openwrt/staging/blogic.git
5 years agoswiotlb-xen: merge xen_unmap_single into xen_swiotlb_unmap_page
Christoph Hellwig [Wed, 24 Jul 2019 14:18:41 +0000 (16:18 +0200)]
swiotlb-xen: merge xen_unmap_single into xen_swiotlb_unmap_page

No need for a no-op wrapper.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoswiotlb-xen: simplify cache maintainance
Christoph Hellwig [Thu, 5 Sep 2019 08:04:30 +0000 (10:04 +0200)]
swiotlb-xen: simplify cache maintainance

Now that we know we always have the dma-noncoherent.h helpers available
if we are on an architecture with support for non-coherent devices,
we can just call them directly, and remove the calls to the dma-direct
routines, including the fact that we call the dma_direct_map_page
routines but ignore the value returned from it.  Instead we now have
Xen wrappers for the arch_sync_dma_for_{device,cpu} helpers that call
the special Xen versions of those routines for foreign pages.

Note that the new helpers get the physical address passed in addition
to the dma address to avoid another translation for the local cache
maintainance.  The pfn_valid checks remain on the dma address as in
the old code, even if that looks a little funny.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
Acked-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
5 years agoswiotlb-xen: use the same foreign page check everywhere
Christoph Hellwig [Mon, 26 Aug 2019 09:23:44 +0000 (11:23 +0200)]
swiotlb-xen: use the same foreign page check everywhere

xen_dma_map_page uses a different and more complicated check for foreign
pages than the other three cache maintainance helpers.  Switch it to the
simpler pfn_valid method a well, and document the scheme with a single
improved comment in xen_dma_map_page.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoswiotlb-xen: remove xen_swiotlb_dma_mmap and xen_swiotlb_dma_get_sgtable
Christoph Hellwig [Mon, 2 Sep 2019 08:45:39 +0000 (10:45 +0200)]
swiotlb-xen: remove xen_swiotlb_dma_mmap and xen_swiotlb_dma_get_sgtable

There is no need to wrap the common version, just wire them up directly.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen: remove the exports for xen_{create,destroy}_contiguous_region
Christoph Hellwig [Wed, 24 Jul 2019 14:01:28 +0000 (16:01 +0200)]
xen: remove the exports for xen_{create,destroy}_contiguous_region

These routines are only used by swiotlb-xen, which cannot be modular.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen/arm: remove xen_dma_ops
Christoph Hellwig [Wed, 24 Jul 2019 12:09:39 +0000 (14:09 +0200)]
xen/arm: remove xen_dma_ops

arm and arm64 can just use xen_swiotlb_dma_ops directly like x86, no
need for a pointer indirection.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen/arm: simplify dma_cache_maint
Christoph Hellwig [Wed, 24 Jul 2019 13:26:08 +0000 (15:26 +0200)]
xen/arm: simplify dma_cache_maint

Calculate the required operation in the caller, and pass it directly
instead of recalculating it for each page, and use simple arithmetics
to get from the physical address to Xen page size aligned chunks.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen/arm: use dev_is_dma_coherent
Christoph Hellwig [Wed, 24 Jul 2019 12:06:54 +0000 (14:06 +0200)]
xen/arm: use dev_is_dma_coherent

Use the dma-noncoherent dev_is_dma_coherent helper instead of the home
grown variant.  Note that both are always initialized to the same
value in arch_setup_dma_ops.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Julien Grall <julien.grall@arm.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen/arm: consolidate page-coherent.h
Christoph Hellwig [Wed, 24 Jul 2019 12:02:10 +0000 (14:02 +0200)]
xen/arm: consolidate page-coherent.h

Shared the duplicate arm/arm64 code in include/xen/arm/page-coherent.h.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoxen/arm: use dma-noncoherent.h calls for xen-swiotlb cache maintainance
Christoph Hellwig [Mon, 2 Sep 2019 08:44:19 +0000 (10:44 +0200)]
xen/arm: use dma-noncoherent.h calls for xen-swiotlb cache maintainance

Copy the arm64 code that uses the dma-direct/swiotlb helpers for DMA
on-coherent devices.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>
5 years agoarm: remove wrappers for the generic dma remap helpers
Christoph Hellwig [Fri, 14 Jun 2019 09:40:28 +0000 (11:40 +0200)]
arm: remove wrappers for the generic dma remap helpers

Remove a few tiny wrappers around the generic dma remap code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: introduce a dma_common_find_pages helper
Christoph Hellwig [Mon, 3 Jun 2019 07:14:31 +0000 (09:14 +0200)]
dma-mapping: introduce a dma_common_find_pages helper

A helper to find the backing page array based on a virtual address.
This also ensures we do the same vm_flags check everywhere instead
of slightly different or missing ones in a few places.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: always use VM_DMA_COHERENT for generic DMA remap
Christoph Hellwig [Fri, 30 Aug 2019 06:51:01 +0000 (08:51 +0200)]
dma-mapping: always use VM_DMA_COHERENT for generic DMA remap

Currently the generic dma remap allocator gets a vm_flags passed by
the caller that is a little confusing.  We just introduced a generic
vmalloc-level flag to identify the dma coherent allocations, so use
that everywhere and remove the now pointless argument.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agovmalloc: lift the arm flag for coherent mappings to common code
Christoph Hellwig [Mon, 3 Jun 2019 06:55:13 +0000 (08:55 +0200)]
vmalloc: lift the arm flag for coherent mappings to common code

The arm architecture had a VM_ARM_DMA_CONSISTENT flag to mark DMA
coherent remapping for a while.  Lift this flag to common code so
that we can use it generically.  We also check it in the only place
VM_USERMAP is directly check so that we can entirely replace that
flag as well (although I'm not even sure why we'd want to allow
remapping DMA appings, but I'd rather not change behavior).

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: provide a better default ->get_required_mask
Christoph Hellwig [Tue, 6 Aug 2019 12:01:38 +0000 (15:01 +0300)]
dma-mapping: provide a better default ->get_required_mask

Most dma_map_ops instances are IOMMUs that work perfectly fine in 32-bits
of IOVA space, and the generic direct mapping code already provides its
own routines that is intelligent based on the amount of memory actually
present.  Wire up the dma-direct routine for the ARM direct mapping code
as well, and otherwise default to the constant 32-bit mask.  This way
we only need to override it for the occasional odd IOMMU that requires
64-bit IOVA support, or IOMMU drivers that are more efficient if they
can fall back to the direct mapping.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: remove the dma_declare_coherent_memory export
Christoph Hellwig [Fri, 30 Aug 2019 06:48:27 +0000 (08:48 +0200)]
dma-mapping: remove the dma_declare_coherent_memory export

dma_declare_coherent_memory is something that the platform setup code
(which pretty much means the device tree these days) need to do so that
drivers can use the memory as declared by the platform.  Drivers
themselves have no business calling this function.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoremoteproc: don't allow modular build
Christoph Hellwig [Fri, 30 Aug 2019 06:43:12 +0000 (08:43 +0200)]
remoteproc: don't allow modular build

Remoteproc started using dma_declare_coherent_memory recently, which is
a bad idea from drivers, and the maintainers agreed to fix that.  But
until that is fixed only allow building the driver built in so that we
can remove the dma_declare_coherent_memory export and prevent other
drivers from "accidentally" using it like remoteproc.  Note that the
driver would also leak the declared coherent memory on unload if it
actually was built as a module at the moment.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Bjorn Andersson <bjorn.andersson@linaro.org>
5 years agodma-mapping: remove the dma_mmap_from_dev_coherent export
Christoph Hellwig [Fri, 30 Aug 2019 06:40:33 +0000 (08:40 +0200)]
dma-mapping: remove the dma_mmap_from_dev_coherent export

dma_mmap_from_dev_coherent is only used by dma_map_ops instances,
none of which is modular.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: remove dma_release_declared_memory
Christoph Hellwig [Fri, 30 Aug 2019 06:46:57 +0000 (08:46 +0200)]
dma-mapping: remove dma_release_declared_memory

This function is entirely unused given that declared memory is
generally provided by platform setup code.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: remove dma_{alloc,free,mmap}_writecombine
Christoph Hellwig [Fri, 26 Jul 2019 06:58:36 +0000 (08:58 +0200)]
dma-mapping: remove dma_{alloc,free,mmap}_writecombine

We can already use DMA_ATTR_WRITE_COMBINE or the _wc prefixed version,
so remove the third way of doing things.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Reviewed-by: Tomi Valkeinen <tomi.valkeinen@ti.com>
5 years agodma-mapping: remove CONFIG_ARCH_NO_COHERENT_DMA_MMAP
Christoph Hellwig [Tue, 6 Aug 2019 12:06:40 +0000 (15:06 +0300)]
dma-mapping: remove CONFIG_ARCH_NO_COHERENT_DMA_MMAP

CONFIG_ARCH_NO_COHERENT_DMA_MMAP is now functionally identical to
!CONFIG_MMU, so remove the separate symbol.  The only difference is that
arm did not set it for !CONFIG_MMU, but arm uses a separate dma mapping
implementation including its own mmap method, which is handled by moving
the CONFIG_MMU check in dma_can_mmap so that is only applies to the
dma-direct case, just as the other ifdefs for it.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org> # m68k
5 years agoparisc: don't set ARCH_NO_COHERENT_DMA_MMAP
Christoph Hellwig [Tue, 6 Aug 2019 11:12:37 +0000 (14:12 +0300)]
parisc: don't set ARCH_NO_COHERENT_DMA_MMAP

parisc is the only architecture that sets ARCH_NO_COHERENT_DMA_MMAP
when an MMU is enabled.  AFAIK this is because parisc CPUs use VIVT
caches, which means exporting normally cachable memory to userspace is
relatively dangrous due to cache aliasing.

