Dual stream(H264 +MJPEG) alignement trap if width more than 2048

Vamsi

Hi all,

We are working on DM368IPNC for an IP camera

We are trying to stream dual streams (H.264 + MJPEG) and we are struck with an Alignment Trap in av_server.out.

The arguments for the exe are ./av_server.out DM368 NTSC APPRO2A AEWB GE_4MP H264 1500000 VBR AUTO MJPEG 75 MENUOFF.

The problem seems to occur when the width of H.264 stream is more than 2048. The width is not a concern if we are configuring for H264 alone , but when configured to stream both H264 and MJPEG we are facing the problem when the H264 stream width is more than 2048. We already upgraded to 256M memory and I don’t think we are having any memory issues for this problem.

I got the Backtrace from GDB after the crash and copying that here.

Program received signal SIGBUS, Bus error.

[Switching to Thread 762.820]

0x000ad964 in ALG_activate ()

(gdb) bt

#0 0x000ad964 in ALG_activate ()

#1 0x000ad164 in Algorithm_activate ()

#2 0x000ab138 in VISA_enter ()

#3 0x000a6c7c in IMGENC1_process ()

#4 0x000694f0 in ALG_jpgEncRun (hndl=0x4c9ac8, prm=0x4557b470, runStatus=0x4557b4b4) at alg_jpgEnc.c:174

#5 0x00061cbc in ALG_vidEncRun (hndl=0x4c9888, prm=0x4557bc8c, runStatus=0x4557bc7c) at alg_vidEnc.c:702

#6 0x00036af4 in VIDEO_encodeTskRun (streamId=1) at videoEncodeThr.c:280

#7 0x00036fec in VIDEO_encodeTskMain (pTsk=0x3db330, pMsg=0x4f52e8, curState=0) at videoEncodeThr.c:385

#8 0x000975f4 in OSA_tskThrMain (pPrm=0x3db330) at osa_tsk.c:17

#9 0x4002a8f8 in ?? ()

Cannot access memory at address 0x1

#10 0x4002a8f8 in ?? ()

Cannot access memory at address 0x1

The trap message on the console is

Alignment trap: av_server.out (814) PC=0x000ad964 Instr=0xe5933004 Address=0x95949497 FSR 0x001

From the backtrace it looks like the problem is with JPEGEncRun, but this dissapears if I reduce the size of H264 from 2056x2056 to 2048x2056 and I could stream out both the stream

The MJpeg dimesnions are 528x480.

Is there any restriction at the codec level we should be aware of. Would apprectiate any thoughts.

Regards,

Vamsi

over 13 years ago

0 Raghu Kudva over 13 years ago

TI__Genius 11000 points

hi Vamsi,

Can you please tell us whether you areusing latest codec for H264 and MJPEg used in V3.0 IPNC RDK.

Those codecs support Megapixel mode more than 2048 width.

Howver, in general can you check how you have writter the driver for 4MP mode.

Please refer to 5MP mode driver in the IPNC RDK and check whether you are using two pass resize operation for width more then 2048.

Resizer has restriction of width 2048 for single pass resizing.

Regards,

Raghu

0 Vamsi over 13 years ago in reply to Raghu Kudva

Prodigy 110 points

Hi Raghu,

Thanks for the reply. It gave me some insights of where the problem can be.

We ae using Version2.6 IPNC in our code. Are you suggesting that using V3.0 might clear this problem?

I believe we a re using two pass as I can stream the 2056x2056 rezsized output when I am configuring it for a single stream.

I am not sure how to check whether it is configured for the twopass or single pass method. I compared with 5MP implementation and it is quite identical.

Could you please let me know if there is any particular variable to check whether the camera is set to two pass or single pass.

PS: 2056x2056 MJPEG + 640 x480 MJPEG is working fine only 2056x2056 H264 + 640x480 MJPEG is failing.

