Re: [libdc] Marlin on 12.2 follow up
Capture and control API for IIDC compliant cameras
Brought to you by:
ddouxchamps,
gordp
From: E C. <e.c...@xt...> - 2012-10-01 20:24:48
|
Hello Stefan, Many thanks for your email > So this means: > [good] Camwire + libdc1394 (git) + openSUSE 12.2 > [good] grab_gray_image + libdc1394 (git) + openSUSE 12.2 > [bad] Coriander (cvs or git?) + libdc1394 (git) + openSUSE 12.2 > [good] Coriander (cvs or git?) + libdc1394 (git) + openSUSE 12.1 > > with > "good" = usable > "bad" = wrong frame rate reported, > "Format 7 is not responding properly", > "When I try the format 7 with external trigger it seems to lag" > > Is this correct? Yes ... but in case A it only works if I use Xv not SDL. > The particular git checkout which you are using can be characterized with > for example "git describe". On current mainline libdc1394 master, this > would say "V_2_2_0-9-gfa2b5b5" for 9 commits past tag V_2_2_0, commit ID > fa2b5b5[...]. V_2_2_0-9-gfa2b5b5 > The particular CVS checkout can be characterized with... well, I don't > know. > Coriander also uses libraw1394 besides libdc1394. Which libraw1394 > versions were/ are installed under openSUSE 12.2 and 12.1? 12.1 : 2.0.7-4.1.2 12.2 : 2.0.7-7.1.2 both from repository where usually I compile the all lot. > You noticed "unsolicited response (source ffc0, tlabel ...)" kernel > messages in the "bad" setup. The source ffc0 is very certainly the > camera, and the tlabel (transaction label) was a rather random sequence of > hexadecimal numbers. Do you get this also with any of the "good" setups > (or at least any of the "good" setups which you can reproduce at this time > or from which you still have logfiles left)? *On 12.1 coriander working fine** * [79008.219499] firewire_core: phy config: card 0, new root=ffc1, gap_count=5 [79008.721543] firewire_core: phy config: card 0, new root=ffc1, gap_count=5 [79013.728310] firewire_core: phy config: card 0, new root=ffc1, gap_count=5 [79023.747547] firewire_core: created device fw2: GUID 000a4701010470e0, S400, 3 config ROM retries starting coriander : nothing more in dmesg > As David already hinted, the most typical circumstance under which these > messages occur is when the camera sends the response packet very late > after it received the respective request packet from the Linux node --- > particularly, after the Split_Timeout on the Linux node elapsed ( = a > timeout which starts when the request was sent). > > This time-out differs between kernel driver versions: > > ieee1394) > default 100 ms, can be configured by remote or local software to > up to 8 s > > firewire-core 2.6.35 and older) > default 100 ms, fixed > > firewire-core 2.6.36...2.6.38) > default 100 ms, can be configured by remote or local software to > up to 8 s, local software needs permission to access the local node's > device file > > firewire-core 2.6.39 and newer) > default 2000 ms, can be configured by remote or local software to > up to 8 s, local software needs permission to access the local node's > device file That sounds like a possibility but I am on 3.1.10-1.16-desktop and it works fine Unless the value is right for 3.1 .. how can I check that / modify that ? > Actually I am not so sure about the configurable upper limit; this is what > the IEEE 1394 spec allows but OHCI 1394 might differ. I would have to > check. The value of Split_Timeout after power reset is 100 ms according > to IEEE 1394, but some video and audio devices have been observed to need > more than that. Hence the current Linux default of 2000 ms. ho ..... I see > According to distrowatch.com, openSUSE is based on the following kernels: > 11.3) 2.6.34 > 11.4) 2.6.37.1 > 12.1) 3.1 > 12.2) 3.4.6 Yes I do confirm > The ieee1394 driver stack was removed from mainline Linux in 2.6.37, IOW > Linux mainline kernel 2.6.36 was the last one to contain them. Linux > distributions switched from ieee1394 etc. to firewire-core etc. at > different points in time. Well when installing 12.1 it clearly states : loading ieee1394 modules loading; e.g. I might have been lucky Is there a way I can check /modify this value of firewire core ? I also need to test a Stingray on both machines and see what happens. Will do that today; hopefully we can get more info from the experience. Many thanks again for the insight. cheers E |