arm64: dts: qcom: sm8450: fix soundwire controllers node name

[ Upstream commit 11fcb81373de52eeb1d3ff135a8d24a4b18978d3 ]

Fix the following dt bindings check:
arch/arm64/boot/dts/qcom/sm8450-hdk.dtb: soundwire-controller@31f0000: $nodename:0: 'soundwire-controller@31f0000' does not match '^soundwire(@.*)?$'
        from schema $id: http://devicetree.org/schemas/soundwire/qcom,soundwire.yaml#

Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20231106-topic-sm8450-upstream-soundwire-bindings-fix-v1-1-41d4844a5a7d@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: dts: qcom: sm8550: fix soundwire controllers node name

[ Upstream commit 07c88da81caf0e72c3690b689d30f0d325cfeff4 ]

Fix the following dt bindings check:
arch/arm64/boot/dts/qcom/sm8550-mtp.dtb: soundwire-controller@6ab0000: $nodename:0: 'soundwire-controller@6ab0000' does not match '^soundwire(@.*)?$'
from schema $id: http://devicetree.org/schemas/soundwire/qcom,soundwire.yaml#

Signed-off-by: Neil Armstrong <neil.armstrong@linaro.org>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20231106-topic-sm8550-upstream-soundwire-bindings-fix-v1-1-4ded91c805a1@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: mvmdio: Avoid excessive sleeps in polled mode

[ Upstream commit 7dd12fe34686d89c332b1a05104d18d728591f0a ]

Before this change, when operating in polled mode, i.e. no IRQ is
available, every individual C45 access would be hit with a 150us sleep
after the bus access.

For example, on a board with a CN9130 SoC connected to an MV88X3310
PHY, a single C45 read would take around 165us:

    root@infix:~$ mdio f212a600.mdio-mii mmd 4:1 bench 0xc003
    Performed 1000 reads in 165ms

By replacing the long sleep with a tighter poll loop, we observe a 10x
increase in bus throughput:

    root@infix:~$ mdio f212a600.mdio-mii mmd 4:1 bench 0xc003
    Performed 1000 reads in 15ms

Signed-off-by: Tobias Waldekranz <tobias@waldekranz.com>
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Tested-by: Andrew Lunn <andrew@lunn.ch>
Link: https://lore.kernel.org/r/20231204100811.2708884-3-tobias@waldekranz.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

minmax: relax check to allow comparison between unsigned arguments and signed constants

commit 867046cc7027703f60a46339ffde91a1970f2901 upstream.

Allow (for example) min(unsigned_var, 20).

The opposite min(signed_var, 20u) is still errored.

Since a comparison between signed and unsigned never makes the unsigned
value negative it is only necessary to adjust the __types_ok() test.

Link: https://lkml.kernel.org/r/633b64e2f39e46bb8234809c5595b8c7@AcuMS.aculab.com
Signed-off-by: David Laight <david.laight@aculab.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

minmax: allow comparisons of 'int' against 'unsigned char/short'

commit 4ead534fba42fc4fd41163297528d2aa731cd121 upstream.

Since 'unsigned char/short' get promoted to 'signed int' it is safe to
compare them against an 'int' value.

Link: https://lkml.kernel.org/r/8732ef5f809c47c28a7be47c938b28d4@AcuMS.aculab.com
Signed-off-by: David Laight <david.laight@aculab.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

minmax: fix indentation of __cmp_once() and __clamp_once()

commit f4b84b2ff851f01d0fac619eadef47eb41648534 upstream.

Remove the extra indentation and align continuation markers.

Link: https://lkml.kernel.org/r/bed41317a05c498ea0209eafbcab45a5@AcuMS.aculab.com
Signed-off-by: David Laight <david.laight@aculab.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

minmax: allow min()/max()/clamp() if the arguments have the same signedness.

commit d03eba99f5bf7cbc6e2fdde3b6fa36954ad58e09 upstream.

The type-check in min()/max() is there to stop unexpected results if a
negative value gets converted to a large unsigned value.  However it also
rejects 'unsigned int' v 'unsigned long' compares which are common and
never problematc.

Replace the 'same type' check with a 'same signedness' check.

The new test isn't itself a compile time error, so use static_assert() to
report the error and give a meaningful error message.

Due to the way builtin_choose_expr() works detecting the error in the
'non-constant' side (where static_assert() can be used) also detects
errors when the arguments are constant.

Link: https://lkml.kernel.org/r/fe7e6c542e094bfca655abcd323c1c98@AcuMS.aculab.com
Signed-off-by: David Laight <david.laight@aculab.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

minmax: add umin(a, b) and umax(a, b)

commit 80fcac55385ccb710d33a20dc1caaef29bd5a921 upstream.

Patch series "minmax: Relax type checks in min() and max()", v4.

The min() (etc) functions in minmax.h require that the arguments have
exactly the same types.

However when the type check fails, rather than look at the types and fix
the type of a variable/constant, everyone seems to jump on min_t().  In
reality min_t() ought to be rare - when something unusual is being done,
not normality.

The orginal min() (added in 2.4.9) replaced several inline functions and
included the type - so matched the implicit casting of the function call.
This was renamed min_t() in 2.4.10 and the current min() added.  There is
no actual indication that the conversion of negatve values to large
unsigned values has ever been an actual problem.

A quick grep shows 5734 min() and 4597 min_t().  Having the casts on
almost half of the calls shows that something is clearly wrong.

If the wrong type is picked (and it is far too easy to pick the type of
the result instead of the larger input) then significant bits can get
discarded.

Pretty much the worst example is in the derived clamp_val(), consider:
        unsigned char x = 200u;
        y = clamp_val(x, 10u, 300u);

I also suspect that many of the min_t(u16, ...) are actually wrong.  For
example copy_data() in printk_ringbuffer.c contains:

        data_size = min_t(u16, buf_size, len);

Here buf_size is 'unsigned int' and len 'u16', pass a 64k buffer (can you
prove that doesn't happen?) and no data is returned.  Apparantly it did -
and has since been fixed.

