Compilation failure with WITH_DTRACE_TESTS on mips/mips and powerpc/powerpc; it's trying to compile a host object on mips/powerpc because MAKEOBJDIRPREFIX is incorrect

Bryan Drewery bdrewery at FreeBSD.org
Fri Oct 30 23:09:58 UTC 2015


On 10/30/2015 4:08 PM, Mark Johnston wrote:
> On Fri, Oct 30, 2015 at 04:01:27PM -0700, Bryan Drewery wrote:
>> On 10/30/2015 3:03 PM, NGie Cooper wrote:
>>> On Fri, Oct 30, 2015 at 2:34 PM, Bryan Drewery <bdrewery at freebsd.org> wrote:
>>>> On 10/30/2015 2:21 PM, Bryan Drewery wrote:
>>>>> On 10/30/2015 1:57 PM, NGie Cooper wrote:
>>>>>> Hi Bryan/Simon!
>>>>>>      I tried doing buildworld on powerpc/powerpc with -DWITH_DTRACE_TESTS and I ran into this linker issue below. I have no idea (yet) why it’s trying to compile an x64 object when I specify powerpc/powerpc — and more importantly, why is the object not being put in obj.powerpc?
>>>>>>      I ran into the same issue on ref11-amd64.freebsd.org when I ran “make tinderbox".
>>>>>> Thanks!
>>>>>> -NGie
>>>>>>
>>>>>
>>>>> Have you modified any of your local toolchain handling, or skipped
>>>>> CLANG_BOOTSTRAP? I would expect this to be failing much more broadly and
>>>>> there to be a lot more reports if there was a problem with buildworld
>>>>> cross compiling.
>>>>>
>>>>>> % make buildworld TARGET=powerpc TARGET_ARCH=powerpc
>>>>>>>>>>>> ===> cddl/usr.sbin/dtrace/tests/common/json (all)
>>>>>> (cd /usr/src/svn/cddl/usr.sbin/dtrace/tests/common/json &&  DEPENDFILE=.depend.tst.usdt.exe  NO_SUBDIR=1 make -f /usr/src/svn/cddl/usr.sbin/dtrace/tests/common/json/Makefile _RECURSING_PROGS=  PROG=tst.usdt.exe )
>>>>>> cc  -O2 -pipe -fno-strict-aliasing -O2 -pipe   -O0 -g -I/usr/obj/powerpc.powerpc/usr/src/svn/cddl/usr.sbin/dtrace/tests/common/json -std=gnu99 -fstack-protector-strong    -c /usr/src/svn/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/json/tst.usdt.c -o tst.usdt.o
>>>>>> dtrace -C -x nolibs -G -o usdt.o -s /usr/src/svn/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/json/usdt.d tst.usdt.o
>>>>>> dtrace: failed to link script /usr/src/svn/cddl/contrib/opensolaris/cmd/dtrace/test/tst/common/json/usdt.d: incorrect ELF machine type for object file: tst.usdt.o
>>>>>> *** Error code 1
>>
>> The problem looks specific to compiling of .d files using dtrace(1). It
>> must not have cross-compile support.
>>
>> The manpage does say: "The D compiler produces programs using the native
>> data model of the operating system kernel.".
>>
>> So these will need to be disabled for non-native builds.
>>
>> I don't know if it would be possible to build a cross-compile version of
>> dtrace(1) and drop it in WORLDTMP/usr/sbin and have it work.
> 
> In the snippet above, tst.usdt.o is generated by cc, not dtrace(1).
> dtrace is complaining that the input file doesn't have the expected
> machine type, which seems valid given the file(1) output below.
> 

The example output must be a mistake as they are correct on ref11:

ref11-amd64% find
/scratch/tmp/ngie/obj/*/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json
-name tst.usdt.o -exec file {} +
/scratch/tmp/ngie/obj/arm.arm/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
          ELF 32-bit LSB relocatable, ARM, EABI5 version 1 (FreeBSD),
not stripped
/scratch/tmp/ngie/obj/arm.armeb/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
        ELF 32-bit MSB relocatable, ARM, EABI5 version 1 (FreeBSD), not
stripped
/scratch/tmp/ngie/obj/arm.armv6/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
        ELF 32-bit LSB relocatable, ARM, EABI5 version 1 (FreeBSD), not
stripped
/scratch/tmp/ngie/obj/arm.armv6hf/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
      ELF 32-bit LSB relocatable, ARM, EABI5 version 1 (FreeBSD), not
stripped
/scratch/tmp/ngie/obj/arm64.aarch64/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
    ELF 64-bit LSB relocatable, ARM aarch64, version 1 (FreeBSD), not
stripped
/scratch/tmp/ngie/obj/i386.i386/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
        ELF 32-bit LSB relocatable, Intel 80386, version 1 (FreeBSD),
not stripped
/scratch/tmp/ngie/obj/pc98.i386/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
        ELF 32-bit LSB relocatable, Intel 80386, version 1 (FreeBSD),
not stripped
/scratch/tmp/ngie/obj/powerpc.powerpc/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
  ELF 32-bit MSB relocatable, PowerPC or cisco 4500, version 1
(FreeBSD), not stripped
/scratch/tmp/ngie/obj/powerpc.powerpc64/scratch/tmp/ngie/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o:
ELF 64-bit MSB relocatable, 64-bit PowerPC or cisco 7500, version 1
(FreeBSD), not stripped

[sorry for bad mail client]



>>
>>>>>> $ find /usr/obj/usr/src/svn/ -name tst.usdt.o
>>>>>> /usr/obj/usr/src/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o
>>>>>> $ file /usr/obj/usr/src/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o
>>>>>> /usr/obj/usr/src/svn/cddl/usr.sbin/dtrace/tests/common/json/tst.usdt.o: ELF 64-bit LSB relocatable, x86-64, version 1 (FreeBSD), not stripped
>>>>>>
>>>>
>>>> I ran a buildworld with TARGET=powerpc just a few days ago and it seemed
>>>> to be fine with PROGS.  Here's a test object built via PROGS:
>>>>
>>>> ~/git/freebsd # find /usr/obj/powerpc.powerpc -name ld_library_pathfds.o
>>>> /usr/obj/powerpc.powerpc/root/git/freebsd/libexec/rtld-elf/tests/ld_library_pathfds.o
>>>> ~/git/freebsd # file
>>>> /usr/obj/powerpc.powerpc/root/git/freebsd/libexec/rtld-elf/tests/ld_library_pathfds.o
>>>> /usr/obj/powerpc.powerpc/root/git/freebsd/libexec/rtld-elf/tests/ld_library_pathfds.o:
>>>> ELF 32-bit MSB relocatable, PowerPC or cisco 4500, version 1 (FreeBSD),
>>>> not stripped
>>>> -rw-r--r--  1 root  wheel  21136 Oct 23 17:08
>>>> /usr/obj/powerpc.powerpc/root/git/freebsd/libexec/rtld-elf/tests/ld_library_pathfds.o
>>>>
>>>> I see nothing special with the DTRACE_TESTS to change any of this.
>>>
>>> I could see there being a possible issue with my host VM, but I
>>> haven't modified my environment in ref11-amd64.freebsd.org at all.
>>>
>>> Could you please try reproing it there with your user?
>>>
>>> Thanks,
>>> -NGie
>>>
>>
>>
>> -- 
>> Regards,
>> Bryan Drewery
>>
> 
> 


-- 
Regards,
Bryan Drewery

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 473 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freebsd.org/pipermail/freebsd-current/attachments/20151030/f3e9495d/attachment.bin>


More information about the freebsd-current mailing list