The intent behind making it a per-request setting was that it would be
set for writes, but not for reads.  As it is, the flag is set for all
fs/ceph requests except for pool perm check stat request (technically
a read).

ceph_osdc_abort_on_full() skips reads since the previous commit and
I don't see a use case for marking individual requests.
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>

Don't consider reads for aborting and use ->base_oloc instead of
->target_oloc, as done in __submit_request().

Strictly speaking, we shouldn't be aborting FULL_TRY/FULL_FORCE writes
either.  But, there is an inconsistency in FULL_TRY/FULL_FORCE handling
on the OSD side [1], so given that neither of these is used in the
kernel client, leave it for when the OSD behaviour is sorted out.

[1] http://tracker.ceph.com/issues/24339Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>

Sending map check after complete_request() was called is not only
useless, but can lead to a use-after-free as req->r_kref decrement in
__complete_request() races with map check code.
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>

The "FULL or reached pool quota" warning is there to explain paused
requests.  No need to emit it if pausing isn't going to occur.
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>

Scanning the trees just to see if there is anything to abort is
unnecessary -- all that is needed here is to update the epoch barrier
first, before we start aborting.  Simplify and do the update inside the
loop before calling abort_request() for the first time.

The switch to for_each_request() also fixes a bug: homeless requests
weren't even considered for aborting.
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>

In the common case, req->r_callback is called by handle_reply() on the
ceph-msgr worker thread without any locks.  If handle_reply() fails, it
is called with both osd->lock and osdc->lock.  In the map check case,
it is called with just osdc->lock but held for write.  Finally, if the
request is aborted because of -ENOSPC or by ceph_osdc_abort_requests(),
it is called directly on the submitter's thread, again with both locks.

req->r_callback on the submitter's thread is relatively new (introduced
in 4.12) and ripe for deadlocks -- e.g. writeback worker thread waiting
on itself:

  inode_wait_for_writeback+0x26/0x40
  evict+0xb5/0x1a0
  iput+0x1d2/0x220
  ceph_put_wrbuffer_cap_refs+0xe0/0x2c0 [ceph]
  writepages_finish+0x2d3/0x410 [ceph]
  __complete_request+0x26/0x60 [libceph]
  complete_request+0x2e/0x70 [libceph]
  __submit_request+0x256/0x330 [libceph]
  submit_request+0x2b/0x30 [libceph]
  ceph_osdc_start_request+0x25/0x40 [libceph]
  ceph_writepages_start+0xdfe/0x1320 [ceph]
  do_writepages+0x1f/0x70
  __writeback_single_inode+0x45/0x330
  writeback_sb_inodes+0x26a/0x600
  __writeback_inodes_wb+0x92/0xc0
  wb_writeback+0x274/0x330
  wb_workfn+0x2d5/0x3b0

Defer __complete_request() to a workqueue in all failure cases so it's
never on the same thread as ceph_osdc_start_request() and always called
with no locks held.

Link: http://tracker.ceph.com/issues/23978Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>

Move req->r_completion wake up and req->r_kref decrement into
__complete_request().
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>

destroy_workqueue() drains the workqueue before proceeding with
destruction.
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

Pending works hold inode references, which cause "Busy inodes after
unmount" warning.
Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

This avoid force umount waiting on page writeback:

  io_schedule+0xd/0x30
  wait_on_page_bit_common+0xc6/0x130
  __filemap_fdatawait_range+0xbd/0x100
  filemap_fdatawait_keep_errors+0x15/0x40
  sync_inodes_sb+0x1cf/0x240
  sync_filesystem+0x52/0x90
  generic_shutdown_super+0x1d/0x110
  ceph_kill_sb+0x28/0x80 [ceph]
  deactivate_locked_super+0x35/0x60
  cleanup_mnt+0x36/0x70
  task_work_run+0x79/0xa0
  exit_to_usermode_loop+0x62/0x70
  do_syscall_64+0xdb/0xf0
  entry_SYSCALL_64_after_hwframe+0x44/0xa9
  0xffffffffffffffff
Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

This will be used by the filesystem for "umount -f".
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

Currently, calling stat on a cephfs directory returns 1 for st_nlink.
This behaviour has recently changed in the fuse client, as some
applications seem to expect this value to be either 0 (if it's
unlinked) or 2 + number of subdirectories.  This behaviour was changed
in the fuse client with commit 67c7e4619188 ("client: use common
interp of st_nlink for dirs").

