Re: [for-4.7] x86/emulate: synchronize LOCKed instruction emulation

[Razvan Cojocaru <rcojocaru@bitdefender.com>]

On 04/14/2016 07:35 AM, Jan Beulich wrote:
>> --- a/xen/arch/x86/hvm/emulate.c
>> >+++ b/xen/arch/x86/hvm/emulate.c
>> >@@ -25,6 +25,8 @@
>  >#include <asm/hvm/svm/svm.h>
>  >#include <asm/vm_event.h>
>  >
>> >+DEFINE_PERCPU_RWLOCK_GLOBAL(emulate_locked_rwlock);
> You should try hard to make this static.

On second though, this would make the code somewhat more convoluted, as
the functions in emulate.c need to access this variable, and so does the
code in domain.c that initializes the lock when the domain is created.

What I've done is similar to what I found in the current source code with:

arch/x86/mm/p2m.c:DEFINE_PERCPU_RWLOCK_GLOBAL(p2m_percpu_rwlock);
common/grant_table.c:DEFINE_PERCPU_RWLOCK_GLOBAL(grant_rwlock);

But I could add a function to emulate.h, for example:

void init_emulate_smp_lock();

and call that from arch/x86/domain.c (as suggested by Andrew Cooper),
which would allow domain.c to stop referencing emulate_locked_rwlock
directly.

Would you prefer this?


Thanks,
Razvan

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
http://lists.xen.org/xen-devel