]> sourceware.org Git - lvm2.git/log
lvm2.git
6 years agobuild: fix build rules for srcdir
Zdenek Kabelac [Sat, 12 May 2018 16:14:53 +0000 (18:14 +0200)]
build: fix build rules for srcdir

It's very hard to use some 'non-recurive' Makefiles with
rest of system running 'recursively'.

So ATM drop inclusion of subdir makefile and add support
for 2 new top-level targets:

unit-test  (builds test/unit dir)
run-unit-test (build & run test/unit/unit-test run)

6 years agobcache: do not use libdm header files
Zdenek Kabelac [Fri, 11 May 2018 16:41:44 +0000 (18:41 +0200)]
bcache: do not use libdm header files

Logging for libdm differs from lvm logging - keep using consisten
logging function calls.

6 years agobuild: make generate
Zdenek Kabelac [Sat, 12 May 2018 15:02:34 +0000 (17:02 +0200)]
build: make generate

6 years agotests: drop cache checking
Zdenek Kabelac [Mon, 7 May 2018 09:53:41 +0000 (11:53 +0200)]
tests: drop cache checking

Just like 52656c89fd6e7655b0599e20f90cc9d70561f1fb
when now cache is compiled in 'unditionally'.

This patch is actually enforce by changes in
commit: 2bc896f2a3a0023a4f45f10da3d0a22295a633ee
where CACHE value is not set anymore.

6 years agotests: add support to run unit test
Zdenek Kabelac [Sat, 12 May 2018 16:09:44 +0000 (18:09 +0200)]
tests: add support to run unit test

6 years agobuild: properly track source file for lmvlockctl
Zdenek Kabelac [Fri, 11 May 2018 23:19:07 +0000 (01:19 +0200)]
build: properly track source file for lmvlockctl

Ensure the source file is tracked by various cleanup functions.

6 years agotests: inittest may run without root
Zdenek Kabelac [Fri, 11 May 2018 22:08:06 +0000 (00:08 +0200)]
tests: inittest may run without root

If the test does not need root, it can use 'SKIP_ROOT_DM_CHECK'.

For such test no actions needed root to initilize DM devices and
nodes will be take and test can check i.e. functional unit tests.

6 years agotests: inittest compare string
Zdenek Kabelac [Fri, 11 May 2018 22:06:44 +0000 (00:06 +0200)]
tests: inittest compare string

Avoid logging warning when compared string is empty with -eq.

6 years agotests: aux extra protection for rm -rf
Zdenek Kabelac [Fri, 11 May 2018 22:05:27 +0000 (00:05 +0200)]
tests: aux extra protection for rm -rf

6 years agotests: again disable this raid test
Zdenek Kabelac [Sat, 12 May 2018 10:02:01 +0000 (12:02 +0200)]
tests: again disable this raid test

Still kills testing machines even with 4.17-rc4 kernels
on reshaping.

6 years agotests: drop delaying
Zdenek Kabelac [Sat, 12 May 2018 16:04:06 +0000 (18:04 +0200)]
tests: drop delaying

Here seem delaying dev has no use.

6 years agotests: using throttling
Zdenek Kabelac [Mon, 7 May 2018 11:09:46 +0000 (13:09 +0200)]
tests: using throttling

6 years agotests: move device discard
Zdenek Kabelac [Mon, 7 May 2018 09:56:00 +0000 (11:56 +0200)]
tests: move device discard

6 years agotests: use throttle_dm_mirror
Zdenek Kabelac [Mon, 7 May 2018 09:45:11 +0000 (11:45 +0200)]
tests: use throttle_dm_mirror

In this case it's better to use throttling of mirror sync,
that delay everything with dm_delay.

6 years agotests: aux support throttling of dm mirror
Zdenek Kabelac [Mon, 7 May 2018 09:39:28 +0000 (11:39 +0200)]
tests: aux support throttling of dm mirror

Usage of dm_delay looks to be slowing not just 'delayed' portion
of device, but due to the fact it's also slows down ANY flush
operation on such device it's overal speed impact is huge.

