Skip to contentSkip to conversation

USB: g_file_storage: fix handling zero-length packet

PATCH In review sangsu 3 participants 3 messages Sep 6, 2008 – Sep 7, 2008 USB

lore
Share & tools
lore

sangsu

author
#0 lore raw

PATCH 1/1 USB: g_file_storage: fix handling zero-length packet

According to "CH 5.5.3. Control Transfer Packet Size Constraints", Zero-length packet should be sent when the last data payload size is equal to the endpoint's MaxPacketSize. This patch helps g_file_storage send zero-length packet properly.

This is tested in s3c2440/s5c7329 architectures and works well.

---
driver/usb/gadget/file_storage.c | 3++-

1 files changed, 2 insertion(+), 1 deletions(-)
drivers/usb/gadget/file_storage.c +0 -0
@@ -1463,7 +1463,8 @@
	if (rc >= 0 && rc != DELAYED_STATUS) {
		rc = min(rc, w_length);
		fsg->ep0req->length = rc;
-		fsg->ep0req->zero = rc < w_length;
+		fsg->ep0req->zero = rc < w_length
+			&& (rc % gadget->ep0->maxpacket) == 0;
		fsg->ep0req_name = (ctrl->bRequestType & USB_DIR_IN ?
				"ep0-in" : "ep0-out");
		rc = ep0_queue(fsg);
#1 lore raw
Quoted 3 lines · sangsu

According to "CH 5.5.3. Control Transfer Packet Size Constraints", Zero-length packet should be sent when the last data payload size is equal to the endpoint's MaxPacketSize.

Not so. 5.5.3 actually says this:

The Data stage of a control transfer from an endpoint to the host is complete when the endpoint does one of the following:

- Has transferred exactly the amount of data specified during
the Setup stage

- Transfers a packet with a payload size less than
wMaxPacketSize or transfers a zero-length packet

This clearly indicates that if the amount of data specified in the Setup stage is a multiple of the wMaxPacketSize then there should not be a zero-length packet, because the transfer ends as soon as the last wMaxPacketSize packet is sent.

Quoted 13 lines

This patch helps
g_file_storage send zero-length packet properly.

This is tested in s3c2440/s5c7329 architectures and works well.

---
driver/usb/gadget/file_storage.c | 3++-

1 files changed, 2 insertion(+), 1 deletions(-)

drivers/usb/gadget/file_storage.c line 1470

Quoted 9 lines
drivers/usb/gadget/file_storage.c +2 -1
@@ -1463,7 +1463,8 @@
 	if (rc >= 0 && rc != DELAYED_STATUS) {
 		rc = min(rc, w_length);
 		fsg->ep0req->length = rc;
-		fsg->ep0req->zero = rc < w_length;
+		fsg->ep0req->zero = rc < w_length
+			&& (rc % gadget->ep0->maxpacket) == 0;
 		fsg->ep0req_name = (ctrl->bRequestType & USB_DIR_IN ?
 				"ep0-in" : "ep0-out");
 		rc = ep0_queue(fsg);

Firstly, this patch does not do what your description says it does.

Secondly, the patch is not needed. g_file_storage doesn't have to check whether the data size is a multiple of MaxPacketSize because the device controller driver already is supposed to make that check.

Alan Stern

#2 lore raw
Quoted 3 lines · Alan Stern

Secondly, the patch is not needed.  g_file_storage doesn't have to check whether the data size is a multiple of MaxPacketSize because the device controller driver already is supposed to make that check.

Exactly.

The only fuzziness here is what to do if the host misbehaves and sends another IN packet when all its data has been sent, instead of the OUT/STATUS packet. (Host error...)

Most peripheral controller hardware is well enough designed that it can detect this and send a zero length packet rather than STALL the transfer. It's polite to do so, because that host bug isn't exactly unknown. (Some old versions of Linux did that...)