For link with maximum pipeline delay of 1 e.g.,
TPG use case or sensors with pipeline delay of 1,
if the request is not submitted before 2
consecutive triggers we do not get chance to
increment rd idx, in the mean time the slot which
was last applied will be reset and we will not be
able to apply request even if new requests are scheduled.
This will cause the camera to not apply any request
further, hence increasing the rd idx if no lower pd
devices are pending will fix the issue.
CRs-Fixed: 2622845
Change-Id: I012e242c7fca22abecc171ef4d7063d851bb5748
Signed-off-by: Tejas Prajapati <tpraja@codeaurora.org>
New usecase for flash driver requires to be trigger at EOF.
This change adds the new operation code support in flash
driver along with EOF enable support in crm and across
other drivers. Also to improve the performance, this change
adds the logic to dynamically update the subscribe event to
CRM, so that CRM can enqueue EOF events when Flash or any other
device add the request for EOF.
CRs-Fixed: 2633194
Change-Id: I2f68ac7fc6a4699debd39b64319728cdf17bbcfa
Signed-off-by: Jigarkumar Zala <jzala@codeaurora.org>
crm applied both sensor and isp packets,sensor packet failed to
apply due to CCI timeout error. In the meantime ISP packet applied
and that request of ISP moved to free_list.In the next epoch crm
trying to reapply both the packets of sensor and isp again.In this
case ISP always reporting an apply failure because its pending queue
doesn’t have that request.To handle this we are skipping to re-apply
the ISP successful buf_done request and trying to apply only sensor
packet in this scenario.
CRs-Fixed: 2599186
Change-Id: Id7e72fc016a75782b57f05558c615b069b416ffe
Signed-off-by: Mangalaram ARCHANA <mangar@codeaurora.org>
Currently CRM waits for a single trigger to apply settings
for all the devices on a given link. This change provides
provision for CRM to wait on two triggers prior to applying
a given setting.
CRs-Fixed: 2627065
Change-Id: If6e4c9281cfd1bf1a8ffa369daee060d79f3c39e
Signed-off-by: Karthik Anantha Ram <kartanan@codeaurora.org>
Creates a function to match expectant types for platform device
comparisons during component registration.
CRs-Fixed: 2584631
Change-Id: Ib07160b8d8f1a370ba535ffdbbc6adbe391c68d5
Signed-off-by: Karthik Jayakumar <kjayakum@codeaurora.org>
Missing video unregister was causing list delete corruption.
Also, there were double free errors in Sync and CRM driver. Fix
v4l2 issues by adding the missing unregister calls and changing
release callback to empty, because we are freeing it during
cleanup. Improve logging during bind/unbind to better reflect what
is happening.
CRs-Fixed: 2584631
Change-Id: Idc5db655d22df54e8bdb470d29896e10f6987796
Signed-off-by: Mukund Madhusudan Atre <matre@codeaurora.org>
Due to the asynchronous nature of platform probes, inter
dependency between drivers needs to be taken care during
kernel boot up. Component helper provides the facility of
adding matching drivers in a list ordered in the way we want
to bind those drivers. The CRM driver acts as component master
to make sure all slave drivers are bound before it returns
from its own bind call. Add support for serializing platform
probes through component framework.
CRs-Fixed: 2584631
Change-Id: I345da1d2b9cccf6021ac6fc899143013b7714ec4
Signed-off-by: Mukund Madhusudan Atre <matre@codeaurora.org>
When user space detects an error or does not receive
response for a request, Lets do a reset(LDAR) is triggered.
Before LDAR, user space sends flush command to the
kernel space.
To debug the cause for this situation and to dump
the information, user space sends a dump command to the
kernel space before sending flush.
As a part of this command, it passes the culprit request id
and the buffer into which the information can be dumped.
Kernel space traverses across the drivers and find the culprit hw
and dumps the relevant information in the buffer.
This data is written to a file for offline processing.
This commit implements the framework for traversal
across the RT and NRT devices.
CRs-Fixed: 2602180
Change-Id: I7e24006c20c23bfab163a2ad13b4ac6e2913bb9e
Signed-off-by: Gaurav Jindal <gjindal@codeaurora.org>
Update header files to reflect changes in functions that are being
documented.
CRs-Fixed: 2584631
Change-Id: I45e2c7830a685b43f25bfc1fb4db6800a2de6150
Signed-off-by: Karthik Jayakumar <kjayakum@codeaurora.org>
When cam request manager gets a request to deactivate a link, we
do pause for each of the device for that link. There is a race
here in current scenario that workqueues can be scheduled even
after the link has been deactivated. This can lead to unexpected
behavior. This change has updated the activate and deactivate
handling to take care of the race.