In some case we can however user other methods to slowdown disk writes,
in case of old dm 'mirror'  target we can throttle  I/O of mirror
synchronisation giving the next commands enough time to test couple
race conditions.

Usage:

throttle_dm_mirror [percentage]

Thtrottle down sync speed (lowest is '1' which is also default when
unspecified)

restore_dm_mirror

Restores the value of throttling before call of  'throttle_dm_mirror'
Usually it should '100'

6 years agotests: try running tests over ramdisk
Zdenek Kabelac [Mon, 7 May 2018 09:35:35 +0000 (11:35 +0200)]
tests: try running tests over ramdisk

Currently usage of loop device over backend file in ramdisk (tmpfs)
is actually causing unnecassary memory consution, since just
reading such loop device is causing RAM provisioning.

This patch add another possible way how to use ramdisk directly
through 'brd' device when possible (and allowed).

This however has it's limitation as well -  brd does not support
TRIM, so the only way how to erase is to remove brd module ??

Alse there is 4K sector size limitation imposed by ramdisk.

Anyway - for some mirror test that were using large amount of
disk space (tens of MB) this brings noticable speed boost.
(But could be worth to solve the slowness of loop in kernel?)

To prevent using 'brd' for testing set LVM_TEST_PREFER_BRD=0
like this:

make check_local LVM_TEST_PREFER_BRD=0

6 years agotests: crypt test cannot run on ramdisk
Zdenek Kabelac [Sat, 12 May 2018 14:00:55 +0000 (16:00 +0200)]
tests: crypt test cannot run on ramdisk

This test can't use brd (ramdisk) as backend since for some
weird reason  lsblk is not listing these device.

TODO: test could be probably rewritten to avoid using lsblk somehow??

6 years agotests: happy using of 4K backend devices
Zdenek Kabelac [Sat, 12 May 2018 13:58:13 +0000 (15:58 +0200)]
tests: happy using of 4K backend devices

When the backend device supports only 4K blocks (like ramdisk)
we cannot use for testing any smaller blocksize.

So recalc test for 4K extent size.

We may possibly introduce one list extra test that
can be executed on devices with 512b sectors to
check lvm2 support those min extent sizes...

6 years agotests: raise min size for XFS
Zdenek Kabelac [Sat, 12 May 2018 13:56:46 +0000 (15:56 +0200)]
tests: raise min size for XFS

Seems XFS now requires at least 1605 blocks.

6 years agopoll: add stdout fflush after poll query
Zdenek Kabelac [Mon, 7 May 2018 09:46:09 +0000 (11:46 +0200)]
poll: add stdout fflush after poll query

ATM it's a bit ugly to enforce flushing of 'stdio' here, but works as quick
hot-fix.

log_print*() is using buffered I/O.

But for pooling with typical 1s interval this may take a while before
buffer about continues progress gets flushed.
So ATM fflush().

TODO: either add  log_print*_with_flush() or maybe directly use just
line buffering with log_print() and only log_debug() keep using buffered
I/O mode.

6 years agoscan: ignore duplicates that are md component devs
David Teigland [Fri, 11 May 2018 20:52:22 +0000 (15:52 -0500)]
scan: ignore duplicates that are md component devs

md devices using an older superblock version have
superblocks at the end of the md device.  For commands
that skip reading the end of devices during filtering,
the md component devs will be scanned, and will appear
as duplicate PVs to the original md device.  Remove
these md components from the list of unused duplicate
devices, so they are treated as if they had been
ignored during filtering.  This avoids the restrictions
that are placed on using PVs with duplicates.

6 years agodev_cache: remove the lvmcache check when closing fd
David Teigland [Fri, 11 May 2018 19:28:46 +0000 (14:28 -0500)]
dev_cache: remove the lvmcache check when closing fd

This is no longer used since devices are not held
open in dev_cache.

