CEPH filesystem development
 help / color / mirror / Atom feed
From: Alex Elder <elder@inktank.com>
To: Sage Weil <sage@inktank.com>
Cc: ceph-devel@vger.kernel.org
Subject: Re: [PATCH, v2 06/11] rbd: add read_only rbd map option
Date: Fri, 07 Sep 2012 15:36:34 -0500	[thread overview]
Message-ID: <504A5AD2.1090107@inktank.com> (raw)
In-Reply-To: <alpine.DEB.2.00.1209070844190.9861@cobra.newdream.net>

On 09/07/2012 10:45 AM, Sage Weil wrote:
> On Thu, 6 Sep 2012, Alex Elder wrote:
>> Add the ability to map an rbd image read-only, by specifying either
>> "read_only" or "ro" as an option on the rbd "command line."  Also
>> allow the inverse to be explicitly specified using "read_write" or
>> "rw".
>>
>> Signed-off-by: Alex Elder <elder@inktank.com>
>> ---
>> v2: - Made the read_only/ro flag explicitly make the mapping
>>       read-only rather than having it toggle.
>>     - Added read_write/rw flags to allow a writable mapping to
>>       be explicitly requested.  Snapshot mappings are silently
>>       mapped read-only (even if read_write was requested).
> 
> If rw is the default, and rw is silently ignored if the image itself is a 
> snapshot, what is the value in rw?  Would it make more sense to fail with 
> EROFS if you specify rw on a snapshot (or otherwise read-only image)?

I explicitly called this out because what I wanted to do was simply too
hard.

Namely:
- default for base image map would be read-write
- default for snapshot map would be read-only
- attempt to map snapshot read-write would be met with a "ignoring"
  warning

But the way the argument parsing is structured, you never have all the
information you want at the right time to do the above.  So I punted
and just decided attempts to read-write map a snapshot would be silently
ignored.

I'll take a look at simply failing on a read-write mount of a snapshot
and if it's workable I'll do that.

					-Alex

PS  As I told to Josh separately, one of the reasons I even put in the
    read-only option in the first place is to have a reason to preserve
    the argument parsing infrastructure.  And that's why it gets added
    at this position in the series--so that code remains after the next
    patch, which removes the only other option (notify_timeout).  But it
    also seemed like a (marginally) useful option either way.

> sage
> 
>>
>>  drivers/block/rbd.c |   28 ++++++++++++++++++++++++----
>>  1 file changed, 24 insertions(+), 4 deletions(-)
>>
>> Index: b/drivers/block/rbd.c
>> ===================================================================
>> --- a/drivers/block/rbd.c
>> +++ b/drivers/block/rbd.c
>> @@ -69,7 +69,8 @@
>>  #define DEV_NAME_LEN		32
>>  #define MAX_INT_FORMAT_WIDTH	((5 * sizeof (int)) / 2 + 1)
>>
>> -#define RBD_NOTIFY_TIMEOUT_DEFAULT 10
>> +#define RBD_NOTIFY_TIMEOUT_DEFAULT	10
>> +#define RBD_READ_ONLY_DEFAULT		false
>>
>>  /*
>>   * block device image metadata (in-memory version)
>> @@ -91,6 +92,7 @@ struct rbd_image_header {
>>
>>  struct rbd_options {
>>  	int	notify_timeout;
>> +	bool	read_only;
>>  };
>>
>>  /*
>> @@ -176,7 +178,7 @@ struct rbd_device {
>>  	u64                     snap_id;	/* current snapshot id */
>>  	/* whether the snap_id this device reads from still exists */
>>  	bool                    snap_exists;
>> -	int                     read_only;
>> +	bool			read_only;
>>
>>  	struct list_head	node;
>>
>> @@ -351,12 +353,21 @@ enum {
>>  	/* int args above */
>>  	Opt_last_string,
>>  	/* string args above */
>> +	Opt_read_only,
>> +	Opt_read_write,
>> +	/* Boolean args above */
>> +	Opt_last_bool,
>>  };
>>
>>  static match_table_t rbd_opts_tokens = {
>>  	{Opt_notify_timeout, "notify_timeout=%d"},
>>  	/* int args above */
>>  	/* string args above */
>> +	{Opt_read_only, "read_only"},
>> +	{Opt_read_only, "ro"},		/* Alternate spelling */
>> +	{Opt_read_write, "read_write"},
>> +	{Opt_read_write, "rw"},		/* Alternate spelling */
>> +	/* Boolean args above */
>>  	{-1, NULL}
>>  };
>>
>> @@ -381,6 +392,8 @@ static int parse_rbd_opts_token(char *c,
>>  	} else if (token > Opt_last_int && token < Opt_last_string) {
>>  		dout("got string token %d val %s\n", token,
>>  		     argstr[0].from);
>> +	} else if (token > Opt_last_string && token < Opt_last_bool) {
>> +		dout("got Boolean token %d\n", token);
>>  	} else {
>>  		dout("got token %d\n", token);
>>  	}
>> @@ -389,6 +402,12 @@ static int parse_rbd_opts_token(char *c,
>>  	case Opt_notify_timeout:
>>  		rbd_opts->notify_timeout = intval;
>>  		break;
>> +	case Opt_read_only:
>> +		rbd_opts->read_only = true;
>> +		break;
>> +	case Opt_read_write:
>> +		rbd_opts->read_only = false;
>> +		break;
>>  	default:
>>  		BUG_ON(token);
>>  	}
>> @@ -407,6 +426,7 @@ static int rbd_get_client(struct rbd_dev
>>  	struct rbd_client *rbdc;
>>
>>  	rbd_opts->notify_timeout = RBD_NOTIFY_TIMEOUT_DEFAULT;
>> +	rbd_opts->read_only = RBD_READ_ONLY_DEFAULT;
>>
>>  	ceph_opts = ceph_parse_options(options, mon_addr,
>>  					mon_addr + mon_addr_len,
>> @@ -620,7 +640,7 @@ static int rbd_header_set_snap(struct rb
>>  		    sizeof (RBD_SNAP_HEAD_NAME))) {
>>  		rbd_dev->snap_id = CEPH_NOSNAP;
>>  		rbd_dev->snap_exists = false;
>> -		rbd_dev->read_only = 0;
>> +		rbd_dev->read_only = rbd_dev->rbd_opts.read_only;
>>  		if (size)
>>  			*size = rbd_dev->header.image_size;
>>  	} else {
>> @@ -632,7 +652,7 @@ static int rbd_header_set_snap(struct rb
>>  			goto done;
>>  		rbd_dev->snap_id = snap_id;
>>  		rbd_dev->snap_exists = true;
>> -		rbd_dev->read_only = 1;
>> +		rbd_dev->read_only = true;	/* No choice for snapshots */
>>  	}
>>
>>  	ret = 0;
>> --
>> To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>>
> 
> 


