commit f3af3651 ("usb: dwc3: gadget: always
enable IOC on bulk/interrupt transfers") ended up
regressing Isochronous endpoints by clearing
DWC3_EP_BUSY flag too early, which resulted in
choppy audio playback over USB.

Fix that by partially reverting original commit and
making sure that we check for isochronous endpoints.

Fixes: f3af3651 ("usb: dwc3: gadget: always enable IOC
		on bulk/interrupt transfers")
Cc: <stable@vger.kernel.org>
Signed-off-by: NKonrad Leszczynski <konrad.leszczynski@intel.com>
Signed-off-by: NRafal Redzimski <rafal.f.redzimski@intel.com>
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

As a micro-power optimization, let's only resume the
USB2 PHY if we're working on <=HIGHSPEED. If we're
gonna work on SUPERSPEED or SUPERSPEED+, there's no
point in resuming the USB2 PHY.

Fixes: 2b0f11df ("usb: dwc3: gadget: clear SUSPHY bit before ep cmds")
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

by holding gadget's IRQ number in dwc->irq_gadget,
it'll be simpler to free_irq() and disable the IRQ
in case an IRQ fires while we are runtime suspended.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

By adding a pointer to endpoint registers' base
address, we can avoid using our controller-wide
struct dwc3 pointer for everything. At some point
this will allow us to have per-endpoint locks which
will, in turn, let us queue requests to separate
endpoints in parallel.

Because of this change our debugfs interface and io
accessors need to be changed accordingly.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

In all call sites of dwc3_send_gadget_ep_cmd() we
already had a valid dep pointer, so instead of
passing dwc and dep->number, which would be used to
fetch the same pointer we already had, just pass dep
directly.

In other words, we're changing:

	struct dwc3_ep *dep = dwc[dep->number];

to just passing struct dwc3_ep *dep as argument.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Instead of using burst size to configure NUMP, we
should be using RxFIFO Size instead. DWC3 is smart
enough to know that it shouldn't burst in case burst
size is 0.
Reported-by: NJohn Youn <johnyoun@synopsys.com>
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

To aid code readability, we're gonna split
__dwc3_gadget_kick_transfer() into its constituent
parts: scatter gather and linear buffers.

That way, it's easier to follow the code and focus
debug effort when one or the other fails.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Instead of returning -EINVAL when someone calls
__dwc3_gadget_wakeup() in speeds > highspeed, let's
return 0. There are no problems for the driver for
calling it in superspeed as we cleanly just return.

This avoids an annoying WARN_ONCE() always
triggering during superspeed enumeration with LPM
enabled.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

When we send an endpoint command, we want that to
complete as soon as possible, so let's remove the
unnecessary udelay(1) call.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

sg_is_last() and list_is_last() will encode the
required information for the driver to make
decisions WRT CHN and LST bits.

While at that, also replace '1' with 'true' for
consistency.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

as it turns out, we don't need the extra 'start_new'
argument as that can be inferred from DWC3_EP_BUSY
flag.

Because of that, we can simplify
__dwc3_gadget_kick_transfer() by quite a bit, even
allowing us to prepare more TRBs unconditionally.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

If we're updating transfers, we can also prepare as
many TRBs as we can fit in the ring. Let's start
doing that.

This patch 'solves' a limitation of how many TRBs we
can prepare when we're getting close the end of the
ring. Instead driver to prepare only up to end of
the ring, we check if we have space to wrap around
the ring properly.

Note that this only happens when our enqueue and
dequeue pointers are equal (which is the case for
bulk endpoints after an XferComplete event).
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Instead of trying hard to stay connected to the
host, it's best (and far easier) to disconnect from
the host already.

Anything relying on KEEP_CONNECT will just have that
ignored, but we don't have proper hibernation
implementation yet, so there are no regressions.

In any case, hibernation is only useful for runtime
PM, not system sleep.

While at that, also remove dwc3.dcfg which has been
rendered unnecessary.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

we will be re-using it for suspend/resume, so
instead of duplicating code, let's just re-factor
the functions so they can be re-used.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

As of core revision 2.60a the recommended programming model is to set
the ClearPendIN bit when issuing a Clear Stall EP command for IN
endpoints. This is to prevent an issue where some (non-compliant) hosts
may not send ACK TPs for pending IN transfers due to a mishandled error
condition. Synopsys STAR 9000614252.
Signed-off-by: NJohn Youn <johnyoun@synopsys.com>
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Now that we calculate DCFG.NUMP, we can disable
dwc3's automatic calculation so we maximize our
chances of very high throughtput through the use of
bursts.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

NumP field of DCFG register is used on NumP field of
ACK TP header and it tells the host how many packets
an endpoint can receive before waiting for
synchronization.

