linux-2.6-microblaze.git
4 years agoMerge tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost
Linus Torvalds [Fri, 7 Feb 2020 20:26:34 +0000 (12:26 -0800)]
Merge tag 'for_linus' of git://git./linux/kernel/git/mst/vhost

Pull virtio updates from Michael Tsirkin:
 "Some bug fixes/cleanups.

  The deprecated scsi passthrough for virtio_blk is removed"

* tag 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/mst/vhost:
  virtio_balloon: Fix memory leaks on errors in virtballoon_probe()
  virtio-balloon: Fix memory leak when unloading while hinting is in progress
  virtio_balloon: prevent pfn array overflow
  virtio-blk: remove VIRTIO_BLK_F_SCSI support
  virtio-pci: check name when counting MSI-X vectors
  virtio-balloon: initialize all vq callbacks
  virtio-mmio: convert to devm_platform_ioremap_resource

4 years agoMerge tag 'xtensa-20200206' of git://github.com/jcmvbkbc/linux-xtensa
Linus Torvalds [Fri, 7 Feb 2020 20:22:29 +0000 (12:22 -0800)]
Merge tag 'xtensa-20200206' of git://github.com/jcmvbkbc/linux-xtensa

Pull xtensa updates from Max Filippov:

 - reorganize exception vectors placement

 - small cleanups (drop unused functions/headers/defconfig entries,
   spelling fixes)

* tag 'xtensa-20200206' of git://github.com/jcmvbkbc/linux-xtensa:
  xtensa: ISS: improve simcall assembly
  xtensa: reorganize vectors placement
  xtensa: separate SMP and XIP support
  xtensa: move fast exception handlers close to vectors
  arch/xtensa: fix Kconfig typos for HAVE_SMP
  xtensa: clean up optional XCHAL_* definitions
  xtensa: drop unused function fast_coprocessor_double
  xtensa: drop empty platform_* functions from platforms
  xtensa: clean up platform headers
  xtensa: drop set_except_vector declaration
  xtensa: configs: Cleanup old Kconfig IO scheduler options

4 years agoMerge tag 'kvm-5.6-2' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Thu, 6 Feb 2020 17:07:45 +0000 (09:07 -0800)]
Merge tag 'kvm-5.6-2' of git://git./virt/kvm/kvm

Pull more KVM updates from Paolo Bonzini:
 "s390:
   - fix register corruption
   - ENOTSUPP/EOPNOTSUPP mixed
   - reset cleanups/fixes
   - selftests

  x86:
   - Bug fixes and cleanups
   - AMD support for APIC virtualization even in combination with
     in-kernel PIT or IOAPIC.

  MIPS:
   - Compilation fix.

  Generic:
   - Fix refcount overflow for zero page"

* tag 'kvm-5.6-2' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (42 commits)
  KVM: vmx: delete meaningless vmx_decache_cr0_guest_bits() declaration
  KVM: x86: Mark CR4.UMIP as reserved based on associated CPUID bit
  x86: vmxfeatures: rename features for consistency with KVM and manual
  KVM: SVM: relax conditions for allowing MSR_IA32_SPEC_CTRL accesses
  KVM: x86: Fix perfctr WRMSR for running counters
  x86/kvm/hyper-v: don't allow to turn on unsupported VMX controls for nested guests
  x86/kvm/hyper-v: move VMX controls sanitization out of nested_enable_evmcs()
  kvm: mmu: Separate generating and setting mmio ptes
  kvm: mmu: Replace unsigned with unsigned int for PTE access
  KVM: nVMX: Remove stale comment from nested_vmx_load_cr3()
  KVM: MIPS: Fold comparecount_func() into comparecount_wakeup()
  KVM: MIPS: Fix a build error due to referencing not-yet-defined function
  x86/kvm: do not setup pv tlb flush when not paravirtualized
  KVM: fix overflow of zero page refcount with ksm running
  KVM: x86: Take a u64 when checking for a valid dr7 value
  KVM: x86: use raw clock values consistently
  KVM: x86: reorganize pvclock_gtod_data members
  KVM: nVMX: delete meaningless nested_vmx_run() declaration
  KVM: SVM: allow AVIC without split irqchip
  kvm: ioapic: Lazy update IOAPIC EOI
  ...

4 years agoMerge tag 'kgdb-fixes-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/danie...
Linus Torvalds [Thu, 6 Feb 2020 17:05:42 +0000 (09:05 -0800)]
Merge tag 'kgdb-fixes-5.6-rc1' of git://git./linux/kernel/git/danielt/linux

Pull kgdb fix from Daniel Thompson:
 "One of the simplifications added for 5.6-rc1 has caused build
  regressions on some platforms (it was reported for sparc64).

  This fixes it with a revert"

* tag 'kgdb-fixes-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/danielt/linux:
  Revert "kdb: Get rid of confusing diag msg from "rd" if current task has no regs"

4 years agoMerge tag 'pci-v5.6-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Linus Torvalds [Thu, 6 Feb 2020 14:17:38 +0000 (14:17 +0000)]
Merge tag 'pci-v5.6-fixes-1' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fixes from Bjorn Helgaas:

 - Define to_pci_sysdata() always to fix build breakage when !CONFIG_PCI
   (Jason A. Donenfeld)

 - Use PF PASID for VFs to fix VF IOMMU bind failures (Kuppuswamy
   Sathyanarayanan)

* tag 'pci-v5.6-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  PCI/ATS: Use PF PASID for VFs
  x86/PCI: Define to_pci_sysdata() even when !CONFIG_PCI

4 years agoMerge tag 'sound-fix-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai...
Linus Torvalds [Thu, 6 Feb 2020 14:15:01 +0000 (14:15 +0000)]
Merge tag 'sound-fix-5.6-rc1' of git://git./linux/kernel/git/tiwai/sound

Pull sound fixes from Takashi Iwai:
 "A collection of pending small fixes:

  ALSA core:
   - PCM memory leak fix

  ASoC:
   - Lots of SOF and Intel driver fixes
   - Addition of COMMON_CLK for wcd934x
   - Regression fixes for AMD and Tegra platforms

  HD-audio:
   - DP-MST HDMI regression fix, Tegra workarounds, HP quirk fix

  Others:
   - A few fixes relevant with the recent uapi-updates
   - Sparse warnings and endianness fixes"

* tag 'sound-fix-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tiwai/sound: (35 commits)
  ALSA: hda: Clear RIRB status before reading WP
  ALSA: hda/realtek - Fixed one of HP ALC671 platform Headset Mic supported
  ASoC: wcd934x: Add missing COMMON_CLK dependency to SND_SOC_ALL_CODECS
  ALSA: hda - Fix DP-MST support for NVIDIA codecs
  ASoC: wcd934x: Add missing COMMON_CLK dependency
  MAINTAINERS: Remove the Bard Liao from the MAINTAINERS of Realtek CODECs
  ASoC: tegra: Revert 24 and 32 bit support
  ASoC: SOF: Intel: add PCI ID for JasperLake
  ALSA: hdsp: Make the firmware loading ioctl a bit more readable
  ALSA: emu10k1: Fix annotation and cast for the recent uapi header change
  ALSA: dummy: Fix PCM format loop in proc output
  ALSA: usb-audio: Annotate endianess in Scarlett gen2 quirk
  ALSA: usb-audio: Fix endianess in descriptor validation
  ALSA: hda: Add JasperLake PCI ID and codec vid
  ALSA: pcm: Fix sparse warnings wrt snd_pcm_state_t
  ALSA: pcm: Fix memory leak at closing a stream without hw_free
  ALSA: uapi: Fix sparse warning
  ASoC: rt715: Add __maybe_unused to PM callbacks
  ASoC: rt711: Add __maybe_unused to PM callbacks
  ASoC: rt700: Add __maybe_unused to PM callbacks
  ...

4 years agoMerge tag 'ceph-for-5.6-rc1' of https://github.com/ceph/ceph-client
Linus Torvalds [Thu, 6 Feb 2020 12:21:01 +0000 (12:21 +0000)]
Merge tag 'ceph-for-5.6-rc1' of https://github.com/ceph/ceph-client

Pull ceph fixes from Ilya Dryomov:

 - a set of patches that fixes various corner cases in mount and umount
   code (Xiubo Li). This has to do with choosing an MDS, distinguishing
   between laggy and down MDSes and parsing the server path.

 - inode initialization fixes (Jeff Layton). The one included here
   mostly concerns things like open_by_handle() and there is another one
   that will come through Al.

 - copy_file_range() now uses the new copy-from2 op (Luis Henriques).
   The existing copy-from op turned out to be infeasible for generic
   filesystem use; we disable the copy offload if OSDs don't support
   copy-from2.

 - a patch to link "rbd" and "block" devices together in sysfs (Hannes
   Reinecke)

... and a smattering of cleanups from Xiubo, Jeff and Chengguang.

* tag 'ceph-for-5.6-rc1' of https://github.com/ceph/ceph-client: (25 commits)
  rbd: set the 'device' link in sysfs
  ceph: move net/ceph/ceph_fs.c to fs/ceph/util.c
  ceph: print name of xattr in __ceph_{get,set}xattr() douts
  ceph: print r_direct_hash in hex in __choose_mds() dout
  ceph: use copy-from2 op in copy_file_range
  ceph: close holes in structs ceph_mds_session and ceph_mds_request
  rbd: work around -Wuninitialized warning
  ceph: allocate the correct amount of extra bytes for the session features
  ceph: rename get_session and switch to use ceph_get_mds_session
  ceph: remove the extra slashes in the server path
  ceph: add possible_max_rank and make the code more readable
  ceph: print dentry offset in hex and fix xattr_version type
  ceph: only touch the caps which have the subset mask requested
  ceph: don't clear I_NEW until inode metadata is fully populated
  ceph: retry the same mds later after the new session is opened
  ceph: check availability of mds cluster on mount after wait timeout
  ceph: keep the session state until it is released
  ceph: add __send_request helper
  ceph: ensure we have a new cap before continuing in fill_inode
  ceph: drop unused ttl_from parameter from fill_inode
  ...

4 years agoRevert "kdb: Get rid of confusing diag msg from "rd" if current task has no regs"
Daniel Thompson [Thu, 6 Feb 2020 11:40:09 +0000 (11:40 +0000)]
Revert "kdb: Get rid of confusing diag msg from "rd" if current task has no regs"

This reverts commit bbfceba15f8d1260c328a254efc2b3f2deae4904.

When DBG_MAX_REG_NUM is zero then a number of symbols are conditionally
defined. It is therefore not possible to check it using C expressions.

Reported-by: Anatoly Pugachev <matorola@gmail.com>
Acked-by: Doug Anderson <dianders@chromium.org>
Signed-off-by: Daniel Thompson <daniel.thompson@linaro.org>
4 years agoMerge tag 'asoc-v5.6-3' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie...
Takashi Iwai [Thu, 6 Feb 2020 10:58:44 +0000 (11:58 +0100)]
Merge tag 'asoc-v5.6-3' of https://git./linux/kernel/git/broonie/sound into for-linus

ASoC: Fix for v5.6

An incremental fix for the Qualcomm COMMON_CLK issue.

4 years agoALSA: hda: Clear RIRB status before reading WP
Mohan Kumar [Thu, 6 Feb 2020 10:10:53 +0000 (15:40 +0530)]
ALSA: hda: Clear RIRB status before reading WP

RIRB interrupt status getting cleared after the write pointer is read
causes a race condition, where last response(s) into RIRB may remain
unserviced by IRQ, eventually causing azx_rirb_get_response to fall
back to polling mode. Clearing the RIRB interrupt status ahead of
write pointer access ensures that this condition is avoided.

