[lvm-devel] [PATCH 0/4]: Change semantics of lvm_vg_remove and rename vg_remove fns
Dave Wysochanski
dwysocha at redhat.com
Mon Aug 10 18:08:21 UTC 2009
This patchset changes the semantics of lvm_vg_remove to be more consistent with
lvm_vg_create, and renames some of the internal lvm library vg_remove functions.
For liblvm, we would like a consistent API that allows modification of LVM
objects in memory before committing to disk. The commit to disk is normally
done in a single call, lvm_vg_write(). Currently, not all liblvm functions
behave this way, and lvm_vg_remove is one of them, as it commits to disk after
completion and does not require lvm_vg_write. This is inconsistent with
lvm_vg_create, which requires lvm_vg_write to commit to disk. This patchset
changes the semantic of lvm_vg_remove to require lvm_vg_write to commit to
disk and make it consistent with lvm_vg_create and other functions. Along
the way, there are a few renames and minor refactorings to the internal
vg_remove functions.
Further cleanup should probably consolidate the various areas where devices are
removed from VGs as there is currently a bit of duplicated code. I've avoided
such changes in this patchset though to keep it as simple and low risk.
Testsuite ok and applies cleanly on:
commit 597986b711175bd90d234e0a6ac001f165aa1274
Author: Dave Wysochanski <dwysocha at redhat.com>
Date: Mon Aug 10 17:23:04 2009 +0000
Update man pages to clarify usage of PE ranges.
More information about the lvm-devel
mailing list