Skip to contentSkip to conversation

[2.6.10] cdc-acm.c patches

PATCH Alexander V. Lukyanov 2 participants 2 messages Feb 24, 2005

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

Hello!

I have ZyXEL UNO modem which has usb acm interface. Starting with kernel version 2.6.8 the driver cdc-acm began to hand when the modem connection was dropped due to being idle. After the hang, no data could be read from the device /dev/usb/ttyACM0, but writting was ok. The led "data" was lit on the modem all the time after the disconnect (which indicates that modem has data to be read).

I tracked down the problem to this: acm_read_bulk was called from usb_submit_urb(acm->readurb), _inside_ acm_tty_open, and that time acm->used was 0 which led to acm_rx_tasklet being not called, and acm_read_bulk was never called again.

The problem was caused by patch in 2.6.8, which moved incrementing of acm->used to bottom of acm_tty_open. My patch to fix it is attached.

When debugging the problem I noticed that debug messages sometimes have double \n and the second one is not KERN_DEBUG level, which causes it to be emitted to display. A second patch to normalize newlines in debug output is attached.

Signature · 1 line
   Alexander.
drivers/usb/class/cdc-acm.c renamed from drivers/usb/class/cdc-acm.c.orig +2 -2
@@ -272,7 +272,7 @@ static int acm_tty_open(struct tty_struc
 
         down(&open_sem);
 
-	if (acm->used) {
+	if (acm->used++) {
 		goto done;
         }
 
@@ -296,7 +296,6 @@ static int acm_tty_open(struct tty_struc
 	tty->low_latency = 1;
 
 done:
-	acm->used++;
 	up(&open_sem);
 	return 0;
 
@@ -305,6 +304,7 @@ full_bailout:
 bail_out_and_unlink:
 	usb_kill_urb(acm->ctrlurb);
 bail_out:
+	acm->used--;
 	up(&open_sem);
 	return -EIO;
 }
drivers/usb/class/cdc-acm.c renamed from drivers/usb/class/cdc-acm.c.orig1 +7 -7
@@ -175,7 +175,7 @@ exit:
 static void acm_read_bulk(struct urb *urb, struct pt_regs *regs)
 {
 	struct acm *acm = urb->context;
-	dbg("Entering acm_read_bulk with status %d\n", urb->status);
+	dbg("Entering acm_read_bulk with status %d", urb->status);
 
 	if (!ACM_READY(acm))
 		return;
@@ -232,7 +232,7 @@ static void acm_rx_tasklet(unsigned long
 static void acm_write_bulk(struct urb *urb, struct pt_regs *regs)
 {
 	struct acm *acm = (struct acm *)urb->context;
-	dbg("Entering acm_write_bulk with status %d\n", urb->status);
+	dbg("Entering acm_write_bulk with status %d", urb->status);
 
 	if (!ACM_READY(acm))
 		goto out;
@@ -248,7 +248,7 @@ out:
 static void acm_softint(void *private)
 {
 	struct acm *acm = private;
-	dbg("Entering acm_softint.\n");
+	dbg("Entering acm_softint.");
 	
 	if (!ACM_READY(acm))
 		return;
@@ -262,7 +262,7 @@ static void acm_softint(void *private)
 static int acm_tty_open(struct tty_struct *tty, struct file *filp)
 {
 	struct acm *acm = acm_table[tty->index];
-	dbg("Entering acm_tty_open.\n");
+	dbg("Entering acm_tty_open.");
 
 	if (!acm || !acm->dev)
 		return -EINVAL;
@@ -339,7 +339,7 @@ static int acm_tty_write(struct tty_stru
 {
 	struct acm *acm = tty->driver_data;
 	int stat;
-	dbg("Entering acm_tty_write to write %d bytes,\n", count);
+	dbg("Entering acm_tty_write to write %d bytes,", count);
 
 	if (!ACM_READY(acm))
 		return -EINVAL;
@@ -352,7 +352,7 @@ static int acm_tty_write(struct tty_stru
 
 	dbg("Get %d bytes...", count);
 	memcpy(acm->write_buffer, buf, count);
-	dbg("  Successfully copied.\n");
+	dbg("  Successfully copied.");
 
 	acm->writeurb->transfer_buffer_length = count;
 	acm->writeurb->dev = acm->dev;
@@ -540,7 +540,7 @@ static int acm_probe (struct usb_interfa
 
 	if (!buflen) {
 		if (intf->cur_altsetting->endpoint->extralen && intf->cur_altsetting->endpoint->extra) {
-			dev_dbg(&intf->dev,"Seeking extra descriptors on endpoint");
+			dev_dbg(&intf->dev,"Seeking extra descriptors on endpoint\n");
 			buflen = intf->cur_altsetting->endpoint->extralen;
 			buffer = intf->cur_altsetting->endpoint->extra;
 		} else {
#1 lore raw
Quoted 3 lines

The problem was caused by patch in 2.6.8, which moved incrementing of acm->used to bottom of acm_tty_open. My patch to fix it is attached.

Your patch has been integrated into the USB tree which is regularly pushed up into the mainline.

Regards
Oliver