maple_tree: fix mas_skip_node() end slot detection

commit 0fa99fdfe1b38da396d0b2d1496a823bcd0ebea0 upstream.

Patch series "Fix mas_skip_node() for mas_empty_area()", v2.

mas_empty_area() was incorrectly returning an error when there was room.
The issue was tracked down to mas_skip_node() using the incorrect
end-of-slot count.  Instead of using the nodes hard limit, the limit of
data should be used.

mas_skip_node() was also setting the min and max to that of the child
node, which was unnecessary.  Within these limits being set, there was
also a bug that corrupted the maple state's max if the offset was set to
the maximum node pivot.  The bug was without consequence unless there was
a sufficient gap in the next child node which would cause an error to be
returned.

This patch set fixes these errors by removing the limit setting from
mas_skip_node() and uses the mas_data_end() for slot limits, and adds
tests for all failures discovered.

This patch (of 2):

mas_skip_node() is used to move the maple state to the node with a higher
limit.  It does this by walking up the tree and increasing the slot count.
Since slot count may not be able to be increased, it may need to walk up
multiple times to find room to walk right to a higher limit node.  The
limit of slots that was being used was the node limit and not the last
location of data in the node.  This would cause the maple state to be
shifted outside actual data and enter an error state, thus returning
-EBUSY.

The result of the incorrect error state means that mas_awalk() would
return an error instead of finding the allocation space.

The fix is to use mas_data_end() in mas_skip_node() to detect the nodes
data end point and continue walking the tree up until it is safe to move
to a node with a higher limit.

The walk up the tree also sets the maple state limits so remove the buggy
code from mas_skip_node().  Setting the limits had the unfortunate side
effect of triggering another bug if the parent node was full and the there
was no suitable gap in the second last child, but room in the next child.

mas_skip_node() may also be passed a maple state in an error state from
mas_anode_descend() when no allocations are available.  Return on such an
error state immediately.

Link: https://lkml.kernel.org/r/20230307180247.2220303-1-Liam.Howlett@oracle.com
Link: https://lkml.kernel.org/r/20230307180247.2220303-2-Liam.Howlett@oracle.com
Fixes: 54a611b60590 ("Maple Tree: add new data structure")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Reported-by: Snild Dolkow <snild@sony.com>
Link: https://lore.kernel.org/linux-mm/cb8dc31a-fef2-1d09-f133-e9f7b9f9e77a@sony.com/
Tested-by: Snild Dolkow <snild@sony.com>
Cc: Peng Zhang <zhangpeng.00@bytedance.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

test_maple_tree: add more testing for mas_empty_area()

commit 4bd6dded6318dc8e2514d74868c1f8fb38b61a60 upstream.

Test robust filling of an entire area of the tree, then test one beyond.
This is to test the walking back up the tree at the end of nodes and error
condition.  Test inspired by the reproducer code provided by Snild Dolkow.

The last test in the function tests for the case of a corrupted maple
state caused by the incorrect limits set during mas_skip_node().  There
needs to be a gap in the second last child and last child, but the search
must rule out the second last child's gap.  This would avoid correcting
the maple state to the correct max limit and return an error.

Link: https://lkml.kernel.org/r/20230307180247.2220303-3-Liam.Howlett@oracle.com
Cc: Snild Dolkow <snild@sony.com>
Link: https://lore.kernel.org/linux-mm/cb8dc31a-fef2-1d09-f133-e9f7b9f9e77a@sony.com/
Fixes: e15e06a83923 ("lib/test_maple_tree: add testing for maple tree")
Signed-off-by: Liam R. Howlett <Liam.Howlett@oracle.com>
Cc: Peng Zhang <zhangpeng.00@bytedance.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Revert "kasan: drop skip_kasan_poison variable in free_pages_prepare"

commit f446883d12b8bfa486f7c98d403054d61d38c989 upstream.

This reverts commit 487a32ec24be819e747af8c2ab0d5c515508086a.

should_skip_kasan_poison() reads the PG_skip_kasan_poison flag from
page->flags.  However, this line of code in free_pages_prepare():

page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP;

clears most of page->flags, including PG_skip_kasan_poison, before calling
should_skip_kasan_poison(), which meant that it would never return true as
a result of the page flag being set.  Therefore, fix the code to call
should_skip_kasan_poison() before clearing the flags, as we were doing
before the reverted patch.

This fixes a measurable performance regression introduced in the reverted
commit, where munmap() takes longer than intended if HW tags KASAN is
supported and enabled at runtime.  Without this patch, we see a
single-digit percentage performance regression in a particular
mmap()-heavy benchmark when enabling HW tags KASAN, and with the patch,
there is no statistically significant performance impact when enabling HW
tags KASAN.

Link: https://lkml.kernel.org/r/20230310042914.3805818-2-pcc@google.com
Fixes: 487a32ec24be ("kasan: drop skip_kasan_poison variable in free_pages_prepare")
Link: https://linux-review.googlesource.com/id/Ic4f13affeebd20548758438bb9ed9ca40e312b79
Signed-off-by: Peter Collingbourne <pcc@google.com>
Reviewed-by: Andrey Konovalov <andreyknvl@gmail.com>
Cc: Andrey Ryabinin <ryabinin.a.a@gmail.com>
Cc: Catalin Marinas <catalin.marinas@arm.com> [arm64]
Cc: Evgenii Stepanov <eugenis@google.com>
Cc: Vincenzo Frascino <vincenzo.frascino@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: <stable@vger.kernel.org> [6.1]
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring/rsrc: fix null-ptr-deref in io_file_bitmap_get()

commit 02a4d923e4400a36d340ea12d8058f69ebf3a383 upstream.

When fixed files are unregistered, file_alloc_end and alloc_hint
are not cleared. This can later cause a NULL pointer dereference in
io_file_bitmap_get() if auto index selection is enabled via
IORING_FILE_INDEX_ALLOC:

[    6.519129] BUG: kernel NULL pointer dereference, address: 0000000000000000
[...]
[    6.541468] RIP: 0010:_find_next_zero_bit+0x1a/0x70
[...]
[    6.560906] Call Trace:
[    6.561322]  <TASK>
[    6.561672]  io_file_bitmap_get+0x38/0x60
[    6.562281]  io_fixed_fd_install+0x63/0xb0
[    6.562851]  ? __pfx_io_socket+0x10/0x10
[    6.563396]  io_socket+0x93/0xf0
[    6.563855]  ? __pfx_io_socket+0x10/0x10
[    6.564411]  io_issue_sqe+0x5b/0x3d0
[    6.564914]  io_submit_sqes+0x1de/0x650
[    6.565452]  __do_sys_io_uring_enter+0x4fc/0xb20
[    6.566083]  ? __do_sys_io_uring_register+0x11e/0xd80
[    6.566779]  do_syscall_64+0x3c/0x90
[    6.567247]  entry_SYSCALL_64_after_hwframe+0x72/0xdc
[...]

To fix the issue, set file alloc range and alloc_hint to zero after
file tables are freed.

Cc: stable@vger.kernel.org
Fixes: 4278a0deb1f6 ("io_uring: defer alloc_hint update to io_file_bitmap_set()")
Signed-off-by: Savino Dicanosa <sd7.dev@pm.me>
[axboe: add explicit bitmap == NULL check as well]
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

io_uring/net: avoid sending -ECONNABORTED on repeated connection requests

commit 74e2e17ee1f8d8a0928b90434ad7e2df70f8483e upstream.

Since io_uring does nonblocking connect requests, if we do two repeated
ones without having a listener, the second will get -ECONNABORTED rather
than the expected -ECONNREFUSED. Treat -ECONNABORTED like a normal retry
condition if we're nonblocking, if we haven't already seen it.

Cc: stable@vger.kernel.org
Fixes: 3fb1bd688172 ("io_uring/net: handle -EINPROGRESS correct for IORING_OP_CONNECT")
Link: https://github.com/axboe/liburing/issues/828
Reported-by: Hui, Chunyang <sanqian.hcy@antgroup.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

kfence: avoid passing -g for test

commit 2e08ca1802441224f5b7cc6bffbb687f7406de95 upstream.

Nathan reported that when building with GNU as and a version of clang that
defaults to DWARF5:

  $ make -skj"$(nproc)" ARCH=riscv CROSS_COMPILE=riscv64-linux-gnu- \
LLVM=1 LLVM_IAS=0 O=build \
mrproper allmodconfig mm/kfence/kfence_test.o
  /tmp/kfence_test-08a0a0.s: Assembler messages:
  /tmp/kfence_test-08a0a0.s:14627: Error: non-constant .uleb128 is not supported
  /tmp/kfence_test-08a0a0.s:14628: Error: non-constant .uleb128 is not supported
  /tmp/kfence_test-08a0a0.s:14632: Error: non-constant .uleb128 is not supported
  /tmp/kfence_test-08a0a0.s:14633: Error: non-constant .uleb128 is not supported
  /tmp/kfence_test-08a0a0.s:14639: Error: non-constant .uleb128 is not supported
  ...

This is because `-g` defaults to the compiler debug info default.  If the
assembler does not support some of the directives used, the above errors
occur.  To fix, remove the explicit passing of `-g`.

All the test wants is that stack traces print valid function names, and
debug info is not required for that.  (I currently cannot recall why I
added the explicit `-g`.)

Link: https://lkml.kernel.org/r/20230316224705.709984-1-elver@google.com
Fixes: bc8fbc5f305a ("kfence: add test suite")
Signed-off-by: Marco Elver <elver@google.com>
Reported-by: Nathan Chancellor <nathan@kernel.org>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm: kfence: fix using kfence_metadata without initialization in show_object()

commit 1c86a188e03156223a34d09ce290b49bd4dd0403 upstream.

The variable kfence_metadata is initialized in kfence_init_pool(), then,
it is not initialized if kfence is disabled after booting.  In this case,
kfence_metadata will be used (e.g.  ->lock and ->state fields) without
initialization when reading /sys/kernel/debug/kfence/objects.  There will
be a warning if you enable CONFIG_DEBUG_SPINLOCK.  Fix it by creating
debugfs files when necessary.

Link: https://lkml.kernel.org/r/20230315034441.44321-1-songmuchun@bytedance.com
Fixes: 0ce20dd84089 ("mm: add Kernel Electric-Fence infrastructure")
Signed-off-by: Muchun Song <songmuchun@bytedance.com>
Tested-by: Marco Elver <elver@google.com>
Reviewed-by: Marco Elver <elver@google.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Jann Horn <jannh@google.com>
Cc: SeongJae Park <sjpark@amazon.de>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: ucsi_acpi: Increase the command completion timeout

commit 02d210f434249a7edbc160969b75df030dc6934d upstream.

Commit 130a96d698d7 ("usb: typec: ucsi: acpi: Increase command
completion timeout value") increased the timeout from 5 seconds
to 60 seconds due to issues related to alternate mode discovery.

After the alternate mode discovery switch to polled mode
the timeout was reduced, but instead of being set back to
5 seconds it was reduced to 1 second.

This is causing problems when using a Lenovo ThinkPad X1 yoga gen7
connected over Type-C to a LG 27UL850-W (charging DP over Type-C).

When the monitor is already connected at boot the following error
is logged: "PPM init failed (-110)", /sys/class/typec is empty and
on unplugging the NULL pointer deref fixed earlier in this series
happens.

When the monitor is connected after boot the following error
is logged instead: "GET_CONNECTOR_STATUS failed (-110)".

Setting the timeout back to 5 seconds fixes both cases.

Fixes: e08065069fc7 ("usb: typec: ucsi: acpi: Reduce the command completion timeout")
Cc: stable@vger.kernel.org
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20230308154244.722337-4-hdegoede@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: ucsi: Fix NULL pointer deref in ucsi_connector_change()

commit f87fb985452ab2083967103ac00bfd68fb182764 upstream.

When ucsi_init() fails, ucsi->connector is NULL, yet in case of
ucsi_acpi we may still get events which cause the ucs_acpi code to call
ucsi_connector_change(), which then derefs the NULL ucsi->connector
pointer.

Fix this by not setting ucsi->ntfy inside ucsi_init() until ucsi_init()
has succeeded, so that ucsi_connector_change() ignores the events
because UCSI_ENABLE_NTFY_CONNECTOR_CHANGE is not set in the ntfy mask.

Fixes: bdc62f2bae8f ("usb: typec: ucsi: Simplified registration and I/O API")
Link: https://bugzilla.kernel.org/show_bug.cgi?id=217106
Cc: stable@vger.kernel.org
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20230308154244.722337-2-hdegoede@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: dwc3: gadget: Add 1ms delay after end transfer command without IOC

commit d8a2bb4eb75866275b5cf7de2e593ac3449643e2 upstream.

Previously, there was a 100uS delay inserted after issuing an end transfer
command for specific controller revisions.  This was due to the fact that
there was a GUCTL2 bit field which enabled synchronous completion of the
end transfer command once the CMDACT bit was cleared in the DEPCMD
register.  Since this bit does not exist for all controller revisions and
the current implementation heavily relies on utizling the EndTransfer
command completion interrupt, add the delay back in for uses where the
interrupt on completion bit is not set, and increase the duration to 1ms
for the controller to complete the command.

