Re: [Qemu-devel] [PATCH v6 1/8] atomic: introduce atomic operations

[liu ping fan]

On Tue, Nov 13, 2012 at 6:11 PM, Paolo Bonzini <address@hidden> wrote:
>> > Il 05/11/2012 06:38, Liu Ping Fan ha scritto:
>> > > From: Liu Ping Fan <address@hidden>
>> > >
>> > > If out of global lock, we will be challenged by SMP in low level,
>> > > so need atomic ops.
>> > >
>> > > This file is a wrapper of GCC atomic builtin.
>> >
>> > I still object to this.
>> >
>> > I know it enforces type-safety, but it is incomplete.  It doesn't
>>
>> Although it is incomplete, but the rest cases are rarely used.  Linux
>> faces such issue, and the "int" version is enough, so I think we can
>> borrow experience from there.
>
> One of the two places that use __sync_* require 64-bit accesses.  My
Yes, these two places are not easy to fix.

> RCU prototype required pointer-sized access, which you cannot make type-
But I think that your RCU prototype should rely on atomic of CPU, not
gcc‘s atomic.
Otherwise, it could be slow (I guess something like spinlock there).

Regards,
pingfan

> safe without C++ templates.
>
>> > provide neither atomic accesses to pointers, nor useful operations such
>> > as exchange.  It won't be used consistently, because in some places you
>> > just do not have an Atomic value (see both current uses of __sync_*
>> > builtins).
>> >
>> > If you can make it complete, and prove it by using it where
>> > __sync_* is
>>
>> For current code, __sync_* is used rarely, I think except the
>> barriers, only two places use it. We can substitute it easily.
>
> No, you cannot.  See above, one doesn't use ints and the other is
> guest state so migration becomes complicated if you use Atomic.  I'm
> happy to be proven wrong, however.
>
> Paolo