x19, 1
+# CHECK: bb.1:
+tracksRegLiveness: true
+frameInfo:
+  isFrameAddressTaken: true
+stack:
+  - { id: 0, type: variable-sized,  alignment: 1 }
+  - { id: 1, size: 8, alignment: 8 }
+  - { id: 2, size: 8, alignment: 8 }
+body: |
+  bb.0:
+    liveins: $x0, $x8, $d0
+    $x9 = LDRXui $x0, 0 :: (load (s64))
+    STRDui $d0, %stack.1, 0 :: (store (s64) into %stack.1)
+    STRXui $x0, %stack.2, 0 :: (store (s64) into %stack.2)
+    B %bb.1
+  bb.1:
+    liveins: $x1, $x2, $x3, $x4, $x5, $x6, $x7, $x8, $x9, $x10, $x11, $x12, $x13, $x14, $x15, $x16, $x17, $x18, $x19, $x20, $x21, $x22, $x23, $x24, $x25, $x26, $x27, $x28, $lr
+    RET_ReallyLR implicit $x19, implicit $x20, implicit $x21, implicit $x22, implicit $x23, implicit $x24, implicit $x25, implicit $x26, implicit $x27, implicit $x28, implicit $lr
+...
+---
+# +------------------+
+# | GPR callee-saves |
+# +------------------+ <- FP
+# | <hazard padding> |
+# +------------------+
+# | FPR locals       |
+# | %stack.1         |
+# +------------------+
+# | <hazard padding> |
+# +------------------+
+# | GPR locals       |
+# | %stack.2         | (very large)
+# | <emergency spill>|
+# +------------------+ <- BP
+# | <VLA>            |
+# +------------------+ <- SP (can't be used due to VLA)
+
+# In this case we need to use the emergency spill slot to access %stack.1 as it
+# is too far from the frame pointer and the base pointer to directly address.
+# Note: This also tests that the <emergency spill> located near the SP/BP.
+
+name: stack_hazard_streaming_compat_emergency_spill_slot
+# CHECK-LABEL: name: stack_hazard_streaming_compat_emergency_spill_slot
+# CHECK: bb.0:
+# CHECK:      STRXui killed $[[SCRATCH:x[0-9]+]], $x19, 0
+# CHECK-NEXT: $[[SCRATCH]] = ADDXri $x19, 1056, 0
+# CHECK-NEXT: STRDui $d0, killed $[[SCRATCH]], 4095
+# CHECK-NEXT: $[[SCRATCH]] = LDRXui $x19, 0
+# CHECK: bb.1:
+tracksRegLiveness: true
+frameInfo:
+  isFrameAddressTaken: true
+stack:
+  - { id: 0, type: variable-sized,  alignment: 1 }
+  - { id: 1, size: 8, alignment: 8 }
+  - { id: 2, size: 32761, alignment: 8 }
+body: |
+  bb.0:
+    liveins: $x0, $x8, $d0
+    $x9 = LDRXui $x0, 0 :: (load (s64))
+    STRDui $d0, %stack.1, 0 :: (store (s64) into %stack.1)
+    B %bb.1
+  bb.1:
+    liveins: $x0, $x1, $x2, $x3, $x4, $x5, $x6, $x7, $x8, $x9, $x10, $x11, $x12, $x13, $x14, $x15, $x16, $x17, $x18, $x19, $x20, $x21, $x22, $x23, $x24, $x25, $x26, $x27, $x28, $lr
+    RET_ReallyLR implicit $x19, implicit $x20, implicit $x21, implicit $x22, implicit $x23, implicit $x24, implicit $x25, implicit $x26, implicit $x27, implicit $x28, implicit $lr

