Add support for basic power management on
the dwc3 driver. While there is still lots
to improve for full PM support, this minimal
patch will already make sure that we survive
suspend-to-ram and suspend-to-disk without
major issues.
Cc: Vikas C Sajjan <vikas.sajjan@linaro.org>
Tested-by: Vivek Gautam <gautam.vivek@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This will be used during resume to verify
if we should reconnect our pullups or not.
Tested-by: Vivek Gautam <gautam.vivek@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
DWC3 controller curretly depends on USB && USB_GADGET.
Some hardware may like to use only host feature on dwc3,
or only gadget feature.
So, removing this dependency of USB_DWC3 on USB and USB_GADGET.
Adding the mode of operaiton of DWC3 also here
HOST/GADGET/DUAL_ROLE based on which features are enabled.
[ balbi@ti.com :
. make sure we have default modes for all possible Kernel
configurations.
. Remove the config -> menuconfig change as it's unnecessary
. switch over to IS_ENABLED() ]
CC: Doug Anderson <dianders@chromium.org>
Signed-off-by: Vivek Gautam <gautam.vivek@samsung.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
To work with scatter gather properly, fixes have been done in number of
functions. I will explain requirement of each fixes one by one.
start_slot: used to retrieve all request of SG during cleanup
dwc3_gadget_giveback: We need to skip link TRB if it was one of the
intermediate TRB of SG.
dwc3_prepare_one_trb: We need to track all submitted TRBs during
cleanup. Since, all TRBs would be serially allocated, so we can just
keep starting slot info and we can always find rest of them. We need to
pass sg node number, so that we cab appropriately program ISOC_FIRST/ISOC,
Chain etc.
dwc3_prepare_trbs: last_one should be set when it is last node
of SG as well as last node of request_list.
__dwc3_cleanup_done_trbs: It has been prepared after re-factorization of
dwc3_cleanup_done_reqs. It is called for each TRB of SG.
Signed-off-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
There are two reasons to generate missed isoc.
1. when the host does not poll for all the data.
2. because of application-side delays that prevent all the data from
being transferred in programmed microframe.
Current code was able to handle first case only. This patch handles
scenario 2 as well.Scenario 2 sometime may occur with complex gadget
application, however it can be easily reproduced for testing purpose as
follows:
a. use isoc binterval as 1 in f_sourcesink.
b. use pattern=0
c. introduce a delay of 150us deliberately in source_sink_complete, so
that after few frames it lands into scenario 2.
d. now run testusb 16 (isoc in test). You will notice that if this
patch is not applied then isoc transfer is not able to recover after
first missed.
Current patch's approach is as under:
If missed isoc occurs and there is no request queued then issue END
TRANSFER, so that core generates next xfernotready and we will issue a
fresh START TRANSFER.
If there are still queued request then wait, do not issue either END or
UPDATE TRANSFER, just attach next request in request_list during giveback.
If any future queued request is successfully transferred then we will issue
UPDATE TRANSFER for all request in the request_list.
Cc: <stable@vger.kernel.org> # v3.6 v3.7 v3.8
Signed-off-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Currently we're allocating an entire page to
serve as our event buffer. Provided our events
are 4 bytes long, it's very unlikely we will
even trigger 1k events at once.
Even in the worst case scenario where every
endpoint triggers one event and we still have
a couple of error events, that would still
be less than 40 events.
In order to cope with future versions of the
IP which could (or could not) increase the
amount of possible events to trigger
simultaneously, we're using an arbitrary size
of 64 events for our event buffer.
We're saving 3840 bytes by doing so.
Signed-off-by: Felipe Balbi <balbi@ti.com>
The lockless implementation of the unique id is quite impressive (:P)
but dirver's core can handle it, we can remove it and make our code a
little smaller.
Cc: Anton Tikhomirov <av.tikhomirov@samsung.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Some much needed changes for our dwc3 driver. First there's a
rework on the ep0 handling due to some Silicon issue we uncovered
which affects all users of this IP core (there's a missing
XferNotReady(DATA) event in some conditions). This issue which
show up as a SETUP transfers which wouldn't complete ever and
we would fail TD 7.06 of the Link Layer Test from USB-IF and
Lecroy's USB3 Exerciser.
We also fix a long standing bug regarding EP0 enable sequencing
where we weren't setting a particular bit (Ignore Sequence
Number). Since we never saw any problems caused by that, it
didn't deserve being sent to stable tree.
