[Linuxppc-users] Toolchain version for RHEL 7.1
Ben Hood
ben at relops.com
Thu Nov 15 00:25:38 AEDT 2018
I’ve now managed to get it to work by adding -L/opt/at12.0/ppc64le/lib64 to LDFLAGS and recompiling the binary. Because I’ve experimented with so many things, I can’t be 100% sure this was the true cause of the problem, so I’m going to try to reproduce the behavior I was seeing beforehand.
> On 13 Nov 2018, at 18:34, Ulrich Weigand <Ulrich.Weigand at de.ibm.com> wrote:
>
> OK, that's good so far. I still don't understand why you're seeing the error then. A number of additional questions:
>
> 1) I see weird names in some of logs below:
> /lib64/ld64.30.2
> libm.30.6
> libc.30.6
> Note the number "30" in place of the string "so" (shared object). Is this actually real, or was this just somehow corrupted in the output?
>
> 2) When executing the "erlc" binary, do you have either LD_PRELOAD or LD_LIBRARY_PATH (or any LD_... variable, really) set to anything? If so, can you try with those variables unset?
>
> 3) Can you send me the output (stderr) of running "erlc" with LD_DEBUG=all ?
>
>
> Mit freundlichen Gruessen / Best Regards
>
> Ulrich Weigand
>
> --
> Dr. Ulrich Weigand | Phone: +49-7031/16-3727
> STSM, GNU/Linux compilers and toolchain
> IBM Deutschland Research & Development GmbH
> Vorsitzende des Aufsichtsrats: Martina Koederitz | Geschäftsführung: Dirk Wittkopp
> Sitz der Gesellschaft: Böblingen | Registergericht: Amtsgericht Stuttgart, HRB 243294
>
> <graycol.gif>Ben Hood ---13.11.2018 18:50:13---Here is the output - it appears to have the line you expected: $ readelf -1 /tmp/c4/erts-10.1.1/bin/
>
> From: Ben Hood <ben at relops.com>
> To: linuxppc-users at lists.ozlabs.org
> Date: 13.11.2018 18:50
> Subject: Re: [Linuxppc-users] Toolchain version for RHEL 7.1
> Sent by: "Linuxppc-users" <linuxppc-users-bounces+ulrich.weigand=de.ibm.com at lists.ozlabs.org>
>
>
>
>
> Here is the output - it appears to have the line you expected:
>
> $ readelf -1 /tmp/c4/erts-10.1.1/bin/erlc
> Elf file type is EXEC (Executable file)
> Entry point OxlOOOOecO
> fhere are 9 program headers, starting at offset 64
> Program Headers:
> Type Offset VirtAddr PhysAddr
> FileSiz MemSiz Flags Align
> PHDR 0x0000000000000040 0x0000000010000040 0x0000000010000040
> OxOOOOOOOOOOOOOlf8 OxOOOOOOOOOOOOOlf8 R 8
> INTERP 0x0000000000000238 0x0000000010000238 0x0000000010000238
> 0x000000000000001c 0x000000000000001c R 1
> [Requesting program interpreter: /opt/atl2.0/lib64/ld64.so.2]
> LOAD 0x0000000000000000 0x0000000010000000 0x0000000010000000
> 0x0000000000008134 0x0000000000008134 R E 10000
> LOAD OxOOOOOOOOOOOOfcbO OxOOOOOOOOlOOlfcbO OxOOOOOOOOlOOlfcbO
> 0x00000000000004b9 0x0000000000000518 RW 10000
> DYNAMIC OxOOOOOOOOOOOOfccO OxOOOOOOOOlOOlfccO OxOOOOOOOOlOOlfccO
> 0x0000000000000240 0x0000000000000240 RW 8
> NOTE 0x0000000000000254 0x0000000010000254 0x0000000010000254
> 0x0000000000000044 0x0000000000000044 R 4
> GNU_EH_FRAME 0x0000000000007838 0x0000000010007838 0x0000000010007838
> 0x00000000000001c4 0x00000000000001c4 R 4
> GNU_STACK 0x0000000000000000 0x0000000000000000 0x0000000000000000
> 0x0000000000000000 0x0000000000000000 RW 10
> GNU_RELRO OxOOOOOOOOOOOOfcbO OxOOOOOOOOlOOlfcbO OxOOOOOOOOlOOlfcbO
> 0x0000000000000350 0x0000000000000350 R 1
> Section to Segment mapping:
> Segment Sections...
> 00
> 01 .interp
> 02 .interp .note.ABI-tag .note.gnu.build-id .hash .gnu.hash .dynsym .dynstr .gnu.version .gnu.version_r .rela.dyn .rela.plt . init .text .fini .rodata .eh_frame_hdr .eh_frame
> 03 .init_array .fini_array .dynamic .got .pit .data .bss
> 04 .dynamic
> 05 .note.ABI-tag .note.gnu.build-id
> 06 .eh_frame_hdr
> 07
> 08 .init_array .fini_array .dynamic .got
>
>
> When I compiled this, I supplied this LD arg:
>
> LD="/opt/at12.0/bin/powerpc64le-linux-gnu-ld"
>
> --
> Ben Hood
> Technical Director | RelOps Ltd
> +44 782 4631586 | ben at relops.com | http://relops.com <http://relops.com/>
> > On 13 Nov 2018, at 17:34, Ulrich Weigand <Ulrich.Weigand at de.ibm.com> wrote:
> >
> > Aha, the loader is actually complaining about missing symbols in *the loader itself*, not in libc.
