Re: [PATCH] bridge: netfilter: spanning tree: Add masked_ether_addr_equal and neatening

[Pablo Neira Ayuso]

On Fri, Jun 24, 2016 at 10:51:28AM +0200, Pablo Neira Ayuso wrote:
> On Thu, Jun 23, 2016 at 12:00:00PM -0700, Joe Perches wrote:
> > On Thu, 2016-06-23 at 19:36 +0200, Pablo Neira Ayuso wrote:
> > > On Wed, Jun 15, 2016 at 01:58:45PM -0700, Joe Perches wrote:
> > > > 
> > > > There is code duplication of a masked ethernet address comparison here
> > > > so make it a separate function instead.
> > > > 
> > > > Miscellanea:
> > > > 
> > > > o Neaten alignment of FWINV macro uses to make it clearer for the reader
> > > Applied, thanks.
> > > 
> > > > 
> > > > Signed-off-by: Joe Perches 
> > > > ---
> > > > 
> > > > This masked_ether_addr_equal function could go into etherdevice.h,
> > > > but I don't see another use like it in kernel code.  Is there one?
> > >
> > > This is specific of iptables, not even nftables would use this. So I
> > > would keep this in the iptables tree.
> > 
> > Did you see the other patch that adds a generic
> > ether_addr_equal_masked() and uses it in a few
> > more files?
> 
> You mean this one:
> 
> http://patchwork.ozlabs.org/patch/636208/
> 
> OK, so I'll toss the previous and will take this one instead.
> 
> As I said my opinion is that ether_addr_equal_masked() is only
> required by netfilter, but thinking it well I don't really mind in
> what header this function is placed given that these are our internal
> headers.

git am reports patch I get from patchwork is corrupt at line 37.
Tried a couple of tricks to fix it but this didn't work.

Would you mind resubmitting this patch?

Sorry for the inconvenience.

Re: [PATCH] bridge: netfilter: spanning tree: Add masked_ether_addr_equal and neatening

[Pablo Neira Ayuso]

On Thu, Jun 23, 2016 at 12:00:00PM -0700, Joe Perches wrote:
> On Thu, 2016-06-23 at 19:36 +0200, Pablo Neira Ayuso wrote:
> > On Wed, Jun 15, 2016 at 01:58:45PM -0700, Joe Perches wrote:
> > > 
> > > There is code duplication of a masked ethernet address comparison here
> > > so make it a separate function instead.
> > > 
> > > Miscellanea:
> > > 
> > > o Neaten alignment of FWINV macro uses to make it clearer for the reader
> > Applied, thanks.
> > 
> > > 
> > > Signed-off-by: Joe Perches 
> > > ---
> > > 
> > > This masked_ether_addr_equal function could go into etherdevice.h,
> > > but I don't see another use like it in kernel code.  Is there one?
> >
> > This is specific of iptables, not even nftables would use this. So I
> > would keep this in the iptables tree.
> 
> Did you see the other patch that adds a generic
> ether_addr_equal_masked() and uses it in a few
> more files?

You mean this one:

http://patchwork.ozlabs.org/patch/636208/

OK, so I'll toss the previous and will take this one instead.

As I said my opinion is that ether_addr_equal_masked() is only
required by netfilter, but thinking it well I don't really mind in
what header this function is placed given that these are our internal
headers.

Thanks.

Re: [PATCH] bridge: netfilter: spanning tree: Add masked_ether_addr_equal and neatening

[Joe Perches]

On Thu, 2016-06-23 at 19:36 +0200, Pablo Neira Ayuso wrote:
> On Wed, Jun 15, 2016 at 01:58:45PM -0700, Joe Perches wrote:
> > 
> > There is code duplication of a masked ethernet address comparison here
> > so make it a separate function instead.
> > 
> > Miscellanea:
> > 
> > o Neaten alignment of FWINV macro uses to make it clearer for the reader
> Applied, thanks.
> 
> > 
> > Signed-off-by: Joe Perches 
> > ---
> > 
> > This masked_ether_addr_equal function could go into etherdevice.h,
> > but I don't see another use like it in kernel code.  Is there one?
> This is specific of iptables, not even nftables would use this. So I
> would keep this in the iptables tree.