An issue was seen where the USB request buffer was unmapped while the DWC3
controller was still accessing the TRB.  However, it was confirmed that the
end transfer command was successfully submitted. (no end transfer timeout)
In situations, such as dwc3_gadget_soft_disconnect() and
__dwc3_gadget_ep_disable(), the dwc3_remove_request() is utilized, which
will issue the end transfer command, and follow up with
dwc3_gadget_giveback().  At least for the USB ep disable path, it is
required for any pending and started requests to be completed and returned
to the function driver in the same context of the disable call.  Without
the GUCTL2 bit, it is not ensured that the end transfer is completed before
the buffers are unmapped.

Fixes: cf2f8b63f7f1 ("usb: dwc3: gadget: Remove END_TRANSFER delay")
Cc: stable <stable@kernel.org>
Signed-off-by: Wesley Cheng <quic_wcheng@quicinc.com>
Acked-by: Thinh Nguyen <Thinh.Nguyen@synopsys.com>
Link: https://lore.kernel.org/r/20230306200557.29387-1-quic_wcheng@quicinc.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: chipidea: core: fix possible concurrent when switch role

commit 451b15ed138ec15bffbebb58a00ebdd884c3e659 upstream.

The user may call role_store() when driver is handling
ci_handle_id_switch() which is triggerred by otg event or power lost
event. Unfortunately, the controller may go into chaos in this case.
Fix this by protecting it with mutex lock.

Fixes: a932a8041ff9 ("usb: chipidea: core: add sysfs group")
cc: <stable@vger.kernel.org>
Acked-by: Peter Chen <peter.chen@kernel.org>
Signed-off-by: Xu Yang <xu.yang_2@nxp.com>
Link: https://lore.kernel.org/r/20230317061516.2451728-2-xu.yang_2@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: chipdea: core: fix return -EINVAL if request role is the same with current role

commit 3670de80678961eda7fa2220883fc77c16868951 upstream.

It should not return -EINVAL if the request role is the same with current
role, return non-error and without do anything instead.

Fixes: a932a8041ff9 ("usb: chipidea: core: add sysfs group")
cc: <stable@vger.kernel.org>
Acked-by: Peter Chen <peter.chen@kernel.org>
Signed-off-by: Xu Yang <xu.yang_2@nxp.com>
Link: https://lore.kernel.org/r/20230317061516.2451728-1-xu.yang_2@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: cdnsp: changes PCI Device ID to fix conflict with CNDS3 driver

commit 96b96b2a567fb34dd41c87e6cf01f6902ce8cae4 upstream.

Patch changes CDNS_DEVICE_ID in USBSSP PCI Glue driver to remove
the conflict with Cadence USBSS driver.

cc: <stable@vger.kernel.org>
Fixes: 3d82904559f4 ("usb: cdnsp: cdns3 Add main part of Cadence USBSSP DRD Driver")
Signed-off-by: Pawel Laszczak <pawell@cadence.com>
Link: https://lore.kernel.org/r/20230309063048.299378-1-pawell@cadence.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: cdnsp: Fixes issue with redundant Status Stage

commit 5bc38d33a5a1209fd4de65101d1ae8255ea12c6e upstream.

In some cases, driver trees to send Status Stage twice.
The first one from upper layer of gadget usb subsystem and
second time from controller driver.
This patch fixes this issue and remove tricky handling of
SET_INTERFACE from controller driver which is no longer
needed.

cc: <stable@vger.kernel.org>
Fixes: 3d82904559f4 ("usb: cdnsp: cdns3 Add main part of Cadence USBSSP DRD Driver")
Signed-off-by: Pawel Laszczak <pawell@cadence.com>
Link: https://lore.kernel.org/r/20230307111420.376056-1-pawell@cadence.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: cdns3: Fix issue with using incorrect PCI device function

commit 1272fd652a226ccb34e9f47371b6121948048438 upstream.

PCI based platform can have more than two PCI functions.
USBSS PCI Glue driver during initialization should
consider only DRD/HOST/DEVICE PCI functions and
all other should be ignored. This patch adds additional
condition which causes that only DRD and HOST/DEVICE
function will be accepted.

cc: <stable@vger.kernel.org>
Fixes: 7733f6c32e36 ("usb: cdns3: Add Cadence USB3 DRD Driver")
Signed-off-by: Pawel Laszczak <pawell@cadence.com>
Link: https://lore.kernel.org/r/20230308124427.311245-1-pawell@cadence.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: typec: tcpm: fix warning when handle discover_identity message

commit abfc4fa28f0160df61c7149567da4f6494dfb488 upstream.

Since both source and sink device can send discover_identity message in
PD3, kernel may dump below warning:

------------[ cut here ]------------
WARNING: CPU: 0 PID: 169 at drivers/usb/typec/tcpm/tcpm.c:1446 tcpm_queue_vdm+0xe0/0xf0
Modules linked in:
CPU: 0 PID: 169 Comm: 1-0050 Not tainted 6.1.1-00038-g6a3c36cf1da2-dirty #567
Hardware name: NXP i.MX8MPlus EVK board (DT)
pstate: 20000005 (nzCv daif -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
pc : tcpm_queue_vdm+0xe0/0xf0
lr : tcpm_queue_vdm+0x2c/0xf0
sp : ffff80000c19bcd0
x29: ffff80000c19bcd0 x28: 0000000000000001 x27: ffff0000d11c8ab8
x26: ffff0000d11cc000 x25: 0000000000000000 x24: 00000000ff008081
x23: 0000000000000001 x22: 00000000ff00a081 x21: ffff80000c19bdbc
x20: 0000000000000000 x19: ffff0000d11c8080 x18: ffffffffffffffff
x17: 0000000000000000 x16: 0000000000000000 x15: ffff0000d716f580
x14: 0000000000000001 x13: ffff0000d716f507 x12: 0000000000000001
x11: 0000000000000000 x10: 0000000000000020 x9 : 00000000000ee098
x8 : 00000000ffffffff x7 : 000000000000001c x6 : ffff0000d716f580
x5 : 0000000000000000 x4 : 0000000000000000 x3 : 0000000000000000
x2 : ffff80000c19bdbc x1 : 00000000ff00a081 x0 : 0000000000000004
Call trace:
tcpm_queue_vdm+0xe0/0xf0
tcpm_pd_rx_handler+0x340/0x1ab0
kthread_worker_fn+0xcc/0x18c
kthread+0x10c/0x110
ret_from_fork+0x10/0x20
---[ end trace 0000000000000000 ]---

Below sequences may trigger this warning:

tcpm_send_discover_work(work)
  tcpm_send_vdm(port, USB_SID_PD, CMD_DISCOVER_IDENT, NULL, 0);
   tcpm_queue_vdm(port, header, data, count);
    port->vdm_state = VDM_STATE_READY;

vdm_state_machine_work(work);
<-- received discover_identity from partner
 vdm_run_state_machine(port);
  port->vdm_state = VDM_STATE_SEND_MESSAGE;
   mod_vdm_delayed_work(port, x);

tcpm_pd_rx_handler(work);
 tcpm_pd_data_request(port, msg);
  tcpm_handle_vdm_request(port, msg->payload, cnt);
   tcpm_queue_vdm(port, response[0], &response[1], rlen - 1);
--> WARN_ON(port->vdm_state > VDM_STATE_DONE);

For this case, the state machine could still send out discover
identity message later if we skip current discover_identity message.
So we should handle the received message firstly and override the pending
discover_identity message without warning in this case. Then, a delayed
send_discover work will send discover_identity message again.

Fixes: e00943e91678 ("usb: typec: tcpm: PD3.0 sinks can send Discover Identity even in device mode")
cc: <stable@vger.kernel.org>
Signed-off-by: Xu Yang <xu.yang_2@nxp.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20230216031515.4151117-1-xu.yang_2@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: typec: tcpm: fix create duplicate source-capabilities file

commit a826492fc9dfe32afd70fff93955ae8174bbf14b upstream.

The kernel will dump in the below cases:
sysfs: cannot create duplicate filename
'/devices/virtual/usb_power_delivery/pd1/source-capabilities'

1. After soft reset has completed, an Explicit Contract negotiation occurs.
The sink device will receive source capabilitys again. This will cause
a duplicate source-capabilities file be created.
2. Power swap twice on a device that is initailly sink role.

This will unregister existing capabilities when above cases occurs.

Fixes: 8203d26905ee ("usb: typec: tcpm: Register USB Power Delivery Capabilities")
cc: <stable@vger.kernel.org>
Signed-off-by: Xu Yang <xu.yang_2@nxp.com>
Reviewed-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Reviewed-by: Guenter Roeck <linux@roeck-us.net>
Link: https://lore.kernel.org/r/20230215054951.238394-1-xu.yang_2@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

dm thin: fix deadlock when swapping to thin device

commit 9bbf5feecc7eab2c370496c1c161bbfe62084028 upstream.

This is an already known issue that dm-thin volume cannot be used as
swap, otherwise a deadlock may happen when dm-thin internal memory
demand triggers swap I/O on the dm-thin volume itself.

But thanks to commit a666e5c05e7c ("dm: fix deadlock when swapping to
encrypted device"), the limit_swap_bios target flag can also be used
for dm-thin to avoid the recursive I/O when it is used as swap.

Fix is to simply set ti->limit_swap_bios to true in both pool_ctr()
and thin_ctr().

In my test, I create a dm-thin volume /dev/vg/swap and use it as swap
device. Then I run fio on another dm-thin volume /dev/vg/main and use
large --blocksize to trigger swap I/O onto /dev/vg/swap.

The following fio command line is used in my test,
  fio --name recursive-swap-io --lockmem 1 --iodepth 128 \
     --ioengine libaio --filename /dev/vg/main --rw randrw \
    --blocksize 1M --numjobs 32 --time_based --runtime=12h

Without this fix, the whole system can be locked up within 15 seconds.

With this fix, there is no any deadlock or hung task observed after
2 hours of running fio.

Furthermore, if blocksize is changed from 1M to 128M, after around 30
seconds fio has no visible I/O, and the out-of-memory killer message
shows up in kernel message. After around 20 minutes all fio processes
are killed and the whole system is back to being alive.

This is exactly what is expected when recursive I/O happens on dm-thin
volume when it is used as swap.

Depends-on: a666e5c05e7c ("dm: fix deadlock when swapping to encrypted device")
Cc: stable@vger.kernel.org
Signed-off-by: Coly Li <colyli@suse.de>
Acked-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Mike Snitzer <snitzer@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

igb: revert rtnl_lock() that causes deadlock

commit 65f69851e44d71248b952a687e44759a7abb5016 upstream.

The commit 6faee3d4ee8b ("igb: Add lock to avoid data race") adds
rtnl_lock to eliminate a false data race shown below

 (FREE from device detaching)      |   (USE from netdev core)
igb_remove                         |  igb_ndo_get_vf_config
 igb_disable_sriov                 |  vf >= adapter->vfs_allocated_count?
  kfree(adapter->vf_data)          |
  adapter->vfs_allocated_count = 0 |
                                   |    memcpy(... adapter->vf_data[vf]

The above race will never happen and the extra rtnl_lock causes deadlock
below

[  141.420169]  <TASK>
[  141.420672]  __schedule+0x2dd/0x840
[  141.421427]  schedule+0x50/0xc0
[  141.422041]  schedule_preempt_disabled+0x11/0x20
[  141.422678]  __mutex_lock.isra.13+0x431/0x6b0
[  141.423324]  unregister_netdev+0xe/0x20
[  141.423578]  igbvf_remove+0x45/0xe0 [igbvf]
[  141.423791]  pci_device_remove+0x36/0xb0
[  141.423990]  device_release_driver_internal+0xc1/0x160
[  141.424270]  pci_stop_bus_device+0x6d/0x90
[  141.424507]  pci_stop_and_remove_bus_device+0xe/0x20
[  141.424789]  pci_iov_remove_virtfn+0xba/0x120
[  141.425452]  sriov_disable+0x2f/0xf0
[  141.425679]  igb_disable_sriov+0x4e/0x100 [igb]
[  141.426353]  igb_remove+0xa0/0x130 [igb]
[  141.426599]  pci_device_remove+0x36/0xb0
[  141.426796]  device_release_driver_internal+0xc1/0x160
[  141.427060]  driver_detach+0x44/0x90
[  141.427253]  bus_remove_driver+0x55/0xe0
[  141.427477]  pci_unregister_driver+0x2a/0xa0
[  141.428296]  __x64_sys_delete_module+0x141/0x2b0
[  141.429126]  ? mntput_no_expire+0x4a/0x240
[  141.429363]  ? syscall_trace_enter.isra.19+0x126/0x1a0
[  141.429653]  do_syscall_64+0x5b/0x80
[  141.429847]  ? exit_to_user_mode_prepare+0x14d/0x1c0
[  141.430109]  ? syscall_exit_to_user_mode+0x12/0x30
[  141.430849]  ? do_syscall_64+0x67/0x80
[  141.431083]  ? syscall_exit_to_user_mode_prepare+0x183/0x1b0
[  141.431770]  ? syscall_exit_to_user_mode+0x12/0x30
[  141.432482]  ? do_syscall_64+0x67/0x80
[  141.432714]  ? exc_page_fault+0x64/0x140
[  141.432911]  entry_SYSCALL_64_after_hwframe+0x72/0xdc

Since the igb_disable_sriov() will call pci_disable_sriov() before
releasing any resources, the netdev core will synchronize the cleanup to
avoid any races. This patch removes the useless rtnl_(un)lock to guarantee
correctness.

