qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Max Reitz <mreitz@redhat.com>
To: Wen Congyang <wency@cn.fujitsu.com>,
	qemu devel <qemu-devel@nongnu.org>,
	Eric Blake <eblake@redhat.com>,
	Markus Armbruster <armbru@redhat.com>,
	Alberto Garcia <berto@igalia.com>,
	Stefan Hajnoczi <stefanha@redhat.com>
Cc: Kevin Wolf <kwolf@redhat.com>,
	zhanghailiang <zhang.zhanghailiang@huawei.com>,
	qemu block <qemu-block@nongnu.org>,
	Jiang Yunhong <yunhong.jiang@intel.com>,
	Dong Eddie <eddie.dong@intel.com>,
	"Dr. David Alan Gilbert" <dgilbert@redhat.com>,
	Gonglei <arei.gonglei@huawei.com>,
	Yang Hongyang <yanghy@cn.fujitsu.com>
Subject: Re: [Qemu-devel] [Qemu-block] [PATCH v5 2/4] quorum: implement bdrv_add_child() and bdrv_del_child()
Date: Wed, 7 Oct 2015 20:51:21 +0200	[thread overview]
Message-ID: <561569A9.50609@redhat.com> (raw)
In-Reply-To: <1442907862-21376-3-git-send-email-wency@cn.fujitsu.com>

[-- Attachment #1: Type: text/plain, Size: 7322 bytes --]

On 22.09.2015 09:44, Wen Congyang wrote:
> Signed-off-by: Wen Congyang <wency@cn.fujitsu.com>
> Signed-off-by: zhanghailiang <zhang.zhanghailiang@huawei.com>
> Signed-off-by: Gonglei <arei.gonglei@huawei.com>
> ---
>  block.c               |  6 ++---
>  block/quorum.c        | 72 +++++++++++++++++++++++++++++++++++++++++++++++++--
>  include/block/block.h |  3 +++
>  3 files changed, 76 insertions(+), 5 deletions(-)
> 
> diff --git a/block.c b/block.c
> index 1b25e43..01f6d69 100644
> --- a/block.c
> +++ b/block.c
> @@ -1079,9 +1079,9 @@ static int bdrv_fill_options(QDict **options, const char **pfilename,
>      return 0;
>  }
>  
> -static BdrvChild *bdrv_attach_child(BlockDriverState *parent_bs,
> -                                    BlockDriverState *child_bs,
> -                                    const BdrvChildRole *child_role)
> +BdrvChild *bdrv_attach_child(BlockDriverState *parent_bs,
> +                             BlockDriverState *child_bs,
> +                             const BdrvChildRole *child_role)
>  {
>      BdrvChild *child = g_new(BdrvChild, 1);
>      *child = (BdrvChild) {
> diff --git a/block/quorum.c b/block/quorum.c
> index 8fe53b4..111a57b 100644
> --- a/block/quorum.c
> +++ b/block/quorum.c
> @@ -66,6 +66,9 @@ typedef struct QuorumVotes {
>  typedef struct BDRVQuorumState {
>      BlockDriverState **bs; /* children BlockDriverStates */
>      int num_children;      /* children count */
> +    int max_children;      /* The maximum children count, we need to reallocate
> +                            * bs if num_children grows larger than maximum.
> +                            */
>      int threshold;         /* if less than threshold children reads gave the
>                              * same result a quorum error occurs.
>                              */
> @@ -874,9 +877,9 @@ static int quorum_open(BlockDriverState *bs, QDict *options, int flags,
>          ret = -EINVAL;
>          goto exit;
>      }
> -    if (s->num_children < 2) {
> +    if (s->num_children < 1) {
>          error_setg(&local_err,
> -                   "Number of provided children must be greater than 1");
> +                   "Number of provided children must be 1 or more");
>          ret = -EINVAL;
>          goto exit;
>      }
> @@ -925,6 +928,7 @@ static int quorum_open(BlockDriverState *bs, QDict *options, int flags,
>      /* allocate the children BlockDriverState array */
>      s->bs = g_new0(BlockDriverState *, s->num_children);
>      opened = g_new0(bool, s->num_children);
> +    s->max_children = s->num_children;
>  
>      for (i = 0; i < s->num_children; i++) {
>          char indexstr[32];
> @@ -995,6 +999,67 @@ static void quorum_attach_aio_context(BlockDriverState *bs,
>      }
>  }
>  
> +static void quorum_add_child(BlockDriverState *bs, BlockDriverState *child_bs,
> +                             Error **errp)
> +{
> +    BDRVQuorumState *s = bs->opaque;
> +
> +    bdrv_drain(bs);
> +
> +    if (s->num_children == s->max_children) {
> +        if (s->max_children >= INT_MAX) {