Signed-off-by: Mohan Kumar <mkumard@nvidia.com>
Signed-off-by: Viswanath L <viswanathl@nvidia.com>
Link: https://lore.kernel.org/r/1580983853-351-1-git-send-email-viswanathl@nvidia.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agovirtio_balloon: Fix memory leaks on errors in virtballoon_probe()
David Hildenbrand [Wed, 5 Feb 2020 16:34:01 +0000 (17:34 +0100)]
virtio_balloon: Fix memory leaks on errors in virtballoon_probe()

We forget to put the inode and unmount the kernfs used for compaction.

Fixes: 71994620bb25 ("virtio_balloon: replace oom notifier with shrinker")
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Wei Wang <wei.w.wang@intel.com>
Cc: Liang Li <liang.z.li@intel.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20200205163402.42627-3-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
4 years agovirtio-balloon: Fix memory leak when unloading while hinting is in progress
David Hildenbrand [Wed, 5 Feb 2020 16:34:00 +0000 (17:34 +0100)]
virtio-balloon: Fix memory leak when unloading while hinting is in progress

When unloading the driver while hinting is in progress, we will not
release the free page blocks back to MM, resulting in a memory leak.

Fixes: 86a559787e6f ("virtio-balloon: VIRTIO_BALLOON_F_FREE_PAGE_HINT")
Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Jason Wang <jasowang@redhat.com>
Cc: Wei Wang <wei.w.wang@intel.com>
Cc: Liang Li <liang.z.li@intel.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20200205163402.42627-2-david@redhat.com
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
4 years agovirtio_balloon: prevent pfn array overflow
Michael S. Tsirkin [Thu, 6 Feb 2020 07:40:58 +0000 (02:40 -0500)]
virtio_balloon: prevent pfn array overflow

Make sure, at build time, that pfn array is big enough to hold a single
page.  It happens to be true since the PAGE_SHIFT value at the moment is
20, which is 1M - exactly 256 4K balloon pages.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
4 years agovirtio-blk: remove VIRTIO_BLK_F_SCSI support
Christoph Hellwig [Thu, 12 Dec 2019 16:37:19 +0000 (17:37 +0100)]
virtio-blk: remove VIRTIO_BLK_F_SCSI support

Since the need for a special flag to support SCSI passthrough on a
block device was added in May 2017 the SCSI passthrough support in
virtio-blk has been disabled.  It has always been a bad idea
(just ask the original author..) and we have virtio-scsi for proper
passthrough.  The feature also never made it into the virtio 1.0
or later specifications.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Hannes Reinecke <hare@suse.de>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
4 years agovirtio-pci: check name when counting MSI-X vectors
Daniel Verkamp [Fri, 3 Jan 2020 18:40:45 +0000 (10:40 -0800)]
virtio-pci: check name when counting MSI-X vectors

VQs without a name specified are not valid; they are skipped in the
later loop that assigns MSI-X vectors to queues, but the per_vq_vectors
loop above that counts the required number of vectors previously still
counted any queue with a non-NULL callback as needing a vector.

Add a check to the per_vq_vectors loop so that vectors with no name are
not counted to make the two loops consistent.  This prevents
over-counting unnecessary vectors (e.g. for features which were not
negotiated with the device).

Cc: stable@vger.kernel.org
Fixes: 86a559787e6f ("virtio-balloon: VIRTIO_BALLOON_F_FREE_PAGE_HINT")
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Wang, Wei W <wei.w.wang@intel.com>
4 years agovirtio-balloon: initialize all vq callbacks
Daniel Verkamp [Fri, 3 Jan 2020 18:40:43 +0000 (10:40 -0800)]
virtio-balloon: initialize all vq callbacks

Ensure that elements of the callbacks array that correspond to
unavailable features are set to NULL; previously, they would be left
uninitialized.

Since the corresponding names array elements were explicitly set to
NULL, the uninitialized callback pointers would not actually be
dereferenced; however, the uninitialized callbacks elements would still
be read in vp_find_vqs_msix() and used to calculate the number of MSI-X
vectors required.

Cc: stable@vger.kernel.org
Fixes: 86a559787e6f ("virtio-balloon: VIRTIO_BALLOON_F_FREE_PAGE_HINT")
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Daniel Verkamp <dverkamp@chromium.org>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
4 years agovirtio-mmio: convert to devm_platform_ioremap_resource
Yangtao Li [Sun, 22 Dec 2019 19:08:39 +0000 (19:08 +0000)]
virtio-mmio: convert to devm_platform_ioremap_resource

Use devm_platform_ioremap_resource() to simplify code, which
contains platform_get_resource, devm_request_mem_region and
devm_ioremap.

Signed-off-by: Yangtao Li <tiny.windzz@gmail.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
4 years agoMerge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu
Linus Torvalds [Thu, 6 Feb 2020 08:13:23 +0000 (08:13 +0000)]
Merge branch 'for-next' of git://git./linux/kernel/git/gerg/m68knommu

Pull m68knommu updates from Greg Ungerer:
 "A couple of changes:

   - remove old CONFIG options from the m68knommu defconfig files

   - fix a warning in the m68k non-MMU get_user() macro"

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/gerg/m68knommu:
  m68knommu: fix memcpy() out of bounds warning in get_user()
  m68k: configs: Cleanup old Kconfig IO scheduler options

4 years agoMerge tag 'Smack-for-5.6' of git://github.com/cschaufler/smack-next
Linus Torvalds [Thu, 6 Feb 2020 08:08:59 +0000 (08:08 +0000)]
Merge tag 'Smack-for-5.6' of git://github.com/cschaufler/smack-next

Pull smack fix from Casey Schaufler:
 "One fix for an obscure error found using an old version of ping(1)
  that did not use IPv6 sockets in the documented way"

* tag 'Smack-for-5.6' of git://github.com/cschaufler/smack-next:
  broken ping to ipv6 linklocal addresses on debian buster

4 years agoMerge tag 'xfs-5.6-merge-8' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Thu, 6 Feb 2020 07:58:38 +0000 (07:58 +0000)]
Merge tag 'xfs-5.6-merge-8' of git://git./fs/xfs/xfs-linux

Pull moar xfs updates from Darrick Wong:
 "This contains the buffer error code refactoring I mentioned last week,
  now that it has had extra time to complete the full xfs fuzz testing
  suite to make sure there aren't any obvious new bugs"

* tag 'xfs-5.6-merge-8' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: fix xfs_buf_ioerror_alert location reporting
  xfs: remove unnecessary null pointer checks from _read_agf callers
  xfs: make xfs_*read_agf return EAGAIN to ALLOC_FLAG_TRYLOCK callers
  xfs: remove the xfs_btree_get_buf[ls] functions
  xfs: make xfs_trans_get_buf return an error code
  xfs: make xfs_trans_get_buf_map return an error code
  xfs: make xfs_buf_read return an error code
  xfs: make xfs_buf_get_uncached return an error code
  xfs: make xfs_buf_get return an error code
  xfs: make xfs_buf_read_map return an error code
  xfs: make xfs_buf_get_map return an error code
  xfs: make xfs_buf_alloc return an error code

4 years agoMerge tag 'trace-v5.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt...
Linus Torvalds [Thu, 6 Feb 2020 07:12:11 +0000 (07:12 +0000)]
Merge tag 'trace-v5.6-2' of git://git./linux/kernel/git/rostedt/linux-trace

Pull tracing updates from Steven Rostedt:

 - Added new "bootconfig".

   This looks for a file appended to initrd to add boot config options,
   and has been discussed thoroughly at Linux Plumbers.

   Very useful for adding kprobes at bootup.

   Only enabled if "bootconfig" is on the real kernel command line.

 - Created dynamic event creation.

   Merges common code between creating synthetic events and kprobe
   events.

 - Rename perf "ring_buffer" structure to "perf_buffer"

 - Rename ftrace "ring_buffer" structure to "trace_buffer"

   Had to rename existing "trace_buffer" to "array_buffer"

 - Allow trace_printk() to work withing (some) tracing code.

 - Sort of tracing configs to be a little better organized

 - Fixed bug where ftrace_graph hash was not being protected properly

 - Various other small fixes and clean ups

* tag 'trace-v5.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/rostedt/linux-trace: (88 commits)
  bootconfig: Show the number of nodes on boot message
  tools/bootconfig: Show the number of bootconfig nodes
  bootconfig: Add more parse error messages
  bootconfig: Use bootconfig instead of boot config
  ftrace: Protect ftrace_graph_hash with ftrace_sync
  ftrace: Add comment to why rcu_dereference_sched() is open coded
  tracing: Annotate ftrace_graph_notrace_hash pointer with __rcu
  tracing: Annotate ftrace_graph_hash pointer with __rcu
  bootconfig: Only load bootconfig if "bootconfig" is on the kernel cmdline
  tracing: Use seq_buf for building dynevent_cmd string
  tracing: Remove useless code in dynevent_arg_pair_add()
  tracing: Remove check_arg() callbacks from dynevent args
  tracing: Consolidate some synth_event_trace code
  tracing: Fix now invalid var_ref_vals assumption in trace action
  tracing: Change trace_boot to use synth_event interface
  tracing: Move tracing selftests to bottom of menu
  tracing: Move mmio tracer config up with the other tracers
  tracing: Move tracing test module configs together
  tracing: Move all function tracing configs together
  tracing: Documentation for in-kernel synthetic event API
  ...

4 years agoALSA: hda/realtek - Fixed one of HP ALC671 platform Headset Mic supported
Kailang Yang [Wed, 5 Feb 2020 07:40:01 +0000 (15:40 +0800)]
ALSA: hda/realtek - Fixed one of HP ALC671 platform Headset Mic supported

HP want to keep BIOS verb table for release platform.
So, it need to add 0x19 pin for quirk.

