Re: [Patch v2] mm/vmscan: fix unsequenced modification and access warning

[Michal Hocko]

I have discussed this with our gcc guys and here is what they say:

On Wed 10-05-17 10:38:44, Michal Hocko wrote:
[...]
> But I
> still do not understand which part of the code is undefined and why. My
> reading and understanding of the C specification is that
> struct A {
> 	int a;
> 	int b;
> };
> 
> struct A f = { .a = c = foo(c), .b = c};
> 
> as long as foo(c) doesn't have any side effects because because .a is
> initialized before b and the assignment ordering will make sure that c
> is initialized before a.
> 
> 6.7.8 par 19 (ISO/IEC 9899)
> 19 The initialization shall occur in initializer list order, each
>    initializer provided for a particular subobject overriding any
>    previously listed initializer for the same subobject; all subobjects
>    that are not initialized explicitly shall be initialized implicitly
>    the same as objects that have static storage duration.
> 
> So is my understanding of the specification wrong or is this a bug in
> -Wunsequenced in Clang?

: This is not the reason why the above is okay.  The following part:
:    { .a = c = ..., .b = c }
: is okay because there's a sequence point after each full expression, and 
: an initializer is a full expression, so there's a sequence point between 
: both initializers.  The following part:
:    { ... c = foo(c) ... }
: is okay as well, because there's a sequence point after evaluating all 
: arguments and before the actual call (otherwise the common 'i=next(i)' 
: idiom doesn't work).  So both constructs that potentially could be sources 
: of sequence point violations actually aren't and hence okay.  clangs 
: warning is invalid.

I guess it is worth reporting this to clang bugzilla. Could you take
care of that Nick?
-- 
Michal Hocko
SUSE Labs