CRs-Fixed: 2601863
Change-Id: I7ff03c74c240fc3250618db518d586531d87369f
Signed-off-by: Vikram Sharma <vikramsa@codeaurora.org>
This change reduces the maximum attempts to apply a particular
request from three to two. After two failed attempts to apply
to ISP device, IRQs for ISP device will be overlapped from
two requests. Thereafter apply request at every alternate
frame will fail without a chance of recovery. Therefore we
must notify userspace to trigger recover on the link.
CRs-Fixed: 2606911
Change-Id: I526bb837a496fe1e67786b854c5afb062dddb918
Signed-off-by: Venkat Chinta <vchinta@codeaurora.org>
This change increases the watchdog timeout value from one second
to five.
CRs-Fixed: 2606911
Change-Id: I353f78369c31a513dfd94dd1988bc42c2f75b167
Signed-off-by: Venkat Chinta <vchinta@codeaurora.org>
Expire timer after destroying workqueues so that we
do not refer to watchdog timer while the link is
getting unlinked from session handle.
CRs-Fixed: 2585098
Change-Id: Ife2450ae66bd52ec704ac7d593b2daaeb20ba54d
Signed-off-by: Vikram Sharma <vikramsa@codeaurora.org>
The EPOCH at which bubble is detected no setting is applied
to any device. This change will trigger applying the bubbled
setting to the device with highest pipeline delay.
CRs-Fixed: 2564669
Change-Id: I180b4a1d2d29267f330546b8860a099baf1688e9
Signed-off-by: Karthik Anantha Ram <kartanan@codeaurora.org>
Reset the slot with the valid request on receiving
the buf_done which is reported on next EPOCH.
In case of back to back irqs slot is marked as skip idx
it will apply the request and try to reset a slot based
on MAX_PD + 1 calculation which might have the request
for which we have got a bubble. To make sure this does not
happen, reset a slot only on the buf_done for valid request.
If STOP ioctl is called from UMD for ISP, it will flush all
the lists. But this is not notified to CRM which might have
a stale entry of the half applied request, so on STOP ioctl
clear all the slots.
In case of SAT mode if one of the link has generated bubble,
while the sync_link has the sync_request slot reset sync_link
will not let the link recover of the bubble. To make sure the
recovery sync_mode is set to 0 for two consecutive slots of
the link.
CRs-Fixed: 2551701
Change-Id: If4b6f8a4a831ffddcef2cae6292d066778c18b04
Signed-off-by: Tejas Prajapati <tpraja@codeaurora.org>
Due to scheduling delays, process trigger workq can be delayed.
In the meantime, link state can be reset to IDLE. This can cause
abnormal behaviour resulting in stability issues. Also, at the
time of apply fail for flash, failed_dev is not updated.
This causes crash while notifying the error on link.
This commit prevents the execution of workq process trigger if
the link has been reset to IDLE state.
Also, failed_dev is updated if the apply for flash fails.
CRs-Fixed: 2572511
Change-Id: Iaea1e0a7a24afc9e408a1530a5875f6b6c41a45b
Signed-off-by: Gaurav Jindal <gjindal@codeaurora.org>
Due to deferred probes being dpendent on request manager to register
v4l2 subdev, failures are seen if probes are delayed. Change logic
for subdev registration to avoid failures due to delay.
CRs-Fixed: 2584631
Change-Id: Ia01d5d27ff6a6e6dbbb8e3b25b9aa53ac2fa9e38
Signed-off-by: Mukund Madhusudan Atre <matre@codeaurora.org>
Signed-off-by: Jigarkumar Zala <jzala@codeaurora.org>
Add interface to umd to give an option whether to
disable Delayed Unamp feature for a given buffer
mapping. DelayedUnmap is enabled by default if
umd doesn't explicitly asks for disable.
CRs-Fixed: 2580128
Change-Id: I66f87a9dbdfc4d9cecdc02eb24c1c670c9985cae
Signed-off-by: Pavan Kumar Chilamkurthi <pchilamk@codeaurora.org>
The v4l2 registration requires all devices that are not subdevs
to have device caps field to be non zero. Add device caps for
CRM and SYNC devices during probe.
CRs-Fixed: 2582994
Change-Id: I804e7bc00d597debc39d1f71259cf607c60ed069
Signed-off-by: Mukund Madhusudan Atre <matre@codeaurora.org>
When loading the camera FW, use the appropriate APIs
depending on the kernel version to map the FW region.
CRs-Fixed: 2564857
Change-Id: I2104b63b54bce799c8e44f36c49042de86f78c7e
Signed-off-by: Isaac J. Manjarres <isaacm@codeaurora.org>
Signed-off-by: Karthik Jayakumar <kjayakum@codeaurora.org>
Currently camera modules platform registration/remove function
is being called as a part of module_init function individually.
Building camera as module, current infrastructure cannot support
it, as DLKM only support single module_init/exit function.
This change enables that support by combining all platform
entry/remove functions centrally. Also, remove all submodule
makefiles and combined into one root makefile in order to generate
single kernel object file to dynamically loading the kernel module
in case of DLKM compilation.