6 years agodev_cache: fix close in utility functions
David Teigland [Fri, 11 May 2018 19:25:08 +0000 (14:25 -0500)]
dev_cache: fix close in utility functions

All these functions are now used as utilities,
e.g. for ioctl (not for io), and need to
open/close the device each time they are called.
(Many of the opens can probably be eliminated by
just using the bcache fd for the ioctl.)

6 years agoscan: remove unused args from label_read
David Teigland [Fri, 11 May 2018 19:16:49 +0000 (14:16 -0500)]
scan: remove unused args from label_read

6 years agodevs: recognize md devices in subsystem check
David Teigland [Fri, 11 May 2018 19:00:19 +0000 (14:00 -0500)]
devs: recognize md devices in subsystem check

If md components appear as duplicate PVs, let the
existing subsystem check recognize the md device.

6 years agodev_cache: fix close in dev_get_block_size
David Teigland [Fri, 11 May 2018 18:53:19 +0000 (13:53 -0500)]
dev_cache: fix close in dev_get_block_size

6 years agodev_cache: drop open_list
David Teigland [Fri, 11 May 2018 16:59:44 +0000 (11:59 -0500)]
dev_cache: drop open_list

devices are now held open only in bcache,
so drop the dev_cache list of open devices
which is unused.

6 years agobcache: disable fallback to old io
David Teigland [Fri, 11 May 2018 14:50:25 +0000 (09:50 -0500)]
bcache: disable fallback to old io

All io has been converted to bcache.

6 years agopvck: allow checking at user specified offsets
David Teigland [Thu, 10 May 2018 21:27:34 +0000 (16:27 -0500)]
pvck: allow checking at user specified offsets

with the --labelsector option.  We probably don't
need all this code to support any value for this
option; it's unclear how, when, why it would be
used.

6 years agoradix-tree: fix a function decl
Joe Thornber [Fri, 11 May 2018 10:40:47 +0000 (11:40 +0100)]
radix-tree: fix a function decl

6 years agoradix-tree: remove some unneccessary includes
Joe Thornber [Fri, 11 May 2018 08:46:34 +0000 (09:46 +0100)]
radix-tree: remove some unneccessary includes

6 years agodoc: add a little document describing new directory structure.
Joe Thornber [Fri, 11 May 2018 05:46:25 +0000 (06:46 +0100)]
doc: add a little document describing new directory structure.

6 years agoradix-tree: First drop of radix tree.
Joe Thornber [Fri, 11 May 2018 05:10:01 +0000 (06:10 +0100)]
radix-tree: First drop of radix tree.

An implementation of an adaptive radix tree.  Has the following nice
properties:

  - At least as fast as the hash table
  - Uses less memory
  - You don't need to give an expected size when you create
  - It scales nicely (ie. no large reallocations like the hash table).
  - You can iterate the keys in lexicographical order.

Only insert and lookup are implemented so far.  Plus there's a lot
more performance to come.

6 years agoMerge branch 'master' of git+ssh://sourceware.org/git/lvm2
Joe Thornber [Fri, 11 May 2018 04:39:27 +0000 (05:39 +0100)]
Merge branch 'master' of git+ssh://sourceware.org/git/lvm2

6 years agobcache/sync io engine: handle short ios
Joe Thornber [Fri, 11 May 2018 04:37:47 +0000 (05:37 +0100)]
bcache/sync io engine: handle short ios

6 years agoio: write log header with bcache
David Teigland [Thu, 10 May 2018 21:25:33 +0000 (16:25 -0500)]
io: write log header with bcache

6 years agopvscan: remove unused var warning
David Teigland [Thu, 10 May 2018 21:18:36 +0000 (16:18 -0500)]
pvscan: remove unused var warning

6 years agofilter: use bcache for filter reads
David Teigland [Thu, 3 May 2018 22:12:07 +0000 (17:12 -0500)]
filter: use bcache for filter reads

Filters are still applied before any device reading or
the label scan, but any filter checks that want to read
the device are skipped and the device is flagged.