Fixes: 5af29028fd6d ("ALSA: hda/realtek - Add Headset Mic supported for HP cPC")
Signed-off-by: Kailang Yang <kailang@realtek.com>
Link: https://lore.kernel.org/r/74636ccb700a4cbda24c58a99dc430ce@realtek.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agoMerge tag 'io_uring-5.6-2020-02-05' of git://git.kernel.dk/linux-block
Linus Torvalds [Thu, 6 Feb 2020 06:33:17 +0000 (06:33 +0000)]
Merge tag 'io_uring-5.6-2020-02-05' of git://git.kernel.dk/linux-block

Pull io_uring updates from Jens Axboe:
 "Some later fixes for io_uring:

   - Small cleanup series from Pavel

   - Belt and suspenders build time check of sqe size and layout
     (Stefan)

   - Addition of ->show_fdinfo() on request of Jann Horn, to aid in
     understanding mapped personalities

   - eventfd recursion/deadlock fix, for both io_uring and aio

   - Fixup for send/recv handling

   - Fixup for double deferral of read/write request

   - Fix for potential double completion event for close request

   - Adjust fadvise advice async/inline behavior

   - Fix for shutdown hang with SQPOLL thread

   - Fix for potential use-after-free of fixed file table"

* tag 'io_uring-5.6-2020-02-05' of git://git.kernel.dk/linux-block:
  io_uring: cleanup fixed file data table references
  io_uring: spin for sq thread to idle on shutdown
  aio: prevent potential eventfd recursion on poll
  io_uring: put the flag changing code in the same spot
  io_uring: iterate req cache backwards
  io_uring: punt even fadvise() WILLNEED to async context
  io_uring: fix sporadic double CQE entry for close
  io_uring: remove extra ->file check
  io_uring: don't map read/write iovec potentially twice
  io_uring: use the proper helpers for io_send/recv
  io_uring: prevent potential eventfd recursion on poll
  eventfd: track eventfd_signal() recursion depth
  io_uring: add BUILD_BUG_ON() to assert the layout of struct io_uring_sqe
  io_uring: add ->show_fdinfo() for the io_uring file descriptor

4 years agoMerge tag 'block-5.6-2020-02-05' of git://git.kernel.dk/linux-block
Linus Torvalds [Thu, 6 Feb 2020 06:15:23 +0000 (06:15 +0000)]
Merge tag 'block-5.6-2020-02-05' of git://git.kernel.dk/linux-block

Pull more block updates from Jens Axboe:
 "Some later arrivals, but all fixes at this point:

   - bcache fix series (Coly)

   - Series of BFQ fixes (Paolo)

   - NVMe pull request from Keith with a few minor NVMe fixes

   - Various little tweaks"

* tag 'block-5.6-2020-02-05' of git://git.kernel.dk/linux-block: (23 commits)
  nvmet: update AEN list and array at one place
  nvmet: Fix controller use after free
  nvmet: Fix error print message at nvmet_install_queue function
  brd: check and limit max_part par
  nvme-pci: remove nvmeq->tags
  nvmet: fix dsm failure when payload does not match sgl descriptor
  nvmet: Pass lockdep expression to RCU lists
  block, bfq: clarify the goal of bfq_split_bfqq()
  block, bfq: get a ref to a group when adding it to a service tree
  block, bfq: remove ifdefs from around gets/puts of bfq groups
  block, bfq: extend incomplete name of field on_st
  block, bfq: get extra ref to prevent a queue from being freed during a group move
  block, bfq: do not insert oom queue into position tree
  block, bfq: do not plug I/O for bfq_queues with no proc refs
  bcache: check return value of prio_read()
  bcache: fix incorrect data type usage in btree_flush_write()
  bcache: add readahead cache policy options via sysfs interface
  bcache: explicity type cast in bset_bkey_last()
  bcache: fix memory corruption in bch_cache_accounting_clear()
  xen/blkfront: limit allocated memory size to actual use case
  ...

4 years agoMerge tag 'libata-5.6-2020-02-05' of git://git.kernel.dk/linux-block
Linus Torvalds [Thu, 6 Feb 2020 06:11:50 +0000 (06:11 +0000)]
Merge tag 'libata-5.6-2020-02-05' of git://git.kernel.dk/linux-block

Pull libata updates from Jens Axboe:

 - Add a Sandisk CF card to supported pata_pcmcia list (Christian)

 - Move pata_arasan_cf away from legacy API (Peter)

 - Ensure ahci DMA/ints are shut down on shutdown (Prabhakar)

* tag 'libata-5.6-2020-02-05' of git://git.kernel.dk/linux-block:
  ata: pata_arasan_cf: Use dma_request_chan() instead dma_request_slave_channel()
  ata: ahci: Add shutdown to freeze hardware resources of ahci
  pata_pcmia: add SanDisk High (>8G) CF card to supported list

4 years agobootconfig: Show the number of nodes on boot message
Masami Hiramatsu [Wed, 5 Feb 2020 13:50:23 +0000 (22:50 +0900)]
bootconfig: Show the number of nodes on boot message

Show the number of bootconfig nodes on boot message.

Link: http://lkml.kernel.org/r/158091062297.27924.9051634676068550285.stgit@devnote2
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agotools/bootconfig: Show the number of bootconfig nodes
Masami Hiramatsu [Wed, 5 Feb 2020 13:50:13 +0000 (22:50 +0900)]
tools/bootconfig: Show the number of bootconfig nodes

Show the number of bootconfig nodes when applying new bootconfig to
initrd.

Since there are limitations of bootconfig not only in its filesize,
but also the number of nodes, the number should be shown when applying
so that user can get the feeling of scale of current bootconfig.

Link: http://lkml.kernel.org/r/158091061337.27924.10886706631693823982.stgit@devnote2
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agobootconfig: Add more parse error messages
Masami Hiramatsu [Wed, 5 Feb 2020 13:50:04 +0000 (22:50 +0900)]
bootconfig: Add more parse error messages

Add more error messages for following cases.
 - Exceeding max number of nodes
 - Config tree data is empty (e.g. comment only)
 - Config data is empty or exceeding max size
 - bootconfig is already initialized

Link: http://lkml.kernel.org/r/158091060401.27924.9024818742827122764.stgit@devnote2
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agobootconfig: Use bootconfig instead of boot config
Masami Hiramatsu [Wed, 5 Feb 2020 13:49:54 +0000 (22:49 +0900)]
bootconfig: Use bootconfig instead of boot config

Use "bootconfig" (1 word) instead of "boot config" (2 words)
in the boot message.

Link: http://lkml.kernel.org/r/158091059459.27924.14414336187441539879.stgit@devnote2
Signed-off-by: Masami Hiramatsu <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agoftrace: Protect ftrace_graph_hash with ftrace_sync
Steven Rostedt (VMware) [Wed, 5 Feb 2020 14:20:32 +0000 (09:20 -0500)]
ftrace: Protect ftrace_graph_hash with ftrace_sync

As function_graph tracer can run when RCU is not "watching", it can not be
protected by synchronize_rcu() it requires running a task on each CPU before
it can be freed. Calling schedule_on_each_cpu(ftrace_sync) needs to be used.

Link: https://lore.kernel.org/r/20200205131110.GT2935@paulmck-ThinkPad-P72
Cc: stable@vger.kernel.org
Fixes: b9b0c831bed26 ("ftrace: Convert graph filter to use hash tables")
Reported-by: "Paul E. McKenney" <paulmck@kernel.org>
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agobroken ping to ipv6 linklocal addresses on debian buster
Casey Schaufler [Mon, 3 Feb 2020 17:15:00 +0000 (09:15 -0800)]
broken ping to ipv6 linklocal addresses on debian buster

I am seeing ping failures to IPv6 linklocal addresses with Debian
buster. Easiest example to reproduce is:

$ ping -c1 -w1 ff02::1%eth1
connect: Invalid argument

$ ping -c1 -w1 ff02::1%eth1
PING ff02::01%eth1(ff02::1%eth1) 56 data bytes
64 bytes from fe80::e0:f9ff:fe0c:37%eth1: icmp_seq=1 ttl=64 time=0.059 ms

git bisect traced the failure to
commit b9ef5513c99b ("smack: Check address length before reading address family")

Arguably ping is being stupid since the buster version is not setting
the address family properly (ping on stretch for example does):

$ strace -e connect ping6 -c1 -w1 ff02::1%eth1
connect(5, {sa_family=AF_UNSPEC,
sa_data="\4\1\0\0\0\0\377\2\0\0\0\0\0\0\0\0\0\0\0\0\0\1\3\0\0\0"}, 28)
= -1 EINVAL (Invalid argument)

but the command works fine on kernels prior to this commit, so this is
breakage which goes against the Linux paradigm of "don't break userspace"

Cc: stable@vger.kernel.org
Reported-by: David Ahern <dsahern@gmail.com>
Suggested-by: Tetsuo Handa <penguin-kernel@i-love.sakura.ne.jp>
Signed-off-by: Casey Schaufler <casey@schaufler-ca.com>
 security/smack/smack_lsm.c | 41 +++++++++++++++++++----------------------
 1 file changed, 19 insertions(+), 22 deletions(-)

4 years agoftrace: Add comment to why rcu_dereference_sched() is open coded
Steven Rostedt (VMware) [Wed, 5 Feb 2020 07:17:57 +0000 (02:17 -0500)]
ftrace: Add comment to why rcu_dereference_sched() is open coded

Because the function graph tracer can execute in sections where RCU is not
"watching", the rcu_dereference_sched() for the has needs to be open coded.
This is fine because the RCU "flavor" of the ftrace hash is protected by
its own RCU handling (it does its own little synchronization on every CPU
and does not rely on RCU sched).

Acked-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agotracing: Annotate ftrace_graph_notrace_hash pointer with __rcu
Amol Grover [Wed, 5 Feb 2020 05:57:02 +0000 (11:27 +0530)]
tracing: Annotate ftrace_graph_notrace_hash pointer with __rcu

Fix following instances of sparse error
kernel/trace/ftrace.c:5667:29: error: incompatible types in comparison
kernel/trace/ftrace.c:5813:21: error: incompatible types in comparison
kernel/trace/ftrace.c:5868:36: error: incompatible types in comparison
kernel/trace/ftrace.c:5870:25: error: incompatible types in comparison

Use rcu_dereference_protected to dereference the newly annotated pointer.

Link: http://lkml.kernel.org/r/20200205055701.30195-1-frextrite@gmail.com
Signed-off-by: Amol Grover <frextrite@gmail.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agotracing: Annotate ftrace_graph_hash pointer with __rcu
Amol Grover [Sat, 1 Feb 2020 07:27:04 +0000 (12:57 +0530)]
tracing: Annotate ftrace_graph_hash pointer with __rcu

Fix following instances of sparse error
kernel/trace/ftrace.c:5664:29: error: incompatible types in comparison
kernel/trace/ftrace.c:5785:21: error: incompatible types in comparison
kernel/trace/ftrace.c:5864:36: error: incompatible types in comparison
kernel/trace/ftrace.c:5866:25: error: incompatible types in comparison

Use rcu_dereference_protected to access the __rcu annotated pointer.

Link: http://lkml.kernel.org/r/20200201072703.17330-1-frextrite@gmail.com
Reviewed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Amol Grover <frextrite@gmail.com>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agoMerge tag 'pwm/for-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry...
Linus Torvalds [Wed, 5 Feb 2020 18:11:51 +0000 (18:11 +0000)]
Merge tag 'pwm/for-5.6-rc1' of git://git./linux/kernel/git/thierry.reding/linux-pwm

Pull pwm updates from Thierry Reding:
 "Mostly cleanups and minor improvements with some new chip support for
  some drivers"

* tag 'pwm/for-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/thierry.reding/linux-pwm: (37 commits)
  pwm: Remove set but not set variable 'pwm'
  pwm: sun4i: Initialize variables before use
  pwm: stm32: Remove automatic output enable
  pwm: sun4i: Narrow scope of local variable
  pwm: bcm2835: Allow building for ARCH_BRCMSTB
  pwm: imx27: Eliminate error message for defer probe
  pwm: sun4i: Fix inconsistent IS_ERR and PTR_ERR
  pwm: sun4i: Move pwm_calculate() out of spin_lock()
  pwm: omap-dmtimer: Allow compiling with COMPILE_TEST
  pwm: omap-dmtimer: put_device() after of_find_device_by_node()
  pwm: omap-dmtimer: Simplify error handling
  pwm: omap-dmtimer: Remove PWM chip in .remove before making it unfunctional
  pwm: Implement tracing for .get_state() and .apply_state()
  pwm: rcar: Document inability to set duty_cycle = 0
  pwm: rcar: Drop useless call to pwm_get_state()
  pwm: Fix minor Kconfig whitespace issues
  pwm: atmel: Implement .get_state()
  pwm: atmel: Use register accessors for channels
  pwm: atmel: Document known weaknesses of both hardware and software
  pwm: atmel: Replace loop in prescale calculation by ad-hoc calculation
  ...

