client: in UserPerm, correctly use delete[] and don't leak gids on copy

gids is an array. We also need to clean it up when deep_copy()ing, which
requires initializing our member fields in the copy-constructor.

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

mds: sort GID lists in MDSAuthCaps

Fixes: http://tracker.ceph.com/issues/17368
Signed-off-by: Greg Farnum <gfarnum@redhat.com>

test: temporarily disable tests which relied on MDS' incorrect GID-adding behavior

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

Merge remote-tracking branch 'origin/master' into wip-getuid

Conflicts:
src/client/Client.cc
src/client/Client.h
src/client/Makefile.am
src/client/fuse_ll.cc

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

client: use uid_t/gid_t in _getgrouplist

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

test: basic checks on gid list is_capable

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

mds: do not make use of GIDs which aren't included by the Client in its request

We were previously iterating through all the GIDs allowed by a particular
MDSAuthCap::match, but there are plenty of situations where a GID might
be *allowed* but not actually assigned. Take the set intersection of what's
allowed and what's specified instead.

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

mds: pass full gid list to MDSAuthCap checks, when present

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

libcephfs: repeatedly use new default_perms instead of pick_my_perms()

There's no need to invoke it each time for library-using clients. Either they've
set the default uid/gid or they haven't, and they aren't going to be switching
dynamic uid/gid a bunch either. (Right? Anyway, we want them to use the
not-here-yet functions specifying uid/gid on each call.)

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

client: do a shallow copy for MetaRequest::perms

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

client: include UserPerm contents in may_* debug output

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

Client: always pass a Fh to _do_filelock

We weren't passing in the Fh as a signal we were releasing locks (ie, don't
remove the lock ranges in question rather than adding htem to our accounting).
But that was a little weird and doesn't work when we're pulling perms out of
the Fh.

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

client: add a static pick_my_perms taking a CephContext

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

fuse: optionally (off by default) set the group lists on calls

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

client: give UserPerm a move constructor

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

MClientRequest: include the full gid list of the caller

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

client: remove the now-unused UserGroups entirely

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

client: stop using UserGroups for may_setattr() now that UserPerm does it all

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

client: Actually handle gid lists in UserPerm

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

client: stop using UserGroups for posix ACL checks

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

client: push UserGroups into _posix_acl_permission(), instead of passing it

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

client: switch Inode::check_mode() to UserPerm

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

client: switch inode_permission() to UserPerm-based (over UserGroups)

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

UserPerm: add trivial implementation of UserGroup interface equivalents

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

client: remove get_(uid|gid) functions

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

client: remove now-unused uid-based may_* functions

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

client: pass UserPerm to mkdirs, rmdir, symlink, mknod

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

client: remove now-unused owner_(uid|gid) from dir_result_t

client: replace Inode's "best" logic with a UserPerm over gid/uid pair

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

client: include UserPerm.h in MetaRequest.h

client: move UserPerm into its own header file

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

client: add a UserPerm to MetaRequest

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

client: remove uid-based _create()

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

client: remove uid-based _geattr() functions

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

client: call _getattr() with UserPerm in posix checkers

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

client: switch _posix_acl_create() to UserPerm

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

client: use Fh::actor_perms for _read() and _write()

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

client: remove uid-based path_walk()

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

client: pass UserPerm to get_caps_issued()

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

client: pass UserPerm to describe_layout()

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

client: pass UserPerm to ll_walk()

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

client: pass UserPerm to all open() variants

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

client: pass UserPerm to readlink()

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

client: use UserPerm for _posix_acl_chmod

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

client: pass UserPerm to get_quota_root() and quota helpers

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

client: use Fh::actor_perm for _write()

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

client: fill in _interrupt_filelock() with MetaRequest's uid/gid

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

client: store "actor" UserPerm in Fh, use for requesting file locks and things

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

client: fill in cap renewals with the "best" uid/gid perms

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

client: track which request uid/gid granted a cap

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

client: pass UserPerm to lookup_name()

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

client: pass UserPerm to lookup_parent()

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

client: pass UserPerm to lookup_ino()

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

client: pass UserPerm to lookup_hash()

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

client: store UserPerms in dir_result_t and use on MDS requests

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

client: pass UserPerm to statfs() variants

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

client: pass UserPerm to chdir() and getcwd()

