#define RX_LE_BYTES (RX_LE_SIZE*sizeof(struct sky2_rx_le))
#define RX_MAX_PENDING (RX_LE_SIZE/2 - 2)
#define RX_DEF_PENDING RX_MAX_PENDING
+#define RX_SKB_ALIGN 8
#define TX_RING_SIZE 512
#define TX_DEF_PENDING (TX_RING_SIZE - 1)
}
#endif
+/*
+ * It appears the hardware has a bug in the FIFO logic that
+ * cause it to hang if the FIFO gets overrun and the receive buffer
+ * is not aligned. ALso alloc_skb() won't align properly if slab
+ * debugging is enabled.
+ */
+static inline struct sk_buff *sky2_alloc_skb(unsigned int size, gfp_t gfp_mask)
+{
+ struct sk_buff *skb;
+
+ skb = alloc_skb(size + RX_SKB_ALIGN, gfp_mask);
+ if (likely(skb)) {
+ unsigned long p = (unsigned long) skb->data;
+ skb_reserve(skb,
+ ((p + RX_SKB_ALIGN - 1) & ~(RX_SKB_ALIGN - 1)) - p);
+ }
+
+ return skb;
+}
+
/*
* Allocate and setup receiver buffer pool.
* In case of 64 bit dma, there are 2X as many list elements
* available as ring entries
* and need to reserve one list element so we don't wrap around.
- *
- * It appears the hardware has a bug in the FIFO logic that
- * cause it to hang if the FIFO gets overrun and the receive buffer
- * is not aligned. This means we can't use skb_reserve to align
- * the IP header.
*/
static int sky2_rx_start(struct sky2_port *sky2)
{
for (i = 0; i < sky2->rx_pending; i++) {
struct ring_info *re = sky2->rx_ring + i;
- re->skb = dev_alloc_skb(sky2->rx_bufsize);
+ re->skb = sky2_alloc_skb(sky2->rx_bufsize, GFP_KERNEL);
if (!re->skb)
goto nomem;
} else {
struct sk_buff *nskb;
- nskb = dev_alloc_skb(sky2->rx_bufsize);
+ nskb = sky2_alloc_skb(sky2->rx_bufsize, GFP_ATOMIC);
if (!nskb)
goto resubmit;