mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
* [PATCH 1/2] staging: et131x: Avoid unnecessary calculations in for loop
@ 2012-11-12 21:16 Mark Einon
  2012-11-12 21:16 ` [PATCH 2/2] staging: et131x: Remove unnecessary DMA address alignment code Mark Einon
  0 siblings, 1 reply; 2+ messages in thread
From: Mark Einon @ 2012-11-12 21:16 UTC (permalink / raw)
  To: gregkh; +Cc: devel, linux-kernel, Mark Einon

In et131x_rx_dma_memory_alloc(), we loop over fbr_entries - the values of
fbr_align and fbr_chunksize calculated in the loop do not depend on the
loop counter and are the same for all loop iterations.

Take the calculations of these values out of the loop, as we only
need to calculate them once per loop, not once per loop iteration.

Signed-off-by: Mark Einon <mark.einon@gmail.com>
---
 drivers/staging/et131x/et131x.c |   20 +++++++++++---------
 1 file changed, 11 insertions(+), 9 deletions(-)

diff --git a/drivers/staging/et131x/et131x.c b/drivers/staging/et131x/et131x.c
index 69a0729..4f68646 100644
--- a/drivers/staging/et131x/et131x.c
+++ b/drivers/staging/et131x/et131x.c
@@ -2266,7 +2266,9 @@ static int et131x_rx_dma_memory_alloc(struct et131x_adapter *adapter)
 	u8 id;
 	u32 i, j;
 	u32 bufsize;
-	u32 pktstat_ringsize, fbr_chunksize;
+	u32 pktstat_ringsize;
+	u32 fbr_chunksize;
+	u32 fbr_align;
 	struct rx_ring *rx_ring;
 
 	/* Setup some convenience pointers */
@@ -2341,10 +2343,17 @@ static int et131x_rx_dma_memory_alloc(struct et131x_adapter *adapter)
 	}
 
 	for (id = 0; id < NUM_FBRS; id++) {
+		if (id == 0 && rx_ring->fbr[id]->buffsize > 4096)
+				fbr_align = 4096;
+		else
+				fbr_align = rx_ring->fbr[id]->buffsize;
+
+		fbr_chunksize = (FBR_CHUNKS *
+				 rx_ring->fbr[id]->buffsize) + fbr_align - 1;
+
 		for (i = 0; i < (rx_ring->fbr[id]->num_entries / FBR_CHUNKS); i++) {
 			dma_addr_t fbr_tmp_physaddr;
 			dma_addr_t fbr_offset;
-			u32 fbr_align;
 
 			/* This code allocates an area of memory big enough for
 			 * N free buffers + (buffer_size - 1) so that the
@@ -2353,13 +2362,6 @@ static int et131x_rx_dma_memory_alloc(struct et131x_adapter *adapter)
 			 * effect would be to double the size of FBR0. By
 			 * allocating N buffers at once, we reduce this overhead
 			 */
-			if (id == 0 && rx_ring->fbr[id]->buffsize > 4096)
-					fbr_align = 4096;
-			else
-					fbr_align = rx_ring->fbr[id]->buffsize;
-
-			fbr_chunksize = (FBR_CHUNKS *
-				rx_ring->fbr[id]->buffsize) + fbr_align - 1;
 			rx_ring->fbr[id]->mem_virtaddrs[i] = dma_alloc_coherent(
 					&adapter->pdev->dev, fbr_chunksize,
 					&rx_ring->fbr[id]->mem_physaddrs[i],
-- 
1.7.9.5


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2012-11-12 21:16 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2012-11-12 21:16 [PATCH 1/2] staging: et131x: Avoid unnecessary calculations in for loop Mark Einon
2012-11-12 21:16 ` [PATCH 2/2] staging: et131x: Remove unnecessary DMA address alignment code Mark Einon

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox

all inboxes | Powered by JetHome®