On Tue, 2023-08-01 at 19:35 +0200, Christoph Hellwig wrote:

It has recently come to my attention that nvidia is circumventing the
protection added in 262e6ae7081d ("modules: inherit
TAINT_PROPRIETARY_MODULE") by importing exports from their proprietary
modules into an allegedly GPL licensed module and then rexporting them.

Given that symbol_get was only ever intended for tightly cooperating
modules using very internal symbols it is logical to restrict it to
being used on EXPORT_SYMBOL_GPL and prevent nvidia from costly DMCA
Circumvention of Access Controls law suites.

I'm all for insisting that everything be exported with
EXPORT_SYMBOL_GPL and nothing at all ever be exported with just
EXPORT_SYMBOL.

But if we're going to tolerate the core kernel still exporting some
stuff with EXPORT_SYMBOL, why isn't OK for a GPL-licensed module do to
the same? Even an *in-tree* GPL-licensed module now can't export
functionality with EXPORT_SYMBOL and have it used with symbol_get().

We're forced to *either* allow direct linking by non-GPL modules, or
allow symbol_get(), but not both?

Fixes: 262e6ae7081d ("modules: inherit TAINT_PROPRIETARY_MODULE")

Hm, the condition we really need to fix *that* is "symbol_get() will
only import symbols from GPL-licensed modules", isn't it?

As long as that property is correctly transitive, why does the symbol
itself have to be EXPORT_SYMBOL_GPL instead of EXPORT_SYMBOL? Am I
missing another potential loophole?

I suppose there's now scope for a different type of shim which
*directly* imports an EXPORT_SYMBOL function in order to export it
again as EXPORT_SYMBOL_GPL and thus allow the GPL export to be found
with symbol_get()?

That's the *converse* of the problematic shim that was being used
before, and from a licensing point of view it seems fine... it's just
working around the unintended side-effects of this patch?