Fix alignment of byte arrays inside tuples

CHANGELOG.md

@@ -19,6 +19,8 @@ and this project adheres to
   - [#2631](https://github.com/iovisor/bpftrace/pull/2631)
 - Fix crashes when maps are concurrently modified
   - [#2623](https://github.com/iovisor/bpftrace/pull/2623)
+- Fix alignment of byte arrays inside tuples
+  - [#2625](https://github.com/iovisor/bpftrace/pull/2625)
 
 
 ## [0.18.0] 2023-05-15

src/struct.cpp

@@ -65,7 +65,7 @@ std::unique_ptr<Struct> Struct::CreateTuple(std::vector<SizedType> fields)
 
   for (auto &field : fields)
   {
-    auto align = field.GetAlignment();
+    auto align = field.GetInTupleAlignment();
     struct_align = std::max(align, struct_align);
     auto size = field.GetSize();
 

src/types.cpp

@@ -516,9 +516,9 @@ void SizedType::DumpStructure(std::ostream &os)
   return inner_struct_.lock()->Dump(os);
 }
 
-ssize_t SizedType::GetAlignment() const
+ssize_t SizedType::GetInTupleAlignment() const
 {
-  if (IsStringTy())
+  if (IsByteArray())
     return 1;
 
   if (IsTupleTy() || IsRecordTy())
@@ -527,8 +527,8 @@ ssize_t SizedType::GetAlignment() const
   if (GetSize() <= 2)
     return GetSize();
   else if (IsArrayTy())
-    return element_type_->GetAlignment();
-  else if (IsByteArray() || GetSize() <= 4)
+    return element_type_->GetInTupleAlignment();
+  else if (GetSize() <= 4)
     return 4;
   else
     return 8;

src/types.h

@@ -167,9 +167,9 @@ class SizedType
   std::weak_ptr<const Struct> GetStruct() const;
 
   /**
-     Required alignment for this type
+     Required alignment for this type when used inside a tuple
    */
-  ssize_t GetAlignment() const;
+  ssize_t GetInTupleAlignment() const;
 
