| Commit message (Collapse) | Author | Age |
|---|
| ... | |
| | | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
commit 3158a8d416f4e1b79dcc867d67cb50013140772c upstream.
$ lsusb:
Bus 001 Device 101: ID 1e0e:9001 Qualcomm / Option
$ usb-devices:
T: Bus=01 Lev=02 Prnt=02 Port=00 Cnt=01 Dev#=101 Spd=480 MxCh= 0
D: Ver= 2.00 Cls=00(>ifc ) Sub=00 Prot=00 MxPS=64 #Cfgs= 2
P: Vendor=1e0e ProdID=9001 Rev= 2.32
S: Manufacturer=SimTech, Incorporated
S: Product=SimTech, Incorporated
S: SerialNumber=0123456789ABCDEF
C:* #Ifs= 7 Cfg#= 1 Atr=80 MxPwr=500mA
I:* If#= 0 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=ff Prot=ff Driver=option
I:* If#= 1 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
I:* If#= 2 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
I:* If#= 3 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
I:* If#= 4 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=00 Prot=00 Driver=option
I:* If#= 5 Alt= 0 #EPs= 3 Cls=ff(vend.) Sub=ff Prot=ff Driver=qmi_wwan
I:* If#= 6 Alt= 0 #EPs= 2 Cls=ff(vend.) Sub=42 Prot=01 Driver=(none)
The last interface (6) is used for Android Composite ADB interface.
Serial port layout:
0: QCDM/DIAG
1: NMEA
2: AT
3: AT/PPP
4: audio
Signed-off-by: Andrey Skvortsov <andrej.skvortzov@gmail.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
| | | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | |
| | | | | | | | |
commit 6627ae19385283b89356a199d7f03c75ba35fb29 upstream.
Add USB ID for cp2104/5 devices on GE B650v3 and B850v3 boards.
Signed-off-by: Ken Lin <ken.lin@advantech.com.tw>
Signed-off-by: Akshay Bhat <akshay.bhat@timesys.com>
Signed-off-by: Johan Hovold <johan@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
| | | | |_|_|_|/
| | |/| | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
commit c450960187f45d4260db87c7dd4fc0bceb5565d8 upstream.
The assignement of EP transfer resources was not handled properly in the
dwc3 driver. Commit aebda6187181 ("usb: dwc3: Reset the transfer
resource index on SET_INTERFACE") previously fixed one aspect of this
where resources may be exhausted with multiple calls to SET_INTERFACE.
However, it introduced an issue where composite devices with multiple
interfaces can be assigned the same transfer resources for different
endpoints. This patch solves both issues.
The assignment of transfer resources cannot perfectly follow the data
book due to the fact that the controller driver does not have all
knowledge of the configuration in advance. It is given this information
piecemeal by the composite gadget framework after every
SET_CONFIGURATION and SET_INTERFACE. Trying to follow the databook
programming model in this scenario can cause errors. For two reasons:
1) The databook says to do DEPSTARTCFG for every SET_CONFIGURATION and
SET_INTERFACE (8.1.5). This is incorrect in the scenario of multiple
interfaces.
2) The databook does not mention doing more DEPXFERCFG for new endpoint
on alt setting (8.1.6).
The following simplified method is used instead:
All hardware endpoints can be assigned a transfer resource and this
setting will stay persistent until either a core reset or hibernation.
So whenever we do a DEPSTARTCFG(0) we can go ahead and do DEPXFERCFG for
every hardware endpoint as well. We are guaranteed that there are as
many transfer resources as endpoints.
This patch triggers off of the calling dwc3_gadget_start_config() for
EP0-out, which always happens first, and which should only happen in one
of the above conditions.
