ceph_context: remove unsafe cast for singletons

It was previously assumed that a CephContext singleton would
inherit from CephContext::AssociatedSingletonObject, but it was
not enforced.  This could result in unknown behavior when the
singleton is destroyed due to the implied virtual destructor.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit fb62c78637d7092f48871d943282f45029bd6d29)

osd: conditionally initialize the tracepoint provider

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 0009f343a5daf28358a669836d95660c0d9068d8)

librados: conditionally initialize the tracepoint provider

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 6368c281b53d6175a564725b85516de4b6ae54de)

librbd: conditionally initialize the tracepoint provider

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 1a6eea95bd1ccef53f01cf53e493e652faa203ee)

Conflicts:
src/librbd/librbd.cc: trivial resolution

common: new tracing config options

Config options to enable LTTng-UST tracepoint providers for
the OSD, OSD objectstore, librados, and librbd.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 2ccef89badb1f7a63721a0bf48d05eb4c2d0a97a)

Conflicts:
src/common/config_opts.h: trivial resolution

tracing: merge tracepoint provider makefiles into main makefile

This results in the tracepoint provider shared libraries being
placed in the library path for unittests.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit e11d8d68da27a1c224aa55fc40a72ac6aaa0441c)

Conflicts:
configure.ac: trivial resolution
src/Makefile.am: trivial resolution

tracing: dynamic tracepoint provider helper

The TracepointProvider class is a configuration observer.  When
tracing is enabled, it will dynamically load the associated
tracepoint provider.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit b3d02cc21b4cc40a194f84a9bbbf82cf9e7956d1)

Conflicts:
src/common/Makefile.am: trivial resolution

packaging: add new tracepoint probe shared libraries

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit a7ed8e161b27c7852bf0f3a23a977ebd39137943)

ceph.spec.in: add new tracepoint probe shared libraries

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit f4feee2f9dcd7b94476701b66d1f0bdf6fb6e0c2)

 Conflicts:
ceph.spec.in: trivial resolution

lttng: move tracepoint probes to dynamic libraries

LTTng-UST initializes itself at program load, which means it is
currently always enabled.  This can lead to issues with SElinux
and AppArmor which might restrict access to the necessary device
files.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 4a5305e41677885d45b2b74299113a3d00189c66)

 Conflicts:
src/test/Makefile-client.am: trivial resolution
src/tracing/Makefile.am: trivial resolution

Merge pull request #6580 from dachary/wip-13786-hammer

rbd-replay-* moved from ceph-test-dbg to ceph-common-dbg as well

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6286 from dillaman/wip-13460-hammer

rbd-replay-prep and rbd-replay improvements

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #5790 from SUSE/wip-12932-hammer

Miscellaneous spec file fixes

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6644 from dachary/wip-13812-upgrade

revert: osd: use GMT time for hitsets

Reviewed-by: Kefu Chai <kchai@redhat.com>

revert: osd: use GMT time for hitsets

"Merge pull request #5825 from tchaikov/wip-12848-hammer"

This reverts commit 39544718dc2f09bcfdc632ac72fd2a3cda87687e, reversing
changes made to 4ad97162026e1eb6e6e948ddf3eb39f711431e45.

http://tracker.ceph.com/issues/13812 Fixes: #13812

Signed-off-by: Loic Dachary <loic@dachary.org>

Merge pull request #6402 from SUSE/wip-13621-hammer

CephFS restriction on removing cache tiers is overly strict

Reviewed-by: Greg Farnum <gfarnum@redhat.com>

Merge pull request #6354 from theanalyst/wip-13387-hammer

librbd: reads larger than cache size hang

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

Merge pull request #6289 from dillaman/wip-13461-hammer

librbd: invalidate object map on error even w/o holding lock

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6153 from dachary/wip-13205-hammer

ReplicatedBackend: populate recovery_info.size for clone (bug symptom is size mismatch on replicated backend on a clone in scrub)

