Re: [syzbot] Monthly trace report (Oct 2024)

[Steven Rostedt]

On Tue, 15 Oct 2024 12:05:29 +0200
Aleksandr Nogikh <nogikh@xxxxxxxxxx> wrote:

> > None of these look like they are tracing infrastructure related.  
> 
> Like get_maintainer.pl, syzbot relies on the MAINTAINERS file to
> attribute bugs to the individual kernel subsystems. If several ones
> are suitable, the bug is assigned several labels at once. It's now
> actually the case for all open "trace" findings:
> 
> https://syzkaller.appspot.com/upstream/s/trace
> 
> (FWIW it's also possible to manually overwrite these labels and remove
> specific bugs from the monthly reports).
> 
> I could make syzbot set "trace" only if there's no other good
> candidate, but I wonder if that could hide the findings in the trace
> infrastructure that manifested themselves in some specific traced
> subsystem.
> 

I don't mind being Cc'd to these bugs. What I do mind is that only the
tracing maintainers are Cc'd. I still care about these, because they do
depend on the tracing code, and it could be the tracing infrastructure's
fault. But if an error is in a file that is explicitly called out in the
maintainers file, such as, blktrace.c and bpf_trace.c, then PLEASE also Cc
the maintainers of those files!

I had to manually add those maintainers when I replied to the initial
email. That is something I shouldn't need to do.

-- Steve