Thanks again for the reply

Regards,

Vamsi

0 Raghu Kudva over 13 years ago in reply to Vamsi

TI__Genius 11000 points

From the resolution it looks the width and height are not multiple of 16, can you make that and check this

Also, can you please share the complete log?

Regards,

Raghu

0 Vamsi over 13 years ago in reply to Raghu Kudva

Prodigy 110 points

Hi Raghu,

I tried with 2064x2064 resolution and I am still getting the same alignment trap. Please find attached the log I got from my console. Please let me know if you are looking for some other log and the steps to produce that. Thanks for you help .

Regards,

Vamsi

Fullscreen 7824.AvserverAlignmentlog.txt Download

DM36x initialization passed!
TI UBL Base Version: 1.50
Boot Loader BootMode = NAND
Starting NAND Copy...
Valid magicnum 0xA1ACED66, found in block 0x00000008
entryPoint:0x81080000
numPage   :0x00000050
block     :0x00000009
page      :0x00000000
ldAddress :0x81080000
Boot Mode Task Completed

IPNC UBL Version: 1.1.3
Platform: DM368

Jumping to entry point at 0x81080000

U-Boot 1.3.4 (Feb  7 2012 - 14:06:00) DM368-IPNC-1.0.2

I2C:   ready
DRAM:  256 MB
NAND:  NAND device: Manufacturer ID: 0xec, Chip ID: 0xf1 (Samsung NAND 128MiB 3,3V 8-bit)
Bad block table found at page 65472, version 0x01
Bad block table found at page 65408, version 0x01
128 MiB
In:    serial
Out:   serial
Err:   serial
ARM Clock :- 432MHz
DDR Clock :- 340MHz
Ethernet PHY: GENERIC @ 0x01
Hit any key to stop autoboot:  0

Loading from NAND 128MiB 3,3V 8-bit, offset 0x500000
   Image Name:   Linux-2.6.18_pro500-davinci_IPNC
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    1825924 Bytes =  1.7 MB
   Load Address: 80008000
   Entry Point:  80008000
## Booting kernel from Legacy Image at 80700000 ...
   Image Name:   Linux-2.6.18_pro500-davinci_IPNC
   Image Type:   ARM Linux Kernel Image (uncompressed)
   Data Size:    1825924 Bytes =  1.7 MB
   Load Address: 80008000
   Entry Point:  80008000
   Verifying Checksum ... OK
   Loading Kernel Image ... OK
OK

Starting kernel ...