After bcache is populated, but before lvm looks for
devices (i.e. before label scan), the filters are
reapplied to the devices that were flagged above.
The filters will then find the data they need in
bcache.

6 years agolabel/lv_manip: squash some warnings
Joe Thornber [Thu, 10 May 2018 14:14:39 +0000 (15:14 +0100)]
label/lv_manip: squash some warnings

6 years agofunctional-tests/vdo: fix mem leak in test
Joe Thornber [Thu, 10 May 2018 13:31:16 +0000 (14:31 +0100)]
functional-tests/vdo: fix mem leak in test

6 years agobcache: Add sync io engine
Joe Thornber [Thu, 10 May 2018 13:29:26 +0000 (14:29 +0100)]
bcache: Add sync io engine

Something to fall back to when testing.

6 years agobcache: knock out err param.
Joe Thornber [Thu, 10 May 2018 12:26:08 +0000 (13:26 +0100)]
bcache: knock out err param.

Dave used this for debugging.  Not needed in general.

6 years agoMerge branch 'master' of git+ssh://sourceware.org/git/lvm2
Joe Thornber [Thu, 10 May 2018 12:02:27 +0000 (13:02 +0100)]
Merge branch 'master' of git+ssh://sourceware.org/git/lvm2

6 years agovdo status: Unit tests + fix bugs
Joe Thornber [Thu, 10 May 2018 12:01:26 +0000 (13:01 +0100)]
vdo status: Unit tests + fix bugs

6 years agoMerge branch 'master' into 2018-04-30-vdo-support
Joe Thornber [Thu, 10 May 2018 11:34:04 +0000 (12:34 +0100)]
Merge branch 'master' into 2018-04-30-vdo-support

6 years agodoc: add filter info to scanning
David Teigland [Wed, 9 May 2018 16:14:05 +0000 (11:14 -0500)]
doc: add filter info to scanning

6 years agoio: replace dev_set with bcache equivalents
David Teigland [Wed, 9 May 2018 15:34:28 +0000 (10:34 -0500)]
io: replace dev_set with bcache equivalents

6 years agoMerge branch 'master' of git+ssh://sourceware.org/git/lvm2
Joe Thornber [Wed, 9 May 2018 10:07:24 +0000 (11:07 +0100)]
Merge branch 'master' of git+ssh://sourceware.org/git/lvm2

6 years agobcache-utils: bcache_set_bytes()
Joe Thornber [Wed, 9 May 2018 10:05:29 +0000 (11:05 +0100)]
bcache-utils: bcache_set_bytes()

6 years agounit-test/bcache-utils: Tweak zero tests
Joe Thornber [Wed, 9 May 2018 09:50:31 +0000 (10:50 +0100)]
unit-test/bcache-utils: Tweak zero tests

6 years agodoc: lvm disk reading
David Teigland [Fri, 26 Jan 2018 12:50:52 +0000 (06:50 -0600)]
doc: lvm disk reading

6 years agotests: bump raid target version in reshape tests
Heinz Mauelshagen [Fri, 4 May 2018 14:56:38 +0000 (16:56 +0200)]
tests: bump raid target version in reshape tests

Adjust to target version allowing tests to succeed.

6 years agotests: don't rely on cache target in component-raid.sh
Heinz Mauelshagen [Fri, 4 May 2018 14:53:16 +0000 (16:53 +0200)]
tests: don't rely on cache target in component-raid.sh

Lead to unnecessary skips of the test.

6 years agoMerge branch 'master' into 2018-04-30-vdo-support
Joe Thornber [Fri, 4 May 2018 12:32:07 +0000 (13:32 +0100)]
Merge branch 'master' into 2018-04-30-vdo-support

6 years agoMerge branch 'master' of git+ssh://sourceware.org/git/lvm2
Joe Thornber [Fri, 4 May 2018 08:49:55 +0000 (09:49 +0100)]
Merge branch 'master' of git+ssh://sourceware.org/git/lvm2

6 years agoRevert "build: Calculate dependencies at same time as compiling."
Joe Thornber [Fri, 4 May 2018 08:48:40 +0000 (09:48 +0100)]
Revert "build: Calculate dependencies at same time as compiling."