Reviewed-by: Samuel Just <sjust@redhat.com>

Merge pull request #6158 from dachary/wip-13336-hammer

osd: we do not ignore notify from down osds

Reviewed-by: Samuel Just <sjust@redhat.com>

Merge pull request #6335 from Abhishekvrshny/wip-13488-hammer

object_info_t::decode() has wrong version

Reviewed-by: Samuel Just <sjust@redhat.com>

Merge pull request #6401 from SUSE/wip-13620-hammer

osd: pg stuck in replay

Reviewed-by: Samuel Just <sjust@redhat.com>

build/ops: rbd-replay moved from ceph-test-dbg to ceph-common-dbg

http://tracker.ceph.com/issues/13785 Fixes: #13785

Signed-off-by: Loic Dachary <loic@dachary.org>
(cherry picked from commit a0204dc47acc0c50223e449fd23a2fc97bfd111a)

    Conflicts:
     debian/control: << 0.94.5-2 is appropriate for hammer

ceph.spec.in: lttng in SLES12 only

Over in the SUSE sector, we are trying to enable the SLE_12 and openSUSE_13.2
build targets. The lttng/babeltrace stuff is currently available only in
SLE_12.

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit bf9ca1eb107c4462d8768faf3ff4db7972384ffd)

Conflicts:
ceph.spec.in
            trivial resolution

ceph.spec.in: fix lttng/babeltrace conditionals

lttng and babeltrace are build dependencies for rbd-replay-prep. Make
sure the right package names are used. Enable for SLE12, as well as
for openSUSE 13.1 and higher.

Move the BuildRequires out of the ceph-test subpackage and into the
distro-conditional dependencies section.

Make ordering of BuildRequires a little more alphabetical.

http://tracker.ceph.com/issues/12360 Fixes: #12360

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit e267128ce22d8b8cd83f6d6d82f24f496600e678)

Conflicts:
ceph.spec.in
            Dropped Requires: xmlstarlet in ceph-test that had inadvertently
            been grabbed from d2cc2b1

packaging: move rbd-replay* to ceph-common

The rbd-replay* utilities are useful for Ceph users with RBD clients.
Currently the rbd-replay* utilities ship in the "ceph-test" package, and
we intend this ceph-test package for Ceph developers and contributors,
not normal users.

Move the rbd-replay* utilities to "ceph-common".

http://tracker.ceph.com/issues/12994 Fixes: #12994

Signed-off-by: Ken Dreyer <kdreyer@redhat.com>
(cherry picked from commit 96486fd6f284ca3ab67f4f94631896d41c529e85)

Conflicts:
ceph.spec.in
            50a33dea has not been backported to hammer
debian/ceph-test.install
            50a33dea has not been backported to hammer
debian/control
            different ceph-test Replaces: and Breaks: version

Merge pull request #6213 from SUSE/wip-13425-hammer

wrong conditional for boolean function KeyServer::get_auth()

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #6336 from Abhishekvrshny/wip-13535-hammer

LibRadosWatchNotify.WatchNotify2Timeout

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6391 from SUSE/wip-13590-hammer

mon: should not set isvalid = true when cephx_verify_authorizer return false

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6398 from SUSE/wip-13588-hammer

OSD::build_past_intervals_parallel() shall reset primary and up_primary when begin a new past_interval.

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6199 from liewegas/wip-randomize-scrub-hammer

osd: randomize scrub times

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #5825 from tchaikov/wip-12848-hammer

ReplicatedPG::hit_set_trim osd/ReplicatedPG.cc: 11006: FAILED assert(obc)

Reviewed-by: Loic Dachary <ldachary@redhat.com>

ReplicatedBackend::prepare_pull: set recover_info.size for clones

Fixes: #12828
Signed-off-by: Samuel Just <sjust@redhat.com>
(cherry picked from commit 1d3e46bedb9a8c0d2200e39f62f4e2c3337619f3)

Merge pull request #6157 from dachary/wip-13040-hammer

common/Thread:pthread_attr_destroy(thread_attr) when done with it

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6155 from dachary/wip-13171-hammer

objecter: cancellation bugs

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6152 from dachary/wip-13233-hammer

mon: include min_last_epoch_clean as part of PGMap::print_summary and PGMap::dump