4 years agoMerge tag 'dmaengine-fix-5.6-rc1' of git://git.infradead.org/users/vkoul/slave-dma
Linus Torvalds [Wed, 5 Feb 2020 18:07:39 +0000 (18:07 +0000)]
Merge tag 'dmaengine-fix-5.6-rc1' of git://git.infradead.org/users/vkoul/slave-dma

Pull dmaengine fixes from Vinod Koul:
 "Fixes for:

   - Documentation build error fix

   - Fix dma_request_chan() error return

   - Remove unneeded conversion in idxd driver

   - Fix pointer check for dma_async_device_channel_register()

   - Fix slave-channel symlink cleanup"

* tag 'dmaengine-fix-5.6-rc1' of git://git.infradead.org/users/vkoul/slave-dma:
  dmaengine: Cleanups for the slave <-> channel symlink support
  dmaengine: fix null ptr check for __dma_async_device_channel_register()
  dmaengine: idxd: fix boolconv.cocci warnings
  dmaengine: Fix return value for dma_request_chan() in case of failure
  dmaengine: doc: Properly indent metadata title

4 years agoPCI/ATS: Use PF PASID for VFs
Kuppuswamy Sathyanarayanan [Wed, 29 Jan 2020 19:14:00 +0000 (11:14 -0800)]
PCI/ATS: Use PF PASID for VFs

Per PCIe r5.0, sec 9.3.7.14, if a PF implements the PASID Capability, the
PF PASID configuration is shared by its VFs, and VFs must not implement
their own PASID Capability.  But commit 751035b8dc06 ("PCI/ATS: Cache PASID
Capability offset") changed pci_max_pasids() and pci_pasid_features() to
use the PASID Capability of the VF device instead of the associated PF
device.  This leads to IOMMU bind failures when pci_max_pasids() and
pci_pasid_features() are called for VFs.

In pci_max_pasids() and pci_pasid_features(), always use the PF PASID
Capability.

Fixes: 751035b8dc06 ("PCI/ATS: Cache PASID Capability offset")
Link: https://lore.kernel.org/r/fe891f9755cb18349389609e7fed9940fc5b081a.1580325170.git.sathyanarayanan.kuppuswamy@linux.intel.com
Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@linux.intel.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
CC: stable@vger.kernel.org # v5.5+
4 years agoMerge tag 'iommu-updates-v5.6' of git://git.kernel.org/pub/scm/linux/kernel/git/joro...
Linus Torvalds [Wed, 5 Feb 2020 17:49:54 +0000 (17:49 +0000)]
Merge tag 'iommu-updates-v5.6' of git://git./linux/kernel/git/joro/iommu

Pull iommu updates from Joerg Roedel:

 - Allow compiling the ARM-SMMU drivers as modules.

 - Fixes and cleanups for the ARM-SMMU drivers and io-pgtable code
   collected by Will Deacon. The merge-commit (6855d1ba7537) has all the
   details.

 - Cleanup of the iommu_put_resv_regions() call-backs in various
   drivers.

 - AMD IOMMU driver cleanups.

 - Update for the x2APIC support in the AMD IOMMU driver.

 - Preparation patches for Intel VT-d nested mode support.

 - RMRR and identity domain handling fixes for the Intel VT-d driver.

 - More small fixes and cleanups.

* tag 'iommu-updates-v5.6' of git://git.kernel.org/pub/scm/linux/kernel/git/joro/iommu: (87 commits)
  iommu/amd: Remove the unnecessary assignment
  iommu/vt-d: Remove unnecessary WARN_ON_ONCE()
  iommu/vt-d: Unnecessary to handle default identity domain
  iommu/vt-d: Allow devices with RMRRs to use identity domain
  iommu/vt-d: Add RMRR base and end addresses sanity check
  iommu/vt-d: Mark firmware tainted if RMRR fails sanity check
  iommu/amd: Remove unused struct member
  iommu/amd: Replace two consecutive readl calls with one readq
  iommu/vt-d: Don't reject Host Bridge due to scope mismatch
  PCI/ATS: Add PASID stubs
  iommu/arm-smmu-v3: Return -EBUSY when trying to re-add a device
  iommu/arm-smmu-v3: Improve add_device() error handling
  iommu/arm-smmu-v3: Use WRITE_ONCE() when changing validity of an STE
  iommu/arm-smmu-v3: Add second level of context descriptor table
  iommu/arm-smmu-v3: Prepare for handling arm_smmu_write_ctx_desc() failure
  iommu/arm-smmu-v3: Propagate ssid_bits
  iommu/arm-smmu-v3: Add support for Substream IDs
  iommu/arm-smmu-v3: Add context descriptor tables allocators
  iommu/arm-smmu-v3: Prepare arm_smmu_s1_cfg for SSID support
  ACPI/IORT: Parse SSID property of named component node
  ...

4 years agoMerge tag 'for-linus-5.6-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 5 Feb 2020 17:44:14 +0000 (17:44 +0000)]
Merge tag 'for-linus-5.6-rc1-tag' of git://git./linux/kernel/git/xen/tip

Pull xen updates from Juergen Gross:

 - fix a bug introduced in 5.5 in the Xen gntdev driver

 - fix the Xen balloon driver when running on ancient Xen versions

 - allow Xen stubdoms to control interrupt enable flags of
   passed-through PCI cards

 - release resources in Xen backends under memory pressure

* tag 'for-linus-5.6-rc1-tag' of git://git.kernel.org/pub/scm/linux/kernel/git/xen/tip:
  xen/blkback: Consistently insert one empty line between functions
  xen/blkback: Remove unnecessary static variable name prefixes
  xen/blkback: Squeeze page pools if a memory pressure is detected
  xenbus/backend: Protect xenbus callback with lock
  xenbus/backend: Add memory pressure handler callback
  xen/gntdev: Do not use mm notifiers with autotranslating guests
  xen/balloon: Support xend-based toolstack take two
  xen-pciback: optionally allow interrupt enable flag writes

4 years agoMerge tag 'devicetree-fixes-for-5.6' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Wed, 5 Feb 2020 17:37:25 +0000 (17:37 +0000)]
Merge tag 'devicetree-fixes-for-5.6' of git://git./linux/kernel/git/robh/linux

Pull devicetree fixes from Rob Herring:

 - Fix incorrect $id paths in schemas

 - Two fixes for Intel LGM SoC binding schemas

* tag 'devicetree-fixes-for-5.6' of git://git.kernel.org/pub/scm/linux/kernel/git/robh/linux:
  dt-bindings: Fix paths in schema $id fields
  dt-bindings: PCI: intel: Fix dt_binding_check compilation failure
  dt-bindings: phy: Fix errors in intel,lgm-emmc-phy example

4 years agoMerge tag 's390-5.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Wed, 5 Feb 2020 17:33:35 +0000 (17:33 +0000)]
Merge tag 's390-5.6-2' of git://git./linux/kernel/git/s390/linux

Pull more s390 updates from Vasily Gorbik:
 "The second round of s390 fixes and features for 5.6:

   - Add KPROBES_ON_FTRACE support

   - Add EP11 AES secure keys support

   - PAES rework and prerequisites for paes-s390 ciphers selftests

   - Fix page table upgrade for hugetlbfs"

* tag 's390-5.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390/pkey/zcrypt: Support EP11 AES secure keys
  s390/zcrypt: extend EP11 card and queue sysfs attributes
  s390/zcrypt: add new low level ep11 functions support file
  s390/zcrypt: ep11 structs rework, export zcrypt_send_ep11_cprb
  s390/zcrypt: enable card/domain autoselect on ep11 cprbs
  s390/crypto: enable clear key values for paes ciphers
  s390/pkey: Add support for key blob with clear key value
  s390/crypto: Rework on paes implementation
  s390: support KPROBES_ON_FTRACE
  s390/mm: fix dynamic pagetable upgrade for hugetlbfs

4 years agoKVM: vmx: delete meaningless vmx_decache_cr0_guest_bits() declaration
Miaohe Lin [Wed, 5 Feb 2020 15:33:53 +0000 (23:33 +0800)]
KVM: vmx: delete meaningless vmx_decache_cr0_guest_bits() declaration

The function vmx_decache_cr0_guest_bits() is only called below its
implementation. So this is meaningless and should be removed.

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: Mark CR4.UMIP as reserved based on associated CPUID bit
Sean Christopherson [Tue, 28 Jan 2020 23:53:44 +0000 (15:53 -0800)]
KVM: x86: Mark CR4.UMIP as reserved based on associated CPUID bit

Re-add code to mark CR4.UMIP as reserved if UMIP is not supported by the
host.  The UMIP handling was unintentionally dropped during a recent
refactoring.

Not flagging CR4.UMIP allows the guest to set its CR4.UMIP regardless of
host support or userspace desires.  On CPUs with UMIP support, including
emulated UMIP, this allows the guest to enable UMIP against the wishes
of the userspace VMM.  On CPUs without any form of UMIP, this results in
a failed VM-Enter due to invalid guest state.

Fixes: 345599f9a2928 ("KVM: x86: Add macro to ensure reserved cr4 bits checks stay in sync")
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agox86: vmxfeatures: rename features for consistency with KVM and manual
Paolo Bonzini [Wed, 5 Feb 2020 15:20:23 +0000 (16:20 +0100)]
x86: vmxfeatures: rename features for consistency with KVM and manual

Three of the feature bits in vmxfeatures.h have names that are different
from the Intel SDM.  The names have been adjusted recently in KVM but they
were using the old name in the tip tree's x86/cpu branch.  Adjust for
consistency.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoMerge tag 'kvm-s390-next-5.6-1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Paolo Bonzini [Wed, 5 Feb 2020 15:15:05 +0000 (16:15 +0100)]
Merge tag 'kvm-s390-next-5.6-1' of git://git./linux/kernel/git/kvms390/linux into HEAD

KVM: s390: Fixes and cleanups for 5.6
- fix register corruption
- ENOTSUPP/EOPNOTSUPP mixed
- reset cleanups/fixes
- selftests

4 years agoKVM: SVM: relax conditions for allowing MSR_IA32_SPEC_CTRL accesses
Paolo Bonzini [Wed, 5 Feb 2020 15:10:52 +0000 (16:10 +0100)]
KVM: SVM: relax conditions for allowing MSR_IA32_SPEC_CTRL accesses

Userspace that does not know about the AMD_IBRS bit might still
allow the guest to protect itself with MSR_IA32_SPEC_CTRL using
the Intel SPEC_CTRL bit.  However, svm.c disallows this and will
cause a #GP in the guest when writing to the MSR.  Fix this by
loosening the test and allowing the Intel CPUID bit, and in fact
allow the AMD_STIBP bit as well since it allows writing to
MSR_IA32_SPEC_CTRL too.

Reported-by: Zhiyi Guo <zhguo@redhat.com>
Analyzed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Analyzed-by: Laszlo Ersek <lersek@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: Fix perfctr WRMSR for running counters
Eric Hankland [Mon, 27 Jan 2020 21:22:56 +0000 (13:22 -0800)]
KVM: x86: Fix perfctr WRMSR for running counters

Correct the logic in intel_pmu_set_msr() for fixed and general purpose
counters. This was recently changed to set pmc->counter without taking
in to account the value of pmc_read_counter() which will be incorrect if
the counter is currently running and non-zero; this changes back to the
old logic which accounted for the value of currently running counters.

