[PATCH 2/6] netdev: octeon_mgmt: Fix race condition freeing TX buffers.

[David Daney <ddaney@caviumnetworks.com>]

Under heavy load the TX cleanup tasklet and xmit threads would race
and try to free too many buffers.

Signed-off-by: David Daney <ddaney@caviumnetworks.com>
---
 drivers/net/octeon/octeon_mgmt.c |   11 +++++++++--
 1 files changed, 9 insertions(+), 2 deletions(-)

diff --git a/drivers/net/octeon/octeon_mgmt.c b/drivers/net/octeon/octeon_mgmt.c
index bbbd737..b975a2f 100644
--- a/drivers/net/octeon/octeon_mgmt.c
+++ b/drivers/net/octeon/octeon_mgmt.c
@@ -189,12 +189,19 @@ static void octeon_mgmt_clean_tx_buffers(struct 
octeon_mgmt *p)
 
        mix_orcnt.u64 = cvmx_read_csr(CVMX_MIXX_ORCNT(port));
        while (mix_orcnt.s.orcnt) {
+               spin_lock_irqsave(&p->tx_list.lock, flags);
+
+               mix_orcnt.u64 = cvmx_read_csr(CVMX_MIXX_ORCNT(port));
+
+               if (mix_orcnt.s.orcnt == 0) {
+                       spin_unlock_irqrestore(&p->tx_list.lock, flags);
+                       break;
+               }
+
                dma_sync_single_for_cpu(p->dev, p->tx_ring_handle,
                                        
ring_size_to_bytes(OCTEON_MGMT_TX_RING_SIZE),
                                        DMA_BIDIRECTIONAL);
 
-               spin_lock_irqsave(&p->tx_list.lock, flags);
-
                re.d64 = p->tx_ring[p->tx_next_clean];
                p->tx_next_clean =
                        (p->tx_next_clean + 1) % OCTEON_MGMT_TX_RING_SIZE;
-- 
1.6.6.1