linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jaegeuk Kim <jaegeuk@kernel.org>
To: Chao Yu <yuchao0@huawei.com>
Cc: linux-kernel@vger.kernel.org, linux-f2fs-devel@lists.sourceforge.net
Subject: Re: [f2fs-dev] [PATCH v2] f2fs: fix to keep isolation of atomic write
Date: Mon, 11 Jan 2021 08:32:17 -0800	[thread overview]
Message-ID: <X/x9kTlL8E1Wj4Dd@google.com> (raw)
In-Reply-To: <X/ZAS6oyFiudshe2@google.com>

On 01/06, Jaegeuk Kim wrote:
> On 01/06, Jaegeuk Kim wrote:
> > Hi Chao,
> > 
> > With a quick test, this patch causes down_write failure resulting in blocking
> > process. I didn't dig in the bug so, please check the code again. :P
> 
> nvm. I can see it works now.

Hmm, this gives a huge perf regression when running sqlite. :(
We may need to check the lock coverage. Thoughts?

> 
> > 
> > On 12/30, Chao Yu wrote:
> > > ThreadA					ThreadB
> > > - f2fs_ioc_start_atomic_write
> > > - write
> > > - f2fs_ioc_commit_atomic_write
> > >  - f2fs_commit_inmem_pages
> > >  - f2fs_drop_inmem_pages
> > >  - f2fs_drop_inmem_pages
> > >   - __revoke_inmem_pages
> > > 					- f2fs_vm_page_mkwrite
> > > 					 - set_page_dirty
> > > 					  - tag ATOMIC_WRITTEN_PAGE and add page
> > > 					    to inmem_pages list
> > >   - clear_inode_flag(FI_ATOMIC_FILE)
> > > 					- f2fs_vm_page_mkwrite
> > > 					  - set_page_dirty
> > > 					   - f2fs_update_dirty_page
> > > 					    - f2fs_trace_pid
> > > 					     - tag inmem page private to pid
> > > 					- truncate
> > > 					 - f2fs_invalidate_page
> > > 					 - set page->mapping to NULL
> > > 					  then it will cause panic once we
> > > 					  access page->mapping
> > > 
> > > The root cause is we missed to keep isolation of atomic write in the case
> > > of commit_atomic_write vs mkwrite, let commit_atomic_write helds i_mmap_sem
> > > lock to avoid this issue.
> > > 
> > > Signed-off-by: Chao Yu <yuchao0@huawei.com>
> > > ---
> > > v2:
> > > - use i_mmap_sem to avoid mkwrite racing with below flows:
> > >  * f2fs_ioc_start_atomic_write
> > >  * f2fs_drop_inmem_pages
> > >  * f2fs_commit_inmem_pages
> > > 
> > >  fs/f2fs/file.c    | 3 +++
> > >  fs/f2fs/segment.c | 7 +++++++
> > >  2 files changed, 10 insertions(+)
> > > 
> > > diff --git a/fs/f2fs/file.c b/fs/f2fs/file.c
> > > index 4e6d4b9120a8..a48ec650d691 100644
> > > --- a/fs/f2fs/file.c
> > > +++ b/fs/f2fs/file.c
> > > @@ -2050,6 +2050,7 @@ static int f2fs_ioc_start_atomic_write(struct file *filp)
> > >  		goto out;
> > >  
> > >  	down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
> > > +	down_write(&F2FS_I(inode)->i_mmap_sem);
> > >  
> > >  	/*
> > >  	 * Should wait end_io to count F2FS_WB_CP_DATA correctly by
> > > @@ -2060,6 +2061,7 @@ static int f2fs_ioc_start_atomic_write(struct file *filp)
> > >  			  inode->i_ino, get_dirty_pages(inode));
> > >  	ret = filemap_write_and_wait_range(inode->i_mapping, 0, LLONG_MAX);
> > >  	if (ret) {
> > > +		up_write(&F2FS_I(inode)->i_mmap_sem);
> > >  		up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
> > >  		goto out;
> > >  	}
> > > @@ -2073,6 +2075,7 @@ static int f2fs_ioc_start_atomic_write(struct file *filp)
