mail archive of the barebox mailing list
 help / color / mirror / Atom feed
From: Markus Pargmann <mpa@pengutronix.de>
To: Michael Grzeschik <mgr@pengutronix.de>
Cc: barebox@lists.infradead.org
Subject: Re: [PATCH v2 5/6] state: backend_storage_direct: also use cached data on write
Date: Wed, 29 Jun 2016 07:56:58 +0200	[thread overview]
Message-ID: <4661321.Y6GvhYOqll@adelgunde> (raw)
In-Reply-To: <20160628065532.GE13410@pengutronix.de>


[-- Attachment #1.1: Type: text/plain, Size: 5113 bytes --]

On Tuesday 28 June 2016 08:55:32 Michael Grzeschik wrote:
> On Tue, Jun 28, 2016 at 08:18:47AM +0200, Sascha Hauer wrote:
> > On Fri, Jun 24, 2016 at 12:06:01PM +0200, Markus Pargmann wrote:
> > > From: Michael Grzeschik <m.grzeschik@pengutronix.de>
> > > 
> > > Signed-off-by: Michael Grzeschik <m.grzeschik@pengutronix.de>
> > > Signed-off-by: Markus Pargmann <mpa@pengutronix.de>
> > > ---
> > >  common/state/backend_bucket_direct.c | 55 ++++++++++++++++++++++++++++++++++++
> > >  1 file changed, 55 insertions(+)
> > 
> > What's the problem this patch tries to solve?
> 
> The backend storage is writing the first correct bucket data it finds
> back to each copy. Therefor every bucket got written on every boot. Even
> if the content did not change. The direct backend was missing an cache
> that represent the current data in the buckets. This patch adds the
> cache and improves the bootup time on systems with direct bucket storage
> like eeproms.

I talked with Sascha. He suggested to use a separate cache that
implements caching for any bucket type. That would remove all the
caching code from bucket-direct and bucket-circular and make it simpler.

