mm/damon/core.c: avoid unintentional filtering out of schemes

The function '__damos_filter_out()' causes DAMON to always filter out
schemes whose filter type is anon or memcg if its matching value is set
to false.

This commit addresses the issue by ensuring that '__damos_filter_out()'
no longer applies to filters whose type is 'anon' or 'memcg'.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: ab9bda001b681 ("mm/damon/core: introduce address range type damos filter")
Signed-off-by: Hyeongtak Ji <[email protected]>
Reviewed-by: SeongJae Park <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

mm: kmem: drop __GFP_NOFAIL when allocating objcg vectors

Objcg vectors attached to slab pages to store slab object ownership
information are allocated using gfp flags for the original slab
allocation.  Depending on slab page order and the size of slab objects,
objcg vector can take several pages.

If the original allocation was done with the __GFP_NOFAIL flag, it
triggered a warning in the page allocation code.  Indeed, order > 1 pages
should not been allocated with the __GFP_NOFAIL flag.

Fix this by simply dropping the __GFP_NOFAIL flag when allocating the
objcg vector.  It effectively allows to skip the accounting of a single
slab object under a heavy memory pressure.

An alternative would be to implement the mechanism to fallback to order-0
allocations for accounting metadata, which is also not perfect because it
will increase performance penalty and memory footprint of the kernel
memory accounting under memory pressure.

Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Roman Gushchin <[email protected]>
Reported-by: Christoph Lameter <[email protected]>
Closes: https://lkml.kernel.org/r/[email protected]
Acked-by: Shakeel Butt <[email protected]>
Cc: Matthew Wilcox <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

mm/damon/sysfs-schemes: handle tried region directory allocation failure

DAMON sysfs interface's before_damos_apply callback
(damon_sysfs_before_damos_apply()), which creates the DAMOS tried regions
for each DAMOS action applied region, is not handling the allocation
failure for the sysfs directory data.  As a result, NULL pointer
derefeence is possible.  Fix it by handling the case.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: f1d13cacabe1 ("mm/damon/sysfs: implement DAMOS tried regions update command")
Signed-off-by: SeongJae Park <[email protected]>
Cc: <[email protected]> [6.2+]
Signed-off-by: Andrew Morton <[email protected]>

mm/damon/sysfs-schemes: handle tried regions sysfs directory allocation failure

DAMOS tried regions sysfs directory allocation function
(damon_sysfs_scheme_regions_alloc()) is not handling the memory allocation
failure.  In the case, the code will dereference NULL pointer.  Handle the
failure to avoid such invalid access.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 9277d0367ba1 ("mm/damon/sysfs-schemes: implement scheme region directory")
Signed-off-by: SeongJae Park <[email protected]>
Cc: <[email protected]> [6.2+]
Signed-off-by: Andrew Morton <[email protected]>

mm/damon/sysfs: check error from damon_sysfs_update_target()

Patch series "mm/damon/sysfs: fix unhandled return values".

Some of DAMON sysfs interface code is not handling return values from some
functions.  As a result, confusing user input handling or NULL-dereference
is possible.  Check those properly.

This patch (of 3):

damon_sysfs_update_target() returns error code for failures, but its
caller, damon_sysfs_set_targets() is ignoring that.  The update function
seems making no critical change in case of such failures, but the behavior
will look like DAMON sysfs is silently ignoring or only partially
accepting the user input.  Fix it.

Link: https://lkml.kernel.org/r/[email protected]
Link: https://lkml.kernel.org/r/[email protected]
Fixes: 19467a950b49 ("mm/damon/sysfs: remove requested targets when online-commit inputs")
Signed-off-by: SeongJae Park <[email protected]>
Cc: <[email protected]> [5.19+]
Signed-off-by: Andrew Morton <[email protected]>

mm: fix for negative counter: nr_file_hugepages

While qualifiying the 6.4 release, the following warning was detected in
messages:

vmstat_refresh: nr_file_hugepages -15664

The warning is caused by the incorrect updating of the NR_FILE_THPS
counter in the function split_huge_page_to_list.  The if case is checking
for folio_test_swapbacked, but the else case is missing the check for
folio_test_pmd_mappable.  The other functions that manipulate the counter
like __filemap_add_folio and filemap_unaccount_folio have the
corresponding check.

I have a test case, which reproduces the problem. It can be found here:
  https://github.com/sroeschus/testcase/blob/main/vmstat_refresh/madv.c

The test case reproduces on an XFS filesystem. Running the same test
case on a BTRFS filesystem does not reproduce the problem.

AFAIK version 6.1 until 6.6 are affected by this problem.

[[email protected]: whitespace fix]
[[email protected]: test for folio_test_pmd_mappable()]
Link: https://lkml.kernel.org/r/[email protected]
Link: https://lkml.kernel.org/r/[email protected]
Signed-off-by: Stefan Roesch <[email protected]>
Co-debugged-by: Johannes Weiner <[email protected]>
Acked-by: Johannes Weiner <[email protected]>
Reviewed-by: Matthew Wilcox (Oracle) <[email protected]>
Reviewed-by: David Hildenbrand <[email protected]>
Reviewed-by: Yang Shi <[email protected]>
Cc: Rik van Riel <[email protected]>
Cc: <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

selftests/mm: add hugetlb_fault_after_madv to .gitignore

commit 116d57303a05 ("selftests/mm: add a new test for madv and hugetlb")
added a new test case, but, it didn't add the binary name in
tools/testing/selftests/mm/.gitignore.

Add hugetlb_fault_after_madv to tools/testing/selftests/mm/.gitignore.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 116d57303a05 ("selftests/mm: add a new test for madv and hugetlb")
Signed-off-by: Breno Leitao <[email protected]>
Reported-by: Ryan Roberts <[email protected]>
Closes: https://lore.kernel.org/all/[email protected]/
Signed-off-by: Andrew Morton <[email protected]>

selftests/mm: restore number of hugepages

The test mm `hugetlb_fault_after_madv` selftest needs one and only one
huge page to run, thus it sets `/proc/sys/vm/nr_hugepages` to 1.

The problem is that further tests require the previous number of hugepages
allocated in order to succeed.

Save the number of huge pages before changing it, and restore it once the
test finishes, so, further tests could run successfully.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 116d57303a05 ("selftests/mm: add a new test for madv and hugetlb")
Signed-off-by: Breno Leitao <[email protected]>
Reported-by: Ryan Roberts <[email protected]>
Closes: https://lore.kernel.org/all/[email protected]/
Signed-off-by: Andrew Morton <[email protected]>

selftests: mm: fix some build warnings

Fix build warnings:
pagemap_ioctl.c:1154:38: warning: format `%s' expects a matching `char *' argument [-Wformat=]
pagemap_ioctl.c:1162:51: warning: format `%ld' expects argument of type `long int', but argument 2 has type `int' [-Wformat=]
pagemap_ioctl.c:1192:51: warning: format `%ld' expects argument of type `long int', but argument 2 has type `int' [-Wformat=]
pagemap_ioctl.c:1600:51: warning: format `%ld' expects argument of type `long int', but argument 2 has type `int' [-Wformat=]
pagemap_ioctl.c:1628:51: warning: format `%ld' expects argument of type `long int', but argument 2 has type `int' [-Wformat=]

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 46fd75d4a3c9 ("selftests: mm: add pagemap ioctl tests")
Signed-off-by: Muhammad Usama Anjum <[email protected]>
Cc: Shuah Khan <[email protected]>
Cc: Ryan Roberts <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

selftests: mm: skip whole test instead of failure

Some architectures don't support userfaultfd.  Skip running the whole test
on them instead of registering the failure.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 46fd75d4a3c9 ("selftests: mm: add pagemap ioctl tests")
Reported-by: Ryan Roberts <[email protected]>
Closes: https://lore.kernel.org/all/[email protected]
Signed-off-by: Muhammad Usama Anjum <[email protected]>
Cc: Shuah Khan <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

mm/damon/sysfs: eliminate potential uninitialized variable warning

The "err" variable is not initialized if damon_target_has_pid(ctx) is
false and sys_target->regions->nr is zero.

Link: https://lkml.kernel.org/r/[email protected]
Fixes: 0bcd216c4741 ("mm/damon/sysfs: update monitoring target regions for online input commit")
Signed-off-by: Dan Carpenter <[email protected]>
Reviewed-by: SeongJae Park <[email protected]>
Signed-off-by: Andrew Morton <[email protected]>

ALSA: hda/realtek: Add quirks for HP Laptops

These HP laptops use Realtek HDA codec combined with 2 or 4 CS35L41
Amplifiers using SPI with Internal Boost.

Signed-off-by: Stefan Binding <[email protected]>
Cc: <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Takashi Iwai <[email protected]>

ALSA: hda/realtek: Add quirks for ASUS 2024 Zenbooks

These ASUS Zenbook laptops use Realtek HDA codec combined with
2xCS35L41 Amplifiers using SPI or I2C with External Boost or
Internal Boost.

Signed-off-by: Stefan Binding <[email protected]>
Cc: <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Takashi Iwai <[email protected]>

nouveau: don't fail driver load if no display hw present.

If we get back ENODEV don't fail load. There are nvidia devices
that don't have display blocks and the driver should work on those.

Fixes: 15740541e8f0 ("drm/nouveau/devinit/tu102-: prepare for GSP-RM")
Link: https://gitlab.freedesktop.org/drm/nouveau/-/issues/270
Signed-off-by: Dave Airlie <[email protected]>
Signed-off-by: Danilo Krummrich <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]

drm: panel-orientation-quirks: Add quirk for Lenovo Legion Go

The Legion Go has a 2560x1600 portrait screen, with the native "up" facing
the right controller (90° CW from the rest of the device).

Signed-off-by: Brenton Simpson <[email protected]>
Signed-off-by: Hans de Goede <[email protected]>
Link: https://lore.kernel.org/r/[email protected]

bpf: Do not allocate percpu memory at init stage

Kirill Shutemov reported significant percpu memory consumption increase after
booting in 288-cpu VM ([1]) due to commit 41a5db8d8161 ("bpf: Add support for
non-fix-size percpu mem allocation"). The percpu memory consumption is
increased from 111MB to 969MB. The number is from /proc/meminfo.

I tried to reproduce the issue with my local VM which at most supports upto
255 cpus. With 252 cpus, without the above commit, the percpu memory
consumption immediately after boot is 57MB while with the above commit the
percpu memory consumption is 231MB.

This is not good since so far percpu memory from bpf memory allocator is not
widely used yet. Let us change pre-allocation in init stage to on-demand
allocation when verifier detects there is a need of percpu memory for bpf
program. With this change, percpu memory consumption after boot can be reduced
signicantly.

  [1] https://lore.kernel.org/lkml/20231109154934[email protected]/

Fixes: 41a5db8d8161 ("bpf: Add support for non-fix-size percpu mem allocation")
Reported-and-tested-by: Kirill A. Shutemov <[email protected]>
Signed-off-by: Yonghong Song <[email protected]>
Acked-by: Hou Tao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Alexei Starovoitov <[email protected]>

io_uring/fdinfo: remove need for sqpoll lock for thread/pid retrieval

A previous commit added a trylock for getting the SQPOLL thread info via
fdinfo, but this introduced a regression where we often fail to get it if
the thread is busy. For that case, we end up not printing the current CPU
and PID info.

Rather than rely on this lock, just print the pid we already stored in
the io_sq_data struct, and ensure we update the current CPU every time
we've slept or potentially rescheduled. The latter won't potentially be
100% accurate, but that wasn't the case before either as the task can
get migrated at any time unless it has been pinned at creation time.

We retain keeping the io_sq_data dereference inside the ctx->uring_lock,
as it has always been, as destruction of the thread and data happen below
that. We could make this RCU safe, but there's little point in doing that.

With this, we always print the last valid information we had, rather than
have spurious outputs with missing information.

Fixes: 7644b1a1c9a7 ("io_uring/fdinfo: lock SQ thread while retrieving thread cpu/pid")
Signed-off-by: Jens Axboe <[email protected]>

ALSA: hda: i915: Alays handle -EPROBE_DEFER

It turns out that even if the comment says that the driver can load
fine, it's not really the case and no codecs are detected.
Specifically for -EPROBE_DEFER, always fail the probe.

This fixes a regression when HDA-intel is loaded before i915.

