Re: PATCH libatomic

[Joerg Sonnenberger <joerg%bec.de@localhost>]

On Fri, May 08, 2020 at 02:44:14PM +0200, Martin Husemann wrote:
> On Fri, May 08, 2020 at 02:26:45PM +0200, Kamil Rytarowski wrote:
> > With _Atomic() we can mark any arbitrary struct to have serialized
> > accesses. As I said, with your attitude we shall remove FPU support (and
> > softfloat) as they are not async safe, not safe in virtualization for
> > MMU accesses and more just to mention 2 issues.
> 
> Now come on, that certainly is apples and oranges.

Actually, it is not. Take a look at the history of
pkgsrc/x11/xorg-libs/patches/patch-bl for how ironically similar apples
and oranges can be. I've dealt with code using atomics in ways that fail
for any of the three given reasons in the wild. Async signal safety is
an edge case, relatively speaking. The other two are not. Take atomic access
to shared memory is not used in real world software? Name one insanely
large and "essential" software application by memory that has been used
as justification for pulling Futexes into the kernel. You have one try
and 5 seconds. Now tell me how sure you are that they only use Futexes
and not other forms of atomics.

Joerg