Re: Moving pthread_(cond|mutex) to libc

[manu%netbsd.org@localhost (Emmanuel Dreyfus)]

David Laight <david%l8s.co.uk@localhost> wrote:

> > Another idea:  let libpthread be dlopen'ed, but abort if
> > pthread_create() is called after libc mutex or condvar stubs have been
> > called.
> That will cause similar grief later on - and probably be harder to
> diagnose than had the dlopen() fail.

Not quite: it solves the problem when libpthread is loaded as a
dependency but is not wanted, and not used at all. It worked fine prior
6.0 and is now broken. This is that regression I want to address. 

For diagnosing purpose, we may have more explicit error message.

-- 
Emmanuel Dreyfus
http://hcpnet.free.fr/pubz
manu%netbsd.org@localhost