From: Jan Kara <jack@suse.cz>
To: "Marciniszyn, Mike" <mike.marciniszyn@intel.com>
Cc: Jan Kara <jack@suse.cz>, LKML <linux-kernel@vger.kernel.org>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
infinipath <infinipath@intel.com>,
Roland Dreier <roland@kernel.org>,
"linux-rdma@vger.kernel.org" <linux-rdma@vger.kernel.org>
Subject: Re: [PATCH 23/26] ib: Convert qib_get_user_pages() to get_user_pages_unlocked()
Date: Tue, 8 Oct 2013 21:06:04 +0200 [thread overview]
Message-ID: <20131008190604.GB14223@quack.suse.cz> (raw)
In-Reply-To: <20131007172604.GD30441@quack.suse.cz>
[-- Attachment #1: Type: text/plain, Size: 2440 bytes --]
On Mon 07-10-13 19:26:04, Jan Kara wrote:
> On Mon 07-10-13 15:38:24, Marciniszyn, Mike wrote:
> > > > 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()?
> >
> > The old code looked like:
> > __qib_get_user_pages()
> > (broken) ulimit test
> > for (...)
> > get_user_pages()
> >
> > qib_get_user_pages()
> > mmap_sem lock
> > __qib_get_user_pages()
> > mmap_sem() unlock
> >
> > The new code is:
> >
> > get_user_pages_unlocked()
> > mmap_sem lock
> > get_user_pages()
> > mmap_sem unlock
> >
> > qib_get_user_pages()
> > mmap_sem lock
> > ulimit test and locked pages maintenance
> > mmap_sem unlock
> > for (...)
> > get_user_pages_unlocked()
> >
> > I count an additional pair of mmap_sem transactions in the normal case.
> Ah, sorry, you are right.
>
> > > > 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...
> > >
> >
> > This is what I had in mind:
> >
> > get_user_pages_ulimit_unlocked()
> > mmap_sem lock
> > ulimit test and locked pages maintenance (from qib/ipath)
> > for (...)
> > get_user_pages_unlocked()
> > mmap_sem unlock
> >
> > qib_get_user_pages()
> > get_user_pages_ulimit_unlocked()
> >
> > This really pushes the code into a new wrapper common to ipath/qib and
> > any others that might want to combine locking with ulimit enforcement.
> We could do that but frankly, I'd rather change ulimit enforcement to not
> require mmap_sem and use atomic counter instead. I'll see what I can do.
OK, so something like the attached patch (compile tested only). What do
you think? I'm just not 100% sure removing mmap_sem surrounding stuff like
__ipath_release_user_pages() is safe. I don't see a reason why it shouldn't
be - we have references to the pages and we only mark them dirty and put the
reference - but maybe I miss something subtle...
Honza
--
Jan Kara <jack@suse.cz>
SUSE Labs, CR
[-- Attachment #2: 0001-mm-Switch-mm-pinned_vm-to-atomic_long_t.patch --]
[-- Type: text/x-patch, Size: 0 bytes --]
next prev parent reply other threads:[~2013-10-08 19:06 UTC|newest]
Thread overview: 63+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-10-02 14:27 [PATCH 0/26] get_user_pages() cleanup Jan Kara
2013-10-02 14:27 ` [PATCH 01/26] cris: Convert cryptocop to use get_user_pages_fast() Jan Kara
2013-10-02 14:27 ` [PATCH 02/26] ia64: Use get_user_pages_fast() in err_inject.c Jan Kara
2013-10-02 14:27 ` [PATCH 03/26] dma: Use get_user_pages_fast() in dma_pin_iovec_pages() Jan Kara
2013-10-02 14:27 ` [PATCH 04/26] drm: Convert via driver to use get_user_pages_fast() Jan Kara
2013-10-02 14:27 ` [PATCH 05/26] omap3isp: Make isp_video_buffer_prepare_user() " Jan Kara
2013-10-02 19:41 ` Laurent Pinchart
2013-10-02 20:18 ` Jan Kara