Linux version 2.6.18_pro500-davinci_IPNC_DM368_2.6.0-gc2a0433d-dirty (marks@curium) (gcc version 4.2.0 (MontaVista 4.2.0-16.0.32.0801914 2008-08-30)) #1 PREEMPT Thu Jan 26 08:36:55 GMT 2012
CPU: ARM926EJ-S [41069265] revision 5 (ARMv5TEJ), cr=00053177
Machine: DaVinci DM368 IPNC
Memory policy: ECC disabled, Data cache writeback
DaVinci DM0365 variant 0x8
PLL0: fixedrate: 24000000, commonrate: 170000000, vpssrate: 340000000
PLL0: vencrate_sd: 75555555, ddrrate: 340000000 mmcsdrate: 48571428
PLL1: armrate: 432000000, voicerate: 20571428, vencrate_hd: 27000000
CPU0: D VIVT write-back cache
CPU0: I cache: 16384 bytes, associativity 4, 32 byte lines, 128 sets
CPU0: D cache: 8192 bytes, associativity 4, 32 byte lines, 64 sets
Built 1 zonelists.  Total pages: 12288
Kernel command line: eth=00:11:35:01:00:47 console=ttyS1,115200n8 noinitrd rw eth=00:11:35:01:00:47 ip=dhcp root=/dev/nfs nfsroot=192.168.0.105:/home/vamsiy/alta/Alta-head/ipnc/target/filesys, nolock mem=48M
TI DaVinci EMAC: kernel boot params Ethernet address: 00:11:35:01:00:47
TI DaVinci EMAC: kernel boot params Ethernet address: 00:11:35:01:00:47
PID hash table entries: 256 (order: 8, 1024 bytes)
Clock event device timer0_0 configured with caps set: 07
Console: colour dummy device 80x30
Dentry cache hash table entries: 8192 (order: 3, 32768 bytes)
Inode-cache hash table entries: 4096 (order: 2, 16384 bytes)
Memory: 48MB = 48MB total
Memory: 44672KB available (3129K code, 645K data, 176K init)
Security Framework v1.0.0 initialized
Capability LSM initialized
Mount-cache hash table entries: 512
CPU: Testing write buffer coherency: ok
NET: Registered protocol family 16
DaVinci: 104 gpio irqs
MUX: initialized GPIO20
MUX: initialized I2C_SCL
Generic PHY: Registered new driver
ch0 default output "COMPOSITE", mode "NTSC"
VPBE Encoder Initialized
SCSI subsystem initialized
NET: Registered protocol family 2
IP route cache hash table entries: 512 (order: -1, 2048 bytes)
TCP established hash table entries: 2048 (order: 1, 8192 bytes)
TCP bind hash table entries: 1024 (order: 0, 4096 bytes)
TCP: Hash tables configured (established 2048 bind 1024)
TCP reno registered
VFS: Disk quotas dquot_6.5.1
Dquot-cache hash table entries: 1024 (order 0, 4096 bytes)
squashfs: version 3.1 (2006/08/19) Phillip Lougher
JFFS2 version 2.2. (NAND) (C) 2001-2006 Red Hat, Inc.
yaffs Jan 26 2012 08:35:09 Installing.
SGI XFS with no debug enabled
Initializing Cryptographic API
io scheduler noop registered
io scheduler anticipatory registered (default)
LTT : ltt-facilities init
LTT : ltt-facility-core init in kernel
DAVINCI-WDT: DaVinci Watchdog Timer: heartbeat 60 sec
Serial: 8250/16550 driver $Revision: 1.90 $ 2 ports, IRQ sharing disabled
serial8250.0: ttyS0 at MMIO map 0x1c20000 mem 0xfbc20000 (irq = 40) is a 16550A
serial8250.0: ttyS1 at MMIO map 0x1d06000 mem 0xfbd06000 (irq = 41) is a 16550A
RAMDISK driver initialized: 1 RAM disks of 32768K size 1024 blocksize
Davinci EMAC MII Bus: probed
MAC address is 00:11:35:01:00:47
TI DaVinci EMAC Linux version updated 4.0
Linux video capture interface: v2.00
Trying to register davinci display video device.
layer=c0545800,layer->video_dev=c0545960
Trying to register davinci display video device.
layer=c0545600,layer->video_dev=c0545760
davinci_init:DaVinci V4L2 Display Driver V1.0 loaded
i2c /dev entries driver
