This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: The current egcs is broken on x86


> 
> | Why this? Have you looked my version? It doesn't make any senses
> | to call single_set nor reg_mentioned_p if REG_LIVE_LENGTH is
> | negative.
> 
> No I didn't see your version.  Yeah, I might change the order, but its harmless
> to do the tests in either order (unlike the validate test which changes the
> insn).  The single_set BTW will always be true at present, since the caller has
> called single_set before setting the variables.  I kept it in the function,
> since I don't like dependencies like that (ie, if somebody else calls
> copy_src_or_dest, it may not be the case that single_set is true).  If you look
> at the comments in regs.h, you will see that it is expected to be rare that
> REG_LIVE_LENGTH is negative.
> 

It may be rare. But it is not that rare. Change the order should help
egcs performance.


-- 
H.J. Lu (hjl@gnu.org)


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]