> >
> > This seems strange:
> > /opt/atl2.0/lib64/ld64.so.2 => /Iib64/ld64.so.2
> >
> > /opt/atl12.0/lib64/ld64.so.2 should be the dynamic loader provided by the AT, it should definitely not fall back to using the system dynamic loader (/lib64/ld64.so.2).
> >
> > If you run "readelf -l" on your binary, what does it say under the INTERP segment (it should have something like "Requesting program interpreter: ...")?
> >
> > If this does *not* show /opt/atl12.0/lib64/ld64.so.2, then most likely you did not use the AT to *link* your application. Note, you specifically need to use the AT gcc also for the *link* step, not just for the compile steps.
> >
> >
> > Mit freundlichen Gruessen / Best Regards
> >
> > Ulrich Weigand
> >
> > --
> > Dr. Ulrich Weigand | Phone: +49-7031/16-3727
> > STSM, GNU/Linux compilers and toolchain
> > IBM Deutschland Research & Development GmbH
> > Vorsitzende des Aufsichtsrats: Martina Koederitz | Geschäftsführung: Dirk Wittkopp
> > Sitz der Gesellschaft: Böblingen | Registergericht: Amtsgericht Stuttgart, HRB 243294
> >
> > <graycol.gif>Ben Hood ---13.11.2018 18:25:17---Here’s the output from ldd: $ ldd /tmp/c4/erts-10.1.1/bin/erlc
> >
> > From: Ben Hood <ben at relops.com>
> > To: linuxppc-users at lists.ozlabs.org
> > Date: 13.11.2018 18:25
> > Subject: Re: [Linuxppc-users] Toolchain version for RHEL 7.1
> > Sent by: "Linuxppc-users" <linuxppc-users-bounces+ulrich.weigand=de.ibm.com at lists.ozlabs.org>
> >
> >
> >
> >
> > Here’s the output from ldd:
> >
> > $ ldd /tmp/c4/erts-10.1.1/bin/erlc
> > /tmp/c4/erts-10.1.1/bin/erlc: /lib64/ld64.30.2: version 'GLIBC2.23' not found (required by /opt/atl2.0/lib64/power8/libc.so.6)
> > /tmp/c4/erts-10.1.1/bin/erlc: /lib64/ld64.so.2: version 'GLIBC2.22' not found (required by /opt/atl2.0/lib64/power8/libc.so.6)
> > linux-vdso64.so.1 => (0x00003fff7acb0000)
> > libutil.so.l => /opt/atl2.0/lib64/power8/libutil.so.1 (0x00003fff7ac80000)
> > libdl.so.2 => /opt/atl2.0/lib64/power8/libdl.so.2 (0x00003fff7ac50000)
> > libm.30.6 => /opt/atl2.0/lib64/power8/libm.so.6 (0x00003fff7aaf0000)
> > libc.30.6 => /opt/atl2.0/lib64/power8/libc.so.6 (0x00003fff7a8a0000)
> > /opt/atl2.0/lib64/ld64.so.2 => /Iib64/ld64.so.2 (0x000000002d750000
> >
> > I’m wondering whether when I compiled the binary, I had somehow compiled/linked against a glibc that is only available on the build machine, rather than instructing the compiler to 100% link against the libraries supplied by AT?
> > On 13 Nov 2018, at 16:33, Ulrich Weigand <Ulrich.Weigand at de.ibm.com> wrote:
> > That runtime package is supposed to install the AT glibc (under /opt/atX.X/lib64/...).
> >
> > The AT dynamic loader (/opt/atX.X/lib64/ld64.so.2) is supposed to automatically use the AT libraries, including glibc, from /opt/atX.X/lib64/...
> >
> > AT-built binaries should have the AT dynamic loader pre-configured so they always use it.
> >
> > If this doesn't work for you, one of the above three steps must have failed somehow. What is the output when running your binary under "ldd" ?
> >
> >
> > Mit freundlichen Gruessen / Best Regards
> >
> > Ulrich Weigand
> >
> > --
> > Dr. Ulrich Weigand | Phone: +49-7031/16-3727
> > STSM, GNU/Linux compilers and toolchain
> > IBM Deutschland Research & Development GmbH
> > Vorsitzende des Aufsichtsrats: Martina Koederitz | Geschäftsführung: Dirk Wittkopp
> > Sitz der Gesellschaft: Böblingen | Registergericht: Amtsgericht Stuttgart, HRB 243294
> >
> > <graycol.gif>Ben Hood ---13.11.2018 16:28:15---Many thanks for the heads up. I’ve installed advance-toolchain-at12.0-runtime on the target machine,
> >
> > From: Ben Hood <ben at relops.com>
> > To: linuxppc-users at lists.ozlabs.org
> > Date: 13.11.2018 16:28
> > Subject: Re: [Linuxppc-users] Toolchain version for RHEL 7.1
> > Sent by: "Linuxppc-users" <linuxppc-users-bounces+ulrich.weigand=de.ibm.com at lists.ozlabs.org>
> >
> >
> >
> >
> >
> > Many thanks for the heads up.
