ia64/xen-unstable

changeset 11169:4b145f81fce4

[NET] front: Fix indentation of a comment.
Signed-off-by: Keir Fraser <keir@xensource.com>
author kfraser@localhost.localdomain
date Wed Aug 16 16:14:03 2006 +0100 (2006-08-16)
parents d4efff1beedb
children 00a70786e550
files linux-2.6-xen-sparse/drivers/xen/netfront/netfront.c
line diff
     1.1 --- a/linux-2.6-xen-sparse/drivers/xen/netfront/netfront.c	Wed Aug 16 16:11:53 2006 +0100
     1.2 +++ b/linux-2.6-xen-sparse/drivers/xen/netfront/netfront.c	Wed Aug 16 16:14:03 2006 +0100
     1.3 @@ -1259,26 +1259,26 @@ err:
     1.4  
     1.5  		i = xennet_fill_frags(np, skb, &tmpq);
     1.6  
     1.7 -        /*
     1.8 -         * Truesize must approximates the size of true data plus
     1.9 -         * any supervisor overheads. Adding hypervisor overheads
    1.10 -         * has been shown to significantly reduce achievable
    1.11 -         * bandwidth with the default receive buffer size. It is
    1.12 -         * therefore not wise to account for it here.
    1.13 -         *
    1.14 -         * After alloc_skb(RX_COPY_THRESHOLD), truesize is set to
    1.15 -         * RX_COPY_THRESHOLD + the supervisor overheads. Here, we
    1.16 -         * add the size of the data pulled in xennet_fill_frags().
    1.17 -         *
    1.18 -         * We also adjust for any unused space in the main data
    1.19 -         * area by subtracting (RX_COPY_THRESHOLD - len). This is
    1.20 -         * especially important with drivers which split incoming
    1.21 -         * packets into header and data, using only 66 bytes of
    1.22 -         * the main data area (see the e1000 driver for example.)
    1.23 -         * On such systems, without this last adjustement, our
    1.24 -         * achievable receive throughout using the standard receive
    1.25 -         * buffer size was cut by 25%(!!!).
    1.26 -         */
    1.27 +		/*
    1.28 +		 * Truesize must approximates the size of true data plus
    1.29 +		 * any supervisor overheads. Adding hypervisor overheads
    1.30 +		 * has been shown to significantly reduce achievable
    1.31 +		 * bandwidth with the default receive buffer size. It is
    1.32 +		 * therefore not wise to account for it here.
    1.33 +		 *
    1.34 +		 * After alloc_skb(RX_COPY_THRESHOLD), truesize is set to
    1.35 +		 * RX_COPY_THRESHOLD + the supervisor overheads. Here, we
    1.36 +		 * add the size of the data pulled in xennet_fill_frags().
    1.37 +		 *
    1.38 +		 * We also adjust for any unused space in the main data
    1.39 +		 * area by subtracting (RX_COPY_THRESHOLD - len). This is
    1.40 +		 * especially important with drivers which split incoming
    1.41 +		 * packets into header and data, using only 66 bytes of
    1.42 +		 * the main data area (see the e1000 driver for example.)
    1.43 +		 * On such systems, without this last adjustement, our
    1.44 +		 * achievable receive throughout using the standard receive
    1.45 +		 * buffer size was cut by 25%(!!!).
    1.46 +		 */
    1.47  		skb->truesize += skb->data_len - (RX_COPY_THRESHOLD - len);
    1.48  		skb->len += skb->data_len;
    1.49