But normally cachable memory is only allocated by dma_alloc_coherent
on parisc when using the sba_iommu or ccio_iommu drivers, so just
remove the .mmap implementation for them so that we don't have to set
ARCH_NO_COHERENT_DMA_MMAP, which I plan to get rid of.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoarm-nommu: call dma_mmap_from_dev_coherent directly
Christoph Hellwig [Wed, 7 Aug 2019 10:55:01 +0000 (13:55 +0300)]
arm-nommu: call dma_mmap_from_dev_coherent directly

There is no need to go through dma_common_mmap for the arm-nommu
dma mmap implementation as the only possible memory not handled above
could be that from the per-device coherent pool.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoALSA: pcm: use dma_can_mmap() to check if a device supports dma_mmap_*
Christoph Hellwig [Sat, 3 Aug 2019 10:31:59 +0000 (13:31 +0300)]
ALSA: pcm: use dma_can_mmap() to check if a device supports dma_mmap_*

Replace the local hack with the dma_can_mmap helper to check if
a given device supports mapping DMA allocations to userspace.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Takashi Iwai <tiwai@suse.de>
5 years agodma-mapping: add a dma_can_mmap helper
Christoph Hellwig [Sat, 3 Aug 2019 10:31:25 +0000 (13:31 +0300)]
dma-mapping: add a dma_can_mmap helper

Add a helper to check if DMA allocations for a specific device can be
mapped to userspace using dma_mmap_*.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: explicitly wire up ->mmap and ->get_sgtable
Christoph Hellwig [Tue, 6 Aug 2019 12:01:50 +0000 (15:01 +0300)]
dma-mapping: explicitly wire up ->mmap and ->get_sgtable

While the default ->mmap and ->get_sgtable implementations work for the
majority of our dma_map_ops impementations they are inherently safe
for others that don't use the page allocator or CMA and/or use their
own way of remapping not covered by the common code.  So remove the
defaults if these methods are not wired up, but instead wire up the
default implementations for all safe instances.

Fixes: e1c7e324539a ("dma-mapping: always provide the dma_map_ops based implementation")
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: move the dma_get_sgtable API comments from arm to common code
Christoph Hellwig [Fri, 15 Mar 2019 16:56:43 +0000 (17:56 +0100)]
dma-mapping: move the dma_get_sgtable API comments from arm to common code

The comments are spot on and should be near the central API, not just
near a single implementation.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: fix filename references
Andy Shevchenko [Wed, 19 Jun 2019 14:19:55 +0000 (17:19 +0300)]
dma-mapping: fix filename references

After commit cf65a0f6f6ff ("dma-mapping: move all DMA mapping code to
kernel/dma") some of the files are referring to outdated information,
i.e. old file names of DMA mapping sources. Fix it here.

Note, the lines with "Glue code for..." have been removed completely.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoiommu/dma: add a new dma_map_ops of get_merge_boundary()
Yoshihiro Shimoda [Wed, 28 Aug 2019 12:35:41 +0000 (21:35 +0900)]
iommu/dma: add a new dma_map_ops of get_merge_boundary()

This patch adds a new dma_map_ops of get_merge_boundary() to
expose the DMA merge boundary if the domain type is IOMMU_DOMAIN_DMA.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Acked-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: introduce dma_get_merge_boundary()
Yoshihiro Shimoda [Wed, 28 Aug 2019 12:35:40 +0000 (21:35 +0900)]
dma-mapping: introduce dma_get_merge_boundary()

This patch adds a new DMA API "dma_get_merge_boundary". This function
returns the DMA merge boundary if the DMA layer can merge the segments.
This patch also adds the implementation for a new dma_map_ops pointer.

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agommc: queue: use bigger segments if DMA MAP layer can merge the segments
Yoshihiro Shimoda [Wed, 28 Aug 2019 12:35:43 +0000 (21:35 +0900)]
mmc: queue: use bigger segments if DMA MAP layer can merge the segments

When the max_segs of a mmc host is smaller than 512, the mmc
subsystem tries to use 512 segments if DMA MAP layer can merge
the segments, and then the mmc subsystem exposes such information
to the block layer by using blk_queue_can_use_dma_map_merging().

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Ulf Hansson <ulf.hansson@linaro.org>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au>
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoblock: add a helper function to merge the segments
Yoshihiro Shimoda [Wed, 28 Aug 2019 12:35:42 +0000 (21:35 +0900)]
block: add a helper function to merge the segments

This patch adds a helper function whether a queue can merge
the segments by the DMA MAP layer (e.g. via IOMMU).

Signed-off-by: Yoshihiro Shimoda <yoshihiro.shimoda.uh@renesas.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Reviewed-by: Simon Horman <horms+renesas@verge.net.au
Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoMIPS: document mixing "slightly different CCAs"
Christoph Hellwig [Mon, 26 Aug 2019 07:22:13 +0000 (09:22 +0200)]
MIPS: document mixing "slightly different CCAs"

Based on an email from Paul Burton, quoting section 4.8 "Cacheability and
Coherency Attributes and Access Types" of "MIPS Architecture Volume 1:
Introduction to the MIPS32 Architecture" (MD00080, revision 6.01).

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Paul Burton <paul.burton@mips.com>
5 years agoarm64: document the choice of page attributes for pgprot_dmacoherent
Christoph Hellwig [Sat, 3 Aug 2019 09:38:31 +0000 (12:38 +0300)]
arm64: document the choice of page attributes for pgprot_dmacoherent

Based on an email from Will Deacon.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Will Deacon <will@kernel.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
5 years agodma-mapping: make dma_atomic_pool_init self-contained
Christoph Hellwig [Sat, 3 Aug 2019 09:42:15 +0000 (12:42 +0300)]
dma-mapping: make dma_atomic_pool_init self-contained

The memory allocated for the atomic pool needs to have the same
mapping attributes that we use for remapping, so use
pgprot_dmacoherent instead of open coding it.  Also deduct a
suitable zone to allocate the memory from based on the presence
of the DMA zones.

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agodma-mapping: remove arch_dma_mmap_pgprot
Christoph Hellwig [Mon, 26 Aug 2019 07:03:44 +0000 (09:03 +0200)]
dma-mapping: remove arch_dma_mmap_pgprot

arch_dma_mmap_pgprot is used for two things:

 1) to override the "normal" uncached page attributes for mapping
    memory coherent to devices that can't snoop the CPU caches
 2) to provide the special DMA_ATTR_WRITE_COMBINE semantics on older
    arm systems and some mips platforms

Replace one with the pgprot_dmacoherent macro that is already provided
by arm and much simpler to use, and lift the DMA_ATTR_WRITE_COMBINE
handling to common code with an explicit arch opt-in.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Acked-by: Geert Uytterhoeven <geert@linux-m68k.org> # m68k
Acked-by: Paul Burton <paul.burton@mips.com> # mips
5 years agoarm-nommu: remove the unused pgprot_dmacoherent define
Christoph Hellwig [Thu, 1 Aug 2019 14:14:14 +0000 (17:14 +0300)]
arm-nommu: remove the unused pgprot_dmacoherent define

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agounicore32: remove the unused pgprot_dmacoherent define
Christoph Hellwig [Thu, 1 Aug 2019 14:13:51 +0000 (17:13 +0300)]
unicore32: remove the unused pgprot_dmacoherent define

Signed-off-by: Christoph Hellwig <hch@lst.de>
5 years agoLinux 5.3-rc6
Linus Torvalds [Sun, 25 Aug 2019 19:01:23 +0000 (12:01 -0700)]
Linux 5.3-rc6

5 years agoMerge tag 'auxdisplay-for-linus-v5.3-rc7' of git://github.com/ojeda/linux
Linus Torvalds [Sun, 25 Aug 2019 18:43:17 +0000 (11:43 -0700)]
Merge tag 'auxdisplay-for-linus-v5.3-rc7' of git://github.com/ojeda/linux

Pull auxdisplay cleanup from Miguel Ojeda:
 "Make ht16k33_fb_fix and ht16k33_fb_var constant (Nishka Dasgupta)"

* tag 'auxdisplay-for-linus-v5.3-rc7' of git://github.com/ojeda/linux:
  auxdisplay: ht16k33: Make ht16k33_fb_fix and ht16k33_fb_var constant

5 years agoMerge tag 'for-linus-5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/uml
Linus Torvalds [Sun, 25 Aug 2019 18:40:24 +0000 (11:40 -0700)]
Merge tag 'for-linus-5.3-rc6' of git://git./linux/kernel/git/rw/uml

Pull UML fix from Richard Weinberger:
 "Fix time travel mode"

* tag 'for-linus-5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/uml:
  um: fix time travel mode

5 years agoMerge tag 'for-linus-5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rw...
Linus Torvalds [Sun, 25 Aug 2019 18:29:27 +0000 (11:29 -0700)]
Merge tag 'for-linus-5.3-rc6' of git://git./linux/kernel/git/rw/ubifs

Pull UBIFS and JFFS2 fixes from Richard Weinberger:
 "UBIFS:
   - Don't block too long in writeback_inodes_sb()
   - Fix for a possible overrun of the log head
   - Fix double unlock in orphan_delete()

  JFFS2:
   - Remove C++ style from UAPI header and unbreak picky toolchains"

* tag 'for-linus-5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/rw/ubifs:
  ubifs: Limit the number of pages in shrink_liability
  ubifs: Correctly initialize c->min_log_bytes
  ubifs: Fix double unlock around orphan_delete()
  jffs2: Remove C++ style comments from uapi header

5 years agoMerge branch 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Aug 2019 17:10:15 +0000 (10:10 -0700)]
Merge branch 'x86-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:
 "A few fixes for x86:

   - Fix a boot regression caused by the recent bootparam sanitizing
     change, which escaped the attention of all people who reviewed that
     code.

   - Address a boot problem on machines with broken E820 tables caused
     by an underflow which ended up placing the trampoline start at
     physical address 0.

   - Handle machines which do not advertise a legacy timer of any form,
     but need calibration of the local APIC timer gracefully by making
     the calibration routine independent from the tick interrupt. Marked
     for stable as well as there seems to be quite some new laptops
     rolled out which expose this.

   - Clear the RDRAND CPUID bit on AMD family 15h and 16h CPUs which are
     affected by broken firmware which does not initialize RDRAND
     correctly after resume. Add a command line parameter to override
     this for machine which either do not use suspend/resume or have a
     fixed BIOS. Unfortunately there is no way to detect this on boot,
     so the only safe decision is to turn it off by default.

   - Prevent RFLAGS from being clobbers in CALL_NOSPEC on 32bit which
     caused fast KVM instruction emulation to break.

