| Message ID | 20211011142416.26798-1-fw@strlen.de (mailing list archive) |
|---|---|
| State | Changes Requested |
| Delegated to: | Paul Moore |
| Headers | show |
| Series | [v2] selinux: remove unneeded ipv6 hook wrappers | expand |
On Mon, Oct 11, 2021 at 10:25 AM Florian Westphal <fw@strlen.de> wrote: > > Netfilter places the protocol number the hook function is getting called > from in state->pf, so we can use that instead of an extra wrapper. > > Signed-off-by: Florian Westphal <fw@strlen.de> > --- > v2: add back '#endif /* CONFIG_NETFILTER */' erronously axed in v1. > Applies to 'next' branch of > https://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git/ > > security/selinux/hooks.c | 52 ++++++++++------------------------------ > 1 file changed, 12 insertions(+), 40 deletions(-) ... > diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c > index e7ebd45ca345..831b857d5dd7 100644 > --- a/security/selinux/hooks.c > +++ b/security/selinux/hooks.c > @@ -7470,38 +7442,38 @@ DEFINE_LSM(selinux) = { > > static const struct nf_hook_ops selinux_nf_ops[] = { > { > - .hook = selinux_ipv4_postroute, > + .hook = selinux_hook_postroute, > .pf = NFPROTO_IPV4, > .hooknum = NF_INET_POST_ROUTING, > .priority = NF_IP_PRI_SELINUX_LAST, > }, Thanks for the patch Florian, although the name "selinux_hook_*" seems a bit ambiguous to me, after all we have a little more than 200 "hooks" in the SELinux LSM implementation. Would you be okay with calling the netfilter hook functions "selinux_nf_*" or something similar? If you don't have time I can do the rename during the merge assuming we can all agree on a name.
On Mon, Oct 11, 2021 at 7:10 PM Paul Moore <paul@paul-moore.com> wrote: > On Mon, Oct 11, 2021 at 10:25 AM Florian Westphal <fw@strlen.de> wrote: > > > > Netfilter places the protocol number the hook function is getting called > > from in state->pf, so we can use that instead of an extra wrapper. > > > > Signed-off-by: Florian Westphal <fw@strlen.de> > > --- > > v2: add back '#endif /* CONFIG_NETFILTER */' erronously axed in v1. > > Applies to 'next' branch of > > https://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git/ > > > > security/selinux/hooks.c | 52 ++++++++++------------------------------ > > 1 file changed, 12 insertions(+), 40 deletions(-) > > ... > > > diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c > > index e7ebd45ca345..831b857d5dd7 100644 > > --- a/security/selinux/hooks.c > > +++ b/security/selinux/hooks.c > > @@ -7470,38 +7442,38 @@ DEFINE_LSM(selinux) = { > > > > static const struct nf_hook_ops selinux_nf_ops[] = { > > { > > - .hook = selinux_ipv4_postroute, > > + .hook = selinux_hook_postroute, > > .pf = NFPROTO_IPV4, > > .hooknum = NF_INET_POST_ROUTING, > > .priority = NF_IP_PRI_SELINUX_LAST, > > }, > > Thanks for the patch Florian, although the name "selinux_hook_*" seems > a bit ambiguous to me, after all we have a little more than 200 > "hooks" in the SELinux LSM implementation. Would you be okay with > calling the netfilter hook functions "selinux_nf_*" or something > similar? If you don't have time I can do the rename during the merge > assuming we can all agree on a name. Since selinux_ip_forward() and selinux_ip_postroute() are used only in the hook functions, how about changing their signature and using them as hooks directly? That would solve the naming and also remove a few extra lines of boilerplate.
