Currently as-base classes lack DECL_BIT_FIELD_REPRESENTATIVEs which
means RTL expansion doesn't honor the C++ memory model for bitfields
in them thus for the following testcase
struct B {
B() {}
int x;
int a : 6;
int b : 6;
int c : 6;
};
struct C : B {
char d;
};
C c;
int main()
{
c.c = 1;
c.d = 2;
}
on x86 we happily store to c.c in a way creating a store data race with
c.d:
main:
.LFB6:
.cfi_startproc
movl c+4(%rip), %eax
andl $-258049, %eax
orb $16, %ah
movl %eax, c+4(%rip)
movb $2, c+7(%rip)
xorl %eax, %eax
ret
Fixing the lack of DECL_BIT_FIELD_REPRESENTATIVEs in as-base
classes doesn't help though as the C++ FE builds access trees
for c.c using the non-as-base class FIELD_DECLs which is because
of layout_class_type doing
/* Now that we're done with layout, give the base fields the real types. */
for (field = TYPE_FIELDS (t); field; field = DECL_CHAIN (field))
if (DECL_ARTIFICIAL (field) && IS_FAKE_BASE_TYPE (TREE_TYPE (field)))
TREE_TYPE (field) = TYPE_CONTEXT (TREE_TYPE (field));
this would basically require us to always treat tail-padding in a
struct conservatively in finish_bitfield_representative (according
to the doubt by the ??? comment I patch out below).
Simply commenting out the above makes fixing build_simple_base_path
necessary but even after that it then complains in the verifier
later ("type mismatch in component reference" - as-base to class
assignment).
But it still somehow ends up using the wrong FIELD_DECL in the end.
Now I think we need to fix the wrong-code issue somehow and
doing so in stor-layout.c by conservatively treating tail-padding
is a possibility. But that will pessimize PODs and other languages
unless we have a way to know whether a RECORD_TYPE possibly can
have its tail-padding re-used (I'd hate to put a lang_hooks.name
check there and even that would pessimize C++ PODs).
Any guidance here?
I opened PR71694 for this.
Richard.
2016-06-29 Richard Biener <rguent...@suse.de>
* stor-layout.c (finish_bitfield_representative): Clarify comment.
cp/
* class.c (layout_class_type): Call finish_bitfield_layout on
the as-base variant.
Index: gcc/stor-layout.c
===================================================================
*** gcc/stor-layout.c (revision 237840)
--- gcc/stor-layout.c (working copy)
*************** finish_bitfield_representative (tree rep
*** 1866,1878 ****
}
else
{
! /* ??? If you consider that tail-padding of this struct might be
! re-used when deriving from it we cannot really do the following
! and thus need to set maxsize to bitsize? Also we cannot
! generally rely on maxsize to fold to an integer constant, so
! use bitsize as fallback for this case. */
tree maxsize = size_diffop (TYPE_SIZE_UNIT (DECL_CONTEXT (field)),
DECL_FIELD_OFFSET (repr));
if (tree_fits_uhwi_p (maxsize))
maxbitsize = (tree_to_uhwi (maxsize) * BITS_PER_UNIT
- tree_to_uhwi (DECL_FIELD_BIT_OFFSET (repr)));
--- 1866,1878 ----
}
else
{
! /* Note that if the C++ FE sets up tail-padding to be re-used it
! creates a as-base variant of the type with TYPE_SIZE adjusted
! accordingly. So it is safe to include tail-padding here. */
tree maxsize = size_diffop (TYPE_SIZE_UNIT (DECL_CONTEXT (field)),
DECL_FIELD_OFFSET (repr));
+ /* We cannot generally rely on maxsize to fold to an integer constant,
+ so use bitsize as fallback for this case. */
if (tree_fits_uhwi_p (maxsize))
maxbitsize = (tree_to_uhwi (maxsize) * BITS_PER_UNIT
- tree_to_uhwi (DECL_FIELD_BIT_OFFSET (repr)));
Index: gcc/cp/class.c
===================================================================
*** gcc/cp/class.c (revision 237840)
--- gcc/cp/class.c (working copy)
*************** layout_class_type (tree t, tree *virtual
*** 6556,6561 ****
--- 6556,6564 ----
}
*next_field = NULL_TREE;
+ /* Finish bitfield layout of the type. */
+ finish_bitfield_layout (base_t);
+
/* Record the base version of the type. */
CLASSTYPE_AS_BASE (t) = base_t;
TYPE_CONTEXT (base_t) = t;
