I have an AMD Opteron/Athlon64 in a VIA K8T800Pro mobo. Installing "stock" gentoo more or less straight from the Handbook. (Taking mostly conservative / suggested options even.) If I choose the i686 stage3, openrc stops before init, and dumps me into a shell, as described here: http://forums.gentoo.org/viewtopic-p-6879892.html Note that the last comment is to use i486 stage3. I also found that using the i486 stage3 worked, as daft as that sounds. I was able to scp the .config to save it, so it was the *same* .config file in both cases: http://l-i-e.com/kernel.config NOTE: My permissions were okay on the various binaries as described here: http://dev.gentoo.org/~a3li/openrc.txt I did the devtmpfs thing described here as well: http://forums.gentoo.org/viewtopic-t-881317-start-0.html but that made no difference. Only the i486 stage3 worked. Really. I'm not sure you'll be able to do anything with this report without the hardware in hand, but I could maybe arrange root access at a pre-determined window if needed... I'll need to image the drive somehow so I can restore it though...
Oh yeah: init 3 in the shell for the i686 indicated that /bin/initctl was file not found. ls /bin/initctl confirmed that fact. Dunno if that's cause or symptom, but there it is.
Please give us the complete stage names that you used for running these tests.
(In reply to comment #2) > Please give us the complete stage names that you used for running these tests. stage3-i686-20111213.tar.bz2 stage3-i486-20111213.tar.bz2 portage-latest.tar.bz2 I believe I used the Argonne and/or Hoobly Classifieds mirrors, as I'm in Chicago, and those are nearest. It was usually ANL, but ANL was unavailable/over-loaded at some point in the last few days, so I saw hoobly scroll by at least once. I confess I did not run the MD5 checker on any of them.
Technically, the mobo is "VIA K8T NEO V" That's what is printed on the board itself. lspci was only identifying some portion of the chipset. I don't think this matters, but just in case...
/bin/initctl is part of upstart, not anything in gentoo. It's not in either of the tarballs, and it's not supposed to be. Your initial link to the forums is for a SC1100, that is NOT i686 class hardware, but i486 class instead. Please provide a capture of the kernel boot output from i686. Also, on that hardware, you should be using an amd64 stage ideally (but i686 should work). Please specify what CPU you actually have. Ideally with the output from /proc/cpuinfo. Also, after you boot another env, we need your emerge --info output from that machine. Reopen with the info requested.
(In reply to comment #5) > /bin/initctl is part of upstart, not anything in gentoo. It's not in either of > the tarballs, and it's not supposed to be. I don't know where it should come from. I just know OpenRC won't get any farther without it :-) > Your initial link to the forums is for a SC1100, that is NOT i686 class > hardware, but i486 class instead. So the symptoms make sense there... > Please provide a capture of the kernel boot output from i686. I don't know if I can manage to get it off the box, but can try later. Will attach dmesg from i386 working for now. > Also, on that hardware, you should be using an amd64 stage ideally (but i686 > should work). I got burned badly by 64-bit builds long ago, and never tried again. Probably should... > Please specify what CPU you actually have. Ideally with the output from > /proc/cpuinfo. Will attach. > Also, after you boot another env, we need your emerge --info output from that > machine. Will attach.
Created attachment 298969 [details] cat /proc/cpuinfo
Created attachment 298971 [details] emerge --info This is from the working i386 build, after a lot of post-install emerge and tweaks...
Created attachment 298973 [details] dmesg This is also from the i386 working build, which wasn't what you asked for, but it's what I have handy now...
(In reply to comment #6) > (In reply to comment #5) > > /bin/initctl is part of upstart, not anything in gentoo. It's not in either of > > the tarballs, and it's not supposed to be. > > I don't know where it should come from. > I just know OpenRC won't get any farther without it :-) NOTHING in OpenRC needs /bin/initctl. Why do you keep claiming it does? Did you have upstart installed at some point? Or are you misread the message as /sbin/init or /dev/initctl? > > Your initial link to the forums is for a SC1100, that is NOT i686 class > > hardware, but i486 class instead. > So the symptoms make sense there... No, they don't, esp know that your Sempron cpuinfo is attached. Sempron is 100% an i686 system. Sempron is NOT an Opteron or Athlon64 like you claimed in your initial description. > > Please provide a capture of the kernel boot output from i686. > > I don't know if I can manage to get it off the box, but can try later. > Will attach dmesg from i386 working for now. That should work w/ i686. If you can boot an i686 media, and chroot to a i686 stage, then the problem is NOT that your system cannot handle i686, but rather something else. > > Also, on that hardware, you should be using an amd64 stage ideally (but i686 > > should work). > > I got burned badly by 64-bit builds long ago, and never tried again. > Probably should... My statement of amd64 stage was based on your assertion it was Opteron/Athlon64, which you were not correct about. > > Also, after you boot another env, we need your emerge --info output from that > > machine. > > Will attach. Your emerge --info says i686, so what are you on?
(In reply to comment #10) > (In reply to comment #6) > > (In reply to comment #5) > > > /bin/initctl is part of upstart, not anything in gentoo. It's not in either of > > > the tarballs, and it's not supposed to be. > > > > I don't know where it should come from. > > I just know OpenRC won't get any farther without it :-) > NOTHING in OpenRC needs /bin/initctl. Why do you keep claiming it does? > Did you have upstart installed at some point? Or are you misread the message as > /sbin/init or /dev/initctl? When I was dumped into a shell before OpenRC started init scripts, I was told in IRC#gentoo to try to run init 3 by hand. It failed, and printed a file not found for /bin/initctl, which was confirmed with ls > > > Your initial link to the forums is for a SC1100, that is NOT i686 class > > > hardware, but i486 class instead. > > So the symptoms make sense there... > No, they don't, esp know that your Sempron cpuinfo is attached. > Sempron is 100% an i686 system. > > Sempron is NOT an Opteron or Athlon64 like you claimed in your initial > description. I'm afraid various CPU detection tools are telling me different things. lspci versus dmesg versus ... My comment here was that the SC1100 symptoms / solution matched, not mine. > > > Please provide a capture of the kernel boot output from i686. > > > > I don't know if I can manage to get it off the box, but can try later. > > Will attach dmesg from i386 working for now. > That should work w/ i686. If you can boot an i686 media, and chroot to a i686 > stage, then the problem is NOT that your system cannot handle i686, but rather > something else. I can boot from SystemRescueCD or gentoo minimum install CD. I can go through the process of building the kernel with the current i686 gentoo tarballs and fail to get to the init scripts. I can use the current i486 tarballs tarballs and have a working system. IRC #gentoo folk suggested that it was an OpenRC issue. I have NO IDEA what Upstart is. > > > Also, on that hardware, you should be using an amd64 stage ideally (but i686 > > > should work). > > > > I got burned badly by 64-bit builds long ago, and never tried again. > > Probably should... > My statement of amd64 stage was based on your assertion it was > Opteron/Athlon64, which you were not correct about. If the software supposed to be telling me what the CPU is lies to me, I can't really fix that... I may have the box the CPU came in from over 5 years ago, but I doubt it... Unlike the mobo, I don't think the CPU has its version printed on it... > > > Also, after you boot another env, we need your emerge --info output from that > > > machine. > > > > Will attach. > Your emerge --info says i686, so what are you on? I used the i486 tarballs to compile the kernel. If emerge --info is claiming otherwise, I cannot fix that.
I'm closing this bug as it has been sitting for over 6 months, we never got any confirmation about it and we can't do anything with the current information.