CC: stable@vger.kernel.org
Fixes: 6faee3d4ee8b ("igb: Add lock to avoid data race")
Reported-by: Corinna Vinschen <vinschen@redhat.com>
Link: https://lore.kernel.org/intel-wired-lan/ZAcJvkEPqWeJHO2r@calimero.vinschen.de/
Signed-off-by: Lin Ma <linma@zju.edu.cn>
Tested-by: Corinna Vinschen <vinschen@redhat.com>
Reviewed-by: Jacob Keller <jacob.e.keller@intel.com>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

arm64: dts: imx8mm-nitrogen-r2: fix WM8960 clock name

commit 32f86da7c86b27ebed31c24453a0713f612e43fb upstream.

The WM8960 Linux driver expects the clock to be named "mclk".  Otherwise
the clock will be ignored and not prepared/enabled by the driver.

Fixes: 40ba2eda0a7b ("arm64: dts: imx8mm-nitrogen-r2: add audio")
Cc: <stable@vger.kernel.org>
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

lockd: set file_lock start and end when decoding nlm4 testargs

commit 7ff84910c66c9144cc0de9d9deed9fb84c03aff0 upstream.

Commit 6930bcbfb6ce dropped the setting of the file_lock range when
decoding a nlm_lock off the wire. This causes the client side grant
callback to miss matching blocks and reject the lock, only to rerequest
it 30s later.

Add a helper function to set the file_lock range from the start and end
values that the protocol uses, and have the nlm_lock decoder call that to
set up the file_lock args properly.

Fixes: 6930bcbfb6ce ("lockd: detect and reject lock arguments that overflow")
Reported-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Jeff Layton <jlayton@kernel.org>
Tested-by: Amir Goldstein <amir73il@gmail.com>
Cc: stable@vger.kernel.org #6.0
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fsverity: Remove WQ_UNBOUND from fsverity read workqueue

commit f959325e6ac3f499450088b8d9c626d1177be160 upstream.

WQ_UNBOUND causes significant scheduler latency on ARM64/Android.  This
is problematic for latency sensitive workloads, like I/O
post-processing.

Removing WQ_UNBOUND gives a 96% reduction in fsverity workqueue related
scheduler latency and improves app cold startup times by ~30ms.
WQ_UNBOUND was also removed from the dm-verity workqueue for the same
reason [1].

This code was tested by running Android app startup benchmarks and
measuring how long the fsverity workqueue spent in the runnable state.

Before
Total workqueue scheduler latency: 553800us
After
Total workqueue scheduler latency: 18962us

[1]: https://lore.kernel.org/all/20230202012348.885402-1-nhuck@google.com/

Signed-off-by: Nathan Huckleberry <nhuck@google.com>
Fixes: 8a1d0f9cacc9 ("fs-verity: add data verification hooks for ->readpages()")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20230310193325.620493-1-nhuck@google.com
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

fscrypt: destroy keyring after security_sb_delete()

commit ccb820dc7d2236b1af0d54ae038a27b5b6d5ae5a upstream.

fscrypt_destroy_keyring() must be called after all potentially-encrypted
inodes were evicted; otherwise it cannot safely destroy the keyring.
Since inodes that are in-use by the Landlock LSM don't get evicted until
security_sb_delete(), this means that fscrypt_destroy_keyring() must be
called *after* security_sb_delete().

This fixes a WARN_ON followed by a NULL dereference, only possible if
Landlock was being used on encrypted files.

Fixes: d7e7b9af104c ("fscrypt: stop using keyrings subsystem for fscrypt_master_key")
Cc: stable@vger.kernel.org
Reported-by: syzbot+93e495f6a4f748827c88@syzkaller.appspotmail.com
Link: https://lore.kernel.org/r/00000000000044651705f6ca1e30@google.com
Reviewed-by: Christian Brauner <brauner@kernel.org>
Link: https://lore.kernel.org/r/20230313221231.272498-2-ebiggers@kernel.org
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mm/slab: Fix undefined init_cache_node_node() for NUMA and !SMP

commit 66a1c22b709178e7b823d44465d0c2e5ed7492fb upstream.

sh/migor_defconfig:

    mm/slab.c: In function ‘slab_memory_callback’:
    mm/slab.c:1127:23: error: implicit declaration of function ‘init_cache_node_node’; did you mean ‘drain_cache_node_node’? [-Werror=implicit-function-declaration]
     1127 |                 ret = init_cache_node_node(nid);
  |                       ^~~~~~~~~~~~~~~~~~~~
  |                       drain_cache_node_node

The #ifdef condition protecting the definition of init_cache_node_node()
no longer matches the conditions protecting the (multiple) users.

Fix this by syncing the conditions.

Fixes: 76af6a054da40553 ("mm/migrate: add CPU hotplug to demotion #ifdef")
Reported-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/b5bdea22-ed2f-3187-6efe-0c72330270a4@infradead.org
Signed-off-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: John Paul Adrian Glaubitz <glaubitz@physik.fu-berlin.de>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Cc: <stable@vger.kernel.org>
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

efi: sysfb_efi: Fix DMI quirks not working for simpledrm

commit 3615c78673c332b69aaacefbcde5937c5c706686 upstream.

Commit 8633ef82f101 ("drivers/firmware: consolidate EFI framebuffer setup
for all arches") moved the sysfb_apply_efi_quirks() call in sysfb_init()
from before the [sysfb_]parse_mode() call to after it.
But sysfb_apply_efi_quirks() modifies the global screen_info struct which
[sysfb_]parse_mode() parses, so doing it later is too late.

This has broken all DMI based quirks for correcting wrong firmware efifb
settings when simpledrm is used.

To fix this move the sysfb_apply_efi_quirks() call back to its old place
and split the new setup of the efifb_fwnode (which requires
the platform_device) into its own function and call that at
the place of the moved sysfb_apply_efi_quirks(pd) calls.

Fixes: 8633ef82f101 ("drivers/firmware: consolidate EFI framebuffer setup for all arches")
Cc: stable@vger.kernel.org
Cc: Javier Martinez Canillas <javierm@redhat.com>
Cc: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Reviewed-by: Javier Martinez Canillas <javierm@redhat.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Bluetooth: Fix race condition in hci_cmd_sync_clear

commit 1c66bee492a5fe00ae3fe890bb693bfc99f994c6 upstream.

There is a potential race condition in hci_cmd_sync_work and
hci_cmd_sync_clear, and could lead to use-after-free. For instance,
hci_cmd_sync_work is added to the 'req_workqueue' after cancel_work_sync
The entry of 'cmd_sync_work_list' may be freed in hci_cmd_sync_clear, and
causing kernel panic when it is used in 'hci_cmd_sync_work'.

Here's the call trace:

dump_stack_lvl+0x49/0x63
print_report.cold+0x5e/0x5d3
? hci_cmd_sync_work+0x282/0x320
kasan_report+0xaa/0x120
? hci_cmd_sync_work+0x282/0x320
__asan_report_load8_noabort+0x14/0x20
hci_cmd_sync_work+0x282/0x320
process_one_work+0x77b/0x11c0
? _raw_spin_lock_irq+0x8e/0xf0
worker_thread+0x544/0x1180
? poll_idle+0x1e0/0x1e0
kthread+0x285/0x320
? process_one_work+0x11c0/0x11c0
? kthread_complete_and_exit+0x30/0x30
ret_from_fork+0x22/0x30
</TASK>

Allocated by task 266:
kasan_save_stack+0x26/0x50
__kasan_kmalloc+0xae/0xe0
kmem_cache_alloc_trace+0x191/0x350
hci_cmd_sync_queue+0x97/0x2b0
hci_update_passive_scan+0x176/0x1d0
le_conn_complete_evt+0x1b5/0x1a00
hci_le_conn_complete_evt+0x234/0x340
hci_le_meta_evt+0x231/0x4e0
hci_event_packet+0x4c5/0xf00
hci_rx_work+0x37d/0x880
process_one_work+0x77b/0x11c0
worker_thread+0x544/0x1180
kthread+0x285/0x320
ret_from_fork+0x22/0x30

Freed by task 269:
kasan_save_stack+0x26/0x50
kasan_set_track+0x25/0x40
kasan_set_free_info+0x24/0x40
____kasan_slab_free+0x176/0x1c0
__kasan_slab_free+0x12/0x20
slab_free_freelist_hook+0x95/0x1a0
kfree+0xba/0x2f0
hci_cmd_sync_clear+0x14c/0x210
hci_unregister_dev+0xff/0x440
vhci_release+0x7b/0xf0
__fput+0x1f3/0x970
____fput+0xe/0x20
task_work_run+0xd4/0x160
do_exit+0x8b0/0x22a0
do_group_exit+0xba/0x2a0
get_signal+0x1e4a/0x25b0
arch_do_signal_or_restart+0x93/0x1f80
exit_to_user_mode_prepare+0xf5/0x1a0
syscall_exit_to_user_mode+0x26/0x50
ret_from_fork+0x15/0x30

Fixes: 6a98e3836fa2 ("Bluetooth: Add helper for serialized HCI command execution")
Cc: stable@vger.kernel.org
Signed-off-by: Min Li <lm0963hack@gmail.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

btrfs: zoned: fix btrfs_can_activate_zone() to support DUP profile

commit 9e1cdf0c354e46e428c0e0cab008abbe81b6013d upstream.

btrfs_can_activate_zone() returns true if at least one device has one zone
available for activation. This is OK for the single profile, but not OK for
DUP profile. We need two zones to create a DUP block group. Fix it by
properly handling the case with the profile flags.

Fixes: 265f7237dd25 ("btrfs: zoned: allow DUP on meta-data block groups")
CC: stable@vger.kernel.org # 6.1+
Reviewed-by: Johannes Thumshirn <johannes.thumshirn@wdc.com>
Signed-off-by: Naohiro Aota <naohiro.aota@wdc.com>
Reviewed-by: David Sterba <dsterba@suse.com>
Signed-off-by: David Sterba <dsterba@suse.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: gadget: u_audio: don't let userspace block driver unbind

commit 6c67ed9ad9b83e453e808f9b31a931a20a25629b upstream.

In the unbind callback for f_uac1 and f_uac2, a call to snd_card_free()
via g_audio_cleanup() will disconnect the card and then wait for all
resources to be released, which happens when the refcount falls to zero.
Since userspace can keep the refcount incremented by not closing the
relevant file descriptor, the call to unbind may block indefinitely.
This can cause a deadlock during reboot, as evidenced by the following
blocked task observed on my machine:

  task:reboot  state:D stack:0   pid:2827  ppid:569    flags:0x0000000c
  Call trace:
   __switch_to+0xc8/0x140
   __schedule+0x2f0/0x7c0
   schedule+0x60/0xd0
   schedule_timeout+0x180/0x1d4
   wait_for_completion+0x78/0x180
   snd_card_free+0x90/0xa0
   g_audio_cleanup+0x2c/0x64
   afunc_unbind+0x28/0x60
   ...
   kernel_restart+0x4c/0xac
   __do_sys_reboot+0xcc/0x1ec
   __arm64_sys_reboot+0x28/0x30
   invoke_syscall+0x4c/0x110
   ...

The issue can also be observed by opening the card with arecord and
then stopping the process through the shell before unbinding:

  # arecord -D hw:UAC2Gadget -f S32_LE -c 2 -r 48000 /dev/null
  Recording WAVE '/dev/null' : Signed 32 bit Little Endian, Rate 48000 Hz, Stereo
  ^Z[1]+  Stopped                    arecord -D hw:UAC2Gadget -f S32_LE -c 2 -r 48000 /dev/null
  # echo gadget.0 > /sys/bus/gadget/drivers/configfs-gadget/unbind
  (observe that the unbind command never finishes)

Fix the problem by using snd_card_free_when_closed() instead, which will
still disconnect the card as desired, but defer the task of freeing the
resources to the core once userspace closes its file descriptor.

Fixes: 132fcb460839 ("usb: gadget: Add Audio Class 2.0 Driver")
Cc: stable@vger.kernel.org
Signed-off-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Reviewed-by: Ruslan Bilovol <ruslan.bilovol@gmail.com>
Reviewed-by: John Keeping <john@metanate.com>
Link: https://lore.kernel.org/r/20230302163648.3349669-1-alvin@pqrs.dk
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: dwc2: fix a devres leak in hw_enable upon suspend resume

commit f747313249b74f323ddf841a9c8db14d989f296a upstream.