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6146 from dachary/wip-13337-hammer

segfault in agent_work

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6145 from dachary/wip-13338-hammer

filestore: fix peek_queue for OpSequencer

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6144 from dachary/wip-13339-hammer

mon: check for store writeablility before participating in election

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #5891 from Abhishekvrshny/wip-13035-hammer

requeue_scrub when kick_object_context_blocked

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #5890 from Abhishekvrshny/wip-13037-hammer

hit set clear repops fired in same epoch as map change -- segfault since they fall into the new interval even though the repops are cleared

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6322 from sponce/WIP-13210-hammer

tests : BACKPORT #13210 Fixed broken Makefiles after integration of ttng into rados

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #5810 from Abhishekvrshny/wip-12948-hammer

Heavy memory shuffling in rados bench

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #5530 from SUSE/wip-12587-hammer

FileStore calls syncfs(2) even it is not supported

Reviewed-by: Kefu Chai <kchai@redhat.com>

Merge pull request #5889 from Abhishekvrshny/wip-13042-hammer

ThreadPool add/remove work queue methods not thread safe

Reviewed-by: Josh Durgin <jdurgin@redhat.com>

Merge pull request #6151 from dachary/wip-13245-hammer

client nonce collision due to unshared pid namespaces

Reviewed-by: Josh Durgin <jdurgin@redhat.com>

Merge pull request #6159 from dachary/wip-13341-hammer

ceph upstart script rbdmap.conf incorrectly processes parameters

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6154 from dachary/wip-13195-hammer

should recalc the min_last_epoch_clean when decode PGMap

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6132 from SUSE/wip-13307-hammer

dumpling incrementals do not work properly on hammer and newer

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6097 from dillaman/wip-13045-hammer

librbd: diff_iterate needs to handle holes in parent images

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #6060 from XinzeChi/wip-hammer-outdata-set

bug fix: osd: avoid multi set osd_op.outdata in tier pool

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #5897 from tchaikov/wip-12940-hammer

IO error on kvm/rbd with an erasure coded pool tier

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #5787 from SUSE/wip-12923-hammer

logrotate reload error on Ubuntu 14.04

Reviewed-by: Loic Dachary <ldachary@redhat.com>

Merge pull request #5766 from dachary/wip-12590-hammer

ceph mds add_data_pool check for EC pool is wrong

Reviewed-by: Greg Farnum <gfarnum@redhat.com>

Merge pull request #5361 from ceph/wip-11786.hammer

mon: MonitorDBStore: get_next_key() only if prefix matches

Reviewed-by: Loic Dachary <ldachary@redhat.com>

qa: update cephtool test for CephFS tier cases

1. Creating a filesystem using a
   readonly tier on an EC pool (should be forbidden)
2. Removing a tier from a replicated base pool (should
   be permitted)

Signed-off-by: John Spray <john.spray@redhat.com>
(cherry picked from commit f9b11b0abe7c2b0dcdfc5802336be8a0426ccb3a)

mon: forbid readonly cache tiers for CephFS

For EC pools, we need a tier that is not just forwarding
writes, in order to provide the ops required
by CephFS.

Signed-off-by: John Spray <john.spray@redhat.com>
(cherry picked from commit 11e5faf3c1d4fa28a8b41e3a92f0f684a844e5f5)

mon: refine check_remove_tier checks

Fixes: #11504
Signed-off-by: John Spray <john.spray@redhat.com>
(cherry picked from commit a50c8f1f2ad8845c7f77110868f9376f2d1ff883)

osd: fix requeue of replay requests during activating

If the replay period expires while we are still in the activating
state, we can simply insert our list of requests at the front of
the waiting_for_active list.

Fixes: #13116
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit d18cf51d9419819cdda3782b188b010969288911)

OSD:shall reset primary and up_primary fields when beginning a new past_interval.

Shall reset primary and up_primary fields when we start over a new past_interval in OSD::build_past_intervals_parallel().
Fixes: #13471
Signed-off-by: xie.xingguo@zte.com.cn
(cherry picked from commit 65064ca05bc7f8b6ef424806d1fd14b87add62a4)

mon: should not set isvalid = true when cephx_verify_authorizer return false

Fixes: #13525
Signed-off-by: Ruifeng Yang <yangruifeng.09209@h3c.com>
(cherry picked from commit c7f75b8f7c0a773148ec16141941efd00ee76626)

0.94.5

ceph.spec.in: rm EOL Fedoras; add OBS RHEL5 instead

Fedora 12 has been EOL for a long time. Remove the reference in the
RPM .spec file.