Signed-off-by: Eric Hankland <ehankland@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agox86/kvm/hyper-v: don't allow to turn on unsupported VMX controls for nested guests
Vitaly Kuznetsov [Wed, 5 Feb 2020 12:30:34 +0000 (13:30 +0100)]
x86/kvm/hyper-v: don't allow to turn on unsupported VMX controls for nested guests

Sane L1 hypervisors are not supposed to turn any of the unsupported VMX
controls on for its guests and nested_vmx_check_controls() checks for
that. This is, however, not the case for the controls which are supported
on the host but are missing in enlightened VMCS and when eVMCS is in use.

It would certainly be possible to add these missing checks to
nested_check_vm_execution_controls()/_vm_exit_controls()/.. but it seems
preferable to keep eVMCS-specific stuff in eVMCS and reduce the impact on
non-eVMCS guests by doing less unrelated checks. Create a separate
nested_evmcs_check_controls() for this purpose.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agox86/kvm/hyper-v: move VMX controls sanitization out of nested_enable_evmcs()
Vitaly Kuznetsov [Wed, 5 Feb 2020 12:30:33 +0000 (13:30 +0100)]
x86/kvm/hyper-v: move VMX controls sanitization out of nested_enable_evmcs()

With fine grained VMX feature enablement QEMU>=4.2 tries to do KVM_SET_MSRS
with default (matching CPU model) values and in case eVMCS is also enabled,
fails.

It would be possible to drop VMX feature filtering completely and make
this a guest's responsibility: if it decides to use eVMCS it should know
which fields are available and which are not. Hyper-V mostly complies to
this, however, there are some problematic controls:
SECONDARY_EXEC_VIRTUALIZE_APIC_ACCESSES
VM_{ENTRY,EXIT}_LOAD_IA32_PERF_GLOBAL_CTRL

which Hyper-V enables. As there are no corresponding fields in eVMCS, we
can't handle this properly in KVM. This is a Hyper-V issue.

Move VMX controls sanitization from nested_enable_evmcs() to vmx_get_msr(),
and do the bare minimum (only clear controls which are known to cause issues).
This allows userspace to keep setting controls it wants and at the same
time hides them from the guest.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: mmu: Separate generating and setting mmio ptes
Ben Gardon [Mon, 3 Feb 2020 23:09:10 +0000 (15:09 -0800)]
kvm: mmu: Separate generating and setting mmio ptes

Separate the functions for generating MMIO page table entries from the
function that inserts them into the paging structure. This refactoring
will facilitate changes to the MMU sychronization model to use atomic
compare / exchanges (which are not guaranteed to succeed) instead of a
monolithic MMU lock.

No functional change expected.

Tested by running kvm-unit-tests on an Intel Haswell machine. This
commit introduced no new failures.

Signed-off-by: Ben Gardon <bgardon@google.com>
Reviewed-by: Oliver Upton <oupton@google.com>
Reviewed-by: Peter Shier <pshier@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: mmu: Replace unsigned with unsigned int for PTE access
Ben Gardon [Mon, 3 Feb 2020 23:09:09 +0000 (15:09 -0800)]
kvm: mmu: Replace unsigned with unsigned int for PTE access

There are several functions which pass an access permission mask for
SPTEs as an unsigned. This works, but checkpatch complains about it.
Switch the occurrences of unsigned to unsigned int to satisfy checkpatch.

No functional change expected.

Tested by running kvm-unit-tests on an Intel Haswell machine. This
commit introduced no new failures.

Signed-off-by: Ben Gardon <bgardon@google.com>
Reviewed-by: Oliver Upton <oupton@google.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Remove stale comment from nested_vmx_load_cr3()
Sean Christopherson [Tue, 4 Feb 2020 15:32:59 +0000 (07:32 -0800)]
KVM: nVMX: Remove stale comment from nested_vmx_load_cr3()

The blurb pertaining to the return value of nested_vmx_load_cr3() no
longer matches reality, remove it entirely as the behavior it is
attempting to document is quite obvious when reading the actual code.

Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Reviewed-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: MIPS: Fold comparecount_func() into comparecount_wakeup()
Sean Christopherson [Mon, 3 Feb 2020 18:42:00 +0000 (10:42 -0800)]
KVM: MIPS: Fold comparecount_func() into comparecount_wakeup()

Fold kvm_mips_comparecount_func() into kvm_mips_comparecount_wakeup() to
eliminate the nondescript function name as well as its unnecessary cast
of a vcpu to "unsigned long" and back to a vcpu.  Presumably func() was
used as a callback at some point during pre-upstream development, as
wakeup() is the only user of func() and has been the only user since
both with introduced by commit 669e846e6c4e ("KVM/MIPS32: MIPS arch
specific APIs for KVM").

Cc: Davidlohr Bueso <dbueso@suse.de>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: MIPS: Fix a build error due to referencing not-yet-defined function
Sean Christopherson [Mon, 3 Feb 2020 18:41:59 +0000 (10:41 -0800)]
KVM: MIPS: Fix a build error due to referencing not-yet-defined function

Hoist kvm_mips_comparecount_wakeup() above its only user,
kvm_arch_vcpu_create() to fix a compilation error due to referencing an
undefined function.

Fixes: d11dfed5d700 ("KVM: MIPS: Move all vcpu init code into kvm_arch_vcpu_create()")
Reported-by: kbuild test robot <lkp@intel.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agox86/kvm: do not setup pv tlb flush when not paravirtualized
Thadeu Lima de Souza Cascardo [Fri, 31 Jan 2020 15:56:55 +0000 (12:56 -0300)]
x86/kvm: do not setup pv tlb flush when not paravirtualized

kvm_setup_pv_tlb_flush will waste memory and print a misguiding message
when KVM paravirtualization is not available.

Intel SDM says that the when cpuid is used with EAX higher than the
maximum supported value for basic of extended function, the data for the
highest supported basic function will be returned.

So, in some systems, kvm_arch_para_features will return bogus data,
causing kvm_setup_pv_tlb_flush to detect support for pv tlb flush.

Testing for kvm_para_available will work as it checks for the hypervisor
signature.

Besides, when the "nopv" command line parameter is used, it should not
continue as well, as kvm_guest_init will no be called in that case.

Signed-off-by: Thadeu Lima de Souza Cascardo <cascardo@canonical.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: fix overflow of zero page refcount with ksm running
Zhuang Yanying [Sat, 12 Oct 2019 03:37:31 +0000 (11:37 +0800)]
KVM: fix overflow of zero page refcount with ksm running

We are testing Virtual Machine with KSM on v5.4-rc2 kernel,
and found the zero_page refcount overflow.
The cause of refcount overflow is increased in try_async_pf
(get_user_page) without being decreased in mmu_set_spte()
while handling ept violation.
In kvm_release_pfn_clean(), only unreserved page will call
put_page. However, zero page is reserved.
So, as well as creating and destroy vm, the refcount of
zero page will continue to increase until it overflows.

step1:
echo 10000 > /sys/kernel/pages_to_scan/pages_to_scan
echo 1 > /sys/kernel/pages_to_scan/run
echo 1 > /sys/kernel/pages_to_scan/use_zero_pages

step2:
just create several normal qemu kvm vms.
And destroy it after 10s.
Repeat this action all the time.

After a long period of time, all domains hang because
of the refcount of zero page overflow.

Qemu print error log as follow:
 â€¦
 error: kvm run failed Bad address
 EAX=00006cdc EBX=00000008 ECX=80202001 EDX=078bfbfd
 ESI=ffffffff EDI=00000000 EBP=00000008 ESP=00006cc4
 EIP=000efd75 EFL=00010002 [-------] CPL=0 II=0 A20=1 SMM=0 HLT=0
 ES =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
 CS =0008 00000000 ffffffff 00c09b00 DPL=0 CS32 [-RA]
 SS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
 DS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
 FS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
 GS =0010 00000000 ffffffff 00c09300 DPL=0 DS   [-WA]
 LDT=0000 00000000 0000ffff 00008200 DPL=0 LDT
 TR =0000 00000000 0000ffff 00008b00 DPL=0 TSS32-busy
 GDT=     000f7070 00000037
 IDT=     000f70ae 00000000
 CR0=00000011 CR2=00000000 CR3=00000000 CR4=00000000
 DR0=0000000000000000 DR1=0000000000000000 DR2=0000000000000000 DR3=0000000000000000
 DR6=00000000ffff0ff0 DR7=0000000000000400
 EFER=0000000000000000
 Code=00 01 00 00 00 e9 e8 00 00 00 c7 05 4c 55 0f 00 01 00 00 00 <8b> 35 00 00 01 00 8b 3d 04 00 01 00 b8 d8 d3 00 00 c1 e0 08 0c ea a3 00 00 01 00 c7 05 04
 â€¦

Meanwhile, a kernel warning is departed.

 [40914.836375] WARNING: CPU: 3 PID: 82067 at ./include/linux/mm.h:987 try_get_page+0x1f/0x30
 [40914.836412] CPU: 3 PID: 82067 Comm: CPU 0/KVM Kdump: loaded Tainted: G           OE     5.2.0-rc2 #5
 [40914.836415] RIP: 0010:try_get_page+0x1f/0x30
 [40914.836417] Code: 40 00 c3 0f 1f 84 00 00 00 00 00 48 8b 47 08 a8 01 75 11 8b 47 34 85 c0 7e 10 f0 ff 47 34 b8 01 00 00 00 c3 48 8d 78 ff eb e9 <0f> 0b 31 c0 c3 66 90 66 2e 0f 1f 84 00 0
 0 00 00 00 48 8b 47 08 a8
 [40914.836418] RSP: 0018:ffffb4144e523988 EFLAGS: 00010286
 [40914.836419] RAX: 0000000080000000 RBX: 0000000000000326 RCX: 0000000000000000
 [40914.836420] RDX: 0000000000000000 RSI: 00004ffdeba10000 RDI: ffffdf07093f6440
 [40914.836421] RBP: ffffdf07093f6440 R08: 800000424fd91225 R09: 0000000000000000
 [40914.836421] R10: ffff9eb41bfeebb8 R11: 0000000000000000 R12: ffffdf06bbd1e8a8
 [40914.836422] R13: 0000000000000080 R14: 800000424fd91225 R15: ffffdf07093f6440
 [40914.836423] FS:  00007fb60ffff700(0000) GS:ffff9eb4802c0000(0000) knlGS:0000000000000000
 [40914.836425] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
 [40914.836426] CR2: 0000000000000000 CR3: 0000002f220e6002 CR4: 00000000003626e0
 [40914.836427] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
 [40914.836427] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
 [40914.836428] Call Trace:
 [40914.836433]  follow_page_pte+0x302/0x47b
 [40914.836437]  __get_user_pages+0xf1/0x7d0
 [40914.836441]  ? irq_work_queue+0x9/0x70
 [40914.836443]  get_user_pages_unlocked+0x13f/0x1e0
 [40914.836469]  __gfn_to_pfn_memslot+0x10e/0x400 [kvm]
 [40914.836486]  try_async_pf+0x87/0x240 [kvm]
 [40914.836503]  tdp_page_fault+0x139/0x270 [kvm]
 [40914.836523]  kvm_mmu_page_fault+0x76/0x5e0 [kvm]
 [40914.836588]  vcpu_enter_guest+0xb45/0x1570 [kvm]
 [40914.836632]  kvm_arch_vcpu_ioctl_run+0x35d/0x580 [kvm]
 [40914.836645]  kvm_vcpu_ioctl+0x26e/0x5d0 [kvm]
 [40914.836650]  do_vfs_ioctl+0xa9/0x620
 [40914.836653]  ksys_ioctl+0x60/0x90
 [40914.836654]  __x64_sys_ioctl+0x16/0x20
 [40914.836658]  do_syscall_64+0x5b/0x180
 [40914.836664]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
 [40914.836666] RIP: 0033:0x7fb61cb6bfc7

Signed-off-by: LinFeng <linfeng23@huawei.com>
Signed-off-by: Zhuang Yanying <ann.zhuangyanying@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: Take a u64 when checking for a valid dr7 value
Sean Christopherson [Fri, 24 Jan 2020 23:07:22 +0000 (15:07 -0800)]
KVM: x86: Take a u64 when checking for a valid dr7 value

Take a u64 instead of an unsigned long in kvm_dr7_valid() to fix a build
warning on i386 due to right-shifting a 32-bit value by 32 when checking
for bits being set in dr7[63:32].