CRs-Fixed: 2569808
Change-Id: I59db3ed6cef60123e474db2f222c39836056bd6e
Signed-off-by: Jigarkumar Zala <jzala@codeaurora.org>
Signed-off-by: Karthik Jayakumar <kjayakum@codeaurora.org>
If link is not ready on master epoch, but it becomes ready on slave
epoch, then master skips apply and slave applies request and goes ahead
of master.
Fix this by skipping apply on slave if master slot in not in applied
state.
CRs-Fixed: 2562008
Change-Id: Ic612eedfeedf2a6ea50737a49a6f1f31a5de1dc2
Signed-off-by: Trishansh Bhardwaj <tbhardwa@codeaurora.org>
Adding CRM interface to stop SOF timer from isp during flush.
During flush hardware is getting stop and will not send SOF
notification to CRM so need to pause SOF timer. Whenever SOF
timer will get expire,do not need to send error to UMD during
pause time.
CRs-Fixed: 2564389
Change-Id: I6d85f2c658c30dbe211f0ec9d83bca323a5c265b
Signed-off-by: Chandan Kumar Jha <cjha@codeaurora.org>
In case userspace fails to dequeue a v4l2 event with a
valid request, kernel will dump that message as error and
rest will be rate limited.
CRs-Fixed: 2558548
Change-Id: I1c6769f47d0e1ca4a3ce5c6467f6c280104ae2b8
Signed-off-by: Karthik Anantha Ram <kartanan@codeaurora.org>
Adds fixes to makefile in order to support legacy kernel build systems.
CRs-Fixed: 2560543
Change-Id: Iab571871e5171aab501c41496cc09e3c5d942985
Signed-off-by: Karthik Jayakumar <kjayakum@codeaurora.org>
In case of long exposure shots, watchdog timer needs to be modified
according to the requested value to avoid trigger. Add support for
modifying timer value and changing it back to normal.
CRs-Fixed: 2530691
Change-Id: I97bdd92d2ed461066bbf746bc6293094a444f8a5
Signed-off-by: Mukund Madhusudan Atre <matre@codeaurora.org>
Signed-off-by: Karthik Anantha Ram <kartanan@codeaurora.org>
Arm arch does not support dividing 64 bit integer, replacing it with
do_div call.
CRs-Fixed: 2543730
Change-Id: I86f8c706d20978b4c7bfd78c499dacb482c4d37a
Signed-off-by: Trishansh Bhardwaj <tbhardwa@codeaurora.org>
Fix improper use of divide and modulo operator on 64 bit numbers.
Fix variable type to work with both 32/64 bit arch.
CRs-Fixed: 2543730
Change-Id: Ifa52d46dece3434d41308d284982a0cd8e17cd1b
Signed-off-by: Trishansh Bhardwaj <tbhardwa@codeaurora.org>
As part of flush only the request slots were reset, we need
to reset the slots in each of the pd tables.
CRs-Fixed: 2545130
Change-Id: I34054dea2e5849739df9dd0a77d908a9af9fe3be
Signed-off-by: Karthik Anantha Ram <kartanan@codeaurora.org>
Kernel uses v4l2 events to communicate with userspace.
If userspace is not dequeuing events and kernel is pushing more events
then v4l2_event_queue fails silently.
This change uses v4l2 callbacks to log v4l2_event_queue failure.
CRs-Fixed: 2541894
Change-Id: I98f8f609f615104ec6dde569bf1f39de1eb368fc
Signed-off-by: Trishansh Bhardwaj <tbhardwa@codeaurora.org>
Return different error number when new requests and update
packets are rejected due to bad request ID. This allows
userspace to differentiate this specific reason for failure.
CRs-Fixed: 2518451
Change-Id: I0400c6f2b2e0baf99bee4b4a3736bb1eab646627
Signed-off-by: Venkat Chinta <vchinta@codeaurora.org>
The req of two links should be same if current
frame sync with previous frame of sync link, CRM can
do self-correction based on this condition.
CRs-Fixed: 2492019
Change-Id: Idb208958a850ec1fe12f12443e7ecc2e126970d8
Signed-off-by: Depeng Shao <depengs@codeaurora.org>
Do the initial sync by SOF timestamp, if the gap
between the initial SOF of link and the last SOF
of sync link is less than half of frame druation,
CRM should abandon this frame since this frame
should sync with next frame of sync link.
CRs-Fixed: 2492019
Change-Id: I204e3ed49bcf4ac7424aaf5109ad5ce3bc3a2789
Signed-off-by: Depeng Shao <depengs@codeaurora.org>
When a request setting could not be applied by any device
on the link, it does not indicate a fatal error. It only
leads to a frame drop. Changing the apply failure logs to
warn as opposed to error.
CRs-Fixed: 2514752
Change-Id: Idbca2dbb06b0bbb4c380264cf31b665188306396
Signed-off-by: Karthik Anantha Ram <kartanan@codeaurora.org>