This patch modifies the kernel client to have a similar behaviour.

Link: https://tracker.ceph.com/issues/23873Signed-off-by: NLuis Henriques <lhenriques@suse.com>
Reviewed-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

Link: http://tracker.ceph.com/issues/24028Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

This is how it was before commit 95cca2b4 ("ceph: limit osd write
size") went in.
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

Without these new fields, stale st_size is returned in following
case.

1. MDS modifies a directory
2. MDS issues CEPH_CAP_ANY_SHARED to client
3. The client satifies stat(2) by its cached metadata. set st_size
   to "i_files + i_subdirs".

Link: http://tracker.ceph.com/issues/23855Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

The data structure includes the versioned feilds of cap message.
Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

In MDS, file/subdir counts of a directory inode are protected by
filelock. In request reply without Fs cap, nfiles/nsubdirs can be
stale.
Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

rstat is not tracked by capability. client can't know if rstat from
non-auth mds is uptodate or not.

Link: http://tracker.ceph.com/issues/23538Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

Signed-off-by: N"Yan, Zheng" <zyan@redhat.com>
Acked-by: NJeff Layton <jlayton@redhat.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

Avoid a copy into the "skip buffer".
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

All gotos to "more" are conditioned on con->state == OPEN, but the only
thing "more" does is opening the socket if con->state == PREOPEN.  Kill
that label and rename "more_kvec" to "more".
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>
Reviewed-by: NJason Dillaman <dillaman@redhat.com>

Add proper error handling for osd_req_op_cls_init() to replace
BUG_ON statement when failing from memory allocation.
Signed-off-by: NChengguang Xu <cgxu519@gmx.com>
Reviewed-by: NIlya Dryomov <idryomov@gmail.com>
Signed-off-by: NIlya Dryomov <idryomov@gmail.com>

Pull vfs fixes from Al Viro.

 - fix io_destroy()/aio_complete() race

 - the vfs_open() change to get rid of open_check_o_direct() boilerplate
   was nice, but buggy. Al has a patch avoiding a revert, but that's
   definitely not a last-day fodder, so for now revert it is...

* 'fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  Revert "fs: fold open_check_o_direct into do_dentry_open"
  fix io_destroy()/aio_complete() race

This reverts commit cab64df1.

Having vfs_open() in some cases drop the reference to
struct file combined with

	error = vfs_open(path, f, cred);
	if (error) {
		put_filp(f);
		return ERR_PTR(error);
	}
	return f;

is flat-out wrong.  It used to be

		error = vfs_open(path, f, cred);
		if (!error) {
			/* from now on we need fput() to dispose of f */
			error = open_check_o_direct(f);
			if (error) {
				fput(f);
				f = ERR_PTR(error);
			}
		} else {
			put_filp(f);
			f = ERR_PTR(error);
		}

and sure, having that open_check_o_direct() boilerplate gotten rid of is
nice, but not that way...

Worse, another call chain (via finish_open()) is FUBAR now wrt
FILE_OPENED handling - in that case we get error returned, with file
already hit by fput() *AND* FILE_OPENED not set.  Guess what happens in
path_openat(), when it hits

	if (!(opened & FILE_OPENED)) {
		BUG_ON(!error);
		put_filp(file);
	}

The root cause of all that crap is that the callers of do_dentry_open()
have no way to tell which way did it fail; while that could be fixed up
(by passing something like int *opened to do_dentry_open() and have it
marked if we'd called ->open()), it's probably much too late in the
cycle to do so right now.
Signed-off-by: NAl Viro <viro@zeniv.linux.org.uk>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Pull scheduler fixes from Thomas Gleixner:

