[cfe-commits] [patch] Fix PR6353 and some related issues

[Rafael Espindola]

The attached patch adds asserts to check that if we are doing codegen
for a implicit constructor, destructor or copy assignment, then we
should have marked it as used in Sema. The patch also fixes all cases
I found in a linux x86 bootstrap. The bootstrap fails when linking opt
and clang, but it also fails without the test.

One part of the patch I am not very happy with is the change to
SemaExpr.  It is there to match the logic in  EmitBaseInitializer. The
problem is that it is always marking the destructor as used. We should
do so only if it is actually being used in a base initializer and
exceptions are enabled. Suggestions on how to do so are welcome.

Is the patch OK if a bootstrap on Darwin x86-64 is OK?

Cheers,
-- 
Rafael Ávila de Espíndola
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PR6353.patch
Type: text/x-diff
Size: 10474 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20100228/bffd986a/attachment.patch>