Subject: Re: cvs commit: src/lib/libc/db/hash hash_buf.c
To: None <thorpej@nas.nasa.gov>
From: Jon Ribbens <jon@oaktree.co.uk>
List: tech-kern
Date: 10/19/1996 18:06:24
> ftp'd to localhost, logged in as myself.  Send that process a SIGSEGV
> with kill(1).  It definitely did not drop a core file.  Just to make
> sure, tried again with SIGABRT.  Nup.

Odd. I used SIGILL, but SIGSEGV ought to work too, according to
'man signal'.

> Oh .. I'm running a much more recent NetBSD kernel than your are, I guess.
> The coredump() function in kern_sig.c wasn't as picky in NetBSD 1.1.
> The behavior I'm describing is the way it appears in NetBSD 1.2.

> ! 	if (pcred->p_svuid != pcred->p_ruid ||
> ! 	    cred->cr_uid != pcred->p_ruid ||
> ! 	    pcred->p_svgid != pcred->p_rgid ||
> ! 	    cred->cr_gid != pcred->p_rgid)
>   		return (EFAULT);

None of those checks should prevent the ftpd dumping core, surely?
I think the only change between this and 1.1 is that it's also checking
if effective [ug]id != real [ug]id (assuming I'm gusesing all those
cryptic variable names right ;-) ).

Cheers


Jon
____
\  //    Jon Ribbens    //
 \// jon@oaktree.co.uk //