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

client: remove uid/gid-based make_request, push up logic

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

client: remove client reference from RequestUserGroups

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

client: remove spurious RequestUserGroups from xattr_permission

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

client: switch xattr_permission() implementation to UserPerm

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

client: switch _create() implementation to UserPerm

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

client: switch _open() implementation to UserPerm

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

client: switch _getattr() implementation to UserPerm

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

client: switch path_walk() implementation to UserPerm

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

client: switch _lookup() implementation to UserPerm

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

client: switch may_*() function implementations to UserPerm

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

client: switch make_request implementation and verify_reply_trace to UserPerm

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

client: switch _do_lookup() to UserPerm

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

client: add a UserPerm param to mount

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

client: remove unused ll_listxattr_chunks declaration

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

client: switch ll_open and ll_create to be UserPerm-based

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

client: generate UserPerm _open and _create variants

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

client: always pass a UserPerm to seek variants

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

client: always pass a UserPerm to stat variants

Also remove the lstatlite() declaration, since it's not defined.

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

client: remove non-UserPerm _setattr implementation

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

client: always pass a UserPerm to truncate variants

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

client: always pass a UserPerm to utime variants

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

client: always pass a UserPerm to chown variants

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

Merge pull request #11122 from branch-predictor/bp-zlib-improvements

compressor/zlib: switch to raw deflate

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

Merge pull request #11108 from liewegas/wip-bluefs-dirty-fnodes

os/bluestore/BlueFS: log dirty files at sync time

Merge pull request #11173 from tchaikov/wip-17306

test/cli/crushtool: fix the test of compile-decompile-recompile.t

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

test/cli/crushtool: fix the test of compile-decompile-recompile.t

should read the map from $TESTDIR,
it's a regression introduced by b2c0a07

Fixes: http://tracker.ceph.com/issues/17306
Signed-off-by: Kefu Chai <kchai@redhat.com>

Merge pull request #11166 from xiexingguo/xxg-wip-bluestore-0921

os/bluestore: simplify calculation of collection key range

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

Merge pull request #11131 from ceph/rgw-lifecycle-testing

Rgw lifecycle testing

Reviewed-by: Casey Bodley <cbodley@redhat.com>

Lifecycle testing

Add new config option - rgw_lc_debug_interval - that modifies the
lifecycle algorithm.  It treats days as being the configured number of
seconds.  This allows lifecycle to be tested in minutes, rather than
days.

Signed-off-by: Daniel Gryniewicz <dang@redhat.com>

Lifecycle - Fix config locking

The lifecycle config lock was not setting a cookie.  This means that all
instances of that lock conflict, causing errors to be returned to the
client.

Set a unique cookie on each lock instance.

Signed-off-by: Daniel Gryniewicz <dang@redhat.com>

Merge pull request #11031 from yangdongsheng/du

rbd: return error if we specified a wrong image name for rbd du

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

Merge pull request #11163 from jcsp/wip-cmake-vstart

doc: Update developer docs for cmake paths

Reviewed-by: Brad Hubbard <bhubbard@redhat.com>

doc: update vstart guide for cmake-ish paths

Signed-off-by: John Spray <john.spray@redhat.com>

doc: update build instructions for cmake

Signed-off-by: John Spray <john.spray@redhat.com>

os/bluestore: simplify calculation of collection key range

Signed-off-by: xie xingguo <xie.xingguo@zte.com.cn>

Merge pull request #11083 from ifed01/wip-load-plugins-fix

common/compressor: add libcommon as a dependency for zlib and snappy p…

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

Merge pull request #10975 from stiopaa1/osd_replicatedbackend_moveSomeClassesToCC

osd/PGBackend.h: move structs to .cc file

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

Merge pull request #11073 from badone/wip-remove-check_disk_write_cache

FileJournal: Remove obsolete _check_disk_write_cache function

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

Merge pull request #11060 from stiopaa1/mon_osdmonitor_removeAssign

mon/OSDMonitor.cc: remove use of boost assign

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

Merge pull request #11039 from stiopaa1/osd_inprogress_removingCount

osd/OSD.cc: remove unneeded searching of maps

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

Merge branch 'wip-rocksdb-librados-test'

qa: add test of envlibrados for rocksdb

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

add qa test of envlibrados for rocksdb

Signed-off-by: Ryne Li <lizhenqiangsnake@gmail.com>

Merge pull request #11146 from tchaikov/wip-17307

os/filestore/FileJournal: error out if FileJournal is not a file

Reviewed-by: David Zafman <dzafman@redhat.com>