DRA72XEVM: Streaming multiple cameras in QML application throws segmentation fault.

Sam George39

Part Number: DRA72XEVM

Hi there,

My application is supposed to stream 4 cameras simultaneously. But after few seconds of streaming, the Qt app exits unexpectedly throwing the following error:

MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115:   dce_buf_lock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_release: Error: unable to release buffer
../git/libdce_linux.c:145:   dce_buf_unlock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115:   dce_buf_lock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115:   dce_buf_lock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115:   dce_buf_lock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_release: Error: unable to release buffer
../git/libdce_linux.c:145:   dce_buf_unlock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115:   dce_buf_lock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115:   dce_buf_lock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_release: Error: unable to release buffer
../git/libdce_linux.c:145:   dce_buf_unlock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115:   dce_buf_lock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_release: Error: unable to release buffer
../git/libdce_linux.c:145:   dce_buf_unlock   ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
[ 1539.100785] rpmsg_rpc rpmsg-dce: unwinding UVA to RDA translations! translation = 1
[ 1539.140317] rpmsg_rpc rpmsg-dce: failed to translate all pointers for remote core!
../git/libdce_linux.c:115:d   ce_buf_lock   ERROR: Failed eError == DE_EOK error val -5MmRpc_call: Error: write failed
QObject::~QObject: Timers cannot be stopped from another thread
../git/libdce.c:965:   process   ERROR: Failed eError == DCE_EOK error val -5root@dra7xx-evm:~/ITS_HMI_Resltn_1920# [ 1597.783268] omap-iommu 55082000.mmu: 55082000.mmu: version 2.1

Is it because ducatidecoder instances which my application needs is 4 but only 2 can be created at a time. If so what can be the solution to this??

over 5 years ago

0 Ramprasad over 5 years ago

TI__Mastermind 30480 points

Hi Sam,
Is your SDK PSDKLA3.04? Please share the gstreamer pipeline you are using.

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

Hi Ram,

SDK version PSDKLA 3.04.0.03

I am using qml Mediaplayer object.

MediaPlayer {
id: mediaplayer1
source: "rtsp://admin:password@192.168.15.51:554/cam/realmonitor?channel=1&subtype=1"
autoPlay: true
}

and there are 4 of these kind. I don't know the underlying gstreamer pipeline.

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,

You had reported this issue
e2e.ti.com/.../697194
Have you made the changes required?

There is limit on number of ducati instances that can be created from one process. Currently it is set to 5.
I suspect if this is the issue if the issue happens during creation of 3rd streaming.

I have few questions
1) What is the capture resolution for each capture?
2) What is the format of the stream? jpeg or h264 ?
3) Is there any encoding involved here ?
4) Are you seeing the issue as soon as you start third capture?

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

We made the changes in libdce and ipumm. Is it enough to run the following command

./build-core-sdk.sh dra7xx-evm

to rebuild both ipumm and libdce and get this changes in rootfs?

1. resolution can be 720p or 352x288.

2. streaming is in h264

3. don't know how qt is implementing gstreamer but i am sure that there is some decoding involved here.

4. yes as soon as the third stream was introduced the error was observed.

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
Refer this
processors.wiki.ti.com/index.php

to rebuild ipumm and libdce after modifications.

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

Hi Ram,

We made the changes in libdce and ipumm, generated and installed the ipk files. But still on running the qt app, we are facing the same issue..
We did modification through following commands:

host $ ./build-specific-recipe.sh dra7xx-evm -f -c configure ipumm-fw
host $ ./build-specific-recipe.sh dra7xx-evm -f -c compile ipumm-fw

host $ ./build-specific-recipe.sh dra7xx-evm -f -c install ipumm-fw
host $ ./build-specific-recipe.sh dra7xx-evm -f -c package ipumm-fw
host $ ./build-specific-recipe.sh dra7xx-evm -f -c package_write_ipk ipumm-fw

and same thing i did for libdce and installed in the system.

And how to increase the heap memory for ducati encoders?

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
Can you add few prints in ipumm and confirm your changes are taken?
You can use cat /sys/kernel/debug/remoteproc/remoteproc0/trace0 to get the prints on target from ipumm.

