> "P. Christeas" <[EMAIL PROTECTED]> writes: > > ? I have : > > -rwsr-s--- 1 root cdwriter 318348 Aug 22 21:46 > > /usr/bin/cdrecord* > > -rwsr-s--- 1 root cdwriter 183324 Aug 2 > > 2001 /usr/bin/cdrecord.old* > > > > where cdrecord.old (v. 1.1.x) manages to set the priority, while cdrecord > > doesn't. Is there a bug in the code (perhaps about eid elevation?). > > I'm running on Linus's 2.4.22 kernel. I could test with 2.6.0 as well > > (although I don't have a recorder on the other box). > > you need to be in the cdwriter group
I am. You see, even though the crecord.old binary has exactly the same permissions, it is allowed to set the scheduling, while the v2.0 'cdrecord' isn't. What's more interesting is that I can execute the 'cdrecord' as root. Then, it *does* set the scheduling. However the buffer *is not 100% full* , as it should have been. The 'nice' value (eid=0=root) is still 0, but no warning. I haven't gone into the source code yet (I believe I must fetch your tree just to keep in sync) but I have the feeling that cdrecord 2.0 does not perform the correct sequence of actions needed to set its priority. Did v1.1 have the same threads v2.0 has?
