From: Matthew Wilcox <willy@linux.intel.com>
To: Yigal Korman <yigal@plexistor.com>
Cc: Matthew Wilcox <matthew.r.wilcox@intel.com>,
linux-fsdevel@vger.kernel.org,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
willy@linux.intel.com, Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH v12 04/20] mm: Allow page fault handlers to perform the COW
Date: Thu, 5 Feb 2015 16:39:39 -0500 [thread overview]
Message-ID: <20150205213939.GA3364@wil.cx> (raw)
In-Reply-To: <CACTTzNbZ2K824aoPqXe4Q8WDRuc72ch5+B9J3GZQ2Z4Kwia56A@mail.gmail.com>
On Thu, Feb 05, 2015 at 11:16:53AM +0200, Yigal Korman wrote:
> I have a question on a related issue (I think).
> I've noticed that for pfn-only mappings (VM_FAULT_NOPAGE)
> do_shared_fault only maps the pfn with r/o permissions.
> So if I use DAX to write the mmap()-ed pfn I get two faults - first
> handled by do_shared_fault and then again for making it r/w in
> do_wp_page.
> Is this simply a missing optimization like was done here with the
> cow_page? or am I missing something?
I have also noticed this behaviour. I tracked down why it's happening:
DAX calls:
error = vm_insert_mixed(vma, vaddr, pfn);
which calls:
return insert_pfn(vma, addr, pfn, vma->vm_page_prot);
If you insert some debugging, you'll notice here that vm_page_prot does
not include PROT_WRITE.
That got cleared during mmap_region() where it does:
if (vma_wants_writenotify(vma)) {
pgprot_t pprot = vma->vm_page_prot;
...
vma->vm_page_prot = vm_get_page_prot(vm_flags & ~VM_SHARED);
And why do we want writenotify (according to the VM)? Because we have:
/* The backer wishes to know when pages are first written to? */
if (vma->vm_ops && vma->vm_ops->page_mkwrite)
return 1;
We don't really want to be notified on a first write; we want the page to be
inserted write-enabled. But in the case where we've covered a hole with a
read-only zero page, we need to be notified so we can allocate a page of
storage.
So, how to fix? We could adjust vm_page_prot to include PROT_WRITE.
I think that should work, since we'll only insert zeroed pages for read
faults, and so the maybe_mkwrite() won't be called in do_set_pte().
I'm just not entirely sure where to set it. Perhaps a MM person could
make a helpful suggestion?
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
WARNING: multiple messages have this Message-ID (diff)
From: Matthew Wilcox <willy@linux.intel.com>
To: Yigal Korman <yigal@plexistor.com>
Cc: Matthew Wilcox <matthew.r.wilcox@intel.com>,
linux-fsdevel@vger.kernel.org,
"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
"linux-mm@kvack.org" <linux-mm@kvack.org>,
willy@linux.intel.com, Andrew Morton <akpm@linux-foundation.org>
Subject: Re: [PATCH v12 04/20] mm: Allow page fault handlers to perform the COW
Date: Thu, 5 Feb 2015 16:39:39 -0500 [thread overview]
Message-ID: <20150205213939.GA3364@wil.cx> (raw)
In-Reply-To: <CACTTzNbZ2K824aoPqXe4Q8WDRuc72ch5+B9J3GZQ2Z4Kwia56A@mail.gmail.com>
On Thu, Feb 05, 2015 at 11:16:53AM +0200, Yigal Korman wrote:
> I have a question on a related issue (I think).
> I've noticed that for pfn-only mappings (VM_FAULT_NOPAGE)
> do_shared_fault only maps the pfn with r/o permissions.
> So if I use DAX to write the mmap()-ed pfn I get two faults - first
> handled by do_shared_fault and then again for making it r/w in
> do_wp_page.
> Is this simply a missing optimization like was done here with the
> cow_page? or am I missing something?
I have also noticed this behaviour. I tracked down why it's happening:
DAX calls:
error = vm_insert_mixed(vma, vaddr, pfn);
which calls:
return insert_pfn(vma, addr, pfn, vma->vm_page_prot);
If you insert some debugging, you'll notice here that vm_page_prot does
not include PROT_WRITE.
That got cleared during mmap_region() where it does:
if (vma_wants_writenotify(vma)) {
pgprot_t pprot = vma->vm_page_prot;
...
vma->vm_page_prot = vm_get_page_prot(vm_flags & ~VM_SHARED);
And why do we want writenotify (according to the VM)? Because we have:
/* The backer wishes to know when pages are first written to? */
if (vma->vm_ops && vma->vm_ops->page_mkwrite)
return 1;
We don't really want to be notified on a first write; we want the page to be
inserted write-enabled. But in the case where we've covered a hole with a
read-only zero page, we need to be notified so we can allocate a page of
storage.
So, how to fix? We could adjust vm_page_prot to include PROT_WRITE.
I think that should work, since we'll only insert zeroed pages for read
faults, and so the maybe_mkwrite() won't be called in do_set_pte().
