Re: RFC: initial TLS patch

[Daniel Berlin <dberlin@dberlin.org>]

On 2 Jul 2002, Jim Blandy wrote:

> 
> This is a work-in-progress, posted for comments.  This code has never
> been executed, since I only last night got glibc and GCC to actually
> compile and run a program that uses `__thread', and GCC doesn't yet
> emit the Dwarf 2 extension for thread-local variables.

Maybe you could review the patch from april that implements the suggested 
location function mechanism, and use that here to handle thread local 
variable locations, rather than reuse LOC_THREAD_LOCAL_STATIC (Which, 
correct me if i'm wrong, is only currently used by hpread, and pretty 
esay to get rid of before your patch)?
After all, aren't we trying to get rid of all the location types eventually?
It would seem to make sense to not add more uses of them if it can be 
avoided.
Of course, if you have major problems with that patch, it's understandable 
that you'd rather not have this work depend on it, but if it's just minor 
issues, it would seem to make sense to me.