Problem:
--------
before48(a, b) returns the same as before48(b, a) for all a, b which
are 2^47 apart (modulo-2^48). The reason is that the difference a-b is used
and `true' is returned whenever 2^47 <= a-b <= 2^48-1. This has the
disadvantage that a positive difference a-b = 2^47 can not be distinguished
from a negative difference b-a = 2^47. As a result, an ambiguity arises in
2^47 cases.
Fix:
----
The fix is the same as the one suggested for TCP: define a `before' b
whenever 1 <= b-a <= 2^47-1 (positive signed 48-bit numbers). As a result,
the ambiguity disappears. (Note: we could have used dccp_delta_seqno, but
Arnaldo's concept of using shift and subtraction requires fewer operations).
The patch further defines after48 as a macro, since it is in actual fact just
before48 with the parameters swapped (this mirrors the TCP solution).
Signed-off-by: Gerrit Renker <[EMAIL PROTECTED]>
Acked-by: Ian McDonald <[EMAIL PROTECTED]>
Signed-off-by: Arnaldo Carvalho de Melo <[EMAIL PROTECTED]>
---
net/dccp/dccp.h | 7 ++-----
1 files changed, 2 insertions(+), 5 deletions(-)
diff --git a/net/dccp/dccp.h b/net/dccp/dccp.h
index e6c95e9..287a040 100644
--- a/net/dccp/dccp.h
+++ b/net/dccp/dccp.h
@@ -124,14 +124,11 @@ static inline s64 dccp_delta_seqno(const u64 seqno1,
const u64 seqno2)
/* is seq1 < seq2 ? */
static inline int before48(const u64 seq1, const u64 seq2)
{
- return (s64)((seq1 << 16) - (seq2 << 16)) < 0;
+ return (s64)((seq2 << 16) - (seq1 << 16)) > 0;
}
/* is seq1 > seq2 ? */
-static inline int after48(const u64 seq1, const u64 seq2)
-{
- return (s64)((seq2 << 16) - (seq1 << 16)) < 0;
-}
+#define after48(seq1, seq2) before48(seq2, seq1)
/* is seq2 <= seq1 <= seq3 ? */
static inline int between48(const u64 seq1, const u64 seq2, const u64 seq3)
--
1.5.0.3
-
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to [EMAIL PROTECTED]
More majordomo info at http://vger.kernel.org/majordomo-info.html
