| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Date: Tue, 22 Sep 2020 00:31:13 +0200
From: Hauke Mehrtens <hauke@...ke-m.de>
To: davem@...emloft.net
Cc: kuba@...nel.org, netdev@...r.kernel.org,
martin.blumenstingl@...glemail.com,
Hauke Mehrtens <hauke@...ke-m.de>
Subject: [PATCH] net: lantiq: Add locking for TX DMA channel
The TX DMA channel data is accessed by the xrx200_start_xmit() and the
xrx200_tx_housekeeping() function from different threads. Make sure the
accesses are synchronized by using locking around the accesses.
Signed-off-by: Hauke Mehrtens <hauke@...ke-m.de>
---
drivers/net/ethernet/lantiq_xrx200.c | 19 +++++++++++++++++--
1 file changed, 17 insertions(+), 2 deletions(-)
diff --git a/drivers/net/ethernet/lantiq_xrx200.c b/drivers/net/ethernet/lantiq_xrx200.c
index 635ff3a5dcfb..f4de09d1f582 100644
--- a/drivers/net/ethernet/lantiq_xrx200.c
+++ b/drivers/net/ethernet/lantiq_xrx200.c
@@ -59,6 +59,7 @@ struct xrx200_chan {
struct ltq_dma_channel dma;
struct sk_buff *skb[LTQ_DESC_NUM];
+ spinlock_t lock;
struct xrx200_priv *priv;
};
@@ -242,9 +243,11 @@ static int xrx200_tx_housekeeping(struct napi_struct *napi, int budget)
struct xrx200_chan *ch = container_of(napi,
struct xrx200_chan, napi);
struct net_device *net_dev = ch->priv->net_dev;
+ unsigned long flags;
int pkts = 0;
int bytes = 0;
+ spin_lock_irqsave(&ch->lock, flags);
while (pkts < budget) {
struct ltq_dma_desc *desc = &ch->dma.desc_base[ch->tx_free];
@@ -268,6 +271,8 @@ static int xrx200_tx_housekeeping(struct napi_struct *napi, int budget)
net_dev->stats.tx_bytes += bytes;
netdev_completed_queue(ch->priv->net_dev, pkts, bytes);
+ spin_unlock_irqrestore(&ch->lock, flags);
+
if (netif_queue_stopped(net_dev))
netif_wake_queue(net_dev);
@@ -284,7 +289,8 @@ static netdev_tx_t xrx200_start_xmit(struct sk_buff *skb,
{
struct xrx200_priv *priv = netdev_priv(net_dev);
struct xrx200_chan *ch = &priv->chan_tx;
- struct ltq_dma_desc *desc = &ch->dma.desc_base[ch->dma.desc];
+ struct ltq_dma_desc *desc;
+ unsigned long flags;
u32 byte_offset;
dma_addr_t mapping;
int len;
@@ -297,8 +303,11 @@ static netdev_tx_t xrx200_start_xmit(struct sk_buff *skb,
len = skb->len;
+ spin_lock_irqsave(&ch->lock, flags);
+ desc = &ch->dma.desc_base[ch->dma.desc];
if ((desc->ctl & (LTQ_DMA_OWN | LTQ_DMA_C)) || ch->skb[ch->dma.desc]) {
netdev_err(net_dev, "tx ring full\n");
+ spin_unlock_irqrestore(&ch->lock, flags);
netif_stop_queue(net_dev);
return NETDEV_TX_BUSY;
}
@@ -306,8 +315,10 @@ static netdev_tx_t xrx200_start_xmit(struct sk_buff *skb,
ch->skb[ch->dma.desc] = skb;
mapping = dma_map_single(priv->dev, skb->data, len, DMA_TO_DEVICE);
- if (unlikely(dma_mapping_error(priv->dev, mapping)))
+ if (unlikely(dma_mapping_error(priv->dev, mapping))) {
+ spin_unlock_irqrestore(&ch->lock, flags);
goto err_drop;
+ }
/* dma needs to start on a 16 byte aligned address */
byte_offset = mapping % 16;
@@ -324,6 +335,8 @@ static netdev_tx_t xrx200_start_xmit(struct sk_buff *skb,
netdev_sent_queue(net_dev, len);
+ spin_unlock_irqrestore(&ch->lock, flags);
+
return NETDEV_TX_OK;
err_drop:
@@ -367,6 +380,7 @@ static int xrx200_dma_init(struct xrx200_priv *priv)
ch_rx->dma.nr = XRX200_DMA_RX;
ch_rx->dma.dev = priv->dev;
ch_rx->priv = priv;
+ spin_lock_init(&ch_rx->lock);
ltq_dma_alloc_rx(&ch_rx->dma);
for (ch_rx->dma.desc = 0; ch_rx->dma.desc < LTQ_DESC_NUM;
@@ -387,6 +401,7 @@ static int xrx200_dma_init(struct xrx200_priv *priv)
ch_tx->dma.nr = XRX200_DMA_TX;
ch_tx->dma.dev = priv->dev;
ch_tx->priv = priv;
+ spin_lock_init(&ch_tx->lock);
ltq_dma_alloc_tx(&ch_tx->dma);
ret = devm_request_irq(priv->dev, ch_tx->dma.irq, xrx200_dma_irq, 0,
--
2.20.1
Powered by blists - more mailing lists