On this pull request we also fix Burst Size initialization which
should be done only in SuperSpeed and we were mistakenly setting
Burst Size to the maximum value on non-SuperSpeed mode. Again,
since we never saw any problems caused by that, we're not sending
this patch to stable.
There's also a memory ordering fix regarding usage of bitmaps in
dwc3 driver.
You will also find some sparse warnings fix, a fix for missed
isochronous packets when the endpoint is already busy, and a
fix for synchronization delay on dwc3_stop_active_transfer().
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)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=Lqwl
-----END PGP SIGNATURE-----
Merge tag 'dwc3-for-v3.7' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb into usb-next
usb: dwc3: patches for v3.7 merge window
Some much needed changes for our dwc3 driver. First there's a
rework on the ep0 handling due to some Silicon issue we uncovered
which affects all users of this IP core (there's a missing
XferNotReady(DATA) event in some conditions). This issue which
show up as a SETUP transfers which wouldn't complete ever and
we would fail TD 7.06 of the Link Layer Test from USB-IF and
Lecroy's USB3 Exerciser.
We also fix a long standing bug regarding EP0 enable sequencing
where we weren't setting a particular bit (Ignore Sequence
Number). Since we never saw any problems caused by that, it
didn't deserve being sent to stable tree.
On this pull request we also fix Burst Size initialization which
should be done only in SuperSpeed and we were mistakenly setting
Burst Size to the maximum value on non-SuperSpeed mode. Again,
since we never saw any problems caused by that, we're not sending
this patch to stable.
There's also a memory ordering fix regarding usage of bitmaps in
dwc3 driver.
You will also find some sparse warnings fix, a fix for missed
isochronous packets when the endpoint is already busy, and a
fix for synchronization delay on dwc3_stop_active_transfer().
The definition of DWC3_DCTL_HIRD_THRES macro is
completely wrong. It will only work for when we
want to read the register's contents for that bitfield.
Change the macro so that it can be used to writing to
the register, and when we need to read, we can add
extra right shift of 24 bits.
Signed-off-by: Pratyush Anand <pratyush.anand@st.com>
[ balbi@ti.com: add a commit log ]
Signed-off-by: Felipe Balbi <balbi@ti.com>
If an IN transfer is missed on isoc endpoint, then driver must insure
that next ep_queue is properly handled.
This patch fixes this issue by starting a new transfer for next queued
request.
Signed-off-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
SOF Number is bit16:3 of DSTS. Correct the mask accordingly.
Signed-off-by: Pratyush Anand <pratyush.anand@st.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch adds definitions for some new registers that have been
added to later versions of the controller, up to v2.10a.
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This pull request contains one workaround for a Silicon
Issue found on all RTL releases prior to 2.20a, which
would cause a metastability state on Run/Stop bit.
We also have some patches implementing a few extra Standard
requests introduced by USB3 spec (Set SEL and Set Isoch Delay),
as well as one patch, which has been pending for a long time,
implementing LPM support.
Last, but not least, we are splitting the host address space
out of the dwc3 core driver otherwise xHCI won't be able to
request_mem_region() its own address space. This patch is
only needed because we are (as we should) re-using the xHCI
driver, which is a completely separate module.
Together with these three big changes, come a few extra preparatory
patches which most move code around, define macros and so on, as
well as a fix for Isochronous transfers which hasn't been triggered
before.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (GNU/Linux)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=qgf/
-----END PGP SIGNATURE-----
Merge tag 'dwc3-for-v3.5' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb into usb-next
usb: dwc3: patches for v3.5 merge window
This pull request contains one workaround for a Silicon
Issue found on all RTL releases prior to 2.20a, which
would cause a metastability state on Run/Stop bit.
We also have some patches implementing a few extra Standard
requests introduced by USB3 spec (Set SEL and Set Isoch Delay),
as well as one patch, which has been pending for a long time,
implementing LPM support.
Last, but not least, we are splitting the host address space
out of the dwc3 core driver otherwise xHCI won't be able to
request_mem_region() its own address space. This patch is
only needed because we are (as we should) re-using the xHCI
driver, which is a completely separate module.
Together with these three big changes, come a few extra preparatory
patches which most move code around, define macros and so on, as
well as a fix for Isochronous transfers which hasn't been triggered
before.
[ resolved conflicts and build error in drivers/usb/dwc3/gadget.c - gregkh]
to avoid sprinkling magic constants on the driver
we define a constant to be used when allocating
setup_buffer and ep0_bounce buffer.
