Re: [PATCH v2] lib: parser: Fix match_wildcard to correctly handle trailing stars

[Inseob Kim]

On Thu, Mar 26, 2026 at 1:45 PM Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
>
> On Thu, 26 Mar 2026 13:12:26 +0900 Inseob Kim <inseob@xxxxxxxxxx> wrote:
>
> > On Thu, Mar 26, 2026 at 11:25 AM Andrew Morton
> > <akpm@xxxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > > On Thu, 26 Mar 2026 11:06:04 +0900 Inseob Kim <inseob@xxxxxxxxxx> wrote:
> > >
> > > > This fixes a bug of match_wildcard that incorrectly handles trailing
> > > > asterisks. For example, `match_wildcard("abc**", "abc")` must return
> > > > true, but it returns false.
> > > >
> > > > Signed-off-by: Inseob Kim <inseob@xxxxxxxxxx>
> > > > ---
> > > > v2:
> > > >   - Added Cc. No changes to the code.
> > > > ---
> > > >  lib/parser.c | 2 +-
> > > >  1 file changed, 1 insertion(+), 1 deletion(-)
> > > >
> > > > diff --git a/lib/parser.c b/lib/parser.c
> > > > index 73e8f8e5be73..62da0ac0d438 100644
> > > > --- a/lib/parser.c
> > > > +++ b/lib/parser.c
> > > > @@ -315,7 +315,7 @@ bool match_wildcard(const char *pattern, const char *str)
> > > >               }
> > > >       }
> > > >
> > > > -     if (*p == '*')
> > > > +     while (*p == '*')
> > > >               ++p;
> > > >       return !*p;
> > > >  }
> > >
> > > Thanks, looks right.
> > >
> > > We don't appear to have any selftesting for this code.
> >
> > Would you prefer test cases for match_wildcard?
>
> That would of course be wonderful, but such a contribution is unrelated
> to this bugfix.  Up to you.
>
> > >
> > > Should all of parser.c actually exist?  Some of it is a subset of
> > > lib/glob.c?
> >
> > Wildcard is definitely a subset of glob, but we're intentionally using
> > wildcard for genfscon for example
> > (https://lore.kernel.org/selinux/20250318083139.1515253-1-takayas@xxxxxxxxxxxx/).
> > I'd like to leave the parser.c as is.
>
> That's a nice boot-time improvement, but I don't understand from that
> why wildcard is preferable to glob?


Sorry for the confusion; let me correct myself. It's not that wildcard
is preferable to glob, but genfs_seclabel_wildcard is already merged
and in use so moving from wildcard to glob would introduce breaking
changes. Let me invite Takaya who possibly has more rationale for the
wildcard.