On 18. jan 2003 20:39, Götz Waschk wrote: > Am Samstag, 18. Januar 2003, 00:00:14 Uhr MET, schrieb Tibor Pittich: > > i have a small (but i think that useful) draft. > > what about putting suid bit to mplayer binary? if mplayer doesn't have > > it, then after starting write at stdout about failed this ioctl > > rtc_irqp_set 1024. suid bit should fix this problem.
> I think the risk is too high that this could cause security problems. > If someone wants to setuid root the mplayer binary, he can do this himself. i agree with you, this can be a possible security problem, therefore i don't recommend this solution. > > but there is another way to "fix" this problem. set via system's sysctl > > config file (/etc/sysctl.conf) during starting system this value: > > dev.rtc.max-user-freq = 1024 > > and mplayer can set this rtc frequency without suid bit. > > > i vote for second solution, and add into post script check, if this > > value is presented in sysctl.conf, if not, then add it and run sysctl -p > > to refresh this new setting. > I don't think userland packages should mess around with such system > settings. Do you really need the rtc stuff? MPlayer is working fine > for me without it. hm, of course, mplayer works without this properly, but not perfectly :) look at documentation: http://www.mplayerhq.hu/DOCS/#rtc we can see, that rtc is described as new and better (precisious) method for timing, and older usleep() as old. increasing maximum frequency of rtc we can named as "messing", but i'd rather called it as "tuning for better functionality" :) -- virtual.phuture.sk, Linux Version 2.4.20-2mdk, Mandrake Cooker One 225MHz Cyrix MII Processor, 112M RAM, 448.92 Bogomips Uptime 4 days 6 hours 24 minutes, Load Average 0.04, 0.02, 0.00