This reverts commit 0931067dc5549c443e144d599683b53220021f44.

The dep files should be in the build dir, which is not necc. the src dir.

Easy to fix, but reverting for now until I have time to revisit.

6 years agoclvmd: saved_vg code and comment formatting
David Teigland [Thu, 3 May 2018 16:41:10 +0000 (11:41 -0500)]
clvmd: saved_vg code and comment formatting

6 years agoclvmd: don't save cft and buf for saved_vg
David Teigland [Thu, 3 May 2018 15:50:38 +0000 (10:50 -0500)]
clvmd: don't save cft and buf for saved_vg

6 years agoclvmd: separate saved_vg from vginfo
David Teigland [Wed, 2 May 2018 21:58:49 +0000 (16:58 -0500)]
clvmd: separate saved_vg from vginfo

The clvmd saved_vg data is independent from the normal lvm
lvmcache vginfo data, so separate saved_vg from vginfo.
Normal lvm doesn't need to use save_vg at all, and in clvmd,
lvmcache changes on vginfo can be made without worrying
about unwanted effects on saved_vg.

6 years agoclvmd: defer freeing saved vgs
David Teigland [Mon, 30 Apr 2018 21:48:53 +0000 (16:48 -0500)]
clvmd: defer freeing saved vgs

To avoid the chance of freeing a saved vg while another
code path is using it, defer freeing saved vgs until
all the lvmcache content is dropped for the vg.

6 years agoraid: use new internal APIs
Heinz Mauelshagen [Thu, 3 May 2018 19:35:32 +0000 (21:35 +0200)]
raid: use new internal APIs

Use APIs introduced with commit 4ebfd8e8eb68442efc334b35bc1f22eda3e4dd3d
where appropriate to minimize redundant code.

6 years agoMerge branch '2018-05-03-improve-bcache-utils'
Joe Thornber [Thu, 3 May 2018 19:15:13 +0000 (20:15 +0100)]
Merge branch '2018-05-03-improve-bcache-utils'

6 years agounit-test: a bunch of tests for bcache-utils
Joe Thornber [Thu, 3 May 2018 19:13:13 +0000 (20:13 +0100)]
unit-test: a bunch of tests for bcache-utils

6 years agolvconvert: don't return success on degraded -m raid1 conversion
Heinz Mauelshagen [Thu, 3 May 2018 16:00:21 +0000 (18:00 +0200)]
lvconvert: don't return success on degraded -m raid1 conversion

In case "lvconvert -mN RaidLV" was used on a degraded
raid1 LV, success was returned instead of an error.

Provide message to inform about the need to repair first
before changing number of mirrors and exit with error.

Add new lvconvert-m-raid1-degraded.sh test.

Resolves: rhbz1573960

6 years agoconfigure.ac: bad configure generated due to stray ;;
Joe Thornber [Thu, 3 May 2018 14:38:05 +0000 (15:38 +0100)]
configure.ac: bad configure generated due to stray ;;

6 years agoMerge branch 'master' of git+ssh://sourceware.org/git/lvm2
Joe Thornber [Thu, 3 May 2018 13:23:12 +0000 (14:23 +0100)]
Merge branch 'master' of git+ssh://sourceware.org/git/lvm2

6 years agofunctional tests: Update have_raid function
Joe Thornber [Thu, 3 May 2018 13:22:24 +0000 (14:22 +0100)]
functional tests: Update have_raid function

No need to check if it's built in.

6 years agoconfigure.ac: Remove some more remnants of optional RAID
Joe Thornber [Thu, 3 May 2018 13:21:21 +0000 (14:21 +0100)]
configure.ac: Remove some more remnants of optional RAID

(It's now always 'internal')

6 years agobcache-utils: rewrite
Joe Thornber [Thu, 3 May 2018 10:36:15 +0000 (11:36 +0100)]
bcache-utils: rewrite

They take care to avoid redundant reads now.