Since RHEL 5 support for Ceph is a work in progress, we won't remove
this entire python_sitelib / python_sitearch conditional for now, since
those are still needed on RHEL 5.

Add the rhel_version macro to make the conditional compatible with
SUSE's OBS.

Signed-off-by: Ken Dreyer <kdreyer@redhat.com>
(cherry picked from commit 353a3258968f76deaea4542d2165e9ab32b88de8)

packaging: Add rgw placeholder dirs in proper packages

Automatically create /var/lib/ceph/radosgw with the
ceph-radosgw package, and /var/lib/ceph/bootstrap-rgw
via the ceph package.

Signed-off-by: Travis Rhoden <trhoden@redhat.com>
(cherry picked from commit bf93128f664ac74383806530a1f974edde8df5b6)

Conflicts:
    ceph.spec.in
        hammer uses %ghost

ceph.spec.in: consolidate centos/rhel macros

0%{?rhel} || 0%{?centos} can be simplified to just 0%{?rhel}, since
CentOS defines that macro for compatibility with RHEL.

Signed-off-by: Nathan Cutler <ncutler@suse.cz>
(cherry picked from commit 46404dd381e240634d224e77adc69f0ba4807045)

Conflicts:
ceph.spec.in
            https://github.com/ceph/ceph/pull/4068 has not been backported

packaging: make logrotate a hard dependency across all distros

Given ceph's reputation for voluminous logs, it makes sense for
logrotate to be a hard dependency for all distros.

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit 0734cd1ec3ca3bd012686dc1bdda675909224a49)

ceph.spec.in: add missing -%{release}

We have it everywhere else and I can't think of any reason why
it should be omitted here.

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit d952d59ae95d5f66aa4b98fd7aa4f4ecb36a4848)

ceph.spec.in: remove duplicate BuildRequires: sharutils

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit c418bd9615e6f23f3c14ae9e71003c66d31d4b78)

ceph.spec.in: fix python-flask dependency for SUSE

In SLE and openSUSE, the package is called python-Flask with an upper-case F.

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit 39c28b9faf0fb069d91ac069a4d8df7bf03fb490)

Conflicts:
ceph.spec.in
            add missing SUSE-specific build deps

ceph.spec.in: clarify two important comments

First, the terms "common" and "specific" are vague. Second,
"common" can easily be confused with the ceph-common subpackage.

Fix this by rephrasing to "distro-unconditional dependencies" and
"distro-conditional dependencies", respectively.

Third, move the "distro-unconditional dependencies" header so it
is above the part where these dependencies are actually defined.

Signed-off-by: Nathan Cutler <ncutler@suse.com>
(cherry picked from commit dd212fd1d367f70c61f65a97599959668b44aa53)

ceph.spec.in: rm reference to EOL Fedoras (< 20)

Fedora 19 is EOL, so there's no need to refer to Fedoras less than 20.
"%{fedora}" effectively implies "Fedora 20 and above".

Reported-by: Ira Cooper <ira@samba.org>
Signed-off-by: Ken Dreyer <kdreyer@redhat.com>
(cherry picked from commit f3e4a91c458ca7b11706fe4608cb739499f5d3cc)

Conflicts:
ceph.spec.in
           added xmlstarlet dependency to ceph-test package

ceph.spec.in: package rbd-replay-prep on all Fedoras

This reverts the change in commit
85517d611b7bf4cb6cbffcd2c65303be0d038264. Since we BuildRequire:
libbabeltrace-devel, autoconf will see that babeltrace is available
during the build, and make will build/install the rbd-replay-prep
utility.

This change also simplifies Fedora selection logic, because Fedora 19 is
EOL, so "%{fedora}" implies "Fedora 20 and above".