The only reason that most of the min_t() are 'fine' is that pretty much
all the values in the kernel are between 0 and INT_MAX.

Patch 1 adds umin(), this uses integer promotions to convert both
arguments to 'unsigned long long'.  It can be used to compare a signed
type that is known to contain a non-negative value with an unsigned type.
The compiler typically optimises it all away.  Added first so that it can
be referred to in patch 2.

Patch 2 replaces the 'same type' check with a 'same signedness' one.  This
makes min(unsigned_int_var, sizeof()) be ok.  The error message is also
improved and will contain the expanded form of both arguments (useful for
seeing how constants are defined).

Patch 3 just fixes some whitespace.

Patch 4 allows comparisons of 'unsigned char' and 'unsigned short' to
signed types.  The integer promotion rules convert them both to 'signed
int' prior to the comparison so they can never cause a negative value be
converted to a large positive one.

Patch 5 (rewritted for v4) allows comparisons of unsigned values against
non-negative constant integer expressions.  This makes
min(unsigned_int_var, 4) be ok.

The only common case that is still errored is the comparison of signed
values against unsigned constant integer expressions below __INT_MAX__.
Typcally min(int_val, sizeof (foo)), the real fix for this is casting the
constant: min(int_var, (int)sizeof (foo)).

With all the patches applied pretty much all the min_t() could be replaced
by min(), and most of the rest by umin().  However they all need careful
inspection due to code like:

        sz = min_t(unsigned char, sz - 1, LIM - 1) + 1;

which converts 0 to LIM.

This patch (of 6):

umin() and umax() can be used when min()/max() errors a signed v unsigned
compare when the signed value is known to be non-negative.

Unlike min_t(some_unsigned_type, a, b) umin() will never mask off high
bits if an inappropriate type is selected.

The '+ 0u + 0ul + 0ull' may look strange.
The '+ 0u' is needed for 'signed int' on 64bit systems.
The '+ 0ul' is needed for 'signed long' on 32bit systems.
The '+ 0ull' is needed for 'signed long long'.

Link: https://lkml.kernel.org/r/b97faef60ad24922b530241c5d7c933c@AcuMS.aculab.com
Link: https://lkml.kernel.org/r/41d93ca827a248698ec64bf57e0c05a5@AcuMS.aculab.com
Signed-off-by: David Laight <david.laight@aculab.com>
Cc: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: Jason A. Donenfeld <Jason@zx2c4.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Matthew Wilcox (Oracle) <willy@infradead.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

minmax: fix header inclusions

commit f6e9d38f8eb00ac8b52e6d15f6aa9bcecacb081b upstream.

BUILD_BUG_ON*() macros are defined in build_bug.h.  Include it.  Replace
compiler_types.h by compiler.h, which provides the former, to have a
definition of the __UNIQUE_ID().

Link: https://lkml.kernel.org/r/20230912092355.79280-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Herve Codina <herve.codina@bootlin.com>
Cc: Rasmus Villemoes <linux@rasmusvillemoes.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>

minmax: deduplicate __unconst_integer_typeof()

commit 5e57418a2031cd5e1863efdf3d7447a16a368172 upstream.

It appears that compiler_types.h already have an implementation of the
__unconst_integer_typeof() called __unqual_scalar_typeof().  Use it
instead of the copy.

Link: https://lkml.kernel.org/r/20230911154913.4176033-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Acked-by: Herve Codina <herve.codina@bootlin.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

scsi: libfc: Fix up timeout error in fc_fcp_rec_error()

[ Upstream commit 53122a49f49796beb2c4a1bb702303b66347e29f ]

We should set the status to FC_TIMED_OUT when a timeout error is passed to
fc_fcp_rec_error().

Signed-off-by: Hannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/20231129165832.224100-3-hare@kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: libfc: Don't schedule abort twice

[ Upstream commit b57c4db5d23b9df0118a25e2441c9288edd73710 ]

The current FC error recovery is sending up to three REC (recovery) frames
in 10 second intervals, and as a final step sending an ABTS after 30
seconds for the command itself.  Unfortunately sending an ABTS is also the
action for the SCSI abort handler, and the default timeout for SCSI
commands is also 30 seconds. This causes two ABTS to be scheduled, with the
libfc one slightly earlier. The ABTS scheduled by SCSI EH then sees the
command to be already aborted, and will always return with a 'GOOD' status
irrespective on the actual result from the first ABTS.  This causes the
SCSI EH abort handler to always succeed, and SCSI EH never to be engaged.
Fix this by not issuing an ABTS when a SCSI command is present for the
exchange, but rather wait for the abort scheduled from SCSI EH.  And warn
if an abort is already scheduled to avoid similar errors in the future.

Signed-off-by: Hannes Reinecke <hare@suse.de>
Link: https://lore.kernel.org/r/20231129165832.224100-2-hare@kernel.org
Reviewed-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath12k: fix and enable AP mode for WCN7850

[ Upstream commit ed7e818a7b501012038d6bc6fedadaf7375a380a ]

For AP mode, the peer is created earlier in ath12k_mac_op_add_interface() but
ath12k_mac_op_assign_vif_chanctx() will try to create peer again.  Then an
error will return which makes AP mode startup fail.

Kernel log:

[ 5017.665006] ath12k_pci 0000:04:00.0: failed to create peer after vdev start delay: -22

wpa_supplicant log:

Failed to set beacon parameters
Interface initialization failed
wls1: interface state UNINITIALIZED->DISABLED
wls1: AP-DISABLED
wls1: Unable to setup interface.
Failed to initialize AP interface
wls1: interface state DISABLED->DISABLED
wls1: AP-DISABLED

So fix this check and enable AP mode for WCN7850, as now AP mode works normally.