   - Explain the Intel CPU model naming convention so that the repeating
     discussions come to an end"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/retpoline: Don't clobber RFLAGS during CALL_NOSPEC on i386
  x86/boot: Fix boot regression caused by bootparam sanitizing
  x86/CPU/AMD: Clear RDRAND CPUID bit on AMD family 15h/16h
  x86/boot/compressed/64: Fix boot on machines with broken E820 table
  x86/apic: Handle missing global clockevent gracefully
  x86/cpu: Explain Intel model naming convention

5 years agoMerge branch 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Aug 2019 17:08:01 +0000 (10:08 -0700)]
Merge branch 'timers-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull timekeeping fix from Thomas Gleixner:
 "A single fix for a regression caused by the generic VDSO
  implementation where a math overflow causes CLOCK_BOOTTIME to become a
  random number generator"

* 'timers-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  timekeeping/vsyscall: Prevent math overflow in BOOTTIME update

5 years agoMerge branch 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Aug 2019 17:06:12 +0000 (10:06 -0700)]
Merge branch 'sched-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull scheduler fix from Thomas Gleixner:
 "Handle the worker management in situations where a task is scheduled
  out on a PI lock contention correctly and schedule a new worker if
  possible"

* 'sched-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  sched/core: Schedule new worker even if PI-blocked

5 years agoMerge branch 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Aug 2019 17:03:32 +0000 (10:03 -0700)]
Merge branch 'perf-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Thomas Gleixner:
 "Two small fixes for kprobes and perf:

   - Prevent a deadlock in kprobe_optimizer() causes by reverse lock
     ordering

   - Fix a comment typo"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  kprobes: Fix potential deadlock in kprobe_optimizer()
  perf/x86: Fix typo in comment

5 years agoMerge branch 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 25 Aug 2019 17:00:21 +0000 (10:00 -0700)]
Merge branch 'irq-urgent-for-linus' of git://git./linux/kernel/git/tip/tip

Pull irq fix from Thomas Gleixner:
 "A single fix for a imbalanced kobject operation in the irq decriptor
  code which was unearthed by the new warnings in the kobject code"

* 'irq-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  genirq: Properly pair kobject_del() with kobject_add()

5 years agoMerge branch 'akpm' (patches from Andrew)
Linus Torvalds [Sun, 25 Aug 2019 16:56:27 +0000 (09:56 -0700)]
Merge branch 'akpm' (patches from Andrew)

Mergr misc fixes from Andrew Morton:
 "11 fixes"

Mostly VM fixes, one psi polling fix, and one parisc build fix.

* emailed patches from Andrew Morton <akpm@linux-foundation.org>:
  mm/kasan: fix false positive invalid-free reports with CONFIG_KASAN_SW_TAGS=y
  mm/zsmalloc.c: fix race condition in zs_destroy_pool
  mm/zsmalloc.c: migration can leave pages in ZS_EMPTY indefinitely
  mm, page_owner: handle THP splits correctly
  userfaultfd_release: always remove uffd flags and clear vm_userfaultfd_ctx
  psi: get poll_work to run when calling poll syscall next time
  mm: memcontrol: flush percpu vmevents before releasing memcg
  mm: memcontrol: flush percpu vmstats before releasing memcg
  parisc: fix compilation errrors
  mm, page_alloc: move_freepages should not examine struct page of reserved memory
  mm/z3fold.c: fix race between migration and destruction

5 years agoMerge tag 'dma-mapping-5.3-5' of git://git.infradead.org/users/hch/dma-mapping
Linus Torvalds [Sun, 25 Aug 2019 03:00:11 +0000 (20:00 -0700)]
Merge tag 'dma-mapping-5.3-5' of git://git.infradead.org/users/hch/dma-mapping

Pull dma-mapping fixes from Christoph Hellwig:
 "Two fixes for regressions in this merge window:

   - select the Kconfig symbols for the noncoherent dma arch helpers on
     arm if swiotlb is selected, not just for LPAE to not break then Xen
     build, that uses swiotlb indirectly through swiotlb-xen

   - fix the page allocator fallback in dma_alloc_contiguous if the CMA
     allocation fails"

* tag 'dma-mapping-5.3-5' of git://git.infradead.org/users/hch/dma-mapping:
  dma-direct: fix zone selection after an unaddressable CMA allocation
  arm: select the dma-noncoherent symbols for all swiotlb builds

5 years agomm/kasan: fix false positive invalid-free reports with CONFIG_KASAN_SW_TAGS=y
Andrey Ryabinin [Sun, 25 Aug 2019 00:55:09 +0000 (17:55 -0700)]
mm/kasan: fix false positive invalid-free reports with CONFIG_KASAN_SW_TAGS=y

The code like this:

ptr = kmalloc(size, GFP_KERNEL);
page = virt_to_page(ptr);
offset = offset_in_page(ptr);
kfree(page_address(page) + offset);

may produce false-positive invalid-free reports on the kernel with
CONFIG_KASAN_SW_TAGS=y.

In the example above we lose the original tag assigned to 'ptr', so
kfree() gets the pointer with 0xFF tag.  In kfree() we check that 0xFF
tag is different from the tag in shadow hence print false report.

Instead of just comparing tags, do the following:

1) Check that shadow doesn't contain KASAN_TAG_INVALID.  Otherwise it's
   double-free and it doesn't matter what tag the pointer have.

2) If pointer tag is different from 0xFF, make sure that tag in the
   shadow is the same as in the pointer.

Link: http://lkml.kernel.org/r/20190819172540.19581-1-aryabinin@virtuozzo.com
Fixes: 7f94ffbc4c6a ("kasan: add hooks implementation for tag-based mode")
Signed-off-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reported-by: Walter Wu <walter-zh.wu@mediatek.com>
Reported-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Andrey Konovalov <andreyknvl@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will.deacon@arm.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/zsmalloc.c: fix race condition in zs_destroy_pool
Henry Burns [Sun, 25 Aug 2019 00:55:06 +0000 (17:55 -0700)]
mm/zsmalloc.c: fix race condition in zs_destroy_pool

In zs_destroy_pool() we call flush_work(&pool->free_work).  However, we
have no guarantee that migration isn't happening in the background at
that time.

Since migration can't directly free pages, it relies on free_work being
scheduled to free the pages.  But there's nothing preventing an
in-progress migrate from queuing the work *after*
zs_unregister_migration() has called flush_work().  Which would mean
pages still pointing at the inode when we free it.

Since we know at destroy time all objects should be free, no new
migrations can come in (since zs_page_isolate() fails for fully-free
zspages).  This means it is sufficient to track a "# isolated zspages"
count by class, and have the destroy logic ensure all such pages have
drained before proceeding.  Keeping that state under the class spinlock
keeps the logic straightforward.

In this case a memory leak could lead to an eventual crash if compaction
hits the leaked page.  This crash would only occur if people are
changing their zswap backend at runtime (which eventually starts
destruction).

Link: http://lkml.kernel.org/r/20190809181751.219326-2-henryburns@google.com
Fixes: 48b4800a1c6a ("zsmalloc: page migration support")
Signed-off-by: Henry Burns <henryburns@google.com>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Henry Burns <henrywolfeburns@gmail.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Jonathan Adams <jwadams@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/zsmalloc.c: migration can leave pages in ZS_EMPTY indefinitely
Henry Burns [Sun, 25 Aug 2019 00:55:03 +0000 (17:55 -0700)]
mm/zsmalloc.c: migration can leave pages in ZS_EMPTY indefinitely

In zs_page_migrate() we call putback_zspage() after we have finished
migrating all pages in this zspage.  However, the return value is
ignored.  If a zs_free() races in between zs_page_isolate() and
zs_page_migrate(), freeing the last object in the zspage,
putback_zspage() will leave the page in ZS_EMPTY for potentially an
unbounded amount of time.

To fix this, we need to do the same thing as zs_page_putback() does:
schedule free_work to occur.

To avoid duplicated code, move the sequence to a new
putback_zspage_deferred() function which both zs_page_migrate() and
zs_page_putback() call.

Link: http://lkml.kernel.org/r/20190809181751.219326-1-henryburns@google.com
Fixes: 48b4800a1c6a ("zsmalloc: page migration support")
Signed-off-by: Henry Burns <henryburns@google.com>
Reviewed-by: Sergey Senozhatsky <sergey.senozhatsky@gmail.com>
Cc: Henry Burns <henrywolfeburns@gmail.com>
Cc: Minchan Kim <minchan@kernel.org>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Jonathan Adams <jwadams@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm, page_owner: handle THP splits correctly
Vlastimil Babka [Sun, 25 Aug 2019 00:54:59 +0000 (17:54 -0700)]
mm, page_owner: handle THP splits correctly

THP splitting path is missing the split_page_owner() call that
split_page() has.

As a result, split THP pages are wrongly reported in the page_owner file
as order-9 pages.  Furthermore when the former head page is freed, the
remaining former tail pages are not listed in the page_owner file at
all.  This patch fixes that by adding the split_page_owner() call into
__split_huge_page().

Link: http://lkml.kernel.org/r/20190820131828.22684-2-vbabka@suse.cz
Fixes: a9627bc5e34e ("mm/page_owner: introduce split_page_owner and replace manual handling")
Reported-by: Kirill A. Shutemov <kirill@shutemov.name>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: Mel Gorman <mgorman@techsingularity.net>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agouserfaultfd_release: always remove uffd flags and clear vm_userfaultfd_ctx
Oleg Nesterov [Sun, 25 Aug 2019 00:54:56 +0000 (17:54 -0700)]
userfaultfd_release: always remove uffd flags and clear vm_userfaultfd_ctx

userfaultfd_release() should clear vm_flags/vm_userfaultfd_ctx even if
mm->core_state != NULL.

Otherwise a page fault can see userfaultfd_missing() == T and use an
already freed userfaultfd_ctx.

Link: http://lkml.kernel.org/r/20190820160237.GB4983@redhat.com
Fixes: 04f5866e41fb ("coredump: fix race condition between mmget_not_zero()/get_task_mm() and core dumping")
Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Reported-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Andrea Arcangeli <aarcange@redhat.com>
Tested-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Mike Rapoport <rppt@linux.ibm.com>
Cc: Jann Horn <jannh@google.com>
Cc: Jason Gunthorpe <jgg@mellanox.com>
Cc: Michal Hocko <mhocko@suse.com>
Cc: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agopsi: get poll_work to run when calling poll syscall next time
Jason Xing [Sun, 25 Aug 2019 00:54:53 +0000 (17:54 -0700)]
psi: get poll_work to run when calling poll syscall next time

Only when calling the poll syscall the first time can user receive
POLLPRI correctly.  After that, user always fails to acquire the event
signal.

Reproduce case:
 1. Get the monitor code in Documentation/accounting/psi.txt
 2. Run it, and wait for the event triggered.
 3. Kill and restart the process.

The question is why we can end up with poll_scheduled = 1 but the work
not running (which would reset it to 0).  And the answer is because the
scheduling side sees group->poll_kworker under RCU protection and then
schedules it, but here we cancel the work and destroy the worker.  The
cancel needs to pair with resetting the poll_scheduled flag.

