RGB24 not working but BGR24 working on Kodicom 8800 clone.

[m1ke]

I have added:
ZM_CPU_EXTENSIONS=true
ZM_FAST_IMAGE_BLENDS=true

to /usr/local/etc/zm.conf.

I think this brought CPU usage % down a bit more.

Only major issue I have now is that I can no longer view events I just get a blank white screen. No errors seem to be in the logs.

Any idea why that could be ?

Mike.

[m1ke]

I have added:
ZM_CPU_EXTENSIONS=true
ZM_FAST_IMAGE_BLENDS=true

to /usr/local/etc/zm.conf.

I think this brought CPU usage % down a bit more.

Only major issue I have now is that I can no longer view events I just get a blank white screen. No errors seem to be in the logs.

Any idea why that could be ?

Mike.

OK I've found ZM_CPU_EXTENSIONS & ZM_FAST_IMAGE_BLENDS in the options screen. Didn't look hard enough

I still have a blank event screen when try to play back events however. I have cambozola.jar in /var/www/html/zm & OPT_CAMBOZOLA ticked & RELOAD_CAMBOZOLA set to 0.

What else could be wrong ?

Mike.

[mastertheknife]

Hi,

Glad to hear its working. From my testing, 32bit is only faster on processors with SSSE3.
About the blank events. Whats in the syslog? and where are the events stored?

mastertheknife.

[m1ke]

Hi Kfir,
Yes the events are stored OK. I can convert them to video & download & view. I can also view the stills. If I use IE to view I just get a white screen with cambozola writen in the middle. If I use Firefox I get only a bank white screen.

In /var/log/messages (syslog) I get the following when I click to view an event:

Jun 28 17:06:19 zm1 zmc_dvideo1[24704]: INF [Monitor-2: 77000 - Capturing at 25.00 fps]
Jun 28 17:06:21 zm1 zmc_dvideo0[24696]: INF [Monitor-1: 77000 - Capturing at 25.00 fps]
Jun 28 17:06:25 zm1 zms[24917]: INF [Debug Level = 0, Debug Log = <none>]
Jun 28 17:06:25 zm1 zms[24917]: INF [Authenticated user 'admin']
Jun 28 17:06:25 zm1 zms[24917]: ERR [Unable to send raw frame 1: Invalid argument]

What do you think ?

Mike

[m1ke]

Sorry miss-read your post you asked "where are the events stored" not "are the events stored"!
OK - they seem to be stored in '/var/www/html/zm/events' with a sub dir for each cam 1, 2 etc.

Mike

[m1ke]

Option DIR_EVENTS is set to 'events' which seems correct also...

Mike.

[m1ke]

Anyone got any ideas on why I get :
Jun 28 17:06:25 zm1 zms[24917]: ERR [Unable to send raw frame 1: Invalid argument]

whenever I go to view a recorded event ??

Thanks in advance,

Mike.

[mastertheknife]

Hi mike,

Sorry for the delay. It seems the sendfile() function doesn't work on your platform correctly but is broken. I will soon attach a modified zm_event.cpp that switches back to the old and slower way if sendfile() fails.

mastertheknife

[m1ke]

Hi mike,

Sorry for the delay. Anyways, it seems the sendfile() function doesn't work on your platform correctly but is broken. I will soon attach a modified zm_event.cpp that doesn't use sendfile but the old and slower way.

mastertheknife

Ahh, OK.
Wonder why that function it doesn't work on my platform - I suppose that function assumes something my Prescott CPU doesn't have or something...

Look forward to the modified zm_event.cpp

Cheers, Mike.

[mastertheknife]

Hi,

Here is zm_event.cpp that falls back to the standard way if sendfile() fails. I tested it on my machine and it appears to be working. if you also confirm it to be working, i'll update the patch.

Thank you,
mastertheknife

[m1ke]

Fantastic - it works !!

I replaced the original ../src/zm_event.cpp with your new one did the following:
service zm stop
autoconf
./configure --with-webdir=/var/www/html/zm --with-cgidir=/var/www/cgi-bin --with-ffmpeg=/usr/local/share/ffmpeg CFLAGS="-g -O3 -march=prescott -mtune=prescott" CXXFLAGS="-g -O3 -march=prescott -mtune=prescott" --enable-mmap --enable-debug=yes --enable-crashtrace=yes CPPFLAGS="-D__STDC_CONSTANT_MACROS -msse2 -I/opt/libjpeg-turbo/include"
make
make install
service zm start

Bingo all working.
Why do you think it didn't work on my platform ?

Cheers, Mike.

[mastertheknife]

"make" and "make install" in the src directory is also enough

Not sure why it didn't work, maybe you are using a bugged glibc version. There should be no noticeable performance difference between sendfile() and the other way, so it doesn't really matter

Thank you for testing. I will update the patch in my thread soon to include this.

mastertheknife