Francesco Romani wrote: > On Wed, 2008-10-22 at 15:44 -0500, Carl Karsten wrote: >> Francesco Romani wrote: >>> On Wed, 2008-10-22 at 11:37 -0500, Carl Karsten wrote: >>>> Getting back to my vga2usb project that I left off 6 months ago. I am >>>> starting >>>> with what is simple, and expecting to be told "build from current source" >>>> and >>>> "those options don't make sense" >>> It may be the case (and it can be the case) but one way or another a >>> segfault or anything else as serious as that is not acceptable, >>> especially in beta or later releases. >>> >>> As always, please be patient if my feedback comes late and don't be >>> afraid to remind to the list/bugtracker (berlios!)/me the still unsolved >>> issues. >> http://developer.berlios.de/bugs/?group_id=10094 > > > OK, got the backtrace. > > The battle plan follows. > > - first and foremost, let's use CVS HEAD as a testbed.
Is there a good way to tell what version I is being run? pretty sure the backtrace was from head - here is my process (it leans on how Christian set things up, which helps make sure his setup stays sane.) # based on https://help.ubuntu.com/community/UpdatingADeb PACKAGE=transcode OLDVER=transcode-1.1.0~beta1 NEWVER=transcode sudo apt-get --assume-yes --force-yes build-dep $PACKAGE apt-get source $PACKAGE cvs -z3 -d :pserver:[EMAIL PROTECTED]:/cvstc login cvs -z3 -d :pserver:[EMAIL PROTECTED]:/cvstc co -r HEAD transcode cd $NEWVER/ autoreconf -i cd .. cp -a $OLDVER/debian $NEWVER/ cd $NEWVER/ # bump the version number debchange --nmu $0 DEB_BUILD_OPTIONS="nostrip" dpkg-buildpackage -rfakeroot -uc -b cd .. gdebi $PACKAGE*.deb I just cvs up and see a change to import/v4l/import_v4l.c - Ill build and make sure something changed. > I'll be happier > testing 1.1.0-RCs, but after a bit of thinking CVS HEAD is better in the > end, we have more liberty to modify the code; only the fixes will be > backported. > [- as a sidenote, looks like I'll be able a reproduce a segfault > reported by you on my box using 1.1.0 and vivi. I'll work on that ASAP]. > - I'm updating the transcode's import_v4l module, mainly but not only > for this task. I'm afraid v4l v1 will stay around for a while, and I've > some interest in offering some support. > - I've some tests and fixes on my queue, but I'm having hard time in > finding some time for applying. > - A fair amount of try-report-fix-retry loops will likely be required. > I hope this will not be a problem too big. I expect a few loops too. good news - vga2usb driver dev says: > Hi Carl, > I can enable debug output for all V4L ioctls in the driver if you want. > However, I don't think (I need to recheck) that it's easy to enable > "hardwareless" mode of operation... too much in the driver rely on > correct initialization from kernel's probe() function. > > We are definitely planning to implement V4L2. Nowadays kernel have very > nice way to enable V4L2 functionality. The only show stopper for us was > the need to support 2.4.x kernels in the single 2.4.x & 2.6.x code base. > It seems nobody is asking for 2.4.x drivers anymore, so we can schedule > transition to V4L2 soon. > > Misha. Carl K
