[PATCH 1/2] toplevel: do not make libstdc++ configure wait for libgomp

[Kyrylo Tkachov]

> On 17 Aug 2026, at 13:22, Jonathan Wakely <jwakely@redhat.com> wrote:
> 
> On Wed, 12 Aug 2026 at 14:56, Jeffrey Law <jeffrey.law@oss.qualcomm.com> wrote:
>> 
>> 
>> 
>> On 8/12/2026 7:50 AM, Kyrylo Tkachov wrote:
>>> 
>>>> On 12 Aug 2026, at 15:30, Jeffrey Law <jeffrey.law@oss.qualcomm.com> wrote:
>>>> 
>>>> 
>>>> 
>>>> On 8/11/2026 9:24 AM, Sam James wrote:
>>>>> Kyrylo Tkachov <ktkachov@nvidia.com> writes:
>>>>> 
>>>>>>> On 11 Aug 2026, at 15:11, Sam James <sam@gentoo.org> wrote:
>>>>>>> 
>>>>>>> <ktkachov@nvidia.com> writes:
>>>>>>> 
>>>>>>>> From: Kyrylo Tkachov <ktkachov@nvidia.com>
>>>>>>>> 
>>>>>>>> libstdc++'s configure was made to depend on libgomp's configure in 2007, when
>>>>>>>> GLIBCXX_ENABLE_PARALLEL decided whether to enable parallel mode by looking for
>>>>>>>> a configured libgomp/omp.h on disk.  Since 2010 the test has instead grepped
>>>>>>> Could you include commit refs for these? It's nice for us who go digging
>>>>>>> fairly often (also makes it easier to verify this is right).
>>>>>> Yeah, sure. Here it is with the references in the commit message.
>>>>> Thank you!
>>>>> 
>>>>> I can't approve it but both patches LGTM.
>>>> OK for the trunk.  Thanks!
>>> Thanks, does that include patch 2 (the one that gives the biggest bootstrap gain)? Or do we want a libstdc++ for that?
>> Sorry I wasn't clear.  Just patch #1. I'd prefer one of the libstdc++
>> guys chime in on patch #2 since it's clearly in their space.
> 
> 
> I approved patch 2 as well, thanks for working on this.

Thanks all, I’ve pushed both to master with the extra comment Jonathan requested.
Kyrill