Tested-on: WCN7850 hw2.0 PCI WLAN.HMT.1.0-03427-QCAHMTSWPL_V1.0_V2.0_SILICONZ-1.15378.4

Signed-off-by: Kang Yang <quic_kangyang@quicinc.com>
Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20231121022459.17209-1-quic_kangyang@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Set need_defer as false when clearing fd array during map free

[ Upstream commit 79d93b3c6ffd79abcd8e43345980aa1e904879c4 ]

Both map deletion operation, map release and map free operation use
fd_array_map_delete_elem() to remove the element from fd array and
need_defer is always true in fd_array_map_delete_elem(). For the map
deletion operation and map release operation, need_defer=true is
necessary, because the bpf program, which accesses the element in fd
array, may still alive. However for map free operation, it is certain
that the bpf program which owns the fd array has already been exited, so
setting need_defer as false is appropriate for map free operation.

So fix it by adding need_defer parameter to bpf_fd_array_map_clear() and
adding a new helper __fd_array_map_delete_elem() to handle the map
deletion, map release and map free operations correspondingly.

Signed-off-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/r/20231204140425.1480317-4-houtao@huaweicloud.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Check rcu_read_lock_trace_held() before calling bpf map helpers

[ Upstream commit 169410eba271afc9f0fb476d996795aa26770c6d ]

These three bpf_map_{lookup,update,delete}_elem() helpers are also
available for sleepable bpf program, so add the corresponding lock
assertion for sleepable bpf program, otherwise the following warning
will be reported when a sleepable bpf program manipulates bpf map under
interpreter mode (aka bpf_jit_enable=0):

  WARNING: CPU: 3 PID: 4985 at kernel/bpf/helpers.c:40 ......
  CPU: 3 PID: 4985 Comm: test_progs Not tainted 6.6.0+ #2
  Hardware name: QEMU Standard PC (i440FX + PIIX, 1996) ......
  RIP: 0010:bpf_map_lookup_elem+0x54/0x60
  ......
  Call Trace:
   <TASK>
   ? __warn+0xa5/0x240
   ? bpf_map_lookup_elem+0x54/0x60
   ? report_bug+0x1ba/0x1f0
   ? handle_bug+0x40/0x80
   ? exc_invalid_op+0x18/0x50
   ? asm_exc_invalid_op+0x1b/0x20
   ? __pfx_bpf_map_lookup_elem+0x10/0x10
   ? rcu_lockdep_current_cpu_online+0x65/0xb0
   ? rcu_is_watching+0x23/0x50
   ? bpf_map_lookup_elem+0x54/0x60
   ? __pfx_bpf_map_lookup_elem+0x10/0x10
   ___bpf_prog_run+0x513/0x3b70
   __bpf_prog_run32+0x9d/0xd0
   ? __bpf_prog_enter_sleepable_recur+0xad/0x120
   ? __bpf_prog_enter_sleepable_recur+0x3e/0x120
   bpf_trampoline_6442580665+0x4d/0x1000
   __x64_sys_getpgid+0x5/0x30
   ? do_syscall_64+0x36/0xb0
   entry_SYSCALL_64_after_hwframe+0x6e/0x76
   </TASK>

Signed-off-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/r/20231204140425.1480317-2-houtao@huaweicloud.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtw89: fix misbehavior of TX beacon in concurrent mode

[ Upstream commit 756b31203d482d2dd1aa6c208978b0410dc7530f ]

In concurrent mode, when STA interface is scanning, it causes
AP interface TX beacon on wrong channel. We modified it to scan
with the operating channel when one of the interfaces is already
connected. Additionally, STA interface need to stop scan when AP
interface is starting to avoid TX beacon on wrong channel. Finally,
AP interface need to stop TX beacon when STA interface is scanning
and switching to non-OP channel,This prevent other device to get
beacons on wrong channel.

Signed-off-by: Chih-Kang Chang <gary.chang@realtek.com>
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20231129070046.18443-5-pkshih@realtek.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath11k: fix race due to setting ATH11K_FLAG_EXT_IRQ_ENABLED too early

[ Upstream commit 5082b3e3027eae393a4e86874bffb4ce3f83c26e ]

We are seeing below error randomly in the case where only
one MSI vector is configured:

kernel: ath11k_pci 0000:03:00.0: wmi command 16387 timeout

The reason is, currently, in ath11k_pcic_ext_irq_enable(),
ATH11K_FLAG_EXT_IRQ_ENABLED is set before NAPI is enabled.
This results in a race condition: after
ATH11K_FLAG_EXT_IRQ_ENABLED is set but before NAPI enabled,
CE interrupt breaks in. Since IRQ is shared by CE and data
path, ath11k_pcic_ext_interrupt_handler() is also called
where we call disable_irq_nosync() to disable IRQ. Then
napi_schedule() is called but it does nothing because NAPI
is not enabled at that time, meaning
ath11k_pcic_ext_grp_napi_poll() will never run, so we have
no chance to call enable_irq() to enable IRQ back. Finally
we get above error.

Fix it by setting ATH11K_FLAG_EXT_IRQ_ENABLED after all
NAPI and IRQ work are done. With the fix, we are sure that
by the time ATH11K_FLAG_EXT_IRQ_ENABLED is set, NAPI is
enabled.

Note that the fix above also introduce some side effects:
if ath11k_pcic_ext_interrupt_handler() breaks in after NAPI
enabled but before ATH11K_FLAG_EXT_IRQ_ENABLED set, nothing
will be done by the handler this time, the work will be
postponed till the next time the IRQ fires.

