Re: mmap and MAP_FIXED

[Johnny Billquist <bqt%update.uu.se@localhost>]

On 2016-05-12 16:33, Michael wrote:
> Hello,
>
> On Thu, 12 May 2016 11:26:08 +0200
> Johnny Billquist <bqt%update.uu.se@localhost> wrote:
>
> > I haven't tested, or analyzed this to any depth, but I wonder... What is
> > the pagesize of the machine? 8K? I wonder if you might have some issue
> > when you try to map only 4K fixed. The man-page says that if MAP_FIXED
> > is used, the address must be page aligned, but it does not say anything
> > about the size, but it makes me wonder...
>
> Page size is hardware dependent, I have machines here that use 4K
> ( sparc, powerpc, etc. ), 8K ( sparc64, arm ) or 16K ( some MIPS )
> pages.
> But the start address in the test program is aligned to something much
> larger, so the first test should succeed if that was the problem. I
> wouldn't bet on the address range used being available though, but
> that's just a hunch.

Uh, my point was that the mmap call is only requesting a 4K-sized map, 
which I wonder what could be the effect on a machine with a larger page 
size, and an anonymous mapping with a fixed address.

> Also, why MMAP_FIXED in the first place?

That is a different question, but a very good one. In general, I think 
it's a bad idea to try and force the address.

> Finally, mmap()ing 1TB will probably run into all sorts of other
> problems before hitting mmap() itself. If the underlying filesystem is
> ffsv1 it probably doesn't even support files that big.

You commented yourself on this later, the size is actually only 4K, and 
not 1TB. :-)
Also, note that there is no underlying file system in this case.

	Johnny