Reported-by: Ville Syrjälä <[email protected]>
Closes: https://lore.kernel.org/r/[email protected]
Signed-off-by: Maarten Lankhorst <[email protected]>
Tested-by: Kai Vehmanen <[email protected]>
Fixes: e6d0c13e9f46 ("ALSA: hda: i915: Remove extra argument from snd_hdac_i915_init")
Link: https://gitlab.freedesktop.org/drm/intel/-/issues/9671
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Takashi Iwai <[email protected]>

net: Fix undefined behavior in netdev name allocation

Cited commit removed the strscpy() call and kept the snprintf() only.

It is common to use 'dev->name' as the format string before a netdev is
registered, this results in 'res' and 'name' pointers being equal.
According to POSIX, if copying takes place between objects that overlap
as a result of a call to sprintf() or snprintf(), the results are
undefined.

Add back the strscpy() and use 'buf' as an intermediate buffer.

Fixes: 7ad17b04dc7b ("net: trust the bitmap in __dev_alloc_name()")
Cc: Jakub Kicinski <[email protected]>
Reviewed-by: Vlad Buslov <[email protected]>
Signed-off-by: Gal Pressman <[email protected]>
Reviewed-by: Jakub Kicinski <[email protected]>
Reviewed-by: Simon Horman <[email protected]>
Reviewed-by: Jiri Pirko <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

accel/ivpu: avoid build failure with CONFIG_PM=n

The usage count of struct dev_pm_info is an implementation detail that
is only available if CONFIG_PM is enabled, so printing it in a debug message
causes a build failure in configurations without PM:

In file included from include/linux/device.h:15,
                 from include/linux/pci.h:37,
                 from drivers/accel/ivpu/ivpu_pm.c:8:
drivers/accel/ivpu/ivpu_pm.c: In function 'ivpu_rpm_get_if_active':
drivers/accel/ivpu/ivpu_pm.c:254:51: error: 'struct dev_pm_info' has no member named 'usage_count'
  254 |                  atomic_read(&vdev->drm.dev->power.usage_count));
      |                                                   ^
include/linux/dev_printk.h:129:48: note: in definition of macro 'dev_printk'
  129 |                 _dev_printk(level, dev, fmt, ##__VA_ARGS__);            \
      |                                                ^~~~~~~~~~~
drivers/accel/ivpu/ivpu_drv.h:75:17: note: in expansion of macro 'dev_dbg'
   75 |                 dev_dbg((vdev)->drm.dev, "[%s] " fmt, #type, ##args);          \
      |                 ^~~~~~~
drivers/accel/ivpu/ivpu_pm.c:253:9: note: in expansion of macro 'ivpu_dbg'
  253 |         ivpu_dbg(vdev, RPM, "rpm_get_if_active count %d\n",
      |         ^~~~~~~~

The print message does not seem essential, so the easiest workaround is
to just remove it.

Fixes: c39dc15191c4 ("accel/ivpu: Read clock rate only if device is up")
Signed-off-by: Arnd Bergmann <[email protected]>
Reviewed-by: Stanislaw Gruszka <[email protected]>
Signed-off-by: Stanislaw Gruszka <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]
(cherry picked from commit 1470acbef122c7e2e588f6346ce459c26d0568a2)
Signed-off-by: Jacek Lawrynowicz <[email protected]>

dt-bindings: net: ethernet-controller: Fix formatting error

When moving the *-internal-delay-ps properties to only apply for RGMII
interface modes there where a typo in the text formatting.

Signed-off-by: Niklas Söderlund <[email protected]>
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

Revert ncsi: Propagate carrier gain/loss events to the NCSI controller

This reverts commit 3780bb29311eccb7a1c9641032a112eed237f7e3.

The cited commit introduced unwanted behavior.

The intent for the commit was to be able to detect carrier loss/gain
for just the NIC connected to the BMC. The unwanted effect is a
carrier loss for auxiliary paths also causes the BMC to lose
carrier. The BMC never regains carrier despite the secondary NIC
regaining a link.

This change, when merged, needs to be backported to stable kernels.
5.4-stable, 5.10-stable, 5.15-stable, 6.1-stable, 6.5-stable

Fixes: 3780bb29311e ("ncsi: Propagate carrier gain/loss events to the NCSI controller")
CC: [email protected]
Signed-off-by: Johnathan Mantey <[email protected]>
Reviewed-by: Simon Horman <[email protected]>
Signed-off-by: David S. Miller <[email protected]>

xen/events: remove some info_for_irq() calls in pirq handling

Instead of the IRQ number user the struct irq_info pointer as parameter
in the internal pirq related functions. This allows to drop some calls
of info_for_irq().

Signed-off-by: Juergen Gross <[email protected]>
Reviewed-by: Oleksandr Tyshchenko <[email protected]>
Signed-off-by: Juergen Gross <[email protected]>

xen/events: modify internal [un]bind interfaces

Modify the internal bind- and unbind-interfaces to take a struct
irq_info parameter. When allocating a new IRQ pass the pointer from
the allocating function further up.

This will reduce the number of info_for_irq() calls and make the code
more efficient.

Signed-off-by: Juergen Gross <[email protected]>
Reviewed-by: Oleksandr Tyshchenko <[email protected]>
Signed-off-by: Juergen Gross <[email protected]>

xen/events: drop xen_allocate_irqs_dynamic()

Instead of having a common function for allocating a single IRQ or a
consecutive number of IRQs, split up the functionality into the callers
of xen_allocate_irqs_dynamic().

This allows to handle any allocation error in xen_irq_init() gracefully
instead of panicing the system. Let xen_irq_init() return the irq_info
pointer or NULL in case of an allocation error.

Additionally set the IRQ into irq_info already at allocation time, as
otherwise the IRQ would be '0' (which is a valid IRQ number) until
being set.

Signed-off-by: Juergen Gross <[email protected]>
Reviewed-by: Oleksandr Tyshchenko <[email protected]>
Signed-off-by: Juergen Gross <[email protected]>

Merge tag 'hardening-v6.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux

Pull hardening fixes from Kees Cook:

 - stackleak: add declarations for global functions (Arnd Bergmann)

 - gcc-plugins: randstruct: Only warn about true flexible arrays (Kees
   Cook)

 - gcc-plugins: latent_entropy: Fix description typo (Konstantin Runov)

* tag 'hardening-v6.7-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/kees/linux:
  gcc-plugins: latent_entropy: Fix typo (args -> argc) in plugin description
  gcc-plugins: randstruct: Only warn about true flexible arrays
  stackleak: add declarations for global functions

bcachefs: six locks: Fix lost wakeup

In percpu reader mode, trylock() for read had a lost wakeup: on failure
to get the lock, we may have caused a writer to fail to get the lock,
because we temporarily elevated the reader count.

We need to check for waiters after decrementing the read count - not
before.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix no_data_io mode checksum check

In no_data_io mode, we expect data checksums to be wrong - don't want to
spew the log with them.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix bch2_check_nlinks() for snapshots

When searching the link table for the matching inode, we were searching
for a specific - incorrect - snapshot ID as well, causing us to fail to
find the inode.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Don't decrease BTREE_ITER_MAX when LOCKDEP=y

Running with fewer max btree paths doesn't work anymore when replication
is enabled - as we've added e.g. the freespace and bucket gens btrees,
we naturally end up needing more btree paths.

This is an issue with lockdep, we end up taking more locks than lockdep
will track (the MAX_LOCKD_DEPTH constant). But bcachefs as merged does
not yet support lockdep anyways, so we can leave that for later.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Disable debug log statements

The journal read path had some informational log statements preperatory
for ZNS support - they're not of interest to users, so we can turn them
off.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix missing transaction commit

In may_delete_deleted_inode(), there's a corner case when a snapshot was
taken while we had an unlinked inode: we don't want to delete the inode
in the internal (shared) snapshot node, since it might have been
reattached in a descendent snapshot.

Instead we propagate the key to any snapshot leaves it doesn't exist in,
so that it can be deleted there if necessary, and then clear the
unlinked flag in the internal node.

But we forgot to commit after clearing the unlinked flag, causing us to
go into an infinite loop.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix error path in bch2_mount()

This fixes a bug discovered by generic/388 where sb->s_fs_info was NULL
while the superblock was still active - the error path was entirely
fubar, and was trying to do something unclear and unecessary.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix potential sleeping during mount

During mount, bcachefs mount option processing may sleep while allocating a string buffer.

Fix this by reference counting in order to take the atomic path.

Signed-off-by: Daniel J Blueman <[email protected]>
Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix iterator leak in may_delete_deleted_inode()

may_delete_deleted_inode() was returning without exiting a btree
iterator, eventually causing propagate_key_to_snaphot_leaves() to go
into an infinite loop hitting btree_trans_too_many_iters().

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Kill journal pre-reservations

This deletes the complicated and somewhat expensive journal
pre-reservation machinery in favor of just using journal watermarks:
when the journal is more than half full, we run journal reclaim more
aggressively, and when the journal is more than 3/4s full we only allow
journal reclaim to get new journal reservations.

Signed-off-by: Kent Overstreet <[email protected]>

Merge tag 'zstd-linus-v6.7-rc2' of https://github.com/terrelln/linux

Pull Zstd fix from Nick Terrell:
 "Only a single line change to fix a benign UBSAN warning"

* tag 'zstd-linus-v6.7-rc2' of https://github.com/terrelln/linux:
  zstd: Fix array-index-out-of-bounds UBSAN warning

Merge branch 'mptcp-misc-fixes-for-v6-7'

Matthieu Baerts says:

====================
mptcp: misc. fixes for v6.7

Here are a few fixes related to MPTCP:

- Patch 1 limits GSO max size to ~64K when MPTCP is being used due to a
  spec limit. 'gso_max_size' can exceed the max value supported by MPTCP
  since v5.19.

- Patch 2 fixes a possible NULL pointer dereference on close that can
  happen since v6.7-rc1.

- Patch 3 avoids sending a RM_ADDR when the corresponding address is no
  longer tracked locally. A regression for a fix backported to v5.19.

- Patch 4 adds a missing lock when changing the IP TOS with setsockopt().
  A fix for v5.17.

- Patch 5 fixes an expectation when running MPTCP Join selftest with the
  checksum option (-C). An issue present since v6.1.
====================

Link: https://lore.kernel.org/r/20231114-upstream-net-20231113-mptcp-misc-fixes-6-7-rc2-v1-0-7b9cd6a7b7f4@kernel.org
Signed-off-by: Jakub Kicinski <[email protected]>

selftests: mptcp: fix fastclose with csum failure

Running the mp_join selftest manually with the following command line:

  ./mptcp_join.sh -z -C

leads to some failures:

  002 fastclose server test
  # ...
  rtx                                 [fail] got 1 MP_RST[s] TX expected 0
  # ...
  rstrx                               [fail] got 1 MP_RST[s] RX expected 0

The problem is really in the wrong expectations for the RST checks
implied by the csum validation. Note that the same check is repeated
explicitly in the same test-case, with the correct expectation and
pass successfully.

Address the issue explicitly setting the correct expectation for
the failing checks.

Reported-by: Xiumei Mu <[email protected]>
Fixes: 6bf41020b72b ("selftests: mptcp: update and extend fastclose test-cases")
Cc: [email protected]
Signed-off-by: Paolo Abeni <[email protected]>
Reviewed-by: Matthieu Baerts <[email protected]>
Signed-off-by: Matthieu Baerts <[email protected]>
Link: https://lore.kernel.org/r/20231114-upstream-net-20231113-mptcp-misc-fixes-6-7-rc2-v1-5-7b9cd6a7b7f4@kernel.org
Signed-off-by: Jakub Kicinski <[email protected]>

mptcp: fix setsockopt(IP_TOS) subflow locking

The MPTCP implementation of the IP_TOS socket option uses the lockless
variant of the TOS manipulation helper and does not hold such lock at
the helper invocation time.

Add the required locking.

Fixes: ffcacff87cd6 ("mptcp: Support for IP_TOS for MPTCP setsockopt()")
Cc: [email protected]
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/457
Signed-off-by: Paolo Abeni <[email protected]>
Reviewed-by: Mat Martineau <[email protected]>
Signed-off-by: Matthieu Baerts <[email protected]>
Link: https://lore.kernel.org/r/20231114-upstream-net-20231113-mptcp-misc-fixes-6-7-rc2-v1-4-7b9cd6a7b7f4@kernel.org
Signed-off-by: Jakub Kicinski <[email protected]>

mptcp: add validity check for sending RM_ADDR

This patch adds the validity check for sending RM_ADDRs for userspace PM
in mptcp_pm_remove_addrs(), only send a RM_ADDR when the address is in the
anno_list or conn_list.

