Re: VIRT_ADDR_VARIES

[Eli Zaretskii]

> Date: Thu, 10 Nov 2011 08:50:39 -0800
> From: Paul Eggert <address@hidden>
> CC: address@hidden
> 
> On 11/10/11 08:43, Eli Zaretskii wrote:
> > But why would someone do such a thing?
> 
> Performance.
> 
> Comparing against my_edata takes just one instruction.
> Comparing against both the start and the end of 'pure'
> takes two comparisons and some glue instructions and
> typically includes a conditional branch.

I'm surprised.  Saving a few instructions in something that is never
called inside any time-critical code would be more than just premature
optimization in my book.