Date: Tue, 20 Jan 2004 19:44:00 -0800 From: Tim Kientzle <kientzle@acm.org> To: Bruce Evans <bde@zeta.org.au> Cc: cvs-src@FreeBSD.org Subject: Re: cvs commit: src/sys/alpha/alpha support.s src/sys/i386/i386 swtch.s src/sys/kern kern_shutdown.c src/sys/sys systm.h Message-ID: <400DF580.2040404@acm.org> In-Reply-To: <20040121123903.W6693@gamplex.bde.org> References: <20040120182929.4573C16A4CF@hub.freebsd.org> <20040121123903.W6693@gamplex.bde.org>
next in thread | previous in thread | raw e-mail | index | archive | help
On Tue, 20 Jan 2004, Bill Paul wrote: > >Anyway, to reiterate: I don't think the lineno/sourcefile additions >to panic() are worth the bloat. I think the lineno/sourcefile additions are very valuable, but for a reason that noone has so far mentioned: Grooming new developers. In short, I think a wannabe kernel hacker is much more likely to dive in and try to understand a problem if you give them some rough guidance about where to start looking. For someone who doesn't already know the codebase, that means a source file and line number, not a function name or panic string that requires them to search the entire codebase. Sure, lineno/filename doesn't help with all kinds of panics, and it's not as good as having every kernel panic immediately open a remote GDB window over the Internet directly on the appropriate developer's desktop. But if it engages the interest of a couple of future FreeBSD developers, then it is worth a lot more than 32k of data space. I would like for tracebacks to have sourcefile/lineno information for the same reason, but that's an awful lot more bloat than Poul-Henning's quite modest proposal. Tim
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?400DF580.2040404>