[PATCH] Handle atomic types in Type::hasIntegerRepresentation()
John McCall
rjmccall at apple.com
Mon Apr 15 15:45:47 PDT 2013
On Apr 15, 2013, at 3:15 PM, Joey Gouly <joey.gouly at arm.com> wrote:
> I don't think these should be accepted. Reading 6.5.7.2 of n1570, it says:
> // Each of the operands shall have integer type.//
An operand of _Atomic(int) type should always be an l-value and therefore
implicitly converted to type 'int' before analyzing the shift per 6.3.2.1p2.
I think we probably don't want to have a general explosion of needing to
handle atomic types absolutely everywhere; these analyses should be
working with non-l-value types.
John.
More information about the cfe-commits
mailing list