Re: perf build broken in 5.1-rc7

[Thomas Backlund <tmb@mageia.org>]

Den 01-05-2019 kl. 17:09, skrev Thomas Backlund:
>
> Den 01-05-2019 kl. 16:07, skrev Arnaldo Carvalho de Melo:
>> Em Tue, Apr 30, 2019 at 04:31:14PM +0300, Thomas Backlund escreveu:
>>> Den 30-04-2019 kl. 16:06, skrev Song Liu:
>>>> On Tue, Apr 30, 2019 at 12:55 AM Thomas Backlund <tmb@mageia.org> 
>>>> wrote:
>>>>> Den 30-04-2019 kl. 10:26, skrev Thomas Backlund:
>>>>>> Building perf in 5.1-rc5/6/7 fails:
>>>>>> Build start:
>>>>>>     make -s -C tools/perf NO_PERF_READ_VDSO32=1 
>>>>>> NO_PERF_READ_VDSOX32=1
>>>>>> WERROR=0 NO_LIBUNWIND=1 HAVE_CPLUS_DEMANGLE=1 NO_GTK2=1 NO_STRLCPY=1
>>>>>> NO_BIONIC=1 NO_JVMTI=1 prefix=/usr lib=lib64 all
>>>>>>      BUILD:   Doing 'make -j32' parallel build
>>>>>>      HOSTCC   fixdep.o
>>>>>>      HOSTLD   fixdep-in.o
>>>>>>      LINK     fixdep
>>>>>> Warning: Kernel ABI header at 
>>>>>> 'tools/arch/x86/include/uapi/asm/vmx.h'
>>>>>> differs from latest version at 'arch/x86/include/uapi/asm/vmx.h'
>>>>>> diff -u tools/arch/x86/include/uapi/asm/vmx.h
>>>>>> arch/x86/include/uapi/asm/vmx.h
>>>>>>
>>>>>> Auto-detecting system features:
>>>>>> ...                         dwarf: [ on  ]
>>>>>> ...            dwarf_getlocations: [ on  ]
>>>>>> ...                         glibc: [ on  ]
>>>>>> ...                          gtk2: [ on  ]
>>>>>> ...                      libaudit: [ on  ]
>>>>>> ...                        libbfd: [ on  ]
>>>>>> ...                        libelf: [ on  ]
>>>>>> ...                       libnuma: [ on  ]
>>>>>> ...        numa_num_possible_cpus: [ on  ]
>>>>>> ...                       libperl: [ on  ]
>>>>>> ...                     libpython: [ on  ]
>>>>>> ...                      libslang: [ on  ]
>>>>>> ...                     libcrypto: [ on  ]
>>>>>> ...                     libunwind: [ on  ]
>>>>>> ...            libdw-dwarf-unwind: [ on  ]
>>>>>> ...                          zlib: [ on  ]
>>>>>> ...                          lzma: [ on  ]
>>>>>> ...                     get_cpuid: [ on  ]
>>>>>> ...                           bpf: [ on  ]
>>>>>> ...                        libaio: [ on  ]
>>>>>> ...        disassembler-four-args: [ OFF ]
>>>>>>
>>>>>> Makefile.config:473: No sys/sdt.h found, no SDT events are defined,
>>>>>> please install systemtap-sdt-devel or systemtap-sdt-dev
>>>>>> Makefile.config:853: No libbabeltrace found, disables 'perf data' 
>>>>>> CTF
>>>>>> format support, please install 
>>>>>> libbabeltrace-dev[el]/libbabeltrace-ctf-dev
>>>>>>
>>>>>>
>>>>>> And breaks with:
>>>>>>
>>>>>>
>>>>>> CC       ui/setup.o
>>>>>> util/annotate.c: In function 'symbol__disassemble_bpf':
>>>>>> util/annotate.c:1767:29: error: incompatible type for argument 1 of
>>>>>> 'disassembler'
>>>>>>      disassemble = disassembler(bfdf);
>>>>>>                                 ^~~~
>>>>>> In file included from util/annotate.c:1689:
>>>>>> /usr/include/dis-asm.h:325:63: note: expected 'enum 
>>>>>> bfd_architecture'
>>>>>> but argument is of type 'bfd *' {aka 'struct bfd *'}
