diff options
author | Douglas Anderson <dianders@chromium.org> | 2017-12-12 10:30:31 -0800 |
---|---|---|
committer | Felipe Balbi <felipe.balbi@linux.intel.com> | 2017-12-13 11:27:53 +0200 |
commit | 38d2b5fb75c15923fb89c32134516a623515bce4 (patch) | |
tree | 42186c421271b50e0431e9fab9e5141417512905 /drivers/usb/dwc2/hcd_queue.c | |
parent | f2830ad455ec0fdc386baeb9d654f7095bf849da (diff) | |
download | linux-38d2b5fb75c15923fb89c32134516a623515bce4.tar.gz linux-38d2b5fb75c15923fb89c32134516a623515bce4.tar.bz2 linux-38d2b5fb75c15923fb89c32134516a623515bce4.zip |
usb: dwc2: host: Don't retry NAKed transactions right away
On rk3288-veyron devices on Chrome OS it was found that plugging in an
Arduino-based USB device could cause the system to lockup, especially
if the CPU Frequency was at one of the slower operating points (like
100 MHz / 200 MHz).
Upon tracing, I found that the following was happening:
* The USB device (full speed) was connected to a high speed hub and
then to the rk3288. Thus, we were dealing with split transactions,
which is all handled in software on dwc2.
* Userspace was initiating a BULK IN transfer
* When we sent the SSPLIT (to start the split transaction), we got an
ACK. Good. Then we issued the CSPLIT.
* When we sent the CSPLIT, we got back a NAK. We immediately (from
the interrupt handler) started to retry and sent another SSPLIT.
* The device kept NAKing our CSPLIT, so we kept ping-ponging between
sending a SSPLIT and a CSPLIT, each time sending from the interrupt
handler.
* The handling of the interrupts was (because of the low CPU speed and
the inefficiency of the dwc2 interrupt handler) was actually taking
_longer_ than it took the other side to send the ACK/NAK. Thus we
were _always_ in the USB interrupt routine.
* The fact that USB interrupts were always going off was preventing
other things from happening in the system. This included preventing
the system from being able to transition to a higher CPU frequency.
As I understand it, there is no requirement to retry super quickly
after a NAK, we just have to retry sometime in the future. Thus one
solution to the above is to just add a delay between getting a NAK and
retrying the transmission. If this delay is sufficiently long to get
out of the interrupt routine then the rest of the system will be able
to make forward progress. Even a 25 us delay would probably be
enough, but we'll be extra conservative and try to delay 1 ms (the
exact amount depends on HZ and the accuracy of the jiffy and how close
the current jiffy is to ticking, but could be as much as 20 ms or as
little as 1 ms).
Presumably adding a delay like this could impact the USB throughput,
so we only add the delay with repeated NAKs.
NOTE: Upon further testing of a pl2303 serial adapter, I found that
this fix may help with problems there. Specifically I found that the
pl2303 serial adapters tend to respond with a NAK when they have
nothing to say and thus we end with this same sequence.
Signed-off-by: Douglas Anderson <dianders@chromium.org>
Reviewed-by: Julius Werner <jwerner@chromium.org>
Tested-by: Stefan Wahren <stefan.wahren@i2se.com>
Acked-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Felipe Balbi <felipe.balbi@linux.intel.com>
Diffstat (limited to 'drivers/usb/dwc2/hcd_queue.c')
-rw-r--r-- | drivers/usb/dwc2/hcd_queue.c | 81 |
1 files changed, 77 insertions, 4 deletions