Reported-by: Ira Cooper <ira@samba.org>
Signed-off-by: Ken Dreyer <kdreyer@redhat.com>
(cherry picked from commit 2db9480fa2646de776f0076d82747e202e574ddd)

ceph.spec.in: rbd-replay-prep is not being built on f21+

Signed-off-by: Boris Ranto <branto@redhat.com>
(cherry picked from commit 85517d611b7bf4cb6cbffcd2c65303be0d038264)

ceph.spec.in:BuildRequires sharutils

The uudecode binary is used to build Java-related components, and
uudecode is provided by the sharutils package on all supported
RPM platforms. When building with "--without=cephfs_java",
sharutils is not needed.

Thanks to Nathan Cutler <ncutler@suse.cz> for going into the
details with me.

On OBS without this patch we get the error message:

[  170s] -----------------------------------------------------------------
[  170s] ----- building ceph.spec (user abuild)
[  170s] -----------------------------------------------------------------
[  170s] -----------------------------------------------------------------
[  170s] + exec rpmbuild -ba --define '_srcdefattr (-,root,root)' --nosignature --define 'disturl obs://build.opensuse.org/home:osynge:ceph:wip:wip_obs_fedora/Fedora_20/c0bbbc1e62228ca956ac3d367edc4fba-master' /home/abuild/rpmbuild/SOURCES/ceph.spec
[  170s] error: Failed build dependencies:
[  170s]    sharutils is needed by ceph-1:2+git.1435043747.c1bd02c-1.1.x86_64

With this patch we can build fedora 22 and fedora 20 rpms fine.

Signed-off-by: Owen Synge <osynge@suse.com>
(cherry picked from commit 43c1784640538b813f0575c50f721ac911a30b3e)

Conflicts:
ceph.spec.in
            _with_systemd -> trivial resolution

ceph.spec.in: fix:Add missing directories breaking build

SUSE builds on OBS are failing with the missing dir entries:

    /usr/share/ceph
    /usr/lib/ceph

On suse these correspond to:

    %dir %{_datarootdir}/ceph/
    %dir %{_libexecdir}/ceph/

Signed-off-by: Owen Synge <osynge@suse.com>
(cherry picked from commit f1ff3e4dcb5dd7eddd2c346a05cfa78d860a5ce6)

Merge pull request #6348 from dillaman/wip-13567-hammer

librbd: potential assertion failure during cache read

Reviewed-by: Sage Weil <sage@redhat.com>

Merge pull request #6362 from liewegas/wip-shut-up-osd-hammer

osd/ReplicatedPG: remove stray debug line

Reviewed-by: Loic Dachary <ldachary@redhat.com>

osd/ReplicatedPG: remove stray debug line

This snuck in

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit ff9600a6cef613d40e875597b6392778df1bb04c)

librbd: potential assertion failure during cache read

It's possible for a cache read from a clone to trigger a writeback if a
previous read op determined the object doesn't exist in the clone,
followed by a cached write to the non-existent clone object, followed
by another read request to the same object.  This causes the cache to
flush the pending writeback ops while not holding the owner lock.

Fixes: #13559
Backport: hammer
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 4692c330bd992a06b97b5b8975ab71952b22477a)

tests: reproduce crash during read-induced CoW

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 382219b6bba09156f8cf3d420c6348738e7ed4c6)

Merge pull request #6330 from dachary/wip-13550-hammer

qemu workunit refers to apt-mirror.front.sepia.ceph.com

Reviewed-by: Jason Dillaman <dillaman@redhat.com>

librbd:reads larger than cache size hang.

Fixes:#13164

Signed-off-by: Lu Shi <shi.lu@h3c.com>
(cherry picked from commit 9c8200bb5d1ac9359803a182df03298b565b8479)

radosstriper : Fixed broken Makefiles after integration of lttng

Fixes: #13210, backport of #6322 into hammer
Signed-off-by: Sebastien Ponce <sebastien.ponce@cern.ch>

osdc/Objecter: distinguish between multiple notify completions

We may send a notify to the cluster multiple times due to OSDMap
changes.  In some cases, earlier notify attempts may complete with
an error, while later attempts succeed.  We need to only pay
attention to the most-recently send notify's completion.

Do this by making note of the notify_id in the initial ACK (only
present when talking to newer OSDs).  When we get a notify
completion, match it against our expected notify_id (if we have
one) or else discard it.