6 years agobcache: rename bcache_write_zeroes() -> bcache_zero_bytes()
Joe Thornber [Thu, 3 May 2018 09:21:14 +0000 (10:21 +0100)]
bcache: rename bcache_write_zeroes() -> bcache_zero_bytes()

Now matches the other util functions:

bcache_{prefetch,read,write,zero}_bytes()

6 years agobcache: rewrite bcache_write_zeros()
Joe Thornber [Thu, 3 May 2018 09:14:56 +0000 (10:14 +0100)]
bcache: rewrite bcache_write_zeros()

It now uses GF_ZERO to avoid reading blocks that are going to be
completely zeroed.

6 years agobcache: switch off_t -> uint64_t
Joe Thornber [Thu, 3 May 2018 08:37:43 +0000 (09:37 +0100)]
bcache: switch off_t -> uint64_t

We always want it to be 64bit

6 years agobcache: Move the utils to a separate file.
Joe Thornber [Thu, 3 May 2018 08:34:41 +0000 (09:34 +0100)]
bcache: Move the utils to a separate file.

This makes it clearer that they don't access the cache internals.

6 years agobcache: add bcache_block_sectors() query fn
Joe Thornber [Thu, 3 May 2018 08:33:55 +0000 (09:33 +0100)]
bcache: add bcache_block_sectors() query fn

6 years agobcache: add a comment
Joe Thornber [Thu, 3 May 2018 08:14:59 +0000 (09:14 +0100)]
bcache: add a comment

6 years agofilters: increase MAX_FILTERS for new filter
David Teigland [Wed, 2 May 2018 19:10:30 +0000 (14:10 -0500)]
filters: increase MAX_FILTERS for new filter

The new signature filter was added without increasing this.

6 years agobcache: reorder includes in .c file too
Joe Thornber [Wed, 2 May 2018 18:45:06 +0000 (19:45 +0100)]
bcache: reorder includes in .c file too

6 years agodevice/bcache: reorder includes
Joe Thornber [Wed, 2 May 2018 17:59:43 +0000 (18:59 +0100)]
device/bcache: reorder includes

6 years agobuild: add -D_FILE_OFFSET_BITS=64
Joe Thornber [Wed, 2 May 2018 17:40:38 +0000 (18:40 +0100)]
build: add -D_FILE_OFFSET_BITS=64

I don't like having this in a common header because it means you end
up including too much and causing unneccessary dependencies.  eg,
lib/misc/lib.h includes libdevmapper.h, internationalisation, and
logging stuff.

6 years agounit-test/matcher_t: Fixup Kabi's test
Joe Thornber [Wed, 2 May 2018 12:53:43 +0000 (13:53 +0100)]
unit-test/matcher_t: Fixup Kabi's test

The matcher matches the regexes in reverse order.

6 years agounit-test/matcher_t: add another (failing!) test for Kabi
Joe Thornber [Wed, 2 May 2018 12:31:57 +0000 (13:31 +0100)]
unit-test/matcher_t: add another (failing!) test for Kabi

6 years agovdo: get status parser compiling
Joe Thornber [Wed, 2 May 2018 10:15:35 +0000 (11:15 +0100)]
vdo: get status parser compiling

6 years agoMerge branch 'master' into 2018-04-30-vdo-support
Joe Thornber [Wed, 2 May 2018 09:09:20 +0000 (10:09 +0100)]
Merge branch 'master' into 2018-04-30-vdo-support

6 years agotests: add gfs-pool test
David Teigland [Tue, 1 May 2018 20:24:42 +0000 (15:24 -0500)]
tests: add gfs-pool test

Put back a test like the old one that was removed
in d709d8445f206b67cfc06266ce7553f9cb6f3df5.

It verifies that lvm will ignore and not use a
gfs-pool device.

