Re: Bug#888478: systemd FTBFS on mipsel: /usr/bin/ld: BFD (GNU Binutils for Debian) 2.29.90.20180122 assertion fail ../../bfd/elflink.c:9757

James Cowgill <[email protected]> Fri, 26 Jan 2018 17:00:33 +0000
Newsgroups	gmane.linux.debian.ports.mips
Message-ID	<[email protected]>
Control: forwarded -1 https://sourceware.org/bugzilla/show_bug.cgi?id=22751
Control: retitle -1 binutils: lto broken for libgcc libcalls

Hi,

On 26/01/18 15:38, Aurelien Jarno wrote:
> On 2018-01-26 15:19, James Cowgill wrote:
>> Hi,
>>
>> On 26/01/18 08:57, Aurelien Jarno wrote:
>>>> The build was performed on eller.d.o.
>>>>
>>>> If you reassign to binutils, please set affects.
>>>>
>>>> I put [email protected] into X-Debbugs-Cc. During cross builds I was able to
>>>> reproduce a very similar build failure for mips and armhf. A native
>>>> build on amd64 seems unaffected. Thus I have included [email protected] as
>>>> well.
>>>
>>> Note that the build also fails on i386 on the same files, but in a
>>> different way:
>>>
>>> | FAILED: test-ipv4ll
>>> | cc  -o test-ipv4ll 'test-ipv4ll@exe/src_libsystemd-network_test-ipv4ll.c.o' -flto -Wl,--no-undefined -Wl,--as-needed -Wl,-z,relro -Wl,-z,now -pie -Wl,--gc-sections -g -O2 -fdebug-prefix-map=/tmp/systemd-236=. -fstack-protector-strong -Wformat -Werror=format-security -Wl,-z,relro -Wl,--start-group src/shared/libsystemd-shared-236.so src/libsystemd-network/libsystemd-network.a -Wl,--end-group '-Wl,-rpath,$ORIGIN/src/shared' -Wl,-rpath-link,/tmp/systemd-236/build-deb/src/shared 
>>> | /usr/bin/ld: /tmp/ccFNIZSq.ltrans0.ltrans.o(.text+0x99f): unresolvable R_386_PLT32 relocation against symbol `__umoddi3'
>>> | /usr/bin/ld: final link failed: Nonrepresentable section on output
>>> | collect2: error: ld returned 1 exit status
>>>
>>> That might help debugging the issue as it is more understandable than
>>> a simple assert.
>>
>> Fairly minimal testcase which triggers both bugs:
>>
>> test.c:
>> volatile unsigned long long a = 42;
>> volatile unsigned long long b = 1;
>>
>> int main(void)
>> {
>>         return (int) (a / b);
>> }
>>
>> build with:
>> gcc -flto -O2 -c test.c
>> ar rcs libtest.a test.o
>> gcc -flto -Wl,--whole-archive libtest.a -Wl,--no-whole-archive
>>
>> On mipsel:
>> /usr/bin/ld: BFD (GNU Binutils for Debian) 2.29.90.20180122 assertion fail ../../bfd/elflink.c:9757
>> collect2: error: ld returned 1 exit status
>>
>> On i386:
>> /usr/bin/ld: /tmp/cc5ZOEfW.ltrans0.ltrans.o(.text.startup+0x3d): unresolvable R_386_PLT32 relocation against symbol `__udivdi3'
>> /usr/bin/ld: final link failed: Nonrepresentable section on output
>> collect2: error: ld returned 1 exit status
> 
> Thanks for the reduced testcase. It means it can also be easily
> reproduced on amd64 with the following command but slightly different
> code:
> 
> volatile  __int128 a = 42;
> volatile  __int128 b = 1;
> 
> int main(void)
> {
>         return (int) (a / b);
> }
> 
> /usr/bin/ld: /tmp/ccwdC3bA.ltrans0.ltrans.o(.text.startup+0x21): unresolvable R_X86_64_PLT32 relocation against symbol `__divti3'
> /usr/bin/ld: final link failed: Nonrepresentable section on output

Thanks for that.

I've submitted an upstream bug and bisected it for them. I've guessed a
bug title which is hopefully more accurate (since this is not mipsel
specific).

James
signature.asc (application/pgp-signature, 833 B)