public inbox for gdb-prs@sourceware.org help / color / mirror / Atom feed
From: "rui314 at gmail dot com" <sourceware-bugzilla@sourceware.org> To: gdb-prs@sourceware.org Subject: [Bug gdb/30008] New: gdb fails to get line numbers with gold/mold with --gdb-index Date: Tue, 17 Jan 2023 06:25:50 +0000 [thread overview] Message-ID: <bug-30008-4717@http.sourceware.org/bugzilla/> (raw) https://sourceware.org/bugzilla/show_bug.cgi?id=30008 Bug ID: 30008 Summary: gdb fails to get line numbers with gold/mold with --gdb-index Product: gdb Version: HEAD Status: UNCONFIRMED Severity: normal Priority: P2 Component: gdb Assignee: unassigned at sourceware dot org Reporter: rui314 at gmail dot com Target Milestone: --- I found this gdb's odd behavior when I was debugging a mold-generated executable, but I could reproduce the same issue with GNU gold, so I'll report it as a gdb issue. If it is actually a linker's issue, please let me know so that I can fix it on the linker side. Since the GDB index is a gdb-specific feature and not very documented well, it is very hard for me to debug this issue. So, here is the issue. I built the mold linker (https://github.com/rui314/mold) with GNU ld and ran it under gdb. Almost all functions in mold are actually C++ templates that takes a target type as a type parameter. Therefore, for examle, we have `parse_nonpositional_args<mold::elf::X86_64>`, `parse_nonpositional_args<mold::elf::ARM32>`, etc in the same executable. When I set a breakpoint at just `parse_nonpositional_args`, gdb fails to obtain its line number, as shown below. ruiu@blue:~/mold/build-gold$ gdb ./mold -nx -batch -ex 'b parse_nonpositional_args' -ex 'run --no-fork --no-threads' -ex bt -ex c -ex bt -ex quit Breakpoint 1 at 0x1194e4 (16 locations) [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Breakpoint 1, 0x000055555566d4e4 in std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > > mold::elf::parse_nonpositional_args<mold::elf::X86_64>(mold::elf::Context<mold::elf::X86_64>&) () #0 0x000055555566d4e4 in std::vector<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, std::allocator<std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> > > > mold::elf::parse_nonpositional_args<mold::elf::X86_64>(mold::elf::Context<mold::elf::X86_64>&) () #1 0x0000555555886568 in int mold::elf::elf_main<mold::elf::X86_64>(int, char**) () #2 0x000055555588212b in mold::elf::main(int, char**) () #3 0x0000555556dba344 in main (argc=3, argv=0x7fffffffe178) at /home/ruiu/mold/common/main.cc:155 mold: fatal: -m option is missing [Inferior 1 (process 3523615) exited with code 01] No stack. However, if I specify the exact name of the function (in this case `parse_nonpositional_args<mold::elf::X86_64>`), gdb can obtain its line number without problems. ruiu@blue:~/mold/build-gold$ gdb ./mold -nx -batch -ex 'b parse_nonpositional_args<mold::elf::X86_64>' -ex 'run --no-fork --no-threads' -ex bt -ex c -ex bt -ex quit Breakpoint 1 at 0x119504: file /home/ruiu/mold/elf/cmdline.cc, line 395. [Thread debugging using libthread_db enabled] Using host libthread_db library "/lib/x86_64-linux-gnu/libthread_db.so.1". Breakpoint 1, mold::elf::parse_nonpositional_args<mold::elf::X86_64> (ctx=...) at /home/ruiu/mold/elf/cmdline.cc:395 395 std::span<std::string_view> args = ctx.cmdline_args; #0 mold::elf::parse_nonpositional_args<mold::elf::X86_64> (ctx=...) at /home/ruiu/mold/elf/cmdline.cc:395 #1 0x0000555555886568 in mold::elf::elf_main<mold::elf::X86_64> (argc=3, argv=0x7fffffffe178) at /home/ruiu/mold/elf/elf-main.cc:387 #2 0x000055555588212b in mold::elf::main (argc=3, argv=0x7fffffffe178) at /home/ruiu/mold/elf/elf-main.cc:758 #3 0x0000555556dba344 in main (argc=3, argv=0x7fffffffe178) at /home/ruiu/mold/common/main.cc:155 mold: fatal: -m option is missing [Inferior 1 (process 3524620) exited with code 01] No stack. Why does gdb behave differently for the above two test cases? Uploaded a gold-linked mold executable here: https://drive.google.com/file/d/1YuLUCJFJV-cJ9zQ7NRxvSSqYgoShCF_N/view?usp=sharing -- You are receiving this mail because: You are on the CC list for the bug.
next reply other threads:[~2023-01-17 6:25 UTC|newest] Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top 2023-01-17 6:25 rui314 at gmail dot com [this message] 2023-01-17 18:20 ` [Bug gdb/30008] " simon.marchi at polymtl dot ca 2023-01-18 0:48 ` rui314 at gmail dot com 2023-02-15 15:56 ` tromey at sourceware dot org 2023-02-15 16:28 ` simon.marchi at polymtl dot ca 2023-02-15 18:15 ` tromey at sourceware dot org 2023-02-16 0:07 ` rui314 at gmail dot com 2023-02-16 2:32 ` sam at gentoo dot org
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=bug-30008-4717@http.sourceware.org/bugzilla/ \ --to=sourceware-bugzilla@sourceware.org \ --cc=gdb-prs@sourceware.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).