linux-kernel - Re: [PATCH v6 4/5] dax: for truncate/hole-punch, do zeroing through the driver if possible

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]

Message-ID: <1462988808.29294.26.camel@intel.com>
Date:	Wed, 11 May 2016 17:47:00 +0000
From:	"Verma, Vishal L" <vishal.l.verma@...el.com>
To:	"jack@...e.cz" <jack@...e.cz>
CC:	"linux-kernel@...r.kernel.org" <linux-kernel@...r.kernel.org>,
	"linux-block@...r.kernel.org" <linux-block@...r.kernel.org>,
	"hch@...radead.org" <hch@...radead.org>,
	"xfs@....sgi.com" <xfs@....sgi.com>,
	"jmoyer@...hat.com" <jmoyer@...hat.com>,
	"linux-mm@...ck.org" <linux-mm@...ck.org>,
	"Williams, Dan J" <dan.j.williams@...el.com>,
	"axboe@...com" <axboe@...com>,
	"akpm@...ux-foundation.org" <akpm@...ux-foundation.org>,
	"linux-nvdimm@...ts.01.org" <linux-nvdimm@...ts.01.org>,
	"linux-fsdevel@...r.kernel.org" <linux-fsdevel@...r.kernel.org>,
	"ross.zwisler@...ux.intel.com" <ross.zwisler@...ux.intel.com>,
	"linux-ext4@...r.kernel.org" <linux-ext4@...r.kernel.org>,
	"boaz@...xistor.com" <boaz@...xistor.com>,
	"david@...morbit.com" <david@...morbit.com>
Subject: Re: [PATCH v6 4/5] dax: for truncate/hole-punch, do zeroing through
 the driver if possible

On Wed, 2016-05-11 at 10:15 +0200, Jan Kara wrote:
> On Tue 10-05-16 12:49:15, Vishal Verma wrote:
> > 
> > In the truncate or hole-punch path in dax, we clear out sub-page
> > ranges.
> > If these sub-page ranges are sector aligned and sized, we can do the
> > zeroing through the driver instead so that error-clearing is handled
> > automatically.
> > 
> > For sub-sector ranges, we still have to rely on clear_pmem and have
> > the
> > possibility of tripping over errors.
> > 
> > Cc: Dan Williams <dan.j.williams@...el.com>
> > Cc: Ross Zwisler <ross.zwisler@...ux.intel.com>
> > Cc: Jeff Moyer <jmoyer@...hat.com>
> > Cc: Christoph Hellwig <hch@...radead.org>
> > Cc: Dave Chinner <david@...morbit.com>
> > Cc: Jan Kara <jack@...e.cz>
> > Reviewed-by: Christoph Hellwig <hch@....de>
> > Signed-off-by: Vishal Verma <vishal.l.verma@...el.com>
> ...
> 
> > 
> > +static bool dax_range_is_aligned(struct block_device *bdev,
> > +				 struct blk_dax_ctl *dax, unsigned
> > int offset,
> > +				 unsigned int length)
> > +{
> > +	unsigned short sector_size = bdev_logical_block_size(bdev);
> > +
> > +	if (!IS_ALIGNED(((u64)dax->addr + offset), sector_size))
> One more question: 'dax' is initialized in dax_zero_page_range() and
> dax->addr is going to be always NULL here. So either you forgot to
> call
> dax_map_atomic() to get the addr or the use of dax->addr is just bogus
> (which is what I currently believe since I see no way how the address
> could
> be unaligned with the sector_size)...
> 

Good catch, and you're right. I don't think I actually even want to use
dax->addr for the alignment check here - I want to check if we're
aligned to the block device sector. I'm thinking something like:

	if (!IS_ALIGNED(offset, sector_size))

Technically we want to check if sector * sector_size + offset is
aligned, but the first part of that is already a sector :)