> > Had some playing on the 'old', now obsolete, saa7124-0.2.4 from 20031601 > on2.4.20 md9717. > [snip] hi, in the last message was a typo. It must be saa7134-0.2.4 above. ----------- Build 2.4.20 with the now latest patches from 20021701. config is: md9717, 2.4.20, v4l2 patch-2.4.20-kraxel20031701, saa7134-0.2.4-20031701, RH-8.0, gcc-3.2.1-20021207, make 3.79.2a1. With the latest release xawtv-3.83 the problem with the overlay mode is solved. (see msg03827 from Kyosti Malkki.) xawtv seems to lose sometimes memory handling on (wrong?) key strokes. Also in rare cases, when tuning on a channel fails to give a picture. It then shows in overlay and grabdisplay mode only a black screen. Tried to unload and load the modules again, but it is back only after reboot. Think this is related to that what happens, when only the audio is'nt found. Then only grabdisplay mode is gone and it no longer offers to select audio mode on the md9717. But there breaks something, because when starting other capture apps thereafter, they can bring all down here, what is the privilege of tvtime usually. (Thanks tvtime-devel and sorry to bring this to you.) Are the bugs in v4l1 compat found by gcc 3.x and do builds work with the recommended compilers? (see msg.18398 from Gerd Knorr.) Have warnings on i2c.main, agpgart and some asm and that makes me feel sick, when it is like it is now. Most strange seen here, once and only once, after downgrading the boards settings again from less advanced to more conservative, was that tvtime-cvs quits on startup with the message 'unresolved symbol in libstdc++'. Two, three minutes later the panic came and it had not even shown a window and this version brought it absolute regulary on cleanup after quitting. Really did wait and wait until finally upgraded to RH-8.0, but when saa7134-0.2 came from 2.5 for the md9717 on 2.4.20, could'nt keep out and want to see this on gnu-linux. Have RH-7.1 on a PII and also Mandrake-8.1. May it help to build this there or set up some alias compiler, or is it something else? All pci-slots on the KT133A with all it's workarounds are full here ... hermann (Hermann Pitton)