Link: http://lkml.kernel.org/r/1566357985-97781-1-git-send-email-joseph.qi@linux.alibaba.com
Signed-off-by: Jason Xing <kerneljasonxing@linux.alibaba.com>
Signed-off-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Reviewed-by: Caspar Zhang <caspar@linux.alibaba.com>
Reviewed-by: Suren Baghdasaryan <surenb@google.com>
Acked-by: Johannes Weiner <hannes@cmpxchg.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: flush percpu vmevents before releasing memcg
Roman Gushchin [Sun, 25 Aug 2019 00:54:50 +0000 (17:54 -0700)]
mm: memcontrol: flush percpu vmevents before releasing memcg

Similar to vmstats, percpu caching of local vmevents leads to an
accumulation of errors on non-leaf levels.  This happens because some
leftovers may remain in percpu caches, so that they are never propagated
up by the cgroup tree and just disappear into nonexistence with on
releasing of the memory cgroup.

To fix this issue let's accumulate and propagate percpu vmevents values
before releasing the memory cgroup similar to what we're doing with
vmstats.

Since on cpu hotplug we do flush percpu vmstats anyway, we can iterate
only over online cpus.

Link: http://lkml.kernel.org/r/20190819202338.363363-4-guro@fb.com
Fixes: 42a300353577 ("mm: memcontrol: fix recursive statistics correctness & scalabilty")
Signed-off-by: Roman Gushchin <guro@fb.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm: memcontrol: flush percpu vmstats before releasing memcg
Roman Gushchin [Sun, 25 Aug 2019 00:54:47 +0000 (17:54 -0700)]
mm: memcontrol: flush percpu vmstats before releasing memcg

Percpu caching of local vmstats with the conditional propagation by the
cgroup tree leads to an accumulation of errors on non-leaf levels.

Let's imagine two nested memory cgroups A and A/B.  Say, a process
belonging to A/B allocates 100 pagecache pages on the CPU 0.  The percpu
cache will spill 3 times, so that 32*3=96 pages will be accounted to A/B
and A atomic vmstat counters, 4 pages will remain in the percpu cache.

Imagine A/B is nearby memory.max, so that every following allocation
triggers a direct reclaim on the local CPU.  Say, each such attempt will
free 16 pages on a new cpu.  That means every percpu cache will have -16
pages, except the first one, which will have 4 - 16 = -12.  A/B and A
atomic counters will not be touched at all.

Now a user removes A/B.  All percpu caches are freed and corresponding
vmstat numbers are forgotten.  A has 96 pages more than expected.

As memory cgroups are created and destroyed, errors do accumulate.  Even
1-2 pages differences can accumulate into large numbers.

To fix this issue let's accumulate and propagate percpu vmstat values
before releasing the memory cgroup.  At this point these numbers are
stable and cannot be changed.

Since on cpu hotplug we do flush percpu vmstats anyway, we can iterate
only over online cpus.

Link: http://lkml.kernel.org/r/20190819202338.363363-2-guro@fb.com
Fixes: 42a300353577 ("mm: memcontrol: fix recursive statistics correctness & scalabilty")
Signed-off-by: Roman Gushchin <guro@fb.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: Vladimir Davydov <vdavydov.dev@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoparisc: fix compilation errrors
Qian Cai [Sun, 25 Aug 2019 00:54:43 +0000 (17:54 -0700)]
parisc: fix compilation errrors

Commit 0cfaee2af3a0 ("include/asm-generic/5level-fixup.h: fix variable
'p4d' set but not used") converted a few functions from macros to static
inline, which causes parisc to complain,

  In file included from include/asm-generic/4level-fixup.h:38:0,
                   from arch/parisc/include/asm/pgtable.h:5,
                   from arch/parisc/include/asm/io.h:6,
                   from include/linux/io.h:13,
                   from sound/core/memory.c:9:
  include/asm-generic/5level-fixup.h:14:18: error: unknown type name 'pgd_t'; did you mean 'pid_t'?
   #define p4d_t    pgd_t
                    ^
  include/asm-generic/5level-fixup.h:24:28: note: in expansion of macro 'p4d_t'
   static inline int p4d_none(p4d_t p4d)
                              ^~~~~

It is because "4level-fixup.h" is included before "asm/page.h" where
"pgd_t" is defined.

Link: http://lkml.kernel.org/r/20190815205305.1382-1-cai@lca.pw
Fixes: 0cfaee2af3a0 ("include/asm-generic/5level-fixup.h: fix variable 'p4d' set but not used")
Signed-off-by: Qian Cai <cai@lca.pw>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm, page_alloc: move_freepages should not examine struct page of reserved memory
David Rientjes [Sun, 25 Aug 2019 00:54:40 +0000 (17:54 -0700)]
mm, page_alloc: move_freepages should not examine struct page of reserved memory

After commit 907ec5fca3dc ("mm: zero remaining unavailable struct
pages"), struct page of reserved memory is zeroed.  This causes
page->flags to be 0 and fixes issues related to reading
/proc/kpageflags, for example, of reserved memory.

The VM_BUG_ON() in move_freepages_block(), however, assumes that
page_zone() is meaningful even for reserved memory.  That assumption is
no longer true after the aforementioned commit.

There's no reason why move_freepages_block() should be testing the
legitimacy of page_zone() for reserved memory; its scope is limited only
to pages on the zone's freelist.

Note that pfn_valid() can be true for reserved memory: there is a
backing struct page.  The check for page_to_nid(page) is also buggy but
reserved memory normally only appears on node 0 so the zeroing doesn't
affect this.

Move the debug checks to after verifying PageBuddy is true.  This
isolates the scope of the checks to only be for buddy pages which are on
the zone's freelist which move_freepages_block() is operating on.  In
this case, an incorrect node or zone is a bug worthy of being warned
about (and the examination of struct page is acceptable bcause this
memory is not reserved).

Why does move_freepages_block() gets called on reserved memory? It's
simply math after finding a valid free page from the per-zone free area
to use as fallback.  We find the beginning and end of the pageblock of
the valid page and that can bring us into memory that was reserved per
the e820.  pfn_valid() is still true (it's backed by a struct page), but
since it's zero'd we shouldn't make any inferences here about comparing
its node or zone.  The current node check just happens to succeed most
of the time by luck because reserved memory typically appears on node 0.

The fix here is to validate that we actually have buddy pages before
testing if there's any type of zone or node strangeness going on.

We noticed it almost immediately after bringing 907ec5fca3dc in on
CONFIG_DEBUG_VM builds.  It depends on finding specific free pages in
the per-zone free area where the math in move_freepages() will bring the
start or end pfn into reserved memory and wanting to claim that entire
pageblock as a new migratetype.  So the path will be rare, require
CONFIG_DEBUG_VM, and require fallback to a different migratetype.

Some struct pages were already zeroed from reserve pages before
907ec5fca3c so it theoretically could trigger before this commit.  I
think it's rare enough under a config option that most people don't run
that others may not have noticed.  I wouldn't argue against a stable tag
and the backport should be easy enough, but probably wouldn't single out
a commit that this is fixing.

Mel said:

: The overhead of the debugging check is higher with this patch although
: it'll only affect debug builds and the path is not particularly hot.
: If this was a concern, I think it would be reasonable to simply remove
: the debugging check as the zone boundaries are checked in
: move_freepages_block and we never expect a zone/node to be smaller than
: a pageblock and stuck in the middle of another zone.

Link: http://lkml.kernel.org/r/alpine.DEB.2.21.1908122036560.10779@chino.kir.corp.google.com
Signed-off-by: David Rientjes <rientjes@google.com>
Acked-by: Mel Gorman <mgorman@techsingularity.net>
Cc: Naoya Horiguchi <n-horiguchi@ah.jp.nec.com>
Cc: Masayoshi Mizuma <m.mizuma@jp.fujitsu.com>
Cc: Oscar Salvador <osalvador@suse.de>
Cc: Pavel Tatashin <pavel.tatashin@microsoft.com>
Cc: Vlastimil Babka <vbabka@suse.cz>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agomm/z3fold.c: fix race between migration and destruction
Henry Burns [Sun, 25 Aug 2019 00:54:37 +0000 (17:54 -0700)]
mm/z3fold.c: fix race between migration and destruction

In z3fold_destroy_pool() we call destroy_workqueue(&pool->compact_wq).
However, we have no guarantee that migration isn't happening in the
background at that time.

Migration directly calls queue_work_on(pool->compact_wq), if destruction
wins that race we are using a destroyed workqueue.

Link: http://lkml.kernel.org/r/20190809213828.202833-1-henryburns@google.com
Signed-off-by: Henry Burns <henryburns@google.com>
Cc: Vitaly Wool <vitalywool@gmail.com>
Cc: Shakeel Butt <shakeelb@google.com>
Cc: Jonathan Adams <jwadams@google.com>
Cc: Henry Burns <henrywolfeburns@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
5 years agoMerge tag 'gpio-v5.3-4' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux...
Linus Torvalds [Sat, 24 Aug 2019 21:45:33 +0000 (14:45 -0700)]
Merge tag 'gpio-v5.3-4' of git://git./linux/kernel/git/linusw/linux-gpio

Pull GPIO fixes from Linus Walleij:
 "Here is a (hopefully last) set of GPIO fixes for the v5.3 kernel
  cycle. Two are pretty core:

   - Fix not reporting open drain/source lines to userspace as "input"

   - Fix a minor build error found in randconfigs

   - Fix a chip select quirk on the Freescale SPI

   - Fix the irqchip initialization semantic order to reflect what it
     was using the old API"

* tag 'gpio-v5.3-4' of git://git.kernel.org/pub/scm/linux/kernel/git/linusw/linux-gpio:
  gpio: Fix irqchip initialization order
  gpio: of: fix Freescale SPI CS quirk handling
  gpio: Fix build error of function redefinition
  gpiolib: never report open-drain/source lines as 'input' to user-space

5 years agoMerge tag 'hyperv-fixes-signed' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sat, 24 Aug 2019 18:42:06 +0000 (11:42 -0700)]
Merge tag 'hyperv-fixes-signed' of git://git./linux/kernel/git/hyperv/linux

Pull Hyper-V fixes from Sasha Levin:

 - Fix for panics and network failures on PAE guests by Dexuan Cui.

 - Fix of a memory leak (and related cleanups) in the hyper-v keyboard
   driver by Dexuan Cui.

 - Code cleanups for hyper-v clocksource driver during the merge window
   by Dexuan Cui.

 - Fix for a false positive warning in the userspace hyper-v KVP store
   by Vitaly Kuznetsov.

