Thread (194 messages) 194 messages, 10 authors, 2012-06-27

[ 037/175] USB: ums_realtek: do not use stack memory for DMA in __do_config_autodelink

From: Greg KH <gregkh@linuxfoundation.org>
Date: 2012-03-30 21:56:45
Also in: lkml

3.3-stable review patch.  If anyone has any objections, please let me know.

------------------

From: Josh Boyer <redacted>

commit 4898e07174b79013afd2b422ef6c4336ef8e6783 upstream.

__do_config_autodelink passes the data variable to the transport function.
If the calling functions pass a stack variable, this will eventually trigger
a DMA-API debug backtrace for mapping stack memory in the DMA buffer.  Fix
this by calling kmemdup for the passed data instead.

Signed-off-by: Josh Boyer <redacted>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

---
 drivers/usb/storage/realtek_cr.c |    8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
--- a/drivers/usb/storage/realtek_cr.c
+++ b/drivers/usb/storage/realtek_cr.c
@@ -507,9 +507,14 @@ static int __do_config_autodelink(struct
 {
 	int retval;
 	u8 cmnd[12] = {0};
+	u8 *buf;
 
 	US_DEBUGP("%s, addr = 0xfe47, len = %d\n", __FUNCTION__, len);
 
+	buf = kmemdup(data, len, GFP_NOIO);
+	if (!buf)
+		return USB_STOR_TRANSPORT_ERROR;
+
 	cmnd[0] = 0xF0;
 	cmnd[1] = 0x0E;
 	cmnd[2] = 0xfe;
@@ -517,7 +522,8 @@ static int __do_config_autodelink(struct
 	cmnd[4] = (u8)(len >> 8);
 	cmnd[5] = (u8)len;
 
-	retval = rts51x_bulk_transport_special(us, 0, cmnd, 12, data, len, DMA_TO_DEVICE, NULL);
+	retval = rts51x_bulk_transport_special(us, 0, cmnd, 12, buf, len, DMA_TO_DEVICE, NULL);
+	kfree(buf);
 	if (retval != USB_STOR_TRANSPORT_GOOD) {
 		return -EIO;
 	}

Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help