On Fri, Apr 1, 2016 at 9:54 PM, Patrick Palka <patr...@parcs.ath.cx> wrote:
> On Fri, Apr 1, 2016 at 9:28 PM, Jason Merrill <ja...@redhat.com> wrote:
>> I like this approach a lot. One thing, though:
>>
>> On 04/01/2016 03:13 PM, Patrick Palka wrote:
>>>
>>> +struct GTY((chain_next ("%h.prev"))) bpr_entry
>>> +{
>>> + tree body;
>>> + tree parms;
>>> + tree res;
>>> + struct bpr_entry *prev;
>>> +};
>>> +
>>> /* Representation of entries in the constexpr function definition table.
>>> */
>>>
>>> struct GTY((for_user)) constexpr_fundef {
>>> tree decl;
>>> tree body;
>>> + /* A chain of unused copies of this function's body awaiting reuse for
>>> + CALL_EXPR evaluation. */
>>> + struct bpr_entry *bpr_freelist;
>>> };
>>
>>
>> The freelist should be discarded on GC. I think the way to achieve that is
>> to put all the freelists into a separate deletable hash table rather than
>> hang them off the fundefs.
>>
>> Jason
>>
>
> Would instead marking the bpr_freelist field as GTY ((skip)) have the
> same effect?
Looks like not according to
https://gcc.gnu.org/onlinedocs/gccint/GTY-Options.html. I should have
read that first :)
