"GLIBC_2.3.4 not found" in CPU2006 on X86 KVM in SE mode Hello community,
I want to add fast forwarding into my SE scripts, so first I try to run CPU2006 tests on X86KvmCPU. Some tests (bzip2, mcf, gobmk, sjeng, specrand) pass. One test (hmmer) passes when compiled with older gcc 7.5.0. Most others fail to link libraries, like: ~/cpu2006/benchspec/CPU2006/464.h264ref/run/run_base_test_amd64-m64-gcc42-nn.0000/h264ref_base.amd64-m64-gcc42-nn: /lib/x86_64-linux-gnu/libm.so.6: version `GLIBC_2.3.4' not found (required by ~/cpu2006/benchspec/CPU2006/464.h264ref/run/run_base_test_amd64-m64-gcc42-nn.0000/h264ref_base.amd64-m64-gcc42-nn) Library files and their versions vary. These versions are actually absent in the .so mentioned, which are newer. Static linking (-static option) causes "page fault at addr 0" on X86KvmCPU. While there exist various workarounds (and I will appreciate if someone suggests a good one), one question really bothers me: How come gcc/g++/gfortran compile the tests so they run natively and on regular gem5, but require odd .so versions in SE KVM? Aren't the executables and the .so's same in all three cases? Secondly: Is it possible to prevent such dependencies during compiling? Or remove the check (if it won't cause errors)? Or should I reconfigure KVM somehow? Thirdly: What the hell is that "page fault at addr 0" if compile statically? Should I try other options with/instead of -static? My setup: Ryzen 3500 on B450 chipset Ubuntu 20.04 LTS gem5 20.1.0.4 _______________________________________________ gem5-users mailing list -- gem5-users@gem5.org To unsubscribe send an email to gem5-users-le...@gem5.org %(web_page_url)slistinfo%(cgiext)s/%(_internal_name)s