Fixes: aebda6187181 ("usb: dwc3: Reset the transfer resource index on SET_INTERFACE")
Reported-by: Ravi Babu <ravibabu@ti.com>
Signed-off-by: John Youn <johnyoun@synopsys.com>
Signed-off-by: Felipe Balbi <balbi@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
|
| |\ \ \ \ \ \ \
| |_|_|/ / / /
|/| | | | | | |
|
| | |/ / / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
__dwc3_gadget_ep_disable API doing memset 0 with size
set to DWC3_TRB_NUM. Number of TRBs allocated for gsi
endpoints are less than DWC3_TRB_NUM. This results in
to memory corruption. Fix this bug by introducing
num_trbs member in dwc3_ep structure to save number of
trbs allocated in a dma pool upon dma pool creation.
Ep disable operation will use num_trbs of a dwc3_ep to
perform memset 0.
Change-Id: I94b5865ca22b4e1fde0d2cd8dcb218906327a916
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| |\ \ \ \ \ \ |
|
| | |/ / / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
USB bus suspend with remote wakeup allowed case:
1. SUSPEND interrupt received from controller on USB bus suspend
2. gsi_suspend() blocks ringing of doorbell by USB controller to IPA GSI
and schedule usb_ipa_w work with event EVNT_SUSPEND
3. ipa_work_handler() work handler makes sure that USB GSI interface is
into IDLE state and calls ipa_usb_xdci_suspend() to put IPA GSI related
channel into suspend state and allow turning of IPA's clock
5. ipa_work_handler() decrements PM usage count, allow USB to go into LPM
6. USB controller driver turns off USB clocks and put into LPM
USB bus resume with remote wakeup allowed case:
1. WAKEUP interrupt received from controller on USB bus resume
2. USB controller driver turns on USB clocks
3. gsi_resume() increments PM usage count and schedule usb_ipa_w work with
event EVENT_RESUME
4. ipa_work_handler() work handler calls ipa_usb_xdci_resume() to resume
IPA GSI related channel and turning on IPA's clock
5. ipa_work_handler() unblocks ringing of doorbell by USB controller to
IPA GSI
In above USB bus resume case using specific host with ECM functionality,
host is able to send PING or data packets equal to prepared number of TRBs
(currently 7) on USB EP OUT successfully before above resume step 5 then
GSI ignores rang doorbell as WRITE pointer updating last consumed TRB is
already gone through full cycle. This results into data stall. Hence to fix
this issue increase number of TRBs used with USB OUT endpoint from 7 to 31
for USB ECM function with GSI accelerated path. This helps to have always
pending TRBs with USB controller and ringing of doorbell is not missed by
GSI once above resume step 5 is completed.
Also ECM doesn't support aggregation. Hence fix out aggregation size param.
CRs-Fixed: 1023149
Change-Id: Ife0f265b65fda92a99f5170154da6cb4c6b113b7
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| |\ \ \ \ \ \ |
|
| | | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | |
| | | | | | | |
There is a possibility of dwc3 irq being fired as soon as
it gets enabled by tasklet handler. As a result dwc3 hard irq
handler disables the irq and schedules tasklet. If tasklet is
still running, tasklet_schedule becomes no op and dwc3 irq
remains disabled permanently. Fix this issue by handling irq
in threaded context with IRQF_ONESHOT flag set. Also update
the shared irq flag in xhci platform driver for irq registration
in HCD driver.
CRs-Fixed: 1038421
Change-Id: I9291cb08c4597922131b8c2d420e834a00a72621
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| |\ \ \ \ \ \ \
| |_|/ / / / /
|/| | | | | | |
|
| | |/ / / / /
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
After gsi_unbind() if gsi_bind() function frees speed descriptor due
to an error leads to double freeing of descriptor. Fix this issue by
setting descriptor pointer back to NULL after freeing the descriptor
memory.
CRs-Fixed: 1038866
Change-Id: Ic5ee71389b6b20c8557e0a75f9bbe5a1749cf655
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| |/ / / / /
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
This change adds USB QUSB PHY v2 driver which is compatible
for USB QUSB PHY having major revision as 2.
