Re: perf build broken in 5.1-rc7

From: Thomas Backlund
Date: Wed May 01 2019 - 10:10:24 EST

Next message: Dan Murphy: "Re: [PATCH v11 1/5] can: m_can: Create a m_can platform framework"
Previous message: Jagan Teki: "Re: [DO NOT MERGE] [PATCH 2/2] arm64: rockchip: rk3399: nanopc-t4: Enable FriendlyELEC HD702E eDP panel"
In reply to: Arnaldo Carvalho de Melo: "Re: perf build broken in 5.1-rc7"
Next in thread: Thomas Backlund: "Re: perf build broken in 5.1-rc7"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

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@xxxxxxxxxx> 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@xxxxxx>
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 ?

--

Thomas

Next message: Dan Murphy: "Re: [PATCH v11 1/5] can: m_can: Create a m_can platform framework"
Previous message: Jagan Teki: "Re: [DO NOT MERGE] [PATCH 2/2] arm64: rockchip: rk3399: nanopc-t4: Enable FriendlyELEC HD702E eDP panel"
In reply to: Arnaldo Carvalho de Melo: "Re: perf build broken in 5.1-rc7"
Next in thread: Thomas Backlund: "Re: perf build broken in 5.1-rc7"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]