diff  --git a/llvm/test/CodeGen/AArch64/stack-hazard.ll b/llvm/test/CodeGen/AArch64/stack-hazard.ll
index 791d7580c327d..e169b199733bd 100644
--- a/llvm/test/CodeGen/AArch64/stack-hazard.ll
+++ b/llvm/test/CodeGen/AArch64/stack-hazard.ll
@@ -2905,12 +2905,13 @@ define i32 @vastate(i32 %x) "aarch64_inout_za" "aarch64_pstate_sm_enabled" "targ
 ; CHECK64-NEXT:    mov x9, sp
 ; CHECK64-NEXT:    mov w20, w0
 ; CHECK64-NEXT:    msub x9, x8, x8, x9
+; CHECK64-NEXT:    mov x19, sp
 ; CHECK64-NEXT:    mov sp, x9
-; CHECK64-NEXT:    stur x9, [x29, #-208]
-; CHECK64-NEXT:    sub x9, x29, #208
-; CHECK64-NEXT:    sturh wzr, [x29, #-198]
-; CHECK64-NEXT:    stur wzr, [x29, #-196]
-; CHECK64-NEXT:    sturh w8, [x29, #-200]
+; CHECK64-NEXT:    str x9, [x19]
+; CHECK64-NEXT:    add x9, x19, #0
+; CHECK64-NEXT:    strh wzr, [x19, #10]
+; CHECK64-NEXT:    str wzr, [x19, #12]
+; CHECK64-NEXT:    strh w8, [x19, #8]
 ; CHECK64-NEXT:    msr TPIDR2_EL0, x9
 ; CHECK64-NEXT:    .cfi_offset vg, -32
 ; CHECK64-NEXT:    smstop sm
@@ -2919,7 +2920,7 @@ define i32 @vastate(i32 %x) "aarch64_inout_za" "aarch64_pstate_sm_enabled" "targ
 ; CHECK64-NEXT:    .cfi_restore vg
 ; CHECK64-NEXT:    smstart za
 ; CHECK64-NEXT:    mrs x8, TPIDR2_EL0
-; CHECK64-NEXT:    sub x0, x29, #208
+; CHECK64-NEXT:    add x0, x19, #0
 ; CHECK64-NEXT:    cbnz x8, .LBB33_2
 ; CHECK64-NEXT:  // %bb.1: // %entry
 ; CHECK64-NEXT:    bl __arm_tpidr2_restore
@@ -2985,16 +2986,13 @@ define i32 @vastate(i32 %x) "aarch64_inout_za" "aarch64_pstate_sm_enabled" "targ
 ; CHECK1024-NEXT:    mov x9, sp
 ; CHECK1024-NEXT:    mov w20, w0
 ; CHECK1024-NEXT:    msub x9, x8, x8, x9
+; CHECK1024-NEXT:    mov x19, sp
 ; CHECK1024-NEXT:    mov sp, x9
-; CHECK1024-NEXT:    sub x10, x29, #1872
-; CHECK1024-NEXT:    stur x9, [x10, #-256]
-; CHECK1024-NEXT:    sub x9, x29, #1862
-; CHECK1024-NEXT:    sub x10, x29, #1860
-; CHECK1024-NEXT:    sturh wzr, [x9, #-256]
-; CHECK1024-NEXT:    sub x9, x29, #2128
-; CHECK1024-NEXT:    stur wzr, [x10, #-256]
-; CHECK1024-NEXT:    sub x10, x29, #1864
-; CHECK1024-NEXT:    sturh w8, [x10, #-256]
+; CHECK1024-NEXT:    str x9, [x19]
+; CHECK1024-NEXT:    add x9, x19, #0
+; CHECK1024-NEXT:    strh wzr, [x19, #10]
+; CHECK1024-NEXT:    str wzr, [x19, #12]
+; CHECK1024-NEXT:    strh w8, [x19, #8]
 ; CHECK1024-NEXT:    msr TPIDR2_EL0, x9
 ; CHECK1024-NEXT:    .cfi_offset vg, -32
 ; CHECK1024-NEXT:    smstop sm
@@ -3003,7 +3001,7 @@ define i32 @vastate(i32 %x) "aarch64_inout_za" "aarch64_pstate_sm_enabled" "targ
 ; CHECK1024-NEXT:    .cfi_restore vg
 ; CHECK1024-NEXT:    smstart za
 ; CHECK1024-NEXT:    mrs x8, TPIDR2_EL0
-; CHECK1024-NEXT:    sub x0, x29, #2128
+; CHECK1024-NEXT:    add x0, x19, #0
 ; CHECK1024-NEXT:    cbnz x8, .LBB33_2
 ; CHECK1024-NEXT:  // %bb.1: // %entry
 ; CHECK1024-NEXT:    bl __arm_tpidr2_restore


        


More information about the llvm-commits mailing list