Signed-off-by: Felipe Balbi <balbi@ti.com>
This is basically a noop for DWC3. We don't have
to do anything. Basically we test if the request
parameters are correct, cache the Isochronous
Delay and return success.
Signed-off-by: Felipe Balbi <balbi@ti.com>
This patch implements Set SEL Standard Request
support for dwc3 driver. It needs to issue a command
to the controller passing the timing we received on
the data phase of the Set SEL request.
Signed-off-by: Felipe Balbi <balbi@ti.com>
On device loading the driver enables LPM and the acceptance of U1 and U2
states. The [Set|Clear]Feature requests for "U1/U2" are forwarded
directly to the hardware and allow / forbid the initiation of the low
power links.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
This fix prevents a problem with dwc3 and host mode where
we were requesting the entire memory region in dwc3/core.c,
thus preventing xhci-plat from ever ioremapping its own address space.
Signed-off-by: Ido Shayevitz <idos@codeaurora.org>
Signed-off-by: Felipe Balbi <balbi@ti.com>
We have other revisions already released, let's
define revision macros for those so we can do
runtime detection of known erratas.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Here are the changes for v3.4 merge window.
It includes a new glue layer for Samsung's Exynos platform, a simplification of
memory management on DWC3 driver by using dev_xxx functions, a few
optimizations to IRQ handling by dropping memcpy() and using bitshifts, a fix
for TI's OMAP5430 TX Fifo Allocation, two fixes on USB2 test mode
implementation (one on debugfs and one on ep0), and several minor changes such
as whitespace cleanups, simplification of a few parts of the code, decreasing a
long delay to something a bit saner, dropping a header which was included twice
and so on.
The highlight on this merge is the support for Samsung's Exynos platform,
increasing the number of different users for this driver to three.
Note that Samsung Exynos glue layer will only compile on platforms which
provide implementation for the clk API for now. Once Samsung supports
pm_runtime, that limitation can be dropped from the Makefile.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.11 (GNU/Linux)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=4Z2+
-----END PGP SIGNATURE-----
Merge tag 'dwc3-for-v3.4' of git://git.kernel.org/pub/scm/linux/kernel/git/balbi/usb into usb-next
usb: dwc3: changes for v3.4 merge window
Here are the changes for v3.4 merge window.
It includes a new glue layer for Samsung's Exynos platform, a simplification of
memory management on DWC3 driver by using dev_xxx functions, a few
optimizations to IRQ handling by dropping memcpy() and using bitshifts, a fix
for TI's OMAP5430 TX Fifo Allocation, two fixes on USB2 test mode
implementation (one on debugfs and one on ep0), and several minor changes such
as whitespace cleanups, simplification of a few parts of the code, decreasing a
long delay to something a bit saner, dropping a header which was included twice
and so on.
The highlight on this merge is the support for Samsung's Exynos platform,
increasing the number of different users for this driver to three.
Note that Samsung Exynos glue layer will only compile on platforms which
provide implementation for the clk API for now. Once Samsung supports
pm_runtime, that limitation can be dropped from the Makefile.
Conflicts:
drivers/usb/dwc3/gadget.c
Define DWC3_GCTL_SCALEDOWN_MASK and use it in place of
DWC3_GCTL_SCALEDOWN(3).
Signed-off-by: Paul Zimmerman <paulz@synopsys.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
those routines have everything we need to map/unmap
USB requests and it's better to use them.
In order to achieve that, we had to add a simple
change on how we allocate and use our setup buffer;
we cannot allocate it from coherent anymore otherwise
the generic map/unmap routines won't be able to easily
know that the GetStatus request already has a DMA
address.
Signed-off-by: Felipe Balbi <balbi@ti.com>
this will get rid of a useless memcpy on
IRQ handling, thus improving driver performance.
Tested with OMAP5430 running g_mass_storage on
SuperSpeed and HighSpeed.
Note that we are removing the little endian access
of the TRB and all accesses will be in System endianness,
if there happens to be a system in BE, bit 12 of GSBUSCFG0
should be set so that HW does byte invariant BE accesses
when fetching TRBs.
Signed-off-by: Felipe Balbi <balbi@ti.com>
When host requests us to enter a test mode,
we cannot directly enter the test mode before
Status Phase is completed, otherwise the core
will never be able to deliver the Status ZLP
to host, because it has already entered the
requested Test Mode.
In order to fix the error, we move the actual
start of Test Mode right after we receive
Transfer Complete event of the status phase.
Signed-off-by: Gerard Cauvy <g-cauvy1@ti.com>
Signed-off-by: Felipe Balbi <balbi@ti.com>
When we set Run/Stop bit, we also move the
core to Rx.Detect state so that USB3 handshake
can start from a known location.
