linux-kernel - Re: [PATCH v6 00/10] block atomic writes

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: <ZhcxDOWy9j8mzs_u@bombadil.infradead.org>
Date: Wed, 10 Apr 2024 17:38:36 -0700
From: Luis Chamberlain <mcgrof@...nel.org>
To: Hannes Reinecke <hare@...e.de>
Cc: Matthew Wilcox <willy@...radead.org>,
	John Garry <john.g.garry@...cle.com>,
	Pankaj Raghav <p.raghav@...sung.com>,
	Daniel Gomez <da.gomez@...sung.com>,
	Javier González <javier.gonz@...sung.com>,
	axboe@...nel.dk, kbusch@...nel.org, hch@....de, sagi@...mberg.me,
	jejb@...ux.ibm.com, martin.petersen@...cle.com, djwong@...nel.org,
	viro@...iv.linux.org.uk, brauner@...nel.org, dchinner@...hat.com,
	jack@...e.cz, linux-block@...r.kernel.org,
	linux-kernel@...r.kernel.org, linux-nvme@...ts.infradead.org,
	linux-fsdevel@...r.kernel.org, tytso@....edu, jbongio@...gle.com,
	linux-scsi@...r.kernel.org, ojaswin@...ux.ibm.com,
	linux-aio@...ck.org, linux-btrfs@...r.kernel.org,
	io-uring@...r.kernel.org, nilay@...ux.ibm.com,
	ritesh.list@...il.com
Subject: Re: [PATCH v6 00/10] block atomic writes

On Wed, Apr 10, 2024 at 08:20:37AM +0200, Hannes Reinecke wrote:
> On 4/10/24 06:05, Matthew Wilcox wrote:
> > On Mon, Apr 08, 2024 at 10:50:47AM -0700, Luis Chamberlain wrote:
> > > On Fri, Apr 05, 2024 at 11:06:00AM +0100, John Garry wrote:
> > > > On 04/04/2024 17:48, Matthew Wilcox wrote:
> > > > > > > The thing is that there's no requirement for an interface as complex as
> > > > > > > the one you're proposing here.  I've talked to a few database people
> > > > > > > and all they want is to increase the untorn write boundary from "one
> > > > > > > disc block" to one database block, typically 8kB or 16kB.
> > > > > > > 
> > > > > > > So they would be quite happy with a much simpler interface where they
> > > > > > > set the inode block size at inode creation time,
> > > > > > We want to support untorn writes for bdev file operations - how can we set
> > > > > > the inode block size there? Currently it is based on logical block size.
> > > > > ioctl(BLKBSZSET), I guess?  That currently limits to PAGE_SIZE, but I
> > > > > think we can remove that limitation with the bs>PS patches.
> > > 
> > > I can say a bit more on this, as I explored that. Essentially Matthew,
> > > yes, I got that to work but it requires a set of different patches. We have
> > > what we tried and then based on feedback from Chinner we have a
> > > direction on what to try next. The last effort on that front was having the
> > > iomap aops for bdev be used and lifting the PAGE_SIZE limit up to the
> > > page cache limits. The crux on that front was that we end requiring
> > > disabling BUFFER_HEAD and that is pretty limitting, so my old
> > > implementation had dynamic aops so to let us use the buffer-head aops
> > > only when using filesystems which require it and use iomap aops
> > > otherwise. But as Chinner noted we learned through the DAX experience
> > > that's not a route we want to again try, so the real solution is to
> > > extend iomap bdev aops code with buffer-head compatibility.
> > 
> > Have you tried just using the buffer_head code?  I think you heard bad
> > advice at last LSFMM.  Since then I've landed a bunch of patches which
> > remove PAGE_SIZE assumptions throughout the buffer_head code, and while
> > I haven't tried it, it might work.  And it might be easier to make work
> > than adding more BH hacks to the iomap code.
> > 
> > A quick audit for problems ...
> > 
> > __getblk_slow:
> >         if (unlikely(size & (bdev_logical_block_size(bdev)-1) ||
> >                          (size < 512 || size > PAGE_SIZE))) {
> > 
> > cont_expand_zero (not used by bdev code)
> > cont_write_begin (ditto)
> > 
> > That's all I spot from a quick grep for PAGE, offset_in_page() and kmap.
> > 
> > You can't do a lot of buffer_heads per folio, because you'll overrun
> >          struct buffer_head *bh, *head, *arr[MAX_BUF_PER_PAGE];
> > in block_read_full_folio(), but you can certainly do _one_ buffer_head
> > per folio, and that's all you need for bs>PS.
> > 
> Indeed; I got a patch here to just restart the submission loop if one
> reaches the end of the array. But maybe submitting one bh at a time and
> using plugging should achieve that same thing. Let's see.

That's great to hear, what about a target filesystem? Without a
buffer-head filesystem to test I'm not sure we'd get enough test
coverage.

The block device cache isn't exaclty a great filesystem target to test
correctness.

> > > I suspect this is a use case where perhaps the max folio order could be
> > > set for the bdev in the future, the logical block size the min order,
> > > and max order the large atomic.
> > 
> > No, that's not what we want to do at all!  Minimum writeback size needs
> > to be the atomic size, otherwise we have to keep track of which writes
> > are atomic and which ones aren't.  So, just set the logical block size
> > to the atomic size, and we're done.
> > 
> +1. My thoughts all along.

Oh, hrm yes, but let's test it out then...

  Luis