nand_davinci nand_davinci.0: Using 4-bit hardware ECC
NAND device: Manufacturer ID: 0xec, Chip ID: 0xf1 (Samsung NAND 128MiB 3,3V 8-bit)
Creating 6 MTD partitions on "nand_davinci.0":
0x00000000-0x00300000 : "bootloader"
0x00300000-0x00500000 : "params"
0x00500000-0x00700000 : "kernel"
0x00700000-0x01f00000 : "filesystem"
0x01f00000-0x02100000 : "data1"
0x02100000-0x08000000 : "data2"
nand_davinci nand_davinci.0: hardware revision: 2.3
rtc_davinci_dm365 rtc_davinci_dm365.0: rtc intf: proc
rtc_davinci_dm365 rtc_davinci_dm365.0: rtc intf: dev (254:0)
rtc_davinci_dm365 rtc_davinci_dm365.0: rtc core: registered rtc_davinci_dm365 as rtc0
Advanced Linux Sound Architecture Driver Version 1.0.12rc1 (Thu Jun 22 13:55:50 2006 UTC).
ASoC version 0.13.1
CQ0093 Voice Codec 0.1
asoc: cq93vc <-> davinci-vcif mapping ok
ALSA device list:
  #0: On-chip voice codec (cq93vc)
IPv4 over IPv4 tunneling driver
TCP bic registered
NET: Registered protocol family 1
NET: Registered protocol family 17
rtc_davinci_dm365 rtc_davinci_dm365.0: setting the system clock to 2000-01-08 19:49:11 (947360951)
Time: timer0_1 clocksource has been installed.
Clock event device timer0_0 configured with caps set: 08
Switched to high resolution mode on CPU 0
Sending DHCP requests .., OK
IP-Config: Got DHCP answer from 0.0.0.0, my address is 192.168.0.36
IP-Config: Complete:
      device=eth0, addr=192.168.0.36, mask=255.255.255.0, gw=192.168.0.105,
     host=192.168.0.36, domain=grandeye.com, nis-domain=(none),
     bootserver=0.0.0.0, rootserver=192.168.0.105, rootpath=
Looking up port of RPC 100003/2 on 192.168.0.105
Looking up port of RPC 100005/1 on 192.168.0.105
VFS: Mounted root (nfs filesystem).
Freeing init memory: 176K
INIT: version 2.86 booting
wd_identify: No such file or directory
wd_identify: No such file or directory
0
mkdir: Cannot create directory `/dev/.test': Read-only file system
wd_identify: No such file or directory
Starting the hotplug events dispatcher: udevd.
Synthesizing the initial hotplug events...done.
Waiting for /dev to be fully populated...done.
0
CSL: Module install successful, device major num = 253
CSL: Module version 0.10.00, built on Mar  5 2012 17:16:23
I2C: Module install successful, device major num = 252
DMA: Module install successful, device major num = 251
DRV: Module install successful
DRV: Module built on Mar  5 2012 17:16:24
DRV: EDMACC.QUEPRI  = 00000777
DRV: SYSTEM.MSTPRI0 = 00440022
DRV: SYSTEM.MSTPRI1 = 00000244
DRV: ISP.BCR        = 00000002
DRV: SYSTEM.MISC    = 00000399
CMEMK module: built on Mar  1 2012 at 12:13:04
  Reference Linux version 2.6.18
  File /home/vamsiy/alta/Alta-head/dvsdk_2_10_01_18/linuxutils_2_24_03/packages/ti/sdo/linuxutils/cmem/src/module/cmemk.c
allocated heap buffer 0xc4000000 of size 0xd000000
CMEM Range Overlaps Kernel Physical - allowing overlap
CMEM phys_start (0x1000) overlaps kernel (0x80000000 -> 0x83000000)
cmemk initialized
EDMAK module: built on Mar  1 2012 at 12:13:05
  Reference Linux version 2.6.18
  File /home/vamsiy/alta/Alta-head/dvsdk_2_10_01_18/linuxutils_2_24_03/packages/ti/sdo/linuxutils/edma/src/module/edmak.c
IRQK module: built on Mar  1 2012 at 12:13:06
  Reference Linux version 2.6.18
  File /home/vamsiy/alta/Alta-head/dvsdk_2_10_01_18/linuxutils_2_24_03/packages/ti/sdo/linuxutils/irq/src/module/irqk.c