Best Regards,

Markus

> 
> Michael
> 
> > Sascha
> > 
> > > 
> > > diff --git a/common/state/backend_bucket_direct.c b/common/state/backend_bucket_direct.c
> > > index 08892f001e25..772267f6dc1d 100644
> > > --- a/common/state/backend_bucket_direct.c
> > > +++ b/common/state/backend_bucket_direct.c
> > > @@ -29,6 +29,10 @@ struct state_backend_storage_bucket_direct {
> > >  
> > >  	int fd;
> > >  
> > > +	/* Cached data of the last read/write */
> > > +	u8 *current_data;
> > > +	ssize_t current_data_len;
> > > +
> > >  	struct device_d *dev;
> > >  };
> > >  
> > > @@ -99,6 +103,29 @@ static int state_backend_bucket_direct_read(struct state_backend_storage_bucket
> > >  	return 0;
> > >  }
> > >  
> > > +
> > > +static int state_backend_bucket_direct_fill_cache(
> > > +		struct state_backend_storage_bucket *bucket)
> > > +{
> > > +	struct state_backend_storage_bucket_direct *direct =
> > > +	    get_bucket_direct(bucket);
> > > +	ssize_t read_len;
> > > +	uint8_t *buf;
> > > +	int ret;
> > > +
> > > +	ret = state_backend_bucket_direct_read(bucket, &buf, &read_len);
> > > +	if (ret < 0) {
> > > +		dev_err(direct->dev, "Failed to read from file, %d\n", ret);
> > > +		free(buf);
> > > +		return ret;
> > > +	}
> > > +
> > > +	direct->current_data = buf;
> > > +	direct->current_data_len = read_len;
> > > +
> > > +	return 0;
> > > +}
> > > +
> > >  static int state_backend_bucket_direct_write(struct state_backend_storage_bucket
> > >  					     *bucket, const uint8_t * buf,
> > >  					     ssize_t len)
> > > @@ -111,6 +138,11 @@ static int state_backend_bucket_direct_write(struct state_backend_storage_bucket
> > >  	if (direct->max_size && len > direct->max_size)
> > >  		return -E2BIG;
> > >  
> > > +	/* Nothing in cache? Then read to see what is on the device currently */
> > > +	if (!direct->current_data) {
> > > +		state_backend_bucket_direct_fill_cache(bucket);
> > > +	}
> > > +
> > >  	ret = lseek(direct->fd, direct->offset, SEEK_SET);
> > >  	if (ret < 0) {
> > >  		dev_err(direct->dev, "Failed to seek file, %d\n", ret);
> > > @@ -119,6 +151,26 @@ static int state_backend_bucket_direct_write(struct state_backend_storage_bucket
> > >  
> > >  	meta.magic = direct_magic;
> > >  	meta.written_length = len;
> > > +
> > > +	/*
> > > +	 * If we would write the same data that is currently on the device, we
> > > +	 * can return successfully without writing.
> > > +	 * Note that the cache may still be empty if the storage is empty or
> > > +	 * errors occured.
> > > +	 */
> > > +	if (direct->current_data) {
> > > +		dev_dbg(direct->dev, "Comparing cached data, writing %zd bytes, cached %zd bytes\n",
> > > +			len, direct->current_data_len);
> > > +		if (len == direct->current_data_len &&
> > > +		    !memcmp(direct->current_data, buf, len)) {
> > > +			dev_dbg(direct->dev, "Data already on device, not writing again\n");
> > > +			return ret;
> > > +		} else {
> > > +			free(direct->current_data);
> > > +			direct->current_data = NULL;
> > > +		}
> > > +	}
> > > +
> > >  	ret = write_full(direct->fd, &meta, sizeof(meta));
> > >  	if (ret < 0) {
> > >  		dev_err(direct->dev, "Failed to write metadata to file, %d\n", ret);
> > > @@ -147,6 +199,9 @@ static void state_backend_bucket_direct_free(struct
> > >  	struct state_backend_storage_bucket_direct *direct =
> > >  	    get_bucket_direct(bucket);
> > >  
> > > +	if (direct->current_data)
> > > +		free(direct->current_data);
> > > +
> > >  	close(direct->fd);
> > >  	free(direct);
> > >  }
> > 
> 
> 

-- 
Pengutronix e.K.                           |                             |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |
Peiner Str. 6-8, 31137 Hildesheim, Germany | Phone: +49-5121-206917-0    |
Amtsgericht Hildesheim, HRA 2686           | Fax:   +49-5121-206917-5555 |

[-- Attachment #1.2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 819 bytes --]

[-- Attachment #2: Type: text/plain, Size: 149 bytes --]

_______________________________________________
barebox mailing list
barebox@lists.infradead.org
http://lists.infradead.org/mailman/listinfo/barebox

  reply	other threads:[~2016-06-29  5:57 UTC|newest]

Thread overview: 10+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2016-06-24 10:05 [PATCH v2 0/6] state: Refactor backend Markus Pargmann
2016-06-24 10:05 ` [PATCH v2 1/6] libfile: Change write_full to be have const buf Markus Pargmann
2016-06-24 10:05 ` [PATCH v2 2/6] state: Refactor state framework Markus Pargmann
2016-06-24 10:05 ` [PATCH v2 3/6] docs: Add/Update state documentation Markus Pargmann
2016-06-24 10:06 ` [PATCH v2 4/6] state: also append backend storage stridesize Markus Pargmann
2016-06-24 10:06 ` [PATCH v2 5/6] state: backend_storage_direct: also use cached data on write Markus Pargmann
2016-06-28  6:18   ` Sascha Hauer
2016-06-28  6:55     ` Michael Grzeschik
2016-06-29  5:56       ` Markus Pargmann [this message]
2016-06-24 10:06 ` [PATCH v2 6/6] barebox-state: handle flush errno correctly Markus Pargmann

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=4661321.Y6GvhYOqll@adelgunde \
    --to=mpa@pengutronix.de \
    --cc=barebox@lists.infradead.org \
    --cc=mgr@pengutronix.de \
    /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