Skip to content

Commit

Permalink
i2c: imx: Only DMA messages with I2C_M_DMA_SAFE flag set
Browse files Browse the repository at this point in the history
[ Upstream commit d36678f ]

Recent changes to the DMA code has resulting in the IMX driver failing
I2C transfers when the buffer has been vmalloc. Only perform DMA
transfers if the message has the I2C_M_DMA_SAFE flag set, indicating
the client is providing a buffer which is DMA safe.

This is a minimal fix for stable. The I2C core provides helpers to
allocate a bounce buffer. For a fuller fix the master should make use
of these helpers.

Fixes: 4544b9f ("dma-mapping: Add vmap checks to dma_map_single()")
Signed-off-by: Andrew Lunn <andrew@lunn.ch>
Acked-by: Oleksij Rempel <o.rempel@pengutronix.de>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
  • Loading branch information
lunn authored and gregkh committed Dec 8, 2022
1 parent 2bfc02b commit 324ec6e
Showing 1 changed file with 4 additions and 2 deletions.
6 changes: 4 additions & 2 deletions drivers/i2c/busses/i2c-imx.c
Original file line number Diff line number Diff line change
Expand Up @@ -1132,7 +1132,8 @@ static int i2c_imx_read(struct imx_i2c_struct *i2c_imx, struct i2c_msg *msgs,
int i, result;
unsigned int temp;
int block_data = msgs->flags & I2C_M_RECV_LEN;
int use_dma = i2c_imx->dma && msgs->len >= DMA_THRESHOLD && !block_data;
int use_dma = i2c_imx->dma && msgs->flags & I2C_M_DMA_SAFE &&
msgs->len >= DMA_THRESHOLD && !block_data;

dev_dbg(&i2c_imx->adapter.dev,
"<%s> write slave address: addr=0x%x\n",
Expand Down Expand Up @@ -1298,7 +1299,8 @@ static int i2c_imx_xfer_common(struct i2c_adapter *adapter,
result = i2c_imx_read(i2c_imx, &msgs[i], is_lastmsg, atomic);
} else {
if (!atomic &&
i2c_imx->dma && msgs[i].len >= DMA_THRESHOLD)
i2c_imx->dma && msgs[i].len >= DMA_THRESHOLD &&
msgs[i].flags & I2C_M_DMA_SAFE)
result = i2c_imx_dma_write(i2c_imx, &msgs[i]);
else
result = i2c_imx_write(i2c_imx, &msgs[i], atomic);
Expand Down

0 comments on commit 324ec6e

Please sign in to comment.