Date: Wed, 18 Oct 2006 22:31:18 +0100 From: Bruce M Simpson <bms@incunabulum.net> To: Mikhail Teterin <mi+mx@aldan.algebra.com> Cc: firewire@freebsd.org Subject: Re: libraw1394 Message-ID: <45369D26.6060807@incunabulum.net> In-Reply-To: <200610181636.12516.mi%2Bmx@aldan.algebra.com> References: <Pine.GSO.4.62.0610181257200.8352@tioga.arc.nasa.gov> <200610181636.12516.mi%2Bmx@aldan.algebra.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi all, Mikhail Teterin wrote: > =D3=C5=D2=C5=C4=C1 18 =D6=CF=D7=D4=C5=CE=D8 2006 16:08, Buzz Slye =CE=C1= =D0=C9=D3=C1=D7: > =20 >> This library was written to conform to an older version of libdc1394. >> The latest version of libdc1394, I have discovered, has many routine >> name changes so there will require some reallignment and cleanup. >> =20 > > Yeah, I figured that out... It seems, however, you understand how Linux= '=20 > ioctls should be translated to FreeBSD's ioctls and other system calls = (read,=20 > write?). Would you be able to help with the reallignments and cleanups?= > =20 From what I understand, Buzz's port does not deal with isochronous=20 streams. There is a rough mapping of Linux raw1394 APIs to=20 firewire(4)'s, based on code inspection, although they are not=20 documented in the FreeBSD man pages. > Right, they don't exist. But libraw1394's own little client utilities c= all=20 > them, and so, likely, will some other clients of the library. This mean= s, our=20 > port will need to provide these functions, even if they are going to be= =20 > simple wrappers around system calls... > =20 It looks like a kqueue event pump will be needed to emulate libraw1394's = callback semantics. > =20 > Thanks, I'm thinking about it too. My Makefile will be using bsd.lib.mk= , of=20 > course :-) > =20 That would reduce the chances of having the patches taken upstream given = the makefile churn. > =20 >> This code does not support multiple cameras. >> =20 > > If the Linux code does, and the FreeBSD kernel code does, the port will= have=20 > to as well... > =20 It looks like a mapping of /dev/raw1394 -> /dev/fwN.N should work; vlc=20 in particular allows you to specify the device used, though I'm not sure = if this is passed down to libraw1394 -- haven't looked that far yet. >> Asychronous tramsmissions only support quadlet (4 byte) read/write for= >> reading and writing camera registers. >> >> Our applications mostly use cameras to take single images at timed >> intervals. If you will be using stream transfer, you may want to >> change the buffering setup. >> =20 spook is going to require isochronous streaming. > > Buzz, I have NO IDEA how to do this :-( I can put the port together, bu= t the=20 > firewire details are not my thing. I just want it to work, so I can: > > a) transfer my own videos from Sony's HDV camera; > =20 This is going to need libavc1394. libraw1394 is certainly a prerequisite.= Regards, BMS
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?45369D26.6060807>