[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <CAGnkfhz+p1o_yHxk2jkY9ggNwLSO-Jk4BcxPuWhSHw1YXoJsSw@mail.gmail.com>
Date: Wed, 10 Jul 2019 01:18:55 +0200
From: Matteo Croce <mcroce@...hat.com>
To: Stephen Hemminger <stephen@...workplumber.org>
Cc: netdev <netdev@...r.kernel.org>, David Ahern <dsahern@...nel.org>
Subject: Re: [PATCH iproute2] utils: don't match empty strings as prefixes
On Tue, Jul 9, 2019 at 11:38 PM Stephen Hemminger
<stephen@...workplumber.org> wrote:
>
> On Tue, 9 Jul 2019 22:40:40 +0200
> Matteo Croce <mcroce@...hat.com> wrote:
>
> > iproute has an utility function which checks if a string is a prefix for
> > another one, to allow use of abbreviated commands, e.g. 'addr' or 'a'
> > instead of 'address'.
> >
> > This routine unfortunately considers an empty string as prefix
> > of any pattern, leading to undefined behaviour when an empty
> > argument is passed to ip:
> >
> > # ip ''
> > 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 65536 qdisc noqueue state UNKNOWN group default qlen 1000
> > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00
> > inet 127.0.0.1/8 scope host lo
> > valid_lft forever preferred_lft forever
> > inet6 ::1/128 scope host
> > valid_lft forever preferred_lft forever
> >
> > # tc ''
> > qdisc noqueue 0: dev lo root refcnt 2
> >
> > # ip address add 192.0.2.0/24 '' 198.51.100.1 dev dummy0
> > # ip addr show dev dummy0
> > 6: dummy0: <BROADCAST,NOARP> mtu 1500 qdisc noop state DOWN group default qlen 1000
> > link/ether 02:9d:5e:e9:3f:c0 brd ff:ff:ff:ff:ff:ff
> > inet 192.0.2.0/24 brd 198.51.100.1 scope global dummy0
> > valid_lft forever preferred_lft forever
> >
> > Rewrite matches() so it takes care of an empty input, and doesn't
> > scan the input strings three times: the actual implementation
> > does 2 strlen and a memcpy to accomplish the same task.
> >
> > Signed-off-by: Matteo Croce <mcroce@...hat.com>
> > ---
> > include/utils.h | 2 +-
> > lib/utils.c | 14 +++++++++-----
> > 2 files changed, 10 insertions(+), 6 deletions(-)
> >
> > diff --git a/include/utils.h b/include/utils.h
> > index 927fdc17..f4d12abb 100644
> > --- a/include/utils.h
> > +++ b/include/utils.h
> > @@ -198,7 +198,7 @@ int nodev(const char *dev);
> > int check_ifname(const char *);
> > int get_ifname(char *, const char *);
> > const char *get_ifname_rta(int ifindex, const struct rtattr *rta);
> > -int matches(const char *arg, const char *pattern);
> > +int matches(const char *prefix, const char *string);
> > int inet_addr_match(const inet_prefix *a, const inet_prefix *b, int bits);
> > int inet_addr_match_rta(const inet_prefix *m, const struct rtattr *rta);
> >
> > diff --git a/lib/utils.c b/lib/utils.c
> > index be0f11b0..73ce19bb 100644
> > --- a/lib/utils.c
> > +++ b/lib/utils.c
> > @@ -887,13 +887,17 @@ const char *get_ifname_rta(int ifindex, const struct rtattr *rta)
> > return name;
> > }
> >
> > -int matches(const char *cmd, const char *pattern)
> > +/* Check if 'prefix' is a non empty prefix of 'string' */
> > +int matches(const char *prefix, const char *string)
> > {
> > - int len = strlen(cmd);
> > + if (!*prefix)
> > + return 1;
> > + while(*string && *prefix == *string) {
> > + prefix++;
> > + string++;
> > + }
> >
> > - if (len > strlen(pattern))
> > - return -1;
> > - return memcmp(pattern, cmd, len);
> > + return *prefix;
> > }
> >
> > int inet_addr_match(const inet_prefix *a, const inet_prefix *b, int bits)
>
> ERROR: space required before the open parenthesis '('
> #134: FILE: lib/utils.c:895:
> + while(*string && *prefix == *string) {
>
> total: 1 errors, 1 warnings, 30 lines checked
>
> The empty prefix string is a bug and should not be allowed.
> Also return value should be same as old code (yours isn't).
>
>
>
The old return value was the difference between the first pair of
bytes, according to the memcmp manpage.
All calls only checks if the matches() return value is 0 or not 0:
iproute2$ git grep 'matches(' |grep -v -e '== 0' -e '= 0' -e '!matches('
include/utils.h:int matches(const char *prefix, const char *string);
include/xtables.h:extern void xtables_register_matches(struct
xtables_match *, unsigned int);
lib/color.c: if (matches(dup, "-color"))
lib/utils.c:int matches(const char *prefix, const char *string)
tc/tc.c: if (matches(argv[0], iter->c))
Is it a problem if it returns a non negative value for non matching strings?
Regards,
--
Matteo Croce
per aspera ad upstream
Powered by blists - more mailing lists