Each time the platform goes to low power, PM suspend / resume routines
call: __dwc2_lowlevel_hw_enable -> devm_add_action_or_reset().
This adds a new devres each time.
This may also happen at runtime, as dwc2_lowlevel_hw_enable() can be
called from udc_start().

This can be seen with tracing:
- echo 1 > /sys/kernel/debug/tracing/events/dev/devres_log/enable
- go to low power
- cat /sys/kernel/debug/tracing/trace

A new "ADD" entry is found upon each low power cycle:
... devres_log: 49000000.usb-otg ADD 82a13bba devm_action_release (8 bytes)
... devres_log: 49000000.usb-otg ADD 49889daf devm_action_release (8 bytes)
...

A second issue is addressed here:
- regulator_bulk_enable() is called upon each PM cycle (suspend/resume).
- regulator_bulk_disable() never gets called.

So the reference count for these regulators constantly increase, by one
upon each low power cycle, due to missing regulator_bulk_disable() call
in __dwc2_lowlevel_hw_disable().

The original fix that introduced the devm_add_action_or_reset() call,
fixed an issue during probe, that happens due to other errors in
dwc2_driver_probe() -> dwc2_core_reset(). Then the probe fails without
disabling regulators, when dr_mode == USB_DR_MODE_PERIPHERAL.

Rather fix the error path: disable all the low level hardware in the
error path, by using the "hsotg->ll_hw_enabled" flag. Checking dr_mode
has been introduced to avoid a dual call to dwc2_lowlevel_hw_disable().
"ll_hw_enabled" should achieve the same (and is used currently in the
remove() routine).

Fixes: 54c196060510 ("usb: dwc2: Always disable regulators on driver teardown")
Fixes: 33a06f1300a7 ("usb: dwc2: Fix error path in gadget registration")
Cc: stable <stable@kernel.org>
Signed-off-by: Fabrice Gasnier <fabrice.gasnier@foss.st.com>
Link: https://lore.kernel.org/r/20230316084127.126084-1-fabrice.gasnier@foss.st.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: dwc2: drd: fix inconsistent mode if role-switch-default-mode="host"

commit 82f5332d3b9872ab5b287e85c57b76d8bb640cd1 upstream.

Some boards might use USB-A female connector for USB ports, however,
the port could be connected to a dual-mode USB controller, making it
also behaves as a peripheral device if male-to-male cable is connected.

In this case, the dts looks like this:

&usb0 {
status = "okay";
dr_mode = "otg";
usb-role-switch;
role-switch-default-mode = "host";
};

After boot, dwc2_ovr_init() sets GOTGCTL to GOTGCTL_AVALOVAL and call
dwc2_force_mode() with parameter host=false, which causes inconsistent
mode - The hardware is in peripheral mode while the kernel status is
in host mode.

What we can do now is to call dwc2_drd_role_sw_set() to switch to
device mode, and everything should work just fine now, even switching
back to none(default) mode afterwards.

Fixes: e14acb876985 ("usb: dwc2: drd: add role-switch-default-node support")
Cc: stable <stable@kernel.org>
Signed-off-by: Ziyang Huang <hzyitc@outlook.com>
Tested-by: Fabrice Gasnier <fabrice.gasnier@foss.st.com>
Acked-by: Minas Harutyunyan <hminas@synopsys.com>
Reviewed-by: Amelie Delaunay <amelie.delaunay@foss.st.com>
Link: https://lore.kernel.org/r/SG2PR01MB204837BF68EDB0E343D2A375C9A59@SG2PR01MB2048.apcprd01.prod.exchangelabs.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: misc: onboard-hub: add support for Microchip USB2517 USB 2.0 hub

commit f7c13cb48e85538709850589b496c4ddb3d3898e upstream.

Add support for Microchip USB2517 USB 2.0 hub to the onboard usb hub
driver. Adopt the generic usb-device compatible ("usbVID,PID").
This hub has the same reset timings as USB2514, so reuse that one.
There is also an USB2517I which just has industrial temperature range.

Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Cc: stable <stable@kernel.org>
Acked-by: Matthias Kaehlcke <mka@chromium.org>
Link: https://lore.kernel.org/r/20230223073920.2912298-1-alexander.stein@ew.tq-group.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: core: Add BLIST_SKIP_VPD_PAGES for SKhynix H28U74301AMR

commit a204b490595de71016b2360a1886ec8c12d0afac upstream.

Xiaomi Poco F1 (qcom/sdm845-xiaomi-beryllium*.dts) comes with a SKhynix
H28U74301AMR UFS. The sd_read_cpr() operation leads to a 120 second
timeout, making the device bootup very slow:

[  121.457736] sd 0:0:0:1: [sdb] tag#23 timing out command, waited 120s

Setting the BLIST_SKIP_VPD_PAGES allows the device to skip the failing
sd_read_cpr operation and boot normally.

Signed-off-by: Joel Selvaraj <joelselvaraj.oss@gmail.com>
Link: https://lore.kernel.org/r/20230313041402.39330-1-joelselvaraj.oss@gmail.com
Cc: stable@vger.kernel.org
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests/x86/amx: Add a ptrace test

commit 62faca1ca10cc84e99ae7f38aa28df2bc945369b upstream.

Include a test case to validate the XTILEDATA injection to the target.

Also, it ensures the kernel's ability to copy states between different
XSAVE formats.

Refactor the memcmp() code to be usable for the state validation.

Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/20230227210504.18520-3-chang.seok.bae%40intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

x86/fpu/xstate: Prevent false-positive warning in __copy_xstate_uabi_buf()

commit b15888840207c2bfe678dd1f68a32db54315e71f upstream.

__copy_xstate_to_uabi_buf() copies either from the tasks XSAVE buffer
or from init_fpstate into the ptrace buffer. Dynamic features, like
XTILEDATA, have an all zeroes init state and are not saved in
init_fpstate, which means the corresponding bit is not set in the
xfeatures bitmap of the init_fpstate header.

But __copy_xstate_to_uabi_buf() retrieves addresses for both the tasks
xstate and init_fpstate unconditionally via __raw_xsave_addr().

So if the tasks XSAVE buffer has a dynamic feature set, then the
address retrieval for init_fpstate triggers the warning in
__raw_xsave_addr() which checks the feature bit in the init_fpstate
header.

Remove the address retrieval from init_fpstate for extended features.
They have an all zeroes init state so init_fpstate has zeros for them.
Then zeroing the user buffer for the init state is the same as copying
them from init_fpstate.

Fixes: 2308ee57d93d ("x86/fpu/amx: Enable the AMX feature in 64-bit mode")
Reported-by: Mingwei Zhang <mizhang@google.com>
Link: https://lore.kernel.org/kvm/20230221163655.920289-2-mizhang@google.com/
Signed-off-by: Chang S. Bae <chang.seok.bae@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Tested-by: Mingwei Zhang <mizhang@google.com>
Link: https://lore.kernel.org/all/20230227210504.18520-2-chang.seok.bae%40intel.com
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: fix dentry lookups in directory handle cache

commit be4fde79812f02914e350bde0bc4cfeae8429378 upstream.

Get rid of any prefix paths in @path before lookup_positive_unlocked()
as it will call ->lookup() which already adds those prefix paths
through build_path_from_dentry().

This has caused a performance regression when mounting shares with a
prefix path where readdir(2) would end up retrying several times to
open bad directory names that contained duplicate prefix paths.

Fix this by skipping any prefix paths in @path before calling
lookup_positive_unlocked().

Fixes: e4029e072673 ("cifs: find and use the dentry for cached non-root directories also")
Cc: stable@vger.kernel.org # 6.1+
Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: print session id while listing open files

commit 175b54abc443b6965e9379b71ec05f7c73c192e9 upstream.

In the output of /proc/fs/cifs/open_files, we only print
the tree id for the tcon of each open file. It becomes
difficult to know which tcon these files belong to with
just the tree id.

This change dumps ses id in addition to all other data today.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: dump pending mids for all channels in DebugData

commit d12bc6d26f92c51b28e8f4a146ffcc630b688198 upstream.

Currently, we only dump the pending mid information only
on the primary channel in /proc/fs/cifs/DebugData.
If multichannel is active, we do not print the pending MID
list on secondary channels.

This change will dump the pending mids for all the channels
based on server->conn_id.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: empty interface list when server doesn't support query interfaces

commit 896cd316b841053f6df95ab77b5f1322c16a8e18 upstream.

When querying server interfaces returns -EOPNOTSUPP,
clear the list of interfaces. Assumption is that multichannel
would be disabled too.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: do not poll server interfaces too regularly

commit 072a28c8907c841f7d4b56c78bce46d3ee211e73 upstream.

We have the server interface list hanging off the tcon
structure today for reasons unknown. So each tcon which is
connected to a file server can query them separately,
which is really unnecessary. To avoid this, in the query
function, we will check the time of last update of the
interface list, and avoid querying the server if it is
within a certain range.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: append path to open_enter trace event

commit fddc6ccc487e5de07b98df8d04118d5dcb5e0407 upstream.

We do not dump the file path for smb3_open_enter ftrace
calls, which is a severe handicap while debugging
using ftrace evens. This change adds that info.

Unfortunately, we're not updating the path in open params
in many places; which I had to do as a part of this change.
SMB2_open gets path in utf16 format, but it's easier of
path is supplied as char pointer in oparms.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

cifs: lock chan_lock outside match_session

commit 2f4e429c846972c8405951a9ff7a82aceeca7461 upstream.

Coverity had rightly indicated a possible deadlock
due to chan_lock being done inside match_session.
All callers of match_* functions should pick up the
necessary locks and call them.

Signed-off-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Cc: stable@vger.kernel.org
Fixes: 724244cdb382 ("cifs: protect session channel fields with chan_lock")
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

act_mirred: use the backlog for nested calls to mirred ingress

[ Upstream commit ca22da2fbd693b54dc8e3b7b54ccc9f7e9ba3640 ]

William reports kernel soft-lockups on some OVS topologies when TC mirred
egress->ingress action is hit by local TCP traffic [1].
The same can also be reproduced with SCTP (thanks Xin for verifying), when
client and server reach themselves through mirred egress to ingress, and
one of the two peers sends a "heartbeat" packet (from within a timer).

Enqueueing to backlog proved to fix this soft lockup; however, as Cong
noticed [2], we should preserve - when possible - the current mirred
behavior that counts as "overlimits" any eventual packet drop subsequent to
the mirred forwarding action [3]. A compromise solution might use the
backlog only when tcf_mirred_act() has a nest level greater than one:
change tcf_mirred_forward() accordingly.

Also, add a kselftest that can reproduce the lockup and verifies TC mirred
ability to account for further packet drops after TC mirred egress->ingress
(when the nest level is 1).

 [1] https://lore.kernel.org/netdev/33dc43f587ec1388ba456b4915c75f02a8aae226.1663945716.git.dcaratti@redhat.com/
 [2] https://lore.kernel.org/netdev/Y0w%2FWWY60gqrtGLp@pop-os.localdomain/
 [3] such behavior is not guaranteed: for example, if RPS or skb RX
     timestamping is enabled on the mirred target device, the kernel
     can defer receiving the skb and return NET_RX_SUCCESS inside
     tcf_mirred_forward().

Reported-by: William Zhao <wizhao@redhat.com>
CC: Xin Long <lucien.xin@gmail.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Reviewed-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net/sched: act_mirred: better wording on protection against excessive stack growth

[ Upstream commit 78dcdffe0418ac8f3f057f26fe71ccf4d8ed851f ]

with commit e2ca070f89ec ("net: sched: protect against stack overflow in
TC act_mirred"), act_mirred protected itself against excessive stack growth
using per_cpu counter of nested calls to tcf_mirred_act(), and capping it
to MIRRED_RECURSION_LIMIT. However, such protection does not detect
recursion/loops in case the packet is enqueued to the backlog (for example,
when the mirred target device has RPS or skb timestamping enabled). Change
the wording from "recursion" to "nesting" to make it more clear to readers.

CC: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Davide Caratti <dcaratti@redhat.com>
Reviewed-by: Marcelo Ricardo Leitner <marcelo.leitner@gmail.com>
Acked-by: Jamal Hadi Salim <jhs@mojatatu.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Stable-dep-of: ca22da2fbd69 ("act_mirred: use the backlog for nested calls to mirred ingress")
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd: Fix initialization mistake for NBIO 7.3.0

[ Upstream commit 1717cc5f2962a4652c76ed3858b499ccae6c277c ]

The same strapping initialization issue that happened on NBIO 7.5.1
appears to be happening on NBIO 7.3.0.
Apply the same fix to 7.3.0 as well.

Note: This workaround relies upon the integrated GPU being enabled
in BIOS. If the integrated GPU is disabled in BIOS a different
workaround will be required.

