Re: Link time optimizations

jaidi@ubd.edu.bn (Nor Jaidi)
10 Feb 1996 11:53:00 -0500

          From comp.compilers

Related articles
Link time optimizations jacob@jacob.remcomp.fr (1996-02-09)
Re: Link time optimizations jaidi@ubd.edu.bn (1996-02-10)
Re: Link time optimizations richard@atheist.tamu.edu (1996-02-13)
Re: Link time optimizations jaidi@ubd.edu.bn (1996-02-14)
| List of all articles for this month |

From: jaidi@ubd.edu.bn (Nor Jaidi)
Newsgroups: comp.compilers
Date: 10 Feb 1996 11:53:00 -0500
Organization: Compilers Central
References: 96-02-088
Keywords: linker, optimize

> ... Normally all compilers generate a jump PC relative instruction
> that generates no relocation or fixup... Invisible damm it!


Procedure calls are segment-relative and this is also invisible.
Anyway, you also have to consider the possibility of indirect call
(via pointer to function). (Another "damn it!" I guess)


I wonder how Topspeed linker does it. Or perhaps I
misread/misunderstood their promotional article. Or it only applies to
their own object file which might include extra stuff. (May be I
should use past tense here because I haven't seen anything that spells
Topspeed for quite a while).


Nor Jaidi jaidi@ubd.edu.bn
Department of Mathematics
Universiti Brunei Darussalam
[Function pointers are easy, since they show up as relocation items.
It's long seemed to me that either linkers should be a lot smarter, so
we can do more link-time optimizations, or they should be a lot
dumber, with all the work done in the assembler or above. -John]


--


Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.