On Friday, September 10, 2021 5:19:58 PM CEST Fabio M. De Francesco wrote:

On Monday, September 6, 2021 4:07:26 PM CEST Greg Kroah-Hartman wrote:

quoted

On Sun, Sep 05, 2021 at 12:00:47AM +0200, Fabio M. De Francesco wrote:

quoted

Shorten the calls chain of rtw_read8/16/32() down to the actual reads.
For this purpose unify the three usb_read8/16/32 into the new
usb_read(); make the latter parameterizable with 'size'; embed most of
the code of usbctrl_vendorreq() into usb_read() and use in it the new
usb_control_msg_recv() API of USB Core.

Suggested-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Co-developed-by: Pavel Skripkin <redacted>
Signed-off-by: Pavel Skripkin <redacted>
Signed-off-by: Fabio M. De Francesco <redacted>
---

[...]

+	while (++vendorreq_times <= MAX_USBCTRL_VENDORREQ_TIMES) {
+		status = usb_control_msg_recv(udev, 0, 

REALTEK_USB_VENQT_CMD_REQ,

quoted

quoted

+					      

REALTEK_USB_VENQT_READ, value,

quoted

quoted

+					      

REALTEK_USB_VENQT_CMD_IDX, io_buf,

quoted

quoted

+					      size, 

RTW_USB_CONTROL_MSG_TIMEOUT,

quoted

quoted

+					      GFP_KERNEL);
+		if (!status) {   /*  Success this control transfer. */

Comments go on the next line.

quoted

+			rtw_reset_continual_urb_error(dvobjpriv);
+			memcpy(data, io_buf, size);
+		} else { /*  error cases */

Again, next line for the comment.

quoted

+			DBG_88E("reg 0x%x, usb %s %u fail, status:

%d vendorreq_times:%d\n",

quoted

quoted

+				value, "read", size, status, 

vendorreq_times);

quoted

These should be removed eventually...

quoted

+
+			if (status == (-ESHUTDOWN) || status == -

ENODEV) {

quoted

quoted

+				adapt->bSurpriseRemoved = true;

Odd, but ok...

I'm not so sure that it is OK. Please correct me if I'm wrong...

The calls chain from usb_control_msg_recv() seems to be the following:

usb_control_msg_recv/send()
        -> usb_control_msg()
                -> usb_internal_control_msg()
                        -> usb_start_wait_urb()
                                -> usb_submit_urb()

Each of the above functions could fail for different reasons and if so they 
return the errors up to the first caller into "status". I can find no lines 
of code where the above-mentioned functions set and return -ESHUTDOWN.

Unless I'm missing something obvious, "status" is a non-shared variable. 

The 

variables that are assigned with errors in all five of the above-mentioned 
functions are also local (non shared) variables.

To summarize: how could "status" be assigned -ESHUTDOWN? Is any point in 

the 

chain that value assigned by a concurrent thread to a shared variable and 
then returned up to the caller (i.e., usb_control_msg_recv())?

Since the code has this "if (status == (-ESHUTDOWN) || ...)" it expects 

that 

sometimes it could be 'true', so I'm 100% sure that I can't see where my 
argument is not valid... :(

Sorry, please disregard my previous message.

I found that, somewhere about a couple of function deeper in the chain,  the 
-ESHUTDOWN error code can indeed be returned. I had to read again and again 
every line of the chain until I saw that.

Fabio

Can someone please help me to understand this topic?

Thanks,

Fabio

quoted

quoted

[...]