diff options
author | Vladislav Zolotarov <vladz@broadcom.com> | 2010-02-28 00:12:02 +0000 |
---|---|---|
committer | David S. Miller <davem@davemloft.net> | 2010-02-28 18:37:12 -0800 |
commit | c16cc0b464b8876cfd57ce1c1dbcb6f9a6a0bce3 (patch) | |
tree | 8f70b0b1a48a5e9e2ad44d7d446936a75a936215 /drivers/net | |
parent | 76dadd76c265a0cdb5a76aa4eef03fcc9639b388 (diff) | |
download | linux-stable-c16cc0b464b8876cfd57ce1c1dbcb6f9a6a0bce3.tar.gz linux-stable-c16cc0b464b8876cfd57ce1c1dbcb6f9a6a0bce3.tar.bz2 linux-stable-c16cc0b464b8876cfd57ce1c1dbcb6f9a6a0bce3.zip |
bnx2x: Tx barriers and locks
[Resending with the proper subject. Sorry for the mess. ]
This patch is based on the RFC of Stanislaw Gruszka.
More specifically it fixes two possible races:
- One, described by Stanislaw, may lead to permanent disabling of the Tx
queue.
This is fixed by adding the smp_wmb() to propagate the BD consumer
change towards the memory.
- Second may lead to bnx2x_start_xmit() returning NETDEV_TX_BUSY.
This is fixed by taking a tx_lock() before rechecking the number of
available Tx BDs.
thanks,
vlad
Signed-off-by: Stanislaw Gruszka <sgruszka@redhat.com>
Signed-off-by: Vladislav Zolotarov <vladz@broadcom.com>
Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'drivers/net')
-rw-r--r-- | drivers/net/bnx2x_main.c | 31 |
1 files changed, 22 insertions, 9 deletions
diff --git a/drivers/net/bnx2x_main.c b/drivers/net/bnx2x_main.c index 5adf2a05246f..ed785a30e98b 100644 --- a/drivers/net/bnx2x_main.c +++ b/drivers/net/bnx2x_main.c @@ -57,8 +57,8 @@ #include "bnx2x_init_ops.h" #include "bnx2x_dump.h" -#define DRV_MODULE_VERSION "1.52.1-6" -#define DRV_MODULE_RELDATE "2010/02/16" +#define DRV_MODULE_VERSION "1.52.1-7" +#define DRV_MODULE_RELDATE "2010/02/28" #define BNX2X_BC_VER 0x040200 #include <linux/firmware.h> @@ -957,21 +957,34 @@ static int bnx2x_tx_int(struct bnx2x_fastpath *fp) fp->tx_pkt_cons = sw_cons; fp->tx_bd_cons = bd_cons; + /* Need to make the tx_bd_cons update visible to start_xmit() + * before checking for netif_tx_queue_stopped(). Without the + * memory barrier, there is a small possibility that + * start_xmit() will miss it and cause the queue to be stopped + * forever. + */ + smp_wmb(); + /* TBD need a thresh? */ if (unlikely(netif_tx_queue_stopped(txq))) { - - /* Need to make the tx_bd_cons update visible to start_xmit() - * before checking for netif_tx_queue_stopped(). Without the - * memory barrier, there is a small possibility that - * start_xmit() will miss it and cause the queue to be stopped - * forever. + /* Taking tx_lock() is needed to prevent reenabling the queue + * while it's empty. This could have happen if rx_action() gets + * suspended in bnx2x_tx_int() after the condition before + * netif_tx_wake_queue(), while tx_action (bnx2x_start_xmit()): + * + * stops the queue->sees fresh tx_bd_cons->releases the queue-> + * sends some packets consuming the whole queue again-> + * stops the queue */ - smp_mb(); + + __netif_tx_lock(txq, smp_processor_id()); if ((netif_tx_queue_stopped(txq)) && (bp->state == BNX2X_STATE_OPEN) && (bnx2x_tx_avail(fp) >= MAX_SKB_FRAGS + 3)) netif_tx_wake_queue(txq); + + __netif_tx_unlock(txq); } return 0; } |