igc: don't reserve excessive XDP_PACKET_HEADROOM on XSK Rx to skb
authorAlexander Lobakin <alexandr.lobakin@intel.com>
Wed, 8 Dec 2021 14:06:59 +0000 (15:06 +0100)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Fri, 8 Apr 2022 12:23:27 +0000 (14:23 +0200)
commit28e561f3660d1a55bf01e320b6c75b3fd0a33bea
treea24d19d0f225c6becb288a475fded17a95fee431
parent0e2f6a7f59f1c12a028cba930a87d74391b35f9c
igc: don't reserve excessive XDP_PACKET_HEADROOM on XSK Rx to skb

[ Upstream commit f9e61d365bafdee40fe2586fc6be490c3e824dad ]

{__,}napi_alloc_skb() allocates and reserves additional NET_SKB_PAD
+ NET_IP_ALIGN for any skb.
OTOH, igc_construct_skb_zc() currently allocates and reserves
additional `xdp->data_meta - xdp->data_hard_start`, which is about
XDP_PACKET_HEADROOM for XSK frames.
There's no need for that at all as the frame is post-XDP and will
go only to the networking stack core.
Pass the size of the actual data only (+ meta) to
__napi_alloc_skb() and don't reserve anything. This will give
enough headroom for stack processing.
Also, net_prefetch() xdp->data_meta and align the copy size to
speed-up memcpy() a little and better match igc_construct_skb().

Fixes: fc9df2a0b520 ("igc: Enable RX via AF_XDP zero-copy")
Signed-off-by: Alexander Lobakin <alexandr.lobakin@intel.com>
Reviewed-by: Michal Swiatkowski <michal.swiatkowski@linux.intel.com>
Tested-by: Nechama Kraus <nechamax.kraus@linux.intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>
drivers/net/ethernet/intel/igc/igc_main.c