This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: PAGESIZE
- To: Bruce Korb <bkorb at sco dot COM>
- Subject: Re: PAGESIZE
- From: Zack Weinberg <zack at wolery dot cumb dot org>
- Date: Tue, 5 Sep 2000 15:09:25 -0700
- Cc: DJ Delorie <dj at delorie dot com>, GNU Compiler <gcc at gcc dot gnu dot org>
- References: <200009051939.PAA27092@envy.delorie.com> <39B56BE2.FF856877@sco.com>
On Tue, Sep 05, 2000 at 02:55:46PM -0700, Bruce Korb wrote:
> DJ Delorie wrote:
> >
> > 2000-09-05 Bruce Korb <bkorb@gnu.org>
> >
> > * gcc/fixinc/fixincl.c(load_file): always read header files
> > with sizes that are a multiple of the page size.
> >
> > "PAGESIZE" isn't always defined. It isn't on Linux, for example.
>
> I presume that that is the case so that it can be varied?
> >From what to what, I wonder. Surely not from one memory
> manager to another within the same OS image. :-).
Different ABIs on the same platform; different memory models;
different versions of the same processor.
In practice, yes, it's a constant on any given architecture.
> At any rate, there is no sysconf(2) attribute for it either.
> I propose disabling mmap altogether, unless someone can point
> me to a portable solution. (Always assume that a memory page
> will never be smaller than 0x1000, for example? I never get
> warm fuzzies from such assumptions.)
This is overkill. getpagesize() is in libiberty, you don't have to
worry about it.
zw