Change-Id: I1751352ebbe38d4b8c7886085d15043c2e5244f5
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
dma address of secondary event ring, transfer ring and
device context base address are required to pass to
remote entity. Remote entity uses these addresses to
program xhci controller registers.
Change-Id: Ie0756d646a396a11b41b93e886bca9aff636ee5d
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Implement APIs to dynamically allocate and free secondary
event rings based upon interrupter number. Also add exported
APIs in usb core layer which allows secondary event ring
management via remote processor entity.
Change-Id: I5ee7d44d6cad8e35e22d3c1a027a1eec5d208585
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | |/ / /
|/| | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
There is limitation when IPA is performing de/aggregation with MBIM
in scatter-gather mode. Hence instead of 2KB buffer use 16KB buffer
(i.e. size of aggregated frame) with each TRB for MBIM DL case.
CRs-Fixed: 1032467
Change-Id: I4f73d30711b5606c2ac88ce9d05a8483d64b665d
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The test results indicate that larger transfers improve write speeds.
The Rx request transfer length is now 16K bytes. Increase the Rx
transfer length from 16K to 1MB to get good throughputs.
Change-Id: I3c64e6b96d569b4dc61805a53a0835db9142d55e
Signed-off-by: ChandanaKishori Chiluveru <cchilu@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Interfaces like MBIM or ECM is having multiple data interfaces.
In this case, SET_CONFIG() happens before set_alt with data interface 1.
Due to this, TXFIFO of GSI IN endpoint is not resized causing low
throughput in DL direction. Fix this issue by using mult as 3 for
GSI related USB IN endpoint irrespective of super-speed or high-speed
mode.
CRs-Fixed: 1025031
Change-Id: I10de98ae57284699af3abcd90bafac63ba03844e
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
commit 19bacdc925055f020a ("usb: dwc3: core: only setting the
dma_mask when needed") does not allow to modify dma mask if it
is already set. Since the platform device has default 32-bit
dma mask set change prevents to update the dma mask to 64-bit.
This leads to kernel panic due to out of SW-IOMMU space when a
function driver tries to map more than 32-bit wide address.
Change-Id: I38b178f38277f9a2fa40735f4e15385638403ae6
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If usb_ep_queue() fails on queueing zero-length packet, driver is not
decrementing dpkts_tolaptop_pending counter which may results into seeing
count mismatch. Hence decrement dpkts_tolaptop_pending if zero length
packet queueing fails.
CRs-Fixed: 1027031
Change-Id: Id3c7c2627bdf37524067512db51d3180c570106d
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Currently driver is reading and logging usage count of gadget
device but it is required to use usage count of gadget's parent
device which is used to prevent USB controller's low power mode.
Hence fix reporting of USB devices' usage count.
Also mark sm_state to initialized and remove queueing of
ipa_usb_wq as ipa_work_handler() is running without USB gadget
is initialized i.e. gadget is NULL.
CRs-Fixed: 1021499
Change-Id: Ia64afa3adb769674f6a9a60fde2c7397b7e4fe49
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Roothub's (udev->dev) can be asynchronously suspended due to
power.async_suspend is set to true i.e. at time of system suspend and
resume, these devices' (usb1 and usb2) kobject count is decremented
and incremented respectively. hcd_release() API expects that shared_hcd
is being released first before hcd. Due to additional reference count
at time of system resume, it results into hcd_release() is called
first with hcd (primary) and then shared_hcd. With this,
usb_hcd_is_primary_hcd() API is returning true for both hcd and
shared_hcd which results into double free of bandwidth_mutex. Fix this
issue by identifying hcd to release bandwidth_mutex without depending
on which order hcd_release() is called with hcd and share_hcd.
CRs-Fixed: 955531
Change-Id: I6bfcfd54525fa6472bd848d4c112fff0c9462355
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Currently MTP driver is resetting string id during unbind, when
cable is disconnected. Due to this, composite driver might run out
of string ids, when it tries to allocate string id during next
composition switch to MTP. This results in composition switch failure
when MTP is selected and eventually causing USB enumeration failure.
