[DTrace-devel] [PATCH v2] Add support for inet_ntop() subroutine

[Kris Van Hees]

On Thu, Sep 14, 2023 at 06:44:30PM -0400, Eugene Loh via DTrace-devel wrote:
> Reviewed-by: Eugene Loh <eugene.loh at oracle.com>
> I confess I don't really understand this stuff but oh well.

It is manual construction of a node tree, cooking it (finalizing it), and
then having the code generator having its way with it.  So, I am essentially
performing a code transformation on the fly.

> Also, test/unittest/funcs/tst.inet_ntop.d fails.  Should the .r file be
> changed so that the AF_INET6 cases with embedded IPV4 no longer have IPv6
> prefixes?

Yes and no.  I plan to post a patch later today to implement the proper prefix
handling for IPv4 in inet_ntop vs inet_ntoa6.

> And...
> 
> On 9/14/23 15:56, Eugene Loh via DTrace-devel wrote:
> > diff --git a/libdtrace/dt_cg.c b/libdtrace/dt_cg.c
> > +static void
> > +dt_cg_subr_inet_ntop(dt_node_t *dnp, dt_irlist_t *dlp, dt_regset_t *drp)
> > +{
> > +	dt_node_t	*af = dnp->dn_args;
> > +	dt_node_t	*addr = af->dn_list;
> > +	dt_node_t	*tnp, *cnp, *lnp, *rnp, *anp, *xnp;;
> 
> s/;;/;/

Thanks.

> > [...]
> > +	xnp = dt_node_type(ddp);		/* frees ddp */
> > +	/* Create a node to represent: (type)addr */
> > +	xnp = dt_node_op2(DT_TOK_LPAR, xnp, addr);
> > [...]
> > +	xnp = dt_node_type(ddp);		/* frees ddp */
> > +
> > +	xnp = dt_node_op2(DT_TOK_LPAR, xnp, addr); /* (type)addr */
> 
> Might as well make those two sections of code as boringly similar as
> possible, comments and all.

Oops, meant to do that.  Thanks.