> > >  	/* add inode in inmem_list first and set atomic_file */
> > >  	set_inode_flag(inode, FI_ATOMIC_FILE);
> > >  	clear_inode_flag(inode, FI_ATOMIC_REVOKE_REQUEST);
> > > +	up_write(&F2FS_I(inode)->i_mmap_sem);
> > >  	up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
> > >  
> > >  	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
> > > diff --git a/fs/f2fs/segment.c b/fs/f2fs/segment.c
> > > index d8570b0359f5..dab870d9faf6 100644
> > > --- a/fs/f2fs/segment.c
> > > +++ b/fs/f2fs/segment.c
> > > @@ -327,6 +327,8 @@ void f2fs_drop_inmem_pages(struct inode *inode)
> > >  	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
> > >  	struct f2fs_inode_info *fi = F2FS_I(inode);
> > >  
> > > +	down_write(&F2FS_I(inode)->i_mmap_sem);
> > > +
> > >  	while (!list_empty(&fi->inmem_pages)) {
> > >  		mutex_lock(&fi->inmem_lock);
> > >  		__revoke_inmem_pages(inode, &fi->inmem_pages,
> > > @@ -344,6 +346,8 @@ void f2fs_drop_inmem_pages(struct inode *inode)
> > >  		sbi->atomic_files--;
> > >  	}
> > >  	spin_unlock(&sbi->inode_lock[ATOMIC_FILE]);
> > > +
> > > +	up_write(&F2FS_I(inode)->i_mmap_sem);
> > >  }
> > >  
> > >  void f2fs_drop_inmem_page(struct inode *inode, struct page *page)
> > > @@ -467,6 +471,7 @@ int f2fs_commit_inmem_pages(struct inode *inode)
> > >  	f2fs_balance_fs(sbi, true);
> > >  
> > >  	down_write(&fi->i_gc_rwsem[WRITE]);
> > > +	down_write(&F2FS_I(inode)->i_mmap_sem);
> > >  
> > >  	f2fs_lock_op(sbi);
> > >  	set_inode_flag(inode, FI_ATOMIC_COMMIT);
> > > @@ -478,6 +483,8 @@ int f2fs_commit_inmem_pages(struct inode *inode)
> > >  	clear_inode_flag(inode, FI_ATOMIC_COMMIT);
> > >  
> > >  	f2fs_unlock_op(sbi);
> > > +
> > > +	up_write(&F2FS_I(inode)->i_mmap_sem);
> > >  	up_write(&fi->i_gc_rwsem[WRITE]);
> > >  
> > >  	return err;
> > > -- 
> > > 2.29.2
> > 
> > 
> > _______________________________________________
> > Linux-f2fs-devel mailing list
> > Linux-f2fs-devel@lists.sourceforge.net
> > https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel
> 
> 
> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2fs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel

  reply	other threads:[~2021-01-11 16:33 UTC|newest]

Thread overview: 13+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-12-30  7:55 [PATCH v2] f2fs: fix to keep isolation of atomic write Chao Yu
2021-01-06 22:28 ` Jaegeuk Kim
2021-01-06 22:57   ` [f2fs-dev] " Jaegeuk Kim
2021-01-11 16:32     ` Jaegeuk Kim [this message]
2021-01-12  2:59       ` Chao Yu
2021-01-12 22:32         ` Jaegeuk Kim
2021-01-13  1:30           ` Chao Yu
2021-01-14 21:53 ` Jaegeuk Kim
2021-01-15  7:59   ` Chao Yu
2021-01-19 19:06     ` Jaegeuk Kim
2021-01-20  1:18       ` Chao Yu
2021-01-28 16:21         ` Jaegeuk Kim
2021-01-29  1:38           ` Chao Yu

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=X/x9kTlL8E1Wj4Dd@google.com \
    --to=jaegeuk@kernel.org \
    --cc=linux-f2fs-devel@lists.sourceforge.net \
    --cc=linux-kernel@vger.kernel.org \
    --cc=yuchao0@huawei.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 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).