>>>>>>     extern disassembler_ftype disassembler (enum bfd_architecture 
>>>>>> arc,
>>>>>> ~~~~~~~~~~~~~~~~~~~~~~^~~
>>>>>> util/annotate.c:1767:16: error: too few arguments to function
>>>>>> 'disassembler'
>>>>>>      disassemble = disassembler(bfdf);
>>>>>>                    ^~~~~~~~~~~~
>>>>>> In file included from util/annotate.c:1689:
>>>>>> /usr/include/dis-asm.h:325:27: note: declared here
>>>>>>     extern disassembler_ftype disassembler (enum bfd_architecture 
>>>>>> arc,
>>>>>>                               ^~~~~~~~~~~~
>>>>>>      CC       arch/x86/util/header.o
>>>>>>      CC       arch/x86/util/tsc.o
>>>>>>      CC       arch/x86/util/pmu.o
>>>>>> mv: cannot stat 'util/.annotate.o.tmp': No such file or directory
>>>>>>      CC       bench/futex-requeue.o
>>>>>>      CC       arch/x86/util/kvm-stat.o
>>>>>> make[4]: ***
>>>>>> [/work/rpmbuild/BUILD/kernel-x86_64/linux-5.0/tools/build/Makefile.build:97: 
>>>>>>
>>>>>> util/annotate.o] Error 1
>>>>>> make[4]: *** Waiting for unfinished jobs....
>>>>>>      CC       util/build-id.o
>>>>>>
>>>>>>
>>>>>>
>>>>> And I forgot...
>>>>>
>>>>> Reverting:
>>>>>    From 6987561c9e86eace45f2dbb0c564964a63f4150a Mon Sep 17 
>>>>> 00:00:00 2001
>>>>> From: Song Liu <songliubraving@fb.com>
>>>>> Date: Mon, 11 Mar 2019 22:30:48 -0700
>>>>> Subject: perf annotate: Enable annotation of BPF programs
>>>>>
>>>>> Makes it build again.
>>>>>
>>>>> -- 
>>>>> Thomas
>>>>>
>>>> Hi Thomas,
>>>>
>>>> Which system are you running this test on? I would like to repro it 
>>>> in a VM.
>>>>
>>>> Thanks,
>>>> Song
>>>
>>> Mageia Cauldron currently stabilizing to become Mageia 7 in ~1 month.
>>>
>>>
>>> Basesystem is:
>>>
>>> binutils-2.32-5.mga7
>>> (includes all fixes from upstream binutils-2_32-branch)
>>>
>>> gcc-8.3.1-0.20190419.2.mga7
>>>
>>> glibc-2.29-7.mga7
>>> (includes all fixes from upstream glibc release/2.29/master branch 
>>> up to
>>> 2019-04-15 for now)
>>>
>>>
>>> kernel-desktop-5.1.0-0.rc7.1.mga7
>>> kernel-userspace-headers-5.1.0-0.rc7.1.mga7
>> Ok, so the steps are:
>>
>> 1) the feature test, the small C program that we try to build is:
>>
>> [acme@quaco perf]$ cat tools/build/feature/test-disassembler-four-args.c
>> // SPDX-License-Identifier: GPL-2.0
>> #include <bfd.h>
>> #include <dis-asm.h>
>>
>> int main(void)
>> {
>>     bfd *abfd = bfd_openr(NULL, NULL);
>>
>>     disassembler(bfd_get_arch(abfd),
>>              bfd_big_endian(abfd),
>>              bfd_get_mach(abfd),
>>              abfd);
>>
>>     return 0;
>> }
>> [acme@quaco perf]$
>>
>> And here in my fedora29 system it ends up producing the following file,
>> when built with:
>>
>> $ make O=/tmp/build/perf  -C tools/perf install-bin
>>
>> [acme@quaco perf]$ cat 
>> /tmp/build/perf/feature/test-disassembler-four-args.make.output
>> [acme@quaco perf]$
>> [acme@quaco perf]$ file 
>> /tmp/build/perf/feature/test-disassembler-four-args.bin
>> /tmp/build/perf/feature/test-disassembler-four-args.bin: ELF 64-bit 
>> LSB executable, x86-64, version 1 (SYSV), dynamically linked, 
>> interpreter /lib64/ld-linux-x86-64.so.2, for GNU/Linux 3.2.0, 