irqk initialized
+ set -e
+ test ! -r /etc/default/dropbear
+ test 0 = 0
+ test -x /usr/sbin/dropbear
+ DROPBEAR_RSAKEY_DEFAULT=/mnt/nand/dropbear_rsa_host_key
+ DROPBEAR_DSSKEY_DEFAULT=/mnt/nand/dropbear_dss_host_key
+ test -z
+ test -n
+ DROPBEAR_RSAKEY=/mnt/nand/dropbear_rsa_host_key
+ test -n
+ DROPBEAR_DSSKEY=/mnt/nand/dropbear_dss_host_key
+ test -n
+ DROPBEAR_KEYTYPES=rsa
+ echo -n Starting Dropbear SSH server:
Starting Dropbear SSH server: + gen_keys
+ test -f /mnt/nand/dropbear_rsa_host_key
+ KEY_ARGS=
+ test -f /mnt/nand/dropbear_dss_host_key
+ test -f /mnt/nand/dropbear_rsa_host_key
+ KEY_ARGS= -r /mnt/nand/dropbear_rsa_host_key
+ start-stop-daemon -S -x /usr/sbin/dropbear -- -r /mnt/nand/dropbear_rsa_host_key -p 22
+ echo dropbear.
dropbear.
+ exit 0
Starting mounting local filesystems: mount none on /dev/shm type tmpfs (rw,size=5M)
none on /var/run type tmpfs (rw)
none on /tmp type tmpfs (rw,sync)
Setting up networking ...rm: unable to remove `/etc/network/run': Read-only file system
Starting hotplug subsystem:
   pci
   pci      [success]
   usb
   usb      [success]
   isapnp
   isapnp   [success]
   ide
   ide      [success]
   input
   input    [success]
   scsi
   scsi     [success]
done.
Starting portmap daemon: portmap.
Starting internet superserver: inetd.
io scheduler deadline registered
io scheduler cfq registered
i2c_emac: exports duplicate symbol davinci_get_macaddr (owned by kernel)
insmod: cannot insert `/opt/ipnc/modules/i2c-emac.ko': Invalid module format (-1): Exec format error
 sbulla: unknown partition table
Installing knfsd (copyright (C) 1996 okir@monad.swb.de).
loop: loaded (max 8 devices)
tun: Universal TUN/TAP device driver, 1.6
tun: (C) 1999-2004 Max Krasnyansky <maxk@qualcomm.com>
CSLIP: code copyright 1989 Regents of the University of California
netconsole: not configured, aborting
PPP generic driver version 2.4.2
PPP Deflate Compression module registered
kobject_add failed for emac_davinci with -EEXIST, don't try to register things with the same name in the same directory.
insmod: cannot insert `/opt/ipnc/modules/davinci_emac.ko': File exists (-1): File exists
davinci-mmc davinci-mmc.0: Supporting 4-bit mode
davinci-mmc davinci-mmc.0: Using DMA mode
usbcore: registered new driver usbfs
usbcore: registered new driver hub
musb_hdrc: version 6.0, cppi-dma, peripheral, debug=0
musb_hdrc musb_hdrc: No DMA interrupt line
musb_hdrc: USB Peripheral mode controller at c38f2000 using DMA, IRQ 12
Loading iSCSI transport class v1.1-646.<5>iscsi: registered transport (tcp)
 sbulla: unknown partition table
mkdosfs 3.0.6 (04 Oct 2009)
unable to get drive geometry, using default 255/63
/dev/sbulla has 255 heads and 63 sectors per track,
logical sector size is 512,
using 0xf8 media descriptor, with 1024 sectors;
file system has 2 12-bit FATs and 4 sectors per cluster.
FAT size is 1 sector, and provides 247 clusters.
Root directory contains 512 slots.
Volume ID is 4c7406eb, no volume label.
Create queue id:0
queue id:0
TimeOut occure in boot_proc.
Program exit.
queue id:0
FileMngThread created
Create queue id:32769
queue id:32769
AlramThread created
Share memory init success
IPNC_DM368_2.6.0-gc2a0433d-dirty (marks@curium) (gcc version 4.
Create queue id:65538
queue id:65538
Create queue id:98307
queue id:98307
queue id:65538
queue id:32769