* tag 'hyperv-fixes-signed' of git://git.kernel.org/pub/scm/linux/kernel/git/hyperv/linux:
  Drivers: hv: vmbus: Fix virt_to_hvpfn() for X86_PAE
  Tools: hv: kvp: eliminate 'may be used uninitialized' warning
  Input: hyperv-keyboard: Use in-place iterator API in the channel callback
  Drivers: hv: vmbus: Remove the unused "tsc_page" from struct hv_context

5 years agoMerge tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Linus Torvalds [Sat, 24 Aug 2019 18:35:25 +0000 (11:35 -0700)]
Merge tag 'arm64-fixes' of git://git./linux/kernel/git/arm64/linux

Pull arm64 fixes from Will Deacon:
 "Two KVM/arm fixes for MMIO emulation and UBSAN.

  Unusually, we're routing them via the arm64 tree as per Paolo's
  request on the list:

    https://lore.kernel.org/kvm/21ae69a2-2546-29d0-bff6-2ea825e3d968@redhat.com/

  We don't actually have any other arm64 fixes pending at the moment
  (touch wood), so I've pulled from Marc, written a merge commit, tagged
  the result and run it through my build/boot/bisect scripts"

* tag 'arm64-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux:
  KVM: arm/arm64: VGIC: Properly initialise private IRQ affinity
  KVM: arm/arm64: Only skip MMIO insn once

5 years agoMerge tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi
Linus Torvalds [Sat, 24 Aug 2019 18:26:51 +0000 (11:26 -0700)]
Merge tag 'scsi-fixes' of git://git./linux/kernel/git/jejb/scsi

Pull SCSI fixes from James Bottomley:
 "Four fixes, three for edge conditions which don't occur very often.
  The lpfc fix mitigates memory exhaustion for some high CPU systems"

* tag 'scsi-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi:
  scsi: lpfc: Mitigate high memory pre-allocation by SCSI-MQ
  scsi: ufs: Fix NULL pointer dereference in ufshcd_config_vreg_hpm()
  scsi: target: tcmu: avoid use-after-free after command timeout
  scsi: qla2xxx: Fix gnl.l memory leak on adapter init failure

5 years agoMerge tag 'xfs-5.3-fixes-6' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Sat, 24 Aug 2019 18:21:26 +0000 (11:21 -0700)]
Merge tag 'xfs-5.3-fixes-6' of git://git./fs/xfs/xfs-linux

Pull xfs fix from Darrick Wong:
 "A single patch that fixes a xfs lockup problem when a chown/chgrp
  operation fails due to running out of quota. It has survived the usual
  xfstests runs and merges cleanly with this morning's master:

   - Fix a forgotten inode unlock when chown/chgrp fail due to quota"

* tag 'xfs-5.3-fixes-6' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  xfs: fix missing ILOCK unlock when xfs_setattr_nonsize fails due to EDQUOT

5 years agoMerge tag 'drm-fixes-2019-08-24' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Sat, 24 Aug 2019 18:16:04 +0000 (11:16 -0700)]
Merge tag 'drm-fixes-2019-08-24' of git://anongit.freedesktop.org/drm/drm

Pull more drm fixes from Dave Airlie:
 "Although the tree built for me fine on arm here, it appears either
  header cleanups in next or some kconfig combo it breaks, so this
  contains a fix to mediatek to include dma-mapping.h explicitly.

  There was also one nouveau fix that came in late that I was going to
  leave until next week, but since I was sending this I thought it may
  as well be in here:

  mediatek:
   - fix build in some cases

  nouveau:
   - fix hang with i2c and mst docks"

* tag 'drm-fixes-2019-08-24' of git://anongit.freedesktop.org/drm/drm:
  drm/mediatek: include dma-mapping header
  drm/nouveau: Don't retry infinitely when receiving no data on i2c over AUX

5 years agoMerge tag 'kvmarm-fixes-for-5.3-3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Will Deacon [Sat, 24 Aug 2019 11:45:20 +0000 (12:45 +0100)]
Merge tag 'kvmarm-fixes-for-5.3-3' of git://git./linux/kernel/git/kvmarm/kvmarm into kvm/fixes

Pull KVM/arm fixes from Marc Zyngier as per Paulo's request at:

  https://lkml.kernel.org/r/21ae69a2-2546-29d0-bff6-2ea825e3d968@redhat.com

  "One (hopefully last) set of fixes for KVM/arm for 5.3: an embarassing
   MMIO emulation regression, and a UBSAN splat. Oh well...

   - Don't overskip instructions on MMIO emulation

   - Fix UBSAN splat when initializing PPI priorities"

* tag 'kvmarm-fixes-for-5.3-3' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmarm/kvmarm:
  KVM: arm/arm64: VGIC: Properly initialise private IRQ affinity
  KVM: arm/arm64: Only skip MMIO insn once

5 years agodrm/mediatek: include dma-mapping header
Dave Airlie [Sat, 24 Aug 2019 05:07:07 +0000 (15:07 +1000)]
drm/mediatek: include dma-mapping header

Although it builds fine here in my arm cross compile, it seems
either via some other patches in -next or some Kconfig combination,
this fails to build for everyone.

Include linux/dma-mapping.h should fix it.

Signed-off-by: Dave Airlie <airlied@redhat.com>
5 years agoMerge tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma
Linus Torvalds [Fri, 23 Aug 2019 21:53:09 +0000 (14:53 -0700)]
Merge tag 'for-linus' of git://git./linux/kernel/git/rdma/rdma

Pull rdma fixes from Doug Ledford:
 "No beating around the bush: this is a monster pull request for an -rc5
  kernel. Intel hit me with a series of fixes for TID processing.
  Mellanox hit me with a series for their UMR memory support.

  And we had one fix for siw that fixes the 32bit build warnings and
  because of the number of casts that had to be changed to properly
  silence the warnings, that one patch alone is a full 40% of the LOC of
  this entire pull request. Given that this is the initial release
  kernel for siw, I'm trying to fix anything in it that we can, so that
  adds to the impetus to take fixes for it like this one.

  I had to do a rebase early in the week. Jason had thought he put a
  patch on the rc queue that he needed to be there so he could base some
  work off of it, and it had actually not been placed there. So he asked
  me (on Tuesday) to fix that up before pushing my wip branch to the
  official rc branch. I did, and that's why the early patches look like
  they were all committed at the same time on Tuesday. That bunch had
  been in my queue prior.

  The various patches all pass my test for being legitimate fixes and
  not attempts to slide new features or development into a late rc.
  Well, they were all fixes with the exception of a couple clean up
  patches people wrote for making the fixes they also wrote better (like
  a cleanup patch to move UMR checking into a function so that the
  remaining UMR fix patches can reference that function), so I left
  those in place too.

  My apologies for the LOC count and the number of patches here, it's
  just how the cards fell this cycle.

  Summary:

   - Fix siw buffer mapping issue

   - Fix siw 32/64 casting issues

   - Fix a KASAN access issue in bnxt_re

   - Fix several memory leaks (hfi1, mlx4)

   - Fix a NULL deref in cma_cleanup

   - Fixes for UMR memory support in mlx5 (4 patch series)

   - Fix namespace check for restrack

   - Fixes for counter support

   - Fixes for hfi1 TID processing (5 patch series)

   - Fix potential NULL deref in siw

   - Fix memory page calculations in mlx5"

* tag 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/rdma/rdma: (21 commits)
  RDMA/siw: Fix 64/32bit pointer inconsistency
  RDMA/siw: Fix SGL mapping issues
  RDMA/bnxt_re: Fix stack-out-of-bounds in bnxt_qplib_rcfw_send_message
  infiniband: hfi1: fix memory leaks
  infiniband: hfi1: fix a memory leak bug
  IB/mlx4: Fix memory leaks
  RDMA/cma: fix null-ptr-deref Read in cma_cleanup
  IB/mlx5: Block MR WR if UMR is not possible
  IB/mlx5: Fix MR re-registration flow to use UMR properly
  IB/mlx5: Report and handle ODP support properly
  IB/mlx5: Consolidate use_umr checks into single function
  RDMA/restrack: Rewrite PID namespace check to be reliable
  RDMA/counters: Properly implement PID checks
  IB/core: Fix NULL pointer dereference when bind QP to counter
  IB/hfi1: Drop stale TID RDMA packets that cause TIDErr
  IB/hfi1: Add additional checks when handling TID RDMA WRITE DATA packet
  IB/hfi1: Add additional checks when handling TID RDMA READ RESP packet
  IB/hfi1: Unsafe PSN checking for TID RDMA READ Resp packet
  IB/hfi1: Drop stale TID RDMA packets
  RDMA/siw: Fix potential NULL de-ref
  ...

5 years agoMerge tag 'for-linus-20190823' of git://git.kernel.dk/linux-block
Linus Torvalds [Fri, 23 Aug 2019 21:45:45 +0000 (14:45 -0700)]
Merge tag 'for-linus-20190823' of git://git.kernel.dk/linux-block

Pull block fixes from Jens Axboe:
 "Here's a set of fixes that should go into this release. This contains:

   - Three minor fixes for NVMe.

   - Three minor tweaks for the io_uring polling logic.

   - Officially mark Song as the MD maintainer, after he's been filling
     that role sucessfully for the last 6 months or so"

* tag 'for-linus-20190823' of git://git.kernel.dk/linux-block:
  io_uring: add need_resched() check in inner poll loop
  md: update MAINTAINERS info
  io_uring: don't enter poll loop if we have CQEs pending
  nvme: Add quirk for LiteON CL1 devices running FW 22301111
  nvme: Fix cntlid validation when not using NVMEoF
  nvme-multipath: fix possible I/O hang when paths are updated
  io_uring: fix potential hang with polled IO

5 years agoMerge tag 'for-5.3/dm-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/devic...
Linus Torvalds [Fri, 23 Aug 2019 17:53:34 +0000 (10:53 -0700)]
Merge tag 'for-5.3/dm-fixes-2' of git://git./linux/kernel/git/device-mapper/linux-dm

Pull device mapper fixes from Mike Snitzer:

 - Revert a DM bufio change from during the 5.3 merge window now that a
   proper fix has been made to the block loopback driver.

 - Fix DM kcopyd to wakeup so failed subjobs get completed.

 - Various fixes to DM zoned target to address error handling, and other
   small tweaks (SPDX license identifiers and fix typos).

 - Fix DM integrity range locking race by tracking whether journal has
   changed.

 - Fix DM dust target to detect reads of badblocks beyond the first 512b
   sector (applicable if blocksize is larger than 512b).

 - Fix DM persistent-data issue in both the DM btree and DM
   space-map-metadata interfaces.

 - Fix out of bounds memory access with certain DM table configurations.

