From: Benny Halevy <bhalevy@panasas.com>
To: Jim Rees <rees@umich.edu>
Cc: linux-nfs@vger.kernel.org, peter honeyman <honey@citi.umich.edu>
Subject: Re: [PATCH] nfs-utils: fix stripe device size calculation for layouts not a multiple of stripe unit
Date: Wed, 23 Feb 2011 18:31:14 -0800 [thread overview]
Message-ID: <4D65C2F2.5080808@panasas.com> (raw)
In-Reply-To: <20110223193016.GA14849-8f4Pc2RrbJmHXe+LvDLADg@public.gmane.org>
Looks good to me.
Merged at pnfs-nfs-utils-1-2-4-rc5-2011-02-23
Benny
On 2011-02-23 11:30, Jim Rees wrote:
> If a complex layout has a stripe volume in it, and that volume has a size
> that is not a multiple of the stripe unit, the total layout size will be
> wrong because it is based on the truncated stripe volume size. Fix this by
> using the truncated size for making the mapped device, and the untruncated
> size in the table for subsequent devices to use.
>
> Also remove the warning for stripe volumes not a multiple of the stripe
> width, since this is legal and apparently common.
>
> While we're at it, use variable names consistent with the terminology in
> rfc5661 section 13.2.
>
> Signed-off-by: Jim Rees <rees@umich.edu>
> ---
> utils/blkmapd/device-process.c | 10 ++++------
> utils/blkmapd/dm-device.c | 21 +++++----------------
> 2 files changed, 9 insertions(+), 22 deletions(-)
>
> diff --git a/utils/blkmapd/device-process.c b/utils/blkmapd/device-process.c
> index 9a78457..79e596d 100644
> --- a/utils/blkmapd/device-process.c
> +++ b/utils/blkmapd/device-process.c
> @@ -295,10 +295,10 @@ decode_blk_volume(uint32_t **pp, uint32_t *end, struct bl_volume *vols, int voln
> case BLOCK_VOLUME_STRIPE:
> BLK_READBUF(p, end, 8);
> READ_SECTOR(vol->param.bv_stripe_unit);
> - off_t chunksize = vol->param.bv_stripe_unit;
> + off_t stripe_unit = vol->param.bv_stripe_unit;
> /* Check limitations imposed by device-mapper */
> - if ((chunksize & (chunksize - 1)) != 0
> - || chunksize < (off_t) (PAGE_SIZE >> 9))
> + if ((stripe_unit & (stripe_unit - 1)) != 0
> + || stripe_unit < (off_t) (PAGE_SIZE >> 9))
> return -EIO;
> BLK_READBUF(p, end, 4);
> READ32(vol->bv_vol_n);
> @@ -306,7 +306,7 @@ decode_blk_volume(uint32_t **pp, uint32_t *end, struct bl_volume *vols, int voln
> return -EIO;
> *array_cnt = vol->bv_vol_n;
> BL_LOG_INFO("%s: stripe %d nvols=%d unit=%ld\n", __func__, voln,
> - vol->bv_vol_n, (long)chunksize);
> + vol->bv_vol_n, (long)stripe_unit);
> status = set_vol_array(&p, end, vols, voln);
> if (status)
> return status;
> @@ -317,9 +317,7 @@ decode_blk_volume(uint32_t **pp, uint32_t *end, struct bl_volume *vols, int voln
> return -EIO;
> }
> }
> - /* Truncate size to a stripe unit boundary */
> vol->bv_size = vol->bv_vols[0]->bv_size * vol->bv_vol_n;
> - vol->bv_size &= ~(chunksize - 1);
> break;
> case BLOCK_VOLUME_CONCAT:
> BLK_READBUF(p, end, 4);
> diff --git a/utils/blkmapd/dm-device.c b/utils/blkmapd/dm-device.c
> index c4fe4e9..0f4f148 100644
> --- a/utils/blkmapd/dm-device.c
> +++ b/utils/blkmapd/dm-device.c
> @@ -369,7 +369,7 @@ static int dm_device_exists(char *dev_name)
> /* TODO: check the value for DM_DEV_NAME_LEN, DM_TYPE_LEN, DM_PARAMS_LEN */
> uint64_t dm_device_create(struct bl_volume *vols, int num_vols)
> {
> - uint64_t size, stripe_unit, stripe_size, nstripes, dev = 0;
> + uint64_t size, stripe_unit, dev = 0;
> unsigned int count = dev_count;
> int volnum, i, pos;
> struct bl_volume *node;
> @@ -414,21 +414,10 @@ uint64_t dm_device_create(struct bl_volume *vols, int num_vols)
> if (!table)
> goto out;
> table->offset = 0;
> - stripe_unit = node->param.bv_stripe_unit << 9;
> - stripe_size = stripe_unit * node->bv_vol_n;
> - nstripes = node->bv_size * node->bv_vol_n / stripe_size;
> - /* Make sure total size is a multiple of stripe size */
> - size = node->bv_size;
> - if (size % stripe_size != 0) {
> - /* XXX Should this be an error? */
> - BL_LOG_WARNING(
> - "%s: %d units of %llu bytes is not a multiple of %lld stripe size\n",
> - __func__, node->bv_vol_n,
> - (long long unsigned) node->bv_size,
> - (long long unsigned) stripe_size);
> - size = nstripes * stripe_size;
> - }
> - table->size = size;
> + /* Truncate size to a stripe unit boundary */
> + stripe_unit = node->param.bv_stripe_unit;
> + table->size =
> + node->bv_size - (node->bv_size % stripe_unit);
> strcpy(table->target_type, "striped");
> sprintf(table->params, "%d %llu %n", node->bv_vol_n,
> (long long unsigned) stripe_unit, &pos);
prev parent reply other threads:[~2011-02-24 2:31 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-02-23 19:30 [PATCH] nfs-utils: fix stripe device size calculation for layouts not a multiple of stripe unit Jim Rees
[not found] ` <20110223193016.GA14849-8f4Pc2RrbJmHXe+LvDLADg@public.gmane.org>
2011-02-24 2:31 ` Benny Halevy [this message]
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=4D65C2F2.5080808@panasas.com \
--to=bhalevy@panasas.com \
--cc=honey@citi.umich.edu \
--cc=linux-nfs@vger.kernel.org \
--cc=rees@umich.edu \
/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).