Did you see the other patch that adds a generic
ether_addr_equal_masked() and uses it in a few
more files?

Re: [PATCH] bridge: netfilter: spanning tree: Add masked_ether_addr_equal and neatening

[Pablo Neira Ayuso]

On Wed, Jun 15, 2016 at 01:58:45PM -0700, Joe Perches wrote:
> There is code duplication of a masked ethernet address comparison here
> so make it a separate function instead.
> 
> Miscellanea:
> 
> o Neaten alignment of FWINV macro uses to make it clearer for the reader

Applied, thanks.

> Signed-off-by: Joe Perches 
> ---
> 
> This masked_ether_addr_equal function could go into etherdevice.h,
> but I don't see another use like it in kernel code.  Is there one?

This is specific of iptables, not even nftables would use this. So I
would keep this in the iptables tree.

Re: [PATCH] bridge: netfilter: spanning tree: Add masked_ether_addr_equal and neatening

[Joe Perches]

On Wed, 2016-06-15 at 13:58 -0700, Joe Perches wrote:
> There is code duplication of a masked ethernet address comparison here
> so make it a separate function instead.
> 
> Miscellanea:
> 
> o Neaten alignment of FWINV macro uses to make it clearer for the reader
> 
> Signed-off-by: Joe Perches 
> ---
> 
> This masked_ether_addr_equal function could go into etherdevice.h,
> but I don't see another use like it in kernel code.  Is there one?

Turns out there are at least a few more uses in bridge/netfilter

 net/bridge/netfilter/ebt_arp.c
 net/bridge/netfilter/ebtables.c

Maybe this?
---
>From 770261c682a745b8de663a5756a66cd00bb5b79b Mon Sep 17 00:00:00 2001
Message-Id: 
<770261c682a745b8de663a5756a66cd00bb5b79b.1466056695.git@perches.com>
From: Joe Perches 
Date: Wed, 15 Jun 2016 13:45:54 -0700
Subject: [PATCH] etherdevice.h & bridge: netfilter: Add and use
 ether_addr_equal_masked

There are code duplications of a masked ethernet address comparison here
so make it a separate function instead.

Miscellanea:

o Neaten alignment of FWINV macro uses to make it clearer for the reader

Signed-off-by: Joe Perches 
---
 include/linux/etherdevice.h | 22 ++
 net/bridge/netfilter/ebt_arp.c  | 17 +-
 net/bridge/netfilter/ebt_stp.c  | 49 ++---
 net/bridge/netfilter/ebtables.c | 17 +-
 4 files changed, 56 insertions(+), 49 deletions(-)

diff --git a/include/linux/etherdevice.h b/include/linux/etherdevice.h
index 37ff4a6..942a24c 100644
--- a/include/linux/etherdevice.h
+++ b/include/linux/etherdevice.h
@@ -374,6 +374,28 @@ static inline bool ether_addr_equal_unaligned(const u8 
*addr1, const u8 *addr2)
 }
 
