hppa dynamic linker fix for g++ 3.0

Alan Modra amodra@bigpond.net.au
Mon Jun 18 01:55:00 GMT 2001


On Mon, Jun 18, 2001 at 12:19:06AM -0700, Ulrich Drepper wrote:
> Alan Modra <amodra@bigpond.net.au> writes:
> 
> > Not true.  You may have missed the fact that byte stores are only used
> > when the address is unaligned.
> 
> No, I did not.

Then you had no reason to claim "This is a horrible performance killer"

> > Adding another reloc type will cost at least one branch in the loop
> > for a number of reloc types, and probably an instruction too.
> 
> The relocation tests happen at the outer level and by correct ordering
> you avoid the tests unless necessary.

That may or may not be possible.  gcc reorders switch case statements to
minimise the total number of comparisons.

Alan



More information about the Libc-alpha mailing list