llvm: improve emitted debug info

[Vexu]

• ensure parameter debug info is in the subroutine di scope
• slit sub file path into basename and dirname

Closes#12257
Closes#12665

Might also resolve the following: #13224 and #13315 but I haven't been able to reproduce those.

[kubkon]

I think I should be able to quickly verify if #13315 is fixed. @nektro could you verify that #13224 is fixed with Vexu's patch?

[kubkon]

The issue #13315 still persists:

❯ ./zld Segmentation fault at address 0x300
???:?:?: 0x7ff80223c572 in ??? (???)
/Users/jakubkonka/dev/tracy/public/common/../client/TracyProfiler.hpp:697:94: 0x109ee8029 in SendString (/Users/jakubkonka/dev/tracy/public/TracyClient.cpp)
void SendString( uint64_t str, const char* ptr, QueueType type ) { SendString( str, ptr, strlen( ptr ), type ); }
^
/Users/jakubkonka/dev/tracy/public/client/TracyProfiler.cpp:3340:9: 0x109ede4f9 in HandleServerQuery (/Users/jakubkonka/dev/tracy/public/TracyClient.cpp)
SendString( ptr, (const char*)ptr, QueueType::StringData );
^
/Users/jakubkonka/dev/tracy/public/client/TracyProfiler.cpp:1865:30: 0x109edbdb4 in Worker (/Users/jakubkonka/dev/tracy/public/TracyClient.cpp)
connActive = HandleServerQuery();
^
/Users/jakubkonka/dev/tracy/public/common/../client/TracyProfiler.hpp:754:63: 0x109eda794 in LaunchWorker (/Users/jakubkonka/dev/tracy/public/TracyClient.cpp)
static void LaunchWorker( void* ptr ) { ((Profiler*)ptr)->Worker(); }
^
/Users/jakubkonka/dev/tracy/public/common/../client/TracyThread.hpp:80:40: 0x109efbc7c in Launch (/Users/jakubkonka/dev/tracy/public/TracyClient.cpp)
static void* Launch( void* ptr ) { ((Thread*)ptr)->m_func( ((Thread*)ptr)->m_ptr ); return nullptr; }
^
???:?:?: 0x7ff802214258 in ??? (???)
???:?:?: 0x7ff80220fc7a in ??? (???)
???:?:?: 0x0 in ??? (???)
./zld: line 3: 24471 Abort trap: 6 /Users/jakubkonka/dev/zld/zig-out/bin/ld64 -dynamic -arch x86_64 -platform_version macos 13.0.0 13.0 -syslibroot /Library/Developer/CommandLineTools/SDKs/MacOSX.sdk -o redis-server -L/usr/local/lib adlist.o quicklist.o ae.o anet.o dict.o server.o sds.o zmalloc.o lzf_c.o lzf_d.o pqsort.o zipmap.o sha1.o ziplist.o release.o networking.o util.o object.o db.o replication.o rdb.o t_string.o t_list.o t_set.o t_zset.o t_hash.o config.o aof.o pubsub.o multi.o debug.o sort.o intset.o syncio.o cluster.o crc16.o endianconv.o slowlog.o scripting.o bio.o rio.o rand.o memtest.o crc64.o bitops.o sentinel.o notify.o setproctitle.o blocked.o hyperloglog.o latency.o sparkline.o redis-check-rdb.o redis-check-aof.o geo.o lazyfree.o module.o evict.o expire.o geohash.o geohash_helper.o childinfo.o defrag.o siphash.o rax.o t_stream.o listpack.o localtime.o lolwut.o lolwut5.o lolwut6.o acl.o gopher.o tracking.o connection.o tls.o sha256.o ../deps/hiredis/libhiredis.a ../deps/lua/src/liblua.a -lm -ldl -lSystem /Library/Developer/CommandLineTools/usr/lib/clang/14.0.0/lib/darwin/libclang_rt.osx.a

[nektro]

13224 currently fails with panic: Zig compiler bug: attempted to destroy declaration with an attached error on 0.11.0.dev

[Vexu]

But do you get a stack trace?

[nektro]

https://bpa.st/raw/2EMA yeah but thats even on master, so the issue no longer repros for me with that input

[micahswitzer]

The change to how the LLVM backend emits the path to the source file leads to a duplicated parent directory with correct DWARF parsing. This can be observed with llvm-dwarfdump and/or my patch to fix std.dwarf's parsing logic (#13540)

There's more context as to my changes to the DWARF parsing code on Discord: https://discord.com/channels/605571803288698900/1041373573908021248/1041386215733215352

Example:

// src/main.zigconstb=@import("a/b.zig");
pubfnmain() void {
b.doTest();
}

// src/a/b.zigpubfndoTest() void {
@panic("test");
}

Before this patch:

$ llvm-dwarfdump -n doTest mainmain: file format elf64-x86-640x000041c6: DW_TAG_subprogram DW_AT_low_pc (0x000000000020bc30) DW_AT_high_pc (0x000000000020bc56) DW_AT_frame_base (DW_OP_reg6 RBP) DW_AT_linkage_name ("a.b.doTest") DW_AT_name ("doTest") DW_AT_decl_file ("src/a/b.zig") DW_AT_decl_line (1) DW_AT_type (0x00003ccd "void")

With this patch:

$ llvm-dwarfdump -n doTest mainmain: file format elf64-x86-640x00004156: DW_TAG_subprogram DW_AT_low_pc (0x000000000020bc30) DW_AT_high_pc (0x000000000020bc56) DW_AT_frame_base (DW_OP_reg6 RBP) DW_AT_linkage_name ("a.b.doTest") DW_AT_name ("doTest") DW_AT_decl_file ("src/src/a/b.zig") DW_AT_decl_line (1) DW_AT_type (0x00003c79 "void")