Signed-off-by: Felipe Balbi <balbi@ti.com>
We need to dynamically re-size TxFifos for the
cases where default values will not do.
While at that, we create a simple function which,
for now, will just allocate one full packet fifo
space for each of the enabled endpoints.
This can be improved later in order to allow for
better throughput by allocating more space for
endpoints which could make good use of that like
isochronous and bulk.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Only bit 3 of the event status bitfield is valid
and the others should not be considered.
Make sure SW matches documentation on that case
to avoid bogus debugging prints which would
confuse an engineer.
Signed-off-by: Felipe Balbi <balbi@ti.com>
ep->max_streams is a mere hint to the gadget
driver that 'ep' supports stream handling. Using
that as a decision variable for enabling streams
was my worst brain-fart to date.
Instead, we should check from the Superspeed
Endpoint Companion Descriptor if the endpoint
has requested streams. For that we need a little
re-factoring but it is now correct.
Debugged-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
Instead of special functions and shortcuts for sending our internal
answers to the host we started doing what the gadget does and used the
public API for this. Since we only were using a few fields the
usb_request was enough. Later added the list handling in order to
synchronize the host / gadget events and now we require to have the
dwc3_request struct around our usb_request or else we touch memory that
does not belong to us. So this patch does this.
Reported-by: Partha Basak <p-basak2@ti.com>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
DWC3 revisions <1.88a have an issue which would
case a missing Disconnect event if cable is
disconnected while there's a Setup packet
pending the FIFO.
Signed-off-by: Felipe Balbi <balbi@ti.com>
RTL revisions <1.83a have an issue where, depending
on the link partner, the USB link might do multiple
entry/exit of low power states before a transfer
takes place causing degraded throughput.
The suggested workaround is to clear bits
12:9 of DCTL register if we see a transition
from U1|U2 to U0 and only re-enable that on
a transfer complete IRQ and we have no pending
transfers on any of the enabled endpoints.
Signed-off-by: Felipe Balbi <balbi@ti.com>
Since the re-worked ep0 handling (which uses HW's hints to recognize the ep0
status) we lost the delayed status handling. This is used by the file and mass
storage gadget to gain some extra time so setup its internal status before it
can proceed further requests.
In particular the storage gadget does nothing on USB_REQ_SET_CONFIGURATION but
wakes up a thread which handles the request. If the udc driver continues ep0
handling before the thread did its work then then endpoint is not yet
configured and further requests will fail. Once the gadget is ready, it will
enqueue an empty packet which is used for synchronization.
In order to fix this issue, the patch does the following:
Set ->delayed_status once the delayed_status has been notices and do not
continue on the next XferNotReady event. We will continues ep0 processing once
the gadget enqueued the zero packet for synchronization.
A cleaner approach would be to enforce the gadget to enqueue an empty
(zero) request even for the status phase but this would do for now.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
There are two where need to set operational mode:
- during initialization while we decide to run in host,device or DRD
mode
- at runtime via the debugfs interface.
This patch provides a new function which sets the operational mode and
moves its initialiation to the mode switch instead in the gadget code
itself.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
hardware will tell us how many event buffers we
need to support, so let's allocate the array
dynamically too.
Signed-off-by: Felipe Balbi <balbi@ti.com>
if we want to support situations where we have
both SoC and PCIe versions of the IP on the same
platform, we need to have sequential numbers between
them, otherwise we will still have name collisions.
Because of that, we need to move dwc3_get/put_device_id()
to core.c and export that symbol to be used by glue
layers.
Signed-off-by: Felipe Balbi <balbi@ti.com>
The GetStatus (STD)-request is handled the driver and uses a tiny hack
to send the two bytes long answer. This patch removes the custom hack
uses the normal usb_ep_queue() for that.
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Felipe Balbi <balbi@ti.com>
The Designware USB3 IP can be configured with
an internal xHCI. If we're running on such a
version, let's start the xHCI stack.
Signed-off-by: Felipe Balbi <balbi@ti.com>
There's no need to add driver_data for something
we can fetch from HW.
This also makes our id_table unnecessary - at least
for now -, so we also remove it on the same patch.
Signed-off-by: Felipe Balbi <balbi@ti.com>
this is mainly for testing. In order to be able
to test if we're enumerating correctly on all
speeds, let that be controlled by a module
parameter.
Signed-off-by: Felipe Balbi <balbi@ti.com>
