On Mon, 17 Jun 2024 at 11:18, Jonathan Wakely wrote:
>
> On Sat, 15 Jun 2024 at 14:04, François Dumont wrote:
> >
> > Here is the simplified patch then.
>
> The use of std::__to_address seems wrong.
>
> The allocator returns a __buckets_ptr, and that function returns a
> __buckets_ptr, so it shou
On Sat, 15 Jun 2024 at 14:04, François Dumont wrote:
>
> Here is the simplified patch then.
The use of std::__to_address seems wrong.
The allocator returns a __buckets_ptr, and that function returns a
__buckets_ptr, so it should just be returned unchanged, not by
converting to a raw pointer with
Here is the simplified patch then.
libstdc++: Do not use memset in _Hashtable buckets allocation
Using memset is incorrect if the __bucket_ptr type is non-trivial, or
does not use an all-zero bit pattern for its null value.
Replace the use of memset with std::__uinitialized_defa
On Thu, 13 Jun 2024 at 19:57, Jonathan Wakely wrote:
>
> On Thu, 13 Jun 2024 at 18:40, François Dumont wrote:
> >
> > Hi
> >
> > Following your recent change here:
> >
> > https://gcc.gnu.org/pipermail/libstdc++/2024-June/058998.html
> >
> > I think we also need to fix the memset at bucket alloca
On Thu, 13 Jun 2024 at 18:40, François Dumont wrote:
>
> Hi
>
> Following your recent change here:
>
> https://gcc.gnu.org/pipermail/libstdc++/2024-June/058998.html
>
> I think we also need to fix the memset at bucket allocation level.
>
> I did it trying also to be more fancy pointer friendly by
Hi
Following your recent change here:
https://gcc.gnu.org/pipermail/libstdc++/2024-June/058998.html
I think we also need to fix the memset at bucket allocation level.
I did it trying also to be more fancy pointer friendly by running
__uninitialized_default_n_a on the allocator returned pointe