IPUMM reserves 40MB of heap for the usecase of 2 Full-HD decoding and one Full-HD encoding. I guess for your usecase of 4 720p , this should be sufficient. but refer the datasheets from ipumm repository for confirmation.

Refer this www.ti.com/.../sprac08a.pdf
to increase the heap memory of IPUMM.

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

Hi Ram,

we added print through DEBUG in ipumm in get_client_instance and get_client functions. And repeated the same building process and we installed in system through opkg --force-reinstall but we didn't observe any prints in trace0 which were to be printed by DEBUG.

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Use System_printf

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

Hi Ram,

There was no new print in trace0, it was same as before. Are we missing any step or are we doing something wrong?

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
Sorry , it is
cat /sys/kernel/debug/remoteproc/remoteproc1/trace0.

remoteproc0 is for IPU1 and it is not used in PSDKLA.

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

Still no luck,

I think we are facing issue while installing the ipk.
What we did is first we tried to remove the package. But there were some errors while doing that.
we used following command to that
opkg remove ipumm-fw*,
opkg remove libcde*
Is this the correct way?

root@dra7xx-evm:~# opkg list|grep ipu
ipumm-fw - 3.00.14.00-r1.5
ipumm-fw - 3.00.14.00-r1.7
ipumm-fw-dev - 3.00.14.00-r1.7
root@dra7xx-evm:~# opkg remove --force-remove ipumm-fw
No packages removed.

And we then tried to force install but still failed,

root@dra7xx-evm:~# opkg install --force-reinstall ipumm-fw_3.00.14.00-r1.7_armv7ahf-neon.ipk
Not selecting ipumm-fw 3.00.14.00 as installing it would break existing dependencies.
Not selecting ipumm-fw 3.00.14.00 as installing it would break existing dependencies.
Installing ipumm-fw (3.00.14.00-r1.7) on root.
Collected errors:
* get_pkg_url: Package ipumm-fw is not available from any configured src.
* opkg_install_pkg: Failed to download ipumm-fw. Perhaps you need to run 'opkg update'?
* opkg_install: Cannot install package ipumm-fw.

I think we did it the wrong way!!!

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam ,
opkg --force_install xxx.ipk should have worked.

Can you directly copy dra7-ipu2-fw.xem4 and libdce.so generated in yocto to /lib/firmware and /usr/lib respectively?

You can first try your first issue
e2e.ti.com/.../697194
works or not.

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

Still on running 4 pipeline we are facing issue, and qmlplayer still show the error:

MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115: dce_buf_lock ERROR: Failed eError == DCE_EOK error val -5MmRpc_release: Error: unable to release buffer
../git/libdce_linux.c:145: dce_buf_unlock ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115: dce_buf_lock ERROR: Failed eError == DCE_EOK error val -5MmRpc_use: Error: unable to declare buffer use
../git/libdce_linux.c:115: dce_buf_lock ERROR: Failed eError == DCE_EOK error val -5MmRpc_call: Error: write failed
../git/libdce.c:965: process ERROR: Failed eError == DCE_EOK error val -5MmRpc_release: Error: unable to release buffer
QObject::~QObject: Timers cannot be stopped from another thread
../git/libdce_linux.c:145: dce_buf_unlock ERROR: Failed eError == DCE_EOK error val -5

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
This error still doesn't confirm that issue happens with creation of third capture.
How long is it taking to reproduce the error?

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

Hi Ram,

We made a simple test app using qt/qml:

Main.cpp:

#include <QGuiApplication>
#include <QQmlApplicationEngine>

int main(int argc, char *argv[])
{
    QGuiApplication app(argc, argv);

    QQmlApplicationEngine engine;
    engine.load(QUrl(QStringLiteral("qrc:/main.qml")));

    return app.exec();
}

main.qml:

import QtQuick 2.3
import QtQuick.Window 2.2
import QtMultimedia 5.5

Window {
    id:cameraTestWindow
    visible: true
    width: 1900
    height:1200

    Rectangle {
        id:mainrect
        anchors.fill: parent
        color:"black"