> >
> > I’ve installed advance-toolchain-at12.0-runtime on the target machine, but this does not appear to supply a glibc of the required version.
> >
> > Is this in another AT RPM package?
> >
> > Perusing the FAQS
> >
> > https://developer.ibm.com/linuxonpower/advance-toolchain/adv-tool-usage/#faq1A <https://developer.ibm.com/linuxonpower/advance-toolchain/adv-tool-usage/#faq1A>
> >
> > Indicates that advance-toolchain-at12.0-runtime is mandatory, but I’m wondering if a GLIBC_2.23 object is supplied by a different package. I guess I could go through and install all of them to find out what works, but it feels like glibc is quite a fundamental building block.
> >
> > Other FAQs indicate that you can use
> >
> > /opt/atX.X/sbin/ldconfig
> >
> > To cache the correct library path, but running this hasn’t resolved the missing glibc object.
> >
> > Am I looking at the wrong FAQs?
> >
> >
> > > On 13 Nov 2018, at 13:52, Ulrich Weigand <Ulrich.Weigand at de.ibm.com> wrote:
> > >
> > > The AT uses a completely separate runtime library which is incompatible with the system library.
> > >
> > > In order to run any binary compiled with AT, you must always install (at least the runtime components of) the AT on the system that is to run the binary.
> > >
> > >
> > > Mit freundlichen Gruessen / Best Regards
> > >
> > > Ulrich Weigand
> > >
> > > --
> > > Dr. Ulrich Weigand | Phone: +49-7031/16-3727
> > > STSM, GNU/Linux compilers and toolchain
> > > IBM Deutschland Research & Development GmbH
> > > Vorsitzende des Aufsichtsrats: Martina Koederitz | Geschäftsführung: Dirk Wittkopp
> > > Sitz der Gesellschaft: Böblingen | Registergericht: Amtsgericht Stuttgart, HRB 243294
> > >
> > > <graycol.gif>Ben Hood ---13.11.2018 14:50:38---Hi list, I’m having issues loading the binary I’ve cross compiled for ppc64le/RHEL 7.1:
> > >
> > > From: Ben Hood <ben at relops.com>
> > > To: linuxppc-users at lists.ozlabs.org
> > > Date: 13.11.2018 14:50
> > > Subject: [Linuxppc-users] Toolchain version for RHEL 7.1
> > > Sent by: "Linuxppc-users" <linuxppc-users-bounces+ulrich.weigand=de.ibm.com at lists.ozlabs.org>
> > >
> > >
> > >
> > >
> > > Hi list,
> > >
> > > I’m having issues loading the binary I’ve cross compiled for ppc64le/RHEL 7.1:
> > >
> > > /lib64/ld64.so.2: version ‘GLIBC_2.23’ not found (required by /opt/at12.0/lib64/power8/libm.so.6)
> > >
> > > The onboard version of glibc appears to be 2.17-157.el7.
> > >
> > > What is the idiomatic way to supply the required glibc version?
> > >
> > > Is there some kind of incompatibility between the AT runtime version (and the choice of toolchain version on the build machine)? Can I solve this issue by using a different version of AT?
> > >
> > > Thanks in advance,
> > >
> > > Ben
> > >
> > >
> > > _______________________________________________
> > > Linuxppc-users mailing list
> > > Linuxppc-users at lists.ozlabs.org
> > > https://lists.ozlabs.org/listinfo/linuxppc-users <https://lists.ozlabs.org/listinfo/linuxppc-users>
> > >
> > >
> > >
> > >
> >
> > _______________________________________________
> > Linuxppc-users mailing list
> > Linuxppc-users at lists.ozlabs.org
> > https://lists.ozlabs.org/listinfo/linuxppc-users <https://lists.ozlabs.org/listinfo/linuxppc-users>
> >
> >
> >
> > _______________________________________________
> > Linuxppc-users mailing list
> > Linuxppc-users at lists.ozlabs.org
> > https://lists.ozlabs.org/listinfo/linuxppc-users <https://lists.ozlabs.org/listinfo/linuxppc-users>
> >
> >
> >
>
> _______________________________________________
> Linuxppc-users mailing list
> Linuxppc-users at lists.ozlabs.org
> https://lists.ozlabs.org/listinfo/linuxppc-users <https://lists.ozlabs.org/listinfo/linuxppc-users>
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/linuxppc-users/attachments/20181114/92aecaae/attachment-0001.html>
More information about the Linuxppc-users
mailing list