This is important because in some cases an early notify completion
may be an error while a later one succeeds.

Note that if we are talking to an old cluster we will simply not record a
notify_id and our behavior will be the same as before (we will trust any
notify completion we get).

Fixes: #13114
Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit e86d033854c76f344c678e92016c4e5c5e0385e2)

Conflicts:
src/osdc/Objecter.cc
In Objecter::handle_watch_notify, a conflict was there due to a modified comment by commit 47277c51db7bb2725ea117e4e8834869ae93e006, which was not backported

osd: reply to notify request with our unique notify_id

The OSD assigns a unique ID to each notify it queues for
processing.  Include this in the reply to the notifier so that
they can match it up with the eventual completions they receive.

This is necessary to distinguish between multiple completions
they may receive if there is PG peering and the notify is resent.
In particular, an earlier notify may return an error when a later
attempt succeeds.

This is forwards and backwards compatible: new clients will make use of
this reply payload but older clients ignore it.

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 049ea702b9bf4a8a62ae2770d9ba55c0730b3eef)

osd: Correct the object_info_t::decode() version

Caused by: 13b9dc70

Fixes: #13462
Backport: Hammer, Firefly

Signed-off-by: David Zafman <dzafman@redhat.com>
(cherry picked from commit 2b7ddde685213f3908f2dd9545b35f1d97f35a79)

qa: Use public qemu repo

This would allow some external tests outside of sepia lab

Signed-off-by: Yuan Zhou <yuan.zhou@intel.com>
(cherry picked from commit 4731c1e35539c9506ff3fe3141553fad099d0eee)

0.94.4

librbd: invalidate object map on error even w/o holding lock

If there is a deep bug that prevents the object map from owning
the lock before objects are updated, flag the object map as invalid
regardless.

Fixes: #13372
Backport: hammer
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit e675400df7f8ba796d60e9bac4234857f0cb1392)

 Conflicts:
src/librbd/ObjectMap.h/cc: hammer branch doesn't support rebuild

Merge pull request #6287 from jcsp/wip-test-fixes

git path backports for hammer

use git://git.ceph.com

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit 34c467ba06345eacd2fd0e79c162dfd0b22f0486)

qa: http://ceph.com/qa -> http://download.ceph.com/qa

Signed-off-by: Sage Weil <sage@redhat.com>
(cherry picked from commit d4d65fbd6ad15320339751c604613ac63511e645)

tests: removed obsolete rbd_replay test cases

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit e0d8cb1283e4bdbfc770684b173e477c22558524)

ceph-dencoder: new rbd_replay trace file types

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit c2a83d0b518497dc8bcbbdea77b88febe52a3034)

rbd-replay: added version control to trace output file

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 3ecdae8388d69123b937a40ce614a0b795a757f1)

rbd-replay-prep: added --verbose command line option

This new command-line will dump all the processed events as they
are extracted from the trace file.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 646e50a771c3089121ec2f4369d1a8188001a3eb)

rbd-replay-prep: stream events to the prep file

Avoid building the entire prep event history in memory before
attempting to write the prepped trace file.

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 98f513a113f4e8aff17c83991d2e6f0f3738a9c9)

rbd-replay-prep: simplify IO dependency calculation

Only track read-after-write and write-after-write IO dependencies
via the associated write completions.  All IO events after a write
completion are considered to be dependent and can be pruned down
to at most the number of concurrent IOs.  This reduces the prep
time from a simple 'rbd bench-write' from over 4 hrs down to seconds.

Fixes: #13378, #13384
Backport: hammer

Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 65fb1b86cbab65023f6207798c9e189bce55dcf6)

rbd-replay: improve error messages

Fixes: #13221
Backport: hammer
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit 4f436e10802d7b3c3949b592fec238c189a02c50)

rbd-replay-prep: support new read/write APIs

Added support for read2, write2, aio_read2, and aio_write2 APIs.

Fixes: 13325
Backport: hammer
Signed-off-by: Jason Dillaman <dillaman@redhat.com>
(cherry picked from commit b5b4a9dd6bfe10844327a457849a0dbf7ef597ea)