Tested-on: WCN6855 hw2.1 PCI WLAN.HSP.1.1-03125-QCAHSPSWPL_V1_V2_SILICONZ_LITE-3.6510.23

Signed-off-by: Baochen Qiang <quic_bqiang@quicinc.com>
Acked-by: Jeff Johnson <quic_jjohnson@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20231117003919.26218-1-quic_bqiang@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath9k: Fix potential array-index-out-of-bounds read in ath9k_htc_txstatus()

[ Upstream commit 2adc886244dff60f948497b59affb6c6ebb3c348 ]

Fix an array-index-out-of-bounds read in ath9k_htc_txstatus(). The bug
occurs when txs->cnt, data from a URB provided by a USB device, is
bigger than the size of the array txs->txstatus, which is
HTC_MAX_TX_STATUS. WARN_ON() already checks it, but there is no bug
handling code after the check. Make the function return if that is the
case.

Found by a modified version of syzkaller.

UBSAN: array-index-out-of-bounds in htc_drv_txrx.c
index 13 is out of range for type '__wmi_event_txstatus [12]'
Call Trace:
 ath9k_htc_txstatus
 ath9k_wmi_event_tasklet
 tasklet_action_common
 __do_softirq
 irq_exit_rxu
 sysvec_apic_timer_interrupt

Signed-off-by: Minsuk Kang <linuxlovemin@yonsei.ac.kr>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20231113065756.1491991-1-linuxlovemin@yonsei.ac.kr
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Fix a few selftest failures due to llvm18 change

[ Upstream commit b16904fd9f01b580db357ef2b1cc9e86d89576c2 ]

With latest upstream llvm18, the following test cases failed:

  $ ./test_progs -j
  #13/2    bpf_cookie/multi_kprobe_link_api:FAIL
  #13/3    bpf_cookie/multi_kprobe_attach_api:FAIL
  #13      bpf_cookie:FAIL
  #77      fentry_fexit:FAIL
  #78/1    fentry_test/fentry:FAIL
  #78      fentry_test:FAIL
  #82/1    fexit_test/fexit:FAIL
  #82      fexit_test:FAIL
  #112/1   kprobe_multi_test/skel_api:FAIL
  #112/2   kprobe_multi_test/link_api_addrs:FAIL
  [...]
  #112     kprobe_multi_test:FAIL
  #356/17  test_global_funcs/global_func17:FAIL
  #356     test_global_funcs:FAIL

Further analysis shows llvm upstream patch [1] is responsible for the above
failures. For example, for function bpf_fentry_test7() in net/bpf/test_run.c,
without [1], the asm code is:

  0000000000000400 <bpf_fentry_test7>:
     400: f3 0f 1e fa                   endbr64
     404: e8 00 00 00 00                callq   0x409 <bpf_fentry_test7+0x9>
     409: 48 89 f8                      movq    %rdi, %rax
     40c: c3                            retq
     40d: 0f 1f 00                      nopl    (%rax)

... and with [1], the asm code is:

  0000000000005d20 <bpf_fentry_test7.specialized.1>:
    5d20: e8 00 00 00 00                callq   0x5d25 <bpf_fentry_test7.specialized.1+0x5>
    5d25: c3                            retq

... and <bpf_fentry_test7.specialized.1> is called instead of <bpf_fentry_test7>
and this caused test failures for #13/#77 etc. except #356.

For test case #356/17, with [1] (progs/test_global_func17.c)), the main prog
looks like:

  0000000000000000 <global_func17>:
       0:       b4 00 00 00 2a 00 00 00 w0 = 0x2a
       1:       95 00 00 00 00 00 00 00 exit

... which passed verification while the test itself expects a verification
failure.

Let us add 'barrier_var' style asm code in both places to prevent function
specialization which caused selftests failure.

  [1] https://github.com/llvm/llvm-project/pull/72903

Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20231127050342.1945270-1-yonghong.song@linux.dev
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: imx7s: Fix nand-controller #size-cells

[ Upstream commit 4aadb841ed49bada1415c48c44d21f5b69e01299 ]

nand-controller.yaml bindings says #size-cells shall be set to 0.
Fixes the dtbs_check warning:
arch/arm/boot/dts/nxp/imx/imx7s-mba7.dtb: nand-controller@33002000:
 #size-cells:0:0: 0 was expected
  from schema $id: http://devicetree.org/schemas/mtd/gpmi-nand.yaml#

Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: imx7s: Fix lcdif compatible

[ Upstream commit 5f55da4cc37051cda600ea870ce8cf29f1297715 ]

imx7d-lcdif is compatible to imx6sx-lcdif. MXSFB_V6 supports overlay
by using LCDC_AS_CTRL register. This registers used by overlay plane:
* LCDC_AS_CTRL
* LCDC_AS_BUF
* LCDC_AS_NEXT_BUF
are listed in i.MX7D RM as well.

Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: imx7d: Fix coresight funnel ports

[ Upstream commit 0d4ac04fa7c3f6dc263dba6f575a2ec7a2d4eca8 ]

imx7d uses two ports for 'in-ports', so the syntax port@<num> has to
be used. imx7d has both port and port@1 nodes present, raising these
error:
funnel@30041000: in-ports: More than one condition true in oneOf schema
funnel@30041000: Unevaluated properties are not allowed
('in-ports' was unexpected)

Fix this by also using port@0 for imx7s as well.

Signed-off-by: Alexander Stein <alexander.stein@ew.tq-group.com>
Signed-off-by: Shawn Guo <shawnguo@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: arcmsr: Support new PCI device IDs 1883 and 1886

[ Upstream commit 41c8a1a1e90fa4721f856bf3cf71211fd16d6434 ]

Add support for Areca RAID controllers with PCI device IDs 1883 and 1886.

Signed-off-by: ching Huang <ching2048@areca.com.tw>
Link: https://lore.kernel.org/r/7732e743eaad57681b1552eec9c6a86c76dbe459.camel@areca.com.tw
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: mpi3mr: Add PCI checks where SAS5116 diverges from SAS4116

[ Upstream commit c9260ff28ee561fca5f96425c9328a9698e8427b ]

Add PCI IDs checks for the cases where SAS5116 diverges from SAS4116 in
behavior.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Link: https://lore.kernel.org/r/20231123160132.4155-3-sumit.saxena@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: mpi3mr: Add support for SAS5116 PCI IDs

[ Upstream commit 6fa21eab82be57a3ad2470fac27b982793805336 ]

Add support for Broadcom's SAS5116 IO/RAID controllers PCI IDs.

