On 02/08/10 20:16, Ulrich Weigand wrote:
Now this point is where the suggestion to use something like a bind mount on startup comes in. That way, there would be no violation of the multiarch rules, because /lib/ld.so.1 would not be part of any package, and in fact not even part of any file system on disk, but simply be present in the in-memory mount table.
In many ways, this would be an elegant solution to the problem.
The problem that I foresee is that a) old programs cannot be used as "foreign" binaries on a multiarch system, and b) there's nothing to stop new programs being "accidentally" linked against the deprecated locations, and likewise not working as "foreign" binaries.
I suggest teaching the kernel to rewrite that path when it finds a non-existent interpreter. Presumably the kernel can "know" what multiarch corresponds to the traditional ABI for any given ELF flags.
Just my suggestion.
Andrew