[lldb][test] Add test for no_unique_address when mixed with bitfields #108155
Conversation
This is the root-cause for the LLDB failures that started occurring after llvm#105865. The DWARFASTParserClang has logic to try derive unnamed bitfields from DWARF offsets. In this case we treat `padding` as a 1-byte size field that would overlap with `flag`, and decide we need to introduce an unnamed bitfield into the AST, which is incorrect.
|
@llvm/pr-subscribers-lldb Author: Michael Buch (Michael137) ChangesThis is the root-cause for the LLDB failures that started occurring after #105865. The DWARFASTParserClang has logic to try derive unnamed bitfields from DWARF offsets. In this case we treat Full diff: https://github.com/llvm/llvm-project/pull/108155.diff 1 Files Affected:
diff --git a/lldb/test/Shell/SymbolFile/DWARF/no_unique_address-with-bitfields.cpp b/lldb/test/Shell/SymbolFile/DWARF/no_unique_address-with-bitfields.cpp
new file mode 100644
index 00000000000000..950bd9585baa22
--- /dev/null
+++ b/lldb/test/Shell/SymbolFile/DWARF/no_unique_address-with-bitfields.cpp
@@ -0,0 +1,30 @@
+// LLDB currently erroneously adds an unnamed bitfield
+// into the AST when a overlapping no_unique_address
+// field precedes a bitfield.
+
+// XFAIL: *
+
+// RUN: %clangxx_host -gdwarf -o %t %s
+// RUN: %lldb %t \
+// RUN: -o "target var global" \
+// RUN: -o "image dump ast" \
+// RUN: -o exit | FileCheck %s
+
+// CHECK: (lldb) image dump ast
+// CHECK-NEXT: CXXRecordDecl {{.*}} struct Foo definition
+// CHECK: |-FieldDecl {{.*}} data 'char[5]'
+// CHECK-NEXT: |-FieldDecl {{.*}} padding 'Empty'
+// CHECK-NEXT: |-FieldDecl {{.*}} 'int'
+// CHECK-NEXT: | `-IntegerLiteral {{.*}} 'int' 8
+// CHECK-NEXT: `-FieldDecl {{.*}} sloc> mem 'unsigned long'
+// CHECK-NEXT: `-IntegerLiteral {{.*}} 'int' 1
+
+struct Empty {};
+
+struct Foo {
+ char data[5];
+ [[no_unique_address]] Empty padding;
+ unsigned long flag : 1;
+};
+
+Foo global;
|
There was a problem hiding this comment.
The test will fail on windows because no_unique_address is not a thing there:
$ bin/clang++ -c --target=x86_64-pc-windows -gdwarf -o /tmp/a.out /tmp/a.cc
/tmp/a.cc:23:5: warning: unknown attribute 'no_unique_address' ignored [-Wunknown-attributes]
23 | [[no_unique_address]] Empty padding;
| ^~~~~~~~~~~~~~~~~
I suggest hardcoding the test with your favourite triple.
Seems like that we should spell it as |
Don't think we have coverage for the |
…llvm#108155) This is the root-cause for the LLDB failures that started occurring after llvm#105865. The DWARFASTParserClang has logic to try derive unnamed bitfields from DWARF offsets. In this case we treat `padding` as a 1-byte size field that would overlap with `flag`, and decide we need to introduce an unnamed bitfield into the AST, which is incorrect.
… into helper (llvm#108196) This logic will need adjusting soon for llvm#108155 This patch pulls out the logic for detecting/creating unnamed bitfields out of `ParseSingleMember` to make the latter (in my opinion) more readable. Otherwise we have a large number of similarly named variables in scope.
This is the root-cause for the LLDB failures that started occurring after #105865.
The DWARFASTParserClang has logic to try derive unnamed bitfields from DWARF offsets. In this case we treat
paddingas a 1-byte size field that would overlap withflag, and decide we need to introduce an unnamed bitfield into the AST, which is incorrect.