On Fri, Apr 20, 2018 at 02:42:37PM +0200, Julia Lawall wrote:
>
>
> On Fri, 20 Apr 2018, David Gibson wrote:
>
> > On Fri, Feb 02, 2018 at 09:41:49PM +0100, Julia Lawall wrote:
> > > The parser is extended to record positions, in build_node,
> > > build_node_delete, and build_property.
> > >
> > > srcpos structures are added to the property and node types, and to the
> > > parameter lists of the above functions that construct these types.
> > > Nodes and properties that are created by the compiler rather than from
> > > parsing source code have NULL as the srcpos value.
> > >
> > > merge_nodes, defined in livetree.c uses srcpos_extend to combine
> > > multiple positions, resulting in a list of positions. srcpos_extend is
> > > defined in srcpos.c and removes duplicates. As srcpos_extend is
> > > performing a union, it is O(n^2). The recursion serves to copy the new
> > > list and then add it to the front of the old one. Using recursion keeps
> > > the positions in order from newer to older. On each recursive call, the
> > > current element is compared against each element in the proposed new
> > > tail, to avoid duplicates. Since the list of positions is only extended
> > > in srcpos_extend, and srcpos_extend does not produce lists with
> > > duplicates, it is not necessary to check for the current element in the
> > > tail of its own list.
> > >
> > > Another change to srcpos.c is to make srcpos_copy always do a full copy,
> > > including a copy of the file substructure. This is required because
> > > when dtc is used on the output of cpp, the successive detected file
> > > names overwrite the file name in the file structure. The next field
> > > does not need to be deep copied, because it is only updated in newly
> > > copied positions and the positions to which it points have also been
> > > copied. File names are only updated in uncopied position structures.
> > >
> > > Signed-off-by: Julia Lawall <Julia.Lawall@xxxxxxx>
> > > Signed-off-by: Frank Rowand <frank.rowand@xxxxxxxx>
> >
> > Sorry I've taken so long to look at this properly.
> >
> > This looks pretty good overall. Couple of nits noted below..
> >
> > [snip]
> > > diff --git a/srcpos.c b/srcpos.c
> > > index 9dd3297..c6688e7 100644
> > > --- a/srcpos.c
> > > +++ b/srcpos.c
> > > @@ -207,6 +207,7 @@ struct srcpos srcpos_empty = {
> > > .last_line = 0,
> > > .last_column = 0,
> > > .file = NULL,
> > > + .next = NULL,
> > > };
> > >
> > > void srcpos_update(struct srcpos *pos, const char *text, int len)
> > > @@ -234,13 +235,52 @@ struct srcpos *
> > > srcpos_copy(struct srcpos *pos)
> > > {
> > > struct srcpos *pos_new;
> > > + struct srcfile_state *srcfile_state;
> > > +
> > > + if (!pos)
> > > + return NULL;
> > >
> > > pos_new = xmalloc(sizeof(struct srcpos));
> > > memcpy(pos_new, pos, sizeof(struct srcpos));
> > >
> > > + if (pos_new) {
> >
> > No need for the if - that's what xmalloc is for (plus we've already
> > written to the pointer in the memcpy().
>
> Thanks. I'll fix it up.
In fact you'd already fixed it in a later patch, so just need to move
the hunks about.
> > > + /* allocate without free */
> > > + srcfile_state = xmalloc(sizeof(struct srcfile_state));
> > > + memcpy(srcfile_state, pos->file, sizeof(struct srcfile_state));
> > > +
> > > + pos_new->file = srcfile_state;
> > > + }
> > > +
> > > return pos_new;
> > > }
> > >
> > > +static bool same_pos(struct srcpos *p1, struct srcpos *p2)
> > > +{
> > > + return (p1->first_line == p2->first_line &&
> > > + p1->first_column == p2->first_column &&
> > > + p1->last_line == p2->last_line &&
> > > + p1->last_column == p2->last_column &&
> > > + !strcmp(p1->file->name, p2->file->name));
> > > +}
> > > +
> > > +struct srcpos *srcpos_extend(struct srcpos *pos, struct srcpos *newtail)
> > > +{
> > > + struct srcpos *next, *p;
> > > +
> > > + if (!pos)
> > > + return newtail;
> > > +
> > > + next = srcpos_extend(pos->next, newtail);
> > > +
> > > + for (p = newtail; p != NULL; p = p->next)
> > > + if (same_pos(pos, p))
> > > + return next;
> > > +
> > > + pos = srcpos_copy(pos);
> > > + pos->next = next;
> > > + return pos;
> >
> > This copies pos and doesn't free the original - and the caller doesn't
> > free the original either, so there's a leak here.
>
> I suspect that we can't free it, because someone else may be using it, but
> I will check on it.
Hrm. AIUI the copies of the srcpos stored in nodes/properties are
owned by those nodes and properties and shoulnd't be shared with
anything. And the only caller of srcpos_extend() is, I believe,
merging two nodes, so the input nodes should be logically discarded.
So I think it should be ok.
But, yeah, lifetime management is not very easy to follow in dtc, so
do check.
--
David Gibson | I'll have my music baroque, and my code
david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_
| _way_ _around_!
http://www.ozlabs.org/~dgibson
Attachment:
signature.asc
Description: PGP signature
