From: Anthony Liguori <anthony@codemonkey.ws>
To: qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 3/5] Write table offset and size in one syscall.
Date: Mon, 24 Nov 2008 10:42:24 -0600 [thread overview]
Message-ID: <492AD970.9050805@codemonkey.ws> (raw)
In-Reply-To: <20081123145312.22178.68255.stgit@dhcp-1-237.tlv.redhat.com>
Gleb Natapov wrote:
> Otherwise if VM is killed between two writes data may be lost.
> But if offset and size fields are at the same disk block one
> write should update them both simultaneously.
>
> Signed-off-by: Gleb Natapov <gleb@redhat.com>
> ---
>
> block-qcow2.c | 27 ++++++++++-----------------
> 1 files changed, 10 insertions(+), 17 deletions(-)
>
> diff --git a/block-qcow2.c b/block-qcow2.c
> index 69f6414..7f99921 100644
> --- a/block-qcow2.c
> +++ b/block-qcow2.c
> @@ -429,8 +429,7 @@ static int grow_l1_table(BlockDriverState *bs, int min_size)
> int new_l1_size, new_l1_size2, ret, i;
> uint64_t *new_l1_table;
> uint64_t new_l1_table_offset;
> - uint64_t data64;
> - uint32_t data32;
> + uint8_t data[12];
>
This assumes packing will happen correctly.
>
> new_l1_size = s->l1_size;
> if (min_size <= new_l1_size)
> @@ -460,14 +459,12 @@ static int grow_l1_table(BlockDriverState *bs, int min_size)
> new_l1_table[i] = be64_to_cpu(new_l1_table[i]);
>
> /* set new table */
> - data64 = cpu_to_be64(new_l1_table_offset);
> - if (bdrv_pwrite(s->hd, offsetof(QCowHeader, l1_table_offset),
> - &data64, sizeof(data64)) != sizeof(data64))
> - goto fail;
> - data32 = cpu_to_be32(new_l1_size);
> - if (bdrv_pwrite(s->hd, offsetof(QCowHeader, l1_size),
> - &data32, sizeof(data32)) != sizeof(data32))
> + *(uint32_t*)data = cpu_to_be32(new_l1_size);
> + *(uint64_t*)&data[4] = cpu_to_be64(new_l1_table_offset);
> + if (bdrv_pwrite(s->hd, offsetof(QCowHeader, l1_size), data,
> + sizeof(data)) != sizeof(data))
> goto fail;
>
>
Why not just introduces a uint8_t data[12] in this function, memcpy to
the right offsets, and do one brdv_pwrite? Then you don't need to do
weird things with packing.
> qemu_free(s->l1_table);
> free_clusters(bs, s->l1_table_offset, s->l1_size * sizeof(uint64_t));
> s->l1_table_offset = new_l1_table_offset;
> @@ -2278,8 +2275,7 @@ static int grow_refcount_table(BlockDriverState *bs, int min_size)
> int new_table_size, new_table_size2, refcount_table_clusters, i, ret;
> uint64_t *new_table;
> int64_t table_offset;
> - uint64_t data64;
> - uint32_t data32;
> + uint8_t data[12];
> int old_table_size;
> int64_t old_table_offset;
>
> @@ -2318,13 +2314,10 @@ static int grow_refcount_table(BlockDriverState *bs, int min_size)
> for(i = 0; i < s->refcount_table_size; i++)
> be64_to_cpus(&new_table[i]);
>
> - data64 = cpu_to_be64(table_offset);
> + *(uint64_t*)data = cpu_to_be64(table_offset);
> + *(uint32_t*)&data[8] = cpu_to_be32(refcount_table_clusters);
> if (bdrv_pwrite(s->hd, offsetof(QCowHeader, refcount_table_offset),
> - &data64, sizeof(data64)) != sizeof(data64))
> - goto fail;
> - data32 = cpu_to_be32(refcount_table_clusters);
> - if (bdrv_pwrite(s->hd, offsetof(QCowHeader, refcount_table_clusters),
> - &data32, sizeof(data32)) != sizeof(data32))
> + data, sizeof(data)) != sizeof(data))
> goto fail;
>
Same here. Alternatively, you could use the cpu_to_beXXs() variants and
just pass in pointer offsets.
Regards,
Anthony Liguori
> qemu_free(s->refcount_table);
> old_table_offset = s->refcount_table_offset;
>
>
>
>
next prev parent reply other threads:[~2008-11-24 16:42 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2008-11-23 14:52 [Qemu-devel] [PATCH 0/5] QCOW2 small cleanups and changing metadata update order Gleb Natapov
2008-11-23 14:52 ` [Qemu-devel] [PATCH 1/5] Remove tabs from qcow_aio_read_cb(). Fix indentation Gleb Natapov
2008-11-24 16:31 ` Anthony Liguori
2008-11-24 20:03 ` Gleb Natapov
2008-11-23 14:53 ` [Qemu-devel] [PATCH 2/5] Introduce new helper function qcow_shedule_bh() Gleb Natapov
2008-11-24 16:32 ` Anthony Liguori
2008-11-24 20:05 ` Gleb Natapov
2008-11-23 14:53 ` [Qemu-devel] [PATCH 3/5] Write table offset and size in one syscall Gleb Natapov
2008-11-24 16:42 ` Anthony Liguori [this message]
2008-11-24 20:11 ` Gleb Natapov
2008-11-23 14:53 ` [Qemu-devel] [PATCH 4/5] Cleanup {alloc|get}_cluster_offset() Gleb Natapov
2008-11-24 16:47 ` Anthony Liguori
2008-11-23 14:53 ` [Qemu-devel] [PATCH 5/5] Change order of metadata update to prevent loosing guest data because of unexpected exit Gleb Natapov
2008-11-24 16:49 ` Anthony Liguori
2008-11-24 20:19 ` Gleb Natapov
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=492AD970.9050805@codemonkey.ws \
--to=anthony@codemonkey.ws \
--cc=qemu-devel@nongnu.org \
/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).