LTO builds

LTO builds

[Ian Rogers]

Hi,

I just wanted to post on this as I think others may be interested,
perhaps in the future the standard non-"DEBUG=1" build can be LTO. I
played around a bit with LTO builds for the perf tool. clang worked if
I did:
$ make -C tools/perf O=/tmp/perf EXTRA_CFLAGS="-flto"
EXTRA_CXXFLAGS="-flto" CC=clang CXX=clang++ LD=ld.lld

However, gcc didn't compile without the hacky patch at the end, and
then it failed with link time issues:

```
$ make -C tools/perf O=/tmp/perf EXTRA_CFLAGS="-flto"
EXTRA_CXXFLAGS="-flto" CC=gcc CXX=g++ LD=ld.lld
make: Entering directory '~/kernel.org/tools/perf'
  BUILD:   Doing 'make -j16' parallel build
Warning: Kernel ABI header at 'tools/include/uapi/linux/perf_event.h'
differs from latest version at 'include/uapi/linux/perf_event.h'
diff -u tools/include/uapi/linux/perf_event.h include/uapi/linux/perf_event.h
Warning: Kernel ABI header at
'tools/arch/arm64/include/uapi/asm/perf_regs.h' differs from latest
version at 'arch/arm64/include/uapi/asm/perf_regs.h'
diff -u tools/arch/arm64/include/uapi/asm/perf_regs.h
arch/arm64/include/uapi/asm/perf_regs.h
Warning: Kernel ABI header at 'tools/include/linux/coresight-pmu.h'
differs from latest version at 'include/linux/coresight-pmu.h'
diff -u tools/include/linux/coresight-pmu.h include/linux/coresight-pmu.h
Makefile.config:602: No sys/sdt.h found, no SDT events are defined,
please install systemtap-sdt-devel or systemtap-sdt-dev
Makefile.config:985: No libcap found, disables capability support,
please install libcap-devel/libcap-dev
Makefile.config:998: No numa.h found, disables 'perf bench numa mem'
benchmark, please install numactl-devel/libnuma-devel/libnuma-dev
Makefile.config:1057: No libbabeltrace found, disables 'perf data' CTF
format support, please install
libbabeltrace-dev[el]/libbabeltrace-ctf-dev

Auto-detecting system features:
...                                   dwarf: [ on  ]
...                      dwarf_getlocations: [ on  ]
...                                   glibc: [ on  ]
...                                  libbfd: [ on  ]
...                          libbfd-buildid: [ on  ]
...                                  libcap: [ OFF ]
...                                  libelf: [ on  ]
...                                 libnuma: [ OFF ]
...                  numa_num_possible_cpus: [ OFF ]
...                                 libperl: [ on  ]
...                               libpython: [ on  ]
...                               libcrypto: [ on  ]
...                               libunwind: [ on  ]
...                      libdw-dwarf-unwind: [ on  ]
...                                    zlib: [ on  ]
...                                    lzma: [ on  ]
...                               get_cpuid: [ on  ]
...                                     bpf: [ on  ]
...                                  libaio: [ on  ]
...                                 libzstd: [ on  ]

  INSTALL libsubcmd_headers
  INSTALL libapi_headers
  INSTALL libperf_headers
  INSTALL libsymbol_headers
  INSTALL libbpf_headers
  LINK    /tmp/perf/perf
lto-wrapper: warning: using serial compilation of 50 LTRANS jobs
lto-wrapper: note: see the ‘-flto’ option documentation for more information
/usr/bin/ld: /tmp/cclv2Qid.ltrans3.ltrans.o: in function `script_fetch_insn':
~/kernel.org/tools/perf/arch/x86/util/archinsn.c:23: undefined
reference to `insn_decode'
/usr/bin/ld: /tmp/cclv2Qid.ltrans32.ltrans.o: in function
`inline_list__append_record':
~/kernel.org/tools/perf/util/symbol-elf.c:286: undefined reference to
`cxx_demangle_sym'
/usr/bin/ld: /tmp/cclv2Qid.ltrans37.ltrans.o: in function
`dso__synthesize_plt_symbols':
~/kernel.org/tools/perf/util/symbol-elf.c:286: undefined reference to
`cxx_demangle_sym'
/usr/bin/ld: ~/kernel.org/tools/perf/util/symbol-elf.c:286: undefined
reference to `cxx_demangle_sym'
/usr/bin/ld: /tmp/cclv2Qid.ltrans37.ltrans.o: in function `dso__demangle_sym':
~/kernel.org/tools/perf/util/symbol-elf.c:286: undefined reference to
`cxx_demangle_sym'
/usr/bin/ld: /tmp/cclv2Qid.ltrans37.ltrans.o: in function
`dso__load_sym_internal.lto_priv.0':
~/kernel.org/tools/perf/util/symbol-elf.c:286: undefined reference to
`cxx_demangle_sym'
/usr/bin/ld: /tmp/cclv2Qid.ltrans41.ltrans.o: in function `arch_fetch_insn':
~/kernel.org/tools/perf/arch/x86/util/archinsn.c:23: undefined
reference to `insn_decode'
/usr/bin/ld: /tmp/cclv2Qid.ltrans49.ltrans.o: in function
`perf_sample__fprintf_insn.isra.0':
~/kernel.org/tools/perf/arch/x86/util/archinsn.c:23: undefined
reference to `insn_decode'
/usr/bin/ld: /tmp/cclv2Qid.ltrans10.ltrans.o:(.data.rel.ro+0x28):
undefined reference to `memset_orig'
/usr/bin/ld: /tmp/cclv2Qid.ltrans10.ltrans.o:(.data.rel.ro+0x40):
undefined reference to `__memset'
/usr/bin/ld: /tmp/cclv2Qid.ltrans10.ltrans.o:(.data.rel+0x28):
undefined reference to `memcpy_orig'
/usr/bin/ld: /tmp/cclv2Qid.ltrans10.ltrans.o:(.data.rel+0x40):
undefined reference to `__memcpy'
/usr/bin/ld: /tmp/cclv2Qid.ltrans33.ltrans.o: in function `__intel_pt_log_insn':
~/kernel.org/tools/perf/util/intel-pt-decoder/intel-pt-log.c:236:
undefined reference to `intel_pt_insn_desc'
/usr/bin/ld: /tmp/cclv2Qid.ltrans33.ltrans.o: in function
`__intel_pt_log_insn_no_data':
~/kernel.org/tools/perf/util/intel-pt-decoder/intel-pt-log.c:251:
undefined reference to `intel_pt_insn_desc'
/usr/bin/ld: /tmp/cclv2Qid.ltrans34.ltrans.o: in function
`intel_pt_sample_flags':
~/kernel.org/tools/perf/util/intel-pt.c:1503: undefined reference to
`intel_pt_insn_type'
/usr/bin/ld: /tmp/cclv2Qid.ltrans34.ltrans.o: in function
`intel_pt_walk_next_insn':
~/kernel.org/tools/perf/util/intel-pt.c:860: undefined reference to
`intel_pt_get_insn'
/usr/bin/ld: /tmp/cclv2Qid.ltrans35.ltrans.o: in function
`intel_bts_process_queue':
~/kernel.org/tools/perf/util/intel-bts.c:335: undefined reference to
`intel_pt_get_insn'
/usr/bin/ld: ~/kernel.org/tools/perf/util/intel-bts.c:390: undefined
reference to `intel_pt_insn_type'
/usr/bin/ld: /tmp/cclv2Qid.ltrans42.ltrans.o: in function `test_data_item':
~/kernel.org/tools/perf/arch/x86/tests/insn-x86.c:106: undefined
reference to `insn_decode'
/usr/bin/ld: ~/kernel.org/tools/perf/arch/x86/tests/insn-x86.c:122:
undefined reference to `intel_pt_get_insn'
/usr/bin/ld: /tmp/cclv2Qid.ltrans42.ltrans.o: in function
`test__arch_unwind_sample':
~/kernel.org/tools/perf/arch/x86/tests/dwarf-unwind.c:72: undefined
reference to `perf_regs_load'
collect2: error: ld returned 1 exit status
make[2]: *** [Makefile.perf:682: /tmp/perf/perf] Error 1
make[1]: *** [Makefile.perf:236: sub-make] Error 2
make: *** [Makefile:70: all] Error 2
make: Leaving directory '~/kernel.org/tools/perf'
```