 - two patches addressing the problem that the scheduler allows under
   certain conditions user space tasks to be scheduled on CPUs which are
   not yet fully booted which causes a few subtle and hard to debug
   issue

 - add a missing runqueue clock update in the deadline scheduler which
   triggers a warning under certain circumstances

 - fix a silly typo in the scheduler header file

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/headers: Fix typo
  sched/deadline: Fix missing clock update
  sched/core: Require cpu_active() in select_task_rq(), for user tasks
  sched/core: Fix rules for running on online && !active CPUs

Pull perf tooling fixes from Thomas Gleixner:

 - fix 'perf test Session topology' segfault on s390 (Thomas Richter)

 - fix NULL return handling in bpf__prepare_load() (YueHaibing)

 - fix indexing on Coresight ETM packet queue decoder (Mathieu Poirier)

 - fix perf.data format description of NRCPUS header (Arnaldo Carvalho
   de Melo)

 - update perf.data documentation section on cpu topology

 - handle uncore event aliases in small groups properly (Kan Liang)

 - add missing perf_sample.addr into python sample dictionary (Leo Yan)

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf tools: Fix perf.data format description of NRCPUS header
  perf script python: Add addr into perf sample dict
  perf data: Update documentation section on cpu topology
  perf cs-etm: Fix indexing for decoder packet queue
  perf bpf: Fix NULL return handling in bpf__prepare_load()
  perf test: "Session topology" dumps core on s390
  perf parse-events: Handle uncore event aliases in small groups properly

Pull networking fixes from David Miller:

 1) Infinite loop in _decode_session6(), from Eric Dumazet.

 2) Pass correct argument to nla_strlcpy() in netfilter, also from Eric
    Dumazet.

 3) Out of bounds memory access in ipv6 srh code, from Mathieu Xhonneux.

 4) NULL deref in XDP_REDIRECT handling of tun driver, from Toshiaki
    Makita.

 5) Incorrect idr release in cls_flower, from Paul Blakey.

 6) Probe error handling fix in davinci_emac, from Dan Carpenter.

 7) Memory leak in XPS configuration, from Alexander Duyck.

 8) Use after free with cloned sockets in kcm, from Kirill Tkhai.

 9) MTU handling fixes fo ip_tunnel and ip6_tunnel, from Nicolas
    Dichtel.

10) Fix UAPI hole in bpf data structure for 32-bit compat applications,
    from Daniel Borkmann.

* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net: (33 commits)
  bpf: fix uapi hole for 32 bit compat applications
  net: usb: cdc_mbim: add flag FLAG_SEND_ZLP
  ip6_tunnel: remove magic mtu value 0xFFF8
  ip_tunnel: restore binding to ifaces with a large mtu
  net: dsa: b53: Add BCM5389 support
  kcm: Fix use-after-free caused by clonned sockets
  net-sysfs: Fix memory leak in XPS configuration
  ixgbe: fix parsing of TC actions for HW offload
  net: ethernet: davinci_emac: fix error handling in probe()
  net/ncsi: Fix array size in dumpit handler
  cls_flower: Fix incorrect idr release when failing to modify rule
  net/sonic: Use dma_mapping_error()
  xfrm Fix potential error pointer dereference in xfrm_bundle_create.
  vhost_net: flush batched heads before trying to busy polling
  tun: Fix NULL pointer dereference in XDP redirect
  be2net: Fix error detection logic for BE3
  net: qmi_wwan: Add Netgear Aircard 779S
  mlxsw: spectrum: Forbid creation of VLAN 1 over port/LAG
  atm: zatm: fix memcmp casting
  iwlwifi: pcie: compare with number of IRQs requested for, not number of CPUs
  ...

Pull SCSI fix from James Bottomley:
 "Eve of merge window fix: The original code was so bogus as to be
  casting the wrong generic device to an rport and proceeding to take
  actions based on the bogus values it found.