        MediaPlayer {
            id: mediaplayer1
            source: "rtsp://admin:password@192.168.15.51:554/cam/realmonitor?channel=1&subtype=1"
            autoPlay: true
        }

        VideoOutput {
            id:videoOp1
            x:620//250//90//0
            y:80//30//0
            width:550//530//320
            height:425//250//240
            source: mediaplayer1
            //fillMode: VideoOutput.Stretch
            MouseArea {
                anchors.fill: parent
                onClicked: {

                    videoOp1.width = 1100;
                    videoOp1.height = 850;
                    videoOp2.visible = false
                    videoOp3.visible = false
                    videoOp4.visible = false
                }
                onDoubleClicked: {

                    videoOp1.width = 550;
                    videoOp1.height = 425;
                    videoOp2.visible = true
                    videoOp3.visible = true
                    videoOp4.visible = true

                }
            }

        }

        MediaPlayer {
            id: mediaplayer2
            source: "rtsp://admin:password@192.168.15.52:554/cam/realmonitor?channel=1&subtype=1"
            autoPlay: true
        }

        VideoOutput {
            id:videoOp2
            x:1170//250//90//0
            y:80//30//0
            width:550//530//320
            height:425//250//240
            source: mediaplayer2
           // fillMode: VideoOutput.Stretch
            MouseArea {
                anchors.fill: parent
                onClicked: {

                    videoOp2.width = 1100;
                    videoOp2.height = 850;
                    videoOp2.x = 620
                    videoOp1.visible = false
                    videoOp3.visible = false
                    videoOp4.visible = false
                }
                onDoubleClicked: {

                    videoOp2.width = 550;
                    videoOp2.height = 425;
                    videoOp2.x = 1170
                    videoOp1.visible = true
                    videoOp3.visible = true
                    videoOp4.visible = true

                }
            }
        }

        MediaPlayer {
            id: mediaplayer3
            source: "rtsp://admin:password@192.168.15.53:554/cam/realmonitor?channel=1&subtype=1"
            autoPlay: true
        }

        VideoOutput {
            id:videoOp3
            x:620//250//90//0
            y:505//30//0
            width:550//530//320
            height:425//250//240
            source: mediaplayer3
            //fillMode: VideoOutput.Stretch
            MouseArea {
                anchors.fill: parent
                onClicked: {

                    videoOp3.width = 1100;
                    videoOp3.height = 850;
                    videoOp3.y = 80;
                    videoOp1.visible = false
                    videoOp2.visible = false
                    videoOp4.visible = false
                }
                onDoubleClicked: {

                    videoOp3.width = 550;
                    videoOp3.height = 425;
                    videoOp3.y = 505;
                    videoOp1.visible = true
                    videoOp2.visible = true
                    videoOp4.visible = true

                }
            }
        }

        MediaPlayer {
            id: mediaplayer4
            source: "rtsp://admin:password@192.168.15.54:554/cam/realmonitor?channel=1&subtype=1"
            autoPlay: true
        }

        VideoOutput {
            id:videoOp4
            x:1170//250//90//0
            y:505//30//0
            width:550//530//320
            height:425//250//240
            source: mediaplayer4
           // fillMode: VideoOutput.Stretch
            MouseArea {
                anchors.fill: parent
                onClicked: {

                    videoOp4.width = 1100;
                    videoOp4.height = 850;
                    videoOp4.x = 620
                    videoOp4.y = 80
                    videoOp1.visible = false
                    videoOp2.visible = false
                    videoOp3.visible = false
                }
                onDoubleClicked: {

                    videoOp4.width = 550;
                    videoOp4.height = 425;
                    videoOp4.x = 1170
                    videoOp4.y = 505
                    videoOp1.visible = true
                    videoOp2.visible = true
                    videoOp3.visible = true

                }
            }
        }

    }
}

On running this, we observe the error which is being discussed. Starting few seconds, we see 4 video screens with a green background, on touching any one window the media player should zoom in. On zooming in error commences and then we see frames intermixing into other video screens and after3-4 seconds the app crashes.