Reported-by: Thomas Glanzmann <thomas@glanzmann.de>
Cc: Basavaraj Natikar <Basavaraj.Natikar@amd.com>
Link: https://lore.kernel.org/linux-usb/Y%2Fz9GdHjPyF2rNG3@glanzmann.de/T/#u
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amdgpu: Fix call trace warning and hang when removing amdgpu device

[ Upstream commit 93bb18d2a873d2fa9625c8ea927723660a868b95 ]

On GPUs with RAS enabled, below call trace and hang are observed when
shutting down device.

v2: use DRM device unplugged flag instead of shutdown flag as the check to
prevent memory wipe in shutdown stage.

[ +0.000000] RIP: 0010:amdgpu_vram_mgr_fini+0x18d/0x1c0 [amdgpu]
[ +0.000001] PKRU: 55555554
[ +0.000001] Call Trace:
[ +0.000001] <TASK>
[ +0.000002] amdgpu_ttm_fini+0x140/0x1c0 [amdgpu]
[ +0.000183] amdgpu_bo_fini+0x27/0xa0 [amdgpu]
[ +0.000184] gmc_v11_0_sw_fini+0x2b/0x40 [amdgpu]
[ +0.000163] amdgpu_device_fini_sw+0xb6/0x510 [amdgpu]
[ +0.000152] amdgpu_driver_release_kms+0x16/0x30 [amdgpu]
[ +0.000090] drm_dev_release+0x28/0x50 [drm]
[ +0.000016] devm_drm_dev_init_release+0x38/0x60 [drm]
[ +0.000011] devm_action_release+0x15/0x20
[ +0.000003] release_nodes+0x40/0xc0
[ +0.000001] devres_release_all+0x9e/0xe0
[ +0.000001] device_unbind_cleanup+0x12/0x80
[ +0.000003] device_release_driver_internal+0xff/0x160
[ +0.000001] driver_detach+0x4a/0x90
[ +0.000001] bus_remove_driver+0x6c/0xf0
[ +0.000001] driver_unregister+0x31/0x50
[ +0.000001] pci_unregister_driver+0x40/0x90
[ +0.000003] amdgpu_exit+0x15/0x120 [amdgpu]

Signed-off-by: lyndonli <Lyndon.Li@amd.com>
Reviewed-by: Guchun Chen <guchun.chen@amd.com>
Reviewed-by: Christian König <christian.koenig@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

sh: sanitize the flags on sigreturn

[ Upstream commit 573b22ccb7ce9ab7f0539a2e11a9d3609a8783f5 ]

We fetch %SR value from sigframe; it might have been modified by signal
handler, so we can't trust it with any bits that are not modifiable in
user mode.

Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
Cc: Rich Felker <dalias@libc.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Update clock table to include highest clock setting

[ Upstream commit 2d99a7ec25cf456cd3680eb314d6454138e5aa64 ]

[Why]
Currently, the clk manager matches SocVoltage with voltage from
fused settings (dfPstate clock table). And then corresponding clocks
are selected.

However in certain situations, this leads to clk manager not
including at least one entry with highest supported clock setting.

[How]
Update the clk manager to include at least one entry with highest
supported clock setting.

Reviewed-by: Pavle Kotarac <pavle.kotarac@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Swapnil Patel <Swapnil.Patel@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: usb: qmi_wwan: add Telit 0x1080 composition

[ Upstream commit 382e363d5bed0cec5807b35761d14e55955eee63 ]

Add the following Telit FE990 composition:

0x1080: tty, adb, rmnet, tty, tty, tty, tty

Signed-off-by: Enrico Sau <enrico.sau@gmail.com>
Link: https://lore.kernel.org/r/20230306120528.198842-1-enrico.sau@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: usb: cdc_mbim: avoid altsetting toggling for Telit FE990

[ Upstream commit 418383e6ed6b4624a54ec05c535f13d184fbf33b ]

Add quirk CDC_MBIM_FLAG_AVOID_ALTSETTING_TOGGLE for Telit FE990
0x1081 composition in order to avoid bind error.

Signed-off-by: Enrico Sau <enrico.sau@gmail.com>
Link: https://lore.kernel.org/r/20230306115933.198259-1-enrico.sau@gmail.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/x86: int3472: Add GPIOs to Surface Go 3 Board data

[ Upstream commit e8059d393158e927e36898aca89986a52025b9f3 ]

Add the INT347E GPIO lookup table to the board data for the Surface
Go 3. This is necessary to allow the ov7251 IR camera to probe
properly on that platform.

Signed-off-by: Daniel Scally <dan.scally@ideasonboard.com>
Link: https://lore.kernel.org/r/20230302102611.314341-1-dan.scally@ideasonboard.com
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: storvsc: Handle BlockSize change in Hyper-V VHD/VHDX file

[ Upstream commit 11d9874c4204a785f43d899a1ab12f9dc8d9de3e ]

Hyper-V uses a VHD or VHDX file on the host as the underlying storage for a
virtual disk.  The VHD/VHDX file format is a sparse format where real disk
space on the host is assigned in chunks that the VHD/VHDX file format calls
the BlockSize.  This BlockSize is not to be confused with the 512-byte (or
4096-byte) sector size of the underlying storage device.  The default block
size for a new VHD/VHDX file is 32 Mbytes.  When a guest VM touches any
disk space within a 32 Mbyte chunk of the VHD/VHDX file, Hyper-V allocates
32 Mbytes of real disk space for that section of the VHD/VHDX. Similarly,
if a discard operation is done that covers an entire 32 Mbyte chunk,
Hyper-V will free the real disk space for that portion of the VHD/VHDX.
This BlockSize is surfaced in Linux as the "discard_granularity" in
/sys/block/sd<x>/queue, which makes sense.

Hyper-V also has differencing disks that can overlay a VHD/VHDX file to
capture changes to the VHD/VHDX while preserving the original VHD/VHDX.
One example of this differencing functionality is for VM snapshots.  When a
snapshot is created, a differencing disk is created.  If the snapshot is
rolled back, Hyper-V can just delete the differencing disk, and the VM will
see the original disk contents at the time the snapshot was taken.
Differencing disks are used in other scenarios as well.

The BlockSize for a differencing disk defaults to 2 Mbytes, not 32 Mbytes.
The smaller default is used because changes to differencing disks are
typically scattered all over, and Hyper-V doesn't want to allocate 32
Mbytes of real disk space for a stray write here or there.  The smaller
BlockSize provides more efficient use of real disk space.

When a differencing disk is added to a VHD/VHDX, Hyper-V reports
UNIT_ATTENTION with a sense code indicating "Operating parameters have
changed", because the value of discard_granularity should be changed to 2
Mbytes. When the differencing disk is removed, discard_granularity should
be changed back to 32 Mbytes.  However, current code simply reports a
message from scsi_report_sense() and the value of
/sys/block/sd<x>/queue/discard_granularity is not updated. The message
isn't very actionable by a sysadmin.

Fix this by having the storvsc driver check for the sense code indicating
that the underly VHD/VHDX block size has changed, and do a rescan of the
device to pick up the new discard_granularity.  With this change the entire
transition to/from differencing disks is handled automatically and
transparently, with no confusing messages being output.

Link: https://lore.kernel.org/r/1677516514-86060-1-git-send-email-mikelley@microsoft.com
Signed-off-by: Michael Kelley <mikelley@microsoft.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: mpi3mr: Bad drive in topology results kernel crash

[ Upstream commit 8e45183978d64699df639e795235433a60f35047 ]

When the SAS Transport Layer support is enabled and a device exposed to
the OS by the driver fails INQUIRY commands, the driver frees up the memory
allocated for an internal HBA port data structure. However, in some places,
the reference to the freed memory is not cleared. When the firmware sends
the Device Info change event for the same device again, the freed memory is
accessed and that leads to memory corruption and OS crash.

Signed-off-by: Ranjan Kumar <ranjan.kumar@broadcom.com>
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Link: https://lore.kernel.org/r/20230228140835.4075-7-ranjan.kumar@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: mpi3mr: NVMe command size greater than 8K fails

[ Upstream commit 4f297e856a7b5da2f2c66a12e739666e23943560 ]

A wrong variable is checked while populating PRP entries in the PRP page
and this results in failure. No PRP entries in the PRP page were
successfully created and any NVMe Encapsulated commands with PRP of size
greater than 8K failed.

Signed-off-by: Ranjan Kumar <ranjan.kumar@broadcom.com>
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Link: https://lore.kernel.org/r/20230228140835.4075-6-ranjan.kumar@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: mpi3mr: Wait for diagnostic save during controller init

[ Upstream commit 0a319f1629495d27879b7ebf6eee62b8cf6e4c37 ]

If a controller reset operation is triggered to recover the controller from
a fault state, then wait for the snapdump to be saved in the firmware
region before proceeding to reset the controller.

Signed-off-by: Ranjan Kumar <ranjan.kumar@broadcom.com>
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Link: https://lore.kernel.org/r/20230228140835.4075-4-ranjan.kumar@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: mpi3mr: Driver unload crashes host when enhanced logging is enabled

[ Upstream commit 5b06a7169c59ce0c77ef8b9c82aa07c478f82aac ]

Prevent driver from trying to dereference a NULL pointer in a debug print
while removing a device during driver unload.

Signed-off-by: Ranjan Kumar <ranjan.kumar@broadcom.com>
Signed-off-by: Sreekanth Reddy <sreekanth.reddy@broadcom.com>
Link: https://lore.kernel.org/r/20230228140835.4075-3-ranjan.kumar@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: lpfc: Avoid usage of list iterator variable after loop

[ Upstream commit 2850b23e9f9ae3696e472d2883ea1b43aafa884e ]

If the &epd_pool->list is empty when executing
lpfc_get_io_buf_from_expedite_pool() the function would return an invalid
pointer. Even in the case if the list is guaranteed to be populated, the
iterator variable should not be used after the loop to be more robust for
future changes.

Linus proposed to avoid any use of the list iterator variable after the
loop, in the attempt to move the list iterator variable declaration into
the macro to avoid any potential misuse after the loop [1].

Link: https://lore.kernel.org/all/CAHk-=wgRr_D8CB-D9Kg-c=EHreAsk5SqXPwr9Y7k9sA6cWXJ6w@mail.gmail.com/
Signed-off-by: Jakob Koschel <jkl820.git@gmail.com>
Link: https://lore.kernel.org/r/20230301-scsi-lpfc-avoid-list-iterator-after-loop-v1-1-325578ae7561@gmail.com
Reviewed-by: Justin Tee <justin.tee@broadcom.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: lpfc: Check kzalloc() in lpfc_sli4_cgn_params_read()

[ Upstream commit 312320b0e0ec21249a17645683fe5304d796aec1 ]

If kzalloc() fails in lpfc_sli4_cgn_params_read(), then we rely on
lpfc_read_object()'s routine to NULL check pdata.

Currently, an early return error is thrown from lpfc_read_object() to
protect us from NULL ptr dereference, but the errno code is -ENODEV.

Change the errno code to a more appropriate -ENOMEM.

Reported-by: Kang Chen <void0red@gmail.com>
Link: https://lore.kernel.org/all/20230226102338.3362585-1-void0red@gmail.com
Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20230228044336.5195-1-justintee8345@gmail.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: ufs: core: Add soft dependency on governor_simpleondemand

[ Upstream commit 2ebe16155dc8bd4e602cad5b5f65458d2eaa1a75 ]

The ufshcd driver uses simpleondemand governor for devfreq. Add it to the
list of ufshcd softdeps to allow userspace initramfs tools like dracut to
automatically pull the governor module into the initramfs together with UFS
drivers.

Link: https://lore.kernel.org/r/20230220140740.14379-1-athierry@redhat.com
Signed-off-by: Adrien Thierry <athierry@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: hisi_sas: Check devm_add_action() return value

[ Upstream commit 06d1a90de60208054cca15ef200138cfdbb642a9 ]

In case devm_add_action() fails, check it in the caller of
interrupt_preinit_v3_hw().

Link: https://lore.kernel.org/r/20230227031030.893324-1-void0red@gmail.com
Signed-off-by: Kang Chen <void0red@gmail.com>
Acked-by: Xiang Chen <chenxiang66@hisilicon.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: qla2xxx: Add option to disable FC2 Target support

[ Upstream commit 877b03795fcf29ff2e2351f7e574ecc9b9c51732 ]

Commit 44c57f205876 ("scsi: qla2xxx: Changes to support FCP2 Target") added
support for FC2 Targets. Unfortunately, there are older setups which break
with this new feature enabled.

Allow to disable it via module option.

Link: https://lore.kernel.org/r/20230208152014.109214-1-dwagner@suse.de
Signed-off-by: Daniel Wagner <dwagner@suse.de>
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: target: iscsi: Fix an error message in iscsi_check_key()

[ Upstream commit 6cc55c969b7ce8d85e09a636693d4126c3676c11 ]

The first half of the error message is printed by pr_err(), the second half
is printed by pr_debug(). The user will therefore see only the first part
of the message and will miss some useful information.