Fixes: 8b1c94da1e48 ("mptcp: only send RM_ADDR in nl_cmd_remove")
Cc: [email protected]
Signed-off-by: Geliang Tang <[email protected]>
Reviewed-by: Mat Martineau <[email protected]>
Signed-off-by: Matthieu Baerts <[email protected]>
Link: https://lore.kernel.org/r/20231114-upstream-net-20231113-mptcp-misc-fixes-6-7-rc2-v1-3-7b9cd6a7b7f4@kernel.org
Signed-off-by: Jakub Kicinski <[email protected]>

mptcp: fix possible NULL pointer dereference on close

After the blamed commit below, the MPTCP release callback can
dereference the first subflow pointer via __mptcp_set_connected()
and send buffer auto-tuning. Such pointer is always expected to be
valid, except at socket destruction time, when the first subflow is
deleted and the pointer zeroed.

If the connect event is handled by the release callback while the
msk socket is finally released, MPTCP hits the following splat:

  general protection fault, probably for non-canonical address 0xdffffc00000000f2: 0000 [#1] PREEMPT SMP KASAN
  KASAN: null-ptr-deref in range [0x0000000000000790-0x0000000000000797]
  CPU: 1 PID: 26719 Comm: syz-executor.2 Not tainted 6.6.0-syzkaller-10102-gff269e2cd5ad #0
  Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/09/2023
  RIP: 0010:mptcp_subflow_ctx net/mptcp/protocol.h:542 [inline]
  RIP: 0010:__mptcp_propagate_sndbuf net/mptcp/protocol.h:813 [inline]
  RIP: 0010:__mptcp_set_connected+0x57/0x3e0 net/mptcp/subflow.c:424
  RAX: dffffc0000000000 RBX: 0000000000000000 RCX: ffffffff8a62323c
  RDX: 00000000000000f2 RSI: ffffffff8a630116 RDI: 0000000000000790
  RBP: ffff88803334b100 R08: 0000000000000001 R09: 0000000000000000
  R10: 0000000000000001 R11: 0000000000000034 R12: ffff88803334b198
  R13: ffff888054f0b018 R14: 0000000000000000 R15: ffff88803334b100
  FS:  0000000000000000(0000) GS:ffff8880b9900000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 00007fbcb4f75198 CR3: 000000006afb5000 CR4: 00000000003506f0
  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
  Call Trace:
   <TASK>
   mptcp_release_cb+0xa2c/0xc40 net/mptcp/protocol.c:3405
   release_sock+0xba/0x1f0 net/core/sock.c:3537
   mptcp_close+0x32/0xf0 net/mptcp/protocol.c:3084
   inet_release+0x132/0x270 net/ipv4/af_inet.c:433
   inet6_release+0x4f/0x70 net/ipv6/af_inet6.c:485
   __sock_release+0xae/0x260 net/socket.c:659
   sock_close+0x1c/0x20 net/socket.c:1419
   __fput+0x270/0xbb0 fs/file_table.c:394
   task_work_run+0x14d/0x240 kernel/task_work.c:180
   exit_task_work include/linux/task_work.h:38 [inline]
   do_exit+0xa92/0x2a20 kernel/exit.c:876
   do_group_exit+0xd4/0x2a0 kernel/exit.c:1026
   get_signal+0x23ba/0x2790 kernel/signal.c:2900
   arch_do_signal_or_restart+0x90/0x7f0 arch/x86/kernel/signal.c:309
   exit_to_user_mode_loop kernel/entry/common.c:168 [inline]
   exit_to_user_mode_prepare+0x11f/0x240 kernel/entry/common.c:204
   __syscall_exit_to_user_mode_work kernel/entry/common.c:285 [inline]
   syscall_exit_to_user_mode+0x1d/0x60 kernel/entry/common.c:296
   do_syscall_64+0x4b/0x110 arch/x86/entry/common.c:88
   entry_SYSCALL_64_after_hwframe+0x63/0x6b
  RIP: 0033:0x7fb515e7cae9
  Code: Unable to access opcode bytes at 0x7fb515e7cabf.
  RSP: 002b:00007fb516c560c8 EFLAGS: 00000246 ORIG_RAX: 000000000000002e
  RAX: 000000000000003c RBX: 00007fb515f9c120 RCX: 00007fb515e7cae9
  RDX: 0000000000000000 RSI: 0000000020000140 RDI: 0000000000000006
  RBP: 00007fb515ec847a R08: 0000000000000000 R09: 0000000000000000
  R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000000
  R13: 000000000000006e R14: 00007fb515f9c120 R15: 00007ffc631eb968
   </TASK>

To avoid sparkling unneeded conditionals, address the issue explicitly
checking msk->first only in the critical place.

Fixes: 8005184fd1ca ("mptcp: refactor sndbuf auto-tuning")
Cc: [email protected]
Reported-by: <[email protected]>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/454
Reported-by: Eric Dumazet <[email protected]>
Closes: https://lore.kernel.org/netdev/CANn89iLZUA6S2a=K8GObnS62KK6Jt4B7PsAs7meMFooM8xaTgw@mail.gmail.com/
Signed-off-by: Paolo Abeni <[email protected]>
Reviewed-by: Eric Dumazet <[email protected]>
Reviewed-by: Mat Martineau <[email protected]>
Signed-off-by: Matthieu Baerts <[email protected]>
Link: https://lore.kernel.org/r/20231114-upstream-net-20231113-mptcp-misc-fixes-6-7-rc2-v1-2-7b9cd6a7b7f4@kernel.org
Signed-off-by: Jakub Kicinski <[email protected]>

mptcp: deal with large GSO size

After the blamed commit below, the TCP sockets (and the MPTCP subflows)
can build egress packets larger than 64K. That exceeds the maximum DSS
data size, the length being misrepresent on the wire and the stream being
corrupted, as later observed on the receiver:

  WARNING: CPU: 0 PID: 9696 at net/mptcp/protocol.c:705 __mptcp_move_skbs_from_subflow+0x2604/0x26e0
  CPU: 0 PID: 9696 Comm: syz-executor.7 Not tainted 6.6.0-rc5-gcd8bdf563d46 #45
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.11.0-2.el7 04/01/2014
  netlink: 8 bytes leftover after parsing attributes in process `syz-executor.4'.
  RIP: 0010:__mptcp_move_skbs_from_subflow+0x2604/0x26e0 net/mptcp/protocol.c:705
  RSP: 0018:ffffc90000006e80 EFLAGS: 00010246
  RAX: ffffffff83e9f674 RBX: ffff88802f45d870 RCX: ffff888102ad0000
  netlink: 8 bytes leftover after parsing attributes in process `syz-executor.4'.
  RDX: 0000000080000303 RSI: 0000000000013908 RDI: 0000000000003908
  RBP: ffffc90000007110 R08: ffffffff83e9e078 R09: 1ffff1100e548c8a
  R10: dffffc0000000000 R11: ffffed100e548c8b R12: 0000000000013908
  R13: dffffc0000000000 R14: 0000000000003908 R15: 000000000031cf29
  FS:  00007f239c47e700(0000) GS:ffff88811b200000(0000) knlGS:0000000000000000
  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
  CR2: 00007f239c45cd78 CR3: 000000006a66c006 CR4: 0000000000770ef0
  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
  DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000600
  PKRU: 55555554
  Call Trace:
   <IRQ>
   mptcp_data_ready+0x263/0xac0 net/mptcp/protocol.c:819
   subflow_data_ready+0x268/0x6d0 net/mptcp/subflow.c:1409
   tcp_data_queue+0x21a1/0x7a60 net/ipv4/tcp_input.c:5151
   tcp_rcv_established+0x950/0x1d90 net/ipv4/tcp_input.c:6098
   tcp_v6_do_rcv+0x554/0x12f0 net/ipv6/tcp_ipv6.c:1483
   tcp_v6_rcv+0x2e26/0x3810 net/ipv6/tcp_ipv6.c:1749
   ip6_protocol_deliver_rcu+0xd6b/0x1ae0 net/ipv6/ip6_input.c:438
   ip6_input+0x1c5/0x470 net/ipv6/ip6_input.c:483
   ipv6_rcv+0xef/0x2c0 include/linux/netfilter.h:304
   __netif_receive_skb+0x1ea/0x6a0 net/core/dev.c:5532
   process_backlog+0x353/0x660 net/core/dev.c:5974
   __napi_poll+0xc6/0x5a0 net/core/dev.c:6536
   net_rx_action+0x6a0/0xfd0 net/core/dev.c:6603
   __do_softirq+0x184/0x524 kernel/softirq.c:553
   do_softirq+0xdd/0x130 kernel/softirq.c:454

Address the issue explicitly bounding the maximum GSO size to what MPTCP
actually allows.

Reported-by: Christoph Paasch <[email protected]>
Closes: https://github.com/multipath-tcp/mptcp_net-next/issues/450
Fixes: 7c4e983c4f3c ("net: allow gso_max_size to exceed 65536")
Cc: [email protected]
Signed-off-by: Paolo Abeni <[email protected]>
Reviewed-by: Mat Martineau <[email protected]>
Signed-off-by: Matthieu Baerts <[email protected]>
Link: https://lore.kernel.org/r/20231114-upstream-net-20231113-mptcp-misc-fixes-6-7-rc2-v1-1-7b9cd6a7b7f4@kernel.org
Signed-off-by: Jakub Kicinski <[email protected]>

gve: Fixes for napi_poll when budget is 0

Netpoll will explicilty pass the polling call with a budget of 0 to
indicate it's clearing the Tx path only. For the gve_rx_poll and
gve_xdp_poll, they were mistakenly taking the 0 budget as the indication
to do all the work. Add check to avoid the rx path and xdp path being
called when budget is 0. And also avoid napi_complete_done being called
when budget is 0 for netpoll.

Fixes: f5cedc84a30d ("gve: Add transmit and receive support")
Signed-off-by: Ziwei Xiao <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Merge branch '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue

Tony Nguyen says:

====================
Intel Wired LAN Driver Updates 2023-11-13 (ice)

This series contains updates to ice driver only.

Arkadiusz ensures the device is initialized with valid lock status
value. He also removes range checking of dpll priority to allow firmware
to process the request; supported values are firmware dependent.
Finally, he removes setting of can change capability for pins that
cannot be changed.

Dan restores ability to load a package which doesn't contain a signature
segment.

* '100GbE' of git://git.kernel.org/pub/scm/linux/kernel/git/tnguy/net-queue:
  ice: fix DDP package download for packages without signature segment
  ice: dpll: fix output pin capabilities
  ice: dpll: fix check for dpll input priority range
  ice: dpll: fix initial lock status of dpll
====================

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Merge branch 'pds_core-fix-irq-index-bug-and-compiler-warnings'

Shannon Nelson says:

====================
pds_core: fix irq index bug and compiler warnings

The first patch fixes a bug in our interrupt masking where we used the
wrong index.  The second patch addresses a couple of kernel test robot
string truncation warnings.
====================

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

pds_core: fix up some format-truncation complaints

Our friendly kernel test robot pointed out a couple of potential
string truncation issues.  None of which were we worried about,
but can be relatively easily fixed to quiet the complaints.

Reported-by: kernel test robot <[email protected]>
Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
Fixes: 45d76f492938 ("pds_core: set up device and adminq")
Signed-off-by: Shannon Nelson <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

pds_core: use correct index to mask irq

Use the qcq's interrupt index, not the irq number, to mask
the interrupt.  Since the irq number can be out of range from
the number of possible interrupts, we can end up accessing
and potentially scribbling on out-of-range and/or unmapped
memory, making the kernel angry.

