From mboxrd@z Thu Jan 1 00:00:00 1970 From: Peter Korsgaard Date: Thu, 24 Feb 2011 12:53:14 +0100 Subject: [Buildroot] Unable to login to buildroot In-Reply-To: <201102241048.48537.dargaud@lpsc.in2p3.fr> (Guillaume Dargaud's message of "Thu, 24 Feb 2011 10:48:48 +0100") References: <201102241048.48537.dargaud@lpsc.in2p3.fr> Message-ID: <87pqqhr6mt.fsf@macbook.be.48ers.dk> List-Id: MIME-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit To: buildroot@busybox.net >>>>> "Guillaume" == Guillaume Dargaud writes: Hi, Guillaume> Welcome to the Gandalf acquisition Guillaume> gandalf login: root Guillaume> Jan 1 00:00:10 login[254]: root login on 'ttyUL0' Guillaume> Welcome to the Gandalf acquisition Guillaume> gandalf login: Guillaume> And it repeats... I also cannot login via ssh (I configured Guillaume> dropbear) and I have a feeling none of the init.d scripts Guillaume> got started. I use the default inittab. The default skeleton doesn't have a password set for root, and dropbear doesn't allow root login without a password. Guillaume> The drive is mounted via nfs, and there are some missing Guillaume> stuff (when looking at them from the NFS server side): Guillaume> $ ll proc tmp Guillaume> proc: Guillaume> total 8 Guillaume> drwxr-xr-x 2 root root 4096 Feb 24 08:31 ./ Guillaume> drwxr-xr-x 16 root root 4096 Feb 24 09:19 ../ Guillaume> tmp: Guillaume> total 12 Guillaume> drwxrwxrwt 3 root root 4096 Feb 24 09:24 ./ Guillaume> drwxr-xr-x 16 root root 4096 Feb 24 09:19 ../ Guillaume> drwxr-xr-x 2 root root 4096 Feb 24 08:54 ldconfig/ Guillaume> -rw-r--r-- 1 root root 0 Feb 24 09:24 utmp Looks normal. Proc is a virtual filesystem and /tmp/ is normally a ramdisk. Guillaume> Anything I overlooked ? Not as far as I can see. You are creating your nfsroot from rootfs.tar and extracting as root, right? Guillaume> Bonus question: if you configure buildroot without an FPU, Guillaume> then what is the difference between hard floats and soft Guillaume> floats ? Why can you have both and what is preferable ? Hard float means let the compiler insert hardware FPU instructions in the code stream whenever floating point calculations are done. Soft float means add calls to software emulation functions instead. On a system without a hardware FPU you CAN run hard float binaries if you have configured the kernel to emulate those instructions, but it will be very slow. Soft float is preferred. -- Bye, Peter Korsgaard