Link: https://lore.kernel.org/r/20230214141556.762047-1-mlombard@redhat.com
Signed-off-by: Maurizio Lombardi <mlombard@redhat.com>
Reviewed-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/bpf: check that modifier resolves after pointer

[ Upstream commit dfdd608c3b365f0fd49d7e13911ebcde06b9865b ]

Add a regression test that ensures that a VAR pointing at a
modifier which follows a PTR (or STRUCT or ARRAY) is resolved
correctly by the datasec validator.

Signed-off-by: Lorenz Bauer <lmb@isovalent.com>
Link: https://lore.kernel.org/r/20230306112138.155352-3-lmb@isovalent.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

m68k: Only force 030 bus error if PC not in exception table

[ Upstream commit e36a82bebbf7da814530d5a179bef9df5934b717 ]

__get_kernel_nofault() does copy data in supervisor mode when
forcing a task backtrace log through /proc/sysrq_trigger.
This is expected cause a bus error exception on e.g. NULL
pointer dereferencing when logging a kernel task has no
workqueue associated. This bus error ought to be ignored.

Our 030 bus error handler is ill equipped to deal with this:

Whenever ssw indicates a kernel mode access on a data fault,
we don't even attempt to handle the fault and instead always
send a SEGV signal (or panic). As a result, the check
for exception handling at the fault PC (buried in
send_sig_fault() which gets called from do_page_fault()
eventually) is never used.

In contrast, both 040 and 060 access error handlers do not
care whether a fault happened on supervisor mode access,
and will call do_page_fault() on those, ultimately honoring
the exception table.

Add a check in bus_error030 to call do_page_fault() in case
we do have an entry for the fault PC in our exception table.

I had attempted a fix for this earlier in 2019 that did rely
on testing pagefault_disabled() (see link below) to achieve
the same thing, but this patch should be more generic.

Tested on 030 Atari Falcon.

Reported-by: Eero Tamminen <oak@helsinkinet.fi>
Link: https://lore.kernel.org/r/alpine.LNX.2.21.1904091023540.25@nippy.intranet
Link: https://lore.kernel.org/r/63130691-1984-c423-c1f2-73bfd8d3dcd3@gmail.com
Signed-off-by: Michael Schmitz <schmitzmic@gmail.com>
Reviewed-by: Geert Uytterhoeven <geert@linux-m68k.org>
Link: https://lore.kernel.org/r/20230301021107.26307-1-schmitzmic@gmail.com
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

m68k: mm: Fix systems with memory at end of 32-bit address space

[ Upstream commit 0d9fad91abfd723ea5070a46d98a9f4496c93ba9 ]

The calculation of end addresses of memory chunks overflowed to 0 when
a memory chunk is located at the end of 32-bit address space.
This is the case for the HP300 architecture.

Link: https://lore.kernel.org/linux-m68k/CACz-3rhUo5pgNwdWHaPWmz+30Qo9xCg70wNxdf7o5x-6tXq8QQ@mail.gmail.com/
Signed-off-by: Kars de Jong <jongk@linux-m68k.org>
Reviewed-by: Geert Uytterhoeven <geert@linux-m68k.org>
Link: https://lore.kernel.org/r/20230223112349.26675-1-jongk@linux-m68k.org
Signed-off-by: Geert Uytterhoeven <geert@linux-m68k.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

HID: intel-ish-hid: ipc: Fix potential use-after-free in work function

[ Upstream commit 8ae2f2b0a28416ed2f6d8478ac8b9f7862f36785 ]

When a reset notify IPC message is received, the ISR schedules a work
function and passes the ISHTP device to it via a global pointer
ishtp_dev. If ish_probe() fails, the devm-managed device resources
including ishtp_dev are freed, but the work is not cancelled, causing a
use-after-free when the work function tries to access ishtp_dev. Use
devm_work_autocancel() instead, so that the work is automatically
cancelled if probe fails.

Signed-off-by: Reka Norman <rekanorman@chromium.org>
Acked-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>

HID: logitech-hidpp: Add support for Logitech MX Master 3S mouse

[ Upstream commit db50f7a3983f0154e730f1147ef729e0c5c2f90c ]

Add signature for the Logitech MX Master 3S mouse over Bluetooth.

Signed-off-by: Rafał Szalecki <perexist7@gmail.com>
Reviewed-by: Bastien Nocera <hadess@hadess.net>
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ca8210: fix mac_len negative array access

[ Upstream commit 6c993779ea1d0cccdb3a5d7d45446dd229e610a3 ]

This patch fixes a buffer overflow access of skb->data if
ieee802154_hdr_peek_addrs() fails.

Reported-by: lianhui tang <bluetlh@gmail.com>
Signed-off-by: Alexander Aring <aahringo@redhat.com>
Link: https://lore.kernel.org/r/20230217042504.3303396-1-aahringo@redhat.com
Signed-off-by: Stefan Schmidt <stefan@datenfreihafen.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

HID: cp2112: Fix driver not registering GPIO IRQ chip as threaded

[ Upstream commit 37f5b858a66543b2b67c0288280af623985abc29 ]

The CP2112 generates interrupts from a polling routine on a thread,
and can only support threaded interrupts. This patch configures the
gpiochip irq chip with this flag, disallowing consumers to request
a hard IRQ from this driver, which resulted in a segfault previously.

Signed-off-by: Danny Kaehn <kaehndan@gmail.com>
Link: https://lore.kernel.org/r/20230210170044.11835-1-kaehndan@gmail.com
Signed-off-by: Benjamin Tissoires <benjamin.tissoires@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/cirrus: NULL-check pipe->plane.state->fb in cirrus_pipe_update()

[ Upstream commit 7245e629dcaaf308f1868aeffa218e9849c77893 ]

After having been compared to NULL value at cirrus.c:455, pointer
'pipe->plane.state->fb' is passed as 1st parameter in call to function
'cirrus_fb_blit_rect' at cirrus.c:461, where it is dereferenced at
cirrus.c:316.

Found by Linux Verification Center (linuxtesting.org) with SVACE.

v2:
* aligned commit message to line-length limits

Signed-off-by: Alexandr Sapozhnikov <alsp705@gmail.com>
Reviewed-by: Thomas Zimmermann <tzimmermann@suse.de>
Signed-off-by: Thomas Zimmermann <tzimmermann@suse.de>
Link: https://patchwork.freedesktop.org/patch/msgid/20230215171549.16305-1-alsp705@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

riscv: Bump COMMAND_LINE_SIZE value to 1024

[ Upstream commit 61fc1ee8be26bc192d691932b0a67eabee45d12f ]

Increase COMMAND_LINE_SIZE as the current default value is too low
for syzbot kernel command line.

There has been considerable discussion on this patch that has led to a
larger patch set removing COMMAND_LINE_SIZE from the uapi headers on all
ports.  That's not quite done yet, but it's gotten far enough we're
confident this is not a uABI change so this is safe.

Reported-by: Dmitry Vyukov <dvyukov@google.com>
Signed-off-by: Alexandre Ghiti <alex@ghiti.fr>
Link: https://lore.kernel.org/r/20210316193420.904-1-alex@ghiti.fr
[Palmer: it's not uabi]
Link: https://lore.kernel.org/linux-riscv/874b8076-b0d1-4aaa-bcd8-05d523060152@app.fastmail.com/#t
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: x86: utils: Add Cezanne to the list for forcing StorageD3Enable

[ Upstream commit e2a56364485e7789e7b8f342637c7f3a219f7ede ]

commit 018d6711c26e4 ("ACPI: x86: Add a quirk for Dell Inspiron 14 2-in-1
for StorageD3Enable") introduced a quirk to allow a system with ambiguous
use of _ADR 0 to force StorageD3Enable.

It was reported that several more Dell systems suffered the same symptoms.
As the list is continuing to grow but these are all Cezanne systems,
instead add Cezanne to the CPU list to apply the StorageD3Enable property
and remove the whole list.

It was also reported that an HP system only has StorageD3Enable on the ACPI
device for the first NVME disk, not the second.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=217003
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216773
Reported-by: David Alvarez Lombardi <dqalombardi@proton.me>
Reported-by: dbilios@stdio.gr
Reported-and-tested-by: Elvis Angelaccio <elvis.angelaccio@kde.org>
Tested-by: victor.bonnelle@proton.me
Tested-by: hurricanepootis@protonmail.com
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: x86: Drop quirk for HP Elitebook

[ Upstream commit e32d546c483a2a0f607687f5b521c2a2f942ffbe ]

There was a quirk in `acpi/x86/s2idle.c` for an HP Elitebook G9
platforms to force AMD GUID codepath instead of Microsoft codepath.

This was due to a bug with WCN6855 WLAN firmware interaction with
the system.

This bug is fixed by WCN6855 firmware:
WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.23

Remove the quirk as it's no longer necessary with this firmware.

Link: https://git.kernel.org/pub/scm/linux/kernel/git/firmware/linux-firmware.git/commit/?id=c7a57ef688f7d99d8338a5d8edddc8836ff0e6de
Tested-by: Anson Tsao <anson.tsao@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: amd: yc: Add DMI entries to support HP OMEN 16-n0xxx (8A43)

[ Upstream commit d52279d5c9204a041e9ba02a66a353573b2f96e4 ]

This model requires an additional detection quirk to enable the internal microphone.

Signed-off-by: Duc Anh Le <lub.the.studio@gmail.com>
Link: https://lore.kernel.org/r/20230227234921.7784-1-lub.the.studio@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ASoC: amd: yp: Add OMEN by HP Gaming Laptop 16z-n000 to quirks

[ Upstream commit 22ce6843abec19270bf69b176d7ee0a4ef781da5 ]

Enables display microphone on the HP OMEN 16z-n000 (8A42) laptop

Signed-off-by: Joseph Hunkeler <jhunkeler@gmail.com>
Link: https://lore.kernel.org/r/20230216155007.26143-1-jhunkeler@gmail.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

thunderbolt: Rename shadowed variables bit to interrupt_bit and auto_clear_bit

commit 58cdfe6f58b35f17f56386f5fcf937168a423ad1 upstream.

cppcheck reports
drivers/thunderbolt/nhi.c:74:7: style: Local variable 'bit' shadows outer variable [shadowVariable]
  int bit;
      ^
drivers/thunderbolt/nhi.c:66:6: note: Shadowed declaration
 int bit = ring_interrupt_index(ring) & 31;
     ^
drivers/thunderbolt/nhi.c:74:7: note: Shadow variable
  int bit;
      ^
For readablity rename the outer to interrupt_bit and the innner
to auto_clear_bit.

Fixes: 468c49f44759 ("thunderbolt: Disable interrupt auto clear for ring")
Cc: stable@vger.kernel.org
Signed-off-by: Tom Rix <trix@redhat.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Use const qualifier for `ring_interrupt_index`

commit 1716efdb07938bd6510e1127d02012799112c433 upstream.

`ring_interrupt_index` doesn't change the data for `ring` so mark it as
const. This is needed by the following patch that disables interrupt
auto clear for rings.

Cc: Sanju Mehta <Sanju.Mehta@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Add missing UNSET_INBOUND_SBTX for retimer access

commit cd0c1e582b055dea615001b8bd8eccaf6f69f7ce upstream.

According to USB4 retimer specification, the process of firmware update
sequence requires issuing a SET_INBOUND_SBTX port operation that later
shall be followed by UNSET_INBOUND_SBTX port operation. This last step
is not currently issued by the driver but it is necessary to make sure
the retimers are put back to passthrough mode even during enumeration.

If this step is missing the link may not come up properly after
soft-reboot for example.

For this reason issue UNSET_INBOUND_SBTX after SET_INBOUND_SBTX for
enumeration and also when the NVM upgrade is run.

Reported-by: Christian Schaubschläger <christian.schaubschlaeger@gmx.at>
Link: https://lore.kernel.org/linux-usb/b556f5ed-5ee8-9990-9910-afd60db93310@gmx.at/
Cc: stable@vger.kernel.org
Signed-off-by: Gil Fine <gil.fine@linux.intel.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Disable interrupt auto clear for rings

commit 468c49f44759720a312e52d44a71c3949ed63d7c upstream.

When interrupt auto clear is programmed, any read to the interrupt
status register will clear all interrupts.  If two interrupts have
come in before one can be serviced then this will cause lost interrupts.

On AMD USB4 routers this has manifested in odd problems particularly
with long strings of control tranfers such as reading the DROM via bit
banging.

Instead of clearing interrupts automatically, clear the bit corresponding
to the given ring's interrupt in the ISR.

Fixes: 7a1808f82a37 ("thunderbolt: Handle ring interrupt by reading interrupt status register")
Cc: Sanju Mehta <Sanju.Mehta@amd.com>
Cc: stable@vger.kernel.org
Tested-by: Anson Tsao <anson.tsao@amd.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Fix memory leak in margining

commit acec726473822bc6b585961f4ca2a11fa7f28341 upstream.

Memory for the usb4->margining needs to be relased for the upstream port
of the router as well, even though the debugfs directory gets released
with the router device removal. Fix this.