    [ 3116.039364] BUG: unable to handle page fault for address: ffffbeea1c3edf84
    [ 3116.047059] #PF: supervisor write access in kernel mode
    [ 3116.052895] #PF: error_code(0x0002) - not-present page
    [ 3116.058636] PGD 100000067 P4D 100000067 PUD 1001f2067 PMD 10f82e067 PTE 0
    [ 3116.066221] Oops: 0002 [#1] SMP NOPTI
    [ 3116.092948] RIP: 0010:iowrite32+0x9/0x76
    [ 3116.190452] Call Trace:
    [ 3116.193185]  <IRQ>
    [ 3116.195430]  ? show_trace_log_lvl+0x1d6/0x2f9
    [ 3116.200298]  ? show_trace_log_lvl+0x1d6/0x2f9
    [ 3116.205166]  ? pdsc_adminq_isr+0x43/0x55 [pds_core]
    [ 3116.210618]  ? __die_body.cold+0x8/0xa
    [ 3116.214806]  ? page_fault_oops+0x16d/0x1ac
    [ 3116.219382]  ? exc_page_fault+0xbe/0x13b
    [ 3116.223764]  ? asm_exc_page_fault+0x22/0x27
    [ 3116.228440]  ? iowrite32+0x9/0x76
    [ 3116.232143]  pdsc_adminq_isr+0x43/0x55 [pds_core]
    [ 3116.237627]  __handle_irq_event_percpu+0x3a/0x184
    [ 3116.243088]  handle_irq_event+0x57/0xb0
    [ 3116.247575]  handle_edge_irq+0x87/0x225
    [ 3116.252062]  __common_interrupt+0x3e/0xbc
    [ 3116.256740]  common_interrupt+0x7b/0x98
    [ 3116.261216]  </IRQ>
    [ 3116.263745]  <TASK>
    [ 3116.266268]  asm_common_interrupt+0x22/0x27

Reported-by: Joao Martins <[email protected]>
Fixes: 01ba61b55b20 ("pds_core: Add adminq processing and commands")
Signed-off-by: Shannon Nelson <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

tg3: Increment tx_dropped in tg3_tso_bug()

tg3_tso_bug() drops a packet if it cannot be segmented for any reason.
The number of discarded frames should be incremented accordingly.

Signed-off-by: Alex Pakhunov <[email protected]>
Signed-off-by: Vincent Wong <[email protected]>
Reviewed-by: Pavan Chebbi <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

tg3: Move the [rt]x_dropped counters to tg3_napi

This change moves [rt]x_dropped counters to tg3_napi so that they can be
updated by a single writer, race-free.

Signed-off-by: Alex Pakhunov <[email protected]>
Signed-off-by: Vincent Wong <[email protected]>
Reviewed-by: Michael Chan <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

net: stmmac: avoid rx queue overrun

dma_rx_size can be set as low as 64. Rx budget might be higher than
that. Make sure to not overrun allocated rx buffers when budget is
larger.

Leave one descriptor unused to avoid wrap around of 'dirty_rx' vs
'cur_rx'.

Signed-off-by: Baruch Siach <[email protected]>
Reviewed-by: Serge Semin <[email protected]>
Fixes: 47dd7a540b8a ("net: add support for STMicroelectronics Ethernet controllers.")
Link: https://lore.kernel.org/r/d95413e44c97d4692e72cec13a75f894abeb6998.1699897370.git.baruch@tkos.co.il
Signed-off-by: Jakub Kicinski <[email protected]>

net: stmmac: fix rx budget limit check

The while loop condition verifies 'count < limit'. Neither value change
before the 'count >= limit' check. As is this check is dead code. But
code inspection reveals a code path that modifies 'count' and then goto
'drain_data' and back to 'read_again'. So there is a need to verify
count value sanity after 'read_again'.

Move 'read_again' up to fix the count limit check.

Fixes: ec222003bd94 ("net: stmmac: Prepare to add Split Header support")
Signed-off-by: Baruch Siach <[email protected]>
Reviewed-by: Serge Semin <[email protected]>
Link: https://lore.kernel.org/r/d9486296c3b6b12ab3a0515fcd47d56447a07bfc.1699897370.git.baruch@tkos.co.il
Signed-off-by: Jakub Kicinski <[email protected]>

perf/core: Fix cpuctx refcounting

Audit of the refcounting turned up that perf_pmu_migrate_context()
fails to migrate the ctx refcount.

Fixes: bd2756811766 ("perf: Rewrite core context handling")
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
Cc: <[email protected]>

futex: Fix hardcoded flags

Xi reported that commit 5694289ce183 ("futex: Flag conversion") broke
glibc's robust futex tests.

This was narrowed down to the change of FLAGS_SHARED from 0x01 to
0x10, at which point Florian noted that handle_futex_death() has a
hardcoded flags argument of 1.

Change this to: FLAGS_SIZE_32 | FLAGS_SHARED, matching how
futex_to_flags() unconditionally sets FLAGS_SIZE_32 for all legacy
futex ops.

Reported-by: Xi Ruoyao <[email protected]>
Reported-by: Florian Weimer <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Signed-off-by: Ingo Molnar <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]
Fixes: 5694289ce183 ("futex: Flag conversion")
Cc: <[email protected]>

zstd: Fix array-index-out-of-bounds UBSAN warning

Zstd used an array of length 1 to mean a flexible array for C89
compatibility. Switch to a C99 flexible array to fix the UBSAN warning.

Tested locally by booting the kernel and writing to and reading from a
BtrFS filesystem with zstd compression enabled. I was unable to reproduce
the issue before the fix, however it is a trivial change.

Link: https://lkml.kernel.org/r/[email protected]
Reported-by: [email protected]
Reported-by: Eric Biggers <[email protected]>
Reported-by: Kees Cook <[email protected]>
Signed-off-by: Nick Terrell <[email protected]>
Reviewed-by: Kees Cook <[email protected]>

audit: don't WARN_ON_ONCE(!current->mm) in audit_exe_compare()

eBPF can end up calling into the audit code from some odd places, and
some of these places don't have @current set properly so we end up
tripping the `WARN_ON_ONCE(!current->mm)` near the top of
`audit_exe_compare()`.  While the basic `!current->mm` check is good,
the `WARN_ON_ONCE()` results in some scary console messages so let's
drop that and just do the regular `!current->mm` check to avoid
problems.

Cc: <[email protected]>
Fixes: 47846d51348d ("audit: don't take task_lock() in audit_exe_compare() code path")
Reported-by: Artem Savkov <[email protected]>
Signed-off-by: Paul Moore <[email protected]>

nouveau: use an rwlock for the event lock.

This allows it to break the following circular locking dependency.

Aug 10 07:01:29 dg1test kernel: ======================================================
Aug 10 07:01:29 dg1test kernel: WARNING: possible circular locking dependency detected
Aug 10 07:01:29 dg1test kernel: 6.4.0-rc7+ #10 Not tainted
Aug 10 07:01:29 dg1test kernel: ------------------------------------------------------
Aug 10 07:01:29 dg1test kernel: wireplumber/2236 is trying to acquire lock:
Aug 10 07:01:29 dg1test kernel: ffff8fca5320da18 (&fctx->lock){-...}-{2:2}, at: nouveau_fence_wait_uevent_handler+0x2b/0x100 [nouveau]
Aug 10 07:01:29 dg1test kernel:
                                but task is already holding lock:
Aug 10 07:01:29 dg1test kernel: ffff8fca41208610 (&event->list_lock#2){-...}-{2:2}, at: nvkm_event_ntfy+0x50/0xf0 [nouveau]
Aug 10 07:01:29 dg1test kernel:
                                which lock already depends on the new lock.
Aug 10 07:01:29 dg1test kernel:
                                the existing dependency chain (in reverse order) is:
Aug 10 07:01:29 dg1test kernel:
                                -> #3 (&event->list_lock#2){-...}-{2:2}:
Aug 10 07:01:29 dg1test kernel:        _raw_spin_lock_irqsave+0x4b/0x70
Aug 10 07:01:29 dg1test kernel:        nvkm_event_ntfy+0x50/0xf0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        ga100_fifo_nonstall_intr+0x24/0x30 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_intr+0x12c/0x240 [nouveau]
Aug 10 07:01:29 dg1test kernel:        __handle_irq_event_percpu+0x88/0x240
Aug 10 07:01:29 dg1test kernel:        handle_irq_event+0x38/0x80
Aug 10 07:01:29 dg1test kernel:        handle_edge_irq+0xa3/0x240
Aug 10 07:01:29 dg1test kernel:        __common_interrupt+0x72/0x160
Aug 10 07:01:29 dg1test kernel:        common_interrupt+0x60/0xe0
Aug 10 07:01:29 dg1test kernel:        asm_common_interrupt+0x26/0x40
Aug 10 07:01:29 dg1test kernel:
                                -> #2 (&device->intr.lock){-...}-{2:2}:
Aug 10 07:01:29 dg1test kernel:        _raw_spin_lock_irqsave+0x4b/0x70
Aug 10 07:01:29 dg1test kernel:        nvkm_inth_allow+0x2c/0x80 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_event_ntfy_state+0x181/0x250 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_event_ntfy_allow+0x63/0xd0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_uevent_mthd+0x4d/0x70 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_ioctl+0x10b/0x250 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvif_object_mthd+0xa8/0x1f0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvif_event_allow+0x2a/0xa0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nouveau_fence_enable_signaling+0x78/0x80 [nouveau]
Aug 10 07:01:29 dg1test kernel:        __dma_fence_enable_signaling+0x5e/0x100
Aug 10 07:01:29 dg1test kernel:        dma_fence_add_callback+0x4b/0xd0
Aug 10 07:01:29 dg1test kernel:        nouveau_cli_work_queue+0xae/0x110 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nouveau_gem_object_close+0x1d1/0x2a0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        drm_gem_handle_delete+0x70/0xe0 [drm]
Aug 10 07:01:29 dg1test kernel:        drm_ioctl_kernel+0xa5/0x150 [drm]
Aug 10 07:01:29 dg1test kernel:        drm_ioctl+0x256/0x490 [drm]
Aug 10 07:01:29 dg1test kernel:        nouveau_drm_ioctl+0x5a/0xb0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        __x64_sys_ioctl+0x91/0xd0
Aug 10 07:01:29 dg1test kernel:        do_syscall_64+0x3c/0x90
Aug 10 07:01:29 dg1test kernel:        entry_SYSCALL_64_after_hwframe+0x72/0xdc
Aug 10 07:01:29 dg1test kernel:
                                -> #1 (&event->refs_lock#4){....}-{2:2}:
Aug 10 07:01:29 dg1test kernel:        _raw_spin_lock_irqsave+0x4b/0x70
Aug 10 07:01:29 dg1test kernel:        nvkm_event_ntfy_state+0x37/0x250 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_event_ntfy_allow+0x63/0xd0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_uevent_mthd+0x4d/0x70 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_ioctl+0x10b/0x250 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvif_object_mthd+0xa8/0x1f0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvif_event_allow+0x2a/0xa0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nouveau_fence_enable_signaling+0x78/0x80 [nouveau]
Aug 10 07:01:29 dg1test kernel:        __dma_fence_enable_signaling+0x5e/0x100
Aug 10 07:01:29 dg1test kernel:        dma_fence_add_callback+0x4b/0xd0
Aug 10 07:01:29 dg1test kernel:        nouveau_cli_work_queue+0xae/0x110 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nouveau_gem_object_close+0x1d1/0x2a0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        drm_gem_handle_delete+0x70/0xe0 [drm]
Aug 10 07:01:29 dg1test kernel:        drm_ioctl_kernel+0xa5/0x150 [drm]
Aug 10 07:01:29 dg1test kernel:        drm_ioctl+0x256/0x490 [drm]
Aug 10 07:01:29 dg1test kernel:        nouveau_drm_ioctl+0x5a/0xb0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        __x64_sys_ioctl+0x91/0xd0
Aug 10 07:01:29 dg1test kernel:        do_syscall_64+0x3c/0x90
Aug 10 07:01:29 dg1test kernel:        entry_SYSCALL_64_after_hwframe+0x72/0xdc
Aug 10 07:01:29 dg1test kernel:
                                -> #0 (&fctx->lock){-...}-{2:2}:
Aug 10 07:01:29 dg1test kernel:        __lock_acquire+0x14e3/0x2240
Aug 10 07:01:29 dg1test kernel:        lock_acquire+0xc8/0x2a0
Aug 10 07:01:29 dg1test kernel:        _raw_spin_lock_irqsave+0x4b/0x70
Aug 10 07:01:29 dg1test kernel:        nouveau_fence_wait_uevent_handler+0x2b/0x100 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_client_event+0xf/0x20 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_event_ntfy+0x9b/0xf0 [nouveau]
Aug 10 07:01:29 dg1test kernel:        ga100_fifo_nonstall_intr+0x24/0x30 [nouveau]
Aug 10 07:01:29 dg1test kernel:        nvkm_intr+0x12c/0x240 [nouveau]
Aug 10 07:01:29 dg1test kernel:        __handle_irq_event_percpu+0x88/0x240
Aug 10 07:01:29 dg1test kernel:        handle_irq_event+0x38/0x80
Aug 10 07:01:29 dg1test kernel:        handle_edge_irq+0xa3/0x240
Aug 10 07:01:29 dg1test kernel:        __common_interrupt+0x72/0x160
Aug 10 07:01:29 dg1test kernel:        common_interrupt+0x60/0xe0
Aug 10 07:01:29 dg1test kernel:        asm_common_interrupt+0x26/0x40
Aug 10 07:01:29 dg1test kernel:
                                other info that might help us debug this:
Aug 10 07:01:29 dg1test kernel: Chain exists of:
                                  &fctx->lock --> &device->intr.lock --> &event->list_lock#2
Aug 10 07:01:29 dg1test kernel:  Possible unsafe locking scenario:
Aug 10 07:01:29 dg1test kernel:        CPU0                    CPU1
Aug 10 07:01:29 dg1test kernel:        ----                    ----
Aug 10 07:01:29 dg1test kernel:   lock(&event->list_lock#2);
Aug 10 07:01:29 dg1test kernel:                                lock(&device->intr.lock);
Aug 10 07:01:29 dg1test kernel:                                lock(&event->list_lock#2);
Aug 10 07:01:29 dg1test kernel:   lock(&fctx->lock);
Aug 10 07:01:29 dg1test kernel:
                                 *** DEADLOCK ***
Aug 10 07:01:29 dg1test kernel: 2 locks held by wireplumber/2236:
Aug 10 07:01:29 dg1test kernel:  #0: ffff8fca53177bf8 (&device->intr.lock){-...}-{2:2}, at: nvkm_intr+0x29/0x240 [nouveau]
Aug 10 07:01:29 dg1test kernel:  #1: ffff8fca41208610 (&event->list_lock#2){-...}-{2:2}, at: nvkm_event_ntfy+0x50/0xf0 [nouveau]
Aug 10 07:01:29 dg1test kernel:
                                stack backtrace:
Aug 10 07:01:29 dg1test kernel: CPU: 6 PID: 2236 Comm: wireplumber Not tainted 6.4.0-rc7+ #10
Aug 10 07:01:29 dg1test kernel: Hardware name: Gigabyte Technology Co., Ltd. Z390 I AORUS PRO WIFI/Z390 I AORUS PRO WIFI-CF, BIOS F8 11/05/2021
Aug 10 07:01:29 dg1test kernel: Call Trace:
Aug 10 07:01:29 dg1test kernel:  <TASK>
Aug 10 07:01:29 dg1test kernel:  dump_stack_lvl+0x5b/0x90
Aug 10 07:01:29 dg1test kernel:  check_noncircular+0xe2/0x110
Aug 10 07:01:29 dg1test kernel:  __lock_acquire+0x14e3/0x2240
Aug 10 07:01:29 dg1test kernel:  lock_acquire+0xc8/0x2a0
Aug 10 07:01:29 dg1test kernel:  ? nouveau_fence_wait_uevent_handler+0x2b/0x100 [nouveau]
Aug 10 07:01:29 dg1test kernel:  ? lock_acquire+0xc8/0x2a0
Aug 10 07:01:29 dg1test kernel:  _raw_spin_lock_irqsave+0x4b/0x70
Aug 10 07:01:29 dg1test kernel:  ? nouveau_fence_wait_uevent_handler+0x2b/0x100 [nouveau]
Aug 10 07:01:29 dg1test kernel:  nouveau_fence_wait_uevent_handler+0x2b/0x100 [nouveau]
Aug 10 07:01:29 dg1test kernel:  nvkm_client_event+0xf/0x20 [nouveau]
Aug 10 07:01:29 dg1test kernel:  nvkm_event_ntfy+0x9b/0xf0 [nouveau]
Aug 10 07:01:29 dg1test kernel:  ga100_fifo_nonstall_intr+0x24/0x30 [nouveau]
Aug 10 07:01:29 dg1test kernel:  nvkm_intr+0x12c/0x240 [nouveau]
Aug 10 07:01:29 dg1test kernel:  __handle_irq_event_percpu+0x88/0x240
Aug 10 07:01:29 dg1test kernel:  handle_irq_event+0x38/0x80
Aug 10 07:01:29 dg1test kernel:  handle_edge_irq+0xa3/0x240
Aug 10 07:01:29 dg1test kernel:  __common_interrupt+0x72/0x160
Aug 10 07:01:29 dg1test kernel:  common_interrupt+0x60/0xe0
Aug 10 07:01:29 dg1test kernel:  asm_common_interrupt+0x26/0x40
Aug 10 07:01:29 dg1test kernel: RIP: 0033:0x7fb66174d700
Aug 10 07:01:29 dg1test kernel: Code: c1 e2 05 29 ca 8d 0c 10 0f be 07 84 c0 75 eb 89 c8 c3 0f 1f 84 00 00 00 00 00 f3 0f 1e fa e9 d7 0f fc ff 0f 1f 80 00 00 00 00 <f3> 0f 1e fa e9 c7 0f fc>
Aug 10 07:01:29 dg1test kernel: RSP: 002b:00007ffdd3c48438 EFLAGS: 00000206
Aug 10 07:01:29 dg1test kernel: RAX: 000055bb758763c0 RBX: 000055bb758752c0 RCX: 00000000000028b0
Aug 10 07:01:29 dg1test kernel: RDX: 000055bb758752c0 RSI: 000055bb75887490 RDI: 000055bb75862950
Aug 10 07:01:29 dg1test kernel: RBP: 00007ffdd3c48490 R08: 000055bb75873b10 R09: 0000000000000001
Aug 10 07:01:29 dg1test kernel: R10: 0000000000000004 R11: 000055bb7587f000 R12: 000055bb75887490
Aug 10 07:01:29 dg1test kernel: R13: 000055bb757f6280 R14: 000055bb758875c0 R15: 000055bb757f6280
Aug 10 07:01:29 dg1test kernel:  </TASK>

