Re: [PATCH 0/8] Fix missing memory barriers on ARM

[Paolo Bonzini]

On 3/6/23 14:35, David Hildenbrand wrote:
> > Note: I have a follow-up set of patches that gets rid completely of
> > atomic_mb_read(); atomic_mb_set() instead can remain and mimic Linux's
> > smp_store_mb() operation.  A glimpse of these changes is already visible
> > in patches 7 and 8.
>
> That sounds interesting. I was briefly confused about atomic_mb_* ...
>
> ... do we want to add some Fixes: tags or is it too hard to come up with 
> something reasonable?

The Fixes tag would often point to

    commit a0aa44b488b3601415d55041e4619aef5f3a4ba8
    Author: Alex Bennée <alex.bennee@linaro.org>
    Date:   Thu Jan 28 10:15:17 2016 +0000

    include/qemu/atomic.h: default to __atomic functions
    
    The __atomic primitives have been available since GCC 4.7 and provide
    a richer interface for describing memory ordering requirements. As a
    bonus by using the primitives instead of hand-rolled functions we can
    use tools such as the ThreadSanitizer which need the use of well
    defined APIs for its analysis.
    
    If we have __ATOMIC defines we exclusively use the __atomic primitives
    for all our atomic access. Otherwise we fall back to the mixture of
    __sync and hand-rolled barrier cases.

(for which I would have sworn I was the sole perpetrator, not just the
committer).  But it pre-dates some of the code that is being fixed, so
I am not sure it makes sense to add it.

Paolo