Alternatively, the warning could be resolved by rewriting the check to
use an i386-friendly method, but taking a u64 fixes another oddity on
32-bit KVM.  Beause KVM implements natural width VMCS fields as u64s to
avoid layout issues between 32-bit and 64-bit, a devious guest can stuff
vmcs12->guest_dr7 with a 64-bit value even when both the guest and host
are 32-bit kernels.  KVM eventually drops vmcs12->guest_dr7[63:32] when
propagating vmcs12->guest_dr7 to vmcs02, but ideally KVM would not rely
on that behavior for correctness.

Cc: Jim Mattson <jmattson@google.com>
Cc: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Fixes: ecb697d10f70 ("KVM: nVMX: Check GUEST_DR7 on vmentry of nested guests")
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: use raw clock values consistently
Paolo Bonzini [Wed, 22 Jan 2020 13:36:09 +0000 (14:36 +0100)]
KVM: x86: use raw clock values consistently

Commit 53fafdbb8b21f ("KVM: x86: switch KVMCLOCK base to monotonic raw
clock") changed kvmclock to use tkr_raw instead of tkr_mono.  However,
the default kvmclock_offset for the VM was still based on the monotonic
clock and, if the raw clock drifted enough from the monotonic clock,
this could cause a negative system_time to be written to the guest's
struct pvclock.  RHEL5 does not like it and (if it boots fast enough to
observe a negative time value) it hangs.

There is another thing to be careful about: getboottime64 returns the
host boot time with tkr_mono frequency, and subtracting the tkr_raw-based
kvmclock value will cause the wallclock to be off if tkr_raw drifts
from tkr_mono.  To avoid this, compute the wallclock delta from the
current time instead of being clever and using getboottime64.

Fixes: 53fafdbb8b21f ("KVM: x86: switch KVMCLOCK base to monotonic raw clock")
Cc: stable@vger.kernel.org
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: reorganize pvclock_gtod_data members
Paolo Bonzini [Wed, 22 Jan 2020 13:32:20 +0000 (14:32 +0100)]
KVM: x86: reorganize pvclock_gtod_data members

We will need a copy of tk->offs_boot in the next patch.  Store it and
cleanup the struct: instead of storing tk->tkr_xxx.base with the tk->offs_boot
included, store the raw value in struct pvclock_clock and sum it in
do_monotonic_raw and do_realtime.   tk->tkr_xxx.xtime_nsec also moves
to struct pvclock_clock.

While at it, fix a (usually harmless) typo in do_monotonic_raw, which
was using gtod->clock.shift instead of gtod->raw_clock.shift.

Fixes: 53fafdbb8b21f ("KVM: x86: switch KVMCLOCK base to monotonic raw clock")
Cc: stable@vger.kernel.org
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: delete meaningless nested_vmx_run() declaration
Miaohe Lin [Thu, 23 Jan 2020 02:08:20 +0000 (10:08 +0800)]
KVM: nVMX: delete meaningless nested_vmx_run() declaration

The function nested_vmx_run() declaration is below its implementation. So
this is meaningless and should be removed.

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: SVM: allow AVIC without split irqchip
Paolo Bonzini [Wed, 22 Jan 2020 16:02:36 +0000 (17:02 +0100)]
KVM: SVM: allow AVIC without split irqchip

SVM is now able to disable AVIC dynamically whenever the in-kernel PIT sets
up an ack notifier, so we can enable it even if in-kernel IOAPIC/PIC/PIT
are in use.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: ioapic: Lazy update IOAPIC EOI
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:19 +0000 (14:15 -0600)]
kvm: ioapic: Lazy update IOAPIC EOI

In-kernel IOAPIC does not receive EOI with AMD SVM AVIC
since the processor accelerate write to APIC EOI register and
does not trap if the interrupt is edge-triggered.

Workaround this by lazy check for pending APIC EOI at the time when
setting new IOPIC irq, and update IOAPIC EOI if no pending APIC EOI.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: ioapic: Refactor kvm_ioapic_update_eoi()
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:18 +0000 (14:15 -0600)]
kvm: ioapic: Refactor kvm_ioapic_update_eoi()

Refactor code for handling IOAPIC EOI for subsequent patch.
There is no functional change.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: i8254: Deactivate APICv when using in-kernel PIT re-injection mode.
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:16 +0000 (14:15 -0600)]
kvm: i8254: Deactivate APICv when using in-kernel PIT re-injection mode.

AMD SVM AVIC accelerates EOI write and does not trap. This causes
in-kernel PIT re-injection mode to fail since it relies on irq-ack
notifier mechanism. So, APICv is activated only when in-kernel PIT
is in discard mode e.g. w/ qemu option:

  -global kvm-pit.lost_tick_policy=discard

Also, introduce APICV_INHIBIT_REASON_PIT_REINJ bit to be used for this
reason.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agosvm: Temporarily deactivate AVIC during ExtINT handling
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:15 +0000 (14:15 -0600)]
svm: Temporarily deactivate AVIC during ExtINT handling

AMD AVIC does not support ExtINT. Therefore, AVIC must be temporary
deactivated and fall back to using legacy interrupt injection via vINTR
and interrupt window.

Also, introduce APICV_INHIBIT_REASON_IRQWIN to be used for this reason.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
[Rename svm_request_update_avic to svm_toggle_avic_for_extint. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agosvm: Deactivate AVIC when launching guest with nested SVM support
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:14 +0000 (14:15 -0600)]
svm: Deactivate AVIC when launching guest with nested SVM support

Since AVIC does not currently work w/ nested virtualization,
deactivate AVIC for the guest if setting CPUID Fn80000001_ECX[SVM]
(i.e. indicate support for SVM, which is needed for nested virtualization).
Also, introduce a new APICV_INHIBIT_REASON_NESTED bit to be used for
this reason.

Suggested-by: Alexander Graf <graf@amazon.com>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: hyperv: Use APICv update request interface
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:13 +0000 (14:15 -0600)]
kvm: x86: hyperv: Use APICv update request interface

Since disabling APICv has to be done for all vcpus on AMD-based
system, adopt the newly introduced kvm_request_apicv_update()
interface, and introduce a new APICV_INHIBIT_REASON_HYPERV.

Also, remove the kvm_vcpu_deactivate_apicv() since no longer used.

Cc: Roman Kagan <rkagan@virtuozzo.com>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agosvm: Add support for dynamic APICv
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:12 +0000 (14:15 -0600)]
svm: Add support for dynamic APICv

Add necessary logics to support (de)activate AVIC at runtime.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: Introduce x86 ops hook for pre-update APICv
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:11 +0000 (14:15 -0600)]
kvm: x86: Introduce x86 ops hook for pre-update APICv

AMD SVM AVIC needs to update APIC backing page mapping before changing
APICv mode. Introduce struct kvm_x86_ops.pre_update_apicv_exec_ctrl
function hook to be called prior KVM APICv update request to each vcpu.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: Introduce APICv x86 ops for checking APIC inhibit reasons
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:10 +0000 (14:15 -0600)]
kvm: x86: Introduce APICv x86 ops for checking APIC inhibit reasons

Inibit reason bits are used to determine if APICv deactivation is
applicable for a particular hardware virtualization architecture.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: svm: avic: Add support for dynamic setup/teardown of virtual APIC backing page
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:09 +0000 (14:15 -0600)]
KVM: svm: avic: Add support for dynamic setup/teardown of virtual APIC backing page

Re-factor avic_init_access_page() to avic_update_access_page() since
activate/deactivate AVIC requires setting/unsetting the memory region used
for virtual APIC backing page (APIC_ACCESS_PAGE_PRIVATE_MEMSLOT).

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: svm: Add support to (de)activate posted interrupts
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:08 +0000 (14:15 -0600)]
kvm: x86: svm: Add support to (de)activate posted interrupts

Introduce interface for (de)activate posted interrupts, and
implement SVM hooks to toggle AMD IOMMU guest virtual APIC mode.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: Add APICv (de)activate request trace points
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:07 +0000 (14:15 -0600)]
kvm: x86: Add APICv (de)activate request trace points

Add trace points when sending request to (de)activate APICv.

Suggested-by: Alexander Graf <graf@amazon.com>
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: Add support for dynamic APICv activation
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:06 +0000 (14:15 -0600)]
kvm: x86: Add support for dynamic APICv activation

Certain runtime conditions require APICv to be temporary deactivated
during runtime.  The current implementation only support run-time
deactivation of APICv when Hyper-V SynIC is enabled, which is not
temporary.

In addition, for AMD, when APICv is (de)activated at runtime,
all vcpus in the VM have to operate in the same mode.  Thus the
requesting vcpu must notify the others.

So, introduce the following:
 * A new KVM_REQ_APICV_UPDATE request bit
 * Interfaces to request all vcpus to update APICv status
 * A new interface to update APICV-related parameters for each vcpu

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: remove get_enable_apicv from kvm_x86_ops
Paolo Bonzini [Wed, 22 Jan 2020 15:54:37 +0000 (16:54 +0100)]
KVM: x86: remove get_enable_apicv from kvm_x86_ops

It is unused now.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: Introduce APICv inhibit reason bits
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:05 +0000 (14:15 -0600)]
kvm: x86: Introduce APICv inhibit reason bits

There are several reasons in which a VM needs to deactivate APICv
e.g. disable APICv via parameter during module loading, or when
enable Hyper-V SynIC support. Additional inhibit reasons will be
introduced later on when dynamic APICv is supported,

Introduce KVM APICv inhibit reason bits along with a new variable,
apicv_inhibit_reasons, to help keep track of APICv state for each VM,

Initially, the APICV_INHIBIT_REASON_DISABLE bit is used to indicate
the case where APICv is disabled during KVM module load.
(e.g. insmod kvm_amd avic=0 or insmod kvm_intel enable_apicv=0).

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
[Do not use get_enable_apicv; consider irqchip_split in svm.c. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: lapic: Introduce APICv update helper function
Suravee Suthikulpanit [Thu, 14 Nov 2019 20:15:04 +0000 (14:15 -0600)]
kvm: lapic: Introduce APICv update helper function

Re-factor code into a helper function for setting lapic parameters when
activate/deactivate APICv, and export the function for subsequent usage.

Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoMerge tag 'asoc-v5.6-2' of https://git.kernel.org/pub/scm/linux/kernel/git/broonie...
Takashi Iwai [Wed, 5 Feb 2020 11:33:45 +0000 (12:33 +0100)]
Merge tag 'asoc-v5.6-2' of https://git./linux/kernel/git/broonie/sound into for-linus

ASoC: Fixes for v5.6

A collection of updates for bugs fixed since the initial pull
request, the most important one being the addition of COMMON_CLK
for wcd934x which is needed for MFD to be merged.