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
Can you check dmesg log and confirm if you are seeing below prints in dmesg?
"omapdrm omapdrm.0: could not remap"

or you can check below command also and confirm if 128MB of TILER memory is exhausted or not when your usecase is running.
cat /sys/kernel/debug/dri/0/tiler_map

I don't have setup with 4 USB cameras to reproduce your issue. Is there any other way to reproduce the issue?

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

There is no such print for "omapdrm".

And while running the app, how to analyse the output of the statement?
we got output like,

096:
097:
098:
099:
100:
101:
102:
103:
104:
105:
106:
107:
108:
109: <===============288K===============><=======288K=======
110:================><===============288K===============><===============288K===============><===============288K===============><==
111:=========================288K===><===============288K===============><===============288K===============><=========288K=========
112:============><===============288K===============><===============288K===============><===============288K===============><=288K=
113:============================><===============288K===============><===============288K===============><===========288K===========
114:========> <===============288K===============><===============288K===============><===288K===
115:========================><===============288K===============><===============288K===============><=============288K=============
116:====><===============288K===============><===============288K===============><===============288K===============><=====288K=====
117:====================><===============288K===============><===============288K===============><===============288K===============
118:><===============288K===============><===============288K===============>XXXXXXX<===============================================
119:================================================================================================================================
120:================================================================================================================================
121:================================================================================================================================
122:=============================================================9600K==============================================================
123:================================================================================================================================
124:================================================================================================================================
125:================================================================================================================================
126:================================================================================================================================
127:===============================================================================================================================>

You can use single camera to achieve the same by giving same address to all mediaplayer. and still you will get same error.

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
This pattern tells how much TILER memory is available . For decoding and encoding using ducati, TILER will be used.
I though with 4 instances of encoding and decoding 128 MB is exhausting and hence it is failing.

Since underlying gstreamer pipeline is not properly known I am not able to comment.

Do you have a yocto build system? If yes, You need to apply few fixes on gstvpe plugin. I hope that should resolve the issue.

You need to apply top 3 commit-IDs on gstvpe plugin. These three are not available on PSDKLA3.04
git://git.ti.com/glsdk/gst-plugin-vpe.git

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

How to apply these patches?? Is there any documentation for that?

1. git.ti.com/.../b9dba5a8bc7d14d55125abdce04672e3dc30bbdf
2. git.ti.com/.../a1fe1f3e3ae0582a4f84ae8fa7ca052fff2e9f76
3. git.ti.com/.../4fe4beaa1498e27878bf04020aef0099c17f2262

Are these the patches?

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
You need to create bbappend recipe file to apply the patch in yocto.
Please make the changes manually from this single patch
git.ti.com/.../a1fe1f3e3ae0582a4f84ae8fa7ca052fff2e9f76

and rebuild gstvpe, copy libgstvpe.so to you rootfs(/usr/lib/gstreamer-1.0) and check if the issue comes or not

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

We re-compiled gstreamer-plugins-vpe with the following steps

First we edited gstvpe.c in following path

ti-processor-sdk-linux-automotive-dra7xx-evm-03_04_00_03/yocto-layers/build/arago-tmp-external-linaro-toolchain/work/dra7xx_evm-linux-gnueabi/gstreamer1.0-plugins-vpe/git-r2.16/git/src/gstvpe.c

then issued yocto commands

./build-specific-recipe.sh dra7xx-evm -f -c configure gstreamer1.0-plugins-vpe
./build-specific-recipe.sh dra7xx-evm -f -c compile gstreamer1.0-plugins-vpe
./build-specific-recipe.sh dra7xx-evm -f -c install gstreamer1.0-plugins-vpe