Fix this by not resetting string id and allocate it once only.
Change-Id: Ieacdc8dd76bc45638002eb749ff87aa95f496fa3
Signed-off-by: Vijayavardhan Vennapusa <vvreddy@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If cable is disconnected while send_file work is in progress,
Endpoints will be disabled and state will be marked as OFFLINE.
As a part of endpoint disable pending requests will be retire out
with -ECONNRESET error. Due to this we will modify the device state
as ERROR in completion callback. If the state is not offline then
we will mark the device state to ready in send_file_work. This results
in mtp driver to not returning any error for next mtp_write operation
and mtp release will not be called.
Hence fix this issue by adding a proper check in completions to not
moving device state to ERROR in case of disconnect.
Change-Id: Idb4e075c89bdf94790a321bc464d30eba546eeaa
Signed-off-by: ChandanaKishori Chiluveru <cchilu@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The existing code shows the phone icon in 'My Computer' of windows host,
for PTP composition.
Create a new Extended Configuration Descriptor for PTP composition and
update the compatibleID to display a camera icon for PTP composition.
Change-Id: If286c5b80874a95be0b5ecc533d0e5c7a14f39d3
Signed-off-by: Azhar Shaikh <azhars@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This change captures time taken by each vfs_read() and vfs_write() call
made from USB MTP gadget driver where data size is equal to USB request
buffer size. It provides debugfs entry to get this information for 100
such call and resetting the same.
How-To use:
To get stats:
cat /sys/kernel/debug/usb_mtp/status
To reset stats:
echo 0 > /sys/kernel/debug/usb_mtp/status
Change-Id: I252e830568704e9e557660c1ae0f7597823e4b17
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
MTP usb device node created as a part of mtp function init call.
Userspace can read/write to MTP device using this node. If MTP is
not enabled in the composition and trying to read mtp_usb dev node
from the userspace leading to null pointer access in mtp_read.
Do not access ep OUT maxpacket size in mtp_read. First block on mtp_read
until the state become online which doesn't wakeup from the thread and
expecting for the read completion or state change which occurs as
a part of set_alt.
Change-Id: Icbee5fe7ae2c02b2bca185a0dc7587eb4940058a
Signed-off-by: ChandanaKishori Chiluveru <cchilu@codeaurora.org>
Signed-off-by: Azhar Shaikh <azhars@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
32-bit userspace calling into 64-bit kernel
cause different ioctl codes to get generated.
The different ioctl code gets generated
because sizeof is used on mtp and accessory
structures which is different for 32/64
compilation. Because of this, 64-bit kernel
can never execute the right ioctl command.
Implement compat_ioctl to handle such
execution environment.
Change-Id: I26cc10986e28a28eab6f3c65f28f4d2b808112d9
Signed-off-by: Sujeet Kumar <ksujeet@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If cable is disconnected while receive_file work is in progress,
there is a case where mtp driver returns -ECANCELLED instead of
-EIO. This results in mtp server thread not getting disconnect
notification and mtp release will not be called. Due to this, mtp
server will not start mtp connection after next USB connection.
Change-Id: I909bebe0d22c19329cebb0a4a76424e08c82328c
Signed-off-by: Vijayavardhan Vennapusa <vvreddy@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Current driver code doesn't handle the memory allocation failure
correctly and ends up leaking memory and trying to free
unallocated one.
Change-Id: I7e3527b7e08ccfe566d85e3009d8a015d8daa707
Signed-off-by: Manu Gautam <mgautam@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Due to a known PMIC HW bug, hard reset or PR swap may result in
the charger notifying a CC disconnect event in sink mode, in which
case the handler ignores it given these conditions. However, this
closely resembles the case of a non-PD capable source getting
physically disconnected as well. Consisdering the most probable
case of this happening is disconnecting from an SDP while hard
reset was just issued (and will fail), this workaround is currently
racy if repeated plugin/plugout is performed as the timeout to
detect failed hard reset takes a few seconds.