 /**
+ * ether_addr_equal_masked - Compare two Ethernet addresses with a mask
+ * @addr1: Pointer to a six-byte array containing the 1st Ethernet address
+ * @addr2: Pointer to a six-byte array containing the 2nd Ethernet address
+ * @mask: Pointer to a six-byte array containing the Ethernet address bitmask
+ *
+ * Compare two Ethernet addresses with a mask, returns true if for every bit
+ * set in the bitmask the equivalent bits in the ethernet addresses are equal.
+ */
+static inline bool ether_addr_equal_masked(const u8 *addr1, const u8 *addr2,
+      const u8 *mask)
+{
+   int i;
+
+   for (i = 0; i < ETH_ALEN; i++) {
+   if ((addr1[i] ^ addr2[i]) & mask[i])
+   return false;
+   }
+
+   return true;
+}
+
+/**
  * is_etherdev_addr - Tell if given Ethernet address belongs to the device.
  * @dev: Pointer to a device structure
  * @addr: Pointer to a six-byte array containing the Ethernet address
diff --git a/net/bridge/netfilter/ebt_arp.c b/net/bridge/netfilter/ebt_arp.c
index cd457b8..cca0a89 100644
--- a/net/bridge/netfilter/ebt_arp.c
+++ b/net/bridge/netfilter/ebt_arp.c
@@ -65,7 +65,6 @@ ebt_arp_mt(const struct sk_buff *skb, struct xt_action_param 
*par)
    if (info->bitmask & (EBT_ARP_SRC_MAC | EBT_ARP_DST_MAC)) {
    const unsigned char *mp;
    unsigned char _mac[ETH_ALEN];
-   uint8_t verdict, i;
 
    if (ah->ar_hln != ETH_ALEN || ah->ar_hrd != htons(ARPHRD_ETHER))
    return false;
@@ -74,11 +73,9 @@ ebt_arp_mt(const struct sk_buff *skb, struct xt_action_param 
*par)
    sizeof(_mac), &_mac);
    if (mp == NULL)
    return false;
-   verdict = 0;
-   for (i = 0; i < 6; i++)
-   verdict |= (mp[i] ^ info->smaddr[i]) &
-      info->smmsk[i];
-   if (FWINV(verdict != 0, EBT_ARP_SRC_MAC))
+   if (FWINV(!ether_addr_equal_masked(mp, info->smaddr,
+      info->smmsk),
+     EBT_ARP_SRC_MAC))
    return false;
    }
 
@@ -88,11 +85,9 @@ ebt_arp_mt(const struct sk_buff *skb, struct xt_action_param 
*par)
    sizeof(_mac), &_mac);
    if (mp == NULL)
    return false;
-   verdict = 0;
-   for (i = 0; i < 6; i++)
-   verdict |= (mp[i] ^ info->dmaddr[i]) &
-   info->dmmsk[i];
-   if (FWINV(verdict != 0, EBT_ARP_DST_MAC))
+   if (FWINV(!ether_addr_equal_masked(mp, info->dmaddr,
+      info->dmmsk),
+     EBT_ARP_DST_MAC))
    return false;
    }
    }
diff --git a/net/bridge/netfilter/ebt_stp.c b/net/bridge/netfilter/ebt_stp.c

[PATCH] bridge: netfilter: spanning tree: Add masked_ether_addr_equal and neatening

[Joe Perches]

There is code duplication of a masked ethernet address comparison here
so make it a separate function instead.

Miscellanea:

o Neaten alignment of FWINV macro uses to make it clearer for the reader

Signed-off-by: Joe Perches 
---

This masked_ether_addr_equal function could go into etherdevice.h,
but I don't see another use like it in kernel code.  Is there one?

 net/bridge/netfilter/ebt_stp.c | 62 --
 1 file changed, 35 insertions(+), 27 deletions(-)

diff --git a/net/bridge/netfilter/ebt_stp.c b/net/bridge/netfilter/ebt_stp.c
index e77f90b..46c3b5d 100644
--- a/net/bridge/netfilter/ebt_stp.c
+++ b/net/bridge/netfilter/ebt_stp.c
@@ -40,13 +40,25 @@ struct stp_config_pdu {
 #define NR16(p) (p[0] << 8 | p[1])
 #define NR32(p) ((p[0] << 24) | (p[1] << 16) | (p[2] << 8) | p[3])
 
