rsi: Fix failure to load firmware after memory leak fix and fix the leak

Fixes commit eae79b4f3e82ca63a53478a161b190a0d38fe526 ("rsi: fix memory leak in rsi_load_ta_instructions()") which stopped the driver from functioning.

Firmware data has been allocated using vmalloc(), resulting in memory that cannot be used for DMA. Hence the firmware was first copied to a buffer allocated with kmalloc() in the original code. This patch reverts the commit and only calls "kfree()" to release the buffer after sending the data. This fixes the memory leak without breaking the driver.

Add a comment to the kmemdup() calls to explain why this is done.

Tested on a Topic Miami-Florida board which contains the rsi SDIO chip.

Also added the same kfree() call to the USB glue driver. This was not tested on actual hardware though, as I only have the SDIO version.

Signed-off-by Mike Looijmans <[email protected]>

Cc [email protected]

---
 drivers/net/wireless/rsi/rsi_91x_sdio_ops.c | 6 +++++-
 drivers/net/wireless/rsi/rsi_91x_usb_ops.c  | 2 ++
 2 files changed, 7 insertions(+), 1 deletion(-)

drivers/net/wireless/rsi/rsi_91x_sdio_ops.c +5 -1

index b6cc9ff..5c37a71 100644
@@ -172,6 +172,7 @@ static int rsi_load_ta_instructions(struct rsi_common *common)
 		(struct rsi_91x_sdiodev *)adapter->rsi_dev;
 	u32 len;
 	u32 num_blocks;
+	const u8 *fw;
 	const struct firmware *fw_entry = NULL;
 	u32 block_size = dev->tx_blk_size;
 	int status = 0;

@@ -200,6 +201,8 @@ static int rsi_load_ta_instructions(struct rsi_common *common)
 		return status;
 	}
 
+	/* Copy firmware into DMA-accessible memory */
+	fw = kmemdup(fw_entry->data, fw_entry->size, GFP_KERNEL);
 	len = fw_entry->size;
 
 	if (len % 4)

@@ -210,7 +213,8 @@ static int rsi_load_ta_instructions(struct rsi_common *common)
 	rsi_dbg(INIT_ZONE, "%s: Instruction size:%d\n", __func__, len);
 	rsi_dbg(INIT_ZONE, "%s: num blocks: %d\n", __func__, num_blocks);
 
-	status = rsi_copy_to_card(common, fw_entry->data, len, num_blocks);
+	status = rsi_copy_to_card(common, fw, len, num_blocks);
+	kfree(fw);
 	release_firmware(fw_entry);
 	return status;
 }

drivers/net/wireless/rsi/rsi_91x_usb_ops.c +2 -0

index 1106ce7..088e28e 100644
@@ -146,6 +146,7 @@ static int rsi_load_ta_instructions(struct rsi_common *common)
 		return status;
 	}
 
+	/* Copy firmware into DMA-accessible memory */
 	fw = kmemdup(fw_entry->data, fw_entry->size, GFP_KERNEL);
 	len = fw_entry->size;
 

@@ -158,6 +159,7 @@ static int rsi_load_ta_instructions(struct rsi_common *common)
 	rsi_dbg(INIT_ZONE, "%s: num blocks: %d\n", __func__, num_blocks);
 
 	status = rsi_copy_to_card(common, fw, len, num_blocks);
+	kfree(fw);
 	release_firmware(fw_entry);
 	return status;
 }

Fixes commit eae79b4f3e82ca63a53478a161b190a0d38fe526 ("rsi: fix memory leak in rsi_load_ta_instructions()") which stopped the driver from functioning.

Firmware data has been allocated using vmalloc(), resulting in memory that cannot be used for DMA. Hence the firmware was first copied to a buffer allocated with kmalloc() in the original code. This patch reverts the commit and only calls "kfree()" to release the buffer after sending the data. This fixes the memory leak without breaking the driver.

Add a comment to the kmemdup() calls to explain why this is done.

Tested on a Topic Miami-Florida board which contains the rsi SDIO chip.

Also added the same kfree() call to the USB glue driver. This was not tested on actual hardware though, as I only have the SDIO version.

Signed-off-by Mike Looijmans <[email protected]>

Cc [email protected]

Quoted 2 lines · Mike Looijmans

Fixes commit eae79b4f3e82ca63a53478a161b190a0d38fe526 ("rsi: fix memory leak in rsi_load_ta_instructions()") which stopped the driver from functioning.

You can abbreviate the commit id:

Fixes commit eae79b4f3e82 ("rsi: fix memory leak in rsi_load_ta_instructions()") which stopped the driver from functioning.

Quoted 6 lines

Firmware data has been allocated using vmalloc(), resulting in memory that cannot be used for DMA. Hence the firmware was first copied to a buffer allocated with kmalloc() in the original code. This patch reverts the commit and only calls "kfree()" to release the buffer after sending the data. This fixes the memory leak without breaking the driver.

Quoted 8 lines

Add a comment to the kmemdup() calls to explain why this is done.

Tested on a Topic Miami-Florida board which contains the rsi SDIO chip.

Also added the same kfree() call to the USB glue driver. This was not tested on actual hardware though, as I only have the SDIO version.

Signed-off-by Mike Looijmans <[email protected]>

Add this before Signed-off-by line:

Fixes eae79b4f3e82 ("rsi: fix memory leak in rsi_load_ta_instructions()")

Quoted 1 line

Cc [email protected]

Also no need to send email to [email protected] list, this line is enough and the stable team will pick the commit automatically.

Fixes commit eae79b4f3e82 ("rsi: fix memory leak in rsi_load_ta_instructions()") which stopped the driver from functioning.

Firmware data has been allocated using vmalloc(), resulting in memory that cannot be used for DMA. Hence the firmware was first copied to a buffer allocated with kmalloc() in the original code. This patch reverts the commit and only calls "kfree()" to release the buffer after sending the data. This fixes the memory leak without breaking the driver.

Add a comment to the kmemdup() calls to explain why this is done, and abort if memory allocation fails.

Tested on a Topic Miami-Florida board which contains the rsi SDIO chip.

Also added the same kfree() call to the USB glue driver. This was not tested on actual hardware though, as I only have the SDIO version.

Fixes eae79b4f3e82 ("rsi: fix memory leak in rsi_load_ta_instructions()")

Signed-off-by Mike Looijmans <[email protected]>

Cc [email protected]