A number of the link time issues are with C++ code and assembly code.
I played around with options like AR=gcc-ar but without success. I
wondered if people on the list had thoughts on how to get this to
work.

Thanks,
Ian

Patch:
```
diff --git a/tools/perf/tests/builtin-test.c b/tools/perf/tests/builtin-test.c
index aa44fdc84763..6bf13e052fff 100644
--- a/tools/perf/tests/builtin-test.c
+++ b/tools/perf/tests/builtin-test.c
@@ -33,9 +33,7 @@
static bool dont_fork;
const char *dso_to_test;

-struct test_suite *__weak arch_tests[] = {
-       NULL,
-};
+extern struct test_suite *__weak arch_tests[];

static struct test_suite *generic_tests[] = {
       &suite__vmlinux_matches_kallsyms,
diff --git a/tools/perf/tests/stat.c b/tools/perf/tests/stat.c
index 500974040fe3..706780fb5695 100644
--- a/tools/perf/tests/stat.c
+++ b/tools/perf/tests/stat.c
@@ -27,7 +27,7 @@ static int process_stat_config_event(struct
perf_tool *tool __maybe_unused,
                                    struct machine *machine __maybe_unused)
{
       struct perf_record_stat_config *config = &event->stat_config;
-       struct perf_stat_config stat_config;
+       struct perf_stat_config stat_config = {};

#define HAS(term, val) \
       has_term(config, PERF_STAT_CONFIG_TERM__##term, val)
diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c
index e0c3f2037477..80ac426b7e6a 100644
--- a/tools/perf/util/parse-events.c
+++ b/tools/perf/util/parse-events.c
@@ -754,7 +754,7 @@ parse_events_config_bpf(struct parse_events_state
*parse_state,
                       struct list_head *head_config)
{
       struct parse_events_term *term;
-       int error_pos;
+       int error_pos = 0;

       if (!head_config || list_empty(head_config))
               return 0;
diff --git a/tools/perf/util/stat.c b/tools/perf/util/stat.c
index 0f7b8a8cdea6..cef1d4a41c05 100644
--- a/tools/perf/util/stat.c
+++ b/tools/perf/util/stat.c
@@ -707,7 +707,7 @@ size_t perf_event__fprintf_stat_round(union
perf_event *event, FILE *fp)

size_t perf_event__fprintf_stat_config(union perf_event *event, FILE *fp)
{
-       struct perf_stat_config sc;
+       struct perf_stat_config sc = {};
       size_t ret;

       perf_event__read_stat_config(&sc, &event->stat_config);
```