Problems with rtc, libstdc++, reboot and dhcpcd [solved]

[sfragis]

I finished to install a minimal system with the 2004.3 profile when the 2005.0 was released, hence I decided to upgrade since very few packages were installed.
No errors came out during the process (I used the easy scripted way, http://amd64.gentoo.org/Makefile-profile_update-2005.0) and everything went right, until I rebooted.
Before reporting any bug I'd like to wait for some opinions of yours. Following are the problems:

1. On boot /etc/init.d/modules complains that cp cannot load the libstc++.so.6 library, even if this libray exists, it's been correctly compiled, it's in the ld cache (I tried ldconfig -p) and lots of programs make use of it (but /etc/init.d/modules only complains).
   
2. I builded the RTC as a kernel module and I load it on boot by /etc/modules.autoload.d/kernel-2.6. The module is correctly loaded (regardless of the problem with /etc/init.d/modules at point 1). Even that, /etc/init.d/clock (which uses hwclock) fails saying that /dev/rtc does not exist. As I'm using udev, I supposed there must be a sort of delay between the instant rtc is loaded and the character device is created. I put a 'sleep 5' into /etc/init.d/clock and now it seems to work (a 3 second delay is not always enough). Ok, I know rtc could be compiled in into the kernel, but I'm wondering why the rtc device creation is deferred that way: after logging in as root 'modproble -r rtc; modprobe rtc' will remove and then create /dev/rtc almost immediately.
   
3. On reboot (and sometimes on boot too) dhcpcd fails bringing up eth0 and asserts another dhcpcd process has been loaded before. I can't really understand why it fails; when /lib/rcscripts/net.modules.d/dhcpcd is called by /etc/init.d/net.eth0, the function clean_pidfile checks for a pid file (which may exist) and for the related process (which must not exist!): what happens is that on boot a process with the pid written in /var/run/dhcpcd-eth0.pid really exists, as if dhcpcd has been called twice.
   
4. Often, the boot process doesn't get to the end and after entering the runlevel 3 the system reboots: sometimes it happens after loading the console fonts, sometimes just after bringing up the loopback interface and sometimes while loading the alsa modules... there's not a specific point of crossover.
   

Where do these problems come from? Has anybody ever seen anything out of these?
Thanks in advance

[sfragis]

Almost everything works now.
problem 1: that's the bug https://bugs.gentoo.org/show_bug.cgi?id=91059, fixed
problem 2: solved by doing nothing but many emerge -auv world
problem 3: that's the (stil unsolved) bug https://bugs.gentoo.org/show_bug.cgi?id=78495
problem 4: no more unaspected reboots, system now completes boot process correctly, maybe some bugs have been fixed (but I've not found anything about them)