+static bool masked_ether_addr_equal(const u8 *addr1, const u8 *addr2,
+   const u8 *mask)
+{
+   int i;
+
+   for (i = 0; i < ETH_ALEN; i++) {
+   if ((addr1[i] ^ addr2[i]) & mask[i])
+   return false;
+   }
+
+   return true;
+}
+
 static bool ebt_filter_config(const struct ebt_stp_info *info,
  const struct stp_config_pdu *stpc)
 {
const struct ebt_stp_config_info *c;
u16 v16;
u32 v32;
-   int verdict, i;
 
c = >config;
if ((info->bitmask & EBT_STP_FLAGS) &&
@@ -54,66 +66,62 @@ static bool ebt_filter_config(const struct ebt_stp_info 
*info,
return false;
if (info->bitmask & EBT_STP_ROOTPRIO) {
v16 = NR16(stpc->root);
-   if (FWINV(v16 < c->root_priol ||
-   v16 > c->root_priou, EBT_STP_ROOTPRIO))
+   if (FWINV(v16 < c->root_priol || v16 > c->root_priou,
+ EBT_STP_ROOTPRIO))
return false;
}
if (info->bitmask & EBT_STP_ROOTADDR) {
-   verdict = 0;
-   for (i = 0; i < 6; i++)
-   verdict |= (stpc->root[2+i] ^ c->root_addr[i]) &
-  c->root_addrmsk[i];
-   if (FWINV(verdict != 0, EBT_STP_ROOTADDR))
+   if (FWINV(!masked_ether_addr_equal(>root[2], c->root_addr,
+  c->root_addrmsk),
+ EBT_STP_ROOTADDR))
return false;
}
if (info->bitmask & EBT_STP_ROOTCOST) {
v32 = NR32(stpc->root_cost);
-   if (FWINV(v32 < c->root_costl ||
-   v32 > c->root_costu, EBT_STP_ROOTCOST))
+   if (FWINV(v32 < c->root_costl || v32 > c->root_costu,
+ EBT_STP_ROOTCOST))
return false;
}
if (info->bitmask & EBT_STP_SENDERPRIO) {
v16 = NR16(stpc->sender);
-   if (FWINV(v16 < c->sender_priol ||
-   v16 > c->sender_priou, EBT_STP_SENDERPRIO))
+   if (FWINV(v16 < c->sender_priol || v16 > c->sender_priou,
+ EBT_STP_SENDERPRIO))
return false;
}
if (info->bitmask & EBT_STP_SENDERADDR) {
-   verdict = 0;
-   for (i = 0; i < 6; i++)
-   verdict |= (stpc->sender[2+i] ^ c->sender_addr[i]) &
-  c->sender_addrmsk[i];
-   if (FWINV(verdict != 0, EBT_STP_SENDERADDR))
+   if (FWINV(!masked_ether_addr_equal(>sender[2],
+  c->sender_addr,
+  c->sender_addrmsk),
+ EBT_STP_SENDERADDR))
return false;
}
if (info->bitmask & EBT_STP_PORT) {
v16 = NR16(stpc->port);
-   if (FWINV(v16 < c->portl ||
-   v16 > c->portu, EBT_STP_PORT))
+   if (FWINV(v16 < c->portl || v16 > c->portu, EBT_STP_PORT))
return false;
}
if (info->bitmask & EBT_STP_MSGAGE) {
v16 = NR16(stpc->msg_age);
-   if (FWINV(v16 < c->msg_agel ||
-   v16 > c->msg_ageu, EBT_STP_MSGAGE))
+   if (FWINV(v16 < c->msg_agel || v16 > c->msg_ageu,
+ EBT_STP_MSGAGE))
return false;
}
if (info->bitmask & EBT_STP_MAXAGE) {
v16 = NR16(stpc->max_age);
-   if (FWINV(v16 < c->max_agel ||
-   v16 > c->max_ageu, EBT_STP_MAXAGE))
+   if (FWINV(v16 < c->max_agel || v16 > c->max_ageu,
+ EBT_STP_MAXAGE))
return false;
}
if (info->bitmask & EBT_STP_HELLOTIME) {
v16 = NR16(stpc->hello_time);
-   if (FWINV(v16 < c->hello_timel ||
-