Signed-off-by: Sumit Saxena <sumit.saxena@broadcom.com>
Link: https://lore.kernel.org/r/20231123160132.4155-2-sumit.saxena@broadcom.com
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: usb: ax88179_178a: avoid two consecutive device resets

[ Upstream commit d2689b6a86b9d23574bd4b654bf770b6034e2c7e ]

The device is always reset two consecutive times (ax88179_reset is called
twice), one from usbnet_probe during the device binding and the other from
usbnet_open.

Remove the non-necessary reset during the device binding and let the reset
operation from open to keep the normal behavior (tested with generic ASIX
Electronics Corp. AX88179 Gigabit Ethernet device).

Reported-by: Herb Wei <weihao.bj@ieisystem.com>
Tested-by: Herb Wei <weihao.bj@ieisystem.com>
Signed-off-by: Jose Ignacio Tornos Martinez <jtornosm@redhat.com>
Link: https://lore.kernel.org/r/20231120121239.54504-1-jtornosm@redhat.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bonding: return -ENOMEM instead of BUG in alb_upper_dev_walk

[ Upstream commit d6b83f1e3707c4d60acfa58afd3515e17e5d5384 ]

If failed to allocate "tags" or could not find the final upper device from
start_dev's upper list in bond_verify_device_path(), only the loopback
detection of the current upper device should be affected, and the system is
no need to be panic.
So return -ENOMEM in alb_upper_dev_walk to stop walking, print some warn
information when failed to allocate memory for vlan tags in
bond_verify_device_path.

I also think that the following function calls
netdev_walk_all_upper_dev_rcu
---->>>alb_upper_dev_walk
---------->>>bond_verify_device_path
From this way, "end device" can eventually be obtained from "start device"
in bond_verify_device_path, IS_ERR(tags) could be instead of
IS_ERR_OR_NULL(tags) in alb_upper_dev_walk.

Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Acked-by: Jay Vosburgh <jay.vosburgh@canonical.com>
Link: https://lore.kernel.org/r/20231118081653.1481260-1-shaozhengchao@huawei.com
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI: Add no PM reset quirk for NVIDIA Spectrum devices

[ Upstream commit 3ed48c80b28d8dcd584d6ddaf00c75b7673e1a05 ]

Spectrum-{1,2,3,4} devices report that a D3hot->D0 transition causes a
reset (i.e., they advertise NoSoftRst-). However, this transition does
not have any effect on the device: It continues to be operational and
network ports remain up. Advertising this support makes it seem as if a
PM reset is viable for these devices. Mark it as unavailable to skip it
when testing reset methods.

Before:

 # cat /sys/bus/pci/devices/0000\:03\:00.0/reset_method
 pm bus

After:

 # cat /sys/bus/pci/devices/0000\:03\:00.0/reset_method
 bus

Signed-off-by: Ido Schimmel <idosch@nvidia.com>
Acked-by: Bjorn Helgaas <bhelgaas@google.com>
Signed-off-by: Petr Machata <petrm@nvidia.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: phy: micrel: fix ts_info value in case of no phc

[ Upstream commit 915d25a9d69be969c1cc6c1dd0c3861f6da7b55e ]

In case of no phc we should not return SOFTWARE TIMESTAMPING flags as we do
not know whether the netdev supports of timestamping.
Remove it from the lan8841_ts_info and simply return 0.

Signed-off-by: Kory Maincent <kory.maincent@bootlin.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: samsung: s5pv210: fix camera unit addresses/ranges

[ Upstream commit 07e6a553c2f1d385edfc9185081dee442a9dd38d ]

The camera node has both unit address and children within the same bus
mapping, thus needs proper ranges property to fix dtc W=1 warnings:

  Warning (unit_address_vs_reg): /soc/camera@fa600000: node has a unit name, but no reg or ranges property
  Warning (simple_bus_reg): /soc/camera@fa600000: missing or empty reg/ranges property

Subtract 0xfa600000 from all its children nodes.  No functional impact
expected.

Link: https://lore.kernel.org/r/20230722121719.150094-3-krzysztof.kozlowski@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: samsung: exynos4: fix camera unit addresses/ranges

[ Upstream commit ba2a45a48503665f7e8eeec51f8b40456566b0cd ]

The camera node has both unit address and children within the same bus
mapping, thus needs proper ranges property to fix dtc W=1 warnings:

  Warning (unit_address_vs_reg): /soc/camera@11800000: node has a unit name, but no reg or ranges property
  Warning (simple_bus_reg): /soc/camera@11800000: missing or empty reg/ranges property

Subtract 0x11800000 from all its children nodes.  No functional impact
expected.

Link: https://lore.kernel.org/r/20230722121719.150094-2-krzysztof.kozlowski@linaro.org
Signed-off-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: lpfc: Fix possible file string name overflow when updating firmware

[ Upstream commit f5779b529240b715f0e358489ad0ed933bf77c97 ]

Because file_name and phba->ModelName are both declared a size 80 bytes,
the extra ".grp" file extension could cause an overflow into file_name.

Define a ELX_FW_NAME_SIZE macro with value 84.  84 incorporates the 4 extra
characters from ".grp".  file_name is changed to be declared as a char and
initialized to zeros i.e. null chars.

Signed-off-by: Justin Tee <justin.tee@broadcom.com>
Link: https://lore.kernel.org/r/20231031191224.150862-3-justintee8345@gmail.com
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>

soc: xilinx: fix unhandled SGI warning message

[ Upstream commit 9c6724abf969251af53cdae525ad8100ec78d3c2 ]

Xen broadcasts SGI to each VM when multiple VMs run on Xen hypervisor. In
such case spurious SGI is expected if one event is registered by one VM and
not registered by another VM. We let users know that Unhandled SGI is not
error and expected if kernel is running on Xen hypervisor.

