Viktor Malík

Sadly, I cannot reproduce. Building on Alt Sisyphus with CMake 3.23.2 generates the file correctly. Are you running `tests/runtime-tests.sh` from the build directory?

I can't reproduce this. Building your program with the provided GCC command and then running: ``` bpftrace -e 'u:./test_trace:func {print(ustack)}' Attaching 1 probe... func+0 __libc_start_call_main+128 func+0 __libc_start_call_main+128 ^C ``` ```...

> it seems bpftrace get wrong pid(kernel pid)? `cat /proc/1087/maps` should be `cat /proc/801/maps` Yes, this is incorrect - `pid` and `cpid` should be the same here. It really seems...

This is definitely a great thing to have. I'm slightly preferring the second option as it seems more intuitive and readable. Even though the "probe-like" syntax is consistent with the...

Our matching of mangled symbols is not very good. Not only we can't distinguish mangled and non-mangled symbols, but adding those implicit leading and trailing wildcards causes some weird behaviour....

I'm fine with allowing the `:c++` prefix, but I wouldn't force users to do so in this case, as name mangling is not a C++ specific feature. For the above...

> > I'm fine with allowing the `:c++` prefix, but I wouldn't force users to do so in this case, as name mangling is not a C++ specific feature. For...

The problem is the `active` field. The DWARF parser does not yet support bitfields and it falls back to the Clang parser which needs the full definition of the type....

I'm not against adding a new builtin function, but we should be cautious about this one. Looking at the helper description from the [bpf-helpers manpage](https://man7.org/linux/man-pages/man7/bpf-helpers.7.html): > Also, note that bpf_trace_printk()...

It could be nice to have `strstr` and since the BPF verifier now allows bounded loops and we have strings of bounded size, it should be possible to implement it....