tcp_bbr: improve arithmetic division in bbr_update_bw()

do_div() does a 64-by-32 division. Use div64_long() instead of it
if the divisor is long, to avoid truncation to 32-bit.
And as a nice side effect also cleans up the function a bit.

Signed-off-by: Wen Yang <wenyang@linux.alibaba.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>
Cc: Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>
Cc: netdev@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

diff --git a/net/ipv4/tcp_bbr.c b/net/ipv4/tcp_bbr.c
index a6545ef0d27b66d310b02affe14f41ab536243eb..6c4d79baff2696b859d214bb63108b9f2f61f457 100644 (file)
--- a/net/ipv4/tcp_bbr.c
+++ b/net/ipv4/tcp_bbr.c
@@ -779,8 +779,7 @@ static void bbr_update_bw(struct sock *sk, const struct rate_sample *rs)
         * bandwidth sample. Delivered is in packets and interval_us in uS and
         * ratio will be <<1 for most connections. So delivered is first scaled.
         */
-       bw = (u64)rs->delivered * BW_UNIT;
-       do_div(bw, rs->interval_us);
+       bw = div64_long((u64)rs->delivered * BW_UNIT, rs->interval_us);
 
        /* If this sample is application-limited, it is likely to have a very
         * low delivered count that represents application behavior rather than