tags 270745 moreinfo thanks At Wed, 08 Sep 2004 22:14:00 -0300, Cesar Eduardo Barros wrote: > However, libc6-i686 did not re-exec init, which is using libc6-i686: > $ memstat -w | fgrep libc- > [...] > 1244k: /lib/tls/i686/cmov/libc-2.3.2.so 1 [...] > [...] > > I had to do a "telinit u" by hand. > > The logs show two attempts to re-exec init; one was probably from > libc6.postinst, and the other was the one I did by hand. Before I > re-execed it by hand, it was using the old copy of libc6 (obvious on > memstat output since it's the one that has almost every pid in it). > > I think libc6-i686 should repeat the daemon-restarting dance of > libc6.postinst, to avoid problems with other daemons (this time it > wasn't needed, since the upgrade was from -13 libc6 decided it only had > to restart init)
init executes with the newer /lib/libc-2.3.2.so after running libc6.postinst. Could you tell me why we need to reexec init? IIRC, init does not get any bonus using /lib/tls/i686/cmov/libc-2.3.2.so. If you don't provide us any valid evidences, I'll close it. Regards, -- gotom -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