Signed-off-by: Dave Airlie <[email protected]>
Tested-by: Danilo Krummrich <[email protected]>
Reviewed-by: Danilo Krummrich <[email protected]>
Signed-off-by: Danilo Krummrich <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]

nouveau/gsp/r535: Fix a NULL vs error pointer bug

The r535_gsp_cmdq_get() function returns error pointers but this code
checks for NULL.  Also we need to propagate the error pointer back to
the callers in r535_gsp_rpc_get().  Returning NULL will lead to a NULL
pointer dereference.

Fixes: 176fdcbddfd2 ("drm/nouveau/gsp/r535: add support for booting GSP-RM")
Signed-off-by: Dan Carpenter <[email protected]>
Reviewed-by: Danilo Krummrich <[email protected]>
Signed-off-by: Danilo Krummrich <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]

nouveau/gsp/r535: uninitialized variable in r535_gsp_acpi_mux_id()

The if we hit the "continue" statement on the first iteration through
the loop then "handle_mux" needs to be set to NULL so we continue
looping.

Fixes: 176fdcbddfd2 ("drm/nouveau/gsp/r535: add support for booting GSP-RM")
Signed-off-by: Dan Carpenter <[email protected]>
Reviewed-by: Danilo Krummrich <[email protected]>
Signed-off-by: Danilo Krummrich <[email protected]>
Link: https://patchwork.freedesktop.org/patch/msgid/[email protected]

sched/fair: Fix the decision for load balance

should_we_balance is called for the decision to do load-balancing.
When sched ticks invoke this function, only one CPU should return
true. However, in the current code, two CPUs can return true. The
following situation, where b means busy and i means idle, is an
example, because CPU 0 and CPU 2 return true.

        [0, 1] [2, 3]
         b  b   i  b

This fix checks if there exists an idle CPU with busy sibling(s)
after looking for a CPU on an idle core. If some idle CPUs with busy
siblings are found, just the first one should do load-balancing.

Fixes: b1bfeab9b002 ("sched/fair: Consider the idle state of the whole core for load balance")
Signed-off-by: Keisuke Nishimura <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Reviewed-by: Chen Yu <[email protected]>
Reviewed-by: Shrikanth Hegde <[email protected]>
Reviewed-by: Vincent Guittot <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]

sched: psi: fix unprivileged polling against cgroups

519fabc7aaba ("psi: remove 500ms min window size limitation for
triggers") breaks unprivileged psi polling on cgroups.

Historically, we had a privilege check for polling in the open() of a
pressure file in /proc, but were erroneously missing it for the open()
of cgroup pressure files.

When unprivileged polling was introduced in d82caa273565 ("sched/psi:
Allow unprivileged polling of N*2s period"), it needed to filter
privileges depending on the exact polling parameters, and as such
moved the CAP_SYS_RESOURCE check from the proc open() callback to
psi_trigger_create(). Both the proc files as well as cgroup files go
through this during write(). This implicitly added the missing check
for privileges required for HT polling for cgroups.

When 519fabc7aaba ("psi: remove 500ms min window size limitation for
triggers") followed right after to remove further restrictions on the
RT polling window, it incorrectly assumed the cgroup privilege check
was still missing and added it to the cgroup open(), mirroring what we
used to do for proc files in the past.

As a result, unprivileged poll requests that would be supported now
get rejected when opening the cgroup pressure file for writing.

Remove the cgroup open() check. psi_trigger_create() handles it.

Fixes: 519fabc7aaba ("psi: remove 500ms min window size limitation for triggers")
Reported-by: Luca Boccassi <[email protected]>
Signed-off-by: Johannes Weiner <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Acked-by: Luca Boccassi <[email protected]>
Acked-by: Suren Baghdasaryan <[email protected]>
Cc: [email protected] # 6.5+
Link: https://lore.kernel.org/r/[email protected]

sched/eevdf: Fix vruntime adjustment on reweight

vruntime of the (on_rq && !0-lag) entity needs to be adjusted when
it gets re-weighted, and the calculations can be simplified based
on the fact that re-weight won't change the w-average of all the
entities. Please check the proofs in comments.

But adjusting vruntime can also cause position change in RB-tree
hence require re-queue to fix up which might be costly. This might
be avoided by deferring adjustment to the time the entity actually
leaves tree (dequeue/pick), but that will negatively affect task
selection and probably not good enough either.

Fixes: 147f3efaa241 ("sched/fair: Implement an EEVDF-like scheduling policy")
Signed-off-by: Abel Wu <[email protected]>
Signed-off-by: Peter Zijlstra (Intel) <[email protected]>
Link: https://lkml.kernel.org/r/[email protected]

cifs: fix lock ordering while disabling multichannel

The code to handle the case of server disabling multichannel
was picking iface_lock with chan_lock held. This goes against
the lock ordering rules, as iface_lock is a higher order lock
(even if it isn't so obvious).

This change fixes the lock ordering by doing the following in
that order for each secondary channel:
1. store iface and server pointers in local variable
2. remove references to iface and server in channels
3. unlock chan_lock
4. lock iface_lock
5. dec ref count for iface
6. unlock iface_lock
7. dec ref count for server
8. lock chan_lock again

Since this function can only be called in smb2_reconnect, and
that cannot be called by two parallel processes, we should not
have races due to dropping chan_lock between steps 3 and 8.

Fixes: ee1d21794e55 ("cifs: handle when server stops supporting multichannel")
Reported-by: Paulo Alcantara <[email protected]>
Signed-off-by: Shyam Prasad N <[email protected]>
Signed-off-by: Steve French <[email protected]>

cifs: fix leak of iface for primary channel

My last change in this area introduced a change which
accounted for primary channel in the interface ref count.
However, it did not reduce this ref count on deallocation
of the primary channel. i.e. during umount.

Fixing this leak here, by dropping this ref count for
primary channel while freeing up the session.

Fixes: fa1d0508bdd4 ("cifs: account for primary channel in the interface list")
Cc: [email protected]
Reported-by: Paulo Alcantara <[email protected]>
Signed-off-by: Shyam Prasad N <[email protected]>
Signed-off-by: Steve French <[email protected]>

gcc-plugins: latent_entropy: Fix typo (args -> argc) in plugin description

Fix the typo in the plugin description comment. Clearly, "argc" should
be used.

Signed-off-by: Konstantin Runov <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Kees Cook <[email protected]>

Remove myself as maintainer of GFS2

I am retiring from Red Hat and will no longer be a maintainer of the
gfs2 file system.

Signed-off-by: Bob Peterson <[email protected]>
Signed-off-by: Linus Torvalds <[email protected]>

ALSA: hda/realtek: Enable Mute LED on HP 255 G10

HP 255 G10 has a mute LED that can be made to work using quirk
ALC236_FIXUP_HP_MUTE_LED_COEFBIT2.
Enable already existing quirk - at correct line to keep order

Signed-off-by: Matus Malych <[email protected]>
Cc: <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Takashi Iwai <[email protected]>

Merge branch '6.7/scsi-staging' into 6.7/scsi-fixes

Pull in queued fixes for 6.7

Signed-off-by: Martin K. Petersen <[email protected]>

netfilter: nf_tables: split async and sync catchall in two functions

list_for_each_entry_safe() does not work for the async case which runs
under RCU, therefore, split GC logic for catchall in two functions
instead, one for each of the sync and async GC variants.

The catchall sync GC variant never sees a _DEAD bit set on ever, thus,
this handling is removed in such case, moreover, allocate GC sync batch
via GFP_KERNEL.

Fixes: 93995bf4af2c ("netfilter: nf_tables: remove catchall element in GC sync path")
Reported-by: Florian Westphal <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>

netfilter: ipset: fix race condition between swap/destroy and kernel side add/del/test

Linkui Xiao reported that there's a race condition when ipset swap and destroy is
called, which can lead to crash in add/del/test element operations. Swap then
destroy are usual operations to replace a set with another one in a production
system. The issue can in some cases be reproduced with the script:

ipset create hash_ip1 hash:net family inet hashsize 1024 maxelem 1048576
ipset add hash_ip1 172.20.0.0/16
ipset add hash_ip1 192.168.0.0/16
iptables -A INPUT -m set --match-set hash_ip1 src -j ACCEPT
while [ 1 ]
do
# ... Ongoing traffic...
        ipset create hash_ip2 hash:net family inet hashsize 1024 maxelem 1048576
        ipset add hash_ip2 172.20.0.0/16
        ipset swap hash_ip1 hash_ip2
        ipset destroy hash_ip2
        sleep 0.05
done

In the race case the possible order of the operations are

CPU0 CPU1
ip_set_test
ipset swap hash_ip1 hash_ip2
ipset destroy hash_ip2
hash_net_kadt

Swap replaces hash_ip1 with hash_ip2 and then destroy removes hash_ip2 which
is the original hash_ip1. ip_set_test was called on hash_ip1 and because destroy
removed it, hash_net_kadt crashes.

The fix is to force ip_set_swap() to wait for all readers to finish accessing the
old set pointers by calling synchronize_rcu().

The first version of the patch was written by Linkui Xiao <[email protected]>.

v2: synchronize_rcu() is moved into ip_set_swap() in order not to burden
    ip_set_destroy() unnecessarily when all sets are destroyed.
v3: Florian Westphal pointed out that all netfilter hooks run with rcu_read_lock() held
    and em_ipset.c wraps the entire ip_set_test() in rcu read lock/unlock pair.
    So there's no need to extend the rcu read locked area in ipset itself.

Closes: https://lore.kernel.org/all/[email protected]/
Reported by: Linkui Xiao <[email protected]>
Signed-off-by: Jozsef Kadlecsik <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>

netfilter: nf_tables: bogus ENOENT when destroying element which does not exist

destroy element command bogusly reports ENOENT in case a set element
does not exist. ENOENT errors are skipped, however, err is still set
and propagated to userspace.

 # nft destroy element ip raw BLACKLIST { 1.2.3.4 }
 Error: Could not process rule: No such file or directory
 destroy element ip raw BLACKLIST { 1.2.3.4 }
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Fixes: f80a612dd77c ("netfilter: nf_tables: add support to destroy operation")
Signed-off-by: Pablo Neira Ayuso <[email protected]>

netfilter: nf_tables: fix pointer math issue in nft_byteorder_eval()

The problem is in nft_byteorder_eval() where we are iterating through a
loop and writing to dst[0], dst[1], dst[2] and so on...  On each
iteration we are writing 8 bytes.  But dst[] is an array of u32 so each
element only has space for 4 bytes.  That means that every iteration
overwrites part of the previous element.

I spotted this bug while reviewing commit caf3ef7468f7 ("netfilter:
nf_tables: prevent OOB access in nft_byteorder_eval") which is a related
issue.  I think that the reason we have not detected this bug in testing
is that most of time we only write one element.

Fixes: ce1e7989d989 ("netfilter: nft_byteorder: provide 64bit le/be conversion")
Signed-off-by: Dan Carpenter <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>

netfilter: nf_conntrack_bridge: initialize err to 0

K2CI reported a problem:

consume_skb(skb);
return err;
[nf_br_ip_fragment() error]  uninitialized symbol 'err'.

err is not initialized, because returning 0 is expected, initialize err
to 0.

Fixes: 3c171f496ef5 ("netfilter: bridge: add connection tracking system")
Reported-by: k2ci <[email protected]>
Signed-off-by: Linkui Xiao <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>

netfilter: nft_set_rbtree: Remove unused variable nft_net

The code that uses nft_net has been removed, and the nft_pernet function
is merely obtaining a reference to shared data through the net pointer.
The content of the net pointer is not modified or changed, so both of
them should be removed.

silence the warning:
net/netfilter/nft_set_rbtree.c:627:26: warning: variable ‘nft_net’ set but not used

Reported-by: Abaci Robot <[email protected]>
Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=7103
Signed-off-by: Yang Li <[email protected]>
Reviewed-by: Simon Horman <[email protected]>
Signed-off-by: Pablo Neira Ayuso <[email protected]>

af_unix: fix use-after-free in unix_stream_read_actor()

syzbot reported the following crash [1]

After releasing unix socket lock, u->oob_skb can be changed
by another thread. We must temporarily increase skb refcount
to make sure this other thread will not free the skb under us.

[1]

BUG: KASAN: slab-use-after-free in unix_stream_read_actor+0xa7/0xc0 net/unix/af_unix.c:2866
Read of size 4 at addr ffff88801f3b9cc4 by task syz-executor107/5297

CPU: 1 PID: 5297 Comm: syz-executor107 Not tainted 6.6.0-syzkaller-15910-gb8e3a87a627b #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 10/09/2023
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0xd9/0x1b0 lib/dump_stack.c:106
print_address_description mm/kasan/report.c:364 [inline]
print_report+0xc4/0x620 mm/kasan/report.c:475
kasan_report+0xda/0x110 mm/kasan/report.c:588
unix_stream_read_actor+0xa7/0xc0 net/unix/af_unix.c:2866
unix_stream_recv_urg net/unix/af_unix.c:2587 [inline]
unix_stream_read_generic+0x19a5/0x2480 net/unix/af_unix.c:2666
unix_stream_recvmsg+0x189/0x1b0 net/unix/af_unix.c:2903
sock_recvmsg_nosec net/socket.c:1044 [inline]
sock_recvmsg+0xe2/0x170 net/socket.c:1066
____sys_recvmsg+0x21f/0x5c0 net/socket.c:2803
___sys_recvmsg+0x115/0x1a0 net/socket.c:2845
__sys_recvmsg+0x114/0x1e0 net/socket.c:2875
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x3f/0x110 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0x6b
RIP: 0033:0x7fc67492c559
Code: 28 00 00 00 75 05 48 83 c4 28 c3 e8 51 18 00 00 90 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b0 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007fc6748ab228 EFLAGS: 00000246 ORIG_RAX: 000000000000002f
RAX: ffffffffffffffda RBX: 000000000000001c RCX: 00007fc67492c559
RDX: 0000000040010083 RSI: 0000000020000140 RDI: 0000000000000004
RBP: 00007fc6749b6348 R08: 00007fc6748ab6c0 R09: 00007fc6748ab6c0
R10: 0000000000000000 R11: 0000000000000246 R12: 00007fc6749b6340
R13: 00007fc6749b634c R14: 00007ffe9fac52a0 R15: 00007ffe9fac5388
</TASK>

Allocated by task 5295:
kasan_save_stack+0x33/0x50 mm/kasan/common.c:45
kasan_set_track+0x25/0x30 mm/kasan/common.c:52
__kasan_slab_alloc+0x81/0x90 mm/kasan/common.c:328
kasan_slab_alloc include/linux/kasan.h:188 [inline]
slab_post_alloc_hook mm/slab.h:763 [inline]
slab_alloc_node mm/slub.c:3478 [inline]
kmem_cache_alloc_node+0x180/0x3c0 mm/slub.c:3523
__alloc_skb+0x287/0x330 net/core/skbuff.c:641
alloc_skb include/linux/skbuff.h:1286 [inline]
alloc_skb_with_frags+0xe4/0x710 net/core/skbuff.c:6331
sock_alloc_send_pskb+0x7e4/0x970 net/core/sock.c:2780
sock_alloc_send_skb include/net/sock.h:1884 [inline]
queue_oob net/unix/af_unix.c:2147 [inline]
unix_stream_sendmsg+0xb5f/0x10a0 net/unix/af_unix.c:2301
sock_sendmsg_nosec net/socket.c:730 [inline]
__sock_sendmsg+0xd5/0x180 net/socket.c:745
____sys_sendmsg+0x6ac/0x940 net/socket.c:2584
___sys_sendmsg+0x135/0x1d0 net/socket.c:2638
__sys_sendmsg+0x117/0x1e0 net/socket.c:2667
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x3f/0x110 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0x6b

Freed by task 5295:
kasan_save_stack+0x33/0x50 mm/kasan/common.c:45
kasan_set_track+0x25/0x30 mm/kasan/common.c:52
kasan_save_free_info+0x2b/0x40 mm/kasan/generic.c:522
____kasan_slab_free mm/kasan/common.c:236 [inline]
____kasan_slab_free+0x15b/0x1b0 mm/kasan/common.c:200
kasan_slab_free include/linux/kasan.h:164 [inline]
slab_free_hook mm/slub.c:1800 [inline]
slab_free_freelist_hook+0x114/0x1e0 mm/slub.c:1826
slab_free mm/slub.c:3809 [inline]
kmem_cache_free+0xf8/0x340 mm/slub.c:3831
kfree_skbmem+0xef/0x1b0 net/core/skbuff.c:1015
__kfree_skb net/core/skbuff.c:1073 [inline]
consume_skb net/core/skbuff.c:1288 [inline]
consume_skb+0xdf/0x170 net/core/skbuff.c:1282
queue_oob net/unix/af_unix.c:2178 [inline]
unix_stream_sendmsg+0xd49/0x10a0 net/unix/af_unix.c:2301
sock_sendmsg_nosec net/socket.c:730 [inline]
__sock_sendmsg+0xd5/0x180 net/socket.c:745
____sys_sendmsg+0x6ac/0x940 net/socket.c:2584
___sys_sendmsg+0x135/0x1d0 net/socket.c:2638
__sys_sendmsg+0x117/0x1e0 net/socket.c:2667
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x3f/0x110 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0x6b

The buggy address belongs to the object at ffff88801f3b9c80
which belongs to the cache skbuff_head_cache of size 240
The buggy address is located 68 bytes inside of
freed 240-byte region [ffff88801f3b9c80, ffff88801f3b9d70)

The buggy address belongs to the physical page:
page:ffffea00007cee40 refcount:1 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x1f3b9
flags: 0xfff00000000800(slab|node=0|zone=1|lastcpupid=0x7ff)
page_type: 0xffffffff()
raw: 00fff00000000800 ffff888142a60640 dead000000000122 0000000000000000
raw: 0000000000000000 00000000000c000c 00000001ffffffff 0000000000000000
page dumped because: kasan: bad access detected
page_owner tracks the page as allocated
page last allocated via order 0, migratetype Unmovable, gfp_mask 0x12cc0(GFP_KERNEL|__GFP_NOWARN|__GFP_NORETRY), pid 5299, tgid 5283 (syz-executor107), ts 103803840339, free_ts 103600093431
set_page_owner include/linux/page_owner.h:31 [inline]
post_alloc_hook+0x2cf/0x340 mm/page_alloc.c:1537
prep_new_page mm/page_alloc.c:1544 [inline]
get_page_from_freelist+0xa25/0x36c0 mm/page_alloc.c:3312
__alloc_pages+0x1d0/0x4a0 mm/page_alloc.c:4568
alloc_pages_mpol+0x258/0x5f0 mm/mempolicy.c:2133
alloc_slab_page mm/slub.c:1870 [inline]
allocate_slab+0x251/0x380 mm/slub.c:2017
new_slab mm/slub.c:2070 [inline]
___slab_alloc+0x8c7/0x1580 mm/slub.c:3223
__slab_alloc.constprop.0+0x56/0xa0 mm/slub.c:3322
__slab_alloc_node mm/slub.c:3375 [inline]
slab_alloc_node mm/slub.c:3468 [inline]
kmem_cache_alloc_node+0x132/0x3c0 mm/slub.c:3523
__alloc_skb+0x287/0x330 net/core/skbuff.c:641
alloc_skb include/linux/skbuff.h:1286 [inline]
alloc_skb_with_frags+0xe4/0x710 net/core/skbuff.c:6331
sock_alloc_send_pskb+0x7e4/0x970 net/core/sock.c:2780
sock_alloc_send_skb include/net/sock.h:1884 [inline]
queue_oob net/unix/af_unix.c:2147 [inline]
unix_stream_sendmsg+0xb5f/0x10a0 net/unix/af_unix.c:2301
sock_sendmsg_nosec net/socket.c:730 [inline]
__sock_sendmsg+0xd5/0x180 net/socket.c:745
____sys_sendmsg+0x6ac/0x940 net/socket.c:2584
___sys_sendmsg+0x135/0x1d0 net/socket.c:2638
__sys_sendmsg+0x117/0x1e0 net/socket.c:2667
page last free stack trace:
reset_page_owner include/linux/page_owner.h:24 [inline]
free_pages_prepare mm/page_alloc.c:1137 [inline]
free_unref_page_prepare+0x4f8/0xa90 mm/page_alloc.c:2347
free_unref_page+0x33/0x3b0 mm/page_alloc.c:2487
__unfreeze_partials+0x21d/0x240 mm/slub.c:2655
qlink_free mm/kasan/quarantine.c:168 [inline]
qlist_free_all+0x6a/0x170 mm/kasan/quarantine.c:187
kasan_quarantine_reduce+0x18e/0x1d0 mm/kasan/quarantine.c:294
__kasan_slab_alloc+0x65/0x90 mm/kasan/common.c:305
kasan_slab_alloc include/linux/kasan.h:188 [inline]
slab_post_alloc_hook mm/slab.h:763 [inline]
slab_alloc_node mm/slub.c:3478 [inline]
slab_alloc mm/slub.c:3486 [inline]
__kmem_cache_alloc_lru mm/slub.c:3493 [inline]
kmem_cache_alloc+0x15d/0x380 mm/slub.c:3502
vm_area_dup+0x21/0x2f0 kernel/fork.c:500
__split_vma+0x17d/0x1070 mm/mmap.c:2365
split_vma mm/mmap.c:2437 [inline]
vma_modify+0x25d/0x450 mm/mmap.c:2472
vma_modify_flags include/linux/mm.h:3271 [inline]
mprotect_fixup+0x228/0xc80 mm/mprotect.c:635
do_mprotect_pkey+0x852/0xd60 mm/mprotect.c:809
__do_sys_mprotect mm/mprotect.c:830 [inline]
__se_sys_mprotect mm/mprotect.c:827 [inline]
__x64_sys_mprotect+0x78/0xb0 mm/mprotect.c:827
do_syscall_x64 arch/x86/entry/common.c:51 [inline]
do_syscall_64+0x3f/0x110 arch/x86/entry/common.c:82
entry_SYSCALL_64_after_hwframe+0x63/0x6b

Memory state around the buggy address:
ffff88801f3b9b80: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
ffff88801f3b9c00: fb fb fb fb fb fb fc fc fc fc fc fc fc fc fc fc
>ffff88801f3b9c80: fa fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
^
ffff88801f3b9d00: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fc fc
ffff88801f3b9d80: fc fc fc fc fc fc fc fc fa fb fb fb fb fb fb fb

Fixes: 876c14ad014d ("af_unix: fix holding spinlock in oob handling")
Reported-and-tested-by: [email protected]
Signed-off-by: Eric Dumazet <[email protected]>
Cc: Rao Shoaib <[email protected]>
Reviewed-by: Rao shoaib <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Paolo Abeni <[email protected]>

xen/events: remove some simple helpers from events_base.c

The helper functions type_from_irq() and cpu_from_irq() are just one
line functions used only internally.

Open code them where needed. At the same time modify and rename
get_evtchn_to_irq() to return a struct irq_info instead of the IRQ
number.

Signed-off-by: Juergen Gross <[email protected]>
Reviewed-by: Oleksandr Tyshchenko <[email protected]>
Signed-off-by: Juergen Gross <[email protected]>

xen/events: reduce externally visible helper functions

get_evtchn_to_irq() has only one external user while irq_from_evtchn()
provides the same functionality and is exported for a wider user base.
Modify the only external user of get_evtchn_to_irq() to use
irq_from_evtchn() instead and make get_evtchn_to_irq() static.

evtchn_from_irq() and irq_from_virq() have a single external user and
can easily be combined to a new helper irq_evtchn_from_virq() allowing
to drop irq_from_virq() and to make evtchn_from_irq() static.

Signed-off-by: Juergen Gross <[email protected]>
Reviewed-by: Oleksandr Tyshchenko <[email protected]>
Signed-off-by: Juergen Gross <[email protected]>

ovl: fix memory leak in ovl_parse_param()

On failure to parse parameters in ovl_parse_param_lowerdir(), it is
necessary to update ctx->nr with the correct nr before using
ovl_reset_lowerdirs() to release l->name.

Reported-and-tested-by: [email protected]
Fixes: c835110b588a ("ovl: remove unused code in lowerdir param parsing")
Co-authored-by: Edward Adam Davis <[email protected]>
Signed-off-by: Amir Goldstein <[email protected]>

ovl: fix misformatted comment

Remove misleading /** prefix from a regular comment.

Reported-by: kernel test robot <[email protected]>
Closes: https://lore.kernel.org/oe-kbuild-all/[email protected]/
Signed-off-by: Amir Goldstein <[email protected]>

Merge branch 'r8169-fix-dash-devices-network-lost-issue'

ChunHao Lin says:

====================
r8169: fix DASH devices network lost issue

This series are used to fix network lost issue on systems that support
DASH. It has been tested on rtl8168ep and rtl8168fp.
====================

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

r8169: fix network lost after resume on DASH systems

Device that support DASH may be reseted or powered off during suspend.
So driver needs to handle DASH during system suspend and resume. Or
DASH firmware will influence device behavior and causes network lost.

Fixes: b646d90053f8 ("r8169: magic.")
Cc: [email protected]
Reviewed-by: Heiner Kallweit <[email protected]>
Signed-off-by: ChunHao Lin <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

r8169: add handling DASH when DASH is disabled

For devices that support DASH, even DASH is disabled, there may still
exist a default firmware that will influence device behavior.
So driver needs to handle DASH for devices that support DASH, no
matter the DASH status is.

This patch also prepares for "fix network lost after resume on DASH
systems".

Fixes: ee7a1beb9759 ("r8169:call "rtl8168_driver_start" "rtl8168_driver_stop" only when hardware dash function is enabled")
Cc: [email protected]
Signed-off-by: ChunHao Lin <[email protected]>
Reviewed-by: Heiner Kallweit <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Merge branch 'fix-large-frames-in-the-gemini-ethernet-driver'

Linus Walleij says:

====================
Fix large frames in the Gemini ethernet driver

This is the result of a bug hunt for a problem with the
RTL8366RB DSA switch leading me wrong all over the place.

I am indebted to Vladimir Oltean who as usual pointed
out where the real problem was, many thanks!

Tryig to actually use big ("jumbo") frames on this
hardware uncovered the real bugs. Then I tested it on
the DSA switch and it indeed fixes the issue.

To make sure it also works fine with big frames on
non-DSA devices I also copied a large video file over
scp to a device with maximum frame size, the data
was transported in large TCP packets ending up in
0x7ff sized frames using software checksumming at
~2.0 MB/s.

If I set down the MTU to the standard 1500 bytes so
that hardware checksumming is used, the scp transfer
of the same file was slightly lower, ~1.8-1.9 MB/s.

Despite this not being the best test it shows that
we can now stress the hardware with large frames
and that software checksum works fine.

v3: https://lore.kernel.org/r/20231107-gemini-largeframe-fix-v3-0-e3803c080b75@linaro.org
v2: https://lore.kernel.org/r/20231105-gemini-largeframe-fix-v2-0-cd3a5aa6c496@linaro.org
v1: https://lore.kernel.org/r/20231104-gemini-largeframe-fix-v1-0-9c5513f22f33@linaro.org
====================

Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

net: ethernet: cortina: Fix MTU max setting

The RX max frame size is over 10000 for the Gemini ethernet,
but the TX max frame size is actually just 2047 (0x7ff after
checking the datasheet). Reflect this in what we offer to Linux,
cap the MTU at the TX max frame minus ethernet headers.

We delete the code disabling the hardware checksum for large
MTUs as netdev->mtu can no longer be larger than
netdev->max_mtu meaning the if()-clause in gmac_fix_features()
is never true.

Fixes: 4d5ae32f5e1e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: Linus Walleij <[email protected]>
Reviewed-by: Vladimir Oltean <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

net: ethernet: cortina: Handle large frames

The Gemini ethernet controller provides hardware checksumming
for frames up to 1514 bytes including ethernet headers but not
FCS.

If we start sending bigger frames (after first bumping up the MTU
on both interfaces sending and receiving the frames), truncated
packets start to appear on the target such as in this tcpdump
resulting from ping -s 1474:

23:34:17.241983 14:d6:4d:a8:3c:4f (oui Unknown) > bc:ae:c5:6b:a8:3d (oui Unknown),
ethertype IPv4 (0x0800), length 1514: truncated-ip - 2 bytes missing!
(tos 0x0, ttl 64, id 32653, offset 0, flags [DF], proto ICMP (1), length 1502)
OpenWrt.lan > Fecusia: ICMP echo request, id 1672, seq 50, length 1482

If we bypass the hardware checksumming and provide a software
fallback, everything starts working fine up to the max TX MTU
of 2047 bytes, for example ping -s2000 192.168.1.2:

00:44:29.587598 bc:ae:c5:6b:a8:3d (oui Unknown) > 14:d6:4d:a8:3c:4f (oui Unknown),
ethertype IPv4 (0x0800), length 2042:
(tos 0x0, ttl 64, id 51828, offset 0, flags [none], proto ICMP (1), length 2028)
Fecusia > OpenWrt.lan: ICMP echo reply, id 1683, seq 4, length 2008

The bit enabling to bypass hardware checksum (or any of the
"TSS" bits) are undocumented in the hardware reference manual.
The entire hardware checksum unit appears undocumented. The
conclusion that we need to use the "bypass" bit was found by
trial-and-error.

Since no hardware checksum will happen, we slot in a software
checksum fallback.

Check for the condition where we need to compute checksum on the
skb with either hardware or software using == CHECKSUM_PARTIAL instead
of != CHECKSUM_NONE which is an incomplete check according to
<linux/skbuff.h>.

On the D-Link DIR-685 router this fixes a bug on the conduit
interface to the RTL8366RB DSA switch: as the switch needs to add
space for its tag it increases the MTU on the conduit interface
to 1504 and that means that when the router sends packages
of 1500 bytes these get an extra 4 bytes of DSA tag and the
transfer fails because of the erroneous hardware checksumming,
affecting such basic functionality as the LuCI web interface.

Fixes: 4d5ae32f5e1e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Signed-off-by: Linus Walleij <[email protected]>
Reviewed-by: Vladimir Oltean <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

net: ethernet: cortina: Fix max RX frame define

Enumerator 3 is 1548 bytes according to the datasheet.
Not 1542.

Fixes: 4d5ae32f5e1e ("net: ethernet: Add a driver for Gemini gigabit ethernet")
Reviewed-by: Andrew Lunn <[email protected]>
Signed-off-by: Linus Walleij <[email protected]>
Reviewed-by: Vladimir Oltean <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

bonding: stop the device in bond_setup_by_slave()

Commit 9eed321cde22 ("net: lapbether: only support ethernet devices")
has been able to keep syzbot away from net/lapb, until today.

In the following splat [1], the issue is that a lapbether device has
been created on a bonding device without members. Then adding a non
ARPHRD_ETHER member forced the bonding master to change its type.

The fix is to make sure we call dev_close() in bond_setup_by_slave()
so that the potential linked lapbether devices (or any other devices
having assumptions on the physical device) are removed.

A similar bug has been addressed in commit 40baec225765
("bonding: fix panic on non-ARPHRD_ETHER enslave failure")

[1]
skbuff: skb_under_panic: text:ffff800089508810 len:44 put:40 head:ffff0000c78e7c00 data:ffff0000c78e7bea tail:0x16 end:0x140 dev:bond0
kernel BUG at net/core/skbuff.c:192 !
Internal error: Oops - BUG: 00000000f2000800 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 PID: 6007 Comm: syz-executor383 Not tainted 6.6.0-rc3-syzkaller-gbf6547d8715b #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/04/2023
pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
pc : skb_panic net/core/skbuff.c:188 [inline]
pc : skb_under_panic+0x13c/0x140 net/core/skbuff.c:202
lr : skb_panic net/core/skbuff.c:188 [inline]
lr : skb_under_panic+0x13c/0x140 net/core/skbuff.c:202
sp : ffff800096a06aa0
x29: ffff800096a06ab0 x28: ffff800096a06ba0 x27: dfff800000000000
x26: ffff0000ce9b9b50 x25: 0000000000000016 x24: ffff0000c78e7bea
x23: ffff0000c78e7c00 x22: 000000000000002c x21: 0000000000000140
x20: 0000000000000028 x19: ffff800089508810 x18: ffff800096a06100
x17: 0000000000000000 x16: ffff80008a629a3c x15: 0000000000000001
x14: 1fffe00036837a32 x13: 0000000000000000 x12: 0000000000000000
x11: 0000000000000201 x10: 0000000000000000 x9 : cb50b496c519aa00
x8 : cb50b496c519aa00 x7 : 0000000000000001 x6 : 0000000000000001
x5 : ffff800096a063b8 x4 : ffff80008e280f80 x3 : ffff8000805ad11c
x2 : 0000000000000001 x1 : 0000000100000201 x0 : 0000000000000086
Call trace:
skb_panic net/core/skbuff.c:188 [inline]
skb_under_panic+0x13c/0x140 net/core/skbuff.c:202
skb_push+0xf0/0x108 net/core/skbuff.c:2446
ip6gre_header+0xbc/0x738 net/ipv6/ip6_gre.c:1384
dev_hard_header include/linux/netdevice.h:3136 [inline]
lapbeth_data_transmit+0x1c4/0x298 drivers/net/wan/lapbether.c:257
lapb_data_transmit+0x8c/0xb0 net/lapb/lapb_iface.c:447
lapb_transmit_buffer+0x178/0x204 net/lapb/lapb_out.c:149
lapb_send_control+0x220/0x320 net/lapb/lapb_subr.c:251
__lapb_disconnect_request+0x9c/0x17c net/lapb/lapb_iface.c:326
lapb_device_event+0x288/0x4e0 net/lapb/lapb_iface.c:492
notifier_call_chain+0x1a4/0x510 kernel/notifier.c:93
raw_notifier_call_chain+0x3c/0x50 kernel/notifier.c:461
call_netdevice_notifiers_info net/core/dev.c:1970 [inline]
call_netdevice_notifiers_extack net/core/dev.c:2008 [inline]
call_netdevice_notifiers net/core/dev.c:2022 [inline]
__dev_close_many+0x1b8/0x3c4 net/core/dev.c:1508
dev_close_many+0x1e0/0x470 net/core/dev.c:1559
dev_close+0x174/0x250 net/core/dev.c:1585
lapbeth_device_event+0x2e4/0x958 drivers/net/wan/lapbether.c:466
notifier_call_chain+0x1a4/0x510 kernel/notifier.c:93
raw_notifier_call_chain+0x3c/0x50 kernel/notifier.c:461
call_netdevice_notifiers_info net/core/dev.c:1970 [inline]
call_netdevice_notifiers_extack net/core/dev.c:2008 [inline]
call_netdevice_notifiers net/core/dev.c:2022 [inline]
__dev_close_many+0x1b8/0x3c4 net/core/dev.c:1508
dev_close_many+0x1e0/0x470 net/core/dev.c:1559
dev_close+0x174/0x250 net/core/dev.c:1585
bond_enslave+0x2298/0x30cc drivers/net/bonding/bond_main.c:2332
bond_do_ioctl+0x268/0xc64 drivers/net/bonding/bond_main.c:4539
dev_ifsioc+0x754/0x9ac
dev_ioctl+0x4d8/0xd34 net/core/dev_ioctl.c:786
sock_do_ioctl+0x1d4/0x2d0 net/socket.c:1217
sock_ioctl+0x4e8/0x834 net/socket.c:1322
vfs_ioctl fs/ioctl.c:51 [inline]
__do_sys_ioctl fs/ioctl.c:871 [inline]
__se_sys_ioctl fs/ioctl.c:857 [inline]
__arm64_sys_ioctl+0x14c/0x1c8 fs/ioctl.c:857
__invoke_syscall arch/arm64/kernel/syscall.c:37 [inline]
invoke_syscall+0x98/0x2b8 arch/arm64/kernel/syscall.c:51
el0_svc_common+0x130/0x23c arch/arm64/kernel/syscall.c:136
do_el0_svc+0x48/0x58 arch/arm64/kernel/syscall.c:155
el0_svc+0x58/0x16c arch/arm64/kernel/entry-common.c:678
el0t_64_sync_handler+0x84/0xfc arch/arm64/kernel/entry-common.c:696
el0t_64_sync+0x190/0x194 arch/arm64/kernel/entry.S:591
Code: aa1803e6 aa1903e7 a90023f5 94785b8b (d4210000)

Fixes: 872254dd6b1f ("net/bonding: Enable bonding to enslave non ARPHRD_ETHER")
Reported-by: syzbot <[email protected]>
Signed-off-by: Eric Dumazet <[email protected]>
Acked-by: Jay Vosburgh <[email protected]>
Reviewed-by: Hangbin Liu <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

ptp: annotate data-race around q->head and q->tail

As I was working on a syzbot report, I found that KCSAN would
probably complain that reading q->head or q->tail without
barriers could lead to invalid results.

Add corresponding READ_ONCE() and WRITE_ONCE() to avoid
load-store tearing.

Fixes: d94ba80ebbea ("ptp: Added a brand new class driver for ptp clocks.")
Signed-off-by: Eric Dumazet <[email protected]>
Acked-by: Richard Cochran <[email protected]>
Link: https://lore.kernel.org/r/[email protected]
Signed-off-by: Jakub Kicinski <[email protected]>

Revert "ptp: Fixes a null pointer dereference in ptp_ioctl"

This reverts commit 8a4f030dbced6fc255cbe67b2d0a129947e18493.

Richard says:

  The test itself is harmless, but keeping it will make people think,
  "oh this pointer can be invalid."

  In fact the core stack ensures that ioctl() can't be invoked after
  release(), otherwise Bad Stuff happens.

Fixes: 8a4f030dbced ("ptp: Fixes a null pointer dereference in ptp_ioctl")
Link: https://lore.kernel.org/all/[email protected]/
Signed-off-by: Jakub Kicinski <[email protected]>

bcachefs: Check for nonce offset inconsistency in data_update path

We've rarely been seeing a nonce offset inconsistency that doesn't show
up in tests: this adds some extra verification code to the data update
path that prints out more relevant info when it occurs.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Make sure to drop/retake btree locks before reclaim

We really don't want to be invoking memory reclaim with btree locks
held: even aside from (solvable, but tricky) recursion issues, it can
cause painful to diagnose performance edge cases.

This fixes a recently reported issue in btree_key_can_insert_cached().

Signed-off-by: Kent Overstreet <[email protected]>
Reported-by: Mateusz Guzik <[email protected]>
Fixes: https://lore.kernel.org/linux-bcachefs/CAGudoHEsb_hGRMeWeXh+UF6po0qQuuq_NKSEo+s1sEb6bDLjpA@mail.gmail.com/T/

bcachefs: btree_trans->write_locked

As prep work for the next patch to fix a key cache reclaim issue, we
need to start tracking whether we're currently holding write locks - so
that we can release and retake the before calling into memory reclaim.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Run btree key cache shrinker less aggressively

The btree key cache maintains lists of items that have been freed, but
can't yet be reclaimed because a bch2_trans_relock() call might find
them - we're waiting for SRCU readers to release.

Previously, we wouldn't count these items against the number we're
attempting to scan for, which would mean we'd evict more live key cache
entries - doing quite a bit of potentially unecessary work.

With recent work to make sure we don't hold SRCU locks for too long, it
should be safe to count all the items on the freelists against number to
scan - even if we can't reclaim them yet, we will be able to soon.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Split out btree_key_cache_types.h

More consistent organization.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Guard against insufficient devices to create stripes

We can't create stripes if we don't have enough devices - this
manifested as an integer underflow bug later.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix null ptr deref in bch2_backpointer_get_node()

bch2_btree_iter_peek_node() can return a NULL ptr (when the tree is
shorter than the search depth); handle this with an early return.

Signed-off-by: Kent Overstreet <[email protected]>
Reported-by: Dan Carpenter <[email protected]>
Fixes: https://lore.kernel.org/linux-bcachefs/[email protected]/T/
Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Fix multiple -Warray-bounds warnings

Transform zero-length array `entries` into a proper flexible-array
member in `struct journal_seq_blacklist_table`; and fix the following
-Warray-bounds warnings:

fs/bcachefs/journal_seq_blacklist.c:148:26: warning: array subscript idx is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]
fs/bcachefs/journal_seq_blacklist.c:150:30: warning: array subscript idx is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]
fs/bcachefs/journal_seq_blacklist.c:154:27: warning: array subscript idx is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]
fs/bcachefs/journal_seq_blacklist.c:176:27: warning: array subscript i is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]
fs/bcachefs/journal_seq_blacklist.c:177:27: warning: array subscript i is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]
fs/bcachefs/journal_seq_blacklist.c:297:34: warning: array subscript i is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]
fs/bcachefs/journal_seq_blacklist.c:298:34: warning: array subscript i is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]
fs/bcachefs/journal_seq_blacklist.c:300:31: warning: array subscript i is outside array bounds of 'struct journal_seq_blacklist_table_entry[0]' [-Warray-bounds=]

This results in no differences in binary output.

This helps with the ongoing efforts to globally enable -Warray-bounds.

Signed-off-by: Gustavo A. R. Silva <[email protected]>
Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Use DECLARE_FLEX_ARRAY() helper and fix multiple -Warray-bounds warnings

Transform zero-length array `s` into a proper flexible-array
member in `struct snapshot_table` via the DECLARE_FLEX_ARRAY()
helper; and fix tons of the following -Warray-bounds warnings:

fs/bcachefs/snapshot.h:36:21: warning: array subscript <unknown> is outside array bounds of 'struct snapshot_t[0]' [-Warray-bounds=]
fs/bcachefs/snapshot.h:36:21: warning: array subscript <unknown> is outside array bounds of 'struct snapshot_t[0]' [-Warray-bounds=]
fs/bcachefs/snapshot.c:135:70: warning: array subscript <unknown> is outside array bounds of 'struct snapshot_t[0]' [-Warray-bounds=]
fs/bcachefs/snapshot.h:36:21: warning: array subscript <unknown> is outside array bounds of 'struct snapshot_t[0]' [-Warray-bounds=]
fs/bcachefs/snapshot.h:36:21: warning: array subscript <unknown> is outside array bounds of 'struct snapshot_t[0]' [-Warray-bounds=]
fs/bcachefs/snapshot.h:36:21: warning: array subscript <unknown> is outside array bounds of 'struct snapshot_t[0]' [-Warray-bounds=]

This helps with the ongoing efforts to globally enable -Warray-bounds.

Signed-off-by: Gustavo A. R. Silva <[email protected]>
Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: Use correct fgf_t type as function argument

This quiets a sparse complaint.

Signed-off-by: Kent Overstreet <[email protected]>

bcachefs: make bch2_target_to_text_sb static

The bch2_target_to_text_sb are not used outside the file disk_groups.c,
so the modification is defined as static.

fs/bcachefs/disk_groups.c:583:6: warning: no previous prototype for ‘bch2_target_to_text_sb’.

Reported-by: Abaci Robot <[email protected]>
Closes: https://bugzilla.openanolis.cn/show_bug.cgi?id=7144
Signed-off-by: Jiapeng Chong <[email protected]>
Signed-off-by: Kent Overstreet <[email protected]>