lkml.org 
[lkml]   [2005]   [Oct]   [6]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
SubjectRe: [PATCH 0/3] Demand faulting for huge pages
From
Date
On Thu, 2005-09-29 at 14:32 +0100, Hugh Dickins wrote:
> On Wed, 28 Sep 2005, Adam Litke wrote:
>
> > Hi Andrew. Can we give hugetlb demand faulting a spin in the mm tree?
> > And could people with alpha, sparc, and ia64 machines give them a good
> > spin? I haven't been able to test those arches yet.
>
> It's going to be a little confusing if these go in while I'm moving
> the page_table_lock inwards. My patches don't make a big difference
> to hugetlb (I've not attempted splitting the lock at all for hugetlb -
> there would be per-arch implementation issues and very little point -
> though more point if we do move to hugetlb faulting). But I'm ill at
> ease with changing the locking at one end while it's unclear whether
> it's right at the other end.
>
> Currently Adam's patches don't include my hugetlb changes already in
> -mm; and I don't see any attention in his patches to the issue of
> hugetlb file truncation, which I was fixing up in those.
>
> The current hugetlb_prefault guards against this with i_sem held:
> which _appears_ to be a lock ordering violation, but may not be,
> since the official lock ordering is determined by the possibility
> of fault within write, whereas hugetlb mmaps were never faulting.
>
> Presumably on-demand hugetlb faulting would entail truncate_count
> checking like do_no_page, and corresponding code in hugetlbfs.
>
> I've no experience of hugetlb use. Personally, I'd be very happy with
> a decision to disallow truncation of hugetlb files (seems odd to allow
> ftruncate when read and write are not allowed, and the size normally
> determined automatically by mmap size); but I have to assume that it's
> been allowed for good reason.

If I were to spend time coding up a patch to remove truncation support
for hugetlbfs, would it be something other people would want to see
merged as well?

--
Adam Litke - (agl at us.ibm.com)
IBM Linux Technology Center

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-10-06 17:45    [W:0.086 / U:1.192 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site