Re: [RFC][PATCH 1/3] radix priority search tree - objrmap complexity fix

[Andrea Arcangeli]

On Fri, Apr 02, 2004 at 05:07:52PM +1000, Paul Mackerras wrote:
> Christoph Hellwig writes:
> > On Fri, Apr 02, 2004 at 04:22:33AM +0200, Andrea Arcangeli wrote:
> > > Well, I doubt anybody could take advantage of this optimization, since
> > > nobody can ship with hugetlbfs disabled anyways (peraphs with the
> > > exception of the embedded people but then I doubt they want to risk
> > 
> > Common. stop smoking that bad stuff.  Almost non-one except the known
> > oracle whores SuSE and RH need it.  Remeber Linux is used much more widely
> > except the known "Enterprise" vendors.  None of the NAS/networking/media
> > applicances or pdas I've seen has the slightest need for hugetlbfs.
> 
> The HPC types also love hugetlbfs since it reduces their tlb miss
> rate.

the point is not the number of people who needs this, the point is that
any distributor will be forced to turn it on, since the distributor
must allow any user to do HPC or database applications. Shipping with
hugetlbfs=n is like shipping with device-mapper=n or sysfs=n or whatever
like that. And having different alloc_pages API depending on a hugetlbfs
compile option was totally broken, plus hugetlbfs=n was buggy, this is
all fixed now.

If we want to make the API not generate compound pages unless you call
with __GFP_COMP that's fine with me, that'll optimize the whole kernel,
and that's a very simple variation of my code. Still the alloc_pages API
must be indipendent of a hugetlbfs compile option.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/