  Fortunately it seems the location that is dereferenced always exists,
  so the code hasn't oopsed yet, but it certainly annoys the memory
  checkers"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: scsi_transport_srp: Fix shost to rport translation

Pull drm fixes from Dave Airlie:
 "A few final fixes:

  i915:
   - fix for potential Spectre vector in the new query uAPI
   - fix NULL pointer deref (FDO #106559)
   - DMI fix to hide LVDS for Radiant P845 (FDO #105468)

  amdgpu:
   - suspend/resume DC regression fix
   - underscan flicker fix on fiji
   - gamma setting fix after dpms

  omap:
   - fix oops regression

  core:
   - fix PSR timing

  dw-hdmi:
   - fix oops regression"

* tag 'drm-fixes-for-v4.17-rc8' of git://people.freedesktop.org/~airlied/linux:
  drm/amd/display: Update color props when modeset is required
  drm/amd/display: Make atomic-check validate underscan changes
  drm/bridge/synopsys: dw-hdmi: fix dw_hdmi_setup_rx_sense
  drm/amd/display: Fix BUG_ON during CRTC atomic check update
  drm/i915/query: nospec expects no more than an unsigned long
  drm/i915/query: Protect tainted function pointer lookup
  drm/i915/lvds: Move acpi lid notification registration to registration phase
  drm/i915: Disable LVDS on Radiant P845
  drm/omap: fix NULL deref crash with SDI displays
  drm/psr: Fix missed entry in PSR setup time table.

Two last minute DC fixes for 4.17.  A fix for underscan on fiji and
a fix for gamma settings getting after dpms.

* 'drm-fixes-4.17' of git://people.freedesktop.org/~agd5f/linux:
  drm/amd/display: Update color props when modeset is required
  drm/amd/display: Make atomic-check validate underscan changes

Pull MIPS fixes from James Hogan:
 "A final few MIPS fixes for 4.17:

   - drop Lantiq gphy reboot/remove reset (4.14)

   - prctl(PR_SET_FP_MODE): Disallow PRE without FR (4.0)

   - ptrace(PTRACE_PEEKUSR): Fix 64-bit FGRs (3.15)"

* tag 'mips_fixes_4.17_3' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux:
  MIPS: ptrace: Fix PTRACE_PEEKUSR requests for 64-bit FGRs
  MIPS: prctl: Disallow FRE without FR with PR_SET_FP_MODE requests
  MIPS: lantiq: gphy: Drop reboot/remove reset asserts

Pull VFIO fix from Alex Williamson:
 "Revert a pfn page mapping optimization identified as introducing a bad
  page state regression (Alex Williamson)"

* tag 'vfio-v4.17' of git://github.com/awilliam/linux-vfio:
  Revert "vfio/type1: Improve memory pinning process for raw PFN mapping"

Pull char/misc driver fixes from Greg KH:
 "Here are four small bugfixes for some char/misc drivers. Well, really
  three fixes and one fix for one of those fixes due to problems found
  by 0-day.

  This resolves some reported issues with the hwtracing drivers, and a
  reported regression for the thunderbolt subsystem. All of these have
  been in linux-next for a while now with no reported problems"

* tag 'char-misc-4.17-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  hwtracing: stm: fix build error on some arches
  intel_th: Use correct device when freeing buffers
  stm class: Use vmalloc for the master map
  thunderbolt: Handle NULL boot ACL entries properly

Pull IIO driver fixes from Greg KH:
 "Here are some old IIO driver fixes that were sitting in my tree for a
  few weeks. Sorry about not getting them to you sooner. They fix a
  number of small IIO driver issues that have been reported.

  All of these have been in linux-next for a while with no reported
  problems"

* tag 'staging-4.17-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/staging:
  iio: adc: select buffer for at91-sama5d2_adc
  iio: hid-sensor-trigger: Fix sometimes not powering up the sensor after resume
  iio: adc: at91-sama5d2_adc: fix channel configuration for differential channels
  iio:kfifo_buf: check for uint overflow
  iio:buffer: make length types match kfifo types
  iio: adc: stm32-dfsdm: fix sample rate for div2 spi clock
  iio: adc: stm32-dfsdm: fix successive oversampling settings
  iio: ad7793: implement IIO_CHAN_INFO_SAMP_FREQ

Pull rdma fixes from Jason Gunthorpe:
 "Just three small last minute regressions that were found in the last
  week. The Broadcom fix is a bit big for rc7, but since it is fixing
  driver crash regressions that were merged via netdev into rc1, I am
  sending it.