On Mon, Oct 11, 2021 at 2:21 PM Ondrej Mosnacek <omosnace@redhat.com> wrote: > > On Mon, Oct 11, 2021 at 7:10 PM Paul Moore <paul@paul-moore.com> wrote: > > On Mon, Oct 11, 2021 at 10:25 AM Florian Westphal <fw@strlen.de> wrote: > > > > > > Netfilter places the protocol number the hook function is getting called > > > from in state->pf, so we can use that instead of an extra wrapper. > > > > > > Signed-off-by: Florian Westphal <fw@strlen.de> > > > --- > > > v2: add back '#endif /* CONFIG_NETFILTER */' erronously axed in v1. > > > Applies to 'next' branch of > > > https://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git/ > > > > > > security/selinux/hooks.c | 52 ++++++++++------------------------------ > > > 1 file changed, 12 insertions(+), 40 deletions(-) > > > > ... > > > > > diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c > > > index e7ebd45ca345..831b857d5dd7 100644 > > > --- a/security/selinux/hooks.c > > > +++ b/security/selinux/hooks.c > > > @@ -7470,38 +7442,38 @@ DEFINE_LSM(selinux) = { > > > > > > static const struct nf_hook_ops selinux_nf_ops[] = { > > > { > > > - .hook = selinux_ipv4_postroute, > > > + .hook = selinux_hook_postroute, > > > .pf = NFPROTO_IPV4, > > > .hooknum = NF_INET_POST_ROUTING, > > > .priority = NF_IP_PRI_SELINUX_LAST, > > > }, > > > > Thanks for the patch Florian, although the name "selinux_hook_*" seems > > a bit ambiguous to me, after all we have a little more than 200 > > "hooks" in the SELinux LSM implementation. Would you be okay with > > calling the netfilter hook functions "selinux_nf_*" or something > > similar? If you don't have time I can do the rename during the merge > > assuming we can all agree on a name. > > Since selinux_ip_forward() and selinux_ip_postroute() are used only in > the hook functions, how about changing their signature and using them > as hooks directly? That would solve the naming and also remove a few > extra lines of boilerplate. No argument against that from me, although you should be able to do the same for selinux_ip_output() as well unless I missed a caller.
Paul Moore <paul@paul-moore.com> wrote: > On Mon, Oct 11, 2021 at 2:21 PM Ondrej Mosnacek <omosnace@redhat.com> wrote: > > On Mon, Oct 11, 2021 at 7:10 PM Paul Moore <paul@paul-moore.com> wrote: > > > On Mon, Oct 11, 2021 at 10:25 AM Florian Westphal <fw@strlen.de> wrote: > > > > Netfilter places the protocol number the hook function is getting called > > > > from in state->pf, so we can use that instead of an extra wrapper. > > > > > > > > Signed-off-by: Florian Westphal <fw@strlen.de> > > > > --- > > > > v2: add back '#endif /* CONFIG_NETFILTER */' erronously axed in v1. > > > > Applies to 'next' branch of > > > > https://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git/ > > > > > > > > security/selinux/hooks.c | 52 ++++++++++------------------------------ > > > > 1 file changed, 12 insertions(+), 40 deletions(-) > > > > > > ... > > > > > > > diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c > > > > index e7ebd45ca345..831b857d5dd7 100644 > > > > --- a/security/selinux/hooks.c > > > > +++ b/security/selinux/hooks.c > > > > @@ -7470,38 +7442,38 @@ DEFINE_LSM(selinux) = { > > > > > > > > static const struct nf_hook_ops selinux_nf_ops[] = { > > > > { > > > > - .hook = selinux_ipv4_postroute, > > > > + .hook = selinux_hook_postroute, > > > > .pf = NFPROTO_IPV4, > > > > .hooknum = NF_INET_POST_ROUTING, > > > > .priority = NF_IP_PRI_SELINUX_LAST, > > > > }, > > > > > > Thanks for the patch Florian, although the name "selinux_hook_*" seems > > > a bit ambiguous to me, after all we have a little more than 200 > > > "hooks" in the SELinux LSM implementation. Would you be okay with > > > calling the netfilter hook functions "selinux_nf_*" or something > > > similar? Absolutely. > > > If you don't have time I can do the rename during the merge > > > assuming we can all agree on a name. I'll submit a v3. > > Since selinux_ip_forward() and selinux_ip_postroute() are used only in > > the hook functions, how about changing their signature and using them > > as hooks directly? That would solve the naming and also remove a few > > extra lines of boilerplate. > > No argument against that from me, although you should be able to do > the same for selinux_ip_output() as well unless I missed a caller. I'll have a look, thanks for the pointers.
