cxgb3: map entire Rx page, feed map+offset to Rx ring.

DMA mapping can be expensive in the presence of iommus.
Reduce the Rx iommu activity by mapping an entire page, and provide the H/W
the mapped address + offset of the current page chunk.
Reserve bits at the end of the page to track mapping references, so the page
can be unmapped.

Signed-off-by: Divy Le Ray <divy@chelsio.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
This commit is contained in:
Divy Le Ray
2009-03-26 16:39:29 +00:00
committed by David S. Miller
parent 952cdf333f
commit 5e68b772e6
2 changed files with 107 additions and 36 deletions

View File

@@ -85,6 +85,8 @@ struct fl_pg_chunk {
struct page *page;
void *va;
unsigned int offset;
u64 *p_cnt;
DECLARE_PCI_UNMAP_ADDR(mapping);
};
struct rx_desc;
@@ -101,6 +103,7 @@ struct sge_fl { /* SGE per free-buffer list state */
struct fl_pg_chunk pg_chunk;/* page chunk cache */
unsigned int use_pages; /* whether FL uses pages or sk_buffs */
unsigned int order; /* order of page allocations */
unsigned int alloc_size; /* size of allocated buffer */
struct rx_desc *desc; /* address of HW Rx descriptor ring */
struct rx_sw_desc *sdesc; /* address of SW Rx descriptor ring */
dma_addr_t phys_addr; /* physical address of HW ring start */