As per bug 2086, the settings currently in /etc/security/limits.conf crate a limit for members of the audio group, where that limit does not exist for non-members. In the case of bug 2086, this caused a problem for the game cultivation, only for ids that are members of the audio group. The patch from that bug report fixes the problem. https://bugs.mageia.org/attachment.cgi?id=669&action=diff
CC: (none) => stormi
@ colin cc'ing you because you are the last one in the changelog for pam-1.1.3-3.mga2.i586.rpm in cauldron, and because I think colin = sound ;) "nobody" is our pam maintainer.
CC: (none) => mageia, marja11
Unfortunately the reasons for this patch are lost in the mists of time... http://svn.mandriva.com/cgi-bin/viewvc.cgi/packages/cooker/pam/current/SOURCES/Linux-PAM-0.99.3.0-enable_rt.patch?view=log The only mention is: * Mon May 16 2005 Thierry Vignaud <tvignaud@mandrakesoft.com> 0.77-28mdk - patch 516: add support for RT/nice rlimit settings (kernel-2.6.12+) - patch 517: enable new RT privileges for audio group in limits.conf - patch 518: fix build with gcc-4.0 I would guess that the reasons for doing this are no longer really valid and thus this patch should simply be dropped. Any app that wants RT privs should use rtkit to request them. As it's generally not the "done thing" to add users to groups, the impact of this problem should be quite low (albeit a tricky one to actually work out the cause). The only cases where this would generally be done (adding users to the audio group) is in the case of headless and "userless" systems such as running mythtv or mpd (such that their respective users could access audio devices). But even still, as PA is the recommended audio system and as it uses rtkit and as we do not add users to the audio groups by default, I propose we just drop this patch. Thierry, WDYT?
CC: (none) => thierry.vignaud
Why not. Anyways those days sharing a sound device between users is painful even with the audio group because PA of the first user "locked" the card
(In reply to comment #3) > Why not. @ Dave Colin and Thierry agree on dropping this patch. If you have a reply to "why not", please comment, else this bug will be closed as wontfix
(In reply to comment #4) > (In reply to comment #3) > > Why not. > > @ Dave > Colin and Thierry agree on dropping this patch. If you have a reply to "why > not", please comment, else this bug will be closed as wontfix I don't understand your comment, Marja. It's about dropping an old patch to the package so that it works better, so there's nothing to wontfix here.
OK, so I've dropped this in Cauldron. I'm not 100% convinced we should bother changing this in mga1 tho'. Is it really worthwhile to push through this update? I'd rather we kept it on Cauldron and it when through the usual cycle of QA as some of the problems resulting from this change could be quite subtle. WDYT?
(In reply to comment #5) > > I don't understand your comment, Marja. > > It's about dropping an old patch to the package so that it works better, so > there's nothing to wontfix here. LOL thanks for telling me, I completely misunderstood :)
(In reply to comment #6) > OK, so I've dropped this in Cauldron. > > I'm not 100% convinced we should bother changing this in mga1 tho'. Is it > really worthwhile to push through this update? I'd rather we kept it on > Cauldron and it when through the usual cycle of QA as some of the problems > resulting from this change could be quite subtle. WDYT? Agreed. Adding my id to all of the groups is a habit I developed before I started using Mandrake, when that was a suitable way of getting access to devices, and log files, etc. It isn't common for people to do that now, so this bug only affects a small number of people, and there is an easy workaround.
Closing this bug report as it's fixed on cauldron, and will not be applied to Mageia 1.
Status: NEW => RESOLVEDResolution: (none) => FIXED
Thanks Dave, but I've actually just seen another update which we need to push out to mga1 for PAM, so while there could be subtle ramifications of this, I think the risk is sufficiently low that we may as well bundle this up too. I'll add a note to bug #3192 about this change too. Will leave this as closed.