Skip to contentSkip to conversation

ioatdma: fix "ioatdma frees DMA memory with wrong function"

PATCH In review Maciej Sosnowski 2 participants 2 messages Apr 23, 2009 – May 12, 2009

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 ioatdma: fix "ioatdma frees DMA memory with wrong function"

as reported by Alexander Beregalov <[email protected]>

ioatdma 0000:00:08.0: DMA-API: device driver frees DMA memory with wrong function [device address=0x000000007f76f800] [size=2000 bytes] [map
ped as single] [unmapped as page]

The ioatdma driver was unmapping all regions
(either allocated as page or single) using unmap_page.
This patch lets dma driver recognize if unmap_single or unmap_page should be used. It introduces two new dma control flags:
DMA_COMPL_SRC_UNMAP_SINGLE and DMA_COMPL_DEST_UNMAP_SINGLE.
They should be set to indicate dma driver to do dma-unmapping as single (first one for the source, tha latter for the destination).
If respective flag is not set, the driver assumes dma-unmapping as page.

---

 drivers/dma/dmaengine.c   |   17 +++++++++++------
 drivers/dma/ioat_dma.c    |   45 ++++++++++++++++++++++++++++-----------------
 include/linux/dmaengine.h |    6 ++++++
 3 files changed, 45 insertions(+), 23 deletions(-)
drivers/dma/dmaengine.c +11 -6
index 92438e9..5a87384 100644
@@ -804,11 +804,14 @@ dma_async_memcpy_buf_to_buf(struct dma_c
 	dma_addr_t dma_dest, dma_src;
 	dma_cookie_t cookie;
 	int cpu;
+	unsigned long flags;
 
 	dma_src = dma_map_single(dev->dev, src, len, DMA_TO_DEVICE);
 	dma_dest = dma_map_single(dev->dev, dest, len, DMA_FROM_DEVICE);
-	tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len,
-					 DMA_CTRL_ACK);
+	flags = DMA_CTRL_ACK |
+		DMA_COMPL_SRC_UNMAP_SINGLE |
+		DMA_COMPL_DEST_UNMAP_SINGLE;
+	tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len, flags);
 
 	if (!tx) {
 		dma_unmap_single(dev->dev, dma_src, len, DMA_TO_DEVICE);
@@ -850,11 +853,12 @@ dma_async_memcpy_buf_to_pg(struct dma_ch
 	dma_addr_t dma_dest, dma_src;
 	dma_cookie_t cookie;
 	int cpu;
+	unsigned long flags;
 
 	dma_src = dma_map_single(dev->dev, kdata, len, DMA_TO_DEVICE);
 	dma_dest = dma_map_page(dev->dev, page, offset, len, DMA_FROM_DEVICE);
-	tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len,
-					 DMA_CTRL_ACK);
+	flags = DMA_CTRL_ACK | DMA_COMPL_SRC_UNMAP_SINGLE;
+	tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len, flags);
 
 	if (!tx) {
 		dma_unmap_single(dev->dev, dma_src, len, DMA_TO_DEVICE);
@@ -898,12 +902,13 @@ dma_async_memcpy_pg_to_pg(struct dma_cha
 	dma_addr_t dma_dest, dma_src;
 	dma_cookie_t cookie;
 	int cpu;
+	unsigned long flags;
 
 	dma_src = dma_map_page(dev->dev, src_pg, src_off, len, DMA_TO_DEVICE);
 	dma_dest = dma_map_page(dev->dev, dest_pg, dest_off, len,
 				DMA_FROM_DEVICE);
-	tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len,
-					 DMA_CTRL_ACK);
+	flags = DMA_CTRL_ACK;
+	tx = dev->device_prep_dma_memcpy(chan, dma_dest, dma_src, len, flags);
 
 	if (!tx) {
 		dma_unmap_page(dev->dev, dma_src, len, DMA_TO_DEVICE);
drivers/dma/ioat_dma.c +28 -17

64 lines not on this page · raw message

include/linux/dmaengine.h +6 -0
index 2e2aa3d..ffefba8 100644
@@ -78,12 +78,18 @@ #define DMA_TX_TYPE_END (DMA_SLAVE + 1)
  * 	dependency chains
  * @DMA_COMPL_SKIP_SRC_UNMAP - set to disable dma-unmapping the source buffer(s)
  * @DMA_COMPL_SKIP_DEST_UNMAP - set to disable dma-unmapping the destination(s)
+ * @DMA_COMPL_SRC_UNMAP_SINGLE - set to do the source dma-unmapping as single
+ * 	(if not set, do the source dma-unmapping as page)
+ * @DMA_COMPL_DEST_UNMAP_SINGLE - set to do the destination dma-unmapping as single
+ * 	(if not set, do the destination dma-unmapping as page)
  */
 enum dma_ctrl_flags {
 	DMA_PREP_INTERRUPT = (1 << 0),
 	DMA_CTRL_ACK = (1 << 1),
 	DMA_COMPL_SKIP_SRC_UNMAP = (1 << 2),
 	DMA_COMPL_SKIP_DEST_UNMAP = (1 << 3),
+	DMA_COMPL_SRC_UNMAP_SINGLE = (1 << 4),
+	DMA_COMPL_DEST_UNMAP_SINGLE = (1 << 5),
 };
 
 /**
#1 lore raw

2009/4/23 Maciej Sosnowski <[email protected]>:

Quoted 16 lines

as reported by Alexander Beregalov <[email protected]>

ioatdma 0000:00:08.0: DMA-API: device driver frees DMA memory with wrong function [device address=0x000000007f76f800] [size=2000 bytes] [map
ped as single] [unmapped as page]

The ioatdma driver was unmapping all regions
(either allocated as page or single) using unmap_page.
This patch lets dma driver recognize if unmap_single or unmap_page should be used. It introduces two new dma control flags:
DMA_COMPL_SRC_UNMAP_SINGLE and DMA_COMPL_DEST_UNMAP_SINGLE.
They should be set to indicate dma driver to do dma-unmapping as single (first one for the source, tha latter for the destination).
If respective flag is not set, the driver assumes dma-unmapping as page.

Quoted 5 lines

---

Will it be sent to Linus ?