For now, until a HW fix is available, optimize this legacy case
by allowing the disconnect to go through if the detected charger
type is SDP. Also fix the SINK_WAIT_CAP_TIME timeout as it was
incorrectly multiplied by 3 during testing and was not removed.
This will significantly reduce the amount of time for max failed
hard reset attempts for non-PD && non-SDP cases and should decrease
the window for racing with an actual disconnect.
Change-Id: Ic57a369ed1e194ab512b4b86ce4d216df46b5f46
Signed-off-by: Jack Pham <jackp@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Currently burst value is set to zero for superspeed in mtp driver.
MTP throughput can be improved by setting max burst value to non
zero value in superspeed mode. Hence set max burst value to two
for both IN and OUT bulk endpoints in superspeed mode.
Change-Id: Ib78b8fec5d1ab934f4d4ec80b7e008149707ce0f
Signed-off-by: Vijayavardhan Vennapusa <vvreddy@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Convert print messages to IPC logging facilities. This
will allow for easier debugging from crash dumps as
well as from debugfs. The print calls are converted to
macros which also print to the console for flexibility.
Change-Id: I6238b0434936fcff562d7a049e5e9d13794f9cdc
Signed-off-by: Jack Pham <jackp@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Update function drivers mtp and accessory to operate
in super speed. The burst capability is not enabled for now.
Change-Id: Ie95cbfc9444c56c8268b70e2916713190699c71a
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Queuing a request on a disabled endpoint during composition switch leads
to prime failure. Hence return -EINVAL if a request is queued on a disabled
endpoint.
Also, in f_mtp, block queuing a request in OUT ep in receive_file_work, if
the device state is STATE_OFFLINE.
Change-Id: I0e706d5280a2460baf6ab05dbf97a09c59b642fb
Signed-off-by: Rajkumar Raghupathy <raghup@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Currently if USB composition switch happens while file transfer is
happening either from/to device, ep_queue fails and sets device state
to ERROR even if device state is already set to OFFLINE in function
disable call.
As part of ioctl call, if the state is not offline, moving the device
state to READY and returning the error. Since the device state is marked
as READY, the next write call tries to queue the request to hardware and
is blocked and due to this, userspace is not calling the release function.
Hence next mtp open fails even after cable reconnection.
Signed-off-by: Vijayavardhan Vennapusa <vvreddy@codeaurora.org>
Change-Id: Ia8cbd1cd8c81b90389900b83744b2bed89068db5
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
If the Host sends CANCEL request when an OUT request is queued,
the mtp_driver should dequeue the request and report the userspace about
the request CANCELLATION. As the mtp_read doesn't wakeup from the thread
which is expecting for the read completion, wakeup the thread using the
state change which occurs as a part of CANCEL request.
Also, when the request is dequeued and so mtp receives read completion
status as error, it moves the state to ERROR. Due to this the application
is not reported about the request cancellation for the next IN(mtp_write)
transfer as the state is not updated from STATE_ERROR to STATE_CANCELED.
So mtp_write proceeds queueing the IN requests for the corresponding sync
which are not processed by the Host and so the transfers initiated later
are observed as error in the HOST side.
So dequeue the OUT(mtp_read) request, if the driver receives a REQ_CANCEL
prior to read completion on mtp_read & update the state to STATE_CANCELED.
Change-Id: I2cd9780f8b3e0ce541472c4d25c8294f11a45d76
Signed-off-by: Rajkumar Raghupathy <raghup@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
DWC3 UDC requires OUT endpoint request length to be multiple of
endpoint max packet size. Hence the OUT endpoint request is
always set to the request buffer size in mtp read. The buffer
can be allocated dynamically and is always multiple of 1024.
If the count passed to read system call is less than the buffer
size, we end up reading more bytes and not giving it to userspace.