6 years agodevices: ignore lvm1 and pool devices
David Teigland [Tue, 1 May 2018 19:32:15 +0000 (14:32 -0500)]
devices: ignore lvm1 and pool devices

6 years agoMerge branch 'master' of git+ssh://sourceware.org/git/lvm2
Joe Thornber [Tue, 1 May 2018 19:04:30 +0000 (20:04 +0100)]
Merge branch 'master' of git+ssh://sourceware.org/git/lvm2

6 years agoRevert "build: Stop creating the symlinks in include/ on the fly."
Joe Thornber [Tue, 1 May 2018 19:03:51 +0000 (20:03 +0100)]
Revert "build: Stop creating the symlinks in include/ on the fly."

This reverts commit cdcea0bf55c170f538c3ffc114792eb3c993f4ac.

6 years agotests: remove use of lvm1 metadatatype
David Teigland [Tue, 1 May 2018 18:28:10 +0000 (13:28 -0500)]
tests: remove use of lvm1 metadatatype

6 years agotests: remove gfs pool test
David Teigland [Tue, 1 May 2018 18:25:40 +0000 (13:25 -0500)]
tests: remove gfs pool test

6 years agotests: update lvmetad-disabled to not use lvm1
David Teigland [Tue, 1 May 2018 16:32:15 +0000 (11:32 -0500)]
tests: update lvmetad-disabled to not use lvm1

6 years agobcache_write_bytes needs to be followed by flush
David Teigland [Tue, 1 May 2018 14:33:55 +0000 (09:33 -0500)]
bcache_write_bytes needs to be followed by flush

The improved bcache_write_bytes is not flushing, so
the caller needs to do that.

6 years agolvmcache: fix typo in lvmcache_get_saved_vg
David Teigland [Tue, 1 May 2018 14:06:04 +0000 (09:06 -0500)]
lvmcache: fix typo in lvmcache_get_saved_vg

6 years agounit-test/bcache_t: test was using too large a block size
Joe Thornber [Tue, 1 May 2018 13:17:12 +0000 (14:17 +0100)]
unit-test/bcache_t: test was using too large a block size

6 years agobcache: squash some warnings on rhel6
Joe Thornber [Tue, 1 May 2018 12:21:53 +0000 (13:21 +0100)]
bcache: squash some warnings on rhel6

6 years agounit-test/bcache_t: Use a stripped down fixture for some tests
Joe Thornber [Tue, 1 May 2018 11:54:57 +0000 (12:54 +0100)]
unit-test/bcache_t: Use a stripped down fixture for some tests

6 years agobuild: update ./configure and configure.h.in
Joe Thornber [Tue, 1 May 2018 11:12:07 +0000 (12:12 +0100)]
build: update ./configure and configure.h.in

Fallout from Dave's removal of format1 and pool.

6 years agobcache: rewrite bcache_{write,zero}_bytes
Joe Thornber [Tue, 1 May 2018 11:07:33 +0000 (12:07 +0100)]
bcache: rewrite bcache_{write,zero}_bytes

These are utility functions so should only use the public interface.

Also write_bytes was flushing, which will kill performance.

6 years agoMerge branch 'master' of git+ssh://sourceware.org/git/lvm2
Joe Thornber [Tue, 1 May 2018 09:48:42 +0000 (10:48 +0100)]
Merge branch 'master' of git+ssh://sourceware.org/git/lvm2

6 years agounit-test/io_engine_t: add a little test for bcache_{read,write}_bytes
Joe Thornber [Tue, 1 May 2018 09:47:40 +0000 (10:47 +0100)]
unit-test/io_engine_t: add a little test for bcache_{read,write}_bytes

6 years agounit-test/bcache_t: fixup a test.
Joe Thornber [Tue, 1 May 2018 08:17:55 +0000 (09:17 +0100)]
unit-test/bcache_t: fixup a test.

Problem found with valgrind.

This page took 0.06998 seconds and 5 git commands to generate.