  reply	other threads:[~2012-09-07 20:36 UTC|newest]

Thread overview: 35+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-08-24 16:26 [PATCH 00/11] rbd: another set of patches Alex Elder
2012-08-24 16:32 ` [PATCH 01/11] rbd: handle locking inside __rbd_client_find() Alex Elder
2012-08-30 16:09   ` Yehuda Sadeh
2012-08-24 16:32 ` [PATCH 02/11] rbd: don't over-allocate space for object prefix Alex Elder
2012-08-30 16:18   ` Yehuda Sadeh
2012-08-24 16:33 ` [PATCH 03/11] rbd: kill incore snap_names_len Alex Elder
2012-08-30 16:24   ` Yehuda Sadeh
2012-08-30 16:41     ` Alex Elder
2012-09-06 15:36   ` [PATCH, v2 " Alex Elder
2012-09-07 21:22     ` Yehuda Sadeh
2012-08-24 16:33 ` [PATCH 04/11] rbd: more cleanup in rbd_header_from_disk() Alex Elder
2012-08-30 16:48   ` Yehuda Sadeh
2012-08-24 16:33 ` [PATCH 05/11] rbd: move rbd_opts to struct rbd_device Alex Elder
2012-08-30 17:07   ` Yehuda Sadeh
2012-09-06 14:21     ` Alex Elder
2012-09-07 21:40       ` Yehuda Sadeh
2012-08-24 16:34 ` [PATCH 06/11] rbd: add read_only rbd map option Alex Elder
2012-08-30 17:29   ` Yehuda Sadeh
2012-08-30 17:39     ` Alex Elder
2012-09-06 15:36   ` [PATCH, v2 " Alex Elder
2012-09-07 15:45     ` Sage Weil
2012-09-07 20:36       ` Alex Elder [this message]
2012-09-07 21:26     ` Yehuda Sadeh Weinraub
2012-08-24 16:34 ` [PATCH 07/11] rbd: kill notify_timeout option Alex Elder
2012-08-30 17:31   ` Yehuda Sadeh
2012-08-24 16:35 ` [PATCH 08/11] rbd: bio_chain_clone() cleanups Alex Elder
2012-08-30 17:40   ` Yehuda Sadeh
2012-08-24 16:35 ` [PATCH 09/11] rbd: drop needless test in rbd_rq_fn() Alex Elder
2012-08-30 17:41   ` Yehuda Sadeh
2012-08-24 16:36 ` [PATCH 10/11] rbd: check for overflow in rbd_get_num_segments() Alex Elder
2012-08-30 17:50   ` Yehuda Sadeh
2012-08-24 16:36 ` [PATCH 11/11] rbd: split up rbd_get_segment() Alex Elder
2012-08-30 18:03   ` Yehuda Sadeh
2012-08-30 12:32 ` [PATCH 00/11] rbd: another set of patches Alex Elder
2012-09-06 15:34 ` Alex Elder

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=504A5AD2.1090107@inktank.com \
    --to=elder@inktank.com \
    --cc=ceph-devel@vger.kernel.org \
    --cc=sage@inktank.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