Re: Interoperability problem with g_file_storage.

On Fri, 24 Aug 2007, Samuel Hangouët wrote:

> Hi,
> 
> I'm currently using the driver g_file_storage on a gumstix product (see 
> http://gumstix.org).
> 
> This driver works fine when the option stall=N is set.
> 
> However, I encountered a problem when I plug it in a JVC micro HiFi.
> 
> With debug mode enabled, g_file_storage prints the following message in 
> kernel log:
> 
> 
>  g_file_storage gadget: bulk-out, length 31:
>       0:  55 53 42 43 03 00 00 00  12 00 00 00 80 00 05 03
>      10:  00 00 00 12 00 00 00 10  00 00 00 00 00 00 00
>  g_file_storage gadget: non-meaningful CBW: lun = 0, flags = 0x80, cmdlen 5
> 
> (see full log at the end of this email)
> 
> In the command send by the HiFi equipment, the length of the CBWCB is 
> equal to 5,
> which makes g_file_storage module reject the message in file_storage.c 
> at line 3009 :
> 
>         /* Is the CBW meaningful? */
>         if (cbw->Lun >= MAX_LUNS || cbw->Flags & ~USB_BULK_IN_FLAG ||
>                         cbw->Length < 6 || cbw->Length > MAX_COMMAND_SIZE) {
>                 DBG(fsg, "non-meaningful CBW: lun = %u, flags = 0x%x, "
>                                 "cmdlen %u\n",
>                                 cbw->Lun, cbw->Flags, cbw->Length);
> 
> I wonder why there is a 6 in here.
> 
> Indeed, the BBB Mass-Storage standard specifies:
> 
> """
> bCBWCBLength:
>    The valid length of the CBWCB in bytes. This defines the valid length 
> of the command block. The
>    only legal values are 1 through 16 (01h through 10h). All other 
> values are reserved.
> """

You looked at the wrong part of the standard.  6.2.2 defines when the 
CBW is meaningful; the third requirement is:

	Both bCBWCBLength and the content of the CBWCB are in
	accordance with bInterfaceSubClass.

In this case the relevant standard is SCSI SPC-2.  It includes the 
definition of the REQUEST SENSE command, and the definition says that 
the command length is 6 bytes, not 5.  In fact, none of the commands 
accepted by g_file_storage have length < 6.

> and if I replaced this 6 by 1, everything works fine.

Really?  It should failed this test in check_command():

	/* Verify the length of the command itself */
	if (cmnd_size != fsg->cmnd_size) {

		/* Special case workaround: MS-Windows issues REQUEST SENSE
		 * with cbw->Length == 12 (it should be 6). */
		if (fsg->cmnd[0] == SC_REQUEST_SENSE && fsg->cmnd_size == 12)
			cmnd_size = fsg->cmnd_size;
		else {
			fsg->phase_error = 1;
			return -EINVAL;
		}
	}

Here cmnd_size would be 6 and fsg->cmnd_size would be 5.

Alan Stern


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/
_______________________________________________
[email protected]
To unsubscribe, use the last form field at:
https://lists.sourceforge.net/lists/listinfo/linux-usb-users