If the swiotlb maps a multi-slab region, swiotlb_sync_single_range()
can be invoked to sync a sub-region which does not include the first
slab. Unfortunately io_tlb_orig_addr[] is only initialised for the
first slab, and hence the call to sync_single() will read a garbage
orig_addr in this case.
This patch fixes the issue by initialising all mapped slabs in
io_tlb_orig_addr[]. It also correctly adjusts the buffer pointer in
sync_single() to handle the case that the given dma_addr is not
aligned on a slab boundary.
Signed-off-by: Keir Fraser <keir.fraser@cl.cam.ac.uk>
(Please Cc me in any response: I am not subscribed to lkml)
diff -urpN linux-2.6.22/lib/swiotlb.c linux-2.6.22-new/lib/swiotlb.c
--- linux-2.6.22/lib/swiotlb.c 2007-07-10 11:38:33.000000000 +0100
+++ linux-2.6.22-new/lib/swiotlb.c 2007-07-10 11:48:44.000000000 +0100
@@ -357,7 +357,8 @@ map_single(struct device *hwdev, char *b
* This is needed when we sync the memory. Then we sync the buffer if
* needed.
*/
- io_tlb_orig_addr[index] = buffer;
+ for (i = 0; i < nslots; i++)
+ io_tlb_orig_addr[index+i] = buffer + (i << IO_TLB_SHIFT);
if (dir == DMA_TO_DEVICE || dir == DMA_BIDIRECTIONAL)
memcpy(dma_addr, buffer, size);
@@ -418,6 +419,8 @@ sync_single(struct device *hwdev, char *
int index = (dma_addr - io_tlb_start) >> IO_TLB_SHIFT;
char *buffer = io_tlb_orig_addr[index];
+ buffer += ((unsigned long)dma_addr & ((1 << IO_TLB_SHIFT) - 1));
+
switch (target) {
case SYNC_FOR_CPU:
if (likely(dir == DMA_FROM_DEVICE || dir == DMA_BIDIRECTIONAL))
-
