Re: [PATCH] polystrap-binfmt for preparing the build host

Hi,

On Tue, Jun 28, 2011 at 11:15:34PM +0200, Yann Dirson wrote:
> As can be seen in your script, the use of binfmt support is not
> without problems.
> 
> +        case $ARCH in
> +                alpha|amd64|arm|armeb|i386|m68k|mips|mipsel\
> +		|powerpc|ppc64|sh4|sh4eb|sparc|sparc64)
> +                	BINFMT_ARCH=$ARCH ;;
> +                armel) 	BINFMT_ARCH=arm ;;
> +                lpia)  	BINFMT_ARCH=i386 ;; # not yet verified FIXME
> +                *) echo "unknown architecture: $ARCH"; exit 1;;
> +        esac
> 
> The most prominent problem IMHO is that it requires a system-wide
> setting, and that in turn ties a qemu arch name to a particular
> distro/ABI version.
Can you elaborate where you see the problem?

> An alternative approach would be to stack another LD_PRELOAD hack,
> whose job would be to wrap calls to exec calls and wrap them into a
> suitable qemu invocation.
> 
> Doing so would require careful wrapping of all those *exec*, system,
> posix_spawn*; it would be tedious to take care not to break the
> semantics of those library calls, and the number of such calls
> suggests that more of them will likely appear, breaking things as they
> appear.  That is, while it is surely workable for the short term, it
> looks like a costly hack, and we may want to avoid that.
> 
> Another approach would be to intercept the execve syscall itself,
> using ptrace.  That would have lower dev and maintainance cost, but
> higher runtime penalty.  This idea was implemented in the past in
> subterfugue (http://subterfugue.org/) - which would provide at least a
> prototyping platform if it had not suffered from bitrot.  This is
> something I'm considering having a look at (I already have some old
> patches addressing some of subterfugue's bitrot).
I'm not sure whether you dont understand the problem that is to be
solved or whether I understand how you suggest something I dont see a
need to be fixed - please explain.

The $ARCH variable is being set in the config file and determines for
what debian architecture a multistrap is done. To figure out which
qemu-*-static binary to copy in the root directory, the above lines are
evaluated - how do you think that can be improved?

cheers, josch