On Mon, Jun 13, 2016 at 11:02:19AM +0200, Oliver Neukum wrote:
> On Mon, 2016-06-13 at 00:37 +0200, Ladislav Michl wrote:
> > On Sun, Jun 12, 2016 at 11:03:45PM +0200, Ladislav Michl wrote:
> > > Once ttyACM0 starts behave strangely, read() returns only what's in buffer before
> > > ttyACM0 was opened and then hangs infinitely. As this bug is hard to trigger, has
> > > anyone clue where to start debugging?
> >
> > Forgot to mention, once this happens "usb 3-1.4: clear tt 1 (9061) error -75"
> > starts showing in the syslog. Also "cdc_acm 3-1.4.3:1.0: failed to set dtr/rts",
> > but this one is there normally.
>
> The translation transactor in the hub is reporting an error.
> That should be reported to the driver, but there is no good
> error handling.
Hmm, assuming that error comes from ehci-q.c:qtd_copy_status() there
is one more disturbing FIXME:
static int qtd_copy_status (
struct ehci_hcd *ehci,
struct urb *urb,
size_t length,
u32 token
)
{
int status = -EINPROGRESS;
/* count IN/OUT bytes, not SETUP (even short packets) */
if (likely (QTD_PID (token) != 2))
urb->actual_length += length - QTD_LENGTH (token);
/* don't modify error codes */
if (unlikely(urb->unlinked))
return status;
/* force cleanup after short read; not always an error */
if (unlikely (IS_SHORT_READ (token)))
status = -EREMOTEIO;
/* serious "can't proceed" faults reported by the hardware */
if (token & QTD_STS_HALT) {
if (token & QTD_STS_BABBLE) {
/* FIXME "must" disable babbling device's port too */
status = -EOVERFLOW;
/* CERR nonzero + halt --> stall */
} else if (QTD_CERR(token)) {
status = -EPIPE;
/* In theory, more than one of the following bits can be set
* since they are sticky and the transaction is retried.
* Which to test first is rather arbitrary.
*/
} else if (token & QTD_STS_MMF) {
/* fs/ls interrupt xfer missed the complete-split */
status = -EPROTO;
} else if (token & QTD_STS_DBE) {
status = (QTD_PID (token) == 1) /* IN ? */
? -ENOSR /* hc couldn't read data */
: -ECOMM; /* hc couldn't write data */
} else if (token & QTD_STS_XACT) {
/* timeout, bad CRC, wrong PID, etc */
ehci_dbg(ehci, "devpath %s ep%d%s 3strikes\n",
urb->dev->devpath,
usb_pipeendpoint(urb->pipe),
usb_pipein(urb->pipe) ? "in" : "out");
status = -EPROTO;
} else { /* unknown */
status = -EPROTO;
}
}
return status;
}
> static void acm_read_bulk_callback(struct urb *urb)
> {
> struct acm_rb *rb = urb->context;
> struct acm *acm = rb->instance;
> unsigned long flags;
> int status = urb->status;
>
> dev_vdbg(&acm->data->dev, "%s - urb %d, len %d\n", __func__,
> rb->index, urb->actual_length);
>
> if (!acm->dev) {
> set_bit(rb->index, &acm->read_urbs_free);
> dev_dbg(&acm->data->dev, "%s - disconnected\n", __func__);
> return;
> }
>
> if (status) {
> set_bit(rb->index, &acm->read_urbs_free);
> dev_dbg(&acm->data->dev, "%s - non-zero urb status: %d\n",
> __func__, status);
> if ((status != -ENOENT) || (urb->actual_length == 0))
> return;
> }
>
> Can you please switch on dynamic debugging for cdc_acm to see what
> is being reported?
Yes, just recompiled with CONFIG_DYNAMIC_DEBUG. It'll take some until
that behaviour shows up.
Thanks,
ladis
--
To unsubscribe from this list: send the line "unsubscribe linux-usb" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