Userspace calls read with 512 bytes for receiving request from the
host. If the request is for receiving a file, the read is called
with 16K bytes for reading the header and some data. The remaining
data is received via ioctl. If 1MB is used as request length for
16K read, the received file will be corrupted.
Change-Id: I2eedf829f267150ff7a303824e95963219a1f555
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The Tx request transfer length is 16K bytes by default. The test
results indicate that larger transfers improve read speeds. Add
provision for specifying Tx transfer length at runtime.
echo -n 1048576 > /sys/module/g_android/parameters/mtp_tx_req_len
The above command can be used to set Tx transfer length to 1MB. If
the memory allocation is failed, fallback to the default length.
Also add provision for specifying no. of Tx requests at runtime.
echo 8 > /sys/module/g_android/parameters/mtp_tx_reqs
Change-Id: I675fc3303be2aba081d2ab59a9efb94aa478f849
Signed-off-by: Vijayavardhan Vennapusa <vvreddy@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The Rx request transfer length is 16K bytes by default. The test
results indicate that larger transfers improve write speeds. Add
provision for specifying Rx transfer length at runtime.
echo -n 1048576 > /sys/module/g_android/parameters/mtp_rx_req_len
The above command can be used to set Rx transfer length to 1MB. If
the memory allocation is failed, fallback to the default length.
Change-Id: I74d5da09437c25e9f1772f7cd49aecb8e4fedcf2
Signed-off-by: Pavankumar Kondeti <pkondeti@codeaurora.org>
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Synopsys USB3 Controller (DWC3) has a restriction where size
of OUT requests (TRB) queued to the controller must be aligned
with the endpoint's max packet size. Generally, MTP userspace
module submits RX requests aligned to 512 bytes which works
fine in High speed mode. But, fails in Super-speed mode where
max packet size can be 1024 bytes.
To overcome this just change the size of RX request buffer
while submitting to DCD as the buffers are already allocated
with the size of 16KB.
Change-Id: I68c32ef3d08b2cf943b52a7269785d0a00146953
Signed-off-by: Manu Gautam <mgautam@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
In case of non-type-c usb cable extcon cable state for
EXTCON_USB_CC returns error and driver bails out. This
causes usb enumeration failure. Hence set the typec_orientation
to ORIENTATION_NONE and continue with usb enumeration.
In case of super speed mode with type-c usb cable even if extcon
cable state return error for EXTCON_USB_CC then also usb device
will get enumerated falling back to high speed mode.
Change-Id: Ibc0b0506b2090f4c090d649f6e3ae7d42b102c93
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
disconnect"
'commit 3a7ce1db5661 ("usb: gadget: gsi: Decrement USB gadget pm usage
count on cable disconnect")' added decrementing USB gadget device's pm
usage count from STATE_INITIALIZED which results into multiple time
decrementing USB gadget device's pm usage count on USB cable disconnect
as below:
1. On USB cable disconnect, USB bus suspend event is being received which
post EVT_DISCONNECTED. If state is STATE_CONNECTED, ipa_work_handles()
decrements USB gadget device's pm usage count, and move state to
STATE_INITIALIZED.
2. Due to USB cable disconnect, gsi_disable() posts EVT_DISCONNECTED.
This event is processed into STATE_CONNECTED causing additional decrement
of USB gadget device's pm usage count.
Due to above case, USB goes into low power mode with dwc3's pm usage
count negative. On connecting USB host mode cable, xhci resume is failing
due to dwc3's (parent of xhci) is having negative usage count which results
into no USB host mode functionality (pm_runtime_get_sync() returns -EBUSY).
Hence revert commit to allow USB Host mode functionality.
CRs-Fixed: 1020388
Change-Id: I853aba1d2d03945ee49adde7f0ea483cd406ce2b
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The USB power_supply now allows PROP_TYPE to be modified. Update it
to 'USB_PD' only after PD communication has been established. This
can happen only in a couple places:
- in sink mode, when a source capabilities message is received
- in source mode, when GoodCRC is received after sending the
source capabilities message successfully
Since PD_ACTIVE=1 is required to be set, also update it in source
mode prior to sending out the capabilities, and clear it if all the
attempts fail. Remove the unreachable/dead code found there as well.
