| Message ID | 20240807105326.86665-4-maciej.fijalkowski@intel.com (mailing list archive) |
|---|---|
| State | Awaiting Upstream |
| Delegated to: | Netdev Maintainers |
| Headers | show |
| Series | ice: fix Rx data path for PAGE_SIZE >= 8192 | expand |
>-----Original Message----- >From: Intel-wired-lan <intel-wired-lan-bounces@osuosl.org> On Behalf Of >Fijalkowski, Maciej >Sent: Wednesday, August 7, 2024 4:23 PM >To: intel-wired-lan@lists.osuosl.org >Cc: Fijalkowski, Maciej <maciej.fijalkowski@intel.com>; >netdev@vger.kernel.org; bjorn@kernel.org; Nguyen, Anthony L ><anthony.l.nguyen@intel.com>; Karlsson, Magnus ><magnus.karlsson@intel.com>; luizcap@redhat.com >Subject: [Intel-wired-lan] [PATCH iwl-net 3/3] ice: fix truesize operations for >PAGE_SIZE >= 8192 > >When working on multi-buffer packet on arch that has PAGE_SIZE >= 8192, >truesize is calculated and stored in xdp_buff::frame_sz per each processed Rx >buffer. This means that frame_sz will contain the truesize based on last received >buffer, but commit 1dc1a7e7f410 ("ice: >Centrallize Rx buffer recycling") assumed this value will be constant for each >buffer, which breaks the page recycling scheme and mess up the way we >update the page::page_offset. > >To fix this, let us work on constant truesize when PAGE_SIZE >= 8192 instead of >basing this on size of a packet read from Rx descriptor. This way we can simplify >the code and avoid calculating truesize per each received frame and on top of >that when using xdp_update_skb_shared_info(), current formula for truesize >update will be valid. > >This means ice_rx_frame_truesize() can be removed altogether. >Furthermore, first call to it within ice_clean_rx_irq() for 4k PAGE_SIZE was >redundant as xdp_buff::frame_sz is initialized via xdp_init_buff() in >ice_vsi_cfg_rxq(). This should have been removed at the point where xdp_buff >struct started to be a member of ice_rx_ring and it was no longer a stack based >variable. > >There are two fixes tags as my understanding is that the first one exposed us to >broken truesize and page_offset handling and then second introduced broken >skb_shared_info update in ice_{construct,build}_skb(). > >Reported-and-tested-by: Luiz Capitulino <luizcap@redhat.com> >Closes: https://lore.kernel.org/netdev/8f9e2a5c-fd30-4206-9311- >946a06d031bb@redhat.com/ >Fixes: 1dc1a7e7f410 ("ice: Centrallize Rx buffer recycling") >Fixes: 2fba7dc5157b ("ice: Add support for XDP multi-buffer on Rx side") >Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> >--- > drivers/net/ethernet/intel/ice/ice_base.c | 21 ++++++++++++++- >drivers/net/ethernet/intel/ice/ice_txrx.c | 33 ----------------------- > 2 files changed, 20 insertions(+), 34 deletions(-) > Tested-by: Chandan Kumar Rout <chandanx.rout@intel.com> (A Contingent Worker at Intel)
diff --git a/drivers/net/ethernet/intel/ice/ice_base.c b/drivers/net/ethernet/intel/ice/ice_base.c index 1facf179a96f..f448d3a84564 100644 --- a/drivers/net/ethernet/intel/ice/ice_base.c +++ b/drivers/net/ethernet/intel/ice/ice_base.c @@ -512,6 +512,25 @@ static void ice_xsk_pool_fill_cb(struct ice_rx_ring *ring) xsk_pool_fill_cb(ring->xsk_pool, &desc); } +/** + * ice_get_frame_sz - calculate xdp_buff::frame_sz + * @rx_ring: the ring being configured + * + * Return frame size based on underlying PAGE_SIZE + */ +static unsigned int ice_get_frame_sz(struct ice_rx_ring *rx_ring) +{ + unsigned int frame_sz; + +#if (PAGE_SIZE >= 8192) + frame_sz = rx_ring->rx_buf_len; +#else + frame_sz = ice_rx_pg_size(rx_ring) / 2; +#endif + + return frame_sz; +} + /** * ice_vsi_cfg_rxq - Configure an Rx queue * @ring: the ring being configured @@ -576,7 +595,7 @@ static int ice_vsi_cfg_rxq(struct ice_rx_ring *ring) } } - xdp_init_buff(&ring->xdp, ice_rx_pg_size(ring) / 2, &ring->xdp_rxq); + xdp_init_buff(&ring->xdp, ice_get_frame_sz(ring), &ring->xdp_rxq); ring->xdp.data = NULL; ring->xdp_ext.pkt_ctx = &ring->pkt_ctx; err = ice_setup_rx_ctx(ring); diff --git a/drivers/net/ethernet/intel/ice/ice_txrx.c b/drivers/net/ethernet/intel/ice/ice_txrx.c index 4b690952bb40..c9bc3f1add5d 100644 --- a/drivers/net/ethernet/intel/ice/ice_txrx.c +++ b/drivers/net/ethernet/intel/ice/ice_txrx.c @@ -521,30 +521,6 @@ int ice_setup_rx_ring(struct ice_rx_ring *rx_ring) return -ENOMEM; } -/** - * ice_rx_frame_truesize - * @rx_ring: ptr to Rx ring - * @size: size - * - * calculate the truesize with taking into the account PAGE_SIZE of - * underlying arch - */ -static unsigned int -ice_rx_frame_truesize(struct ice_rx_ring *rx_ring, const unsigned int size) -{ - unsigned int truesize; - -#if (PAGE_SIZE < 8192) - truesize = ice_rx_pg_size(rx_ring) / 2; /* Must be power-of-2 */ -#else - truesize = rx_ring->rx_offset ? - SKB_DATA_ALIGN(rx_ring->rx_offset + size) + - SKB_DATA_ALIGN(sizeof(struct skb_shared_info)) : - SKB_DATA_ALIGN(size); -#endif - return truesize; -} - /** * ice_run_xdp - Executes an XDP program on initialized xdp_buff * @rx_ring: Rx ring @@ -1154,11 +1130,6 @@ int ice_clean_rx_irq(struct ice_rx_ring *rx_ring, int budget) bool failure; u32 first; - /* Frame size depend on rx_ring setup when PAGE_SIZE=4K */ -#if (PAGE_SIZE < 8192) - xdp->frame_sz = ice_rx_frame_truesize(rx_ring, 0); -#endif - xdp_prog = READ_ONCE(rx_ring->xdp_prog); if (xdp_prog) { xdp_ring = rx_ring->xdp_ring; @@ -1217,10 +1188,6 @@ int ice_clean_rx_irq(struct ice_rx_ring *rx_ring, int budget) hard_start = page_address(rx_buf->page) + rx_buf->page_offset - offset; xdp_prepare_buff(xdp, hard_start, offset, size, !!offset); -#if (PAGE_SIZE > 4096) - /* At larger PAGE_SIZE, frame_sz depend on len size */ - xdp->frame_sz = ice_rx_frame_truesize(rx_ring, size); -#endif xdp_buff_clear_frags_flag(xdp); } else if (ice_add_xdp_frag(rx_ring, xdp, rx_buf, size)) { break;
When working on multi-buffer packet on arch that has PAGE_SIZE >= 8192, truesize is calculated and stored in xdp_buff::frame_sz per each processed Rx buffer. This means that frame_sz will contain the truesize based on last received buffer, but commit 1dc1a7e7f410 ("ice: Centrallize Rx buffer recycling") assumed this value will be constant for each buffer, which breaks the page recycling scheme and mess up the way we update the page::page_offset. To fix this, let us work on constant truesize when PAGE_SIZE >= 8192 instead of basing this on size of a packet read from Rx descriptor. This way we can simplify the code and avoid calculating truesize per each received frame and on top of that when using xdp_update_skb_shared_info(), current formula for truesize update will be valid. This means ice_rx_frame_truesize() can be removed altogether. Furthermore, first call to it within ice_clean_rx_irq() for 4k PAGE_SIZE was redundant as xdp_buff::frame_sz is initialized via xdp_init_buff() in ice_vsi_cfg_rxq(). This should have been removed at the point where xdp_buff struct started to be a member of ice_rx_ring and it was no longer a stack based variable. There are two fixes tags as my understanding is that the first one exposed us to broken truesize and page_offset handling and then second introduced broken skb_shared_info update in ice_{construct,build}_skb(). Reported-and-tested-by: Luiz Capitulino <luizcap@redhat.com> Closes: https://lore.kernel.org/netdev/8f9e2a5c-fd30-4206-9311-946a06d031bb@redhat.com/ Fixes: 1dc1a7e7f410 ("ice: Centrallize Rx buffer recycling") Fixes: 2fba7dc5157b ("ice: Add support for XDP multi-buffer on Rx side") Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com> --- drivers/net/ethernet/intel/ice/ice_base.c | 21 ++++++++++++++- drivers/net/ethernet/intel/ice/ice_txrx.c | 33 ----------------------- 2 files changed, 20 insertions(+), 34 deletions(-)