semi-old -freorder-blocks patch approval?

Richard Henderson rth@cygnus.com
Tue Mar 14 00:05:00 GMT 2000


On Sun, Mar 12, 2000 at 07:30:38PM -0800, Jason Eckhardt wrote:
> i've reviewed this and done some testing. it fixes some -freorder-blocks
> defects.  Would someone approve?
> http://gcc.gnu.org/ml/gcc-patches/2000-02/msg00748.html

!             if (BASIC_BLOCK (i)->local_set)
!               {
!                 OBSTACK_ALLOC_REG_SET (function_obstack);
!                 COPY_REG_SET (nb->local_set, BASIC_BLOCK (i)->local_set);
!               }
!             else
!               BASIC_BLOCK (nb->index)->local_set = 0;

local_set is only valid during calculate_global_regs_live.  There's 
no need to fiddle with it here.

!             make_edge (0, BASIC_BLOCK (i), nb, 0);

First arg should be NULL for my sanity.

Could I get you to factor out all the BASIC_BLOCK (i) invocations
in this section of code into a local variable?

Functionally the patch is fine.


r~


More information about the Gcc-patches mailing list