Create a new getle16 helper to replace Mk16 that deals with
unaligned access. Also, use put_unaligned when writing out
the rc4key result. Opencode the few remaining Hi8/Lo8 calls
and remove the helpers.
Signed-off-by: Harvey Harrison <[email protected]>
---
net/mac80211/tkip.c | 68 ++++++++++++++++++++++----------------------------
1 files changed, 30 insertions(+), 38 deletions(-)
diff --git a/net/mac80211/tkip.c b/net/mac80211/tkip.c
index a3dfa67..dbe80ee 100644
--- a/net/mac80211/tkip.c
+++ b/net/mac80211/tkip.c
@@ -11,6 +11,7 @@
#include <linux/bitops.h>
#include <linux/types.h>
#include <linux/netdevice.h>
+#include <asm/unaligned.h>
#include <net/mac80211.h>
#include "ieee80211_key.h"
@@ -63,21 +64,13 @@ static const u16 tkip_sbox[256] =
};
-static inline u16 Mk16(u8 x, u8 y)
-{
- return ((u16) x << 8) | (u16) y;
-}
-
-
-static inline u8 Hi8(u16 v)
-{
- return v >> 8;
-}
-
-
-static inline u8 Lo8(u16 v)
+/*
+ * Get a cpu-byteorder u16 from a u8 pointer fixing up possible unaligned
+ * accesses.
+ */
+static inline u16 getle16(const u8 *x)
{
- return v & 0xff;
+ return le16_to_cpu(get_unaligned((__le16 *)x));
}
static inline u16 tkip_S(u16 val)
@@ -98,17 +91,17 @@ static void tkip_mixing_phase1(const u8 *ta, const u8 *tk, u32 tsc_IV32,
p1k[0] = tsc_IV32 & 0xFFFF;
p1k[1] = tsc_IV32 >> 16;
- p1k[2] = Mk16(ta[1], ta[0]);
- p1k[3] = Mk16(ta[3], ta[2]);
- p1k[4] = Mk16(ta[5], ta[4]);
+ p1k[2] = getle16(ta + 0);
+ p1k[3] = getle16(ta + 2);
+ p1k[4] = getle16(ta + 4);
for (i = 0; i < PHASE1_LOOP_COUNT; i++) {
j = 2 * (i & 1);
- p1k[0] += tkip_S(p1k[4] ^ Mk16(tk[ 1 + j], tk[ 0 + j]));
- p1k[1] += tkip_S(p1k[0] ^ Mk16(tk[ 5 + j], tk[ 4 + j]));
- p1k[2] += tkip_S(p1k[1] ^ Mk16(tk[ 9 + j], tk[ 8 + j]));
- p1k[3] += tkip_S(p1k[2] ^ Mk16(tk[13 + j], tk[12 + j]));
- p1k[4] += tkip_S(p1k[3] ^ Mk16(tk[ 1 + j], tk[ 0 + j])) + i;
+ p1k[0] += tkip_S(p1k[4] ^ getle16(tk + 0 + j));
+ p1k[1] += tkip_S(p1k[0] ^ getle16(tk + 4 + j));
+ p1k[2] += tkip_S(p1k[1] ^ getle16(tk + 8 + j));
+ p1k[3] += tkip_S(p1k[2] ^ getle16(tk + 12 + j));
+ p1k[4] += tkip_S(p1k[3] ^ getle16(tk + 0 + j)) + i;
}
}
@@ -126,28 +119,27 @@ static void tkip_mixing_phase2(const u16 *p1k, const u8 *tk, u16 tsc_IV16,
ppk[4] = p1k[4];
ppk[5] = p1k[4] + tsc_IV16;
- ppk[0] += tkip_S(ppk[5] ^ Mk16(tk[ 1], tk[ 0]));
- ppk[1] += tkip_S(ppk[0] ^ Mk16(tk[ 3], tk[ 2]));
- ppk[2] += tkip_S(ppk[1] ^ Mk16(tk[ 5], tk[ 4]));
- ppk[3] += tkip_S(ppk[2] ^ Mk16(tk[ 7], tk[ 6]));
- ppk[4] += tkip_S(ppk[3] ^ Mk16(tk[ 9], tk[ 8]));
- ppk[5] += tkip_S(ppk[4] ^ Mk16(tk[11], tk[10]));
- ppk[0] += ror16(ppk[5] ^ Mk16(tk[13], tk[12]), 1);
- ppk[1] += ror16(ppk[0] ^ Mk16(tk[15], tk[14]), 1);
+ ppk[0] += tkip_S(ppk[5] ^ getle16(tk + 0));
+ ppk[1] += tkip_S(ppk[0] ^ getle16(tk + 2));
+ ppk[2] += tkip_S(ppk[1] ^ getle16(tk + 4));
+ ppk[3] += tkip_S(ppk[2] ^ getle16(tk + 6));
+ ppk[4] += tkip_S(ppk[3] ^ getle16(tk + 8));
+ ppk[5] += tkip_S(ppk[4] ^ getle16(tk + 10));
+ ppk[0] += ror16(ppk[5] ^ getle16(tk + 12), 1);
+ ppk[1] += ror16(ppk[0] ^ getle16(tk + 14), 1);
ppk[2] += ror16(ppk[1], 1);
ppk[3] += ror16(ppk[2], 1);
ppk[4] += ror16(ppk[3], 1);
ppk[5] += ror16(ppk[4], 1);
- rc4key[0] = Hi8(tsc_IV16);
- rc4key[1] = (Hi8(tsc_IV16) | 0x20) & 0x7f;
- rc4key[2] = Lo8(tsc_IV16);
- rc4key[3] = Lo8((ppk[5] ^ Mk16(tk[1], tk[0])) >> 1);
+ rc4key[0] = tsc_IV16 >> 8;
+ rc4key[1] = ((tsc_IV16 >> 8) | 0x20) & 0x7f;
+ rc4key[2] = tsc_IV16 & 0xFF;
+ rc4key[3] = ((ppk[5] ^ getle16(tk)) >> 1) & 0xFF;
- for (i = 0; i < 6; i++) {
- rc4key[4 + 2 * i] = Lo8(ppk[i]);
- rc4key[5 + 2 * i] = Hi8(ppk[i]);
- }
+ rc4key += 4;
+ for (i = 0; i < 6; i++)
+ put_unaligned(cpu_to_le16(ppk[i]), (__le16 *)rc4key + i);
}
--
1.5.4.4.684.g0e08