Documentation says it should be set to anything
<=bMaxBurst. Interestingly, however, this setting is
not per-endpoint how it should be (different
endpoints could have different burst sizes), but
things seem to work okay right now.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

if Start Transfer command fails, let's try a little
harder to figure out why the command failed and give
slightly better return codes. This will be usefulf
or isochronous endpoints, at least, which could
decide to retry a given request.
Signed-off-by: NKonrad Leszczynski <konrad.leszczynski@intel.com>
Signed-off-by: NRafal Redzimski <rafal.f.redzimski@intel.com>
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

It makes no sense to interrupt in the middle of
chained transfer. This patch just makes sure we
don't do that.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

trace already adds a newline character for us, we
don't need to do it ourselves.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

instead of limiting link TRB only to Isoc endpoints,
let's use it for all endpoint types, this way we are
more likely to transfer more data before a
XferComplete event.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

By moving our % DWC3_NUM_TRB operation to the
increment helpers, the rest of the driver can be
simplified.

It's also a good practice to make sure we will have
a single place dealing with details about how to
increment our enqueue and dequeue pointers.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Add three little helpers which will aid in making
the code slightly easier to read. One helper
increments enqueue pointer, another increments
dequeue pointer and the last one tests if we're
dealing with the last TRB.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

instead of using a bitwise and, let's rely on the %
operator since that's a lot more clear. Also, GCC
will optimize % 256 to nothing anyway.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

This makes it clear that we're dealing with a queue
of TRBs. No functional changes. While at that, also
rename start_slot to first_trb_index for similar
reasons.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Synopsys Databook says we should move link to U0
before issuing a Start Transfer command. We could
require the gadget driver to call
usb_gadget_wakeup() however I feel that changing all
gadget drivers to keep track of Link State and
conditionally call usb_gadget_wakeup() would be far
too much work. For now we will handle this at the
UDC level, but at some point composite.c should be
one handling this.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

we will need this from StartTransfer to make sure
link is in U0 before starting a transfer.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Synopsys Databook 2.60a has a note that if we're
sending an endpoint command we _must_ make sure that
DWC3_GUSB2PHY(n).SUSPHY bit is cleared.

This patch implements that particular detail.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

dwc3_send_gadget_ep_cmd() had three return
points. That becomes a pain to track when we need to
debug something or if we need to add more code
before returning.

Let's combine all three return points into a single
one just by introducing a local 'ret' variable.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

we don't plan on using multiple event buffers, but
if we find a good use case for it, this little trick
will help us avoid a loop in hardirq handler looping
for each and every event buffer.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Gadget controller might not be always active during system
suspend/resume as gadget driver might not have yet been loaded or
might have been unloaded prior to system suspend.

Check if we're active and only then perform
necessary actions during suspend/resume.
Signed-off-by: NRoger Quadros <rogerq@ti.com>
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

we will be using a single event buffer and that
renders ev_buffs array unnecessary. Let's remove it
in favor of a single pointer to a single event
buffer.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

We never, ever route any of the other event buffers
so we might as well drop support for them.

Until someone has a real, proper benefit for
multiple event buffers, we will rely on a single
one. This also helps reduce memory footprint of
dwc3.ko which won't allocate memory for the extra
event buffers.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

request_list and req_queued were, well, weird naming
choices.

Let's give those better names and call them,
respectively, pending_list and started_list. These
new names better reflect what these lists are
supposed to do.

While at that also rename req->queued to req->started.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

CSP bit of TRB Control is useful for protocols such
CDC EEM/ECM/NCM where we're transferring in blocks
of MTU-sized requests (usually MTU is 1500 bytes).

We know we will always have a short packet after two
(for HS) wMaxPacketSize packets and, usually, we
will have a long(-ish) queue of requests (for our
g_ether gadget, we have at least 10
requests).

Instead of always stopping the queue processing to
interrupt, giveback and restart, let's tell dwc3 to
interrupt but continue processing following request
if we have anything already pending in the queue.

This gave me a considerable improvement of 40% on my
test setup.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

That FIFO resizing logic was added to support OMAP5
ES1.0 which had a bogus default FIFO size. I can't
remember the exact size of default FIFO, but it was
less than one bulk superspeed packet (<1024) which
would prevent USB3 from ever working on OMAP5 ES1.0.

However, OMAP5 ES1.0 support has been dropped by
commit aa2f4b16 ("ARM: OMAP5: id: Remove ES1.0
support") which renders FIFO resizing unnecessary.
Tested-by: NKishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

We were exitting the function before actually
renaming anything. While at that, also always leave
control endpoint un-renamed.
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

It's a requirement that we release controller's lock
while calling gadget API function pointers. This
patch just fixes that long standing bug.
Signed-off-by: NJiebing Li <jiebing.li@intel.com>
Signed-off-by: NFelipe Balbi <felipe.balbi@linux.intel.com>

Check that dwc->maximum_speed is set to a valid value. Also add an error
when we use it later if we encounter an invalid value.
Signed-off-by: NJohn Youn <johnyoun@synopsys.com>
Signed-off-by: NFelipe Balbi <balbi@kernel.org>

Enable SuperSpeedPlus by programming the DCFG.speed and after
enumerating, set gadget->speed appropriately.
Signed-off-by: NJohn Youn <johnyoun@synopsys.com>
Signed-off-by: NFelipe Balbi <balbi@kernel.org>