Re: [Cocci] [PATCH] coccinelle: memdup.cocci: fix matching rules

From: Julia Lawall
Date: Thu Aug 06 2015 - 09:58:43 EST




On Thu, 6 Aug 2015, Nicholas Mc Guire wrote:

> On Thu, 06 Aug 2015, Andrzej Hajda wrote:
>
> > This patch fixes three things, listed in order of importance.
> > 1. Removes matching of kmemdup from !patch rule - it is incorrect and
> > in fact makes report mode unusable.
> > 2. Adds unlikely to if clause. It allows to match more cases - the ones with
> > unlikely and the ones without it.
> > 3. Fixes report message.
> >
> > Signed-off-by: Andrzej Hajda <a.hajda@xxxxxxxxxxx>
> > ---
> > Hi Julia,
> >
> > I guess 1st and 3rd changes are OK. I am not sure about 2nd change, without
> > it I was not able to catch cases containing unlikely macro. For example
> > fs/ntfs/dir.c:1175:
> > ir = kmalloc(rc, GFP_NOFS);
> > if (unlikely(!ir)) {
> > err = -ENOMEM;
> > goto err_out;
> > }
> > /* Copy the index root value (it has been verified in read_inode). */
> > memcpy(ir, (u8*)ctx->attr +
> > le16_to_cpu(ctx->attr->data.resident.value_offset), rc);
> >
> > It seems quite strange for me, as these rules looks to me isomorphic.
> > Is this expected behavior of coccinelle or just some bug?
> >
> > After this fix, cocci finds 46 places to patch, I will send patchset if this
> > change looks OK to you.
> >
> > I have used:
> > spatch version 1.0.1 with Python support and with PCRE support
> > latest linux-next.
> >
> > Regards
> > Andrzej
> > ---
> > scripts/coccinelle/api/memdup.cocci | 9 ++++-----
> > 1 file changed, 4 insertions(+), 5 deletions(-)
> >
> > diff --git a/scripts/coccinelle/api/memdup.cocci b/scripts/coccinelle/api/memdup.cocci
> > index 3d1aa71..2297205 100644
> > --- a/scripts/coccinelle/api/memdup.cocci
> > +++ b/scripts/coccinelle/api/memdup.cocci
> > @@ -39,7 +39,7 @@ statement S;
> >
> > - to = \(kmalloc@p\|kzalloc@p\)(size,flag);
> > + to = kmemdup(from,size,flag);
> > - if (to==NULL || ...) S
> > + if (unlikely(to==NULL) || ...) S
> > - memcpy(to, from, size);
>
> maybe I don't understand isomorphism declarations properly but
> that should actually be covered by standard.iso:
>
> @ unlikely @
> expression E;
> @@
> unlikely(E) <=> likely(E) => E
>
> ??
> so not clear why you would get more cases with this patch.

You have to follow the arrowheads. There is no path from E to
unlikely(E). So adding unlikely catches both the case where it is present
and where it is not.

julia

>
> >
> > @r depends on !patch@
> > @@ -49,18 +49,17 @@ statement S;
> > @@
> >
> > * to = \(kmalloc@p\|kzalloc@p\)(size,flag);
> > - to = kmemdup(from,size,flag);
> > - if (to==NULL || ...) S
> > + if (unlikely(to==NULL) || ...) S
> > * memcpy(to, from, size);
> >
> > @script:python depends on org@
> > p << r.p;
> > @@
> >
> > -coccilib.org.print_todo(p[0], "WARNING opportunity for kmemdep")
> > +coccilib.org.print_todo(p[0], "WARNING opportunity for kmemdup")
> >
> > @script:python depends on report@
> > p << r.p;
> > @@
> >
> > -coccilib.report.print_report(p[0], "WARNING opportunity for kmemdep")
> > +coccilib.report.print_report(p[0], "WARNING opportunity for kmemdup")
> > --
> > 1.9.1
> >
> > _______________________________________________
> > Cocci mailing list
> > Cocci@xxxxxxxxxxxxxxx
> > https://systeme.lip6.fr/mailman/listinfo/cocci
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/