Re: [uml-devel] Can't compile under vanilla 2.6.11

[Rob Landley <rob@landley.net>]

On Thursday 17 March 2005 10:18 pm, Ian McDonald wrote:
> Hi there,
>
> I've been watching the discussion around 2.6.11 and reading the
> archives and I can't find the information with any clarity that I
> need. My problem is that I can't get UML to compile. I can on 2.4.27
> (with the standard patches) and with 2.6.10 also.

All I have to do, on x86 under either Knoppix or Red Hat, is (this is off the 
top of my head, so there might be a typo or two):

tar xvjf linux-2.6.11.tar.bz2
cd linux-2.6.11
# Get a usable .config file, you can use "make ARCH=um menuconfig" for this
# if you know what to select.  This adds a more or less minimal set of options 
# to allnoconfig to provide a usable test system.  Later, you'll want to add
# thinks like loopback support, UBD, ext2, maybe swap support...
make ARCH=um allnoconfig
cat >> .config << EOF
CONFIG_MODE_TT=y
CONFIG_LD_SCRIPT_STATIC=y
CONFIG_BINFMT_ELF=y
CONFIG_HOSTFS=y
CONFIG_STDERR_CONSOLE=y
CONFIG_LBD=y
CONFIG_PROC_FS=y
EOF
yes "" | make ARCH=um oldconfig
# Now build it.
make ARCH=um
# Run it.  Later you might want to add "rootflags" (defaults to rootflags=/)
# and "mem" (defaults to mem=32M).  "quiet" is nice too...
./linux rootfstype=hostfs rw init=/bin/sh
# Show that the new shell prompt is under UML, not the parent system.
mount -t proc /proc /proc
cat /proc/cpuinfo

The polite way to exit is either "shutdown -n" or "halt -f" (both in sbin, 
which isn't int he path), but when pid 1 is /bin/sh then just "exit" will 
work as well if you don't mind a UML panic message about "tried to kill 
init"...

Speaking of path: the default UML path ends with ".", and if you try to 
compile anything this triggers some "holier than thou" code in gcc that 
ignores the entire path without even a warning message because it thinks it 
knows better than you, and thus doesn't find the linker.  Just set your PATH 
to "/sbin:/usr/sbin:/bin:/usr/bin" and it should work fine.  (Why this isn't 
the default, I have no idea...)

> My environment is Debian Woody with 2.4.27 kernel. My gcc version is
> 2.95.4. I am using default options when I do 'make menu_config
> ARCH=um'

There's no underscore in menuconfig.

> Here is the error that I am experiencing once I do 'make linux ARCH=um':
> arch/um/kernel/sys_call_table.c:277: duplicate array index in initializer
> arch/um/kernel/sys_call_table.c:277: (near initialization for
> `sys_call_table') arch/um/kernel/sys_call_table.c:277: duplicate array
> index in initializer arch/um/kernel/sys_call_table.c:277: (near
> initialization for `sys_call_table')

I remember a problem like that in one of the 2.6.11-pre releases, but it was 
fixed in the final 2.6.11.  What compiler version are you using?

> The code at line 277 is:
>        ARCH_SYSCALLS
>        [ LAST_SYSCALL + 1 ... NR_syscalls ] =
>                (syscall_handler_t *) sys_ni_syscall
>
> I don't quite understand this as it surpasses my C knowledge but if I
> do a dirty hack and change this to:

Yeah, I did too.  It was fixed before the 2.6.11 release.  Are you sure you're 
building 2.6.11 sources?

> Can anybody point me in the right direction?

You're sure you're using linux-2.6.11.tar.bz2 on x86?

> Regards,
>
> Ian

Rob


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel