platform/kernel/linux-rpi.git
3 years agomm: introduce page_needs_cow_for_dma() for deciding whether cow
Peter Xu [Sat, 13 Mar 2021 05:07:26 +0000 (21:07 -0800)]
mm: introduce page_needs_cow_for_dma() for deciding whether cow

We've got quite a few places (pte, pmd, pud) that explicitly checked
against whether we should break the cow right now during fork().  It's
easier to provide a helper, especially before we work the same thing on
hugetlbfs.

Since we'll reference is_cow_mapping() in mm.h, move it there too.
Actually it suites mm.h more since internal.h is mm/ only, but mm.h is
exported to the whole kernel.  With that we should expect another patch to
use is_cow_mapping() whenever we can across the kernel since we do use it
quite a lot but it's always done with raw code against VM_* flags.

Link: https://lkml.kernel.org/r/20210217233547.93892-4-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: David Airlie <airlied@linux.ie>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Gal Pressman <galpress@amazon.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Jann Horn <jannh@google.com>
Cc: Kirill Shutemov <kirill@shutemov.name>
Cc: Kirill Tkhai <ktkhai@virtuozzo.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Miaohe Lin <linmiaohe@huawei.com>
Cc: Mike Kravetz <mike.kravetz@oracle.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Roland Scheidegger <sroland@vmware.com>
Cc: VMware Graphics <linux-graphics-maintainer@vmware.com>
Cc: Wei Zhang <wzam@amazon.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agohugetlb: break earlier in add_reservation_in_range() when we can
Peter Xu [Sat, 13 Mar 2021 05:07:22 +0000 (21:07 -0800)]
hugetlb: break earlier in add_reservation_in_range() when we can

All the regions maintained in hugetlb reserved map is inclusive on "from"
but exclusive on "to".  We can break earlier even if rg->from==t because
it already means no possible intersection.

This does not need a Fixes in all cases because when it happens
(rg->from==t) we'll not break out of the loop while we should, however the
next thing we'd do is still add the last file_region we'd need and quit
the loop in the next round.  So this change is not a bugfix (since the old
code should still run okay iiuc), but we'd better still touch it up to
make it logically sane.

Link: https://lkml.kernel.org/r/20210217233547.93892-3-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: David Airlie <airlied@linux.ie>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Gal Pressman <galpress@amazon.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Jann Horn <jannh@google.com>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Kirill Shutemov <kirill@shutemov.name>
Cc: Kirill Tkhai <ktkhai@virtuozzo.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Roland Scheidegger <sroland@vmware.com>
Cc: VMware Graphics <linux-graphics-maintainer@vmware.com>
Cc: Wei Zhang <wzam@amazon.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agohugetlb: dedup the code to add a new file_region
Peter Xu [Sat, 13 Mar 2021 05:07:18 +0000 (21:07 -0800)]
hugetlb: dedup the code to add a new file_region

Patch series "mm/hugetlb: Early cow on fork, and a few cleanups", v5.

As reported by Gal [1], we still miss the code clip to handle early cow
for hugetlb case, which is true.  Again, it still feels odd to fork()
after using a few huge pages, especially if they're privately mapped to
me..  However I do agree with Gal and Jason in that we should still have
that since that'll complete the early cow on fork effort at least, and
it'll still fix issues where buffers are not well under control and not
easy to apply MADV_DONTFORK.

The first two patches (1-2) are some cleanups I noticed when reading into
the hugetlb reserve map code.  I think it's good to have but they're not
necessary for fixing the fork issue.

The last two patches (3-4) are the real fix.

I tested this with a fork() after some vfio-pci assignment, so I'm pretty
sure the page copy path could trigger well (page will be accounted right
after the fork()), but I didn't do data check since the card I assigned is
some random nic.

  https://github.com/xzpeter/linux/tree/fork-cow-pin-huge

[1] https://lore.kernel.org/lkml/27564187-4a08-f187-5a84-3df50009f6ca@amazon.com/

Introduce hugetlb_resv_map_add() helper to add a new file_region rather
than duplication the similar code twice in add_reservation_in_range().

Link: https://lkml.kernel.org/r/20210217233547.93892-1-peterx@redhat.com
Link: https://lkml.kernel.org/r/20210217233547.93892-2-peterx@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Mike Kravetz <mike.kravetz@oracle.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: Gal Pressman <galpress@amazon.com>
Cc: Matthew Wilcox <willy@infradead.org>
Cc: Wei Zhang <wzam@amazon.com>
Cc: Mike Rapoport <rppt@linux.vnet.ibm.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: David Gibson <david@gibson.dropbear.id.au>
Cc: Jason Gunthorpe <jgg@ziepe.ca>
Cc: Jann Horn <jannh@google.com>
Cc: Kirill Tkhai <ktkhai@virtuozzo.com>
Cc: Kirill Shutemov <kirill@shutemov.name>
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Jan Kara <jack@suse.cz>
Cc: Alexey Dobriyan <adobriyan@gmail.com>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: David Airlie <airlied@linux.ie>
Cc: Roland Scheidegger <sroland@vmware.com>
Cc: VMware Graphics <linux-graphics-maintainer@vmware.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm/fork: clear PASID for new mm
Fenghua Yu [Sat, 13 Mar 2021 05:07:15 +0000 (21:07 -0800)]
mm/fork: clear PASID for new mm

When a new mm is created, its PASID should be cleared, i.e.  the PASID is
initialized to its init state 0 on both ARM and X86.

This patch was part of the series introducing mm->pasid, but got lost
along the way [1].  It still makes sense to have it, because each address
space has a different PASID.  And the IOMMU code in
iommu_sva_alloc_pasid() expects the pasid field of a new mm struct to be
cleared.

[1] https://lore.kernel.org/linux-iommu/YDgh53AcQHT+T3L0@otcwcpicx3.sc.intel.com/

Link: https://lkml.kernel.org/r/20210302103837.2562625-1-jean-philippe@linaro.org
Signed-off-by: Fenghua Yu <fenghua.yu@intel.com>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Cc: Jacob Pan <jacob.jun.pan@intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomm/page_alloc.c: refactor initialization of struct page for holes in memory layout
Mike Rapoport [Sat, 13 Mar 2021 05:07:12 +0000 (21:07 -0800)]
mm/page_alloc.c: refactor initialization of struct page for holes in memory layout

There could be struct pages that are not backed by actual physical memory.
This can happen when the actual memory bank is not a multiple of
SECTION_SIZE or when an architecture does not register memory holes
reserved by the firmware as memblock.memory.

Such pages are currently initialized using init_unavailable_mem() function
that iterates through PFNs in holes in memblock.memory and if there is a
struct page corresponding to a PFN, the fields of this page are set to
default values and it is marked as Reserved.

init_unavailable_mem() does not take into account zone and node the page
belongs to and sets both zone and node links in struct page to zero.

Before commit 73a6e474cb37 ("mm: memmap_init: iterate over memblock
regions rather that check each PFN") the holes inside a zone were
re-initialized during memmap_init() and got their zone/node links right.
However, after that commit nothing updates the struct pages representing
such holes.

On a system that has firmware reserved holes in a zone above ZONE_DMA, for
instance in a configuration below:

# grep -A1 E820 /proc/iomem
7a17b000-7a216fff : Unknown E820 type
7a217000-7bffffff : System RAM

unset zone link in struct page will trigger

VM_BUG_ON_PAGE(!zone_spans_pfn(page_zone(page), pfn), page);

in set_pfnblock_flags_mask() when called with a struct page from a range
other than E820_TYPE_RAM because there are pages in the range of
ZONE_DMA32 but the unset zone link in struct page makes them appear as a
part of ZONE_DMA.

Interleave initialization of the unavailable pages with the normal
initialization of memory map, so that zone and node information will be
properly set on struct pages that are not backed by the actual memory.

With this change the pages for holes inside a zone will get proper
zone/node links and the pages that are not spanned by any node will get
links to the adjacent zone/node.  The holes between nodes will be
prepended to the zone/node above the hole and the trailing pages in the
last section that will be appended to the zone/node below.

[akpm@linux-foundation.org: don't initialize static to zero, use %llu for u64]

Link: https://lkml.kernel.org/r/20210225224351.7356-2-rppt@kernel.org
Fixes: 73a6e474cb37 ("mm: memmap_init: iterate over memblock regions rather that check each PFN")
Signed-off-by: Mike Rapoport <rppt@linux.ibm.com>
Reported-by: Qian Cai <cai@lca.pw>
Reported-by: Andrea Arcangeli <aarcange@redhat.com>
Reviewed-by: Baoquan He <bhe@redhat.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Łukasz Majczak <lma@semihalf.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Mel Gorman <mgorman@suse.de>
Cc: Michal Hocko <mhocko@kernel.org>
Cc: "Sarvela, Tomi P" <tomi.p.sarvela@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoinit/Kconfig: make COMPILE_TEST depend on HAS_IOMEM
Masahiro Yamada [Sat, 13 Mar 2021 05:07:08 +0000 (21:07 -0800)]
init/Kconfig: make COMPILE_TEST depend on HAS_IOMEM

I read the commit log of the following two:

bc083a64b6c0 ("init/Kconfig: make COMPILE_TEST depend on !UML")
334ef6ed06fa ("init/Kconfig: make COMPILE_TEST depend on !S390")

Both are talking about HAS_IOMEM dependency missing in many drivers.

So, 'depends on HAS_IOMEM' seems the direct, sensible solution to me.

This does not change the behavior of UML. UML still cannot enable
COMPILE_TEST because it does not provide HAS_IOMEM.

The current dependency for S390 is too strong. Under the condition of
CONFIG_PCI=y, S390 provides HAS_IOMEM, hence can enable COMPILE_TEST.

I also removed the meaningless 'default n'.

Link: https://lkml.kernel.org/r/20210224140809.1067582-1-masahiroy@kernel.org
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Cc: Heiko Carstens <hca@linux.ibm.com>
Cc: Guenter Roeck <linux@roeck-us.net>
Cc: Arnd Bergmann <arnd@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Johannes Weiner <hannes@cmpxchg.org>
Cc: KP Singh <kpsingh@google.com>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Terrell <terrelln@fb.com>
Cc: Quentin Perret <qperret@google.com>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Cc: "Enrico Weigelt, metux IT consult" <lkml@metux.net>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agostop_machine: mark helpers __always_inline
Arnd Bergmann [Sat, 13 Mar 2021 05:07:04 +0000 (21:07 -0800)]
stop_machine: mark helpers __always_inline

With clang-13, some functions only get partially inlined, with a
specialized version referring to a global variable.  This triggers a
harmless build-time check for the intel-rng driver:

WARNING: modpost: drivers/char/hw_random/intel-rng.o(.text+0xe): Section mismatch in reference from the function stop_machine() to the function .init.text:intel_rng_hw_init()
The function stop_machine() references
the function __init intel_rng_hw_init().
This is often because stop_machine lacks a __init
annotation or the annotation of intel_rng_hw_init is wrong.

In this instance, an easy workaround is to force the stop_machine()
function to be inline, along with related interfaces that did not show the
same behavior at the moment, but theoretically could.

The combination of the two patches listed below triggers the behavior in
clang-13, but individually these commits are correct.

Link: https://lkml.kernel.org/r/20210225130153.1956990-1-arnd@kernel.org
Fixes: fe5595c07400 ("stop_machine: Provide stop_machine_cpuslocked()")
Fixes: ee527cd3a20c ("Use stop_machine_run in the Intel RNG driver")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Prarit Bhargava <prarit@redhat.com>
Cc: Daniel Bristot de Oliveira <bristot@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agomemblock: fix section mismatch warning
Arnd Bergmann [Sat, 13 Mar 2021 05:07:01 +0000 (21:07 -0800)]
memblock: fix section mismatch warning

The inlining logic in clang-13 is rewritten to often not inline some
functions that were inlined by all earlier compilers.

In case of the memblock interfaces, this exposed a harmless bug of a
missing __init annotation:

WARNING: modpost: vmlinux.o(.text+0x507c0a): Section mismatch in reference from the function memblock_bottom_up() to the variable .meminit.data:memblock
The function memblock_bottom_up() references
the variable __meminitdata memblock.
This is often because memblock_bottom_up lacks a __meminitdata
annotation or the annotation of memblock is wrong.

Interestingly, these annotations were present originally, but got removed
with the explanation that the __init annotation prevents the function from
getting inlined.  I checked this again and found that while this is the
case with clang, gcc (version 7 through 10, did not test others) does
inline the functions regardless.

As the previous change was apparently intended to help the clang builds,
reverting it to help the newer clang versions seems appropriate as well.
gcc builds don't seem to care either way.

Link: https://lkml.kernel.org/r/20210225133808.2188581-1-arnd@kernel.org
Fixes: 5bdba520c1b3 ("mm: memblock: drop __init from memblock functions to make it inline")
Reference: 2cfb3665e864 ("include/linux/memblock.h: add __init to memblock_set_bottom_up()")
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Mike Rapoport <rppt@linux.ibm.com>
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: Faiyaz Mohammed <faiyazm@codeaurora.org>
Cc: Baoquan He <bhe@redhat.com>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: Aslan Bakirov <aslan@fb.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoMerge tag 'drm-fixes-2021-03-12-1' of git://anongit.freedesktop.org/drm/drm
Linus Torvalds [Fri, 12 Mar 2021 01:38:49 +0000 (17:38 -0800)]
Merge tag 'drm-fixes-2021-03-12-1' of git://anongit.freedesktop.org/drm/drm

Pull drm fixes from Dave Airlie:
 "Regular fixes for rc3. The i915 pull was based on the rc1 tag so I
  just cherry-picked the single fix from there to avoid it. The misc and
  amd trees seem to be on okay bases.

  It's a bunch of fixes across the tree, amdgpu has most of them a few
  ttm fixes around qxl, and nouveau.

  core:
   - Clear holes when converting compat ioctl's between 32-bits and
     64-bits.

  docs:
   - Use gitlab for drm bugzilla now.

  ttm:
   - Fix ttm page pool accounting.

  fbdev:
   - Fix oops in drm_fbdev_cleanup()

  shmem:
   - Assorted fixes for shmem helpers.

  qxl:
   - unpin qxl bos created as pinned when freeing them, and make ttm
     only warn once on this behavior.
   - Zero head.surface_id correctly in qxl.

  atyfb:
   - Use LCD management for atyfb on PPC_MAC.

  meson:
   - Shutdown kms poll helper in meson correctly.

  nouveau:
   - fix regression in bo syncing

  i915:
   - Wedge the GPU if command parser setup fails

  amdgpu:
   - Fix aux backlight control
   - Add a backlight override parameter
   - Various display fixes
   - PCIe DPM fix for vega
   - Polaris watermark fixes
   - Additional S0ix fix

  radeon:
   - Fix GEM regression
   - Fix AGP dependency handling"

* tag 'drm-fixes-2021-03-12-1' of git://anongit.freedesktop.org/drm/drm: (33 commits)
  drm/nouveau: fix dma syncing for loops (v2)
  drm/i915: Wedge the GPU if command parser setup fails
  drm/compat: Clear bounce structures
  drm/shmem-helpers: vunmap: Don't put pages for dma-buf
  drm: meson_drv add shutdown function
  drm/shmem-helper: Don't remove the offset in vm_area_struct pgoff
  drm/shmem-helper: Check for purged buffers in fault handler
  qxl: Fix uninitialised struct field head.surface_id
  drm/ttm: Fix TTM page pool accounting
  drm/ttm: soften TTM warnings
  drm: Use USB controller's DMA mask when importing dmabufs
  MAINTAINERS: update drm bug reporting URL
  fbdev: atyfb: use LCD management functions for PPC_PMAC also
  fbdev: atyfb: always declare aty_{ld,st}_lcd()
  drm/qxl: fix lockdep issue in qxl_alloc_release_reserved
  drm/qxl: unpin release objects
  drm/fb-helper: only unmap if buffer not null
  drm/amdgpu: fix S0ix handling when the CONFIG_AMD_PMC=m
  drm/radeon: fix AGP dependency
  drm/radeon: also init GEM funcs in radeon_gem_prime_import_sg_table
  ...

3 years agodrm/nouveau: fix dma syncing for loops (v2)
Dave Airlie [Thu, 11 Mar 2021 04:35:27 +0000 (14:35 +1000)]
drm/nouveau: fix dma syncing for loops (v2)

The index variable should only be increased in one place.

Noticed this while trying to track down another oops.

v2: use while loop.

Fixes: f295c8cfec83 ("drm/nouveau: fix dma syncing warning with debugging on.")
Signed-off-by: Dave Airlie <airlied@redhat.com>
Reviewed-by: Michael J. Ruhl <michael.j.ruhl@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210311043527.5376-1-airlied@gmail.com
3 years agodrm/i915: Wedge the GPU if command parser setup fails
Tvrtko Ursulin [Tue, 2 Mar 2021 11:42:13 +0000 (11:42 +0000)]
drm/i915: Wedge the GPU if command parser setup fails

Commit 311a50e76a33 ("drm/i915: Add support for mandatory cmdparsing")
introduced mandatory command parsing but setup failures were not
translated into wedging the GPU which was probably the intent.

Possible errors come in two categories. Either the sanity check on
internal tables has failed, which should be caught in CI unless an
affected platform would be missed in testing; or memory allocation failure
happened during driver load, which should be extremely unlikely but for
correctness should still be handled.

v2:
 * Tidy coding style. (Chris)

[airlied: cherry-picked to avoid rc1 base]
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Fixes: 311a50e76a33 ("drm/i915: Add support for mandatory cmdparsing")
Cc: Jon Bloomfield <jon.bloomfield@intel.com>
Cc: Joonas Lahtinen <joonas.lahtinen@linux.intel.com>
Cc: Chris Wilson <chris.p.wilson@intel.com>
Reviewed-by: Chris Wilson <chris.p.wilson@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210302114213.1102223-1-tvrtko.ursulin@linux.intel.com
(cherry picked from commit 5a1a659762d35a6dc51047c9127c011303c77b7f)
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Dave Airlie <airlied@redhat.com>
3 years agoMerge tag 'amd-drm-fixes-5.12-2021-03-10' of https://gitlab.freedesktop.org/agd5f...
Dave Airlie [Fri, 12 Mar 2021 01:20:02 +0000 (11:20 +1000)]
Merge tag 'amd-drm-fixes-5.12-2021-03-10' of https://gitlab.freedesktop.org/agd5f/linux into drm-fixes

amd-drm-fixes-5.12-2021-03-10:

amdgpu:
- Fix aux backlight control
- Add a backlight override parameter
- Various display fixes
- PCIe DPM fix for vega
- Polaris watermark fixes
- Additional S0ix fix

radeon:
- Fix GEM regression
- Fix AGP dependency handling

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Alex Deucher <alexander.deucher@amd.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210310221141.3974-1-alexander.deucher@amd.com
3 years agoMerge tag 'drm-misc-fixes-2021-03-11' of git://anongit.freedesktop.org/drm/drm-misc...
Dave Airlie [Fri, 12 Mar 2021 00:19:55 +0000 (10:19 +1000)]
Merge tag 'drm-misc-fixes-2021-03-11' of git://anongit.freedesktop.org/drm/drm-misc into drm-fixes

drm-misc-fixes for rc3, rebased on rc2:
- Fix oops in drm_fbdev_cleanup()
- unpin qxl bos created as pinned when freeing them,
  and make ttm only warn once on this behavior.
- Use LCD management for atyfb on PPC_MAC.
- Use gitlab for drm bugzilla now.
- Fix ttm page pool accounting.
- Zero head.surface_id correctly in qxl.
- Assorted fixes for shmem helpers.
- Shutdown kms poll helper in meson correctly.
- Clear holes when converting compat ioctl's between 32-bits and 64-bits.

Signed-off-by: Dave Airlie <airlied@redhat.com>
From: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/4606f08e-d0e8-c543-5e96-cee2fd728a41@linux.intel.com
3 years agoMerge tag 'media/v5.12-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab...
Linus Torvalds [Thu, 11 Mar 2021 18:39:18 +0000 (10:39 -0800)]
Merge tag 'media/v5.12-2' of git://git./linux/kernel/git/mchehab/linux-media

Pull media fixes from Mauro Carvalho Chehab:
 "A couple of fixes:

   - fix a build issue with CEC

   - fix a deadlock at usbtv driver

   - fix some null pointer address issues at vsp1 driver

   - fix a wrong bitmap setting at rkisp1 driver"

* tag 'media/v5.12-2' of git://git.kernel.org/pub/scm/linux/kernel/git/mchehab/linux-media:
  media: rkisp1: params: fix wrong bits settings
  media: v4l: vsp1: Fix uif null pointer access
  media: v4l: vsp1: Fix bru null pointer access
  media: usbtv: Fix deadlock on suspend
  media: rc: compile rc-cec.c into rc-core

3 years agomedia: rkisp1: params: fix wrong bits settings
Dafna Hirschfeld [Mon, 1 Mar 2021 17:18:35 +0000 (18:18 +0100)]
media: rkisp1: params: fix wrong bits settings

The histogram mode is set using 'rkisp1_params_set_bits'.
Only the bits of the mode should be the value argument for
that function. Otherwise bits outside the mode mask are
turned on which is not what was intended.

Fixes: bae1155cf579 ("media: staging: rkisp1: add output device for parameters")
Signed-off-by: Dafna Hirschfeld <dafna.hirschfeld@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
3 years agomedia: v4l: vsp1: Fix uif null pointer access
Biju Das [Mon, 1 Mar 2021 12:08:28 +0000 (13:08 +0100)]
media: v4l: vsp1: Fix uif null pointer access

RZ/G2L SoC has no UIF. This patch fixes null pointer access, when UIF
module is not used.

Fixes: 5e824f989e6e8("media: v4l: vsp1: Integrate DISCOM in display pipeline")
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
3 years agomedia: v4l: vsp1: Fix bru null pointer access
Biju Das [Mon, 1 Mar 2021 12:08:27 +0000 (13:08 +0100)]
media: v4l: vsp1: Fix bru null pointer access

RZ/G2L SoC has only BRS. This patch fixes null pointer access,when only
BRS is enabled.

Fixes: cbb7fa49c7466("media: v4l: vsp1: Rename BRU to BRx")
Signed-off-by: Biju Das <biju.das.jz@bp.renesas.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
3 years agomedia: usbtv: Fix deadlock on suspend
Maxim Mikityanskiy [Fri, 5 Feb 2021 22:51:39 +0000 (23:51 +0100)]
media: usbtv: Fix deadlock on suspend

usbtv doesn't support power management, so on system suspend the
.disconnect callback of the driver is called. The teardown sequence
includes a call to snd_card_free. Its implementation waits until the
refcount of the sound card device drops to zero, however, if its file is
open, snd_card_file_add takes a reference, which can't be dropped during
the suspend, because the userspace processes are already frozen at this
point. snd_card_free waits for completion forever, leading to a hang on
suspend.

This commit fixes this deadlock condition by replacing snd_card_free
with snd_card_free_when_closed, that doesn't wait until all references
are released, allowing suspend to progress.

Fixes: 63ddf68de52e ("[media] usbtv: add audio support")
Signed-off-by: Maxim Mikityanskiy <maxtram95@gmail.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
3 years agomedia: rc: compile rc-cec.c into rc-core
Hans Verkuil [Fri, 26 Feb 2021 10:37:47 +0000 (11:37 +0100)]
media: rc: compile rc-cec.c into rc-core

The rc-cec keymap is unusual in that it can't be built as a module,
instead it is registered directly in rc-main.c if CONFIG_MEDIA_CEC_RC
is set. This is because it can be called from drm_dp_cec_set_edid() via
cec_register_adapter() in an asynchronous context, and it is not
allowed to use request_module() to load rc-cec.ko in that case. Trying to
do so results in a 'WARN_ON_ONCE(wait && current_is_async())'.

Since this keymap is only used if CONFIG_MEDIA_CEC_RC is set, we
just compile this keymap into the rc-core module and never as a
separate module.

Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Fixes: 2c6d1fffa1d9 (drm: add support for DisplayPort CEC-Tunneling-over-AUX)
Reported-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sean Young <sean@mess.org>
Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
3 years agodrm/compat: Clear bounce structures
Daniel Vetter [Mon, 22 Feb 2021 10:06:43 +0000 (11:06 +0100)]
drm/compat: Clear bounce structures

Some of them have gaps, or fields we don't clear. Native ioctl code
does full copies plus zero-extends on size mismatch, so nothing can
leak. But compat is more hand-rolled so need to be careful.

None of these matter for performance, so just memset.

Also I didn't fix up the CONFIG_DRM_LEGACY or CONFIG_DRM_AGP ioctl, those
are security holes anyway.

Acked-by: Maxime Ripard <mripard@kernel.org>
Reported-by: syzbot+620cf21140fc7e772a5d@syzkaller.appspotmail.com # vblank ioctl
Cc: syzbot+620cf21140fc7e772a5d@syzkaller.appspotmail.com
Cc: stable@vger.kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210222100643.400935-1-daniel.vetter@ffwll.ch
(cherry picked from commit e926c474ebee404441c838d18224cd6f246a71b7)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/shmem-helpers: vunmap: Don't put pages for dma-buf
Noralf Trønnes [Fri, 19 Feb 2021 12:22:03 +0000 (13:22 +0100)]
drm/shmem-helpers: vunmap: Don't put pages for dma-buf

dma-buf importing was reworked in commit 7d2cd72a9aa3
("drm/shmem-helpers: Simplify dma-buf importing"). Before that commit
drm_gem_shmem_prime_import_sg_table() did set ->pages_use_count=1 and
drm_gem_shmem_vunmap_locked() could call drm_gem_shmem_put_pages()
unconditionally. Now without the use count set, put pages is called also
on dma-bufs. Fix this by only putting pages if it's not imported.

Signed-off-by: Noralf Trønnes <noralf@tronnes.org>
Fixes: 7d2cd72a9aa3 ("drm/shmem-helpers: Simplify dma-buf importing")
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Tested-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210219122203.51130-1-noralf@tronnes.org
(cherry picked from commit cdea72518a2b38207146e92e1c9e2fac15975679)
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm: meson_drv add shutdown function
Artem Lapkin [Tue, 2 Mar 2021 04:22:02 +0000 (12:22 +0800)]
drm: meson_drv add shutdown function

Problem: random stucks on reboot stage about 1/20 stuck/reboots
// debug kernel log
[    4.496660] reboot: kernel restart prepare CMD:(null)
[    4.498114] meson_ee_pwrc c883c000.system-controller:power-controller: shutdown begin
[    4.503949] meson_ee_pwrc c883c000.system-controller:power-controller: shutdown domain 0:VPU...
...STUCK...

Solution: add shutdown function to meson_drm driver
// debug kernel log
[    5.231896] reboot: kernel restart prepare CMD:(null)
[    5.246135] [drm:meson_drv_shutdown]
...
[    5.259271] meson_ee_pwrc c883c000.system-controller:power-controller: shutdown begin
[    5.274688] meson_ee_pwrc c883c000.system-controller:power-controller: shutdown domain 0:VPU...
[    5.338331] reboot: Restarting system
[    5.358293] psci: PSCI_0_2_FN_SYSTEM_RESET reboot_mode:0 cmd:(null)
bl31 reboot reason: 0xd
bl31 reboot reason: 0x0
system cmd  1.
...REBOOT...

Tested: on VIM1 VIM2 VIM3 VIM3L khadas sbcs - 1000+ successful reboots
and Odroid boards, WeTek Play2 (GXBB)

Fixes: bbbe775ec5b5 ("drm: Add support for Amlogic Meson Graphic Controller")
Signed-off-by: Artem Lapkin <art@khadas.com>
Tested-by: Christian Hewitt <christianshewitt@gmail.com>
Acked-by: Neil Armstrong <narmstrong@baylibre.com>
Acked-by: Kevin Hilman <khilman@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210302042202.3728113-1-art@khadas.com
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/shmem-helper: Don't remove the offset in vm_area_struct pgoff
Neil Roberts [Tue, 23 Feb 2021 15:51:25 +0000 (16:51 +0100)]
drm/shmem-helper: Don't remove the offset in vm_area_struct pgoff

When mmapping the shmem, it would previously adjust the pgoff in the
vm_area_struct to remove the fake offset that is added to be able to
identify the buffer. This patch removes the adjustment and makes the
fault handler use the vm_fault address to calculate the page offset
instead. Although using this address is apparently discouraged, several
DRM drivers seem to be doing it anyway.

The problem with removing the pgoff is that it prevents
drm_vma_node_unmap from working because that searches the mapping tree
by address. That doesn't work because all of the mappings are at offset
0. drm_vma_node_unmap is being used by the shmem helpers when purging
the buffer.

This fixes a bug in Panfrost which is using drm_gem_shmem_purge. Without
this the mapping for the purged buffer can still be accessed which might
mean it would access random pages from other buffers

v2: Don't check whether the unsigned page_offset is less than 0.

Cc: stable@vger.kernel.org
Fixes: 17acb9f35ed7 ("drm/shmem: Add madvise state and purge helpers")
Signed-off-by: Neil Roberts <nroberts@igalia.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210223155125.199577-3-nroberts@igalia.com
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/shmem-helper: Check for purged buffers in fault handler
Neil Roberts [Tue, 23 Feb 2021 15:51:24 +0000 (16:51 +0100)]
drm/shmem-helper: Check for purged buffers in fault handler

When a buffer is madvised as not needed and then purged, any attempts to
access the buffer from user-space should cause a bus fault. This patch
adds a check for that.

Cc: stable@vger.kernel.org
Fixes: 17acb9f35ed7 ("drm/shmem: Add madvise state and purge helpers")
Signed-off-by: Neil Roberts <nroberts@igalia.com>
Reviewed-by: Steven Price <steven.price@arm.com>
Signed-off-by: Steven Price <steven.price@arm.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210223155125.199577-2-nroberts@igalia.com
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agoqxl: Fix uninitialised struct field head.surface_id
Colin Ian King [Thu, 4 Mar 2021 09:49:28 +0000 (09:49 +0000)]
qxl: Fix uninitialised struct field head.surface_id

The surface_id struct field in head is not being initialized and
static analysis warns that this is being passed through to
dev->monitors_config->heads[i] on an assignment. Clear up this
warning by initializing it to zero.

Addresses-Coverity: ("Uninitialized scalar variable")
Fixes: a6d3c4d79822 ("qxl: hook monitors_config updates into crtc, not encoder.")
Signed-off-by: Colin Ian King <colin.king@canonical.com>
Link: http://patchwork.freedesktop.org/patch/msgid/20210304094928.2280722-1-colin.king@canonical.com
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/ttm: Fix TTM page pool accounting
Anthony DeRossi [Wed, 3 Mar 2021 01:17:25 +0000 (17:17 -0800)]
drm/ttm: Fix TTM page pool accounting

Freed pages are not subtracted from the allocated_pages counter in
ttm_pool_type_fini(), causing a leak in the count on device removal.
The next shrinker invocation loops forever trying to free pages that are
no longer in the pool:

  rcu: INFO: rcu_sched self-detected stall on CPU
  rcu:  3-....: (9998 ticks this GP) idle=54e/1/0x4000000000000000 softirq=434857/434857 fqs=2237
    (t=10001 jiffies g=2194533 q=49211)
  NMI backtrace for cpu 3
  CPU: 3 PID: 1034 Comm: kswapd0 Tainted: P           O      5.11.0-com #1
  Hardware name: System manufacturer System Product Name/PRIME X570-PRO, BIOS 1405 11/19/2019
  Call Trace:
   <IRQ>
   ...
   </IRQ>
   sysvec_apic_timer_interrupt+0x77/0x80
   asm_sysvec_apic_timer_interrupt+0x12/0x20
  RIP: 0010:mutex_unlock+0x16/0x20
  Code: e7 48 8b 70 10 e8 7a 53 77 ff eb aa e8 43 6c ff ff 0f 1f 00 65 48 8b 14 25 00 6d 01 00 31 c9 48 89 d0 f0 48 0f b1 0f 48 39 c2 <74> 05 e9 e3 fe ff ff c3 66 90 48 8b 47 20 48 85 c0 74 0f 8b 50 10
  RSP: 0018:ffffbdb840797be8 EFLAGS: 00000246
  RAX: ffff9ff445a41c00 RBX: ffffffffc02a9ef8 RCX: 0000000000000000
  RDX: ffff9ff445a41c00 RSI: ffffbdb840797c78 RDI: ffffffffc02a9ac0
  RBP: 0000000000000080 R08: 0000000000000000 R09: ffffbdb840797c80
  R10: 0000000000000000 R11: fffffffffffffff5 R12: 0000000000000000
  R13: 0000000000000000 R14: 0000000000000084 R15: ffffffffc02a9a60
   ttm_pool_shrink+0x7d/0x90 [ttm]
   ttm_pool_shrinker_scan+0x5/0x20 [ttm]
   do_shrink_slab+0x13a/0x1a0
...

debugfs shows the incorrect total:

  $ cat /sys/kernel/debug/dri/0/ttm_page_pool
            --- 0--- --- 1--- --- 2--- --- 3--- --- 4--- --- 5--- --- 6--- --- 7--- --- 8--- --- 9--- ---10---
  wc      :        0        0        0        0        0        0        0        0        0        0        0
  uc      :        0        0        0        0        0        0        0        0        0        0        0
  wc 32   :        0        0        0        0        0        0        0        0        0        0        0
  uc 32   :        0        0        0        0        0        0        0        0        0        0        0
  DMA uc  :        0        0        0        0        0        0        0        0        0        0        0
  DMA wc  :        0        0        0        0        0        0        0        0        0        0        0
  DMA     :        0        0        0        0        0        0        0        0        0        0        0

  total   :     3029 of  8244261

Using ttm_pool_type_take() to remove pages from the pool before freeing
them correctly accounts for the freed pages.

Fixes: d099fc8f540a ("drm/ttm: new TT backend allocation pool v3")
Signed-off-by: Anthony DeRossi <ajderossi@gmail.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210303011723.22512-1-ajderossi@gmail.com
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/ttm: soften TTM warnings
Christian König [Wed, 3 Mar 2021 14:47:14 +0000 (15:47 +0100)]
drm/ttm: soften TTM warnings

QXL indeed unrefs pinned BOs and the warnings are spamming peoples log files.

Make sure we warn only once until the QXL driver is fixed.

Signed-off-by: Christian König <christian.koenig@amd.com>
References: https://lore.kernel.org/lkml/YD+eYcMMcdlXB8PY@alley/
Link: https://patchwork.freedesktop.org/patch/422834/
Reviewed-by: Daniel Vetter <daniel@ffwll.ch>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm: Use USB controller's DMA mask when importing dmabufs
Thomas Zimmermann [Wed, 3 Mar 2021 13:32:29 +0000 (14:32 +0100)]
drm: Use USB controller's DMA mask when importing dmabufs

USB devices cannot perform DMA and hence have no dma_mask set in their
device structure. Therefore importing dmabuf into a USB-based driver
fails, which breaks joining and mirroring of display in X11.

For USB devices, pick the associated USB controller as attachment device.
This allows the DRM import helpers to perform the DMA setup. If the DMA
controller does not support DMA transfers, we're out of luck and cannot
import. Our current USB-based DRM drivers don't use DMA, so the actual
DMA device is not important.

Tested by joining/mirroring displays of udl and radeon under Gnome/X11.

v8:
* release dmadev if device initialization fails (Noralf)
* fix commit description (Noralf)
v7:
* fix use-before-init bug in gm12u320 (Dan)
v6:
* implement workaround in DRM drivers and hold reference to
  DMA device while USB device is in use
* remove dev_is_usb() (Greg)
* collapse USB helper into usb_intf_get_dma_device() (Alan)
* integrate Daniel's TODO statement (Daniel)
* fix typos (Greg)
v5:
* provide a helper for USB interfaces (Alan)
* add FIXME item to documentation and TODO list (Daniel)
v4:
* implement workaround with USB helper functions (Greg)
* use struct usb_device->bus->sysdev as DMA device (Takashi)
v3:
* drop gem_create_object
* use DMA mask of USB controller, if any (Daniel, Christian, Noralf)
v2:
* move fix to importer side (Christian, Daniel)
* update SHMEM and CMA helpers for new PRIME callbacks

Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Fixes: 6eb0233ec2d0 ("usb: don't inherity DMA properties for USB devices")
Tested-by: Pavel Machek <pavel@ucw.cz>
Reviewed-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Acked-by: Christian König <christian.koenig@amd.com>
Acked-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Acked-by: Noralf Trønnes <noralf@tronnes.org>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: <stable@vger.kernel.org> # v5.10+
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210303133229.3288-1-tzimmermann@suse.de
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agoMAINTAINERS: update drm bug reporting URL
Pavel Turinský [Sun, 28 Feb 2021 16:36:58 +0000 (17:36 +0100)]
MAINTAINERS: update drm bug reporting URL

The original bugzilla seems to be read-only now, linking to the gitlab
for new bugs.

Signed-off-by: Pavel Turinský <ledoian@kam.mff.cuni.cz>
Cc: trivial@kernel.org
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
Link: https://patchwork.freedesktop.org/patch/msgid/20210228163658.54962-1-ledoian@kam.mff.cuni.cz
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agofbdev: atyfb: use LCD management functions for PPC_PMAC also
Randy Dunlap [Fri, 26 Feb 2021 17:30:08 +0000 (09:30 -0800)]
fbdev: atyfb: use LCD management functions for PPC_PMAC also

Include PPC_PMAC in the configs that use aty_ld_lcd() and
aty_st_lcd() implementations so that the PM code may work
correctly for PPC_PMAC.

Suggested-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: linux-fbdev@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: David Airlie <airlied@linux.ie>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210226173008.18236-1-rdunlap@infradead.org
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agofbdev: atyfb: always declare aty_{ld,st}_lcd()
Randy Dunlap [Wed, 24 Feb 2021 21:55:28 +0000 (13:55 -0800)]
fbdev: atyfb: always declare aty_{ld,st}_lcd()

The previously added stubs for aty_{ld,}st_lcd() make it
so that these functions are used regardless of the config
options that were guarding them, so remove the #ifdef/#endif
lines and make their declarations always visible.
This fixes build warnings that were reported by clang:

   drivers/video/fbdev/aty/atyfb_base.c:180:6: warning: no previous prototype for function 'aty_st_lcd' [-Wmissing-prototypes]
   void aty_st_lcd(int index, u32 val, const struct atyfb_par *par)
        ^
   drivers/video/fbdev/aty/atyfb_base.c:180:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
   void aty_st_lcd(int index, u32 val, const struct atyfb_par *par)

   drivers/video/fbdev/aty/atyfb_base.c:183:5: warning: no previous prototype for function 'aty_ld_lcd' [-Wmissing-prototypes]
   u32 aty_ld_lcd(int index, const struct atyfb_par *par)
       ^
   drivers/video/fbdev/aty/atyfb_base.c:183:1: note: declare 'static' if the function is not intended to be used outside of this translation unit
   u32 aty_ld_lcd(int index, const struct atyfb_par *par)

They should not be marked as static since they are used in
mach64_ct.c.

Fixes: bfa5782b9caa ("fbdev: atyfb: add stubs for aty_{ld,st}_lcd()")
Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
Reported-by: kernel test robot <lkp@intel.com>
Cc: Daniel Vetter <daniel.vetter@ffwll.ch>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: linux-fbdev@vger.kernel.org
Cc: dri-devel@lists.freedesktop.org
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: David Airlie <airlied@linux.ie>
Cc: Jani Nikula <jani.nikula@linux.intel.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20210224215528.822-1-rdunlap@infradead.org
Acked-by: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/qxl: fix lockdep issue in qxl_alloc_release_reserved
Gerd Hoffmann [Wed, 17 Feb 2021 12:32:06 +0000 (13:32 +0100)]
drm/qxl: fix lockdep issue in qxl_alloc_release_reserved

Call qxl_bo_unpin (which does a reservation) without holding the
release_mutex lock.  Fixes lockdep (correctly) warning on a possible
deadlock.

Fixes: e8dd3506dcf3 ("drm/qxl: unpin release objects")
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: http://patchwork.freedesktop.org/patch/msgid/20210217123213.2199186-5-kraxel@redhat.com
(cherry picked from commit 19089b760e56c97458c272e90e43da761b05cf12)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/qxl: unpin release objects
Gerd Hoffmann [Thu, 4 Feb 2021 14:57:05 +0000 (15:57 +0100)]
drm/qxl: unpin release objects

Balances the qxl_create_bo(..., pinned=true, ...);
call in qxl_release_bo_alloc().

Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Acked-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: http://patchwork.freedesktop.org/patch/msgid/20210204145712.1531203-5-kraxel@redhat.com
(cherry picked from commit 65ffea3c6e738f37bb15ff3ee480415c793df893)
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/fb-helper: only unmap if buffer not null
Tong Zhang [Sun, 28 Feb 2021 04:46:25 +0000 (23:46 -0500)]
drm/fb-helper: only unmap if buffer not null

drm_fbdev_cleanup() can be called when fb_helper->buffer is null, hence
fb_helper->buffer should be checked before calling
drm_client_buffer_vunmap(). This buffer is also checked in
drm_client_framebuffer_delete(), so we should also do the same thing for
drm_client_buffer_vunmap().

[  199.128742] RIP: 0010:drm_client_buffer_vunmap+0xd/0x20
[  199.129031] Code: 43 18 48 8b 53 20 49 89 45 00 49 89 55 08 5b 44 89 e0 41 5c 41 5d 41 5e 5d
c3 0f 1f 00 53 48 89 fb 48 8d 7f 10 e8 73 7d a1 ff <48> 8b 7b 10 48 8d 73 18 5b e9 75 53 fc ff 0
f 1f 44 00 00 48 b8 00
[  199.130041] RSP: 0018:ffff888103f3fc88 EFLAGS: 00010282
[  199.130329] RAX: 0000000000000001 RBX: 0000000000000000 RCX: ffffffff8214d46d
[  199.130733] RDX: 1ffffffff079c6b9 RSI: 0000000000000246 RDI: ffffffff83ce35c8
[  199.131119] RBP: ffff888103d25458 R08: 0000000000000001 R09: fffffbfff0791761
[  199.131505] R10: ffffffff83c8bb07 R11: fffffbfff0791760 R12: 0000000000000000
[  199.131891] R13: ffff888103d25468 R14: ffff888103d25418 R15: ffff888103f18120
[  199.132277] FS:  00007f36fdcbb6a0(0000) GS:ffff88815b400000(0000) knlGS:0000000000000000
[  199.132721] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  199.133033] CR2: 0000000000000010 CR3: 0000000103d26000 CR4: 00000000000006f0
[  199.133420] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
[  199.133807] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
[  199.134195] Call Trace:
[  199.134333]  drm_fbdev_cleanup+0x179/0x1a0
[  199.134562]  drm_fbdev_client_unregister+0x2b/0x40
[  199.134828]  drm_client_dev_unregister+0xa8/0x180
[  199.135088]  drm_dev_unregister+0x61/0x110
[  199.135315]  mgag200_pci_remove+0x38/0x52 [mgag200]
[  199.135586]  pci_device_remove+0x62/0xe0
[  199.135806]  device_release_driver_internal+0x148/0x270
[  199.136094]  driver_detach+0x76/0xe0
[  199.136294]  bus_remove_driver+0x7e/0x100
[  199.136521]  pci_unregister_driver+0x28/0xf0
[  199.136759]  __x64_sys_delete_module+0x268/0x300
[  199.137016]  ? __ia32_sys_delete_module+0x300/0x300
[  199.137285]  ? call_rcu+0x3e4/0x580
[  199.137481]  ? fpregs_assert_state_consistent+0x4d/0x60
[  199.137767]  ? exit_to_user_mode_prepare+0x2f/0x130
[  199.138037]  do_syscall_64+0x33/0x40
[  199.138237]  entry_SYSCALL_64_after_hwframe+0x44/0xae
[  199.138517] RIP: 0033:0x7f36fdc3dcf7

Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Fixes: 763aea17bf57 ("drm/fb-helper: Unmap client buffer during shutdown")
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Cc: Sam Ravnborg <sam@ravnborg.org>
Cc: Maxime Ripard <mripard@kernel.org>
Cc: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
Cc: David Airlie <airlied@linux.ie>
Cc: Daniel Vetter <daniel@ffwll.ch>
Cc: dri-devel@lists.freedesktop.org
Cc: <stable@vger.kernel.org> # v5.11+
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20210228044625.171151-1-ztong0001@gmail.com
Signed-off-by: Maarten Lankhorst <maarten.lankhorst@linux.intel.com>
3 years agodrm/amdgpu: fix S0ix handling when the CONFIG_AMD_PMC=m
Alex Deucher [Wed, 10 Mar 2021 03:58:47 +0000 (22:58 -0500)]
drm/amdgpu: fix S0ix handling when the CONFIG_AMD_PMC=m

Need to check the module variant as well.

Acked-by: Prike Liang <Prike.Liang@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/radeon: fix AGP dependency
Christian König [Mon, 8 Mar 2021 18:22:13 +0000 (19:22 +0100)]
drm/radeon: fix AGP dependency

When AGP is compiled as module radeon must be compiled as module as
well.

Signed-off-by: Christian König <christian.koenig@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
3 years agodrm/radeon: also init GEM funcs in radeon_gem_prime_import_sg_table
Christian König [Mon, 8 Mar 2021 18:35:14 +0000 (19:35 +0100)]
drm/radeon: also init GEM funcs in radeon_gem_prime_import_sg_table

Otherwise we will run into a NULL ptr deref.

Signed-off-by: Christian König <christian.koenig@amd.com>
Bug: https://bugzilla.kernel.org/show_bug.cgi?id=212137
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org # 5.11.x
3 years agodrm/amd/pm: correct the watermark settings for Polaris
Evan Quan [Fri, 5 Mar 2021 06:21:26 +0000 (14:21 +0800)]
drm/amd/pm: correct the watermark settings for Polaris

The "/ 10" should be applied to the right-hand operand instead of
the left-hand one.

Signed-off-by: Evan Quan <evan.quan@amd.com>
Noticed-by: Georgios Toptsidis <gtoptsid@gmail.com>
Reviewed-by: Feifei Xu <Feifei.Xu@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/amd/pm: bug fix for pcie dpm
Kenneth Feng [Tue, 9 Mar 2021 13:10:16 +0000 (21:10 +0800)]
drm/amd/pm: bug fix for pcie dpm

Currently the pcie dpm has two problems.
1. Only the high dpm level speed/width can be overrided
if the requested values are out of the pcie capability.
2. The high dpm level is always overrided though sometimes
it's not necesarry.

Signed-off-by: Kenneth Feng <kenneth.feng@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/amdgpu: fb BO should be ttm_bo_type_device
Nirmoy Das [Mon, 8 Mar 2021 14:22:22 +0000 (15:22 +0100)]
drm/amdgpu: fb BO should be ttm_bo_type_device

FB BO should not be ttm_bo_type_kernel type and
amdgpufb_create_pinned_object() pins the FB BO anyway.

Signed-off-by: Nirmoy Das <nirmoy.das@amd.com>
Acked-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
3 years agodrm/amdgpu/display: Use wm_table.entries for dcn301 calculate_wm
Zhan Liu [Tue, 9 Mar 2021 01:28:22 +0000 (20:28 -0500)]
drm/amdgpu/display: Use wm_table.entries for dcn301 calculate_wm

[Why]
For DGPU Navi, the wm_table.nv_entries are used. These entires are not
populated for DCN301 Vangogh APU, but instead wm_table.entries are.

[How]
Use DCN21 Renoir style wm calculations.

Signed-off-by: Leo Li <sunpeng.li@amd.com>
Signed-off-by: Zhan Liu <zhan.liu@amd.com>
Reviewed-by: Dmytro Laktyushkin <Dmytro.Laktyushkin@amd.com>
Acked-by: Zhan Liu <zhan.liu@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
3 years agodrm/amd/display: Enabled pipe harvesting in dcn30
Dillon Varone [Fri, 19 Feb 2021 23:15:30 +0000 (18:15 -0500)]
drm/amd/display: Enabled pipe harvesting in dcn30

[Why & How]
Ported logic from dcn21 for reading in pipe fusing to dcn30.
Supported configurations are 1 and 6 pipes. Invalid fusing
will revert to 1 pipe being enabled.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Dillon Varone <dillon.varone@amd.com>
Reviewed-by: Jun Lei <Jun.Lei@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
3 years agodrm/amd/display: Revert dram_clock_change_latency for DCN2.1
Sung Lee [Fri, 26 Feb 2021 18:20:43 +0000 (13:20 -0500)]
drm/amd/display: Revert dram_clock_change_latency for DCN2.1

[WHY & HOW]
Using values provided by DF for latency may cause hangs in
multi display configurations. Revert change to previous value.

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Sung Lee <sung.lee@amd.com>
Reviewed-by: Haonan Wang <Haonan.Wang2@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
3 years agoMerge tag 's390-5.12-3' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux
Linus Torvalds [Wed, 10 Mar 2021 21:15:16 +0000 (13:15 -0800)]
Merge tag 's390-5.12-3' of git://git./linux/kernel/git/s390/linux

Pull s390 fixes from Heiko Carstens:

 - fix various user space visible copy_to_user() instances which return
   the number of bytes left to copy instead of -EFAULT

 - make TMPFS_INODE64 available again for s390 and alpha, now that both
   architectures have been switched to 64-bit ino_t (see commit
   96c0a6a72d18: "s390,alpha: switch to 64-bit ino_t")

 - make sure to release a shared hypervisor resource within the zcore
   device driver also on restart and power down; also remove unneeded
   surrounding debugfs_create return value checks

 - for the new hardware counter set device driver rename the uapi header
   file to be a bit more generic; also remove 60 second read limit which
   is not really necessary and without the limit the interface can be
   easier tested

 - some small cleanups, the largest being to convert all long long in
   our time and idle code to longs

 - update defconfigs

* tag 's390-5.12-3' of git://git.kernel.org/pub/scm/linux/kernel/git/s390/linux:
  s390: remove IBM_PARTITION and CONFIGFS_FS from zfcpdump defconfig
  s390: update defconfigs
  s390,alpha: make TMPFS_INODE64 available again
  s390/cio: return -EFAULT if copy_to_user() fails
  s390/tty3270: avoid comma separated statements
  s390/cpumf: remove unneeded semicolon
  s390/crypto: return -EFAULT if copy_to_user() fails
  s390/cio: return -EFAULT if copy_to_user() fails
  s390/cpumf: rename header file to hwctrset.h
  s390/zcore: release dump save area on restart or power down
  s390/zcore: no need to check return value of debugfs_create functions
  s390/cpumf: remove 60 seconds read limit
  s390/topology: remove always false if check
  s390/time,idle: get rid of unsigned long long

3 years agodrm/amd/display: Enable pflip interrupt upon pipe enable
Qingqing Zhuo [Fri, 19 Feb 2021 22:17:50 +0000 (17:17 -0500)]
drm/amd/display: Enable pflip interrupt upon pipe enable

[Why]
pflip interrupt would not be enabled promptly if a pipe is disabled
and re-enabled, causing flip_done timeout error during DP
compliance tests

[How]
Enable pflip interrupt upon pipe enablement

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Reviewed-by: Nicholas Kazlauskas <Nicholas.Kazlauskas@amd.com>
Acked-by: Eryk Brol <eryk.brol@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
3 years agodrm/amdgpu/display: use GFP_ATOMIC in dcn21_validate_bandwidth_fp()
Holger Hoffstätte [Fri, 5 Mar 2021 14:23:18 +0000 (15:23 +0100)]
drm/amdgpu/display: use GFP_ATOMIC in dcn21_validate_bandwidth_fp()

After fixing nested FPU contexts caused by 41401ac67791 we're still seeing
complaints about spurious kernel_fpu_end(). As it turns out this was
already fixed for dcn20 in commit f41ed88cbd ("drm/amdgpu/display:
use GFP_ATOMIC in dcn20_validate_bandwidth_internal") but never moved
forward to dcn21.

Signed-off-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/amd/display: Fix nested FPU context in dcn21_validate_bandwidth()
Holger Hoffstätte [Fri, 5 Mar 2021 11:39:21 +0000 (12:39 +0100)]
drm/amd/display: Fix nested FPU context in dcn21_validate_bandwidth()

Commit 41401ac67791 added FPU wrappers to dcn21_validate_bandwidth(),
which was correct. Unfortunately a nested function alredy contained
DC_FP_START()/DC_FP_END() calls, which results in nested FPU context
enter/exit and complaints by kernel_fpu_begin_mask().
This can be observed e.g. with 5.10.20, which backported 41401ac67791
and now emits the following warning on boot:

WARNING: CPU: 6 PID: 858 at arch/x86/kernel/fpu/core.c:129 kernel_fpu_begin_mask+0xa5/0xc0
Call Trace:
 dcn21_calculate_wm+0x47/0xa90 [amdgpu]
 dcn21_validate_bandwidth_fp+0x15d/0x2b0 [amdgpu]
 dcn21_validate_bandwidth+0x29/0x40 [amdgpu]
 dc_validate_global_state+0x3c7/0x4c0 [amdgpu]

The warning is emitted due to the additional DC_FP_START/END calls in
patch_bounding_box(), which is inlined into dcn21_calculate_wm(),
its only caller. Removing the calls brings the code in line with
dcn20 and makes the warning disappear.

Fixes: 41401ac67791 ("drm/amd/display: Add FPU wrappers to dcn21_validate_bandwidth()")
Signed-off-by: Holger Hoffstätte <holger@applied-asynchrony.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/amd/display: Add a backlight module option
Takashi Iwai [Wed, 3 Feb 2021 12:42:41 +0000 (13:42 +0100)]
drm/amd/display: Add a backlight module option

There seem devices that don't work with the aux channel backlight
control.  For allowing such users to test with the other backlight
control method, provide a new module option, aux_backlight, to specify
enabling or disabling the aux backport support explicitly.  As
default, the aux support is detected by the hardware capability.

v2: make the backlight option generic in case we add future
backlight types (Alex)

BugLink: https://bugzilla.opensuse.org/show_bug.cgi?id=1180749
BugLink: https://gitlab.freedesktop.org/drm/amd/-/issues/1438
Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/amdgpu/display: handle aux backlight in backlight_get_brightness
Alex Deucher [Thu, 10 Dec 2020 06:45:12 +0000 (01:45 -0500)]
drm/amdgpu/display: handle aux backlight in backlight_get_brightness

Need to fetch it via aux.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/amdgpu/display: don't assert in set backlight function
Alex Deucher [Thu, 10 Dec 2020 06:20:08 +0000 (01:20 -0500)]
drm/amdgpu/display: don't assert in set backlight function

It just spams the logs.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agodrm/amdgpu/display: simplify backlight setting
Alex Deucher [Thu, 10 Dec 2020 06:18:40 +0000 (01:18 -0500)]
drm/amdgpu/display: simplify backlight setting

Avoid the extra wrapper function.

Reviewed-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
3 years agoRevert "mm, slub: consider rest of partial list if acquire_slab() fails"
Linus Torvalds [Wed, 10 Mar 2021 18:18:04 +0000 (10:18 -0800)]
Revert "mm, slub: consider rest of partial list if acquire_slab() fails"

This reverts commit 8ff60eb052eeba95cfb3efe16b08c9199f8121cf.

The kernel test robot reports a huge performance regression due to the
commit, and the reason seems fairly straightforward: when there is
contention on the page list (which is what causes acquire_slab() to
fail), we do _not_ want to just loop and try again, because that will
transfer the contention to the 'n->list_lock' spinlock we hold, and
just make things even worse.

This is admittedly likely a problem only on big machines - the kernel
test robot report comes from a 96-thread dual socket Intel Xeon Gold
6252 setup, but the regression there really is quite noticeable:

   -47.9% regression of stress-ng.rawpkt.ops_per_sec

and the commit that was marked as being fixed (7ced37197196: "slub:
Acquire_slab() avoid loop") actually did the loop exit early very
intentionally (the hint being that "avoid loop" part of that commit
message), exactly to avoid this issue.

The correct thing to do may be to pick some kind of reasonable middle
ground: instead of breaking out of the loop on the very first sign of
contention, or trying over and over and over again, the right thing may
be to re-try _once_, and then give up on the second failure (or pick
your favorite value for "once"..).

Reported-by: kernel test robot <oliver.sang@intel.com>
Link: https://lore.kernel.org/lkml/20210301080404.GF12822@xsang-OptiPlex-9020/
Cc: Jann Horn <jannh@google.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Joonsoo Kim <iamjoonsoo.kim@lge.com>
Acked-by: Christoph Lameter <cl@linux.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
3 years agoMerge tag 'for-linus-2021-03-10' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Wed, 10 Mar 2021 18:01:35 +0000 (10:01 -0800)]
Merge tag 'for-linus-2021-03-10' of git://git./linux/kernel/git/brauner/linux

Pull detached mounts fix from Christian Brauner:
 "Creating a series of detached mounts, attaching them to the
  filesystem, and unmounting them can be used to trigger an integer
  overflow in ns->mounts causing the kernel to block any new mounts in
  count_mounts() and returning ENOSPC because it falsely assumes that
  the maximum number of mounts in the mount namespace has been reached,
  i.e. it thinks it can't fit the new mounts into the mount namespace
  anymore.

  Without this fix heavy use of the new mount API with move_mount() will
  cause the host to become unuseable and thus blocks some xfstest
  patches I want to resend.

  Depending on the number of mounts in your system, this can be
  reproduced on any kernel that supportes open_tree() and move_mount().

  A reproducer has been sent for inclusion with xfstests. It takes care
  to do this in another mount namespace, not in the host's mount
  namespace so there shouldn't be any risk in running it but if one did
  run it on the host it would require a reboot in order to be able to
  mount again. See

      https://lore.kernel.org/fstests/20210309121041.753359-1-christian.brauner@ubuntu.com

  The root cause of this is that detached mounts aren't handled
  correctly when source and target mount are identical and reside on a
  shared mount causing a broken mount tree where the detached source
  itself is propagated which propagation prevents for regular
  bind-mounts and new mounts.

  This ultimately leads to a miscalculation of the number of mounts in
  the mount namespace.

  Detached mounts created via 'open_tree(fd, path, OPEN_TREE_CLONE)' are
  essentially like an unattached bind-mount. They can then later on be
  attached to the filesystem via move_mount() which calls into
  attach_recursive_mount().

  Part of attaching it to the filesystem is making sure that mounts get
  correctly propagated in case the destination mountpoint is MS_SHARED,
  i.e. is a shared mountpoint. This is done by calling into
  propagate_mnt() which walks the list of peers calling propagate_one()
  on each mount in this list making sure it receives the propagation
  event. The propagate_one() function thereby skips both new mounts and
  bind mounts to not propagate them "into themselves". Both are
  identified by checking whether the mount is already attached to any
  mount namespace in mnt->mnt_ns. The is what the IS_MNT_NEW() helper is
  responsible for.

  However, detached mounts have an anonymous mount namespace attached to
  them stashed in mnt->mnt_ns which means that IS_MNT_NEW() doesn't
  realize they need to be skipped causing the mount to propagate "into
  itself" breaking the mount table and causing a disconnect between the
  number of mounts recorded as being beneath or reachable from the
  target mountpoint and the number of mounts actually recorded/counted
  in ns->mounts ultimately causing an overflow which in turn prevents
  any new mounts via the ENOSPC issue.

  So teach propagation to handle detached mounts by making it aware of
  them. I've been tracking this issue down for the last couple of days
  and then verifying that the fix is correct by unmounting everything in
  my current mount table leaving only /proc and /sys mounted and running
  the reproducer above overnight verifying the number of mounts counted
  in ns->mounts. With this fix the counts are correct and the ENOSPC
  issue can't be reproduced.

  This change will only have an effect on mounts created with the new
  mount API since detached mounts cannot be created with the old mount
  API so regressions are extremely unlikely.

  Here's an illustration:

    #### mount():
    ubuntu@f1-vm:~$ sudo mount --bind /mnt/ /mnt/
    ubuntu@f1-vm:~$ findmnt  | grep -i mnt
    ├─/mnt                                /dev/sda2[/mnt] ext4       rw,relatime

    #### open_tree(OPEN_TREE_CLONE) + move_mount() with bug:
    ubuntu@f1-vm:~$ sudo ./mount-new /mnt/ /mnt/
    ubuntu@f1-vm:~$ findmnt  | grep -i mnt
    ├─/mnt                                /dev/sda2[/mnt] ext4       rw,relatime
    │ └─/mnt                              /dev/sda2[/mnt] ext4       rw,relatime

    #### open_tree(OPEN_TREE_CLONE) + move_mount() with the fix:
    ubuntu@f1-vm:~$ sudo ./mount-new /mnt /mnt
    ubuntu@f1-vm:~$ findmnt | grep -i mnt
    └─/mnt                                /dev/sda2[/mnt] ext4       rw,relatime"

* tag 'for-linus-2021-03-10' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner/linux:
  mount: fix mounting of detached mounts onto targets that reside on shared mounts

3 years agoMerge tag '5.12-rc2-smb3' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Wed, 10 Mar 2021 17:55:06 +0000 (09:55 -0800)]
Merge tag '5.12-rc2-smb3' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fixes from Steve French:
 "Six cifs/smb3 fixes, three of them for stable, including some
  important mulitchannel crediting fixes, and a fix for statfs error
  handling"

* tag '5.12-rc2-smb3' of git://git.samba.org/sfrench/cifs-2.6:
  cifs: do not send close in compound create+close requests
  cifs: return proper error code in statfs(2)
  cifs: change noisy error message to FYI
  cifs: print MIDs in decimal notation
  cifs: ask for more credit on async read/write code paths
  cifs: fix credit accounting for extra channel

3 years agoMerge git://git.kernel.org:/pub/scm/linux/kernel/git/netdev/net
Linus Torvalds [Wed, 10 Mar 2021 01:15:56 +0000 (17:15 -0800)]
Merge git://git./linux/kernel/git/netdev/net

Pull networking fixes from David Miller:

 1) Fix transmissions in dynamic SMPS mode in ath9k, from Felix Fietkau.

 2) TX skb error handling fix in mt76 driver, also from Felix.

 3) Fix BPF_FETCH atomic in x86 JIT, from Brendan Jackman.

 4) Avoid double free of percpu pointers when freeing a cloned bpf prog.
    From Cong Wang.

 5) Use correct printf format for dma_addr_t in ath11k, from Geert
    Uytterhoeven.

 6) Fix resolve_btfids build with older toolchains, from Kun-Chuan
    Hsieh.

 7) Don't report truncated frames to mac80211 in mt76 driver, from
    Lorenzop Bianconi.

 8) Fix watcdog timeout on suspend/resume of stmmac, from Joakim Zhang.

 9) mscc ocelot needs NET_DEVLINK selct in Kconfig, from Arnd Bergmann.

10) Fix sign comparison bug in TCP_ZEROCOPY_RECEIVE getsockopt(), from
    Arjun Roy.

11) Ignore routes with deleted nexthop object in mlxsw, from Ido
    Schimmel.

12) Need to undo tcp early demux lookup sometimes in nf_nat, from
    Florian Westphal.

13) Fix gro aggregation for udp encaps with zero csum, from Daniel
    Borkmann.

14) Make sure to always use imp*_ndo_send when necessaey, from Jason A.
    Donenfeld.

15) Fix TRSCER masks in sh_eth driver from Sergey Shtylyov.

16) prevent overly huge skb allocationsd in qrtr, from Pavel Skripkin.

17) Prevent rx ring copnsumer index loss of sync in enetc, from Vladimir
    Oltean.

18) Make sure textsearch copntrol block is large enough, from Wilem de
    Bruijn.

19) Revert MAC changes to r8152 leading to instability, from Hates Wang.

20) Advance iov in 9p even for empty reads, from Jissheng Zhang.

21) Double hook unregister in nftables, from PabloNeira Ayuso.

22) Fix memleak in ixgbe, fropm Dinghao Liu.

23) Avoid dups in pkt scheduler class dumps, from Maximilian Heyne.

24) Various mptcp fixes from Florian Westphal, Paolo Abeni, and Geliang
    Tang.

25) Fix DOI refcount bugs in cipso, from Paul Moore.

26) One too many irqsave in ibmvnic, from Junlin Yang.

27) Fix infinite loop with MPLS gso segmenting via virtio_net, from
    Balazs Nemeth.

* git://git.kernel.org:/pub/scm/linux/kernel/git/netdev/net: (164 commits)
  s390/qeth: fix notification for pending buffers during teardown
  s390/qeth: schedule TX NAPI on QAOB completion
  s390/qeth: improve completion of pending TX buffers
  s390/qeth: fix memory leak after failed TX Buffer allocation
  net: avoid infinite loop in mpls_gso_segment when mpls_hlen == 0
  net: check if protocol extracted by virtio_net_hdr_set_proto is correct
  net: dsa: xrs700x: check if partner is same as port in hsr join
  net: lapbether: Remove netif_start_queue / netif_stop_queue
  atm: idt77252: fix null-ptr-dereference
  atm: uPD98402: fix incorrect allocation
  atm: fix a typo in the struct description
  net: qrtr: fix error return code of qrtr_sendmsg()
  mptcp: fix length of ADD_ADDR with port sub-option
  net: bonding: fix error return code of bond_neigh_init()
  net: enetc: allow hardware timestamping on TX queues with tc-etf enabled
  net: enetc: set MAC RX FIFO to recommended value
  net: davicom: Use platform_get_irq_optional()
  net: davicom: Fix regulator not turned off on driver removal
  net: davicom: Fix regulator not turned off on failed probe
  net: dsa: fix switchdev objects on bridge master mistakenly being applied on ports
  ...

3 years agoMerge git://git.kernel.org:/pub/scm/linux/kernel/git/davem/sparc
Linus Torvalds [Wed, 10 Mar 2021 01:08:41 +0000 (17:08 -0800)]
Merge git://git./linux/kernel/git/davem/sparc

Pull sparc fixes from David Miller:
 "Fix opcode filtering for exceptions, and clean up defconfig"

* git://git.kernel.org:/pub/scm/linux/kernel/git/davem/sparc:
  sparc: sparc64_defconfig: remove duplicate CONFIGs
  sparc64: Fix opcode filtering in handling of no fault loads

3 years agosparc: sparc64_defconfig: remove duplicate CONFIGs
Corentin Labbe [Mon, 8 Mar 2021 09:51:26 +0000 (09:51 +0000)]
sparc: sparc64_defconfig: remove duplicate CONFIGs

After my patch there is CONFIG_ATA defined twice.
Remove the duplicate one.
Same problem for CONFIG_HAPPYMEAL, except I added as builtin for boot
test with NFS.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Fixes: a57cdeb369ef ("sparc: sparc64_defconfig: add necessary configs for qemu")
Signed-off-by: Corentin Labbe <clabbe@baylibre.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agosparc64: Fix opcode filtering in handling of no fault loads
Rob Gardner [Mon, 1 Mar 2021 05:48:16 +0000 (22:48 -0700)]
sparc64: Fix opcode filtering in handling of no fault loads

is_no_fault_exception() has two bugs which were discovered via random
opcode testing with stress-ng. Both are caused by improper filtering
of opcodes.

The first bug can be triggered by a floating point store with a no-fault
ASI, for instance "sta %f0, [%g0] #ASI_PNF", opcode C1A01040.

The code first tests op3[5] (0x1000000), which denotes a floating
point instruction, and then tests op3[2] (0x200000), which denotes a
store instruction. But these bits are not mutually exclusive, and the
above mentioned opcode has both bits set. The intent is to filter out
stores, so the test for stores must be done first in order to have
any effect.

The second bug can be triggered by a floating point load with one of
the invalid ASI values 0x8e or 0x8f, which pass this check in
is_no_fault_exception():
     if ((asi & 0xf2) == ASI_PNF)

An example instruction is "ldqa [%l7 + %o7] #ASI 0x8f, %f38",
opcode CF95D1EF. Asi values greater than 0x8b (ASI_SNFL) are fatal
in handle_ldf_stq(), and is_no_fault_exception() must not allow these
invalid asi values to make it that far.

In both of these cases, handle_ldf_stq() reacts by calling
sun4v_data_access_exception() or spitfire_data_access_exception(),
which call is_no_fault_exception() and results in an infinite
recursion.

Signed-off-by: Rob Gardner <rob.gardner@oracle.com>
Tested-by: Anatoly Pugachev <matorola@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 's390-qeth-fixes'
David S. Miller [Wed, 10 Mar 2021 00:14:54 +0000 (16:14 -0800)]
Merge branch 's390-qeth-fixes'

Julian Wiedmann says:

====================
s390/qeth: fixes 2021-03-09

please apply the following patch series to netdev's net tree.

This brings one fix for a memleak in an error path of the setup code.
Also several fixes for dealing with pending TX buffers - two for old
bugs in their completion handling, and one recent regression in a
teardown path.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agos390/qeth: fix notification for pending buffers during teardown
Julian Wiedmann [Tue, 9 Mar 2021 16:52:21 +0000 (17:52 +0100)]
s390/qeth: fix notification for pending buffers during teardown

The cited commit reworked the state machine for pending TX buffers.
In qeth_iqd_tx_complete() it turned PENDING into a transient state, and
uses NEED_QAOB for buffers that get parked while waiting for their QAOB
completion.

But it missed to adjust the check in qeth_tx_complete_buf(). So if
qeth_tx_complete_pending_bufs() is called during teardown to drain
the parked TX buffers, we no longer raise a notification for af_iucv.

Instead of updating the checked state, just move this code into
qeth_tx_complete_pending_bufs() itself. This also gets rid of the
special-case in the common TX completion path.

Fixes: 8908f36d20d8 ("s390/qeth: fix af_iucv notification race")
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agos390/qeth: schedule TX NAPI on QAOB completion
Julian Wiedmann [Tue, 9 Mar 2021 16:52:20 +0000 (17:52 +0100)]
s390/qeth: schedule TX NAPI on QAOB completion

When a QAOB notifies us that a pending TX buffer has been delivered, the
actual TX completion processing by qeth_tx_complete_pending_bufs()
is done within the context of a TX NAPI instance. We shouldn't rely on
this instance being scheduled by some other TX event, but just do it
ourselves.

qeth_qdio_handle_aob() is called from qeth_poll(), ie. our main NAPI
instance. To avoid touching the TX queue's NAPI instance
before/after it is (un-)registered, reorder the code in qeth_open()
and qeth_stop() accordingly.

Fixes: 0da9581ddb0f ("qeth: exploit asynchronous delivery of storage blocks")
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agos390/qeth: improve completion of pending TX buffers
Julian Wiedmann [Tue, 9 Mar 2021 16:52:19 +0000 (17:52 +0100)]
s390/qeth: improve completion of pending TX buffers

The current design attaches a pending TX buffer to a custom
single-linked list, which is anchored at the buffer's slot on the
TX ring. The buffer is then checked for final completion whenever
this slot is processed during a subsequent TX NAPI poll cycle.

But if there's insufficient traffic on the ring, we might never make
enough progress to get back to this ring slot and discover the pending
buffer's final TX completion. In particular if this missing TX
completion blocks the application from sending further traffic.

So convert the custom single-linked list code to a per-queue list_head,
and scan this list on every TX NAPI cycle.

Fixes: 0da9581ddb0f ("qeth: exploit asynchronous delivery of storage blocks")
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agos390/qeth: fix memory leak after failed TX Buffer allocation
Julian Wiedmann [Tue, 9 Mar 2021 16:52:18 +0000 (17:52 +0100)]
s390/qeth: fix memory leak after failed TX Buffer allocation

When qeth_alloc_qdio_queues() fails to allocate one of the buffers that
back an Output Queue, the 'out_freeoutqbufs' path will free all
previously allocated buffers for this queue. But it misses to free the
half-finished queue struct itself.

Move the buffer allocation into qeth_alloc_output_queue(), and deal with
such errors internally.

Fixes: 0da9581ddb0f ("qeth: exploit asynchronous delivery of storage blocks")
Signed-off-by: Julian Wiedmann <jwi@linux.ibm.com>
Reviewed-by: Alexandra Winter <wintera@linux.ibm.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge branch 'virtio_net-infinite-loop'
David S. Miller [Wed, 10 Mar 2021 00:12:20 +0000 (16:12 -0800)]
Merge branch 'virtio_net-infinite-loop'

Balazs Nemeth says:

====================
net: prevent infinite loop caused by incorrect proto from virtio_net_hdr_set_proto

These patches prevent an infinite loop for gso packets with a protocol
from virtio net hdr that doesn't match the protocol in the packet.
Note that packets coming from a device without
header_ops->parse_protocol being implemented will not be caught by
the check in virtio_net_hdr_to_skb, but the infinite loop will still
be prevented by the check in the gso layer.

Changes from v2 to v3:
  - Remove unused *eth.
  - Use MPLS_HLEN to also check if the MPLS header length is a multiple
    of four.
====================

Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: avoid infinite loop in mpls_gso_segment when mpls_hlen == 0
Balazs Nemeth [Tue, 9 Mar 2021 11:31:01 +0000 (12:31 +0100)]
net: avoid infinite loop in mpls_gso_segment when mpls_hlen == 0

A packet with skb_inner_network_header(skb) == skb_network_header(skb)
and ETH_P_MPLS_UC will prevent mpls_gso_segment from pulling any headers
from the packet. Subsequently, the call to skb_mac_gso_segment will
again call mpls_gso_segment with the same packet leading to an infinite
loop. In addition, ensure that the header length is a multiple of four,
which should hold irrespective of the number of stacked labels.

Signed-off-by: Balazs Nemeth <bnemeth@redhat.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Reviewed-by: David Ahern <dsahern@kernel.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: check if protocol extracted by virtio_net_hdr_set_proto is correct
Balazs Nemeth [Tue, 9 Mar 2021 11:31:00 +0000 (12:31 +0100)]
net: check if protocol extracted by virtio_net_hdr_set_proto is correct

For gso packets, virtio_net_hdr_set_proto sets the protocol (if it isn't
set) based on the type in the virtio net hdr, but the skb could contain
anything since it could come from packet_snd through a raw socket. If
there is a mismatch between what virtio_net_hdr_set_proto sets and
the actual protocol, then the skb could be handled incorrectly later
on.

An example where this poses an issue is with the subsequent call to
skb_flow_dissect_flow_keys_basic which relies on skb->protocol being set
correctly. A specially crafted packet could fool
skb_flow_dissect_flow_keys_basic preventing EINVAL to be returned.

Avoid blindly trusting the information provided by the virtio net header
by checking that the protocol in the packet actually matches the
protocol set by virtio_net_hdr_set_proto. Note that since the protocol
is only checked if skb->dev implements header_ops->parse_protocol,
packets from devices without the implementation are not checked at this
stage.

Fixes: 9274124f023b ("net: stricter validation of untrusted gso packets")
Signed-off-by: Balazs Nemeth <bnemeth@redhat.com>
Acked-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: xrs700x: check if partner is same as port in hsr join
George McCollister [Mon, 8 Mar 2021 23:38:22 +0000 (17:38 -0600)]
net: dsa: xrs700x: check if partner is same as port in hsr join

Don't assign dp to partner if it's the same port that xrs700x_hsr_join
was called with. The partner port is supposed to be the other port in
the HSR/PRP redundant pair not the same port. This fixes an issue
observed in testing where forwarding between redundant HSR ports on this
switch didn't work depending on the order the ports were added to the
hsr device.

Fixes: bd62e6f5e6a9 ("net: dsa: xrs700x: add HSR offloading support")
Signed-off-by: George McCollister <george.mccollister@gmail.com>
Reviewed-by: Vladimir Oltean <olteanv@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMerge tag 'gpio-fixes-for-v5.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Tue, 9 Mar 2021 20:02:18 +0000 (12:02 -0800)]
Merge tag 'gpio-fixes-for-v5.12-rc3' of git://git./linux/kernel/git/brgl/linux

Pull gpio fixes from Bartosz Golaszewski:
 "A bunch of fixes for the GPIO subsystem. We have two regressions in
  the core code spotted right after the merge window, a series of fixes
  for ACPI GPIO and a subsequent fix for a related regression in
  gpio-pca953x + a minor tweak in .gitignore and a rework of handling of
  the gpio-line-names to remedy a regression in stm32mp151.

  Summary:

   - fix two regressions in core GPIO subsystem code: one NULL-pointer
     dereference and one list corruption

   - read GPIO line names from fwnode instead of using the generic
     device properties to fix a regression on stm32mp151

   - fixes to ACPI GPIO and gpio-pca953x to handle a regression in IRQ
     handling on Intel Galileo

   - update .gitignore in GPIO selftests"

* tag 'gpio-fixes-for-v5.12-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/brgl/linux:
  gpiolib: Read "gpio-line-names" from a firmware node
  gpio: pca953x: Set IRQ type when handle Intel Galileo Gen 2
  gpiolib: acpi: Allow to find GpioInt() resource by name and index
  gpiolib: acpi: Add ACPI_GPIO_QUIRK_ABSOLUTE_NUMBER quirk
  gpiolib: acpi: Add missing IRQF_ONESHOT
  gpio: fix gpio-device list corruption
  gpio: fix NULL-deref-on-deregistration regression
  selftests: gpio: update .gitignore

3 years agoMerge tag 'mips-fixes_5.12_1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips...
Linus Torvalds [Tue, 9 Mar 2021 19:56:41 +0000 (11:56 -0800)]
Merge tag 'mips-fixes_5.12_1' of git://git./linux/kernel/git/mips/linux

Pull MIPS fixes from Thomas Bogendoerfer:

 - fixes for boot breakage because of misaligned FDTs

 - fix for overwritten exception handlers

 - enable MIPS optimized crypto for all MIPS CPUs to improve wireguard
   performance

* tag 'mips-fixes_5.12_1' of git://git.kernel.org/pub/scm/linux/kernel/git/mips/linux:
  MIPS: kernel: Reserve exception base early to prevent corruption
  MIPS: vmlinux.lds.S: align raw appended dtb to 8 bytes
  crypto: mips/poly1305 - enable for all MIPS processors
  MIPS: boot/compressed: Copy DTB to aligned address

3 years agonet: lapbether: Remove netif_start_queue / netif_stop_queue
Xie He [Sun, 7 Mar 2021 11:33:07 +0000 (03:33 -0800)]
net: lapbether: Remove netif_start_queue / netif_stop_queue

For the devices in this driver, the default qdisc is "noqueue",
because their "tx_queue_len" is 0.

In function "__dev_queue_xmit" in "net/core/dev.c", devices with the
"noqueue" qdisc are specially handled. Packets are transmitted without
being queued after a "dev->flags & IFF_UP" check. However, it's possible
that even if this check succeeds, "ops->ndo_stop" may still have already
been called. This is because in "__dev_close_many", "ops->ndo_stop" is
called before clearing the "IFF_UP" flag.

If we call "netif_stop_queue" in "ops->ndo_stop", then it's possible in
"__dev_queue_xmit", it sees the "IFF_UP" flag is present, and then it
checks "netif_xmit_stopped" and finds that the queue is already stopped.
In this case, it will complain that:
"Virtual device ... asks to queue packet!"

To prevent "__dev_queue_xmit" from generating this complaint, we should
not call "netif_stop_queue" in "ops->ndo_stop".

We also don't need to call "netif_start_queue" in "ops->ndo_open",
because after a netdev is allocated and registered, the
"__QUEUE_STATE_DRV_XOFF" flag is initially not set, so there is no need
to call "netif_start_queue" to clear it.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Signed-off-by: Xie He <xie.he.0141@gmail.com>
Acked-by: Martin Schiller <ms@dev.tdt.de>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMIPS: kernel: Reserve exception base early to prevent corruption
Thomas Bogendoerfer [Mon, 8 Mar 2021 09:24:47 +0000 (10:24 +0100)]
MIPS: kernel: Reserve exception base early to prevent corruption

BMIPS is one of the few platforms that do change the exception base.
After commit 2dcb39645441 ("memblock: do not start bottom-up allocations
with kernel_end") we started seeing BMIPS boards fail to boot with the
built-in FDT being corrupted.

Before the cited commit, early allocations would be in the [kernel_end,
RAM_END] range, but after commit they would be within [RAM_START +
PAGE_SIZE, RAM_END].

The custom exception base handler that is installed by
bmips_ebase_setup() done for BMIPS5000 CPUs ends-up trampling on the
memory region allocated by unflatten_and_copy_device_tree() thus
corrupting the FDT used by the kernel.

To fix this, we need to perform an early reservation of the custom
exception space. Additional we reserve the first 4k (1k for R3k) for
either normal exception vector space (legacy CPUs) or special vectors
like cache exceptions.

Huge thanks to Serge for analysing and proposing a solution to this
issue.

Fixes: 2dcb39645441 ("memblock: do not start bottom-up allocations with kernel_end")
Reported-by: Kamal Dasu <kdasu.kdev@gmail.com>
Debugged-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Acked-by: Mike Rapoport <rppt@linux.ibm.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Reviewed-by: Serge Semin <fancer.lancer@gmail.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
3 years agoMerge git://git.kernel.org:/pub/scm/linux/kernel/git/davem/sparc
Linus Torvalds [Tue, 9 Mar 2021 06:01:58 +0000 (22:01 -0800)]
Merge git://git./linux/kernel/git/davem/sparc

Pull sparc updates from David Miller:
 "Just some more random bits from Al, including a conversion over to
  generic extables"

* git://git.kernel.org:/pub/scm/linux/kernel/git/davem/sparc:
  sparc32: take ->thread.flags out
  sparc32: get rid of fake_swapper_regs
  sparc64: get rid of fake_swapper_regs
  sparc32: switch to generic extables
  sparc32: switch copy_user.S away from range exception table entries
  sparc32: get rid of range exception table entries in checksum_32.S
  sparc32: switch __bzero() away from range exception table entries
  sparc32: kill lookup_fault()
  sparc32: don't bother with lookup_fault() in __bzero()

3 years agocifs: do not send close in compound create+close requests
Paulo Alcantara [Mon, 8 Mar 2021 15:00:50 +0000 (12:00 -0300)]
cifs: do not send close in compound create+close requests

In case of interrupted syscalls, prevent sending CLOSE commands for
compound CREATE+CLOSE requests by introducing an
CIFS_CP_CREATE_CLOSE_OP flag to indicate lower layers that it should
not send a CLOSE command to the MIDs corresponding the compound
CREATE+CLOSE request.

A simple reproducer:

    #!/bin/bash

    mount //server/share /mnt -o username=foo,password=***
    tc qdisc add dev eth0 root netem delay 450ms
    stat -f /mnt &>/dev/null & pid=$!
    sleep 0.01
    kill $pid
    tc qdisc del dev eth0 root
    umount /mnt

Before patch:

    ...
    6 0.256893470 192.168.122.2 → 192.168.122.15 SMB2 402 Create Request File: ;GetInfo Request FS_INFO/FileFsFullSizeInformation;Close Request
    7 0.257144491 192.168.122.15 → 192.168.122.2 SMB2 498 Create Response File: ;GetInfo Response;Close Response
    9 0.260798209 192.168.122.2 → 192.168.122.15 SMB2 146 Close Request File:
   10 0.260841089 192.168.122.15 → 192.168.122.2 SMB2 130 Close Response, Error: STATUS_FILE_CLOSED

Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
3 years agocifs: return proper error code in statfs(2)
Paulo Alcantara [Mon, 8 Mar 2021 15:00:49 +0000 (12:00 -0300)]
cifs: return proper error code in statfs(2)

In cifs_statfs(), if server->ops->queryfs is not NULL, then we should
use its return value rather than always returning 0.  Instead, use rc
variable as it is properly set to 0 in case there is no
server->ops->queryfs.

Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
CC: <stable@vger.kernel.org>
Signed-off-by: Steve French <stfrench@microsoft.com>
3 years agocifs: change noisy error message to FYI
Paulo Alcantara [Mon, 8 Mar 2021 15:00:48 +0000 (12:00 -0300)]
cifs: change noisy error message to FYI

A customer has reported that their dmesg were being flooded by

  CIFS: VFS: \\server Cancelling wait for mid xxx cmd: a
  CIFS: VFS: \\server Cancelling wait for mid yyy cmd: b
  CIFS: VFS: \\server Cancelling wait for mid zzz cmd: c

because some processes that were performing statfs(2) on the share had
been interrupted due to their automount setup when certain users
logged in and out.

Change it to FYI as they should be mostly informative rather than
error messages.

Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
3 years agocifs: print MIDs in decimal notation
Paulo Alcantara [Mon, 8 Mar 2021 15:00:47 +0000 (12:00 -0300)]
cifs: print MIDs in decimal notation

The MIDs are mostly printed as decimal, so let's make it consistent.

Signed-off-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Reviewed-by: Aurelien Aptel <aaptel@suse.com>
Reviewed-by: Ronnie Sahlberg <lsahlber@redhat.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
3 years agoatm: idt77252: fix null-ptr-dereference
Tong Zhang [Mon, 8 Mar 2021 03:25:30 +0000 (22:25 -0500)]
atm: idt77252: fix null-ptr-dereference

this one is similar to the phy_data allocation fix in uPD98402, the
driver allocate the idt77105_priv and store to dev_data but later
dereference using dev->dev_data, which will cause null-ptr-dereference.

fix this issue by changing dev_data to phy_data so that PRIV(dev) can
work correctly.

Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoatm: uPD98402: fix incorrect allocation
Tong Zhang [Mon, 8 Mar 2021 03:25:29 +0000 (22:25 -0500)]
atm: uPD98402: fix incorrect allocation

dev->dev_data is set in zatm.c, calling zatm_start() will overwrite this
dev->dev_data in uPD98402_start() and a subsequent PRIV(dev)->lock
(i.e dev->phy_data->lock) will result in a null-ptr-dereference.

I believe this is a typo and what it actually want to do is to allocate
phy_data instead of dev_data.

Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoatm: fix a typo in the struct description
Tong Zhang [Mon, 8 Mar 2021 03:25:28 +0000 (22:25 -0500)]
atm: fix a typo in the struct description

phy_data means private PHY data not date

Signed-off-by: Tong Zhang <ztong0001@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: qrtr: fix error return code of qrtr_sendmsg()
Jia-Ju Bai [Mon, 8 Mar 2021 09:13:55 +0000 (01:13 -0800)]
net: qrtr: fix error return code of qrtr_sendmsg()

When sock_alloc_send_skb() returns NULL to skb, no error return code of
qrtr_sendmsg() is assigned.
To fix this bug, rc is assigned with -ENOMEM in this case.

Fixes: 194ccc88297a ("net: qrtr: Support decoding incoming v2 packets")
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agomptcp: fix length of ADD_ADDR with port sub-option
Davide Caratti [Mon, 8 Mar 2021 09:00:04 +0000 (10:00 +0100)]
mptcp: fix length of ADD_ADDR with port sub-option

in current Linux, MPTCP peers advertising endpoints with port numbers use
a sub-option length that wrongly accounts for the trailing TCP NOP. Also,
receivers will only process incoming ADD_ADDR with port having such wrong
sub-option length. Fix this, making ADD_ADDR compliant to RFC8684 §3.4.1.

this can be verified running tcpdump on the kselftests artifacts:

 unpatched kernel:
 [root@bottarga mptcp]# tcpdump -tnnr unpatched.pcap | grep add-addr
 reading from file unpatched.pcap, link-type LINUX_SLL (Linux cooked v1), snapshot length 65535
 IP 10.0.1.1.10000 > 10.0.1.2.53078: Flags [.], ack 101, win 509, options [nop,nop,TS val 214459678 ecr 521312851,mptcp add-addr v1 id 1 a00:201:2774:2d88:7436:85c3:17fd:101], length 0
 IP 10.0.1.2.53078 > 10.0.1.1.10000: Flags [.], ack 101, win 502, options [nop,nop,TS val 521312852 ecr 214459678,mptcp add-addr[bad opt]]

 patched kernel:
 [root@bottarga mptcp]# tcpdump -tnnr patched.pcap | grep add-addr
 reading from file patched.pcap, link-type LINUX_SLL (Linux cooked v1), snapshot length 65535
 IP 10.0.1.1.10000 > 10.0.1.2.38178: Flags [.], ack 101, win 509, options [nop,nop,TS val 3728873902 ecr 2732713192,mptcp add-addr v1 id 1 10.0.2.1:10100 hmac 0xbccdfcbe59292a1f,nop,nop], length 0
 IP 10.0.1.2.38178 > 10.0.1.1.10000: Flags [.], ack 101, win 502, options [nop,nop,TS val 2732713195 ecr 3728873902,mptcp add-addr v1-echo id 1 10.0.2.1:10100,nop,nop], length 0

Fixes: 22fb85ffaefb ("mptcp: add port support for ADD_ADDR suboption writing")
CC: stable@vger.kernel.org # 5.11+
Reviewed-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Acked-and-tested-by: Geliang Tang <geliangtang@gmail.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: bonding: fix error return code of bond_neigh_init()
Jia-Ju Bai [Mon, 8 Mar 2021 03:11:02 +0000 (19:11 -0800)]
net: bonding: fix error return code of bond_neigh_init()

When slave is NULL or slave_ops->ndo_neigh_setup is NULL, no error
return code of bond_neigh_init() is assigned.
To fix this bug, ret is assigned with -EINVAL in these cases.

Fixes: 9e99bfefdbce ("bonding: fix bond_neigh_init()")
Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: enetc: allow hardware timestamping on TX queues with tc-etf enabled
Vladimir Oltean [Sun, 7 Mar 2021 13:23:39 +0000 (15:23 +0200)]
net: enetc: allow hardware timestamping on TX queues with tc-etf enabled

The txtime is passed to the driver in skb->skb_mstamp_ns, which is
actually in a union with skb->tstamp (the place where software
timestamps are kept).

Since commit b50a5c70ffa4 ("net: allow simultaneous SW and HW transmit
timestamping"), __sock_recv_timestamp has some logic for making sure
that the two calls to skb_tstamp_tx:

skb_tx_timestamp(skb) # Software timestamp in the driver
-> skb_tstamp_tx(skb, NULL)

and

skb_tstamp_tx(skb, &shhwtstamps) # Hardware timestamp in the driver

will both do the right thing and in a race-free manner, meaning that
skb_tx_timestamp will deliver a cmsg with the software timestamp only,
and skb_tstamp_tx with a non-NULL hwtstamps argument will deliver a cmsg
with the hardware timestamp only.

Why are races even possible? Well, because although the software timestamp
skb->tstamp is private per skb, the hardware timestamp skb_hwtstamps(skb)
lives in skb_shinfo(skb), an area which is shared between skbs and their
clones. And skb_tstamp_tx works by cloning the packets when timestamping
them, therefore attempting to perform hardware timestamping on an skb's
clone will also change the hardware timestamp of the original skb. And
the original skb might have been yet again cloned for software
timestamping, at an earlier stage.

So the logic in __sock_recv_timestamp can't be as simple as saying
"does this skb have a hardware timestamp? if yes I'll send the hardware
timestamp to the socket, otherwise I'll send the software timestamp",
precisely because the hardware timestamp is shared.
Instead, it's quite the other way around: __sock_recv_timestamp says
"does this skb have a software timestamp? if yes, I'll send the software
timestamp, otherwise the hardware one". This works because the software
timestamp is not shared with clones.

But that means we have a problem when we attempt hardware timestamping
with skbs that don't have the skb->tstamp == 0. __sock_recv_timestamp
will say "oh, yeah, this must be some sort of odd clone" and will not
deliver the hardware timestamp to the socket. And this is exactly what
is happening when we have txtime enabled on the socket: as mentioned,
that is put in a union with skb->tstamp, so it is quite easy to mistake
it.

Do what other drivers do (intel igb/igc) and write zero to skb->tstamp
before taking the hardware timestamp. It's of no use to us now (we're
already on the TX confirmation path).

Fixes: 0d08c9ec7d6e ("enetc: add support time specific departure base on the qos etf")
Cc: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Acked-by: Vinicius Costa Gomes <vinicius.gomes@intel.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: enetc: set MAC RX FIFO to recommended value
Alex Marginean [Sun, 7 Mar 2021 13:23:38 +0000 (15:23 +0200)]
net: enetc: set MAC RX FIFO to recommended value

On LS1028A, the MAC RX FIFO defaults to the value 2, which is too high
and may lead to RX lock-up under traffic at a rate higher than 6 Gbps.
Set it to 1 instead, as recommended by the hardware design team and by
later versions of the ENETC block guide.

Signed-off-by: Alex Marginean <alexandru.marginean@nxp.com>
Reviewed-by: Claudiu Manoil <claudiu.manoil@nxp.com>
Reviewed-by: Jason Liu <jason.hui.liu@nxp.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: davicom: Use platform_get_irq_optional()
Paul Cercueil [Sun, 7 Mar 2021 13:17:49 +0000 (13:17 +0000)]
net: davicom: Use platform_get_irq_optional()

The second IRQ line really is optional, so use
platform_get_irq_optional() to obtain it.

Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: davicom: Fix regulator not turned off on driver removal
Paul Cercueil [Sun, 7 Mar 2021 13:17:48 +0000 (13:17 +0000)]
net: davicom: Fix regulator not turned off on driver removal

We must disable the regulator that was enabled in the probe function.

Fixes: 7994fe55a4a2 ("dm9000: Add regulator and reset support to dm9000")
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: davicom: Fix regulator not turned off on failed probe
Paul Cercueil [Sun, 7 Mar 2021 13:17:47 +0000 (13:17 +0000)]
net: davicom: Fix regulator not turned off on failed probe

When the probe fails or requests to be defered, we must disable the
regulator that was previously enabled.

Fixes: 7994fe55a4a2 ("dm9000: Add regulator and reset support to dm9000")
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: fix switchdev objects on bridge master mistakenly being applied on ports
Vladimir Oltean [Sun, 7 Mar 2021 10:21:56 +0000 (12:21 +0200)]
net: dsa: fix switchdev objects on bridge master mistakenly being applied on ports

Tobias reports that after the blamed patch, VLAN objects being added to
a bridge device are being added to all slave ports instead (swp2, swp3).

ip link add br0 type bridge vlan_filtering 1
ip link set swp2 master br0
ip link set swp3 master br0
bridge vlan add dev br0 vid 100 self

This is because the fix was too broad: we made dsa_port_offloads_netdev
say "yes, I offload the br0 bridge" for all slave ports, but we didn't
add the checks whether the switchdev object was in fact meant for the
physical port or for the bridge itself. So we are reacting on events in
a way in which we shouldn't.

The reason why the fix was too broad is because the question itself,
"does this DSA port offload this netdev", was too broad in the first
place. The solution is to disambiguate the question and separate it into
two different functions, one to be called for each switchdev attribute /
object that has an orig_dev == net_bridge (dsa_port_offloads_bridge),
and the other for orig_dev == net_bridge_port (*_offloads_bridge_port).

In the case of VLAN objects on the bridge interface, this solves the
problem because we know that VLAN objects are per bridge port and not
per bridge. And when orig_dev is equal to the net_bridge, we offload it
as a bridge, but not as a bridge port; that's how we are able to skip
reacting on those events. Note that this is compatible with future plans
to have explicit offloading of VLAN objects on the bridge interface as a
bridge port (in DSA, this signifies that we should add that VLAN towards
the CPU port).

Fixes: 99b8202b179f ("net: dsa: fix SWITCHDEV_ATTR_ID_BRIDGE_VLAN_FILTERING getting ignored")
Reported-by: Tobias Waldekranz <tobias@waldekranz.com>
Signed-off-by: Vladimir Oltean <vladimir.oltean@nxp.com>
Reviewed-by: Tobias Waldekranz <tobias@waldekranz.com>
Tested-by: Tobias Waldekranz <tobias@waldekranz.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: wan: fix error return code of uhdlc_init()
Jia-Ju Bai [Sun, 7 Mar 2021 09:12:56 +0000 (01:12 -0800)]
net: wan: fix error return code of uhdlc_init()

When priv->rx_skbuff or priv->tx_skbuff is NULL, no error return code of
uhdlc_init() is assigned.
To fix this bug, ret is assigned with -ENOMEM in these cases.

Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: hisilicon: hns: fix error return code of hns_nic_clear_all_rx_fetch()
Jia-Ju Bai [Sun, 7 Mar 2021 08:40:12 +0000 (00:40 -0800)]
net: hisilicon: hns: fix error return code of hns_nic_clear_all_rx_fetch()

When hns_assemble_skb() returns NULL to skb, no error return code of
hns_nic_clear_all_rx_fetch() is assigned.
To fix this bug, ret is assigned with -ENOMEM in this case.

Reported-by: TOTE Robot <oslab@tsinghua.edu.cn>
Signed-off-by: Jia-Ju Bai <baijiaju1990@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: usb: log errors to dmesg/syslog
Grant Grundler [Sat, 6 Mar 2021 22:12:32 +0000 (14:12 -0800)]
net: usb: log errors to dmesg/syslog

Errors in protocol should be logged when the driver aborts operations.
If the driver can carry on and "humor" the device, then emitting
the message as debug output level is fine.

Signed-off-by: Grant Grundler <grundler@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: usb: cdc_ncm: emit dev_err on error paths
Grant Grundler [Sat, 6 Mar 2021 22:12:31 +0000 (14:12 -0800)]
net: usb: cdc_ncm: emit dev_err on error paths

Several error paths in bind/probe code will only emit
output using dev_dbg. But if we are going to fail the
bind/probe, emit related output with "err" priority.

Signed-off-by: Grant Grundler <grundler@chromium.org>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: ethernet: chelsio: inline_crypto: Mundane typos fixed throughout the file chcr_k...
Bhaskar Chowdhury [Sat, 6 Mar 2021 21:20:28 +0000 (02:50 +0530)]
net: ethernet: chelsio: inline_crypto: Mundane typos fixed throughout the file chcr_ktls.c

Mundane typos fixes throughout the file.

s/establised/established/
s/availbale/available/
s/vaues/values/
s/Incase/In case/

Signed-off-by: Bhaskar Chowdhury <unixbhaskar@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agonet: dsa: bcm_sf2: simplify optional reset handling
Philipp Zabel [Fri, 5 Mar 2021 09:14:48 +0000 (10:14 +0100)]
net: dsa: bcm_sf2: simplify optional reset handling

As of commit bb475230b8e5 ("reset: make optional functions really
optional"), the reset framework API calls use NULL pointers to describe
optional, non-present reset controls.

This allows to unconditionally return errors from
devm_reset_control_get_optional_exclusive.

Signed-off-by: Philipp Zabel <p.zabel@pengutronix.de>
Acked-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
3 years agoMIPS: vmlinux.lds.S: align raw appended dtb to 8 bytes
Bjørn Mork [Sun, 7 Mar 2021 18:23:01 +0000 (19:23 +0100)]
MIPS: vmlinux.lds.S: align raw appended dtb to 8 bytes

The devicetree specification requires 8-byte alignment in
memory. This is now enforced by libfdt since commit 79edff12060f
("scripts/dtc: Update to upstream version v1.6.0-51-g183df9e9c2b9")
which included the upstream commit 5e735860c478 ("libfdt: Check for
8-byte address alignment in fdt_ro_probe_()").

This broke the MIPS raw appended DTBs which would be appended to
the image immediately following the initramfs section.  This ends
with a 32bit size, resulting in a 4-byte alignment of the DTB.

Fix by padding with zeroes to 8-bytes when MIPS_RAW_APPENDED_DTB
is defined.

Fixes: 79edff12060f ("scripts/dtc: Update to upstream version v1.6.0-51-g183df9e9c2b9")
Cc: Rob Herring <robh+dt@kernel.org>
Cc: Frank Rowand <frowand.list@gmail.com>
Signed-off-by: Bjørn Mork <bjorn@mork.no>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
3 years agomount: fix mounting of detached mounts onto targets that reside on shared mounts
Christian Brauner [Sat, 6 Mar 2021 10:10:10 +0000 (11:10 +0100)]
mount: fix mounting of detached mounts onto targets that reside on shared mounts

Creating a series of detached mounts, attaching them to the filesystem,
and unmounting them can be used to trigger an integer overflow in
ns->mounts causing the kernel to block any new mounts in count_mounts()
and returning ENOSPC because it falsely assumes that the maximum number
of mounts in the mount namespace has been reached, i.e. it thinks it
can't fit the new mounts into the mount namespace anymore.