CPU is DM368
./av_server.out DM368 NTSC GE_4MP H264 1500000 VBR AUTO MJPEG 75 MENUOFF &
Check_device_version : DEVICE VER = 1 !!

AVSERVER UI: Initializing.
 AVSERVER API: Creating TSKs.
ApproDrvInit: 7
Create queue id:131076
queue id:131076

 CLK Hz,
 ARM   Hz =  432000000
 DDR   Hz =  340000000
 VPSS  Hz =  340000000
 IMCOP Hz =  340000000

 [OSA_FILE ] Reading file [/mnt/nand/IMG_Paramset.bin] ... ERROR
IMAGE TUNE: Paramset File is not available..... Setting DEFAULT parameter
 AVSERVER API: Creating TSKs...DONE

CONFIGURING AVSERVER FOR DM368 .....
queue id:65538
 AVSERVER API: Sending START.
 AVSERVER MAIN: Recevied CMD = 0x0400, state = 0x0000
 AVSERVER MAIN: AVSERVER_tskConnectInit() ...
 *** WARNING  FD: exceeded max input size for FD, disabling FD (current input 640x480)!!!
 AVSERVER MAIN: Create captureTsk ...
SENSOR FPS VALUE = 10
 AVSERVER MAIN: Create encodeTsk ...
DavinciDisplay DavinciDisplay.1: Before finishing with S_FMT:
layer.pix_fmt.bytesperline = 640,
 layer.pix_fmt.width = 640,
 layer.pix_fmt.height = 480,
 layer.pix_fmt.sizeimage =460800
 AVSERVER MAIN: DavinciDisplay DavinciDisplay.1: pixfmt->width = 640,
 layer->layer_info.config.line_length= 640
Create streamTsk ...

mem_layout         = 9
mem_size[0]       =0x0
mem_blk_size[0]   =0xC800
cache_size[0]     =0x0
cache_blk_size[0] =0xC800
mem_size[1]       =0xA00000
mem_blk_size[1]   =0xC800
cache_size[1]     =0xA00000
cache_blk_size[1] =0xC800
mem_size[2]       =0x0
mem_blk_size[2]   =0x19000
cache_size[2]     =0x0
cache_blk_size[2] =0x19000
mem_size[3]       =0x300000
mem_blk_size[3]   =0xC800
cache_size[3]     =0x300000
cache_blk_size[3] =0xC800
mem_size[4]       =0x80000
mem_blk_size[4]   =0x400
cache_size[4]     =0x80000
cache_blk_size[4] =0x400
totalsize          =0x1B00200
TOTAL_MEM_SIZE     =0x15A0200
queue id:131076
Stream Debug:  Init_Msg_Func
 AVSERVER MAIN: Create fdTsk ...

