Skip libraries consisting of the empty string

[pchintalapudi]

Should fix #50745 (comment) and #6184 (comment), probably.

[gbaraldi]

LGTM

[maleadt]

Not really familiar with the pass, but SGTM.

[pchintalapudi]

The test only fails due to #22318.

[maleadt]

CI looks good, apart from macOS x86_64 which is backlogged. I'm going to go ahead and merge this already, as it's unlikely that this would cause an issue on only that platform, and without it there's many crashes on PkgEval.

[maleadt]

This has not fixed the PkgEval issue I mentioned, see https://s3.amazonaws.com/julialang-reports/nanosoldier/pkgeval/by_date/2023-08/15/report.html

For example:

Failed to precompile SymPy [24249f21-da20-56a4-8eb1-6a02cf4ae2e6] to "/home/pkgeval/.julia/compiled/v1.11/SymPy/jl_wEH7oC".
julia: /source/usr/include/llvm/Support/Casting.h:578: decltype(auto) llvm::cast(From*) [with To = llvm::ConstantExpr; From = llvm::Value]: Assertion `isa<To>(Val) && "cast<Ty>() argument of incompatible type!"' failed.

[496] signal (6.-6): Aborted
in expression starting at /home/pkgeval/.julia/packages/SymPy/AswGW/src/SymPy.jl:39
gsignal at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
abort at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
unknown function (ip: 0x7f33aa10c40e)
__assert_fail at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
cast<llvm::ConstantExpr, llvm::Value> at /source/usr/include/llvm/Support/Casting.h:578 [inlined]
operator() at /source/src/jitlayers.cpp:1597
optimizeDLSyms at /source/src/jitlayers.cpp:2032 [inlined]
_jl_compile_codeinst at /source/src/jitlayers.cpp:227
jl_generate_fptr_impl at /source/src/jitlayers.cpp:525
jl_compile_method_internal at /source/src/gf.c:2475 [inlined]
jl_compile_method_internal at /source/src/gf.c:2363
gen_cfun_wrapper at /source/src/codegen.cpp:6079
emit_cfunction at /source/src/codegen.cpp:6716 [inlined]
emit_expr at /source/src/codegen.cpp:5547
emit_ssaval_assign at /source/src/codegen.cpp:5119
emit_stmtpos at /source/src/codegen.cpp:5354 [inlined]
emit_function at /source/src/codegen.cpp:8467

[gbaraldi]

I can't reproduce this failure, does pkgeval have this PR?

[maleadt]

Yes. You can check the versioninfo() that's part of every log, e.g.: https://s3.amazonaws.com/julialang-reports/nanosoldier/pkgeval/by_date/2023-08/15/Roots.primary.log, which shows 0b190b3, the very commit where this Pr was merged.

[vtjnash]

doesn't seem to have fixed this case either:

julia> ccall(:test, Int, ())
julia: /home/vtjnash/julia1/src/jitlayers.cpp:1597: void JuliaOJIT::DLSymOptimizer::operator()(llvm::Module&): Assertion `cast<ConstantExpr>(libarg)->getOpcode() == Instruction::IntToPtr && "libarg should be either a global variable or a integer index!"' failed.

[2617529] signal (6.-6): Aborted
in expression starting at REPL[1]:1
gsignal at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
abort at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
unknown function (ip: 0x7f3323207728)
__assert_fail at /lib/x86_64-linux-gnu/libc.so.6 (unknown line)
operator() at /home/vtjnash/julia1/src/jitlayers.cpp:1597
optimizeDLSyms at /home/vtjnash/julia1/src/jitlayers.cpp:2032 [inlined]

julia> code_llvm(()) do; ccall(:test, Int, ()); end
define i64 @"julia_#3_496"() #1 {
top:
  %test1 = load atomic i64 ()*, i64 ()** bitcast (void ()** @jlplt_test_499_got to i64 ()**) unordered, align 8
  %0 = call i64 %test1()
  ret i64 %0
}