From: Zdenek Kabelac <zkabelac@redhat.com>
To: lvm-devel@redhat.com
Subject: [PATCH 0/6] VG share v2
Date: Tue, 22 Mar 2011 22:21:54 +0100 [thread overview]
Message-ID: <cover.1300828168.git.zkabelac@redhat.com> (raw)
Second version of VG sharing patch.
Still rather in 'proposal' form as some part of code are not fully
finished (pool-debug code is missing, usage of posix_memalign).
If someone notices some weak points - let me know.
Patchset add support for locking and unlocking memory pool.
Such pool has volatile memory area attached.
VG sharing is made in way the only 1 user could work with such VG.
It may optionaly modify few small parts (only uin64_t is supported)
which are reverted before next reuse.
2 methods to catch writes are implemented -
mprotect - memory consuming, and creates immediate segfault when
expected change of pool content happens.
But it's fast and give exact location of such writer.
pool_crc - simple hash crc checksum - gives notice that after multiple
uses is not in its original state - doesn't segfault, but it's
hard to detect such writer. And checksumming consumes some CPU.
Pool volatile area uses very simle 'exception pair values' - if there
would be a lot of them - more efficent code would be needed.
Also support for storing more differnet types would add some complexity.
Both these extension do not seem to be needed with current code.
Patch set - as is - passes valgrind tests.
Zdenek Kabelac (6):
Pool locking code
Code move vg_mark_partial up in stack
Share VG multiple times
Use dm_pool_set_uint64
lv_postorder using dm_pool_set_uint64
lv_postorder unlock and lock
lib/activate/activate.c | 5 +-
lib/cache/lvmcache.c | 111 ++++++++++++++++++-
lib/cache/lvmcache.h | 3 +
lib/metadata/lv.h | 5 +-
lib/metadata/metadata.c | 58 ++++++++--
lib/metadata/snapshot_manip.c | 15 ++-
lib/metadata/vg.h | 2 +
libdm/libdevmapper.h | 11 ++
libdm/mm/pool-debug.c | 9 ++
libdm/mm/pool-fast.c | 74 ++++++++++++-
libdm/mm/pool.c | 246 ++++++++++++++++++++++++++++++++++++++++-
11 files changed, 517 insertions(+), 22 deletions(-)
--
1.7.4.1
next reply other threads:[~2011-03-22 21:21 UTC|newest]
Thread overview: 20+ messages / expand[flat|nested] mbox.gz Atom feed top
2011-03-22 21:21 Zdenek Kabelac [this message]
2011-03-22 21:21 ` [PATCH 1/6] Pool locking code Zdenek Kabelac
2011-03-23 14:03 ` Joe Thornber
2011-03-23 14:43 ` Zdenek Kabelac
2011-03-23 14:53 ` Joe Thornber
2011-03-23 14:54 ` Zdenek Kabelac
2011-03-23 15:13 ` Joe Thornber
2011-03-23 15:57 ` Zdenek Kabelac
2011-03-23 16:33 ` Zdenek Kabelac
2011-03-23 17:29 ` Zdenek Kabelac
2011-03-23 18:41 ` Joe Thornber
2011-03-23 19:01 ` Zdenek Kabelac
2011-03-24 12:21 ` Joe Thornber
2011-03-24 12:45 ` Zdenek Kabelac
2011-03-24 13:50 ` Zdenek Kabelac
2011-03-22 21:21 ` [PATCH 2/6] Code move vg_mark_partial up in stack Zdenek Kabelac
2011-03-22 21:21 ` [PATCH 3/6] Share VG multiple times Zdenek Kabelac
2011-03-22 21:21 ` [PATCH 4/6] Use dm_pool_set_uint64 Zdenek Kabelac
2011-03-22 21:21 ` [PATCH 5/6] lv_postorder using dm_pool_set_uint64 Zdenek Kabelac
2011-03-22 21:22 ` [PATCH 6/6] lv_postorder unlock and lock Zdenek Kabelac
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=cover.1300828168.git.zkabelac@redhat.com \
--to=zkabelac@redhat.com \
--cc=lvm-devel@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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.