Re: GC seems to kill process

[Hans Boehm <Hans dot Boehm at hp dot com>]

Bryce -

Note that this is starting to get very similar to GC_register_my_thread
in gc7, but that interface is a bit more elaborate, to deal with a
variety of platforms.  Would it make sense to strive for interface
compatibility?

The interface is described in

http://cvs.sourceforge.net/viewcvs.py/bdwgc/bdwgc/include/gc.h?rev=1.5&view=markup

Hans

On Thu, 23 Mar 2006, Bryce McKinlay wrote:

> Andi Vajda wrote:
> > The darwinports build of gcc 4.1.0 I started this morning finally
> > completed and all seems ok. All PyLucene unit tests pass. No crash, no
> > errors.
> >
> > The problem I reported earlier seems to only occur on Linux.
>
> Could you give this (experimental) patch a try? It adds new
> GC_{attach,detach}_thread() functions to the boehm-gc, which register
> threads directly from the invocation API functions instead of attempting
> to intercept pthread_create() at the linker level.
>
> The patch is against gcc-4.1-branch. I'm very interested to know if it
> helps with this problem.
>
> Bryce
>
>