 FR : 0 users registered from exsiting face album
 AVSERVER MAIN: Create displayTsk ...
 AVSERVER MAIN: Create aewbTsk ...
IPNC_DM368_2.6.0-gc2a0433d-dirty (marks@curium) (gcc version 4.
IPNC AUTO_IRIS = 0
 AVSERVER MAIN: Create audioPlayTsk ...
Stream Debug: MSG_CMD_GET_MEM case
 AVSERVER MAIN: Create motionTsk ...
 AVSERVER MAIN: Create swosdTsk ...
Demo composite written.
 AVSERVER MAIN: Allocing Buffers ...
 AVSERVER MAIN: Allocating buffers ...
 AVSERVER MAIN: Stream 0: Allocating LDC buffers: 0 of size 6489728 bytes
 AVSERVER MAIN: Stream 0: Allocating VNF buffers: 0 of size 6489728 bytes
 AVSERVER MAIN: Stream 0: Allocating Encode IN buffers: 3 of size 6489728 bytes
 AVSERVER MAIN: Stream 0: Assigning Resize IN buffers: 0 of size 6489728 bytes
 AVSERVER MAIN: Stream 0: Getting Capture OUT buffers
 AVSERVER MAIN: Stream 0: Capture OUT buffers 0
 AVSERVER MAIN: Stream 0: Capture OUT buffers 1
 AVSERVER MAIN: Stream 0: Capture OUT buffers 2
 AVSERVER MAIN: Stream 1: Allocating LDC buffers: 0 of size 460928 bytes
 AVSERVER MAIN: Stream 1: Allocating VNF buffers: 0 of size 460928 bytes
 AVSERVER MAIN: Stream 1: Allocating Encode IN buffers: 3 of size 460928 bytes
 AVSERVER MAIN: Stream 1: Assigning Resize IN buffers: 0 of size 460928 bytes
 AVSERVER MAIN: Stream 1: Getting Capture OUT buffers
 AVSERVER MAIN: Stream 1: Capture OUT buffers 0
 AVSERVER MAIN: Stream 1: Capture OUT buffers 1
 AVSERVER MAIN: Stream 1: Capture OUT buffers 2
 AVSERVER MAIN: Allocating FD buffers: 3 of size 307328 bytes
 AVSERVER MAIN: Assigning Display buffers: 3 of size 307328 bytes
 AVSERVER MAIN: Stream 0: Allocating Encrypt buffers: 0 of size 2130944 bytes
 AVSERVER MAIN: Stream 0: Allocating Stream buffers: 3 of size 2130944 bytes
 AVSERVER MAIN: Stream 1: Allocating Encrypt buffers: 0 of size 308224 bytes
 AVSERVER MAIN: Stream 1: Allocating Stream buffers: 3 of size 308224 bytes
 AVSERVER MAIN: Allocating MotionDetect buffers: 3 of size 0 bytes
 AVSERVER MAIN: Allocating buffers ...DONE
 AVSERVER MAIN: Start aewbTsk ...
 AVSERVER MAIN: Start captureTsk ...
queue id:0
 AVSERVER MAIN: Start audioPlayTsk ...
 AVSERVER MAIN: Start DONE
 AVSERVER API: Sending START...DONE
Alignment trap: av_server.out (772) PC=0x000b9834 Instr=0xe592e340 Address=0x4a4a4d8a FSR 0x001
TimeOut occure in boot_proc.
Program exit.
ApproDrvInit: 3
queue id:131076
INIT: Entering runlevel: 3
Starting internet superserver: inetd
MontaVista(R) Linux(R) Professional Edition 5.0.0 (0702774)

0 Raghu Kudva over 13 years ago in reply to Vamsi

TI__Genius 11000 points

Hi,

Can you check with only 2056x2056 h264 encoder only without mjpeg and check if this issue exists

Also, you check to set hdvicpreset flag in codec params always for this mode

Regards,

Raghu

0 Vamsi over 13 years ago in reply to Raghu Kudva

Prodigy 110 points

Hi Raghu,

I Checked with 2056x2056 H264 encoder alone and Its working perfectly fine. I also checked the IH264VENC_DynamicParams::resetHDVICPeveryFrame param and its being set to 2 in the dual stream case (H264 + MJPEG) and set to 0 for single stream (i.e H264 alone).

I also tried to get the CE_DEBUG log for the above dual stream run case and am attaching that log . Could you please got hrough taht and let me know if you could find anything suspicious.

Regards,

Vamsi

5430.CEDebug.Dualstream.AlignmentTrap.txt

0 Veeranna Hanchinal over 13 years ago in reply to Vamsi

TI__Expert 8465 points

Hi Vamsi,

Can you please try below things in H264 encoder:

1] Set EnableDDR = 1

2] set resetHDVICPeveryframe = 1

3] use codec debug library and send the logs.[take the back of h264venc_ti_arm926.a and rename h264venc_ti_arm926_debug.a to h264venc_ti_arm926.a]

Thanks,

Veeranna

0 Vamsi over 13 years ago in reply to Veeranna Hanchinal

Prodigy 110 points

Hi Veeranna,

Thank a lot for the tip . In fact setting the EnableDDR to one did the trick and the alignment trap message is no longer coming up.

I can stream out 4M H264 and an MJPEG , But still got some issues.

When I increase the width of the picture beyond 2048 I can see that there is some corruption going on in the H264 encoded output.

from 2048 < Width <= 2112 , this corruption is limited to certain region.

for Width > 2112 , The whole picture is garabage , The scene is live as I can see my movements in the output but it looks like it has alignment problems somewhere.