   /**
      Dump the underlying structure for debug purposes

tests/codegen/llvm/tuple_bytearray.ll

@@ -0,0 +1,66 @@
+; ModuleID = 'bpftrace'
+source_filename = "bpftrace"
+target datalayout = "e-m:e-p:64:64-i64:64-i128:128-n32:64-S128"
+target triple = "bpf-pc-linux"
+
+%usym_t = type { i64, i64 }
+%"unsigned int8_usym_int64__tuple_t" = type { i8, [16 x i8], i64 }
+
+; Function Attrs: nounwind
+declare i64 @llvm.bpf.pseudo(i64 %0, i64 %1) #0
+
+define i64 @"kprobe:f"(i8* %0) section "s_kprobe:f_1" {
+entry:
+  %"@t_key" = alloca i64, align 8
+  %usym = alloca %usym_t, align 8
+  %tuple = alloca %"unsigned int8_usym_int64__tuple_t", align 8
+  %1 = bitcast %"unsigned int8_usym_int64__tuple_t"* %tuple to i8*
+  call void @llvm.lifetime.start.p0i8(i64 -1, i8* %1)
+  %2 = bitcast %"unsigned int8_usym_int64__tuple_t"* %tuple to i8*
+  call void @llvm.memset.p0i8.i64(i8* align 1 %2, i8 0, i64 32, i1 false)
+  %3 = getelementptr %"unsigned int8_usym_int64__tuple_t", %"unsigned int8_usym_int64__tuple_t"* %tuple, i32 0, i32 0
+  store i8 1, i8* %3, align 1
+  %4 = bitcast i8* %0 to i64*
+  %5 = getelementptr i64, i64* %4, i64 16
+  %reg_ip = load volatile i64, i64* %5, align 8
+  %6 = bitcast %usym_t* %usym to i8*
+  call void @llvm.lifetime.start.p0i8(i64 -1, i8* %6)
+  %get_pid_tgid = call i64 inttoptr (i64 14 to i64 ()*)()
+  %7 = lshr i64 %get_pid_tgid, 32
+  %8 = getelementptr %usym_t, %usym_t* %usym, i64 0, i32 0
+  %9 = getelementptr %usym_t, %usym_t* %usym, i64 0, i32 1
+  store i64 %reg_ip, i64* %8, align 8
+  store i64 %7, i64* %9, align 8
+  %10 = getelementptr %"unsigned int8_usym_int64__tuple_t", %"unsigned int8_usym_int64__tuple_t"* %tuple, i32 0, i32 1
+  %11 = bitcast [16 x i8]* %10 to i8*
+  %12 = bitcast %usym_t* %usym to i8*
+  call void @llvm.memcpy.p0i8.p0i8.i64(i8* align 1 %11, i8* align 1 %12, i64 16, i1 false)
+  %13 = getelementptr %"unsigned int8_usym_int64__tuple_t", %"unsigned int8_usym_int64__tuple_t"* %tuple, i32 0, i32 2
+  store i64 10, i64* %13, align 8
+  %14 = bitcast i64* %"@t_key" to i8*
+  call void @llvm.lifetime.start.p0i8(i64 -1, i8* %14)
+  store i64 0, i64* %"@t_key", align 8
+  %pseudo = call i64 @llvm.bpf.pseudo(i64 1, i64 0)
+  %update_elem = call i64 inttoptr (i64 2 to i64 (i64, i64*, %"unsigned int8_usym_int64__tuple_t"*, i64)*)(i64 %pseudo, i64* %"@t_key", %"unsigned int8_usym_int64__tuple_t"* %tuple, i64 0)
+  %15 = bitcast i64* %"@t_key" to i8*
+  call void @llvm.lifetime.end.p0i8(i64 -1, i8* %15)
+  %16 = bitcast %"unsigned int8_usym_int64__tuple_t"* %tuple to i8*
+  call void @llvm.lifetime.end.p0i8(i64 -1, i8* %16)
+  ret i64 0
+}
+
+; Function Attrs: argmemonly nofree nosync nounwind willreturn
+declare void @llvm.lifetime.start.p0i8(i64 immarg %0, i8* nocapture %1) #1
+
+; Function Attrs: argmemonly nofree nosync nounwind willreturn writeonly
+declare void @llvm.memset.p0i8.i64(i8* nocapture writeonly %0, i8 %1, i64 %2, i1 immarg %3) #2
+
+; Function Attrs: argmemonly nofree nosync nounwind willreturn
+declare void @llvm.memcpy.p0i8.p0i8.i64(i8* noalias nocapture writeonly %0, i8* noalias nocapture readonly %1, i64 %2, i1 immarg %3) #1
+
+; Function Attrs: argmemonly nofree nosync nounwind willreturn
+declare void @llvm.lifetime.end.p0i8(i64 immarg %0, i8* nocapture %1) #1
+
+attributes #0 = { nounwind }
+attributes #1 = { argmemonly nofree nosync nounwind willreturn }
+attributes #2 = { argmemonly nofree nosync nounwind willreturn writeonly }

tests/codegen/tuple_bytearray.cpp

@@ -0,0 +1,16 @@
+#include "common.h"
+
+namespace bpftrace {
+namespace test {
+namespace codegen {
+
+TEST(codegen, tuple_bytearray)
+{
+  test(R"_(k:f { @t = ((uint8)1, usym(reg("ip")), 10); })_",
+
+       NAME);
+}
+
+} // namespace codegen
+} // namespace test
+} // namespace bpftrace

tests/runtime/tuples

@@ -88,3 +88,9 @@ NAME tuple strftime type is packed
 PROG BEGIN { @ = (nsecs, strftime("%M:%S", nsecs)); exit(); }
 EXPECT ^@: \(\d+, \d+:\d+\)$
 TIMEOUT 5
+
+NAME bytearray in tuple
+PROG uprobe:./testprogs/uprobe_test:function1 { @ = ((int8)1, usym(reg("ip")), 10); exit(); }
+EXPECT ^@: \(1, 0x[0-9a-f]+, 10\)$
+TIMEOUT 5
+AFTER ./testprogs/uprobe_test