File: symbolize-split-dwarf-no-skel-address.test

package info (click to toggle)
llvm-toolchain-20 1%3A20.1.8-1
  • links: PTS, VCS
  • area: main
  • in suites: experimental
  • size: 2,111,696 kB
  • sloc: cpp: 7,438,781; ansic: 1,393,871; asm: 1,012,926; python: 241,771; f90: 86,635; objc: 75,411; lisp: 42,144; pascal: 17,286; sh: 8,596; ml: 5,082; perl: 4,730; makefile: 3,591; awk: 3,523; javascript: 2,251; xml: 892; fortran: 672
file content (22 lines) | stat: -rw-r--r-- 877 bytes parent folder | download | duplicates (32) 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
 
Copy the DWO and then cd into the output directory, so that the symbolizer would
find the DWO if it attempted to.
RUN: rm -rf %t && mkdir -p %t
RUN: cd %t
RUN: cp %p/Inputs/split-dwarf-no-skel-address.dwo %t
RUN: llvm-symbolizer --inlining --no-demangle \
RUN:    --obj=%p/Inputs/split-dwarf-no-skel-address.o 0x4 | FileCheck %s

Built from the following source:
void f1();
__attribute__((always_inline)) inline void f2() { f1(); }
void f3() { f2(); }

Compiled to assembly with clang, modified the skeleton CU to remove the
high/low pc (& update the CU length field and abbrev to match) & then
compile/objcopy to make the .o and .dwo.

Ensure that the f2 inlined frame is not included - it's inefficient to have to
go and load all the debug info and search for the address ranges, so assume
that a lack of ranges on the CU means the CU covers no addresses.

CHECK-NOT: _Z2f2v