diff --git a/drivers/usb/dwc2/hcd_queue.c b/drivers/usb/dwc2/hcd_queue.c index fcd1676c7f0b..e34ad5e65350 100644 --- a/drivers/usb/dwc2/hcd_queue.c +++ b/drivers/usb/dwc2/hcd_queue.c @@ -58,6 +58,9 @@ /* Wait this long before releasing periodic reservation */ #define DWC2_UNRESERVE_DELAY (msecs_to_jiffies(5)) +/* If we get a NAK, wait this long before retrying */ +#define DWC2_RETRY_WAIT_DELAY (msecs_to_jiffies(1)) + /** * dwc2_periodic_channel_available() - Checks that a channel is available for a * periodic transfer @@ -1441,6 +1444,55 @@ static void dwc2_deschedule_periodic(struct dwc2_hsotg *hsotg, } /** + * dwc2_wait_timer_fn() - Timer function to re-queue after waiting + * + * As per the spec, a NAK indicates that "a function is temporarily unable to + * transmit or receive data, but will eventually be able to do so without need + * of host intervention". + * + * That means that when we encounter a NAK we're supposed to retry. + * + * ...but if we retry right away (from the interrupt handler that saw the NAK) + * then we can end up with an interrupt storm (if the other side keeps NAKing + * us) because on slow enough CPUs it could take us longer to get out of the + * interrupt routine than it takes for the device to send another NAK. That + * leads to a constant stream of NAK interrupts and the CPU locks. + * + * ...so instead of retrying right away in the case of a NAK we'll set a timer + * to retry some time later. This function handles that timer and moves the + * qh back to the "inactive" list, then queues transactions. + * + * @t: Pointer to wait_timer in a qh. + */ +static void dwc2_wait_timer_fn(struct timer_list *t) +{ + struct dwc2_qh *qh = from_timer(qh, t, wait_timer); + struct dwc2_hsotg *hsotg = qh->hsotg; + unsigned long flags; + + spin_lock_irqsave(&hsotg->lock, flags); + + /* + * We'll set wait_timer_cancel to true if we want to cancel this + * operation in dwc2_hcd_qh_unlink(). + */ + if (!qh->wait_timer_cancel) { + enum dwc2_transaction_type tr_type; + + qh->want_wait = false; + + list_move(&qh->qh_list_entry, + &hsotg->non_periodic_sched_inactive); + + tr_type = dwc2_hcd_select_transactions(hsotg); + if (tr_type != DWC2_TRANSACTION_NONE) + dwc2_hcd_queue_transactions(hsotg, tr_type); + } + + spin_unlock_irqrestore(&hsotg->lock, flags); +} + +/** * dwc2_qh_init() - Initializes a QH structure * * @hsotg: The HCD state structure for the DWC OTG controller @@ -1468,6 +1520,7 @@ static void dwc2_qh_init(struct dwc2_hsotg *hsotg, struct dwc2_qh *qh, /* Initialize QH */ qh->hsotg = hsotg; timer_setup(&qh->unreserve_timer, dwc2_unreserve_timer_fn, 0); + timer_setup(&qh->wait_timer, dwc2_wait_timer_fn, 0); qh->ep_type = ep_type; qh->ep_is_in = ep_is_in; @@ -1628,6 +1681,16 @@ void dwc2_hcd_qh_free(struct dwc2_hsotg *hsotg, struct dwc2_qh *qh) dwc2_do_unreserve(hsotg, qh); spin_unlock_irqrestore(&hsotg->lock, flags); } + + /* + * We don't have the lock so we can safely wait until the wait timer + * finishes. Of course, at this point in time we'd better have set + * wait_timer_active to false so if this timer was still pending it + * won't do anything anyway, but we want it to finish before we free + * memory. + */ + del_timer_sync(&qh->wait_timer); + dwc2_host_put_tt_info(hsotg, qh->dwc_tt); if (qh->desc_list) @@ -1663,9 +1726,16 @@ int dwc2_hcd_qh_add(struct dwc2_hsotg *hsotg, struct dwc2_qh *qh) qh->start_active_frame = hsotg->frame_number; qh->next_active_frame = qh->start_active_frame; - /* Always start in inactive schedule */ - list_add_tail(&qh->qh_list_entry, - &hsotg->non_periodic_sched_inactive); + if (qh->want_wait) { + list_add_tail(&qh->qh_list_entry, + &hsotg->non_periodic_sched_waiting); + qh->wait_timer_cancel = false; + mod_timer(&qh->wait_timer, + jiffies + DWC2_RETRY_WAIT_DELAY + 1); + } else { + list_add_tail(&qh->qh_list_entry, + &hsotg->non_periodic_sched_inactive); + } return 0; } @@ -1695,6 +1765,9 @@ void dwc2_hcd_qh_unlink(struct dwc2_hsotg *hsotg, struct dwc2_qh *qh) dev_vdbg(hsotg->dev, "%s()\n", __func__); + /* If the wait_timer is pending, this will stop it from acting */ + qh->wait_timer_cancel = true; + if (list_empty(&qh->qh_list_entry)) /* QH is not in a schedule */ return; @@ -1903,7 +1976,7 @@ void dwc2_hcd_qh_deactivate(struct dwc2_hsotg *hsotg, struct dwc2_qh *qh, if (dwc2_qh_is_non_per(qh)) { dwc2_hcd_qh_unlink(hsotg, qh); if (!list_empty(&qh->qtd_list)) - /* Add back to inactive non-periodic schedule */ + /* Add back to inactive/waiting non-periodic schedule */ dwc2_hcd_qh_add(hsotg, qh); return; } |