Dear Arnau,

I believe that we fixed this issue in 
https://github.com/easybuilders/easybuild-easyconfigs/pull/18250

With XZ built with that change I see:

$ objdump -tT XZ/5.2.5-GCCcore-10.3.0/lib/liblzma.so.5 | grep 
lzma_stream_encoder_mt)
000000000000a690 l     F .text  000000000000006a              
lzma_stream_encoder_mt_52
000000000000a690 l     F .text  000000000000006a              
lzma_stream_encoder_mt_522
000000000000a690 l     F .text  000000000000006a              
lzma_stream_encoder_mt_512a
000000000000a700 l     F .text  00000000000000f8              
lzma_stream_encoder_mt_memusage_52
000000000000a700 l     F .text  00000000000000f8              
lzma_stream_encoder_mt_memusage_522
000000000000a700 l     F .text  00000000000000f8              
lzma_stream_encoder_mt_memusage_512a
000000000000a700 g     F .text  00000000000000f8              
lzma_stream_encoder_mt_memusage
000000000000a690 g     F .text  000000000000006a              
lzma_stream_encoder_mt
000000000000a700 g     F .text  00000000000000f8              
lzma_stream_encoder_mt_memusage
000000000000a690 g     F .text  000000000000006a              
lzma_stream_encoder_mt
000000000000a690 g     F .text  000000000000006a              
lzma_stream_encoder_mt
000000000000a700 g     F .text  00000000000000f8              
lzma_stream_encoder_mt_memusage
000000000000a700 g    DF .text  00000000000000f8  XZ_5.2      
lzma_stream_encoder_mt_memusage
000000000000a700 g    DF .text  00000000000000f8 (XZ_5.2.2)   
lzma_stream_encoder_mt_memusage
000000000000a700 g    DF .text  00000000000000f8 (XZ_5.1.2alpha) 
lzma_stream_encoder_mt_memusage
000000000000a690 g    DF .text  000000000000006a  XZ_5.2      
lzma_stream_encoder_mt
000000000000a690 g    DF .text  000000000000006a (XZ_5.2.2)   
lzma_stream_encoder_mt
000000000000a690 g    DF .text  000000000000006a (XZ_5.1.2alpha) 
lzma_stream_encoder_mt


Regards,
Simon

From: [email protected] <[email protected]> On 
Behalf Of Arnau
Sent: 04 October 2024 11:03
To: [email protected]
Subject: [easybuild] XZ-5.2.5 compilation & nonosymbols

CAUTION: This email originated from outside the organisation. Do not click 
links or open attachments unless you recognise the sender and know the content 
is safe.

Dear all,

I have a problem with XZ compiled with EB and the one from Rocky 9.
When I load my Python/3.11.1-GCCcore-10.3.0-jupyter-4.0 module I get errors 
from the os:

# flatpak
flatpak: symbol lookup error: /lib64/librpmio.so.9: undefined symbol: 
lzma_stream_encoder_mt, version XZ_5.2

# ldd /lib64/librpmio.so.9
[...]
        liblzma.so.5 => 
/apps/prod/easybuild/sl7.x86_64.foss-2021a/software/XZ/5.2.5-GCCcore-10.3.0/lib/liblzma.so.5
 (0x00007f993163c000)

the symbol is there but, as you can see, it's listed twice, first in the static 
table and then in the dynamic static table:

# objdump -tT 
/apps/prod/easybuild/sl7.x86_64.foss-2021a/software/XZ/5.2.5-GCCcore-10.3.0/lib/liblzma.so.5|grep
 lzma_stream_encoder_mt
000000000000a020 g     F .text  0000000000000100              
lzma_stream_encoder_mt_memusage
0000000000009fb0 g     F .text  000000000000006a              
lzma_stream_encoder_mt
000000000000a020 g    DF .text  0000000000000100  XZ_5.1.2alpha 
lzma_stream_encoder_mt_memusage
0000000000009fb0 g    DF .text  000000000000006a  XZ_5.1.2alpha 
lzma_stream_encoder_mt


the problems seems to be that the version from the static table is missing.

if I compare my compilation with the one from Rocky:

#  objdump -tT /lib64/liblzma.so.5|grep lzma_stream_encoder_mt
00000000000101d0 g    DF .text  00000000000000f9  XZ_5.2      
lzma_stream_encoder_mt_memusage
0000000000010150 g    DF .text  0000000000000072  XZ_5.2      
lzma_stream_encoder_mt

they have all the symbols in the dyanmic table (no static table at all) and all 
with vesions.
question here, as I'm not an expert on this, what should be the best approach 
to solve this issue? remove all symbols from static table or add versions to 
the objects in the static table? I don't really understand the implicatiopns of 
one or the other option...
Thanks,
Arnau

Reply via email to