Signed-off-by: Tanmay Shah <tanmay.shah@xilinx.com>
Signed-off-by: Radhey Shyam Pandey <radhey.shyam.pandey@amd.com>
Link: https://lore.kernel.org/r/1698431039-2734260-1-git-send-email-radhey.shyam.pandey@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

soc: xilinx: Fix for call trace due to the usage of smp_processor_id()

[ Upstream commit daed80ed07580e5adc0e6d8bc79933a35154135a ]

When preemption is enabled in kernel and if any task which can be
preempted should not use smp_processor_id() directly, since CPU
switch can happen at any time, the previous value of cpu_id
differs with current cpu_id. As a result we see the below call trace
during xlnx_event_manager_probe.

[ 6.140197] dump_backtrace+0x0/0x190
[ 6.143884] show_stack+0x18/0x40
[ 6.147220] dump_stack_lvl+0x7c/0xa0
[ 6.150907] dump_stack+0x18/0x34
[ 6.154241] check_preemption_disabled+0x124/0x134
[ 6.159068] debug_smp_processor_id+0x20/0x2c
[ 6.163453] xlnx_event_manager_probe+0x48/0x250

To protect cpu_id, It is recommended to use get_cpu()/put_cpu()
to disable preemption, get the cpu_id and enable preemption respectively.
(For Reference, Documentation/locking/preempt-locking.rst and
Documentation/kernel-hacking/hacking.rst)

Use preempt_disable()/smp_processor_id()/preempt_enable()
API's to achieve the same.

Signed-off-by: HariBabu Gattem <haribabu.gattem@xilinx.com>
Signed-off-by: Jay Buddhabhatti <jay.buddhabhatti@amd.com>
Link: https://lore.kernel.org/r/20231027055622.21544-1-jay.buddhabhatti@amd.com
Signed-off-by: Michal Simek <michal.simek@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: qcom: msm8660: fix PMIC node labels

[ Upstream commit c6d86aa8a12194d1c9c2f9108910a46c8a3ddc90 ]

Change PM8058 node labels to start with pm8058_ prefix, following other
Qualcomm PMIC device nodes.

Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20230928110309.1212221-11-dmitry.baryshkov@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: qcom: mdm9615: fix PMIC node labels

[ Upstream commit 0e4688cd4ee6efbeae2b31f75e16961fd7f72735 ]

Change PM8018 node labels to start with pm8018_ prefix, following other
Qualcomm PMIC device nodes.

Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20230928110309.1212221-10-dmitry.baryshkov@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: dts: qcom: strip prefix from PMIC files

[ Upstream commit 455a4c46e023ab84263eae0fc7acca9a5ee8b7ac ]

As the vendor DTS files were moved to per-vendor subdirs, there no need
to use common prefixes. Drop the `qcom-' prefix from PMIC dtsi file.
This makes 32-bit qcom/ dts files closer to arm64 ones.

Reviewed-by: Konrad Dybcio <konrad.dybcio@linaro.org>
Signed-off-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Link: https://lore.kernel.org/r/20230928110309.1212221-8-dmitry.baryshkov@linaro.org
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/bpf: Fix issues in setup_classid_environment()

[ Upstream commit 4849775587844e44d215289c425bcd70f315efe7 ]

If the net_cls subsystem is already mounted, attempting to mount it again
in setup_classid_environment() will result in a failure with the error code
EBUSY. Despite this, tmpfs will have been successfully mounted at
/sys/fs/cgroup/net_cls. Consequently, the /sys/fs/cgroup/net_cls directory
will be empty, causing subsequent setup operations to fail.

Here's an error log excerpt illustrating the issue when net_cls has already
been mounted at /sys/fs/cgroup/net_cls prior to running
setup_classid_environment():

- Before that change

  $ tools/testing/selftests/bpf/test_progs --name=cgroup_v1v2
  test_cgroup_v1v2:PASS:server_fd 0 nsec
  test_cgroup_v1v2:PASS:client_fd 0 nsec
  test_cgroup_v1v2:PASS:cgroup_fd 0 nsec
  test_cgroup_v1v2:PASS:server_fd 0 nsec
  run_test:PASS:skel_open 0 nsec
  run_test:PASS:prog_attach 0 nsec
  test_cgroup_v1v2:PASS:cgroup-v2-only 0 nsec
  (cgroup_helpers.c:248: errno: No such file or directory) Opening Cgroup Procs: /sys/fs/cgroup/net_cls/cgroup.procs
  (cgroup_helpers.c:540: errno: No such file or directory) Opening cgroup classid: /sys/fs/cgroup/net_cls/cgroup-test-work-dir/net_cls.classid
  run_test:PASS:skel_open 0 nsec
  run_test:PASS:prog_attach 0 nsec
  (cgroup_helpers.c:248: errno: No such file or directory) Opening Cgroup Procs: /sys/fs/cgroup/net_cls/cgroup-test-work-dir/cgroup.procs
  run_test:FAIL:join_classid unexpected error: 1 (errno 2)
  test_cgroup_v1v2:FAIL:cgroup-v1v2 unexpected error: -1 (errno 2)
  (cgroup_helpers.c:248: errno: No such file or directory) Opening Cgroup Procs: /sys/fs/cgroup/net_cls/cgroup.procs
  #44      cgroup_v1v2:FAIL
  Summary: 0/0 PASSED, 0 SKIPPED, 1 FAILED

- After that change
  $ tools/testing/selftests/bpf/test_progs --name=cgroup_v1v2
  #44      cgroup_v1v2:OK
  Summary: 1/0 PASSED, 0 SKIPPED, 0 FAILED

Signed-off-by: Yafang Shao <laoar.shao@gmail.com>
Link: https://lore.kernel.org/r/20231111090034.4248-3-laoar.shao@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rt2x00: correct wrong BBP register in RxDCOC calibration

[ Upstream commit 50da74e1e8b682853d1e07fc8bbe3a0774ae5e09 ]

Refer to Mediatek vendor driver RxDCOC_Calibration() function, when
performing gainfreeze calibration, we should write register 140
instead of 141. This fix can reduce the total calibration time from
6 seconds to 1 second.

