Re: [patch] fix PR43602 for emutls targets.

[IainS <developer at sandoe-acoustics dot co dot uk>]

Jakub,

Thanks for the review, and sorry about the formating errors.

On 12 May 2010, at 06:47, Jakub Jelinek wrote:

> On Tue, May 11, 2010 at 08:19:51PM -0400, Jack Howarth wrote:
> > On Tue, May 11, 2010 at 10:13:50AM -0400, Diego Novillo wrote:
> >  It doesn't appear that Jakub's suggested reconfiguration of the  
> > patch is
> > going to bootstrap.

It does for me, on {i686,powerpc}-apple-darwin9.

With the suggested changes it checks without new errors with the race- 
condition changes applied.
(I've also checked with my additional tls checks on ObjC/C++)

===

Jack: could you re-check the attached on x86_64-apple-darwin10?

Jakub, would this be OK to apply (assuming that the x86_64 check  
succeeds) ?

Iain


gcc/Changelog:

	PR 43602
	* varasm.c(get_emutls_init_templ_addr): Copy DECL_VISIBILITY_SPECIFIED
	(emutls_decl): Ditto.
	(emutls_finalize_control_var): New.
	(emutls_finish): Finalize emutls control variables.
	* toplev.c(compile_file): Use emutls_finish () to finalize emutls  
control vars.  Move
	its use ahead of varpool_assemble_pending_decls ().

Attachment: 159280-emutls-diff.txt
Description: Text document