Return-path: Received: from sabertooth01.qualcomm.com ([65.197.215.72]:47089 "EHLO sabertooth01.qualcomm.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1752313AbbBOMDF (ORCPT ); Sun, 15 Feb 2015 07:03:05 -0500 From: Vladimir Kondratiev Cc: Vladimir Kondratiev , linux-wireless@vger.kernel.org, wil6210@qca.qualcomm.com To: Kalle Valo Subject: [PATCH 4/9] wil6210: branch prediction hints Date: Sun, 15 Feb 2015 14:02:33 +0200 Message-Id: <1424001758-29529-5-git-send-email-qca_vkondrat@qca.qualcomm.com> (sfid-20150215_130319_242056_A3FC18A5) In-Reply-To: <1424001758-29529-1-git-send-email-qca_vkondrat@qca.qualcomm.com> References: <1424001758-29529-1-git-send-email-qca_vkondrat@qca.qualcomm.com> Sender: linux-wireless-owner@vger.kernel.org List-ID: Mark expected branches using likely()/unlikely(). Do it on high performance route - data path and interrupts Signed-off-by: Vladimir Kondratiev --- drivers/net/wireless/ath/wil6210/interrupt.c | 25 ++++++++--------- drivers/net/wireless/ath/wil6210/txrx.c | 40 ++++++++++++++-------------- 2 files changed, 33 insertions(+), 32 deletions(-) diff --git a/drivers/net/wireless/ath/wil6210/interrupt.c b/drivers/net/wireless/ath/wil6210/interrupt.c index d5a651b..28ffc18 100644 --- a/drivers/net/wireless/ath/wil6210/interrupt.c +++ b/drivers/net/wireless/ath/wil6210/interrupt.c @@ -226,7 +226,7 @@ static irqreturn_t wil6210_irq_rx(int irq, void *cookie) trace_wil6210_irq_rx(isr); wil_dbg_irq(wil, "ISR RX 0x%08x\n", isr); - if (!isr) { + if (unlikely(!isr)) { wil_err(wil, "spurious IRQ: RX\n"); return IRQ_NONE; } @@ -239,17 +239,18 @@ static irqreturn_t wil6210_irq_rx(int irq, void *cookie) * action is always the same - should empty the accumulated * packets from the RX ring. */ - if (isr & (BIT_DMA_EP_RX_ICR_RX_DONE | BIT_DMA_EP_RX_ICR_RX_HTRSH)) { + if (likely(isr & (BIT_DMA_EP_RX_ICR_RX_DONE | + BIT_DMA_EP_RX_ICR_RX_HTRSH))) { wil_dbg_irq(wil, "RX done\n"); - if (isr & BIT_DMA_EP_RX_ICR_RX_HTRSH) + if (unlikely(isr & BIT_DMA_EP_RX_ICR_RX_HTRSH)) wil_err_ratelimited(wil, "Received \"Rx buffer is in risk of overflow\" interrupt\n"); isr &= ~(BIT_DMA_EP_RX_ICR_RX_DONE | BIT_DMA_EP_RX_ICR_RX_HTRSH); - if (test_bit(wil_status_reset_done, wil->status)) { - if (test_bit(wil_status_napi_en, wil->status)) { + if (likely(test_bit(wil_status_reset_done, wil->status))) { + if (likely(test_bit(wil_status_napi_en, wil->status))) { wil_dbg_txrx(wil, "NAPI(Rx) schedule\n"); need_unmask = false; napi_schedule(&wil->napi_rx); @@ -262,7 +263,7 @@ static irqreturn_t wil6210_irq_rx(int irq, void *cookie) } } - if (isr) + if (unlikely(isr)) wil_err(wil, "un-handled RX ISR bits 0x%08x\n", isr); /* Rx IRQ will be enabled when NAPI processing finished */ @@ -286,19 +287,19 @@ static irqreturn_t wil6210_irq_tx(int irq, void *cookie) trace_wil6210_irq_tx(isr); wil_dbg_irq(wil, "ISR TX 0x%08x\n", isr); - if (!isr) { + if (unlikely(!isr)) { wil_err(wil, "spurious IRQ: TX\n"); return IRQ_NONE; } wil6210_mask_irq_tx(wil); - if (isr & BIT_DMA_EP_TX_ICR_TX_DONE) { + if (likely(isr & BIT_DMA_EP_TX_ICR_TX_DONE)) { wil_dbg_irq(wil, "TX done\n"); isr &= ~BIT_DMA_EP_TX_ICR_TX_DONE; /* clear also all VRING interrupts */ isr &= ~(BIT(25) - 1UL); - if (test_bit(wil_status_reset_done, wil->status)) { + if (likely(test_bit(wil_status_reset_done, wil->status))) { wil_dbg_txrx(wil, "NAPI(Tx) schedule\n"); need_unmask = false; napi_schedule(&wil->napi_tx); @@ -307,7 +308,7 @@ static irqreturn_t wil6210_irq_tx(int irq, void *cookie) } } - if (isr) + if (unlikely(isr)) wil_err(wil, "un-handled TX ISR bits 0x%08x\n", isr); /* Tx IRQ will be enabled when NAPI processing finished */ @@ -496,11 +497,11 @@ static irqreturn_t wil6210_hardirq(int irq, void *cookie) /** * pseudo_cause is Clear-On-Read, no need to ACK */ - if ((pseudo_cause == 0) || ((pseudo_cause & 0xff) == 0xff)) + if (unlikely((pseudo_cause == 0) || ((pseudo_cause & 0xff) == 0xff))) return IRQ_NONE; /* FIXME: IRQ mask debug */ - if (wil6210_debug_irq_mask(wil, pseudo_cause)) + if (unlikely(wil6210_debug_irq_mask(wil, pseudo_cause))) return IRQ_NONE; trace_wil6210_irq_pseudo(pseudo_cause); diff --git a/drivers/net/wireless/ath/wil6210/txrx.c b/drivers/net/wireless/ath/wil6210/txrx.c index 779d836..7e119d0 100644 --- a/drivers/net/wireless/ath/wil6210/txrx.c +++ b/drivers/net/wireless/ath/wil6210/txrx.c @@ -370,11 +370,11 @@ static struct sk_buff *wil_vring_reap_rx(struct wil6210_priv *wil, BUILD_BUG_ON(sizeof(struct vring_rx_desc) > sizeof(skb->cb)); - if (wil_vring_is_empty(vring)) + if (unlikely(wil_vring_is_empty(vring))) return NULL; _d = &vring->va[vring->swhead].rx; - if (!(_d->dma.status & RX_DMA_STATUS_DU)) { + if (unlikely(!(_d->dma.status & RX_DMA_STATUS_DU))) { /* it is not error, we just reached end of Rx done area */ return NULL; } @@ -394,7 +394,7 @@ static struct sk_buff *wil_vring_reap_rx(struct wil6210_priv *wil, wil_hex_dump_txrx("Rx ", DUMP_PREFIX_NONE, 32, 4, (const void *)d, sizeof(*d), false); - if (dmalen > sz) { + if (unlikely(dmalen > sz)) { wil_err(wil, "Rx size too large: %d bytes!\n", dmalen); kfree_skb(skb); return NULL; @@ -423,14 +423,14 @@ static struct sk_buff *wil_vring_reap_rx(struct wil6210_priv *wil, * in Rx descriptor. If type is not data, it is 802.11 frame as is */ ftype = wil_rxdesc_ftype(d) << 2; - if (ftype != IEEE80211_FTYPE_DATA) { + if (unlikely(ftype != IEEE80211_FTYPE_DATA)) { wil_dbg_txrx(wil, "Non-data frame ftype 0x%08x\n", ftype); /* TODO: process it */ kfree_skb(skb); return NULL; } - if (skb->len < ETH_HLEN) { + if (unlikely(skb->len < ETH_HLEN)) { wil_err(wil, "Short frame, len = %d\n", skb->len); /* TODO: process it (i.e. BAR) */ kfree_skb(skb); @@ -441,9 +441,9 @@ static struct sk_buff *wil_vring_reap_rx(struct wil6210_priv *wil, * and in case of error drop the packet * higher stack layers will handle retransmission (if required) */ - if (d->dma.status & RX_DMA_STATUS_L4I) { + if (likely(d->dma.status & RX_DMA_STATUS_L4I)) { /* L4 protocol identified, csum calculated */ - if ((d->dma.error & RX_DMA_ERROR_L4_ERR) == 0) + if (likely((d->dma.error & RX_DMA_ERROR_L4_ERR) == 0)) skb->ip_summed = CHECKSUM_UNNECESSARY; /* If HW reports bad checksum, let IP stack re-check it * For example, HW don't understand Microsoft IP stack that @@ -472,7 +472,7 @@ static int wil_rx_refill(struct wil6210_priv *wil, int count) (next_tail != v->swhead) && (count-- > 0); v->swtail = next_tail) { rc = wil_vring_alloc_skb(wil, v, v->swtail, headroom); - if (rc) { + if (unlikely(rc)) { wil_err(wil, "Error %d in wil_rx_refill[%d]\n", rc, v->swtail); break; @@ -534,7 +534,7 @@ void wil_rx_handle(struct wil6210_priv *wil, int *quota) struct vring *v = &wil->vring_rx; struct sk_buff *skb; - if (!v->va) { + if (unlikely(!v->va)) { wil_err(wil, "Rx IRQ while Rx not yet initialized\n"); return; } @@ -927,7 +927,7 @@ static int __wil_tx_vring(struct wil6210_priv *wil, struct vring *vring, if (unlikely(!txdata->enabled)) return -EINVAL; - if (avail < 1 + nr_frags) { + if (unlikely(avail < 1 + nr_frags)) { wil_err_ratelimited(wil, "Tx ring[%2d] full. No space for %d fragments\n", vring_index, 1 + nr_frags); @@ -948,7 +948,7 @@ static int __wil_tx_vring(struct wil6210_priv *wil, struct vring *vring, /* 1-st segment */ wil_tx_desc_map(d, pa, skb_headlen(skb), vring_index); /* Process TCP/UDP checksum offloading */ - if (wil_tx_desc_offload_cksum_set(wil, d, skb)) { + if (unlikely(wil_tx_desc_offload_cksum_set(wil, d, skb))) { wil_err(wil, "Tx[%2d] Failed to set cksum, drop packet\n", vring_index); goto dma_error; @@ -1051,18 +1051,18 @@ netdev_tx_t wil_start_xmit(struct sk_buff *skb, struct net_device *ndev) int rc; wil_dbg_txrx(wil, "%s()\n", __func__); - if (!test_bit(wil_status_fwready, wil->status)) { + if (unlikely(!test_bit(wil_status_fwready, wil->status))) { if (!pr_once_fw) { wil_err(wil, "FW not ready\n"); pr_once_fw = true; } goto drop; } - if (!test_bit(wil_status_fwconnected, wil->status)) { + if (unlikely(!test_bit(wil_status_fwconnected, wil->status))) { wil_err(wil, "FW not connected\n"); goto drop; } - if (wil->wdev->iftype == NL80211_IFTYPE_MONITOR) { + if (unlikely(wil->wdev->iftype == NL80211_IFTYPE_MONITOR)) { wil_err(wil, "Xmit in monitor mode not supported\n"); goto drop; } @@ -1078,7 +1078,7 @@ netdev_tx_t wil_start_xmit(struct sk_buff *skb, struct net_device *ndev) else vring = wil_tx_bcast(wil, skb); } - if (!vring) { + if (unlikely(!vring)) { wil_dbg_txrx(wil, "No Tx VRING found for %pM\n", eth->h_dest); goto drop; } @@ -1086,7 +1086,7 @@ netdev_tx_t wil_start_xmit(struct sk_buff *skb, struct net_device *ndev) rc = wil_tx_vring(wil, vring, skb); /* do we still have enough room in the vring? */ - if (wil_vring_avail_tx(vring) < wil_vring_wmark_low(vring)) { + if (unlikely(wil_vring_avail_tx(vring) < wil_vring_wmark_low(vring))) { netif_tx_stop_all_queues(wil_to_ndev(wil)); wil_dbg_txrx(wil, "netif_tx_stop : ring full\n"); } @@ -1142,12 +1142,12 @@ int wil_tx_complete(struct wil6210_priv *wil, int ringid) struct wil_net_stats *stats = &wil->sta[cid].stats; volatile struct vring_tx_desc *_d; - if (!vring->va) { + if (unlikely(!vring->va)) { wil_err(wil, "Tx irq[%d]: vring not initialized\n", ringid); return 0; } - if (!txdata->enabled) { + if (unlikely(!txdata->enabled)) { wil_info(wil, "Tx irq[%d]: vring disabled\n", ringid); return 0; } @@ -1165,7 +1165,7 @@ int wil_tx_complete(struct wil6210_priv *wil, int ringid) /* TODO: check we are not past head */ _d = &vring->va[lf].tx; - if (!(_d->dma.status & TX_DMA_STATUS_DU)) + if (unlikely(!(_d->dma.status & TX_DMA_STATUS_DU))) break; new_swtail = (lf + 1) % vring->size; @@ -1193,7 +1193,7 @@ int wil_tx_complete(struct wil6210_priv *wil, int ringid) wil_txdesc_unmap(dev, d, ctx); if (skb) { - if (d->dma.error == 0) { + if (likely(d->dma.error == 0)) { ndev->stats.tx_packets++; stats->tx_packets++; ndev->stats.tx_bytes += skb->len; -- 2.1.0