* tag 'for-5.3/dm-fixes-2' of git://git.kernel.org/pub/scm/linux/kernel/git/device-mapper/linux-dm:
  dm table: fix invalid memory accesses with too high sector number
  dm space map metadata: fix missing store of apply_bops() return value
  dm btree: fix order of block initialization in btree_split_beneath
  dm raid: add missing cleanup in raid_ctr()
  dm zoned: fix potential NULL dereference in dmz_do_reclaim()
  dm dust: use dust block size for badblocklist index
  dm integrity: fix a crash due to BUG_ON in __journal_read_write()
  dm zoned: fix a few typos
  dm zoned: add SPDX license identifiers
  dm zoned: properly handle backing device failure
  dm zoned: improve error handling in i/o map code
  dm zoned: improve error handling in reclaim
  dm kcopyd: always complete failed jobs
  Revert "dm bufio: fix deadlock with loop device"

5 years agoMerge tag 'xfs-5.3-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux
Linus Torvalds [Fri, 23 Aug 2019 17:49:44 +0000 (10:49 -0700)]
Merge tag 'xfs-5.3-fixes-4' of git://git./fs/xfs/xfs-linux

Pull xfs fixes from Darrick Wong:
 "Here are a few more bug fixes that trickled in since the last pull.
  They've survived the usual xfstests runs and merge cleanly with this
  morning's master.

  I expect there to be one more pull request tomorrow for the fix to
  that quota related inode unlock bug that we were reviewing last night,
  but it will continue to soak in the testing machine for several more
  hours.

   - Fix missing compat ioctl handling for get/setlabel

   - Fix missing ioctl pointer sanitization on s390

   - Fix a page locking deadlock in the dedupe comparison code

   - Fix inadequate locking in reflink code w.r.t. concurrent directio

   - Fix broken error detection when breaking layouts"

* tag 'xfs-5.3-fixes-4' of git://git.kernel.org/pub/scm/fs/xfs/xfs-linux:
  fs/xfs: Fix return code of xfs_break_leased_layouts()
  xfs: fix reflink source file racing with directio writes
  vfs: fix page locking deadlocks when deduping files
  xfs: compat_ioctl: use compat_ptr()
  xfs: fall back to native ioctls for unhandled compat ones

5 years agoKVM: arm/arm64: VGIC: Properly initialise private IRQ affinity
Andre Przywara [Fri, 23 Aug 2019 10:34:16 +0000 (11:34 +0100)]
KVM: arm/arm64: VGIC: Properly initialise private IRQ affinity

At the moment we initialise the target *mask* of a virtual IRQ to the
VCPU it belongs to, even though this mask is only defined for GICv2 and
quickly runs out of bits for many GICv3 guests.
This behaviour triggers an UBSAN complaint for more than 32 VCPUs:
------
[ 5659.462377] UBSAN: Undefined behaviour in virt/kvm/arm/vgic/vgic-init.c:223:21
[ 5659.471689] shift exponent 32 is too large for 32-bit type 'unsigned int'
------
Also for GICv3 guests the reporting of TARGET in the "vgic-state" debugfs
dump is wrong, due to this very same problem.

Because there is no requirement to create the VGIC device before the
VCPUs (and QEMU actually does it the other way round), we can't safely
initialise mpidr or targets in kvm_vgic_vcpu_init(). But since we touch
every private IRQ for each VCPU anyway later (in vgic_init()), we can
just move the initialisation of those fields into there, where we
definitely know the VGIC type.

On the way make sure we really have either a VGICv2 or a VGICv3 device,
since the existing code is just checking for "VGICv3 or not", silently
ignoring the uninitialised case.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reported-by: Dave Martin <dave.martin@arm.com>
Tested-by: Julien Grall <julien.grall@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
5 years agoMerge tag 'modules-for-v5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Fri, 23 Aug 2019 16:22:00 +0000 (09:22 -0700)]
Merge tag 'modules-for-v5.3-rc6' of git://git./linux/kernel/git/jeyu/linux

Pull modules fixes from Jessica Yu:
 "Fix BUG_ON() being triggered in frob_text() due to non-page-aligned
  module sections"

* tag 'modules-for-v5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/jeyu/linux:
  modules: page-align module section allocations only for arches supporting strict module rwx
  modules: always page-align module section allocations

5 years agoMerge tag 'ceph-for-5.3-rc6' of git://github.com/ceph/ceph-client
Linus Torvalds [Fri, 23 Aug 2019 16:19:38 +0000 (09:19 -0700)]
Merge tag 'ceph-for-5.3-rc6' of git://github.com/ceph/ceph-client

Pull ceph fixes from Ilya Dryomov:
 "Three important fixes tagged for stable (an indefinite hang, a crash
  on an assert and a NULL pointer dereference) plus a small series from
  Luis fixing instances of vfree() under spinlock"

* tag 'ceph-for-5.3-rc6' of git://github.com/ceph/ceph-client:
  libceph: fix PG split vs OSD (re)connect race
  ceph: don't try fill file_lock on unsuccessful GETFILELOCK reply
  ceph: clear page dirty before invalidate page
  ceph: fix buffer free while holding i_ceph_lock in fill_inode()
  ceph: fix buffer free while holding i_ceph_lock in __ceph_build_xattrs_blob()
  ceph: fix buffer free while holding i_ceph_lock in __ceph_setxattr()
  libceph: allow ceph_buffer_put() to receive a NULL ceph_buffer

5 years agoRDMA/siw: Fix 64/32bit pointer inconsistency
Bernard Metzler [Thu, 22 Aug 2019 17:37:38 +0000 (19:37 +0200)]
RDMA/siw: Fix 64/32bit pointer inconsistency

Fixes improper casting between addresses and unsigned types.
Changes siw_pbl_get_buffer() function to return appropriate
dma_addr_t, and not u64.

Also fixes debug prints. Now any potentially kernel private
pointers are printed formatted as '%pK', to allow keeping that
information secret.

Fixes: d941bfe500be ("RDMA/siw: Change CQ flags from 64->32 bits")
Fixes: b0fff7317bb4 ("rdma/siw: completion queue methods")
Fixes: 8b6a361b8c48 ("rdma/siw: receive path")
Fixes: b9be6f18cf9e ("rdma/siw: transmit path")
Fixes: f29dd55b0236 ("rdma/siw: queue pair methods")
Fixes: 2251334dcac9 ("rdma/siw: application buffer management")
Fixes: 303ae1cdfdf7 ("rdma/siw: application interface")
Fixes: 6c52fdc244b5 ("rdma/siw: connection management")
Fixes: a531975279f3 ("rdma/siw: main include file")
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Reported-by: Jason Gunthorpe <jgg@ziepe.ca>
Reported-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
Link: https://lore.kernel.org/r/20190822173738.26817-1-bmt@zurich.ibm.com
Signed-off-by: Doug Ledford <dledford@redhat.com>
5 years agoMerge tag 'drm-fixes-2019-08-23' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 23 Aug 2019 16:03:06 +0000 (09:03 -0700)]
Merge tag 'drm-fixes-2019-08-23' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Live from the laundromat after my washing machine broke down, we have
  the 5.3-rc6 fixes. Changelog is in the tag below, but nothing too
  noteworthy in here:

  rcar-du:
   - LVDS dual-link mode fix

  mediatek:
   - of node refcount fix
   - prime buffer import fix
   - dma max seg fix

  komeda:
   - output polling fix
   - abfc format fix
   - memory-region DT fix

  amdgpu:
   - bpc display fix
   - ioctl memory leak fix
   - gfxoff fix
   - smu warnings fix

  i915:
   - HDMI mode readout fix"

* tag 'drm-fixes-2019-08-23' of git://anongit.freedesktop.org/drm/drm:
  drm/amdgpu/powerplay: silence a warning in smu_v11_0_setup_pptable
  drm/amd/display: Calculate bpc based on max_requested_bpc
  drm/amdgpu: prevent memory leaks in AMDGPU_CS ioctl
  drm/amd/amdgpu: disable MMHUB PG for navi10
  drm/amd/powerplay: remove duplicate macro smu_get_uclk_dpm_states in amdgpu_smu.h
  drm/amd/powerplay: fix variable type errors in smu_v11_0_setup_pptable
  drm/amdgpu/gfx9: update pg_flags after determining if gfx off is possible
  drm/i915: Fix HW readout for crtc_clock in HDMI mode
  drm/mediatek: mtk_drm_drv.c: Add of_node_put() before goto
  drm: rcar_lvds: Fix dual link mode operations
  drm/mediatek: set DMA max segment size
  drm/mediatek: use correct device to import PRIME buffers
  drm/omap: ensure we have a valid dma_mask
  drm/komeda: Add support for 'memory-region' DT node property
  drm/komeda: Adds internal bpp computing for arm afbc only format YU08 YU10
  drm/komeda: Initialize and enable output polling on Komeda

5 years agox86/retpoline: Don't clobber RFLAGS during CALL_NOSPEC on i386
Sean Christopherson [Thu, 22 Aug 2019 21:11:22 +0000 (14:11 -0700)]
x86/retpoline: Don't clobber RFLAGS during CALL_NOSPEC on i386

Use 'lea' instead of 'add' when adjusting %rsp in CALL_NOSPEC so as to
avoid clobbering flags.

KVM's emulator makes indirect calls into a jump table of sorts, where
the destination of the CALL_NOSPEC is a small blob of code that performs
fast emulation by executing the target instruction with fixed operands.

  adcb_al_dl:
     0x000339f8 <+0>:   adc    %dl,%al
     0x000339fa <+2>:   ret

A major motiviation for doing fast emulation is to leverage the CPU to
handle consumption and manipulation of arithmetic flags, i.e. RFLAGS is
both an input and output to the target of CALL_NOSPEC.  Clobbering flags
results in all sorts of incorrect emulation, e.g. Jcc instructions often
take the wrong path.  Sans the nops...

  asm("push %[flags]; popf; " CALL_NOSPEC " ; pushf; pop %[flags]\n"
     0x0003595a <+58>:  mov    0xc0(%ebx),%eax
     0x00035960 <+64>:  mov    0x60(%ebx),%edx
     0x00035963 <+67>:  mov    0x90(%ebx),%ecx
     0x00035969 <+73>:  push   %edi
     0x0003596a <+74>:  popf
     0x0003596b <+75>:  call   *%esi
     0x000359a0 <+128>: pushf
     0x000359a1 <+129>: pop    %edi
     0x000359a2 <+130>: mov    %eax,0xc0(%ebx)
     0x000359b1 <+145>: mov    %edx,0x60(%ebx)

  ctxt->eflags = (ctxt->eflags & ~EFLAGS_MASK) | (flags & EFLAGS_MASK);
     0x000359a8 <+136>: mov    -0x10(%ebp),%eax
     0x000359ab <+139>: and    $0x8d5,%edi
     0x000359b4 <+148>: and    $0xfffff72a,%eax
     0x000359b9 <+153>: or     %eax,%edi
     0x000359bd <+157>: mov    %edi,0x4(%ebx)

For the most part this has gone unnoticed as emulation of guest code
that can trigger fast emulation is effectively limited to MMIO when
running on modern hardware, and MMIO is rarely, if ever, accessed by
instructions that affect or consume flags.