diff --git a/security/selinux/hooks.c b/security/selinux/hooks.c index e7ebd45ca345..831b857d5dd7 100644 --- a/security/selinux/hooks.c +++ b/security/selinux/hooks.c @@ -5746,22 +5746,13 @@ static unsigned int selinux_ip_forward(struct sk_buff *skb, return NF_ACCEPT; } -static unsigned int selinux_ipv4_forward(void *priv, +static unsigned int selinux_hook_forward(void *priv, struct sk_buff *skb, const struct nf_hook_state *state) { - return selinux_ip_forward(skb, state->in, PF_INET); + return selinux_ip_forward(skb, state->in, state->pf); } -#if IS_ENABLED(CONFIG_IPV6) -static unsigned int selinux_ipv6_forward(void *priv, - struct sk_buff *skb, - const struct nf_hook_state *state) -{ - return selinux_ip_forward(skb, state->in, PF_INET6); -} -#endif /* IPV6 */ - static unsigned int selinux_ip_output(struct sk_buff *skb, u16 family) { @@ -5804,21 +5795,12 @@ static unsigned int selinux_ip_output(struct sk_buff *skb, return NF_ACCEPT; } -static unsigned int selinux_ipv4_output(void *priv, - struct sk_buff *skb, - const struct nf_hook_state *state) -{ - return selinux_ip_output(skb, PF_INET); -} - -#if IS_ENABLED(CONFIG_IPV6) -static unsigned int selinux_ipv6_output(void *priv, +static unsigned int selinux_hook_output(void *priv, struct sk_buff *skb, const struct nf_hook_state *state) { - return selinux_ip_output(skb, PF_INET6); + return selinux_ip_output(skb, state->pf); } -#endif /* IPV6 */ static unsigned int selinux_ip_postroute_compat(struct sk_buff *skb, int ifindex, @@ -5994,22 +5976,12 @@ static unsigned int selinux_ip_postroute(struct sk_buff *skb, return NF_ACCEPT; } -static unsigned int selinux_ipv4_postroute(void *priv, +static unsigned int selinux_hook_postroute(void *priv, struct sk_buff *skb, const struct nf_hook_state *state) { - return selinux_ip_postroute(skb, state->out, PF_INET); + return selinux_ip_postroute(skb, state->out, state->pf); } - -#if IS_ENABLED(CONFIG_IPV6) -static unsigned int selinux_ipv6_postroute(void *priv, - struct sk_buff *skb, - const struct nf_hook_state *state) -{ - return selinux_ip_postroute(skb, state->out, PF_INET6); -} -#endif /* IPV6 */ - #endif /* CONFIG_NETFILTER */ static int selinux_netlink_send(struct sock *sk, struct sk_buff *skb) @@ -7470,38 +7442,38 @@ DEFINE_LSM(selinux) = { static const struct nf_hook_ops selinux_nf_ops[] = { { - .hook = selinux_ipv4_postroute, + .hook = selinux_hook_postroute, .pf = NFPROTO_IPV4, .hooknum = NF_INET_POST_ROUTING, .priority = NF_IP_PRI_SELINUX_LAST, }, { - .hook = selinux_ipv4_forward, + .hook = selinux_hook_forward, .pf = NFPROTO_IPV4, .hooknum = NF_INET_FORWARD, .priority = NF_IP_PRI_SELINUX_FIRST, }, { - .hook = selinux_ipv4_output, + .hook = selinux_hook_output, .pf = NFPROTO_IPV4, .hooknum = NF_INET_LOCAL_OUT, .priority = NF_IP_PRI_SELINUX_FIRST, }, #if IS_ENABLED(CONFIG_IPV6) { - .hook = selinux_ipv6_postroute, + .hook = selinux_hook_postroute, .pf = NFPROTO_IPV6, .hooknum = NF_INET_POST_ROUTING, .priority = NF_IP6_PRI_SELINUX_LAST, }, { - .hook = selinux_ipv6_forward, + .hook = selinux_hook_forward, .pf = NFPROTO_IPV6, .hooknum = NF_INET_FORWARD, .priority = NF_IP6_PRI_SELINUX_FIRST, }, { - .hook = selinux_ipv6_output, + .hook = selinux_hook_output, .pf = NFPROTO_IPV6, .hooknum = NF_INET_LOCAL_OUT, .priority = NF_IP6_PRI_SELINUX_FIRST,
Netfilter places the protocol number the hook function is getting called from in state->pf, so we can use that instead of an extra wrapper. Signed-off-by: Florian Westphal <fw@strlen.de> --- v2: add back '#endif /* CONFIG_NETFILTER */' erronously axed in v1. Applies to 'next' branch of https://git.kernel.org/pub/scm/linux/kernel/git/pcmoore/selinux.git/ security/selinux/hooks.c | 52 ++++++++++------------------------------ 1 file changed, 12 insertions(+), 40 deletions(-)