lkml.org 
[lkml]   [2013]   [Oct]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRE: [PATCH 23/26] ib: Convert qib_get_user_pages() to get_user_pages_unlocked()
Date


> -----Original Message-----
> From: Jan Kara [mailto:jack@suse.cz]
> Sent: Friday, October 04, 2013 2:33 PM
> To: Marciniszyn, Mike
> Cc: Jan Kara; LKML; linux-mm@kvack.org; infinipath; Roland Dreier; linux-
> rdma@vger.kernel.org
> Subject: Re: [PATCH 23/26] ib: Convert qib_get_user_pages() to
> get_user_pages_unlocked()
>
> On Fri 04-10-13 13:52:49, Marciniszyn, Mike wrote:
> > > Convert qib_get_user_pages() to use get_user_pages_unlocked(). This
> > > shortens the section where we hold mmap_sem for writing and also
> > > removes the knowledge about get_user_pages() locking from ipath
> > > driver. We also fix a bug in testing pinned number of pages when
> changing the code.
> > >
> >
> > This patch and the sibling ipath patch will nominally take the mmap_sem
> > twice where the old routine only took it once. This is a performance
> > issue.
> It will take mmap_sem only once during normal operation. Only if
> get_user_pages_unlocked() fail, we have to take mmap_sem again to undo
> the change of mm->pinned_vm.
>
> > Is the intent here to deprecate get_user_pages()?
> Well, as much as I'd like to, there are really places in mm code which need
> to call get_user_pages() while holding mmap_sem to be able to inspect
> corresponding vmas etc. So I want to reduce get_user_pages() use as much
> as possible but I'm not really hoping in completely removing it.
>
> > I agree, the old code's lock limit test is broke and needs to be fixed.
> > I like the elimination of the silly wrapper routine!
> >
> > Could the lock limit test be pushed into another version of the
> > wrapper so that there is only one set of mmap_sem transactions?
> I'm sorry, I don't understand what you mean here...
>
> Honza
> --
> Jan Kara <jack@suse.cz>
> SUSE Labs, CR


\
 
 \ /
  Last update: 2013-10-07 17:41    [W:0.045 / U:1.772 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site