Breakage is almost instantaneous when running with unrestricted guest
disabled, in which case KVM must emulate all instructions when the guest
has invalid state, e.g. when the guest is in Big Real Mode during early
BIOS.

Fixes: 776b043848fd2 ("x86/retpoline: Add initial retpoline support")
Fixes: 1a29b5b7f347a ("KVM: x86: Make indirect calls in emulator speculation safe")
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/20190822211122.27579-1-sean.j.christopherson@intel.com
5 years agodm table: fix invalid memory accesses with too high sector number
Mikulas Patocka [Fri, 23 Aug 2019 13:54:09 +0000 (09:54 -0400)]
dm table: fix invalid memory accesses with too high sector number

If the sector number is too high, dm_table_find_target() should return a
pointer to a zeroed dm_target structure (the caller should test it with
dm_target_is_valid).

However, for some table sizes, the code in dm_table_find_target() that
performs btree lookup will access out of bound memory structures.

Fix this bug by testing the sector number at the beginning of
dm_table_find_target(). Also, add an "inline" keyword to the function
dm_table_get_size() because this is a hot path.

Fixes: 512875bd9661 ("dm: table detect io beyond device")
Cc: stable@vger.kernel.org
Reported-by: Zhang Tao <kontais@zoho.com>
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
5 years agogpio: Fix irqchip initialization order
Linus Walleij [Tue, 20 Aug 2019 08:05:27 +0000 (10:05 +0200)]
gpio: Fix irqchip initialization order

The new API for registering a gpio_irq_chip along with a
gpio_chip has a different semantic ordering than the old
API which added the irqchip explicitly after registering
the gpio_chip.

Move the calls to add the gpio_irq_chip *last* in the
function, so that the different hooks setting up OF and
ACPI and machine gpio_chips are called *before* we try
to register the interrupts, preserving the elder semantic
order.

This cropped up in the PL061 driver which used to work
fine with no special ACPI quirks, but started to misbehave
using the new API.

Fixes: e0d897289813 ("gpio: Implement tighter IRQ chip integration")
Cc: Thierry Reding <treding@nvidia.com>
Cc: Grygorii Strashko <grygorii.strashko@ti.com>
Cc: Andy Shevchenko <andy.shevchenko@gmail.com>
Reported-by: Wei Xu <xuwei5@hisilicon.com>
Tested-by: Wei Xu <xuwei5@hisilicon.com>
Reported-by: Andy Shevchenko <andy.shevchenko@gmail.com>
Signed-off-by: Linus Walleij <linus.walleij@linaro.org>
Link: https://lore.kernel.org/r/20190820080527.11796-1-linus.walleij@linaro.org
5 years agoxfs: fix missing ILOCK unlock when xfs_setattr_nonsize fails due to EDQUOT
Darrick J. Wong [Fri, 23 Aug 2019 03:55:54 +0000 (20:55 -0700)]
xfs: fix missing ILOCK unlock when xfs_setattr_nonsize fails due to EDQUOT

Benjamin Moody reported to Debian that XFS partially wedges when a chgrp
fails on account of being out of disk quota.  I ran his reproducer
script:

# adduser dummy
# adduser dummy plugdev

# dd if=/dev/zero bs=1M count=100 of=test.img
# mkfs.xfs test.img
# mount -t xfs -o gquota test.img /mnt
# mkdir -p /mnt/dummy
# chown -c dummy /mnt/dummy
# xfs_quota -xc 'limit -g bsoft=100k bhard=100k plugdev' /mnt

(and then as user dummy)

$ dd if=/dev/urandom bs=1M count=50 of=/mnt/dummy/foo
$ chgrp plugdev /mnt/dummy/foo

and saw:

================================================
WARNING: lock held when returning to user space!
5.3.0-rc5 #rc5 Tainted: G        W
------------------------------------------------
chgrp/47006 is leaving the kernel with locks still held!
1 lock held by chgrp/47006:
 #0: 000000006664ea2d (&xfs_nondir_ilock_class){++++}, at: xfs_ilock+0xd2/0x290 [xfs]

...which is clearly caused by xfs_setattr_nonsize failing to unlock the
ILOCK after the xfs_qm_vop_chown_reserve call fails.  Add the missing
unlock.

Reported-by: benjamin.moody@gmail.com
Fixes: 253f4911f297 ("xfs: better xfs_trans_alloc interface")
Signed-off-by: Darrick J. Wong <darrick.wong@oracle.com>
Reviewed-by: Dave Chinner <dchinner@redhat.com>
Tested-by: Salvatore Bonaccorso <carnil@debian.org>
5 years agoMerge branch 'linux-5.3' of git://github.com/skeggsb/linux into drm-fixes
Dave Airlie [Fri, 23 Aug 2019 03:53:59 +0000 (13:53 +1000)]
Merge branch 'linux-5.3' of git://github.com/skeggsb/linux into drm-fixes

Fixes i2c on DP with some docks.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Ben Skeggs <skeggsb@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/CACAvsv713t2_BQ44gVV7Lqic6Vwmhq0r4FB5v-t0kD1jzFrbmQ@mail.gmail.com
5 years agodrm/nouveau: Don't retry infinitely when receiving no data on i2c over AUX
Lyude Paul [Thu, 25 Jul 2019 19:40:01 +0000 (15:40 -0400)]
drm/nouveau: Don't retry infinitely when receiving no data on i2c over AUX

While I had thought I had fixed this issue in:

commit 342406e4fbba ("drm/nouveau/i2c: Disable i2c bus access after
->fini()")

It turns out that while I did fix the error messages I was seeing on my
P50 when trying to access i2c busses with the GPU in runtime suspend, I
accidentally had missed one important detail that was mentioned on the
bug report this commit was supposed to fix: that the CPU would only lock
up when trying to access i2c busses _on connected devices_ _while the
GPU is not in runtime suspend_. Whoops. That definitely explains why I
was not able to get my machine to hang with i2c bus interactions until
now, as plugging my P50 into it's dock with an HDMI monitor connected
allowed me to finally reproduce this locally.

Now that I have managed to reproduce this issue properly, it looks like
the problem is much simpler then it looks. It turns out that some
connected devices, such as MST laptop docks, will actually ACK i2c reads
even if no data was actually read:

[  275.063043] nouveau 0000:01:00.0: i2c: aux 000a: 1: 0000004c 1
[  275.063447] nouveau 0000:01:00.0: i2c: aux 000a: 00 01101000 10040000
[  275.063759] nouveau 0000:01:00.0: i2c: aux 000a: rd 00000001
[  275.064024] nouveau 0000:01:00.0: i2c: aux 000a: rd 00000000
[  275.064285] nouveau 0000:01:00.0: i2c: aux 000a: rd 00000000
[  275.064594] nouveau 0000:01:00.0: i2c: aux 000a: rd 00000000

Because we don't handle the situation of i2c ack without any data, we
end up entering an infinite loop in nvkm_i2c_aux_i2c_xfer() since the
value of cnt always remains at 0. This finally properly explains how
this could result in a CPU hang like the ones observed in the
aforementioned commit.

So, fix this by retrying transactions if no data is written or received,
and give up and fail the transaction if we continue to not write or
receive any data after 32 retries.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Ben Skeggs <bskeggs@redhat.com>
5 years agodrm/amdgpu/powerplay: silence a warning in smu_v11_0_setup_pptable
Alex Deucher [Thu, 22 Aug 2019 03:25:27 +0000 (22:25 -0500)]
drm/amdgpu/powerplay: silence a warning in smu_v11_0_setup_pptable

I think gcc is confused as I don't see how size could be used
unitialized, but go ahead and silence the warning.

Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Reviewed-by: Evan Quan <evan.quan@amd.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20190822032527.1376-1-alexander.deucher@amd.com
5 years agoMerge tag 'drm-misc-fixes-2019-08-22' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Fri, 23 Aug 2019 01:43:47 +0000 (11:43 +1000)]
Merge tag 'drm-misc-fixes-2019-08-22' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

Fixes for v5.3-rc6:
- dma fix for omap.
- Make output polling work on komeda.
- Fix bpp computing for AFBC formats in komeda.
- Support the memory-region property in komeda.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/5f1fdfe3-814e-fad1-663c-7279217fc085@linux.intel.com
5 years agoMerge tag 'drm-intel-fixes-2019-08-22' of git://anongit.freedesktop.org/drm/drm-intel...
Dave Airlie [Fri, 23 Aug 2019 01:41:58 +0000 (11:41 +1000)]
Merge tag 'drm-intel-fixes-2019-08-22' of git://anongit.freedesktop.org/drm/drm-intel into drm-fixes

drm/i915 fixes for v5.3-rc6:
- fix hardware state readout for 10 bpc HDMI

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Jani Nikula <jani.nikula@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/87sgptd114.fsf@intel.com
5 years agotimekeeping/vsyscall: Prevent math overflow in BOOTTIME update
Thomas Gleixner [Thu, 22 Aug 2019 11:00:15 +0000 (13:00 +0200)]
timekeeping/vsyscall: Prevent math overflow in BOOTTIME update

The VDSO update for CLOCK_BOOTTIME has a overflow issue as it shifts the
nanoseconds based boot time offset left by the clocksource shift. That
overflows once the boot time offset becomes large enough. As a consequence
CLOCK_BOOTTIME in the VDSO becomes a random number causing applications to
misbehave.

Fix it by storing a timespec64 representation of the offset when boot time
is adjusted and add that to the MONOTONIC base time value in the vdso data
page. Using the timespec64 representation avoids a 64bit division in the
update code.

Fixes: 44f57d788e7d ("timekeeping: Provide a generic update_vsyscall() implementation")
Reported-by: Chris Clayton <chris2553@googlemail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Chris Clayton <chris2553@googlemail.com>
Tested-by: Vincenzo Frascino <vincenzo.frascino@arm.com>
Link: https://lkml.kernel.org/r/alpine.DEB.2.21.1908221257580.1983@nanos.tec.linutronix.de
5 years agoum: fix time travel mode
Johannes Berg [Mon, 22 Jul 2019 07:12:56 +0000 (09:12 +0200)]
um: fix time travel mode

Unfortunately, my build fix for when time travel mode isn't
enabled broke time travel mode, because I forgot that we need
to use the timer time after the timer has been marked disabled,
and thus need to leave the time stored instead of zeroing it.