   - bnxt netdev changes merged this cycle caused the bnxt RDMA driver
     to crash under certain situations

   - Arnd found (several, unfortunately) kconfig problems with the
     patches adding INFINIBAND_ADDR_TRANS. Reverting this last part,
     will fix it more fully outside -rc.

   - Subtle change in error code for a uapi function caused breakage in
     userspace. This was bug was subtly introduced cycle"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma:
  IB/core: Fix error code for invalid GID entry
  IB: Revert "remove redundant INFINIBAND kconfig dependencies"
  RDMA/bnxt_re: Fix broken RoCE driver due to recent L2 driver changes

Pull i2c fixes from Wolfram Sang:
 "A documentation bugfix and a MAINTAINERS addition"

* 'i2c/for-current' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: ocores: update HDL sources URL
  i2c: xlp9xx: Add MAINTAINERS entry

Merge two fixes from Andrew Morton.

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mm: fix the NULL mapping case in __isolate_lru_page()
  mm/huge_memory.c: __split_huge_page() use atomic ClearPageDirty()

George Boole would have noticed a slight error in 4.16 commit
69d763fc ("mm: pin address_space before dereferencing it while
isolating an LRU page").  Fix it, to match both the comment above it,
and the original behaviour.

Although anonymous pages are not marked PageDirty at first, we have an
old habit of calling SetPageDirty when a page is removed from swap
cache: so there's a category of ex-swap pages that are easily
migratable, but were inadvertently excluded from compaction's async
migration in 4.16.

Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1805302014001.12558@eggly.anvils
Fixes: 69d763fc ("mm: pin address_space before dereferencing it while isolating an LRU page")
Signed-off-by: NHugh Dickins <hughd@google.com>
Acked-by: NMinchan Kim <minchan@kernel.org>
Acked-by: NMel Gorman <mgorman@techsingularity.net>
Reported-by: NIvan Kalvachev <ikalvachev@gmail.com>
Cc: "Huang, Ying" <ying.huang@intel.com>
Cc: Jan Kara <jack@suse.cz>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>

Swapping load on huge=always tmpfs (with khugepaged tuned up to be very
eager, but I'm not sure that is relevant) soon hung uninterruptibly,
waiting for page lock in shmem_getpage_gfp()'s find_lock_entry(), most
often when "cp -a" was trying to write to a smallish file.  Debug showed
that the page in question was not locked, and page->mapping NULL by now,
but page->index consistent with having been in a huge page before.

Reproduced in minutes on a 4.15 kernel, even with 4.17's 605ca5ed
("mm/huge_memory.c: reorder operations in __split_huge_page_tail()") added
in; but took hours to reproduce on a 4.17 kernel (no idea why).

The culprit proved to be the __ClearPageDirty() on tails beyond i_size in
__split_huge_page(): the non-atomic __bitoperation may have been safe when
4.8's baa355fd ("thp: file pages support for split_huge_page()")
introduced it, but liable to erase PageWaiters after 4.10's 62906027
("mm: add PageWaiters indicating tasks are waiting for a page bit").

Link: http://lkml.kernel.org/r/alpine.LSU.2.11.1805291841070.3197@eggly.anvils
Fixes: 62906027 ("mm: add PageWaiters indicating tasks are waiting for a page bit")
Signed-off-by: NHugh Dickins <hughd@google.com>
Acked-by: NKirill A. Shutemov <kirill.shutemov@linux.intel.com>
Cc: Konstantin Khlebnikov <khlebnikov@yandex-team.ru>
Cc: Nicholas Piggin <npiggin@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: NAndrew Morton <akpm@linux-foundation.org>
Signed-off-by: NLinus Torvalds <torvalds@linux-foundation.org>