Re: [RFC PATCH 2/4] xen/arm64: bitops: justify uninitialized variable inside a macro

[Nicola Vetrini <nicola.vetrini@xxxxxxxxxxx>]

On 17/07/23 22:45, Julien Grall wrote:
> On 17/07/2023 21:40, Julien Grall wrote:
> > On 14/07/2023 12:49, Nicola Vetrini wrote:
> > > The macro 'testop' expands to a function that declares the local
> > > variable 'oldbit', which is written before being set, but is such a
> > > way that is not amenable to automatic checking.
> > >
> > > Therefore, a deviation comment, is introduced to document this 
> > > situation.
> > > A similar reasoning applies to macro 'guest_testop'.
> > Would you be able to check if the code below (only compile tested so 
> > far) would silence Eclair?
> Hmmm.. I think my e-mail client mangled the diff. Here an unmangled 
> version if needed:
> http://paste.debian.net/1286154/
>
> Cheers,
I have a question: wouldn't this patch also imply an update of 
xen/arch/arm/include/asm/guest_atomics.h ?
This in particular is the bit of code that needs to be reworked to use 
the newer *_timeout definition.

bool succeed;

succeed = name##_timeout(nr, p, &oldbit, 

                         this_cpu(guest_safe_atomic_max));

if ( succeed )
        return oldbit;

Probably something similar to


int succeed;

succeed = name##_timeout(nr, p, false, NULL, [...]);
if ( succeed != 0 )
  return succeed;

I'm not really sure whether the third param should be true or false, but 
I assumed false because it's similar to the use in bitops.c
--
Nicola Vetrini, BSc
Software Engineer, BUGSENG srl (https://bugseng.com)