Fixes: d0f1e0c2a699 ("thunderbolt: Add support for receiver lane margining")
Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Add quirk to disable CLx

commit 7af9da8ce8f9a16221ecd8ba4280582f5bd452fc upstream.

Add QUIRK_NO_CLX to disable the CLx state for hardware which
doesn't supports it.

AMD Yellow Carp and Pink Sardine don't support CLx state,
hence disabling it using QUIRK_NO_CLX.

Cc: stable@vger.kernel.org
Signed-off-by: Sanjay R Mehta <sanju.mehta@amd.com>
Signed-off-by: Basavaraj Natikar <Basavaraj.Natikar@amd.com>
[mw: added debug log when the quirk is run]
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Call tb_check_quirks() after initializing adapters

commit d2d6ddf188f609861489d5d188d545856a3ed399 upstream.

In order to apply quirks based on certain adapter types move call to
tb_check_quirks() happen after the adapters are initialized. This should
not affect the existing quirks.

Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Use scale field when allocating USB3 bandwidth

commit c82510b1d87bdebfe916048857d2ef46f1778aa5 upstream.

When tunneling aggregated USB3 (20 Gb/s) the bandwidth values that are
programmed to the ADP_USB3_CS_2 go higher than 4096 and that does not
fit anymore to the 12-bit field. Fix this by scaling the value using
the scale field accordingly.

Fixes: 3b1d8d577ca8 ("thunderbolt: Implement USB3 bandwidth negotiation routines")
Cc: stable@vger.kernel.org
Signed-off-by: Mika Westerberg <mika.westerberg@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

uas: Add US_FL_NO_REPORT_OPCODES for JMicron JMS583Gen 2

commit a37eb61b6ec064ac794b8a1e89fd33eb582fe51d upstream.

Just like other JMicron JMS5xx enclosures, it chokes on report-opcodes,
let's avoid them.

Signed-off-by: Yaroslav Furman <yaro330@gmail.com>
Cc: stable <stable@kernel.org>
Link: https://lore.kernel.org/r/20230312090745.47962-1-yaro330@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

smb3: fix unusable share after force unmount failure

commit 491eafce1a51c457701351a4bf40733799745314 upstream.

If user does forced unmount ("umount -f") while files are still open
on the share (as was seen in a Kubernetes example running on SMB3.1.1
mount) then we were marking the share as "TID_EXITING" in umount_begin()
which caused all subsequent operations (except write) to fail ... but
unfortunately when umount_begin() is called we do not know yet that
there are open files or active references on the share that would prevent
unmount from succeeding.  Kubernetes had example when they were doing
umount -f when files were open which caused the share to become
unusable until the files were closed (and the umount retried).

Fix this so that TID_EXITING is not set until we are about to send
the tree disconnect (not at the beginning of forced umounts in
umount_begin) so that if "umount -f" fails (due to open files or
references) the mount is still usable.

Cc: stable@vger.kernel.org
Reviewed-by: Shyam Prasad N <sprasad@microsoft.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

smb3: lower default deferred close timeout to address perf regression

commit 7e0e76d99079be13c9961dde7c93b2d1ee665af4 upstream.

Performance tests with large number of threads noted that the change
of the default closetimeo (deferred close timeout between when
close is done by application and when client has to send the close
to the server), to 5 seconds from 1 second, significantly degraded
perf in some cases like this (in the filebench example reported,
the stats show close requests on the wire taking twice as long,
and 50% regression in filebench perf). This is stil configurable
via mount parm closetimeo, but to be safe, decrease default back
to its previous value of 1 second.

Reported-by: Yin Fengwei <fengwei.yin@intel.com>
Reported-by: kernel test robot <yujie.liu@intel.com>
Link: https://lore.kernel.org/lkml/997614df-10d4-af53-9571-edec36b0e2f3@intel.com/
Fixes: 5efdd9122eff ("smb3: allow deferred close timeout to be configurable")
Cc: stable@vger.kernel.org # 6.0+
Tested-by: Yin Fengwei <fengwei.yin@intel.com>
Reviewed-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Reviewed-by: Shyam Prasad N <sprasad@microsoft.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: qla2xxx: Perform lockless command completion in abort path

commit 0367076b0817d5c75dfb83001ce7ce5c64d803a9 upstream.

While adding and removing the controller, the following call trace was
observed:

WARNING: CPU: 3 PID: 623596 at kernel/dma/mapping.c:532 dma_free_attrs+0x33/0x50
CPU: 3 PID: 623596 Comm: sh Kdump: loaded Not tainted 5.14.0-96.el9.x86_64 #1
RIP: 0010:dma_free_attrs+0x33/0x50

Call Trace:
   qla2x00_async_sns_sp_done+0x107/0x1b0 [qla2xxx]
   qla2x00_abort_srb+0x8e/0x250 [qla2xxx]
   ? ql_dbg+0x70/0x100 [qla2xxx]
   __qla2x00_abort_all_cmds+0x108/0x190 [qla2xxx]
   qla2x00_abort_all_cmds+0x24/0x70 [qla2xxx]
   qla2x00_abort_isp_cleanup+0x305/0x3e0 [qla2xxx]
   qla2x00_remove_one+0x364/0x400 [qla2xxx]
   pci_device_remove+0x36/0xa0
   __device_release_driver+0x17a/0x230
   device_release_driver+0x24/0x30
   pci_stop_bus_device+0x68/0x90
   pci_stop_and_remove_bus_device_locked+0x16/0x30
   remove_store+0x75/0x90
   kernfs_fop_write_iter+0x11c/0x1b0
   new_sync_write+0x11f/0x1b0
   vfs_write+0x1eb/0x280
   ksys_write+0x5f/0xe0
   do_syscall_64+0x5c/0x80
   ? do_user_addr_fault+0x1d8/0x680
   ? do_syscall_64+0x69/0x80
   ? exc_page_fault+0x62/0x140
   ? asm_exc_page_fault+0x8/0x30
   entry_SYSCALL_64_after_hwframe+0x44/0xae

The command was completed in the abort path during driver unload with a
lock held, causing the warning in abort path. Hence complete the command
without any lock held.

Reported-by: Lin Li <lilin@redhat.com>
Tested-by: Lin Li <lilin@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Link: https://lore.kernel.org/r/20230313043711.13500-2-njavali@marvell.com
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: John Meneghini <jmeneghi@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: qla2xxx: Synchronize the IOCB count to be in order

commit d3affdeb400f3adc925bd996f3839481f5291839 upstream.

A system hang was observed with the following call trace:

BUG: kernel NULL pointer dereference, address: 0000000000000000
PGD 0 P4D 0
Oops: 0000 [#1] PREEMPT SMP NOPTI
CPU: 15 PID: 86747 Comm: nvme Kdump: loaded Not tainted 6.2.0+ #1
Hardware name: Dell Inc. PowerEdge R6515/04F3CJ, BIOS 2.7.3 03/31/2022
RIP: 0010:__wake_up_common+0x55/0x190
Code: 41 f6 01 04 0f 85 b2 00 00 00 48 8b 43 08 4c 8d
      40 e8 48 8d 43 08 48 89 04 24 48 89 c6\
      49 8d 40 18 48 39 c6 0f 84 e9 00 00 00 <49> 8b 40 18 89 6c 24 14 31
      ed 4c 8d 60 e8 41 8b 18 f6 c3 04 75 5d
RSP: 0018:ffffb05a82afbba0 EFLAGS: 00010082
RAX: 0000000000000000 RBX: ffff8f9b83a00018 RCX: 0000000000000000
RDX: 0000000000000001 RSI: ffff8f9b83a00020 RDI: ffff8f9b83a00018
RBP: 0000000000000001 R08: ffffffffffffffe8 R09: ffffb05a82afbbf8
R10: 70735f7472617473 R11: 5f30307832616c71 R12: 0000000000000001
R13: 0000000000000003 R14: 0000000000000000 R15: 0000000000000000
FS:  00007f815cf4c740(0000) GS:ffff8f9eeed80000(0000)
knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000000000000000 CR3: 000000010633a000 CR4: 0000000000350ee0
Call Trace:
    <TASK>
    __wake_up_common_lock+0x83/0xd0
    qla_nvme_ls_req+0x21b/0x2b0 [qla2xxx]
    __nvme_fc_send_ls_req+0x1b5/0x350 [nvme_fc]
    nvme_fc_xmt_disconnect_assoc+0xca/0x110 [nvme_fc]
    nvme_fc_delete_association+0x1bf/0x220 [nvme_fc]
    ? nvme_remove_namespaces+0x9f/0x140 [nvme_core]
    nvme_do_delete_ctrl+0x5b/0xa0 [nvme_core]
    nvme_sysfs_delete+0x5f/0x70 [nvme_core]
    kernfs_fop_write_iter+0x12b/0x1c0
    vfs_write+0x2a3/0x3b0
    ksys_write+0x5f/0xe0
    do_syscall_64+0x5c/0x90
    ? syscall_exit_work+0x103/0x130
    ? syscall_exit_to_user_mode+0x12/0x30
    ? do_syscall_64+0x69/0x90
    ? exit_to_user_mode_loop+0xd0/0x130
    ? exit_to_user_mode_prepare+0xec/0x100
    ? syscall_exit_to_user_mode+0x12/0x30
    ? do_syscall_64+0x69/0x90
    ? syscall_exit_to_user_mode+0x12/0x30
    ? do_syscall_64+0x69/0x90
    entry_SYSCALL_64_after_hwframe+0x72/0xdc
    RIP: 0033:0x7f815cd3eb97

The IOCB counts are out of order and that would block any commands from
going out and subsequently hang the system. Synchronize the IOCB count to
be in correct order.

Fixes: 5f63a163ed2f ("scsi: qla2xxx: Fix exchange oversubscription for management commands")
Cc: stable@vger.kernel.org
Signed-off-by: Quinn Tran <qutran@marvell.com>
Signed-off-by: Nilesh Javali <njavali@marvell.com>
Link: https://lore.kernel.org/r/20230313043711.13500-3-njavali@marvell.com
Reviewed-by: Himanshu Madhani <himanshu.madhani@oracle.com>
Reviewed-by: John Meneghini <jmeneghi@redhat.com>
Tested-by: Lin Li <lilin@redhat.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

hwmon (it87): Fix voltage scaling for chips with 10.9mV ADCs

[ Upstream commit 968b66ffeb7956acc72836a7797aeb7b2444ec51 ]

Fix voltage scaling for chips that have 10.9mV ADCs, where scaling was
not performed.

Fixes: ead8080351c9 ("hwmon: (it87) Add support for IT8732F")
Signed-off-by: Frank Crawford <frank@crawford.emu.id.au>
Link: https://lore.kernel.org/r/20230318080543.1226700-2-frank@crawford.emu.id.au
[groeck: Update subject and description to focus on bug fix]
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

hwmon: fix potential sensor registration fail if of_node is missing

[ Upstream commit 2315332efcbe7124252f080e03b57d3d2f1f4771 ]

It is not sufficient to check of_node in current device.
In some cases, this would cause the sensor registration to fail.

This patch looks for device's ancestors to find a valid of_node if any.

Fixes: d560168b5d0f ("hwmon: (core) New hwmon registration API")
Signed-off-by: Phinex Hung <phinex@realtek.com>
Link: https://lore.kernel.org/r/20230321060224.3819-1-phinex@realtek.com
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

entry/rcu: Check TIF_RESCHED _after_ delayed RCU wake-up

[ Upstream commit b416514054810cf2d2cc348ae477cea619b64da7 ]

RCU sometimes needs to perform a delayed wake up for specific kthreads
handling offloaded callbacks (RCU_NOCB).  These wakeups are performed
by timers and upon entry to idle (also to guest and to user on nohz_full).

However the delayed wake-up on kernel exit is actually performed after
the thread flags are fetched towards the fast path check for work to
do on exit to user. As a result, and if there is no other pending work
to do upon that kernel exit, the current task will resume to userspace
with TIF_RESCHED set and the pending wake up ignored.

Fix this with fetching the thread flags _after_ the delayed RCU-nocb
kthread wake-up.

Fixes: 47b8ff194c1f ("entry: Explicitly flush pending rcuog wakeup before last rescheduling point")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20230315194349.10798-3-joel@joelfernandes.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf/x86/amd/core: Always clear status for idx

[ Upstream commit 263f5ecaf7080513efc248ec739b6d9e00f4129f ]

The variable 'status' (which contains the unhandled overflow bits) is
not being properly masked in some cases, displaying the following
warning:

  WARNING: CPU: 156 PID: 475601 at arch/x86/events/amd/core.c:972 amd_pmu_v2_handle_irq+0x216/0x270

This seems to be happening because the loop is being continued before
the status bit being unset, in case x86_perf_event_set_period()
returns 0. This is also causing an inconsistency because the "handled"
counter is incremented, but the status bit is not cleaned.

Move the bit cleaning together above, together when the "handled"
counter is incremented.