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/TYAP286MB0315B13B89DF57B6B27BB854BCAFA@TYAP286MB0315.JPNP286.PROD.OUTLOOK.COM
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/bpf: Fix pyperf180 compilation failure with clang18

[ Upstream commit 100888fb6d8a185866b1520031ee7e3182b173de ]

With latest clang18 (main branch of llvm-project repo), when building bpf selftests,
    [~/work/bpf-next (master)]$ make -C tools/testing/selftests/bpf LLVM=1 -j

The following compilation error happens:
    fatal error: error in backend: Branch target out of insn range
    ...
    Stack dump:
    0.      Program arguments: clang -g -Wall -Werror -D__TARGET_ARCH_x86 -mlittle-endian
      -I/home/yhs/work/bpf-next/tools/testing/selftests/bpf/tools/include
      -I/home/yhs/work/bpf-next/tools/testing/selftests/bpf -I/home/yhs/work/bpf-next/tools/include/uapi
      -I/home/yhs/work/bpf-next/tools/testing/selftests/usr/include -idirafter
      /home/yhs/work/llvm-project/llvm/build.18/install/lib/clang/18/include -idirafter /usr/local/include
      -idirafter /usr/include -Wno-compare-distinct-pointer-types -DENABLE_ATOMICS_TESTS -O2 --target=bpf
      -c progs/pyperf180.c -mcpu=v3 -o /home/yhs/work/bpf-next/tools/testing/selftests/bpf/pyperf180.bpf.o
    1.      <eof> parser at end of file
    2.      Code generation
    ...

The compilation failure only happens to cpu=v2 and cpu=v3. cpu=v4 is okay
since cpu=v4 supports 32-bit branch target offset.

The above failure is due to upstream llvm patch [1] where some inlining behavior
are changed in clang18.

To workaround the issue, previously all 180 loop iterations are fully unrolled.
The bpf macro __BPF_CPU_VERSION__ (implemented in clang18 recently) is used to avoid
unrolling changes if cpu=v4. If __BPF_CPU_VERSION__ is not available and the
compiler is clang18, the unrollng amount is unconditionally reduced.

  [1] https://github.com/llvm/llvm-project/commit/1a2e77cf9e11dbf56b5720c607313a566eebb16e

Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Tested-by: Alan Maguire <alan.maguire@oracle.com>
Link: https://lore.kernel.org/bpf/20231110193644.3130906-1-yonghong.song@linux.dev
Signed-off-by: Sasha Levin <sashal@kernel.org>

libbpf: Fix potential uninitialized tail padding with LIBBPF_OPTS_RESET

[ Upstream commit 7f7c43693c1b46652cfafb7af67ba31726d6ec4e ]

Martin reported that there is a libbpf complaining of non-zero-value tail
padding with LIBBPF_OPTS_RESET macro if struct bpf_netkit_opts is modified
to have a 4-byte tail padding. This only happens to clang compiler.
The commend line is: ./test_progs -t tc_netkit_multi_links
Martin and I did some investigation and found this indeed the case and
the following are the investigation details.

Clang:
  clang version 18.0.0
  <I tried clang15/16/17 and they all have similar results>

tools/lib/bpf/libbpf_common.h:
  #define LIBBPF_OPTS_RESET(NAME, ...)                                      \
        do {                                                                \
                memset(&NAME, 0, sizeof(NAME));                             \
                NAME = (typeof(NAME)) {                                     \
                        .sz = sizeof(NAME),                                 \
                        __VA_ARGS__                                         \
                };                                                          \
        } while (0)

  #endif

tools/lib/bpf/libbpf.h:
  struct bpf_netkit_opts {
        /* size of this struct, for forward/backward compatibility */
        size_t sz;
        __u32 flags;
        __u32 relative_fd;
        __u32 relative_id;
        __u64 expected_revision;
        size_t :0;
  };
  #define bpf_netkit_opts__last_field expected_revision
In the above struct bpf_netkit_opts, there is no tail padding.

prog_tests/tc_netkit.c:
  static void serial_test_tc_netkit_multi_links_target(int mode, int target)
  {
        ...
        LIBBPF_OPTS(bpf_netkit_opts, optl);
        ...
        LIBBPF_OPTS_RESET(optl,
                .flags = BPF_F_BEFORE,
                .relative_fd = bpf_program__fd(skel->progs.tc1),
        );
        ...
  }

Let us make the following source change, note that we have a 4-byte
tailing padding now.
#  diff --git a/tools/lib/bpf/libbpf.h b/tools/lib/bpf/libbpf.h
#  index 6cd9c501624f..0dd83910ae9a 100644
#  --- a/tools/lib/bpf/libbpf.h
#  +++ b/tools/lib/bpf/libbpf.h
#  @@ -803,13 +803,13 @@ bpf_program__attach_tcx(const struct bpf_program *prog, int ifindex,
#   struct bpf_netkit_opts {
#        /* size of this struct, for forward/backward compatibility */
#        size_t sz;
#  -       __u32 flags;
#        __u32 relative_fd;
#        __u32 relative_id;
#        __u64 expected_revision;
#  +       __u32 flags;
#        size_t :0;
#   };
#  -#define bpf_netkit_opts__last_field expected_revision
#  +#define bpf_netkit_opts__last_field flags

