x86: export pv_lock_ops non-GPL
Subsystems (2)
PATCH 1/1 x86: export pv_lock_ops non-GPL
None of the spinlock API is exported GPL, so there's no reason for pv_lock_ops to be.
--- arch/x86/kernel/paravirt-spinlocks.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)
===================================================================
arch/x86/kernel/paravirt-spinlocks.c +1 -1
@@ -17,7 +17,7 @@ .spin_unlock = __ticket_spin_unlock,#endif }; -EXPORT_SYMBOL_GPL (pv_lock_ops); +EXPORT_SYMBOL (pv_lock_ops);void __init paravirt_use_bytelocks (void ) {
Quoted 2 lines · Jeremy Fitzhardinge
None of the spinlock API is exported GPL, so there's no reason for
pv_lock_opsto be.
uhm, why is this even a question? All of this stuff is so deep inside the guts of the kernel ... so nope - if then we want to change the others to _GPL exports as well, just to make sure nobody gets funny ideas.
Ingo
Quoted 9 lines · Ingo Molnar
Quoted 2 lines · Jeremy Fitzhardinge
None of the spinlock API is exported GPL, so there's no reason for
pv_lock_opsto be.uhm, why is this even a question? All of this stuff is so deep inside the guts of the kernel ... so nope - if then we want to change the others to
_GPLexports as well, just to make sure nobody gets funny ideas.
i've nevertheless applied it to tip/x86/urgent - but it would be nice to clean all of this up and only export the minimally necessary (ex include file inline functions) APIs and none of the paravirt stuff.
Ingo
Quoted 4 lines · Ingo Molnar
uhm, why is this even a question? All of this stuff is so deep inside the guts of the kernel ... so nope - if then we want to change the others to
_GPLexports as well, just to make sure nobody gets funny ideas.
pv_lock_ops gets accessed by inline functions which get inlined into modules. They should never directly access it, but the generated code will. _spin_lock() is in exactly the same situation.
J