Fix that by splitting the inline into two, so we can call only
the _mode() one in the relevant code path.

Fixes: b482e48d29f1 ("um: fix build without CONFIG_UML_TIME_TRAVEL_SUPPORT")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Richard Weinberger <richard@nod.at>
5 years agoio_uring: add need_resched() check in inner poll loop
Jens Axboe [Thu, 22 Aug 2019 04:19:11 +0000 (22:19 -0600)]
io_uring: add need_resched() check in inner poll loop

The outer poll loop checks for whether we need to reschedule, and
returns to userspace if we do. However, it's possible to get stuck
in the inner loop as well, if the CPU we are running on needs to
reschedule to finish the IO work.

Add the need_resched() check in the inner loop as well. This fixes
a potential hang if the kernel is configured with
CONFIG_PREEMPT_VOLUNTARY=y.

Reported-by: Sagi Grimberg <sagi@grimberg.me>
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Tested-by: Sagi Grimberg <sagi@grimberg.me>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
5 years agoMerge tag 'pci-v5.3-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci
Linus Torvalds [Thu, 22 Aug 2019 21:04:47 +0000 (14:04 -0700)]
Merge tag 'pci-v5.3-fixes-1' of git://git./linux/kernel/git/helgaas/pci

Pull PCI fixes from Bjorn Helgaas:

 - Reset both NVIDIA GPU and HDA in ThinkPad P50 quirk, which was broken
   by another quirk that enabled the HDA device (Lyude Paul)

 - Fix pciebus-howto.rst documentation filename typo (Bjorn Helgaas)

* tag 'pci-v5.3-fixes-1' of git://git.kernel.org/pub/scm/linux/kernel/git/helgaas/pci:
  Documentation PCI: Fix pciebus-howto.rst filename typo
  PCI: Reset both NVIDIA GPU and HDA in ThinkPad P50 workaround

5 years agodm space map metadata: fix missing store of apply_bops() return value
ZhangXiaoxu [Mon, 19 Aug 2019 03:31:21 +0000 (11:31 +0800)]
dm space map metadata: fix missing store of apply_bops() return value

In commit 6096d91af0b6 ("dm space map metadata: fix occasional leak
of a metadata block on resize"), we refactor the commit logic to a new
function 'apply_bops'.  But when that logic was replaced in out() the
return value was not stored.  This may lead out() returning a wrong
value to the caller.

Fixes: 6096d91af0b6 ("dm space map metadata: fix occasional leak of a metadata block on resize")
Cc: stable@vger.kernel.org
Signed-off-by: ZhangXiaoxu <zhangxiaoxu5@huawei.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
5 years agodm btree: fix order of block initialization in btree_split_beneath
ZhangXiaoxu [Sat, 17 Aug 2019 05:32:40 +0000 (13:32 +0800)]
dm btree: fix order of block initialization in btree_split_beneath

When btree_split_beneath() splits a node to two new children, it will
allocate two blocks: left and right.  If right block's allocation
failed, the left block will be unlocked and marked dirty.  If this
happened, the left block'ss content is zero, because it wasn't
initialized with the btree struct before the attempot to allocate the
right block.  Upon return, when flushing the left block to disk, the
validator will fail when check this block.  Then a BUG_ON is raised.

Fix this by completely initializing the left block before allocating and
initializing the right block.

Fixes: 4dcb8b57df359 ("dm btree: fix leak of bufio-backed block in btree_split_beneath error path")
Cc: stable@vger.kernel.org
Signed-off-by: ZhangXiaoxu <zhangxiaoxu5@huawei.com>
Signed-off-by: Mike Snitzer <snitzer@redhat.com>
5 years agoMerge tag 'Wimplicit-fallthrough-5.3-rc6' of git://git.kernel.org/pub/scm/linux/kerne...
Linus Torvalds [Thu, 22 Aug 2019 18:26:10 +0000 (11:26 -0700)]
Merge tag 'Wimplicit-fallthrough-5.3-rc6' of git://git./linux/kernel/git/gustavoars/linux

Pull more fallthrough fixes from Gustavo A. R. Silva:
 "Fix fall-through warnings on arm and mips for multiple configurations"

* tag 'Wimplicit-fallthrough-5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/gustavoars/linux:
  video: fbdev: acornfb: Mark expected switch fall-through
  scsi: libsas: sas_discover: Mark expected switch fall-through
  MIPS: Octeon: Mark expected switch fall-through
  power: supply: ab8500_charger: Mark expected switch fall-through
  watchdog: wdt285: Mark expected switch fall-through
  mtd: sa1100: Mark expected switch fall-through
  drm/sun4i: tcon: Mark expected switch fall-through
  drm/sun4i: sun6i_mipi_dsi: Mark expected switch fall-through
  ARM: riscpc: Mark expected switch fall-through
  dmaengine: fsldma: Mark expected switch fall-through

5 years agoMerge tag 'tag-chrome-platform-fixes-for-v5.3-rc6' of git://git.kernel.org/pub/scm...
Linus Torvalds [Thu, 22 Aug 2019 18:17:20 +0000 (11:17 -0700)]
Merge tag 'tag-chrome-platform-fixes-for-v5.3-rc6' of git://git./linux/kernel/git/chrome-platform/linux

Pull chrome platform fix from Benson Leung:
 "Fix a kernel crash during suspend/resume of cros_ec_ishtp"

* tag 'tag-chrome-platform-fixes-for-v5.3-rc6' of git://git.kernel.org/pub/scm/linux/kernel/git/chrome-platform/linux:
  platform/chrome: cros_ec_ishtp: fix crash during suspend

5 years agoMerge tag 'afs-fixes-20190822' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowe...
Linus Torvalds [Thu, 22 Aug 2019 18:12:33 +0000 (11:12 -0700)]
Merge tag 'afs-fixes-20190822' of git://git./linux/kernel/git/dhowells/linux-fs

Pull AFS fixes from David Howells:

 - Fix a cell record leak due to the default error not being cleared.

 - Fix an oops in tracepoint due to a pointer that may contain an error.

 - Fix the ACL storage op for YFS where the wrong op definition is being
   used. By luck, this only actually affects the information appearing
   in traces.

* tag 'afs-fixes-20190822' of git://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs:
  afs: use correct afs_call_type in yfs_fs_store_opaque_acl2
  afs: Fix possible oops in afs_lookup trace event
  afs: Fix leak in afs_lookup_cell_rcu()

5 years agoubifs: Limit the number of pages in shrink_liability
Liu Song [Tue, 6 Aug 2019 14:21:40 +0000 (22:21 +0800)]
ubifs: Limit the number of pages in shrink_liability

If the number of dirty pages to be written back is large,
then writeback_inodes_sb will block waiting for a long time,
causing hung task detection alarm. Therefore, we should limit
the maximum number of pages written back this time, which let
the budget be completed faster. The remaining dirty pages
tend to rely on the writeback mechanism to complete the
synchronization.

Fixes: b6e51316daed ("writeback: separate starting of sync vs opportunistic writeback")
Signed-off-by: Liu Song <liu.song11@zte.com.cn>
Signed-off-by: Richard Weinberger <richard@nod.at>
5 years agoubifs: Correctly initialize c->min_log_bytes
Richard Weinberger [Tue, 13 Aug 2019 21:55:48 +0000 (23:55 +0200)]
ubifs: Correctly initialize c->min_log_bytes

Currently on a freshly mounted UBIFS, c->min_log_bytes is 0.
This can lead to a log overrun and make commits fail.

Recent kernels will report the following assert:
UBIFS assert failed: c->lhead_lnum != c->ltail_lnum, in fs/ubifs/log.c:412

c->min_log_bytes can have two states, 0 and c->leb_size.
It controls how much bytes of the log area are reserved for non-bud
nodes such as commit nodes.

After a commit it has to be set to c->leb_size such that we have always
enough space for a commit. While a commit runs it can be 0 to make the
remaining bytes of the log available to writers.

Having it set to 0 right after mount is wrong since no space for commits
is reserved.

Fixes: 1e51764a3c2ac ("UBIFS: add new flash file system")
Reported-and-tested-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Richard Weinberger <richard@nod.at>
5 years agoubifs: Fix double unlock around orphan_delete()
Richard Weinberger [Tue, 13 Aug 2019 21:50:51 +0000 (23:50 +0200)]
ubifs: Fix double unlock around orphan_delete()

We unlock after orphan_delete(), so no need to unlock
in the function too.

Reported-by: Han Xu <han.xu@nxp.com>
Fixes: 8009ce956c3d ("ubifs: Don't leak orphans on memory during commit")
Signed-off-by: Richard Weinberger <richard@nod.at>
5 years agojffs2: Remove C++ style comments from uapi header
Masahiro Yamada [Tue, 18 Jun 2019 03:09:26 +0000 (12:09 +0900)]
jffs2: Remove C++ style comments from uapi header

Linux kernel tolerates C++ style comments these days. Actually, the
SPDX License tags for .c files start with //.

On the other hand, uapi headers are written in more strict C, where
the C++ comment style is forbidden.

I simply dropped these lines instead of fixing the comment style.

This code has been always commented out since it was added around
Linux 2.4.9 (i.e. commented out for more than 17 years).

'Maybe later...' will never happen.

Signed-off-by: Masahiro Yamada <yamada.masahiro@socionext.com>
Acked-by: Richard Weinberger <richard@nod.at>
Signed-off-by: Richard Weinberger <richard@nod.at>
5 years agoRDMA/siw: Fix SGL mapping issues
Bernard Metzler [Thu, 22 Aug 2019 15:07:41 +0000 (17:07 +0200)]
RDMA/siw: Fix SGL mapping issues

All user level and most in-kernel applications submit WQEs
where the SG list entries are all of a single type.
iSER in particular, however, will send us WQEs with mixed SG
types: sge[0] = kernel buffer, sge[1] = PBL region.
Check and set is_kva on each SG entry individually instead of
assuming the first SGE type carries through to the last.
This fixes iSER over siw.

Fixes: b9be6f18cf9e ("rdma/siw: transmit path")
Reported-by: Krishnamraju Eraparaju <krishna2@chelsio.com>
Tested-by: Krishnamraju Eraparaju <krishna2@chelsio.com>
Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
Link: https://lore.kernel.org/r/20190822150741.21871-1-bmt@zurich.ibm.com
Signed-off-by: Doug Ledford <dledford@redhat.com>