Change-Id: I77c5875ee8514395a82fac0109b7cff1d507250b
Signed-off-by: Jack Pham <jackp@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Policy engine needs to provide the plug orientation upon
type-c cable plug in. qmp phy driver needs to program phy
lane based upon this information.
Change-Id: Idd236136c9f0a9163b4ae7a8405c412f1d69ca9e
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
In order to support super speed mode using type-c cable, phy
driver needs to programe the phy with appropriate lane based
upon plug orientation.
Change-Id: I893c0b729015cd22791d168453309168246961e2
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Register callbacks for plug orientations in order to
cache the current plug orientation reported by extcon.
This allows super speed phy driver to configure the
appropriate lane upon phy initialization.
Change-Id: I906005680b4cc90cc38dc3d403beebf7aa515ad7
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
User can pass arguments as part of write to requests and endpoint number
will be calculated based on the arguments. There is a chance that driver
can access ep structue that is not allocated due to invalid arguments
passed by user. Hence fix the issue by having check and return error in
case of invalid arguments.
Change-Id: I060ea878b55ce0f9983b91c50e58718c8a2c2fa1
Signed-off-by: Vijayavardhan Vennapusa <vvreddy@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
GSI EPs are assigned to the last 2 IN EPs. While allocating
the TX FIFO sizes we need to iterate over all IN EPs to correctly
allocate larger TX FIFOs for GSI accelerated endpoints.
Update the logic from using min_t(int, dwc->num_in_eps,
cdev->config->num_ineps_used + 1) to just use dwc->num_in_eps.
The EPs that are not enabled will be given the default TX FIFO
size while the ones that are enabled are given TX FIFO sizes
based on the burst size configured for that EP.
Change-Id: Ie9a21544966fb54cf9920e9c719309cc66157846
Signed-off-by: Devdutt Patnaik <dpatnaik@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
qmp phy can run in display port mode or in usb3 mode.
It is recommended to explicitly select the usb3 phy
mode before programming the phy init sequence, since
TCSR_USB3_DP_PHYMODE register is commonly used to
select mode between display port driver as well as
ssphy driver.
Change-Id: I270596868762ccd4f2f2cc9b0daaca647a2bee88
Signed-off-by: Hemant Kumar <hemantk@codeaurora.org>
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
USB GSI function driver uses usb_ipa_w work to queue different events
like EVT_CONNECTED, EVT_DISCONNECTED and more. ipa_event_handler()
uses those events as inputs to make necessary decision about performing
connect and disconnect with IPA driver. It is required that before USB
GSI driver calls ipa_usb_deinit_teth_prot(), it has invoked IPA
disconnect API ipa_usb_xdci_disconnect() if it has called
ipa_usb_xdci_connect() API. Current code is making sure that any
running usb_ipa_w work is being completed before calling
ipa_usb_deinit_teth_prot() but if work is not scheduled and pending,
ipa_usb_xdci_connect() is not called (i.e. later when usb_ipw_w work is
scheduled, EVT_DISCONNECTED is being processed but gsi_unbind() has
changed sm_state as STATE_UNINITIALIZED which results into no-ops.)
which results into USB and IPA driver go out of sync in terms of
expected state machine. Hence calling ipa_usb_init_teth_prot() on next
USB cable connect from gsi_bind_config() fails which results into no
USB GSI functionality. Fix this issue by using drain_workqueue() instead
of flush_workqueue() which makes sure that re-queue work is flushed.
CRs-Fixed: 1005018
Change-Id: I64ff559b85f901688a4abd0110ebb32a5317e34d
Signed-off-by: Mayank Rana <mrana@codeaurora.org>
|