I'm just not entirely sure where to set it. Perhaps a MM person could
make a helpful suggestion?
next prev parent reply other threads:[~2015-02-05 21:39 UTC|newest]
Thread overview: 119+ messages / expand[flat|nested] mbox.gz Atom feed top
2014-10-24 21:20 [PATCH v12 00/20] DAX: Page cache bypass for filesystems on memory storage Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 01/20] axonram: Fix bug in direct_access Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 02/20] block: Change direct_access calling convention Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 03/20] mm: Fix XIP fault vs truncate race Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:09 ` Andrew Morton
2015-01-12 23:09 ` Andrew Morton
2015-01-13 18:50 ` Matthew Wilcox
2015-01-13 18:50 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 04/20] mm: Allow page fault handlers to perform the COW Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:09 ` Andrew Morton
2015-01-12 23:09 ` Andrew Morton
2015-01-13 18:58 ` Matthew Wilcox
2015-01-13 18:58 ` Matthew Wilcox
2015-02-05 9:16 ` Yigal Korman
2015-02-05 9:16 ` Yigal Korman
2015-02-05 21:39 ` Matthew Wilcox [this message]
2015-02-05 21:39 ` Matthew Wilcox
2015-02-08 11:48 ` Yigal Korman
2015-02-08 11:48 ` Yigal Korman
2014-10-24 21:20 ` [PATCH v12 05/20] vfs,ext2: Introduce IS_DAX(inode) Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 06/20] dax,ext2: Replace XIP read and write with DAX I/O Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:09 ` Andrew Morton
2015-01-12 23:09 ` Andrew Morton
2015-01-13 20:59 ` Matthew Wilcox
2015-01-13 20:59 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 07/20] dax,ext2: Replace ext2_clear_xip_target with dax_clear_blocks Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:09 ` Andrew Morton
2015-01-12 23:09 ` Andrew Morton
2015-01-13 21:39 ` Matthew Wilcox
2015-01-13 21:39 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 08/20] dax,ext2: Replace the XIP page fault handler with the DAX page fault handler Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:09 ` Andrew Morton
2015-01-12 23:09 ` Andrew Morton
2015-01-13 21:53 ` Matthew Wilcox
2015-01-13 21:53 ` Matthew Wilcox
2015-01-13 22:47 ` Andrew Morton
2015-01-13 22:47 ` Andrew Morton
2014-10-24 21:20 ` [PATCH v12 09/20] dax,ext2: Replace xip_truncate_page with dax_truncate_page Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:09 ` Andrew Morton
2015-01-12 23:09 ` Andrew Morton
2015-01-13 21:55 ` Matthew Wilcox
2015-01-13 21:55 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 10/20] dax: Replace XIP documentation with DAX documentation Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:10 ` Andrew Morton
2015-01-12 23:10 ` Andrew Morton
2016-01-21 18:38 ` Jared Hulbert
2016-01-21 18:38 ` Jared Hulbert
2016-01-22 13:07 ` Wilcox, Matthew R
2016-01-22 13:48 ` Chris Brandt
2016-01-22 14:39 ` Matthew Wilcox
2016-01-22 14:39 ` Matthew Wilcox
2016-01-24 9:03 ` Jared Hulbert
2016-01-24 9:03 ` Jared Hulbert
2016-01-25 16:52 ` Matthew Wilcox
2016-01-25 16:52 ` Matthew Wilcox
2016-01-25 21:18 ` Jared Hulbert
2016-01-25 21:18 ` Jared Hulbert
2016-01-27 19:51 ` Jared Hulbert
2016-01-27 19:51 ` Jared Hulbert
2014-10-24 21:20 ` [PATCH v12 11/20] vfs: Remove get_xip_mem Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 12/20] ext2: Remove ext2_xip_verify_sb() Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 13/20] ext2: Remove ext2_use_xip Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 14/20] ext2: Remove xip.c and xip.h Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 15/20] vfs,ext2: Remove CONFIG_EXT2_FS_XIP and rename CONFIG_FS_XIP to CONFIG_FS_DAX Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 16/20] ext2: Remove ext2_aops_xip Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 17/20] ext2: Get rid of most mentions of XIP in ext2 Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 18/20] dax: Add dax_zero_page_range Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2015-01-12 23:10 ` Andrew Morton
2015-01-12 23:10 ` Andrew Morton
2015-01-12 23:20 ` Ross Zwisler
2015-01-12 23:20 ` Ross Zwisler
2014-10-24 21:20 ` [PATCH v12 19/20] ext4: Add DAX functionality Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-10-24 21:20 ` [PATCH v12 20/20] brd: Rename XIP to DAX Matthew Wilcox
2014-10-24 21:20 ` Matthew Wilcox
2014-12-10 14:03 ` [PATCH v12 00/20] DAX: Page cache bypass for filesystems on memory storage Christoph Hellwig
2014-12-10 14:03 ` Christoph Hellwig
2014-12-10 14:12 ` Matthew Wilcox
2014-12-10 14:12 ` Matthew Wilcox
2014-12-10 14:28 ` Jeff Moyer
2014-12-10 14:28 ` Jeff Moyer
2014-12-10 20:53 ` Dave Chinner
2014-12-10 20:53 ` Dave Chinner
2015-01-05 18:41 ` Christoph Hellwig
2015-01-05 18:41 ` Christoph Hellwig
2015-01-06 8:47 ` Andrew Morton
2015-01-06 8:47 ` Andrew Morton
2015-01-08 11:49 ` pread2/ pwrite2 Christoph Hellwig
2015-01-08 11:49 ` Christoph Hellwig
2015-01-09 19:30 ` Steve French
2015-01-09 19:30 ` Steve French
2015-01-08 16:27 ` [PATCH v12 00/20] DAX: Page cache bypass for filesystems on memory storage Milosz Tanski
2015-01-08 16:28 ` Milosz Tanski
2015-01-08 16:28 ` Milosz Tanski
2015-01-08 17:36 ` Jeremy Allison
2015-01-08 17:36 ` Jeremy Allison
2015-01-12 14:47 ` Matthew Wilcox
2015-01-12 14:47 ` Matthew Wilcox
2015-01-12 23:09 ` Andrew Morton
2015-01-12 23:09 ` Andrew Morton
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=20150205213939.GA3364@wil.cx \
--to=willy@linux.intel.com \
--cc=akpm@linux-foundation.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-mm@kvack.org \
--cc=matthew.r.wilcox@intel.com \
--cc=yigal@plexistor.com \
/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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.