I am attaching the snapshots I took at different resolutions while streaming the H264. For all these case the MJPEG output is absolutely fine.

Can you please give me inputs on where this might be going wrong.

We are actually interested in 2144x1944 size H264. Can you see any restrictions whcih will not allow us to have another MJPEG (VGA )stream with the above high res H264.

Thanks again for the suggestion.

Do you still want me to send the Debug.

0 Veeranna Hanchinal over 13 years ago in reply to Vamsi

TI__Expert 8465 points

Hi Vamsi,

With EnableDDR = 1, you should not get any corruption. Can you try same thing on stand-alone testapp provided with Release package?. And which version of encoder you are using? If you are not using latest one(02.30.00.04), please upgrade to this.

Currently we don't have ready setup with us(for H264 for HIMP + MJPEG), once we get this, we will give more info. If you find more information please provide us. And have you tried enableDDR=0 and resetHDVICP = 1?

Thanks,

Veeranna

0 Vamsi over 13 years ago in reply to Veeranna Hanchinal

Prodigy 110 points

Hi Veeranna,

Thanks for your reply. I am really sorry that I couldnt reply sooon as I am off for few weeks. To anwser your queries

The encoder version we are using is H264ENC.version.02.20.00.01.wizardversion.0.5.2. Also we are using the Code drop from Appro Version 2.6.0. I will try to upgrade to the latest encoder and see if the problem still exit.

I tried enableDDR=0 and resetHDVICP=1 and the alignement trap came back. I think enableDDR sorted out the problem of alignment trap, but the corruption might be due to somethign else.

I will try to work on a standalone app and let you know if I find any details. Did you try to run at your end, if its working.

Do you suspect the corruption we are observing in the last snapshot I sent is due to any

Please let me know.

Regards,

Vamsi

0 Veeranna Hanchinal over 13 years ago in reply to Vamsi

TI__Expert 8465 points

Please migrate to latest version 02.30.00.04. you can get the codec from http://software-dl.ti.com/dsps/dsps_public_sw/codecs/DM36x/index_FDS.html

Yes we had standalone testcases and we didnt see any corruption. And enableDDR = 1may not be final solution as it reduces performance. It will help us figure out the issue.

0 Vamsi over 13 years ago in reply to Veeranna Hanchinal

Prodigy 110 points

Hi Veeranna,

Thanks for the reply. Is there any chance you could send the standalone testcase code so that I can compare the initialization params with my application.

I will try to migrate to the new encoder.

Vamsi

0 Veeranna Hanchinal over 13 years ago in reply to Vamsi

TI__Expert 8465 points

Hi,

Standalone testapp present in release packagein the path \dm365_h264enc_02_x0_00_0x_production\packages\ti\sdo\codecs\h264enc\apps. Just you need to change max height and width.

Thanks,

Veeranna

0 Vamsi over 13 years ago in reply to Veeranna Hanchinal

Prodigy 110 points

Hi Veeranna and Raghu,

I found the issue related to this. I upgraded to the new codec version from TI. I still got the same problems I had with earlier versions.
After carefully comparing the params of Standalone app and my encoder params, I found that the encoding preset is being set to XDM_HIGH_SPEED for my use case (i.e two streams). I beleive this mode supports resolutions upto 2048 x 2048. When I forced it to use XDM_HIGH_QUALITY, it worked like a charm for the high reslotuions. I need not enable DDR which you mentioned might reduce the performance.

Thanks for your suggestions earlier and pointing me in the right direction.

0 Veeranna Hanchinal over 13 years ago in reply to Vamsi

TI__Expert 8465 points

Thanks for sharing the result Vamsi !!!

Yes High speed mode supports only upto 2048 x 2048. We missed this check inside codec. We will add it , codec creation will fail if you set more than 2kx2k. We have check other encoder mode(encQuality=0/1). For this mode creation will fail when we set res more than 2k.

Regards,

Veeranna

Processors

Processors forum

Dual stream(H264 +MJPEG) alignement trap if width more than 2048