Re: [PATCH] haifa-sched: SCHED_GROUP_P ignored for queued insn

[Vladimir Makarov <vmakarov at redhat dot com>]

Andreas Krebbel wrote:

> Hi,
>
>  
>
> > It could be used for solving the problem.  But it results in inaccuarate 
> > simulated clock and as consequence potentially worse code.  The insn in 
> > the group should be issued on right time.  So the problem should solved 
> > by advacning clock not changing cost. 
> >    
>
> Actually that is exactly what I was trying to accomplish with my patch.
> If insn is a SCHED_GROUP_P insn, the clock is advanced by cost wherein
> cost is the number of cycles the insn otherwise would be stalled.
>
>  
Sorry, I've just read your email and the changelog entry.  I should have 
looked at the patch with more attention.

The patch is ok.  It is safe and solves the bug.  So you can commit it 
into the mainline.

Thanks for the patch.


Vlad