[Mjpeg-developer] bug in lavrec/new driver ?

[Scott M. <sm...@br...>]

Hey all,
   just built 2.4.20-pre7 yesterday with the v4l2 patch from
http://bytesex.org/patches/2.4/11-v4l2-api-2.4.20-pre2.diff.gz
   then built the new driver from cvs for my DC10.

   It doesn't look like lavrec is getting the quality set correctly.

   for instance:

for QUAL in 50 60 70 80 90 100
do
   LAV_VIDEO_DEV=/dev/video1 lavrec -v 1 -i n -q ${QUAL} -f a -t 10 -s
-l 70 -R l -d 12 -U q${QUAL}.avi >log.${QUAL} 2>&1
done

   that outputs the following AVIs:
-rw-r--r--    1 smoser   smoser    6536928 Sep 25 07:17 q100.avi
-rw-r--r--    1 smoser   smoser    6505268 Sep 25 07:16 q50.avi
-rw-r--r--    1 smoser   smoser    6550096 Sep 25 07:16 q60.avi
-rw-r--r--    1 smoser   smoser    6534284 Sep 25 07:16 q70.avi
-rw-r--r--    1 smoser   smoser    6601348 Sep 25 07:12 q75.avi
-rw-r--r--    1 smoser   smoser    6569552 Sep 25 07:16 q80.avi
-rw-r--r--    1 smoser   smoser    6550988 Sep 25 07:17 q90.avi

   each of the log.* files say that their quality is set correctly (
INFO: [lavrec] Quality:            90 ) and there were no error
messages, but it doesn't seem like everything should be the same size.

   dmesg shows:

Sep 25 07:21:50 homer kernel: DC10plus[0]: zoran_open (lavrec -
pid=[2382]), users(-)=0
Sep 25 07:21:50 homer kernel: DC10plus[0]: enable_jpg IDLE
Sep 25 07:21:50 homer kernel: DC10plus[0]: mmap(MJPEG) of
0x40527000-0x41527000 (size=16777216)
Sep 25 07:21:50 homer kernel: DC10plus[0]: enable_jpg MOTION_COMPRESS
Sep 25 07:21:50 homer kernel: DC10plus[0]: first frame ready:
state=0x48000000 odd_even=1 field_per_buff=1 delay=1
Sep 25 07:21:50 homer kernel: DC10plus[0]: stat_com=1000 queue_state=0/0/2/64
Sep 25 07:21:50 homer kernel: DC10plus[0]: seq=0 stat_com: 030074f1 04274000 1d058000 1d058000
Sep 25 07:21:50 homer kernel: DC10plus[0]: seq=0 stat_com: 030074f1 04007c89 1d058000 1d058000
Sep 25 07:21:50 homer kernel: DC10plus[0]: seq=4 stat_com: 114e2000 114e2000 05007b39 1d058000
Sep 25 07:21:50 homer kernel: DC10plus[0]: seq=4 stat_com: 114e2000 114e2000 05007b39 06007aa9
Sep 25 07:21:50 homer kernel: DC10plus[0]: seq=6 stat_com: 07007be1 114e2000 051b9000 051b9000
Sep 25 07:21:50 homer kernel: DC10plus[0]: seq=6 stat_com: 07007be1 08007c49 051b9000 051b9000
Sep 25 07:22:00 homer kernel: DC10plus[0]: enable_jpg IDLE
Sep 25 07:22:00 homer kernel: DC10plus[0]: munmap(MJPEG)
Sep 25 07:22:00 homer kernel: DC10plus[0]: zoran_close, lavrec
pid=[2382]
Sep 25 07:22:00 homer kernel: DC10plus[0]: enable_jpg IDLE
Sep 25 07:22:00 homer kernel: DC10plus[0]: interrupts received:
GIRQ1:603 JPEGRepIRQ:603 JPEG delays: max=1 min=1 queue_state=301/301/303/365
Sep 25 07:22:00 homer kernel: DC10plus[0]: zoran_close done


   any ideas on this?  Can anyone else verify it?  The input is fine,
each of the avis look OK, but obviously aren't varying quality levels.

   Did I do something wrong, or should 2.4.19 patched to 2.4.20-pre7 and
then patched with that patch from bytesex.org and the current CVS driver
checked out with:

cvs -d:pserver:anonymous:@cvs.mjpeg.sourceforge.net:/cvsroot/mjpeg
checkout -r ZORAN_VIDEODEV_2 -d driver-zoran-videodev2 driver-zoran

   ideas ?

   scott