Opposing Berto (:-)), I like >= even if the > part is actually
impossible. I myself like to use constructs such as:

for (i = 0; i < n; i++) {
    /* ... */
}
if (i >= n) {
    /* ... */
}

Even though @i can never exceed @n after the loop. This is because my
way of thinking is "What if it could exceed @n? Then I'd like to take
this branch as well." The same applies here. s->max_children can never
exceed INT_MAX, but if it could, we'd want that to be an error, too.

I do find myself in discussions about that from time to time, especially
since others prefer (for the example above):

for (i = 0; i < n; i++) { /* ... */ }
assert(i <= n);
if (i == n) { /* ... */ }

Or the like.

> +            error_setg(errp, "Too many children");
> +            return;
> +        }
> +
> +        s->bs = g_renew(BlockDriverState *, s->bs, s->max_children + 1);
> +        s->bs[s->num_children] = NULL;
> +        s->max_children++;
> +    }

Just a suggestion, please feel free to ignore it completely:

You can drop the s->max_children field and just always call g_renew()
with s->num_children + 1 as the @count parameter. There shouldn't be any
(visible) performance penalty, but it would simplify the code.

> +
> +    bdrv_ref(child_bs);
> +    bdrv_attach_child(bs, child_bs, &child_format);
> +    s->bs[s->num_children++] = child_bs;
> +}
> +
> +static void quorum_del_child(BlockDriverState *bs, BlockDriverState *child_bs,
> +                             Error **errp)
> +{
> +    BDRVQuorumState *s = bs->opaque;
> +    BdrvChild *child;
> +    int i;
> +
> +    for (i = 0; i < s->num_children; i++) {
> +        if (s->bs[i] == child_bs) {
> +            break;
> +        }
> +    }
> +
> +    QLIST_FOREACH(child, &bs->children, next) {
> +        if (child->bs == child_bs) {
> +            break;
> +        }
> +    }
> +
> +    /* we have checked it in bdrv_del_child() */
> +    assert(i < s->num_children && child);
> +
> +    if (s->num_children <= s->threshold) {
> +        error_setg(errp,
> +            "The number of children cannot be lower than the vote threshold %d",
> +            s->threshold);
> +        return;
> +    }
> +
> +    bdrv_drain(bs);
> +    /* We can safely remove this child now */
> +    memmove(&s->bs[i], &s->bs[i + 1],
> +            (s->num_children - i - 1) * sizeof(void *));

I'd use a matching type here, i.e. sizeof(BlockDriverState *) before
Kevin's bdrv_swap() series or sizeof(BdrvChild *) afterwards.

> +    s->num_children--;
> +    s->bs[s->num_children] = NULL;

In case you do decide to drop s->max_children, you could replace this by
a call to g_renew on s->bs (or s->children).

In any case, as Berto said, this line is unnecessary.

Max

> +    bdrv_unref_child(bs, child);
> +}
> +
>  static void quorum_refresh_filename(BlockDriverState *bs)
>  {
>      BDRVQuorumState *s = bs->opaque;
> @@ -1049,6 +1114,9 @@ static BlockDriver bdrv_quorum = {
>      .bdrv_detach_aio_context            = quorum_detach_aio_context,
>      .bdrv_attach_aio_context            = quorum_attach_aio_context,
>  
> +    .bdrv_add_child                     = quorum_add_child,
> +    .bdrv_del_child                     = quorum_del_child,
> +
>      .is_filter                          = true,
>      .bdrv_recurse_is_first_non_filter   = quorum_recurse_is_first_non_filter,
>  };
> diff --git a/include/block/block.h b/include/block/block.h
> index 665c56f..bd97399 100644
> --- a/include/block/block.h
> +++ b/include/block/block.h
> @@ -514,6 +514,9 @@ void bdrv_disable_copy_on_read(BlockDriverState *bs);
>  void bdrv_ref(BlockDriverState *bs);
>  void bdrv_unref(BlockDriverState *bs);
>  void bdrv_unref_child(BlockDriverState *parent, BdrvChild *child);
> +BdrvChild *bdrv_attach_child(BlockDriverState *parent_bs,
> +                             BlockDriverState *child_bs,
> +                             const BdrvChildRole *child_role);
>  
>  bool bdrv_op_is_blocked(BlockDriverState *bs, BlockOpType op, Error **errp);
>  void bdrv_op_block(BlockDriverState *bs, BlockOpType op, Error *reason);
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

  parent reply	other threads:[~2015-10-07 18:51 UTC|newest]

