pam_tty_audit icanon log switch

[Richard Guy Briggs]

On Mon, Apr 22, 2013 at 01:28:05PM -0400, Miloslav Trmač wrote:
> ----- Original Message -----
> > On Thu, Apr 18, 2013 at 04:07:08PM -0400, Richard Guy Briggs wrote:
> > > On Thu, Apr 18, 2013 at 03:31:36PM -0400, Miloslav Trmač wrote:
> > > > > diff --git a/include/uapi/linux/audit.h b/include/uapi/linux/audit.h
> > > > > index 9f096f1..a863669 100644
> > > > > --- a/include/uapi/linux/audit.h
> > > > > +++ b/include/uapi/linux/audit.h
> > > > > @@ -369,7 +369,8 @@ struct audit_status {
> > > > >  };
> > > > >  
> > > > >  struct audit_tty_status {
> > > > > -	__u32		enabled; /* 1 = enabled, 0 = disabled */
> > > > > +	__u32		enabled;	/* 1 = enabled, 0 = disabled */
> > > > > +	__u32		log_icanon;	/* 1 = enabled, 0 = disabled */
> > > > >  };
> > > > 
> > > > Also, would it make sense for the user-space API to be more general
> > > > about expressing the intent ("log passwords")?  I don't know, being
> > > > precise about the exact effect of the option is also beneficial.
> > > 
> > > Hmmm, I'll have to ponder that...
> > 
> > I am inclined to leave it named as is for precision.
> 
> I suggested it might be better to change the name because with (icanon
> & !echo) being the discriminator, "log_icanon" is no longer precise.

Good point.  I was one thought behind...

> I can't think of an identifier that is both precise and understandable
> - neither log_icanon_noecho or log_passwords are the obvious
> possibilities, neither makes me enthusiastic.

Of these, I prefer log_passwords.

> Perhaps it doesn't matter that much what the audit_tty_status member
> is called - that's an implementation aspect and anybody touching this
> needs to understand the precise effects regardless of the name of the
> member.  The pam_tty_audit option name is user-visible and should be
> easy to understand and use.  (I'm not sure changing the topic like
> this is an improvement - it seems natural to use the same name for
> both, and even the users may need to understand the implications of
> icanon & !echo; I'm just hoping this might lead to better
> suggestions.)

Given the original intent, and the user-visible effects, I'd lean to
log_passwords, and let the kernel implement the details necessary to
carry out that intent.

>     Mirek

- RGB

--
Richard Guy Briggs <rbriggs at redhat.com>
Senior Software Engineer
AMER ENG Base Operating Systems
Remote, Canada, Ottawa
Voice: 1.647.777.2635
Internal: (81) 32635