[PATCH] Fix c++/78621 generic lambda mangling

[Nathan Sidwell]

This patch fixes the generic lambda mangling bug that caused the 
demangler to infinitely recurse (78252).  The generic's auto parms fail 
the is_auto check because that's testing for the 'auto' identifier, 
whereas the parms are distinct template type parms named 'auto:$N'. 
Amending is_auto to do a textual compare of the identifier breaks 
things, and is insufficient anyway.  The mangler's squangling must (a) 
squangle subsequent auto uses as expected and (b) NOT squangle 
subsequent template type parm references to refer the the lambda's autos.

This patch achieves that by tsubsting the generic fn's parameters using 
a consed up array of regular auto types.  The make_auto_1 function is 
further split, so we can control the template LEVEL and ORIG_LEVEL 
values.  When mangling a lambda, we check if it's a generic lambda and 
then (after checking the ABI flag) perform this substitution.  We also 
set the needs_warning flag, if we detect the user's asked about ABI 
compatibility crossing version 11.

This use of tsubst can meet {TYPE,EXPR}_PARAMETER_PACKs, so I had to 
extend tsubst to process those objects.

lambda-mangle-1 is an augmented version of the testcase I included with 
78252.  It checks we retain the old mangling for ABI 10 and below.
lambda-mangle-2 is essentially the same testcase, but checks we use the 
new mangling for ABI=0
lambda-mangle-3 checks we issue a warning for -Wabi=10

I also included a demangler test for the new mangling.  It does demangle 
as expected.

Documentation about -fabi-version=11 also updated.

ok?

nathan
-- 
Nathan Sidwell

-------------- next part --------------
A non-text attachment was scrubbed...
Name: auto-mangle.patch
Type: text/x-patch
Size: 12562 bytes
Desc: not available
URL: <http://gcc.gnu.org/pipermail/gcc-patches/attachments/20161207/00d221c8/attachment.bin>