Thread overview: 45+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-09-22  7:44 [Qemu-devel] [PATCH v5 0/4] qapi: child add/delete support Wen Congyang
2015-09-22  7:44 ` [Qemu-devel] [PATCH v5 1/4] Add new block driver interface to add/delete a BDS's child Wen Congyang
2015-10-07 13:35   ` Alberto Garcia
2015-10-08  2:05     ` Wen Congyang
2015-10-07 18:33   ` [Qemu-devel] [Qemu-block] " Max Reitz
2015-10-08  2:06     ` Wen Congyang
2015-10-07 19:00   ` [Qemu-devel] " Dr. David Alan Gilbert
2015-10-08  2:03     ` Wen Congyang
2015-10-08 18:44       ` Dr. David Alan Gilbert
2015-09-22  7:44 ` [Qemu-devel] [PATCH v5 2/4] quorum: implement bdrv_add_child() and bdrv_del_child() Wen Congyang
2015-10-07 14:12   ` Alberto Garcia
2015-10-08  2:10     ` Wen Congyang
2015-10-07 18:51   ` Max Reitz [this message]
2015-10-08  8:12     ` [Qemu-devel] [Qemu-block] " Alberto Garcia
2015-10-09 15:51       ` Max Reitz
2015-10-12 11:56         ` Alberto Garcia
2015-09-22  7:44 ` [Qemu-devel] [PATCH v5 3/4] qmp: add monitor command to add/remove a child Wen Congyang
2015-10-07 14:33   ` Alberto Garcia
2015-10-07 19:42   ` [Qemu-devel] [Qemu-block] " Max Reitz
2015-10-08  6:15     ` Markus Armbruster
2015-10-08  8:29       ` Alberto Garcia
2015-10-08 10:03         ` Kevin Wolf
2015-10-08 10:13           ` Alberto Garcia
2015-10-09 16:14         ` Max Reitz
2015-10-08 11:02       ` [Qemu-devel] Dynamic reconfiguration (was: qmp: add monitor command to add/remove a child) Kevin Wolf
2015-10-08 11:10         ` [Qemu-devel] [Qemu-block] " Kevin Wolf
2015-10-21  8:27           ` [Qemu-devel] [Qemu-block] Dynamic reconfiguration Markus Armbruster
2015-10-26  2:04             ` Wen Congyang
2015-10-26  7:24               ` Markus Armbruster
2015-10-26  7:25                 ` Wen Congyang
2015-10-09 16:13       ` [Qemu-devel] [Qemu-block] [PATCH v5 3/4] qmp: add monitor command to add/remove a child Max Reitz
2015-10-09 16:42         ` Dr. David Alan Gilbert
2015-10-09 18:24           ` Max Reitz
2015-10-12  8:07             ` Dr. David Alan Gilbert
2015-10-12  8:18             ` Kevin Wolf
2015-10-12  7:58           ` Markus Armbruster
2015-10-12  7:56         ` Markus Armbruster
2015-10-12 16:27     ` Max Reitz
2015-09-22  7:44 ` [Qemu-devel] [PATCH v5 4/4] hmp: " Wen Congyang
2015-10-07 14:38   ` Alberto Garcia
2015-09-22 11:15 ` [Qemu-devel] [PATCH v5 0/4] qapi: child add/delete support Dr. David Alan Gilbert
2015-09-23  1:08   ` Wen Congyang
2015-09-23  9:21     ` Dr. David Alan Gilbert
2015-09-23  9:30       ` Wen Congyang
2015-10-07  6:40 ` Wen Congyang

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=561569A9.50609@redhat.com \
    --to=mreitz@redhat.com \
    --cc=arei.gonglei@huawei.com \
    --cc=armbru@redhat.com \
    --cc=berto@igalia.com \
    --cc=dgilbert@redhat.com \
    --cc=eblake@redhat.com \
    --cc=eddie.dong@intel.com \
    --cc=kwolf@redhat.com \
    --cc=qemu-block@nongnu.org \
    --cc=qemu-devel@nongnu.org \
    --cc=stefanha@redhat.com \
    --cc=wency@cn.fujitsu.com \
    --cc=yanghy@cn.fujitsu.com \
    --cc=yunhong.jiang@intel.com \
    --cc=zhang.zhanghailiang@huawei.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).