`elflint` does rigorous sanity checks on linker-generated ELF files, and their assumption depends on GNU bfd linker

Always add `-flto`

Always add `-flto`

Always add `-flto` to LDFLAGS if a compiler is a recent versions of gcc

Directly invoke ld instead of cc without any -L option. Due to the lack of default search paths, mold failed to find -lc.

dev-perl/HTML-TableParser-0.420.0

dev-perl/Net-Server-Mail-0.280.0

It has an extra check for visibility of linker-synthesized symbols, though it looks like such check is not needed

It uses `nm` to check that only a limited set of symbols are exported from a dso, assuming that `nm` prints out only symbols in .dynsym. In reality, it prints out symbols in .symtab, as well, so the test fails

It uses `nm` to check that only a limited set of symbols are exported from a dso, assuming that `nm` prints out only symbols in .dynsym. In reality, it prints out symbols in .symtab, as well, so the test fails

It uses `nm` to check that only a limited set of symbols are exported from a dso, assuming that `nm` prints out only symbols in .dynsym. In reality, it prints out symbols in .symtab, as well, so the test fails

It uses a linker script

It uses a linker script

It uses a linker script

It uses a linker script

It uses a linker script

sys-kernel/gentoo-kernel-5.10.38

It uses a linker script

Its ./configure script searches for a substring "attempt to open <filename> succeeded" from --verbose output to find the libc path. mold's --verbose output isn't the same as bfd's.

Its test uses `nm` to verify that only a limited set of symbols are exported, but it doesn't look only .dynsym but also .symtab

mold by default do not make the stack area executable to tighten security, but the tool's test assumes that it is by default RWX

app-emulation/dosemu-1.4.1_pre20130107-r5

mold does not support --oformat=binary

mold does not support --oformat=binary

mold does not support `-Ttext=<address>`

mold does not try to resolve undefined symbols in dsos, and becasue of that a generated executable has fewer number of DT_NEEDED entries, which results in a runtime failure

sys-cluster/ampi-0_pre20140616

mold does not try to resolve undefined symbols in dsos, and becasue of that a generated executable has fewer number of DT_NEEDED entries, which results in a runtime failure

Some symbols remain undefined because mold does not pull out an object file from an archive for a common symbol

The package has a check to make sure that `ld` is not gold, since the program doesn't work correctly when linked with gold. mold's --version contains "gold", so it was detected as gold.