Fixes: 7685665c390d ("perf/x86/amd/core: Add PerfMonV2 overflow handling")
Signed-off-by: Breno Leitao <leitao@debian.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Reviewed-by: Sandipan Das <sandipan.das@amd.com>
Link: https://lore.kernel.org/r/20230321113338.1669660-1-leitao@debian.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

entry: Fix noinstr warning in __enter_from_user_mode()

[ Upstream commit f87d28673b71b35b248231a2086f9404afbb7f28 ]

__enter_from_user_mode() is triggering noinstr warnings with
CONFIG_DEBUG_PREEMPT due to its call of preempt_count_add() via
ct_state().

The preemption disable isn't needed as interrupts are already disabled.
And the context_tracking_enabled() check in ct_state() also isn't needed
as that's already being done by the CT_WARN_ON().

Just use __ct_state() instead.

Fixes the following warnings:

  vmlinux.o: warning: objtool: enter_from_user_mode+0xba: call to preempt_count_add() leaves .noinstr.text section
  vmlinux.o: warning: objtool: syscall_enter_from_user_mode+0xf9: call to preempt_count_add() leaves .noinstr.text section
  vmlinux.o: warning: objtool: syscall_enter_from_user_mode_prepare+0xc7: call to preempt_count_add() leaves .noinstr.text section
  vmlinux.o: warning: objtool: irqentry_enter_from_user_mode+0xba: call to preempt_count_add() leaves .noinstr.text section

Fixes: 171476775d32 ("context_tracking: Convert state to atomic_t")
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/d8955fa6d68dc955dda19baf13ae014ae27926f5.1677369694.git.jpoimboe@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

platform/chrome: cros_ec_chardev: fix kernel data leak from ioctl

[ Upstream commit b20cf3f89c56b5f6a38b7f76a8128bf9f291bbd3 ]

It is possible to peep kernel page's data by providing larger `insize`
in struct cros_ec_command[1] when invoking EC host commands.

Fix it by using zeroed memory.

[1]: https://elixir.bootlin.com/linux/v6.2/source/include/linux/platform_data/cros_ec_proto.h#L74

Fixes: eda2e30c6684 ("mfd / platform: cros_ec: Miscellaneous character device to talk with the EC")
Signed-off-by: Tzung-Bi Shih <tzungbi@kernel.org>
Reviewed-by: Guenter Roeck <groeck@chromium.org>
Link: https://lore.kernel.org/r/20230324010658.1082361-1-tzungbi@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: HCI: Fix global-out-of-bounds

[ Upstream commit bce56405201111807cc8e4f47c6de3e10b17c1ac ]

To loop a variable-length array, hci_init_stage_sync(stage) considers
that stage[i] is valid as long as stage[i-1].func is valid.
Thus, the last element of stage[].func should be intentionally invalid
as hci_init0[], le_init2[], and others did.
However, amp_init1[] and amp_init2[] have no invalid element, letting
hci_init_stage_sync() keep accessing amp_init1[] over its valid range.
This patch fixes this by adding {} in the last of amp_init1[] and
amp_init2[].

==================================================================
BUG: KASAN: global-out-of-bounds in hci_dev_open_sync (
/v6.2-bzimage/net/bluetooth/hci_sync.c:3154
/v6.2-bzimage/net/bluetooth/hci_sync.c:3343
/v6.2-bzimage/net/bluetooth/hci_sync.c:4418
/v6.2-bzimage/net/bluetooth/hci_sync.c:4609
/v6.2-bzimage/net/bluetooth/hci_sync.c:4689)
Read of size 8 at addr ffffffffaed1ab70 by task kworker/u5:0/1032
CPU: 0 PID: 1032 Comm: kworker/u5:0 Not tainted 6.2.0 #3
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.15.0-1 04
Workqueue: hci1 hci_power_on
Call Trace:
 <TASK>
dump_stack_lvl (/v6.2-bzimage/lib/dump_stack.c:107 (discriminator 1))
print_report (/v6.2-bzimage/mm/kasan/report.c:307
  /v6.2-bzimage/mm/kasan/report.c:417)
? hci_dev_open_sync (/v6.2-bzimage/net/bluetooth/hci_sync.c:3154
  /v6.2-bzimage/net/bluetooth/hci_sync.c:3343
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4418
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4609
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4689)
kasan_report (/v6.2-bzimage/mm/kasan/report.c:184
  /v6.2-bzimage/mm/kasan/report.c:519)
? hci_dev_open_sync (/v6.2-bzimage/net/bluetooth/hci_sync.c:3154
  /v6.2-bzimage/net/bluetooth/hci_sync.c:3343
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4418
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4609
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4689)
hci_dev_open_sync (/v6.2-bzimage/net/bluetooth/hci_sync.c:3154
  /v6.2-bzimage/net/bluetooth/hci_sync.c:3343
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4418
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4609
  /v6.2-bzimage/net/bluetooth/hci_sync.c:4689)
? __pfx_hci_dev_open_sync (/v6.2-bzimage/net/bluetooth/hci_sync.c:4635)
? mutex_lock (/v6.2-bzimage/./arch/x86/include/asm/atomic64_64.h:190
  /v6.2-bzimage/./include/linux/atomic/atomic-long.h:443
  /v6.2-bzimage/./include/linux/atomic/atomic-instrumented.h:1781
  /v6.2-bzimage/kernel/locking/mutex.c:171
  /v6.2-bzimage/kernel/locking/mutex.c:285)
? __pfx_mutex_lock (/v6.2-bzimage/kernel/locking/mutex.c:282)
hci_power_on (/v6.2-bzimage/net/bluetooth/hci_core.c:485
  /v6.2-bzimage/net/bluetooth/hci_core.c:984)
? __pfx_hci_power_on (/v6.2-bzimage/net/bluetooth/hci_core.c:969)
? read_word_at_a_time (/v6.2-bzimage/./include/asm-generic/rwonce.h:85)
? strscpy (/v6.2-bzimage/./arch/x86/include/asm/word-at-a-time.h:62
  /v6.2-bzimage/lib/string.c:161)
process_one_work (/v6.2-bzimage/kernel/workqueue.c:2294)
worker_thread (/v6.2-bzimage/./include/linux/list.h:292
  /v6.2-bzimage/kernel/workqueue.c:2437)
? __pfx_worker_thread (/v6.2-bzimage/kernel/workqueue.c:2379)
kthread (/v6.2-bzimage/kernel/kthread.c:376)
? __pfx_kthread (/v6.2-bzimage/kernel/kthread.c:331)
ret_from_fork (/v6.2-bzimage/arch/x86/entry/entry_64.S:314)
 </TASK>
The buggy address belongs to the variable:
amp_init1+0x30/0x60
The buggy address belongs to the physical page:
page:000000003a157ec6 refcount:1 mapcount:0 mapping:0000000000000000 ia
flags: 0x200000000001000(reserved|node=0|zone=2)
raw: 0200000000001000 ffffea0005054688 ffffea0005054688 000000000000000
raw: 0000000000000000 0000000000000000 00000001ffffffff 000000000000000
page dumped because: kasan: bad access detected
Memory state around the buggy address:
 ffffffffaed1aa00: f9 f9 f9 f9 00 00 00 00 f9 f9 f9 f9 00 00 00 00
 ffffffffaed1aa80: 00 00 00 00 f9 f9 f9 f9 00 00 00 00 00 00 00 00
>ffffffffaed1ab00: 00 f9 f9 f9 f9 f9 f9 f9 00 00 00 00 00 00 f9 f9
                                                             ^
 ffffffffaed1ab80: f9 f9 f9 f9 00 00 00 00 f9 f9 f9 f9 00 00 00 f9
 ffffffffaed1ac00: f9 f9 f9 f9 00 06 f9 f9 f9 f9 f9 f9 00 00 02 f9

This bug is found by FuzzBT, a modified version of Syzkaller.
Other contributors for this bug are Ruoyu Wu and Peng Hui.

Fixes: d0b137062b2d ("Bluetooth: hci_sync: Rework init stages")
Signed-off-by: Sungwoo Kim <iam@sung-woo.kim>
Reviewed-by: Simon Horman <simon.horman@corigine.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: mgmt: Fix MGMT add advmon with RSSI command

[ Upstream commit 1a0291f81529e8044fb29845a0196ba47af894ce ]

The MGMT command: MGMT_OP_ADD_ADV_PATTERNS_MONITOR_RSSI uses variable
length argument. This causes host not able to register advmon with rssi.

This patch has been locally tested by adding monitor with rssi via
btmgmt on a kernel 6.1 machine.

Reviewed-by: Archie Pusaka <apusaka@chromium.org>
Fixes: b338d91703fa ("Bluetooth: Implement support for Mesh")
Signed-off-by: Howard Chung <howardchung@google.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: btsdio: fix use after free bug in btsdio_remove due to unfinished work

[ Upstream commit 1e9ac114c4428fdb7ff4635b45d4f46017e8916f ]

In btsdio_probe, &data->work was bound with btsdio_work.In
btsdio_send_frame, it was started by schedule_work.

If we call btsdio_remove with an unfinished job, there may
be a race condition and cause UAF bug on hdev.

Fixes: ddbaf13e3609 ("[Bluetooth] Add generic driver for Bluetooth SDIO devices")
Signed-off-by: Zheng Wang <zyytlz.wz@163.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: L2CAP: Fix responding with wrong PDU type

[ Upstream commit 9aa9d9473f1550d1936c31259720b3f1f4690576 ]

L2CAP_ECRED_CONN_REQ shall be responded with L2CAP_ECRED_CONN_RSP not
L2CAP_LE_CONN_RSP:

L2CAP LE EATT Server - Reject - run
  Listening for connections
  New client connection with handle 0x002a
  Sending L2CAP Request from client
  Client received response code 0x15
  Unexpected L2CAP response code (expected 0x18)
L2CAP LE EATT Server - Reject - test failed

> ACL Data RX: Handle 42 flags 0x02 dlen 26
      LE L2CAP: Enhanced Credit Connection Request (0x17) ident 1 len 18
        PSM: 39 (0x0027)
        MTU: 64
        MPS: 64
        Credits: 5
        Source CID: 65
        Source CID: 66
        Source CID: 67
        Source CID: 68
        Source CID: 69
< ACL Data TX: Handle 42 flags 0x00 dlen 16
      LE L2CAP: LE Connection Response (0x15) ident 1 len 8
        invalid size
        00 00 00 00 00 00 06 00

L2CAP LE EATT Server - Reject - run
  Listening for connections
  New client connection with handle 0x002a
  Sending L2CAP Request from client
  Client received response code 0x18
L2CAP LE EATT Server - Reject - test passed

Fixes: 15f02b910562 ("Bluetooth: L2CAP: Add initial code for Enhanced Credit Based Mode")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: btqcomsmd: Fix command timeout after setting BD address

[ Upstream commit 5d44ab9e204200a78ad55cdf185aa2bb109b5950 ]

On most devices using the btqcomsmd driver (e.g. the DragonBoard 410c
and other devices based on the Qualcomm MSM8916/MSM8909/... SoCs)
the Bluetooth firmware seems to become unresponsive for a while after
setting the BD address. On recent kernel versions (at least 5.17+)
this often causes timeouts for subsequent commands, e.g. the HCI reset
sent by the Bluetooth core during initialization:

    Bluetooth: hci0: Opcode 0x c03 failed: -110

Unfortunately this behavior does not seem to be documented anywhere.
Experimentation suggests that the minimum necessary delay to avoid
the problem is ~150us. However, to be sure add a sleep for > 1ms
in case it is a bit longer on other firmware versions.

Older kernel versions are likely also affected, although perhaps with
slightly different errors or less probability. Side effects can easily
hide the issue in most cases, e.g. unrelated incoming interrupts that
cause the necessary delay.

Fixes: 1511cc750c3d ("Bluetooth: Introduce Qualcomm WCNSS SMD based HCI driver")
Signed-off-by: Stephan Gerhold <stephan.gerhold@kernkonzept.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Set dcn32 caps.seamless_odm

[ Upstream commit f9537b1fa7fb51c2162bc15ce469cbbf1ca0fbfe ]

[Why & How]
seamless_odm set was not picked up while
merging commit 2d017189e2b3 ("drm/amd/display:
Blank eDP on enable drv if odm enabled")

Fixes: 2d017189e2b3 ("drm/amd/display: Blank eDP on enable drv if odm enabled")
Reviewed-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Acked-by: Qingqing Zhuo <qingqing.zhuo@amd.com>
Signed-off-by: Hersen Wu <hersenxs.wu@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: mdio: thunder: Add missing fwnode_handle_put()

[ Upstream commit b1de5c78ebe9858ccec9d49af2f76724f1d47e3e ]

In device_for_each_child_node(), we should add fwnode_handle_put()
when break out of the iteration device_for_each_child_node()
as it will automatically increase and decrease the refcounter.

Fixes: 379d7ac7ca31 ("phy: mdio-thunder: Add driver for Cavium Thunder SoC MDIO buses.")
Signed-off-by: Liang He <windhl@126.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>