2013-10-02 14:27 ` [PATCH 06/26] vmw_vmci: Convert driver to " Jan Kara
2013-10-02 14:27 ` [PATCH 07/26] st: Convert sgl_map_user_pages() " Jan Kara
2013-10-02 14:27 ` [PATCH 08/26] ced1401: Convert driver " Jan Kara
2013-10-02 14:27 ` [PATCH 09/26] crystalhd: Convert crystalhd_map_dio() " Jan Kara
2013-10-02 14:27 ` [PATCH 10/26] lustre: Convert ll_get_user_pages() " Jan Kara
2013-10-05 6:27 ` Dilger, Andreas
2013-10-02 14:27 ` [PATCH 11/26] sep: Convert sep_lock_user_pages() to get_user_pages_fast() Jan Kara
2013-10-02 14:27 ` [PATCH 12/26] pvr2fb: Convert pvr2fb_write() to use get_user_pages_fast() Jan Kara
2013-10-02 14:27 ` [PATCH 13/26] fsl_hypervisor: Convert ioctl_memcpy() " Jan Kara
2013-10-04 2:38 ` Timur Tabi
2013-10-02 14:27 ` [PATCH 14/26] nfs: Convert direct IO " Jan Kara
2013-10-02 14:27 ` [PATCH 15/26] ceph: Convert ceph_get_direct_page_vector() to get_user_pages_fast() Jan Kara
2013-10-02 14:27 ` [PATCH 16/26] mm: Provide get_user_pages_unlocked() Jan Kara
2013-10-02 16:25 ` Christoph Hellwig
2013-10-02 16:28 ` KOSAKI Motohiro
2013-10-02 19:39 ` Jan Kara
2013-10-02 14:27 ` [PATCH 17/26] kvm: Use get_user_pages_unlocked() in async_pf_execute() Jan Kara
2013-10-02 14:59 ` Gleb Natapov
2013-10-02 14:27 ` [PATCH 18/26] mm: Convert process_vm_rw_pages() to use get_user_pages_unlocked() Jan Kara
2013-10-02 16:32 ` KOSAKI Motohiro
2013-10-02 19:36 ` Jan Kara
2013-10-03 22:40 ` KOSAKI Motohiro
2013-10-07 20:55 ` Jan Kara
2013-10-08 0:10 ` KOSAKI Motohiro
2013-10-02 14:28 ` [PATCH 19/26] ivtv: Convert driver " Jan Kara
2013-10-05 12:02 ` Andy Walls
2013-10-07 17:22 ` Jan Kara
2013-10-02 14:28 ` [PATCH 20/26] ib: Convert ib_umem_get() to get_user_pages_unlocked() Jan Kara
2013-10-02 14:28 ` [PATCH 21/26] ib: Convert ipath_get_user_pages() " Jan Kara
2013-10-02 14:28 ` [PATCH 22/26] ib: Convert ipath_user_sdma_pin_pages() to use get_user_pages_unlocked() Jan Kara
2013-10-02 14:28 ` [PATCH 23/26] ib: Convert qib_get_user_pages() to get_user_pages_unlocked() Jan Kara
2013-10-02 14:54 ` Marciniszyn, Mike
2013-10-02 15:28 ` Jan Kara
2013-10-02 15:32 ` Marciniszyn, Mike
2013-10-02 15:38 ` Jan Kara
2013-10-04 13:39 ` Marciniszyn, Mike
2013-10-04 13:46 ` Marciniszyn, Mike
2013-10-04 13:44 ` Marciniszyn, Mike
2013-10-04 13:52 ` Marciniszyn, Mike
2013-10-04 18:33 ` Jan Kara
2013-10-07 15:20 ` Marciniszyn, Mike
2013-10-07 15:38 ` Marciniszyn, Mike
2013-10-07 17:26 ` Jan Kara
2013-10-08 19:06 ` Jan Kara [this message]
2013-10-16 21:39 ` Jan Kara
2013-10-02 14:28 ` [PATCH 24/26] ib: Convert qib_user_sdma_pin_pages() to use get_user_pages_unlocked() Jan Kara
2013-10-02 14:28 ` [PATCH 25/26] ib: Convert mthca_map_user_db() to use get_user_pages_fast() Jan Kara
2013-10-02 14:28 ` [PATCH 26/26] aio: Remove useless get_user_pages() call Jan Kara
2013-10-02 16:20 ` [PATCH 0/26] get_user_pages() cleanup Christoph Hellwig
2013-10-02 20:29 ` Jan Kara
2013-10-04 20:31 ` KOSAKI Motohiro
2013-10-04 20:42 ` KOSAKI Motohiro
2013-10-07 21:18 ` Jan Kara
2013-10-08 0:27 ` KOSAKI Motohiro
2013-10-08 6:06 ` Jan Kara
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20131008190604.GB14223@quack.suse.cz \
--to=jack@suse.cz \
--cc=infinipath@intel.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=linux-rdma@vger.kernel.org \
--cc=mike.marciniszyn@intel.com \
--cc=roland@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).