qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Kevin Wolf <kwolf@redhat.com>
To: Stefano Garzarella <sgarzare@redhat.com>
Cc: qemu-devel@nongnu.org, qemu-block@nongnu.org,
	Max Reitz <mreitz@redhat.com>, Josh Durgin <jdurgin@redhat.com>
Subject: Re: [Qemu-devel] [PATCH RFC 1/1] block/rbd: increase dynamically the image size
Date: Mon, 29 Apr 2019 12:25:10 +0200	[thread overview]
Message-ID: <20190429102510.GD8492@localhost.localdomain> (raw)
In-Reply-To: <20190411105025.97397-2-sgarzare@redhat.com>

Am 11.04.2019 um 12:50 hat Stefano Garzarella geschrieben:
> RBD APIs don't allow us to write more than the size set with rbd_create()
> or rbd_resize().
> In order to support growing images (eg. qcow2), we resize the image
> before RW operations that exceed the current size.
> 
> Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1171007
> Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
> ---
>  block/rbd.c | 25 +++++++++++++++++++++++++
>  1 file changed, 25 insertions(+)
> 
> diff --git a/block/rbd.c b/block/rbd.c
> index 0c549c9935..228658e20a 100644
> --- a/block/rbd.c
> +++ b/block/rbd.c
> @@ -102,6 +102,7 @@ typedef struct BDRVRBDState {
>      rbd_image_t image;
>      char *image_name;
>      char *snap;
> +    uint64_t image_size;
>  } BDRVRBDState;

Can't we use bs->total_sectors instead of adding a new image_size field?

>  static int qemu_rbd_connect(rados_t *cluster, rados_ioctx_t *io_ctx,
> @@ -777,6 +778,14 @@ static int qemu_rbd_open(BlockDriverState *bs, QDict *options, int flags,
>          goto failed_open;
>      }
>  
> +    r = rbd_get_size(s->image, &s->image_size);
> +    if (r < 0) {
> +        error_setg_errno(errp, -r, "error reading image size from %s",
> +                         s->image_name);
> +        rbd_close(s->image);
> +        goto failed_open;
> +    }
> +
>      /* If we are using an rbd snapshot, we must be r/o, otherwise
>       * leave as-is */
>      if (s->snap != NULL) {
> @@ -921,6 +930,20 @@ static BlockAIOCB *rbd_start_aio(BlockDriverState *bs,
>          rcb->buf = acb->bounce;
>      }
>  
> +    /*
> +     * RBD APIs don't allow us to write more than actual size, so in order
> +     * to support growing images, we resize the image before RW operations
> +     * that exceed the current size.
> +     */
> +    if (s->image_size < off + size) {
> +        r = rbd_resize(s->image, off + size);
> +        if (r < 0) {
> +            goto failed;
> +        }
> +
> +        s->image_size = off + size;
> +    }

This doesn't check the request type, so it's actually not limited to RW
operations, but even reads will try to resize the image. This is at
least surprising. For regular files, file-posix extends the file for
write requests, but for reads it returns a zeroed buffer without
actually changing the file size.

Kevin

WARNING: multiple messages have this Message-ID (diff)
From: Kevin Wolf <kwolf@redhat.com>
To: Stefano Garzarella <sgarzare@redhat.com>
Cc: Josh Durgin <jdurgin@redhat.com>,
	qemu-devel@nongnu.org, qemu-block@nongnu.org,
	Max Reitz <mreitz@redhat.com>
Subject: Re: [Qemu-devel] [PATCH RFC 1/1] block/rbd: increase dynamically the image size
Date: Mon, 29 Apr 2019 12:25:10 +0200	[thread overview]
Message-ID: <20190429102510.GD8492@localhost.localdomain> (raw)
Message-ID: <20190429102510.Uct8p0Xc680K9mqTMgaSj4xG0EVtQVozsGH21WwYc_o@z> (raw)
In-Reply-To: <20190411105025.97397-2-sgarzare@redhat.com>

Am 11.04.2019 um 12:50 hat Stefano Garzarella geschrieben:
> RBD APIs don't allow us to write more than the size set with rbd_create()
> or rbd_resize().
> In order to support growing images (eg. qcow2), we resize the image
> before RW operations that exceed the current size.
> 
> Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=1171007
> Signed-off-by: Stefano Garzarella <sgarzare@redhat.com>
> ---
>  block/rbd.c | 25 +++++++++++++++++++++++++
>  1 file changed, 25 insertions(+)
> 
> diff --git a/block/rbd.c b/block/rbd.c
> index 0c549c9935..228658e20a 100644
> --- a/block/rbd.c
> +++ b/block/rbd.c
> @@ -102,6 +102,7 @@ typedef struct BDRVRBDState {
>      rbd_image_t image;
>      char *image_name;
>      char *snap;
> +    uint64_t image_size;
>  } BDRVRBDState;

Can't we use bs->total_sectors instead of adding a new image_size field?