we got some errors for package command

./build-specific-recipe.sh dra7xx-evm -f -c package gstreamer1.0-plugins-vpe

NOTE: Executing RunQueue Tasks
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-tests went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-drivers went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-radeon went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-nouveau went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-omap went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-intel went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-exynos went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-kms went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-freedreno went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-amdgpu went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-dbg went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-staticdev went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-dev went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-doc went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm-locale went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdrm-2.4.66-r0.arago1 do_packagedata: QA Issue: Package version for package libdrm went backwards which would break package feeds from (0:2.4.67-r0.arago1.0 to 0:2.4.66-r0.arago1.0) [version-going-backwards]
ERROR: libdce-3.00.08.02-r0 do_packagedata: QA Issue: Package version for package libdce-dbg went backwards which would break package feeds from (0:3.00.14.00-r1.1 to 0:3.00.08.02-r0.0) [version-going-backwards]
ERROR: libdce-3.00.08.02-r0 do_packagedata: QA Issue: Package version for package libdce-staticdev went backwards which would break package feeds from (0:3.00.14.00-r1.1 to 0:3.00.08.02-r0.0) [version-going-backwards]
ERROR: libdce-3.00.08.02-r0 do_packagedata: QA Issue: Package version for package libdce-dev went backwards which would break package feeds from (0:3.00.14.00-r1.1 to 0:3.00.08.02-r0.0) [version-going-backwards]
ERROR: libdce-3.00.08.02-r0 do_packagedata: QA Issue: Package version for package libdce-doc went backwards which would break package feeds from (0:3.00.14.00-r1.1 to 0:3.00.08.02-r0.0) [version-going-backwards]
ERROR: libdce-3.00.08.02-r0 do_packagedata: QA Issue: Package version for package libdce-locale went backwards which would break package feeds from (0:3.00.14.00-r1.1 to 0:3.00.08.02-r0.0) [version-going-backwards]
ERROR: libdce-3.00.08.02-r0 do_packagedata: QA Issue: Package version for package libdce went backwards which would break package feeds from (0:3.00.14.00-r1.1 to 0:3.00.08.02-r0.0) [version-going-backwards]
NOTE: Tasks Summary: Attempted 1321 tasks of which 1298 didn't need to be rerun and all succeeded.
NOTE: Writing buildhistory

Summary: There were 4 WARNING messages shown.
Summary: There were 22 ERROR messages shown, returning a non-zero exit code.
[PSDKLA]>
[PSDKLA]> Completed Yocto build at Tue Jul 24 12:09:00 IST 2018
[PSDKLA]>

But ipk was generated with the last write_ipk command

./build-specific-recipe.sh dra7xx-evm -f -c package_write_ipk gstreamer1.0-plugins-vpe

copied the libstvpe.so to /usr/lib/gstreamer-1.0/

after rebooting the system we had observed the following,

1. when using 3 different cameras in the app, the error probability is around 0.1, the camera screen are perfectly streched and green screen appears rarely.

2. when using 4 cameras the fourth one has green screen an error starts and program crashes. Same buffer issue..

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
>>1. when using 3 different cameras in the app, the error probability is around 0.1, the camera screen are perfectly streched and green screen >>appears rarely.
What was earlier observation without gstvpe patch?

>>2. when using 4 cameras the fourth one has green screen an error starts and program crashes. Same buffer issue..
I didn't get what is green screen here. Can you share the ipumm log cat/sys/kernel/debug/remoteproc/remoteproc1/trace0 when this issue happens? and also dmesg logs when the issue happens.

0 Ramprasad over 5 years ago in reply to Ramprasad

TI__Mastermind 30480 points

Hi Sam,
Are you able to resolve your old issue ?
e2e.ti.com/.../697194

0 Sam George39 over 5 years ago in reply to Ramprasad

Intellectual 840 points

root@dra7xx-evm:~# tail -f /sys/kernel/debug/remoteproc/remoteproc1/trace0

[0][ 0.943] OMAPRPC: function query of type 1

[0][ 0.944] OMAPRPC: Replying with msg type: 3 to addr: 1025 from: 102 len: 216

[0][ 0.951] OMAPRPC: received msg type: 2 len: 216 from addr: 1025

[0][ 0.951] OMAPRPC: function query of type 1

[0][ 0.951] OMAPRPC: Replying with msg type: 3 to addr: 1025 from: 102 len: 216

[0][ 0.951] OMAPRPC: received msg type: 2 len: 216 from addr: 1025

[0][ 0.951] OMAPRPC: function query of type 1

[0][ 0.951] OMAPRPC: Replying with msg type: 3 to addr: 1025 from: 102 len: 216

[0][ 1.000] loadTask: cpu load = 3%

[0][ 2.000] loadTask: cpu load = 0%

[0][ 11.163] OMAPRPC: received msg type: 6 len: 56 from addr: 1026

[0][ 11.163] createService: new OMX Service at endpoint: 103

[0][ 11.163] OMAPRPC: created service instance named: rpmsg-dce (status=0) addr: 103