4 years agoASoC: wcd934x: Add missing COMMON_CLK dependency to SND_SOC_ALL_CODECS
Geert Uytterhoeven [Tue, 4 Feb 2020 13:18:57 +0000 (14:18 +0100)]
ASoC: wcd934x: Add missing COMMON_CLK dependency to SND_SOC_ALL_CODECS

Just adding a dependency on COMMON_CLK to SND_SOC_WCD934X is not
sufficient, as enabling SND_SOC_ALL_CODECS will still select it,
breaking the build later:

    WARNING: unmet direct dependencies detected for SND_SOC_WCD934X
      Depends on [n]: SOUND [=m] && !UML && SND [=m] && SND_SOC [=m] && COMMON_CLK [=n] && MFD_WCD934X [=m]
      Selected by [m]:
      - SND_SOC_ALL_CODECS [=m] && SOUND [=m] && !UML && SND [=m] && SND_SOC [=m] && COMPILE_TEST [=y] && MFD_WCD934X [=m]
    ...
    ERROR: "of_clk_add_provider" [sound/soc/codecs/snd-soc-wcd934x.ko] undefined!
    ERROR: "of_clk_src_simple_get" [sound/soc/codecs/snd-soc-wcd934x.ko] undefined!
    ERROR: "clk_hw_register" [sound/soc/codecs/snd-soc-wcd934x.ko] undefined!
    ERROR: "__clk_get_name" [sound/soc/codecs/snd-soc-wcd934x.ko] undefined!

Fix this by adding the missing dependency to SND_SOC_ALL_CODECS

Fixes: 42b716359beca106 ("ASoC: wcd934x: Add missing COMMON_CLK dependency")
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Tested-by: Stephen Rothwell <sfr@canb.auug.org.au>
Link: https://lore.kernel.org/r/20200204131857.7634-1-geert@linux-m68k.org
Signed-off-by: Mark Brown <broonie@kernel.org>
4 years agobootconfig: Only load bootconfig if "bootconfig" is on the kernel cmdline
Steven Rostedt (VMware) [Tue, 4 Feb 2020 12:33:53 +0000 (07:33 -0500)]
bootconfig: Only load bootconfig if "bootconfig" is on the kernel cmdline

As the bootconfig is appended to the initrd it is not as easy to modify as
the kernel command line. If there's some issue with the kernel, and the
developer wants to boot a pristine kernel, it should not be needed to modify
the initrd to remove the bootconfig for a single boot.

As bootconfig is silently added (if the admin does not know where to look
they may not know it's being loaded). It should be explicitly added to the
kernel cmdline. The loading of the bootconfig is only done if "bootconfig"
is on the kernel command line. This will let admins know that the kernel
command line is extended.

Note, after adding printk()s for when the size is too great or the checksum
is wrong, exposed that the current method always looked for the boot config,
and if this size and checksum matched, it would parse it (as if either is
wrong a printk has been added to show this). It's better to only check this
if the boot config is asked to be looked for.

Link: https://lore.kernel.org/r/CAHk-=wjfjO+h6bQzrTf=YCZA53Y3EDyAs3Z4gEsT7icA3u_Psw@mail.gmail.com
Acked-by: Masami Hiramatsu <mhiramat@kernel.org>
Suggested-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Steven Rostedt (VMware) <rostedt@goodmis.org>
4 years agodt-bindings: Fix paths in schema $id fields
Rob Herring [Fri, 31 Jan 2020 15:27:12 +0000 (09:27 -0600)]
dt-bindings: Fix paths in schema $id fields

The $id path checks were inadequately checking the path part of the $id
value. With the check fixed, there's a number of errors that need to be
fixed. Most of the errors are including 'bindings/' in the path which
should not be as that is considered the root.

Cc: Andy Gross <agross@kernel.org>
Cc: Bjorn Andersson <bjorn.andersson@linaro.org>
Cc: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Cc: Michael Turquette <mturquette@baylibre.com>
Cc: Shawn Guo <shawnguo@kernel.org>
Cc: Sascha Hauer <s.hauer@pengutronix.de>
Cc: Pengutronix Kernel Team <kernel@pengutronix.de>
Cc: Fabio Estevam <festevam@gmail.com>
Cc: NXP Linux Team <linux-imx@nxp.com>
Cc: Maxime Coquelin <mcoquelin.stm32@gmail.com>
Cc: Alexandre Torgue <alexandre.torgue@st.com>
Cc: "Nuno Sá" <nuno.sa@analog.com>
Cc: Jean Delvare <jdelvare@suse.com>
Cc: Stefan Popa <stefan.popa@analog.com>
Cc: Jonathan Cameron <jic23@kernel.org>
Cc: Hartmut Knaack <knaack.h@gmx.de>
Cc: Lars-Peter Clausen <lars@metafoo.de>
Cc: Peter Meerwald-Stadler <pmeerw@pmeerw.net>
Cc: Marcus Folkesson <marcus.folkesson@gmail.com>
Cc: Kent Gustavsson <kent@minoris.se>
Cc: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Cc: linux-clk@vger.kernel.org
Cc: linux-iio@vger.kernel.org
Cc: linux-input@vger.kernel.org
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Rob Herring <robh@kernel.org>
4 years agoMerge branch 'parisc-5.6-1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller...
Linus Torvalds [Wed, 5 Feb 2020 07:29:40 +0000 (07:29 +0000)]
Merge branch 'parisc-5.6-1' of git://git./linux/kernel/git/deller/parisc-linux

Pull parisc updates from Helge Deller:
 "A page table initialization cleanup from Mike Rapoport and regenerated
  defconfig files from Helge Deller"

* 'parisc-5.6-1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Regenerate parisc defconfigs
  parisc: map_pages(): cleanup page table initialization

4 years agoxtensa: ISS: improve simcall assembly
Max Filippov [Tue, 4 Feb 2020 20:51:21 +0000 (12:51 -0800)]
xtensa: ISS: improve simcall assembly

Drop redundant result moving from inline assembly, use a1 and b1 values
as return value and errno value respectively.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
4 years agoxtensa: reorganize vectors placement
Max Filippov [Sat, 1 Feb 2020 04:11:24 +0000 (20:11 -0800)]
xtensa: reorganize vectors placement

Allow vectors to be either merged into the kernel .text or put at a
fixed virtual address independently of XIP option. Drop option that
puts vectors at a fixed offset from the kernel text. Add choice to
Kconfig.
Vectors at fixed virtual address may be useful for XIP-aware MTD support
and for noMMU configurations with available IRAM. Configurations without
VECBASE register must put their vectors at specific locations regardless
of the selected option. All other configurations should happily use
merged vectors.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
4 years agoxtensa: separate SMP and XIP support
Max Filippov [Sat, 1 Feb 2020 05:25:03 +0000 (21:25 -0800)]
xtensa: separate SMP and XIP support

There's no real dependency between SMP and XIP, allow them to be
selected together. Always define 2- and 4-argument SECTION_VECTOR
macros, always use 4-argument macro for the secondary reset vector and
always define relocation entry for it.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
4 years agoxtensa: move fast exception handlers close to vectors
Max Filippov [Sat, 1 Feb 2020 02:48:43 +0000 (18:48 -0800)]
xtensa: move fast exception handlers close to vectors

On XIP kernels it makes sense to have exception vectors and fast
exception handlers together (in a fast memory). In addition, with MTD
XIP support both vectors and fast exception handlers must be outside of
the FLASH.

Add section .exception.text and move fast exception handlers to it.
Put it together with vectors when vectors are outside of the .text.

Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
4 years agoMerge tag 'jfs-5.6' of git://github.com/kleikamp/linux-shaggy
Linus Torvalds [Wed, 5 Feb 2020 05:28:20 +0000 (05:28 +0000)]
Merge tag 'jfs-5.6' of git://github.com/kleikamp/linux-shaggy

Pull jfs update from David Kleikamp:
 "Trivial cleanup for jfs"

* tag 'jfs-5.6' of git://github.com/kleikamp/linux-shaggy:
  jfs: remove unused MAXL2PAGES

4 years agoMerge branch 'work.recursive_removal' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Wed, 5 Feb 2020 05:09:46 +0000 (05:09 +0000)]
Merge branch 'work.recursive_removal' of git://git./linux/kernel/git/viro/vfs

Pull vfs recursive removal updates from Al Viro:
 "We have quite a few places where synthetic filesystems do an
  equivalent of 'rm -rf', with varying amounts of code duplication,
  wrong locking, etc. That really ought to be a library helper.

  Only debugfs (and very similar tracefs) are converted here - I have
  more conversions, but they'd never been in -next, so they'll have to
  wait"

* 'work.recursive_removal' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  simple_recursive_removal(): kernel-side rm -rf for ramfs-style filesystems

4 years agoMerge branch 'imm.timestamp' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Wed, 5 Feb 2020 05:02:42 +0000 (05:02 +0000)]
Merge branch 'imm.timestamp' of git://git./linux/kernel/git/viro/vfs

Pull vfs timestamp updates from Al Viro:
 "More 64bit timestamp work"

* 'imm.timestamp' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  kernfs: don't bother with timestamp truncation
  fs: Do not overload update_time
  fs: Delete timespec64_trunc()
  fs: ubifs: Eliminate timespec64_trunc() usage
  fs: ceph: Delete timespec64_trunc() usage
  fs: cifs: Delete usage of timespec64_trunc
  fs: fat: Eliminate timespec64_trunc() usage
  utimes: Clamp the timestamps in notify_change()

4 years agoio_uring: cleanup fixed file data table references
Jens Axboe [Wed, 5 Feb 2020 02:54:55 +0000 (19:54 -0700)]
io_uring: cleanup fixed file data table references

syzbot reports a use-after-free in io_ring_file_ref_switch() when it
tries to switch back to percpu mode. When we put the final reference to
the table by calling percpu_ref_kill_and_confirm(), we don't want the
zero reference to queue async work for flushing the potentially queued
up items. We currently do a few flush_work(), but they merely paper
around the issue, since the work item may not have been queued yet
depending on the when the percpu-ref callback gets run.

Coming into the file unregister, we know we have the ring quiesced.
io_ring_file_ref_switch() can check for whether or not the ref is dying
or not, and not queue anything async at that point. Once the ref has
been confirmed killed, flush any potential items manually.

Reported-by: syzbot+7caeaea49c2c8a591e3d@syzkaller.appspotmail.com
Fixes: 05f3fb3c5397 ("io_uring: avoid ring quiesce for fixed file set unregister and update")
Signed-off-by: Jens Axboe <axboe@kernel.dk>
4 years agoio_uring: spin for sq thread to idle on shutdown
Jens Axboe [Tue, 4 Feb 2020 23:48:34 +0000 (16:48 -0700)]
io_uring: spin for sq thread to idle on shutdown

As part of io_uring shutdown, we cancel work that is pending and won't
necessarily complete on its own. That includes requests like poll
commands and timeouts.

If we're using SQPOLL for kernel side submission and we shutdown the
ring immediately after queueing such work, we can race with the sqthread
doing the submission. This means we may miss cancelling some work, which
results in the io_uring shutdown hanging forever.

