Re: syslogd still crapping out

[Petr Cech <cech@atrey.karlin.mff.cuni.cz>]

On Thu, Dec 09, 1999 at 12:47:59AM -0700 , Erik Andersen wrote:
> execve("/sbin/syslogd", ["/sbin/syslogd", "-n"], [/* 6 vars */]) = 0
> [------------snip----------------]
> unlink("/dev/log")                      = -1 ENOENT (No such file or directory)
> socket(PF_UNIX, SOCK_STREAM, 0)         = -1 ENOSYS (Function not implemented)
> socket(PF_UNIX, SOCK_STREAM, 0)         = -1 ENOSYS (Function not implemented)
> socket(PF_UNIX, SOCK_STREAM, 0)         = -1 EINVAL (Invalid argument)
> write(2, "Couldn\'t obtain descriptor for s"..., 65) = 65
> _exit(1)                                = ?
> 
> So it looks like a kernel problem.  The kernel used on 
> the boot floppies does not seem to support Unix domain 
> sockets.  That is why it works in chroot, but not on 

it has that in a _module_ :(( As Xu said, else would the bzImage be bigger
that even bzImage can be (1MB). So we must ship unix.o (??) in root.bin
or compile unix sockets in.

				Petr Cech
--
Debian GNU/Linux maintainer - www.debian.{org,cz}
           cech@atrey.karlin.mff.cuni.cz