[0][ 11.163] OMAPRPC: Replying with msg type: 8 to addr: 1026 from: 102 len: 16

[0][ 12.000] loadTask: cpu load = 34%

[0][ 13.000] loadTask: cpu load = 51%

[0][ 14.000] loadTask: cpu load = 47%

[0][ 16.000] loadTask: cpu load = 44%

[0][ 17.000] loadTask: cpu load = 35%

[0][ 18.000] loadTask: cpu load = 42%

[0][ 19.000] loadTask: cpu load = 39%

[0][ 20.000] loadTask: cpu load = 41%

[0][ 25.000] loadTask: cpu load = 38%

[0][ 26.000] loadTask: cpu load = 42%

[0][ 27.000] loadTask: cpu load = 40%

[0][ 32.000] loadTask: cpu load = 38%

[0][ 34.000] loadTask: cpu load = 44%

[0][ 36.000] loadTask: cpu load = 42%

[1][ 38.000] loadTask: cpu load = 40%

[0][ 40.000] loadTask: cpu load = 43%

[0][ 42.000] loadTask: cpu load = 45%

[0][ 43.000] loadTask: cpu load = 40%

[0][ 44.000] loadTask: cpu load = 43%

[0][ 45.777] OMAPRPC: received msg type: 4 len: 12 from addr: 1026

[0][ 45.777] OMAPRPC: destroying instance addr: 103

[1][ 45.777] RcmServer_serverThrFxn_P: Exiting thread.

[0][ 45.777] deleteService: removed RcmServer at endpoint: 103

[0][ 45.777] OMAPRPC: Replying with msg type: 7 to addr: 1026 from: 102 len: 16

[0][ 46.000] loadTask: cpu load = 31%

[0][ 47.000] loadTask: cpu load = 1%

dmesg

[ 11.609201] tvp5158 1-0058: Unable to of_probe

[ 11.613821] usbcore: registered new interface driver usbhid

[ 11.624238] tvp5158 1-0058: Unable to of_probe

[ 11.629008] usbhid: USB HID core driver

[ 20.116551] omap_hwmod: mmu1_dsp1: _wait_target_disable failed

[ 20.133350] omap_hwmod: mmu0_dsp1: _wait_target_disable failed

[ 20.147540] omap_hwmod: mmu1_dsp2: _wait_target_disable failed

[ 20.161429] omap_hwmod: mmu0_dsp2: _wait_target_disable failed

[ 87.077003] omap-iommu 55082000.mmu: 55082000.mmu: version 2.1

[ 121.674274] rpmsg_rpc rpmsg-dce: unwinding UVA to RDA translations! translation = 1

[ 121.693947] rpmsg_rpc rpmsg-dce: failed to translate all pointers for remote core!

The program crashes at the log section marked bold.

0 Sam George39 over 5 years ago in reply to Sam George39

Intellectual 840 points

Hi Ram,
Currently we are not looking at that old issue, we are concentrating on qml part now.

0 Sam George39 over 5 years ago in reply to Sam George39

Intellectual 840 points

You can reproduce this issue using VLC from another PC and stream the video using rtsp method. And use that address in mediaplayer source in board.

0 Ramprasad over 5 years ago in reply to Sam George39

TI__Mastermind 30480 points

Hi Sam,
I configured VLC to stream a 640x272 stream as rtsp server and on DRA7x EVM, I am able to launch 4 instances of this pipeline
gst-launch-1.0 playbin uri=rtsp://172.24.145.0:8554/stream video-sink=waylandsink

it shows all 4 videos on display

Processors

Processors forum

DRA72XEVM: Streaming multiple cameras in QML application throws segmentation fault.