Messages in this thread | | | Subject | Re: perf build broken in 5.1-rc7 | From | Thomas Backlund <> | Date | Wed, 1 May 2019 17:09:59 +0300 |
| |
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 ?
--
Thomas
| |