Cc: stable@vger.kernel.org
Signed-off-by: Jens Axboe <axboe@kernel.dk>
4 years agodt-bindings: PCI: intel: Fix dt_binding_check compilation failure
Dilip Kota [Fri, 20 Dec 2019 09:53:24 +0000 (17:53 +0800)]
dt-bindings: PCI: intel: Fix dt_binding_check compilation failure

Remove <dt-bindings/clock/intel,lgm-clk.h> dependency as
it is not present in the mainline tree. Use numeric value
instead of LGM_GCLK_PCIE10 macro.

Signed-off-by: Dilip Kota <eswara.kota@linux.intel.com>
[robh: Also drop interrupt-parent from example]
Signed-off-by: Rob Herring <robh@kernel.org>
4 years agodt-bindings: phy: Fix errors in intel,lgm-emmc-phy example
Rob Herring [Fri, 31 Jan 2020 15:25:29 +0000 (09:25 -0600)]
dt-bindings: phy: Fix errors in intel,lgm-emmc-phy example

DT labels can't have '-' in them causing a compile failure in the example.
Fixing that leads to more warnings:

Documentation/devicetree/bindings/phy/intel,lgm-emmc-phy.example.dts:23.13-33: Warning (reg_format): /example-0/chiptop@e0200000/emmc-phy@a8:reg: property has invalid length (8 bytes) (#address-cells == 2, #size-cells == 1)
Documentation/devicetree/bindings/phy/intel,lgm-emmc-phy.example.dt.yaml: Warning (pci_device_bus_num): Failed prerequisite 'reg_format'
Documentation/devicetree/bindings/phy/intel,lgm-emmc-phy.example.dt.yaml: Warning (i2c_bus_reg): Failed prerequisite 'reg_format'
Documentation/devicetree/bindings/phy/intel,lgm-emmc-phy.example.dt.yaml: Warning (spi_bus_reg): Failed prerequisite 'reg_format'
Documentation/devicetree/bindings/phy/intel,lgm-emmc-phy.example.dts:21.33-26.13: Warning (avoid_default_addr_size): /example-0/chiptop@e0200000/emmc-phy@a8: Relying on default #address-cells value
Documentation/devicetree/bindings/phy/intel,lgm-emmc-phy.example.dts:21.33-26.13: Warning (avoid_default_addr_size): /example-0/chiptop@e0200000/emmc-phy@a8: Relying on default #size-cells value

Fixes: 5bc999108025 ("dt-bindings: phy: intel-emmc-phy: Add YAML schema for LGM eMMC PHY")
Cc: Ramuthevar Vadivel Murugan <vadivel.muruganx.ramuthevar@linux.intel.com>
Cc: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Cc: Kishon Vijay Abraham I <kishon@ti.com>
Signed-off-by: Rob Herring <robh@kernel.org>
4 years agoMerge branch 'nvme-5.6' of git://git.infradead.org/nvme into block-5.6
Jens Axboe [Tue, 4 Feb 2020 18:29:36 +0000 (11:29 -0700)]
Merge branch 'nvme-5.6' of git://git.infradead.org/nvme into block-5.6

Pull NVMe fixes from Keith.

* 'nvme-5.6' of git://git.infradead.org/nvme:
  nvmet: update AEN list and array at one place
  nvmet: Fix controller use after free
  nvmet: Fix error print message at nvmet_install_queue function
  nvme-pci: remove nvmeq->tags
  nvmet: fix dsm failure when payload does not match sgl descriptor
  nvmet: Pass lockdep expression to RCU lists

4 years agonvmet: update AEN list and array at one place
Daniel Wagner [Thu, 30 Jan 2020 18:29:34 +0000 (10:29 -0800)]
nvmet: update AEN list and array at one place

All async events are enqueued via nvmet_add_async_event() which
updates the ctrl->async_event_cmds[] array and additionally an struct
nvmet_async_event is added to the ctrl->async_events list.

Under normal operations the nvmet_async_event_work() updates again
the ctrl->async_event_cmds and removes the corresponding struct
nvmet_async_event from the list again. Though nvmet_sq_destroy() could
be called which calls nvmet_async_events_free() which only updates the
ctrl->async_event_cmds[] array.

Add new functions nvmet_async_events_process() and
nvmet_async_events_free() to process async events, update an array and
the list.

When we destroy submission queue after clearing the aen present on
the ctrl->async list we also loop over ctrl->async_event_cmds[] for
any requests posted by the host for which we don't have the AEN in
the ctrl->async_events list by calling nvmet_async_event_process()
and nvmet_async_events_free().

Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Daniel Wagner <dwagner@suse.de>
[chaitanya.kulkarni@wdc.com
 * Loop over and clear out outstanding requests
 * Update changelog
]
Signed-off-by: Chaitanya Kulkarni <chaitanya.kulkarni@wdc.com>
Signed-off-by: Keith Busch <kbusch@kernel.org>
4 years agoALSA: hda - Fix DP-MST support for NVIDIA codecs
Nikhil Mahale [Tue, 4 Feb 2020 10:27:46 +0000 (15:57 +0530)]
ALSA: hda - Fix DP-MST support for NVIDIA codecs

If dyn_pcm_assign is set, different jack objects are being created
for pcm and pins.

If dyn_pcm_assign is set, generic_hdmi_build_jack() calls into
add_hdmi_jack_kctl() to create and track separate jack object for
pcm. Like sync_eld_via_acomp(), hdmi_present_sense_via_verbs() also
need to report status change of the pcm jack.

Rename pin_idx_to_jack() to pin_idx_to_pcm_jack(). Update
hdmi_present_sense_via_verbs() to report plug state of pcm jack
object. Unlike sync_eld_via_acomp(), for !acomp drivers the pcm
jack's plug state must be consistent with plug state
of pin's jack.

Fixes: 5398e94fb753 ("ALSA: hda - Add DP-MST support for NVIDIA codecs")
Reported-and-tested-by: Martin Regner <martin@larkos.de>
Signed-off-by: Nikhil Mahale <nmahale@nvidia.com>
Reviewed-by: Kai Vehmanen <kai.vehmanen@linux.intel.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/20200204102746.1356-1-nmahale@nvidia.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
4 years agonvmet: Fix controller use after free
Israel Rukshin [Tue, 4 Feb 2020 12:38:10 +0000 (14:38 +0200)]
nvmet: Fix controller use after free

After nvmet_install_queue() sets sq->ctrl calling to nvmet_sq_destroy()
reduces the controller refcount. In case nvmet_install_queue() fails,
calling to nvmet_ctrl_put() is done twice (at nvmet_sq_destroy and
nvmet_execute_io_connect/nvmet_execute_admin_connect) instead of once for
the queue which leads to use after free of the controller. Fix this by set
NULL at sq->ctrl in case of a failure at nvmet_install_queue().

The bug leads to the following Call Trace:

[65857.994862] refcount_t: underflow; use-after-free.
[65858.108304] Workqueue: events nvmet_rdma_release_queue_work [nvmet_rdma]
[65858.115557] RIP: 0010:refcount_warn_saturate+0xe5/0xf0
[65858.208141] Call Trace:
[65858.211203]  nvmet_sq_destroy+0xe1/0xf0 [nvmet]
[65858.216383]  nvmet_rdma_release_queue_work+0x37/0xf0 [nvmet_rdma]
[65858.223117]  process_one_work+0x167/0x370
[65858.227776]  worker_thread+0x49/0x3e0
[65858.232089]  kthread+0xf5/0x130
[65858.235895]  ? max_active_store+0x80/0x80
[65858.240504]  ? kthread_bind+0x10/0x10
[65858.244832]  ret_from_fork+0x1f/0x30
[65858.249074] ---[ end trace f82d59250b54beb7 ]---

Fixes: bb1cc74790eb ("nvmet: implement valid sqhd values in completions")
Fixes: 1672ddb8d691 ("nvmet: Add install_queue callout")
Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Keith Busch <kbusch@kernel.org>
4 years agonvmet: Fix error print message at nvmet_install_queue function
Israel Rukshin [Tue, 4 Feb 2020 12:38:09 +0000 (14:38 +0200)]
nvmet: Fix error print message at nvmet_install_queue function

Place the arguments in the correct order.

Fixes: 1672ddb8d691 ("nvmet: Add install_queue callout")
Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Reviewed-by: Max Gurtovoy <maxg@mellanox.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Keith Busch <kbusch@kernel.org>
4 years agox86/PCI: Define to_pci_sysdata() even when !CONFIG_PCI
Jason A. Donenfeld [Mon, 3 Feb 2020 21:53:06 +0000 (22:53 +0100)]
x86/PCI: Define to_pci_sysdata() even when !CONFIG_PCI

Recently, the to_pci_sysdata() helper was added inside the CONFIG_PCI
guard, but it is used inside a CONFIG_NUMA guard, which does not require
CONFIG_PCI.  This breaks builds on !CONFIG_PCI machines.  Make
to_pci_sysdata() available in all configurations.

Fixes: aad6aa0cd674 ("x86/PCI: Add to_pci_sysdata() helper")
Link: https://lore.kernel.org/r/20200203215306.172000-1-Jason@zx2c4.com
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Acked-by: Randy Dunlap <rdunlap@infradead.org> # build-tested
4 years agobrd: check and limit max_part par
Zhiqiang Liu [Tue, 4 Feb 2020 11:30:20 +0000 (19:30 +0800)]
brd: check and limit max_part par

In brd_init func, rd_nr num of brd_device are firstly allocated
and add in brd_devices, then brd_devices are traversed to add each
brd_device by calling add_disk func. When allocating brd_device,
the disk->first_minor is set to i * max_part, if rd_nr * max_part
is larger than MINORMASK, two different brd_device may have the same
devt, then only one of them can be successfully added.
when rmmod brd.ko, it will cause oops when calling brd_exit.

Follow those steps:
  # modprobe brd rd_nr=3 rd_size=102400 max_part=1048576
  # rmmod brd
then, the oops will appear.

Oops log:
[  726.613722] Call trace:
[  726.614175]  kernfs_find_ns+0x24/0x130
[  726.614852]  kernfs_find_and_get_ns+0x44/0x68
[  726.615749]  sysfs_remove_group+0x38/0xb0
[  726.616520]  blk_trace_remove_sysfs+0x1c/0x28
[  726.617320]  blk_unregister_queue+0x98/0x100
[  726.618105]  del_gendisk+0x144/0x2b8
[  726.618759]  brd_exit+0x68/0x560 [brd]
[  726.619501]  __arm64_sys_delete_module+0x19c/0x2a0
[  726.620384]  el0_svc_common+0x78/0x130
[  726.621057]  el0_svc_handler+0x38/0x78
[  726.621738]  el0_svc+0x8/0xc
[  726.622259] Code: aa0203f6 aa0103f7 aa1e03e0 d503201f (7940e260)

Here, we add brd_check_and_reset_par func to check and limit max_part par.

--
V5->V6:
 - remove useless code

V4->V5:(suggested by Ming Lei)
 - make sure max_part is not larger than DISK_MAX_PARTS

V3->V4:(suggested by Ming Lei)
 - remove useless change
 - add one limit of max_part

V2->V3: (suggested by Ming Lei)
 - clear .minors when running out of consecutive minor space in brd_alloc
 - remove limit of rd_nr

V1->V2:
 - add more checks in brd_check_par_valid as suggested by Ming Lei.

Signed-off-by: Zhiqiang Liu <liuzhiqiang26@huawei.com>
Reviewed-by: Bob Liu <bob.liu@oracle.com>
Reviewed-by: Ming Lei <ming.lei@redhat.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>