linux-kernel - Re: [rfc][patch 3/3] fs: convert ext2,tmpfs to new truncate

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Mon, 6 Jul 2009 19:59:43 +0200
From:	Nick Piggin <npiggin@...e.de>
To:	Christoph Hellwig <hch@...radead.org>
Cc:	linux-fsdevel@...r.kernel.org, Jan Kara <jack@...e.cz>,
	LKML <linux-kernel@...r.kernel.org>, linux-mm@...ck.org
Subject: Re: [rfc][patch 3/3] fs: convert ext2,tmpfs to new truncate

On Mon, Jul 06, 2009 at 01:28:38PM -0400, Christoph Hellwig wrote:
> On Mon, Jul 06, 2009 at 06:56:29PM +0200, Nick Piggin wrote:
> > 
> > Convert filemap_xip.c, buffer.c, and some filesystems to the new truncate
> > convention. Converting generic helpers is using some ugly code (testing
> > for i_op->ftruncate) to distinguish new and old callers... better
> > alternative might be just define a new function for these guys.
> 
> Splitting generic preparations, ext2 and shmem into separate patch would
> be a tad cleaner I think.

Yes it would.

 
> The testing for the new op is pretty ugly, but this should be just a
> transition help, so it's fine to me.

OK good.

> 
> >  		struct page **pagep, void **fsdata)
> >  {
> > +	int ret;
> > +
> >  	*pagep = NULL;
> > -	return __ext2_write_begin(file, mapping, pos, len, flags, pagep,fsdata);
> > +	ret = __ext2_write_begin(file, mapping, pos, len, flags, pagep,fsdata);
> > +	if (ret < 0) {
> > +		loff_t isize = inode->i_size;
> > +		if (pos + len > isize)
> > +			ext2_ftruncate(NULL, 0, inode, isize);
> > +	}
> > +	return ret;
> > +}
> > +
> > +static int ext2_write_end(struct file *file, struct address_space *mapping,
> > +			loff_t pos, unsigned len, unsigned copied,
> > +			struct page *page, void *fsdata)
> > +{
> > +	int ret;
> > +
> > +	ret = generic_write_end(file, mapping, pos, len, copied, page, fsdata);
> > +	if (ret < len) {
> > +		loff_t isize = inode->i_size;
> > +		if (pos + len > isize)
> > +			ext2_ftruncate(NULL, 0, inode, isize);
> > +	}
> > +	return ret;
> >  }
> >  
> >  static int
> > @@ -770,13 +793,22 @@ ext2_nobh_write_begin(struct file *file,
> >  		loff_t pos, unsigned len, unsigned flags,
> >  		struct page **pagep, void **fsdata)
> >  {
> > +	int ret;
> > +
> >  	/*
> >  	 * Dir-in-pagecache still uses ext2_write_begin. Would have to rework
> >  	 * directory handling code to pass around offsets rather than struct
> >  	 * pages in order to make this work easily.
> >  	 */
> > -	return nobh_write_begin(file, mapping, pos, len, flags, pagep, fsdata,
> > +	ret = nobh_write_begin(file, mapping, pos, len, flags, pagep, fsdata,
> >  							ext2_get_block);
> > +	if (ret < 0) {
> > +		loff_t isize;
> > +		isize = i_size_read(inode);
> > +		if (pos + len > isize)
> > +			ext2_ftruncate(NULL, 0, inode, isize);
> > +	}
> > +	return ret;
> >  }
> >  
> >  static int ext2_nobh_writepage(struct page *page,
> > @@ -796,9 +828,15 @@ ext2_direct_IO(int rw, struct kiocb *ioc
> >  {
> >  	struct file *file = iocb->ki_filp;
> >  	struct inode *inode = file->f_mapping->host;
> > +	ssize_t ret;
> >  
> > -	return blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
> > +	ret = blockdev_direct_IO(rw, iocb, inode, inode->i_sb->s_bdev, iov,
> >  				offset, nr_segs, ext2_get_block, NULL);
> > +	if (ret < 0 && (rw & WRITE)) {
> > +		loff_t isize = i_size_read(inode);
> > +		ext2_ftruncate(NULL, 0, inode, isize);
> 
> These calls don't actually have i_alloc_mutex anymore, do they?

No that's right. But ext2 already had been calling into vmtruncate
without i_alloc_sem (via vmtruncate in write(2) path for trimming
blocks)... hmm, maybe that was a bug itself, though (OTOH, maybe
it doesn't matter because direct IO should not get to blocks
past i_size).


> >  {
> > -	shmem_truncate_range(inode, inode->i_size, (loff_t)-1);
> > +	loff_t oldsize;
> > +	int error;
> > +
> > +	error = inode_truncate_ok(inode, offset);
> > +	if (error)
> > +		return error;
> > +	oldsize = inode->i_size;
> > +	i_size_write(inode, offset);
> > +	truncate_pagecache(inode, oldsize, offset);
> > +	shmem_truncate_range(inode, offset, (loff_t)-1);
> > +
> > +	return error;
> >  }
> 
> Just make this
> 
> 	error = simple_ftruncate(...);
> 	if (!error)
> 		shmem_truncate_range(inode, offset, -1);
> 	return error;

Yes that's better.

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