Depending on the number of mounts in your system, this can be reproduced
on any kernel that supportes open_tree() and move_mount() by compiling
and running the following program:

  /* SPDX-License-Identifier: LGPL-2.1+ */

  #define _GNU_SOURCE
  #include <errno.h>
  #include <fcntl.h>
  #include <getopt.h>
  #include <limits.h>
  #include <stdbool.h>
  #include <stdio.h>
  #include <stdlib.h>
  #include <string.h>
  #include <sys/mount.h>
  #include <sys/stat.h>
  #include <sys/syscall.h>
  #include <sys/types.h>
  #include <unistd.h>

  /* open_tree() */
  #ifndef OPEN_TREE_CLONE
  #define OPEN_TREE_CLONE 1
  #endif

  #ifndef OPEN_TREE_CLOEXEC
  #define OPEN_TREE_CLOEXEC O_CLOEXEC
  #endif

  #ifndef __NR_open_tree
          #if defined __alpha__
                  #define __NR_open_tree 538
          #elif defined _MIPS_SIM
                  #if _MIPS_SIM == _MIPS_SIM_ABI32        /* o32 */
                          #define __NR_open_tree 4428
                  #endif
                  #if _MIPS_SIM == _MIPS_SIM_NABI32       /* n32 */
                          #define __NR_open_tree 6428
                  #endif
                  #if _MIPS_SIM == _MIPS_SIM_ABI64        /* n64 */
                          #define __NR_open_tree 5428
                  #endif
          #elif defined __ia64__
                  #define __NR_open_tree (428 + 1024)
          #else
                  #define __NR_open_tree 428
          #endif
  #endif

  /* move_mount() */
  #ifndef MOVE_MOUNT_F_EMPTY_PATH
  #define MOVE_MOUNT_F_EMPTY_PATH 0x00000004 /* Empty from path permitted */
  #endif

  #ifndef __NR_move_mount
          #if defined __alpha__
                  #define __NR_move_mount 539
          #elif defined _MIPS_SIM
                  #if _MIPS_SIM == _MIPS_SIM_ABI32        /* o32 */
                          #define __NR_move_mount 4429
                  #endif
                  #if _MIPS_SIM == _MIPS_SIM_NABI32       /* n32 */
                          #define __NR_move_mount 6429
                  #endif
                  #if _MIPS_SIM == _MIPS_SIM_ABI64        /* n64 */
                          #define __NR_move_mount 5429
                  #endif
          #elif defined __ia64__
                  #define __NR_move_mount (428 + 1024)
          #else
                  #define __NR_move_mount 429
          #endif
  #endif

  static inline int sys_open_tree(int dfd, const char *filename, unsigned int flags)
  {
          return syscall(__NR_open_tree, dfd, filename, flags);
  }

  static inline int sys_move_mount(int from_dfd, const char *from_pathname, int to_dfd,
                                   const char *to_pathname, unsigned int flags)
  {
          return syscall(__NR_move_mount, from_dfd, from_pathname, to_dfd, to_pathname, flags);
  }

  static bool is_shared_mountpoint(const char *path)
  {
          bool shared = false;
          FILE *f = NULL;
          char *line = NULL;
          int i;
          size_t len = 0;

          f = fopen("/proc/self/mountinfo", "re");
          if (!f)
                  return 0;

          while (getline(&line, &len, f) > 0) {
                  char *slider1, *slider2;

                  for (slider1 = line, i = 0; slider1 && i < 4; i++)
                          slider1 = strchr(slider1 + 1, ' ');

                  if (!slider1)
                          continue;

                  slider2 = strchr(slider1 + 1, ' ');
                  if (!slider2)
                          continue;

                  *slider2 = '\0';
                  if (strcmp(slider1 + 1, path) == 0) {
                          /* This is the path. Is it shared? */
                          slider1 = strchr(slider2 + 1, ' ');
                          if (slider1 && strstr(slider1, "shared:")) {
                                  shared = true;
                                  break;
                          }
                  }
          }
          fclose(f);
          free(line);

          return shared;
  }

  static void usage(void)
  {
          const char *text = "mount-new [--recursive] <base-dir>\n";
          fprintf(stderr, "%s", text);
          _exit(EXIT_SUCCESS);
  }

  #define exit_usage(format, ...)                              \
          ({                                                   \
                  fprintf(stderr, format "\n", ##__VA_ARGS__); \
                  usage();                                     \
          })

  #define exit_log(format, ...)                                \
          ({                                                   \
                  fprintf(stderr, format "\n", ##__VA_ARGS__); \
                  exit(EXIT_FAILURE);                          \
          })

  static const struct option longopts[] = {
          {"help",        no_argument,            0,      'a'},
          { NULL,         no_argument,            0,       0 },
  };

  int main(int argc, char *argv[])
  {
          int exit_code = EXIT_SUCCESS, index = 0;
          int dfd, fd_tree, new_argc, ret;
          char *base_dir;
          char *const *new_argv;
          char target[PATH_MAX];

          while ((ret = getopt_long_only(argc, argv, "", longopts, &index)) != -1) {
                  switch (ret) {
                  case 'a':
                          /* fallthrough */
                  default:
                          usage();
                  }
          }

          new_argv = &argv[optind];
          new_argc = argc - optind;
          if (new_argc < 1)
                  exit_usage("Missing base directory\n");
          base_dir = new_argv[0];

          if (*base_dir != '/')
                  exit_log("Please specify an absolute path");

          /* Ensure that target is a shared mountpoint. */
          if (!is_shared_mountpoint(base_dir))
                  exit_log("Please ensure that \"%s\" is a shared mountpoint", base_dir);

          dfd = open(base_dir, O_RDONLY | O_DIRECTORY | O_CLOEXEC);
          if (dfd < 0)
                  exit_log("%m - Failed to open base directory \"%s\"", base_dir);

          ret = mkdirat(dfd, "detached-move-mount", 0755);
          if (ret < 0)
                  exit_log("%m - Failed to create required temporary directories");

          ret = snprintf(target, sizeof(target), "%s/detached-move-mount", base_dir);
          if (ret < 0 || (size_t)ret >= sizeof(target))
                  exit_log("%m - Failed to assemble target path");

          /*
           * Having a mount table with 10000 mounts is already quite excessive
           * and shoult account even for weird test systems.
           */
          for (size_t i = 0; i < 10000; i++) {
                  fd_tree = sys_open_tree(dfd, "detached-move-mount",
                                          OPEN_TREE_CLONE |
                                          OPEN_TREE_CLOEXEC |
                                          AT_EMPTY_PATH);
                  if (fd_tree < 0) {
                          fprintf(stderr, "%m - Failed to open %d(detached-move-mount)", dfd);
                          exit_code = EXIT_FAILURE;
                          break;
                  }

                  ret = sys_move_mount(fd_tree, "", dfd, "detached-move-mount", MOVE_MOUNT_F_EMPTY_PATH);
                  if (ret < 0) {
                          if (errno == ENOSPC)
                                  fprintf(stderr, "%m - Buggy mount counting");
                          else
                                  fprintf(stderr, "%m - Failed to attach mount to %d(detached-move-mount)", dfd);
                          exit_code = EXIT_FAILURE;
                          break;
                  }
                  close(fd_tree);

                  ret = umount2(target, MNT_DETACH);
                  if (ret < 0) {
                          fprintf(stderr, "%m - Failed to unmount %s", target);
                          exit_code = EXIT_FAILURE;
                          break;
                  }
          }

          (void)unlinkat(dfd, "detached-move-mount", AT_REMOVEDIR);
          close(dfd);

          exit(exit_code);
  }

and wait for the kernel to refuse any new mounts by returning ENOSPC.
How many iterations are needed depends on the number of mounts in your
system. Assuming you have something like 50 mounts on a standard system
it should be almost instantaneous.

The root cause of this is that detached mounts aren't handled correctly
when source and target mount are identical and reside on a shared mount
causing a broken mount tree where the detached source itself is
propagated which propagation prevents for regular bind-mounts and new
mounts. This ultimately leads to a miscalculation of the number of
mounts in the mount namespace.

Detached mounts created via
open_tree(fd, path, OPEN_TREE_CLONE)
are essentially like an unattached new mount, or an unattached
bind-mount. They can then later on be attached to the filesystem via
move_mount() which calls into attach_recursive_mount(). Part of
attaching it to the filesystem is making sure that mounts get correctly
propagated in case the destination mountpoint is MS_SHARED, i.e. is a
shared mountpoint. This is done by calling into propagate_mnt() which
walks the list of peers calling propagate_one() on each mount in this
list making sure it receives the propagation event.
The propagate_one() functions thereby skips both new mounts and bind
mounts to not propagate them "into themselves". Both are identified by
checking whether the mount is already attached to any mount namespace in
mnt->mnt_ns. The is what the IS_MNT_NEW() helper is responsible for.

However, detached mounts have an anonymous mount namespace attached to
them stashed in mnt->mnt_ns which means that IS_MNT_NEW() doesn't
realize they need to be skipped causing the mount to propagate "into
itself" breaking the mount table and causing a disconnect between the
number of mounts recorded as being beneath or reachable from the target
mountpoint and the number of mounts actually recorded/counted in
ns->mounts ultimately causing an overflow which in turn prevents any new
mounts via the ENOSPC issue.

So teach propagation to handle detached mounts by making it aware of
them. I've been tracking this issue down for the last couple of days and
then verifying that the fix is correct by
unmounting everything in my current mount table leaving only /proc and
/sys mounted and running the reproducer above overnight verifying the
number of mounts counted in ns->mounts. With this fix the counts are
correct and the ENOSPC issue can't be reproduced.

This change will only have an effect on mounts created with the new
mount API since detached mounts cannot be created with the old mount API
so regressions are extremely unlikely.

Link: https://lore.kernel.org/r/20210306101010.243666-1-christian.brauner@ubuntu.com
Fixes: 2db154b3ea8e ("vfs: syscall: Add move_mount(2) to move mounts around")
Cc: David Howells <dhowells@redhat.com>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: linux-fsdevel@vger.kernel.org
Cc: <stable@vger.kernel.org>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Christian Brauner <christian.brauner@ubuntu.com>
3 years agogpiolib: Read "gpio-line-names" from a firmware node
Andy Shevchenko [Fri, 5 Mar 2021 12:02:40 +0000 (14:02 +0200)]
gpiolib: Read "gpio-line-names" from a firmware node

On STM32MP1, the GPIO banks are subnodes of pin-controller@50002000,
see arch/arm/boot/dts/stm32mp151.dtsi. The driver for
pin-controller@50002000 is in drivers/pinctrl/stm32/pinctrl-stm32.c
and iterates over all of its DT subnodes when registering each GPIO
bank gpiochip. Each gpiochip has:

  - gpio_chip.parent = dev,
    where dev is the device node of the pin controller
  - gpio_chip.of_node = np,
    which is the OF node of the GPIO bank

Therefore, dev_fwnode(chip->parent) != of_fwnode_handle(chip.of_node),
i.e. pin-controller@50002000 != pin-controller@50002000/gpio@5000*000.

The original code behaved correctly, as it extracted the "gpio-line-names"
from of_fwnode_handle(chip.of_node) = pin-controller@50002000/gpio@5000*000.

To achieve the same behaviour, read property from the firmware node.

Fixes: 7cba1a4d5e162 ("gpiolib: generalize devprop_gpiochip_set_names() for device properties")
Reported-by: Marek Vasut <marex@denx.de>
Reported-by: Roman Guskov <rguskov@dh-electronics.com>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Tested-by: Marek Vasut <marex@denx.de>
Reviewed-by: Marek Vasut <marex@denx.de>
Signed-off-by: Bartosz Golaszewski <bgolaszewski@baylibre.com>
3 years agogpio: pca953x: Set IRQ type when handle Intel Galileo Gen 2
Andy Shevchenko [Thu, 25 Feb 2021 16:33:20 +0000 (18:33 +0200)]
gpio: pca953x: Set IRQ type when handle Intel Galileo Gen 2

The commit 0ea683931adb ("gpio: dwapb: Convert driver to using the
GPIO-lib-based IRQ-chip") indeliberately made a regression on how
IRQ line from GPIO I²C expander is handled. I.e. it reveals that
the quirk for Intel Galileo Gen 2 misses the part of setting IRQ type
which previously was predefined by gpio-dwapb driver. Now, we have to
reorganize the approach to call necessary parts, which can be done via
ACPI_GPIO_QUIRK_ABSOLUTE_NUMBER quirk.

Without this fix and with above mentioned change the kernel hangs
on the first IRQ event with:

    gpio gpiochip3: Persistence not supported for GPIO 1
    irq 32, desc: 62f8fb50, depth: 0, count: 0, unhandled: 0
    ->handle_irq():  41c7b0ab, handle_bad_irq+0x0/0x40
    ->irq_data.chip(): e03f1e72, 0xc2539218
    ->action(): 0ecc7e6f
    ->action->handler(): 8a3db21e, irq_default_primary_handler+0x0/0x10
       IRQ_NOPROBE set
    unexpected IRQ trap at vector 20

Fixes: ba8c90c61847 ("gpio: pca953x: Override IRQ for one of the expanders on Galileo Gen 2")
Depends-on: 0ea683931adb ("gpio: dwapb: Convert driver to using the GPIO-lib-based IRQ-chip")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Reviewed-by: Linus Walleij <linus.walleij@linaro.org>
3 years agogpiolib: acpi: Allow to find GpioInt() resource by name and index
Andy Shevchenko [Thu, 25 Feb 2021 16:33:19 +0000 (18:33 +0200)]
gpiolib: acpi: Allow to find GpioInt() resource by name and index

Currently only search by index is supported. However, in some cases
we might need to pass the quirks to the acpi_dev_gpio_irq_get().

For this, split out acpi_dev_gpio_irq_get_by() and replace
acpi_dev_gpio_irq_get() by calling above with NULL for name parameter.

Fixes: ba8c90c61847 ("gpio: pca953x: Override IRQ for one of the expanders on Galileo Gen 2")
Depends-on: 0ea683931adb ("gpio: dwapb: Convert driver to using the GPIO-lib-based IRQ-chip")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>
3 years agogpiolib: acpi: Add ACPI_GPIO_QUIRK_ABSOLUTE_NUMBER quirk
Andy Shevchenko [Thu, 25 Feb 2021 16:33:18 +0000 (18:33 +0200)]
gpiolib: acpi: Add ACPI_GPIO_QUIRK_ABSOLUTE_NUMBER quirk

On some systems the ACPI tables has wrong pin number and instead of
having a relative one it provides an absolute one in the global GPIO
number space.

Add ACPI_GPIO_QUIRK_ABSOLUTE_NUMBER quirk to cope with such cases.

Fixes: ba8c90c61847 ("gpio: pca953x: Override IRQ for one of the expanders on Galileo Gen 2")
Depends-on: 0ea683931adb ("gpio: dwapb: Convert driver to using the GPIO-lib-based IRQ-chip")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Acked-by: Linus Walleij <linus.walleij@linaro.org>