>  static int qemu_rbd_connect(rados_t *cluster, rados_ioctx_t *io_ctx,
> @@ -777,6 +778,14 @@ static int qemu_rbd_open(BlockDriverState *bs, QDict *options, int flags,
>          goto failed_open;
>      }
>  
> +    r = rbd_get_size(s->image, &s->image_size);
> +    if (r < 0) {
> +        error_setg_errno(errp, -r, "error reading image size from %s",
> +                         s->image_name);
> +        rbd_close(s->image);
> +        goto failed_open;
> +    }
> +
>      /* If we are using an rbd snapshot, we must be r/o, otherwise
>       * leave as-is */
>      if (s->snap != NULL) {
> @@ -921,6 +930,20 @@ static BlockAIOCB *rbd_start_aio(BlockDriverState *bs,
>          rcb->buf = acb->bounce;
>      }
>  
> +    /*
> +     * RBD APIs don't allow us to write more than actual size, so in order
> +     * to support growing images, we resize the image before RW operations
> +     * that exceed the current size.
> +     */
> +    if (s->image_size < off + size) {
> +        r = rbd_resize(s->image, off + size);
> +        if (r < 0) {
> +            goto failed;
> +        }
> +
> +        s->image_size = off + size;
> +    }

This doesn't check the request type, so it's actually not limited to RW
operations, but even reads will try to resize the image. This is at
least surprising. For regular files, file-posix extends the file for
write requests, but for reads it returns a zeroed buffer without
actually changing the file size.

Kevin


  parent reply	other threads:[~2019-04-29 10:25 UTC|newest]

Thread overview: 40+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-04-11 10:50 [Qemu-devel] [PATCH RFC 0/1] block/rbd: increase dynamically the image size Stefano Garzarella
2019-04-11 10:50 ` Stefano Garzarella
2019-04-11 10:50 ` [Qemu-devel] [PATCH RFC 1/1] " Stefano Garzarella
2019-04-11 10:50   ` Stefano Garzarella
2019-04-11 12:35   ` Jason Dillaman
2019-04-11 12:35     ` Jason Dillaman
2019-04-11 13:02     ` Stefano Garzarella
2019-04-11 13:02       ` Stefano Garzarella
2019-04-11 17:06       ` Jason Dillaman
2019-04-11 17:06         ` Jason Dillaman
2019-04-14 13:20         ` Stefano Garzarella
2019-04-14 13:20           ` Stefano Garzarella
2019-04-14 15:14           ` Jason Dillaman
2019-04-14 15:14             ` Jason Dillaman
2019-04-15  8:04             ` Kevin Wolf
2019-04-15  8:04               ` Kevin Wolf
2019-04-17  7:34               ` Stefano Garzarella
2019-04-17  7:34                 ` Stefano Garzarella
2019-04-17  8:04                 ` Kevin Wolf
2019-04-17  8:04                   ` Kevin Wolf
2019-04-19 12:23                   ` Stefano Garzarella
2019-04-19 12:23                     ` Stefano Garzarella
2019-04-23  7:56                     ` Kevin Wolf
2019-04-23  7:56                       ` Kevin Wolf
2019-04-23  8:26                       ` Stefano Garzarella
2019-04-23  8:26                         ` Stefano Garzarella
2019-04-23  8:38                         ` Kevin Wolf
2019-04-23  8:38                           ` Kevin Wolf
2019-04-29  9:58                           ` [Qemu-devel] [Qemu-block] " Kevin Wolf
2019-04-29  9:58                             ` Kevin Wolf
2019-04-29 10:11                             ` Stefano Garzarella
2019-04-29 10:11                               ` Stefano Garzarella
2019-04-29 10:25   ` Kevin Wolf [this message]
2019-04-29 10:25     ` [Qemu-devel] " Kevin Wolf
2019-04-29 14:04     ` Stefano Garzarella
2019-04-29 14:04       ` Stefano Garzarella
2019-04-29 14:30       ` Kevin Wolf
2019-04-29 14:30         ` Kevin Wolf
2019-04-29 15:55         ` Stefano Garzarella
2019-04-29 15:55           ` Stefano Garzarella

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=20190429102510.GD8492@localhost.localdomain \
    --to=kwolf@redhat.com \
    --cc=jdurgin@redhat.com \
    --cc=mreitz@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=sgarzare@redhat.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;
as well as URLs for NNTP newsgroup(s).