>> BuildID[sha1]=c9bd83db766a620c5cb6d756b0cd6991527641ff, not stripped, 
>> too many notes (256)
>> [acme@quaco perf]$ ldd 
>> /tmp/build/perf/feature/test-disassembler-four-args.bin
>>     linux-vdso.so.1 (0x00007ffebc5f9000)
>>     libz.so.1 => /lib64/libz.so.1 (0x00007fed2da04000)
>>     libdl.so.2 => /lib64/libdl.so.2 (0x00007fed2d9fe000)
>>     libc.so.6 => /lib64/libc.so.6 (0x00007fed2d838000)
>>     /lib64/ld-linux-x86-64.so.2 (0x00007fed2da3e000)
>> [acme@quaco perf]$
>>
>> Meaning it built properly, so in this sytem the disassembler() function
>> has indeed four args, so we end up with:
>>
>> [acme@quaco perf]$ grep disassembler /tmp/build/perf/FEATURE-DUMP
>> feature-disassembler-four-args=1
>> [acme@quaco perf]$
>>
>> Can you check the output for
>> /tmp/build/perf/feature/test-disassembler-four-args.make.output in your
>> system? And also check what is the prototype for the disassembler()
>> routine on mageia7?
>
>
> I guess this is what fails the test:
>
> cat /tmp/build/perf/feature/test-disassembler-four-args.make.output
> /usr/bin/ld: /usr/lib64/libbfd.a(plugin.o): in function 
> `try_load_plugin':
> /home/iurt/rpmbuild/BUILD/binutils-2.32/objs/bfd/../../bfd/plugin.c:243: 
> undefined reference to `dlopen'
> /usr/bin/ld: 
> /home/iurt/rpmbuild/BUILD/binutils-2.32/objs/bfd/../../bfd/plugin.c:271: 
> undefined reference to `dlsym'
> /usr/bin/ld: 
> /home/iurt/rpmbuild/BUILD/binutils-2.32/objs/bfd/../../bfd/plugin.c:256: 
> undefined reference to `dlclose'
> /usr/bin/ld: 
> /home/iurt/rpmbuild/BUILD/binutils-2.32/objs/bfd/../../bfd/plugin.c:246: 
> undefined reference to `dlerror'
>
>
> as we allow dynamic linking and loading
>
> And we use linker flags:
>
> rpm --eval %ldflags
>  -Wl,--as-needed -Wl,--no-undefined -Wl,-z,relro -Wl,-O1 
> -Wl,--build-id -Wl,--enable-new-dtags
>
> So it reports:
>
>  disassembler-four-args: [ OFF ]
>
> And falls back to one single command.
>
>> Here I have:
>>
>> [acme@quaco perf]$ rpm -q binutils
>> binutils-2.31.1-25.fc29.x86_64
>> [acme@quaco perf]$
>>
>> Perhaps binutils 2.32 changed that prototype again and instead of
>> falling back to using just one arg we need to use some other number of
>> args, or even a different type for the N args it now maybe have?
>>
>> - Arnaldo
>
>
> There is no change in 2.32
>
>
> disassembler_ftype
> disassembler (enum bfd_architecture a,
>               bfd_boolean big ATTRIBUTE_UNUSED,
>               unsigned long mach ATTRIBUTE_UNUSED,
>               bfd *abfd ATTRIBUTE_UNUSED)
>
>
>
> Is there a way to force it to "detect" / use 4 args ?
>
>

And just to confirm, disabling the test with:

--- linux/tools/perf/Makefile.config.orig    2019-04-29 
11:45:54.168627795 +0300
+++ linux/tools/perf/Makefile.config    2019-05-01 17:18:25.328312186 +0300
@@ -813,9 +813,9 @@ ifdef HAVE_KVM_STAT_SUPPORT
      CFLAGS += -DHAVE_KVM_STAT_SUPPORT
  endif

-ifeq ($(feature-disassembler-four-args), 1)
+#ifeq ($(feature-disassembler-four-args), 1)
      CFLAGS += -DDISASM_FOUR_ARGS_SIGNATURE
-endif
+#endif


makes the perf build work again.


--

Thomas