The clang 18 generated asm code looks like below:
    ;       LIBBPF_OPTS_RESET(optl,
    55e3: 48 8d 7d 98                   leaq    -0x68(%rbp), %rdi
    55e7: 31 f6                         xorl    %esi, %esi
    55e9: ba 20 00 00 00                movl    $0x20, %edx
    55ee: e8 00 00 00 00                callq   0x55f3 <serial_test_tc_netkit_multi_links_target+0x18d3>
    55f3: 48 c7 85 10 fd ff ff 20 00 00 00      movq    $0x20, -0x2f0(%rbp)
    55fe: 48 8b 85 68 ff ff ff          movq    -0x98(%rbp), %rax
    5605: 48 8b 78 18                   movq    0x18(%rax), %rdi
    5609: e8 00 00 00 00                callq   0x560e <serial_test_tc_netkit_multi_links_target+0x18ee>
    560e: 89 85 18 fd ff ff             movl    %eax, -0x2e8(%rbp)
    5614: c7 85 1c fd ff ff 00 00 00 00 movl    $0x0, -0x2e4(%rbp)
    561e: 48 c7 85 20 fd ff ff 00 00 00 00      movq    $0x0, -0x2e0(%rbp)
    5629: c7 85 28 fd ff ff 08 00 00 00 movl    $0x8, -0x2d8(%rbp)
    5633: 48 8b 85 10 fd ff ff          movq    -0x2f0(%rbp), %rax
    563a: 48 89 45 98                   movq    %rax, -0x68(%rbp)
    563e: 48 8b 85 18 fd ff ff          movq    -0x2e8(%rbp), %rax
    5645: 48 89 45 a0                   movq    %rax, -0x60(%rbp)
    5649: 48 8b 85 20 fd ff ff          movq    -0x2e0(%rbp), %rax
    5650: 48 89 45 a8                   movq    %rax, -0x58(%rbp)
    5654: 48 8b 85 28 fd ff ff          movq    -0x2d8(%rbp), %rax
    565b: 48 89 45 b0                   movq    %rax, -0x50(%rbp)
    ;       link = bpf_program__attach_netkit(skel->progs.tc2, ifindex, &optl);

At -O0 level, the clang compiler creates an intermediate copy.
We have below to store 'flags' with 4-byte store and leave another 4 byte
in the same 8-byte-aligned storage undefined,
    5629: c7 85 28 fd ff ff 08 00 00 00 movl    $0x8, -0x2d8(%rbp)
and later we store 8-byte to the original zero'ed buffer
    5654: 48 8b 85 28 fd ff ff          movq    -0x2d8(%rbp), %rax
    565b: 48 89 45 b0                   movq    %rax, -0x50(%rbp)

This caused a problem as the 4-byte value at [%rbp-0x2dc, %rbp-0x2e0)
may be garbage.

gcc (gcc 11.4) does not have this issue as it does zeroing struct first before
doing assignments:
  ;       LIBBPF_OPTS_RESET(optl,
    50fd: 48 8d 85 40 fc ff ff          leaq    -0x3c0(%rbp), %rax
    5104: ba 20 00 00 00                movl    $0x20, %edx
    5109: be 00 00 00 00                movl    $0x0, %esi
    510e: 48 89 c7                      movq    %rax, %rdi
    5111: e8 00 00 00 00                callq   0x5116 <serial_test_tc_netkit_multi_links_target+0x1522>
    5116: 48 8b 45 f0                   movq    -0x10(%rbp), %rax
    511a: 48 8b 40 18                   movq    0x18(%rax), %rax
    511e: 48 89 c7                      movq    %rax, %rdi
    5121: e8 00 00 00 00                callq   0x5126 <serial_test_tc_netkit_multi_links_target+0x1532>
    5126: 48 c7 85 40 fc ff ff 00 00 00 00      movq    $0x0, -0x3c0(%rbp)
    5131: 48 c7 85 48 fc ff ff 00 00 00 00      movq    $0x0, -0x3b8(%rbp)
    513c: 48 c7 85 50 fc ff ff 00 00 00 00      movq    $0x0, -0x3b0(%rbp)
    5147: 48 c7 85 58 fc ff ff 00 00 00 00      movq    $0x0, -0x3a8(%rbp)
    5152: 48 c7 85 40 fc ff ff 20 00 00 00      movq    $0x20, -0x3c0(%rbp)
    515d: 89 85 48 fc ff ff             movl    %eax, -0x3b8(%rbp)
    5163: c7 85 58 fc ff ff 08 00 00 00 movl    $0x8, -0x3a8(%rbp)
  ;       link = bpf_program__attach_netkit(skel->progs.tc2, ifindex, &optl);

It is not clear how to resolve the compiler code generation as the compiler
generates correct code w.r.t. how to handle unnamed padding in C standard.
So this patch changed LIBBPF_OPTS_RESET macro to avoid uninitialized tail
padding. We already knows LIBBPF_OPTS macro works on both gcc and clang,
even with tail padding. So LIBBPF_OPTS_RESET is changed to be a
LIBBPF_OPTS followed by a memcpy(), thus avoiding uninitialized tail padding.

The below is asm code generated with this patch and with clang compiler:
    ;       LIBBPF_OPTS_RESET(optl,
    55e3: 48 8d bd 10 fd ff ff          leaq    -0x2f0(%rbp), %rdi
    55ea: 31 f6                         xorl    %esi, %esi
    55ec: ba 20 00 00 00                movl    $0x20, %edx
    55f1: e8 00 00 00 00                callq   0x55f6 <serial_test_tc_netkit_multi_links_target+0x18d6>
    55f6: 48 c7 85 10 fd ff ff 20 00 00 00      movq    $0x20, -0x2f0(%rbp)
    5601: 48 8b 85 68 ff ff ff          movq    -0x98(%rbp), %rax
    5608: 48 8b 78 18                   movq    0x18(%rax), %rdi
    560c: e8 00 00 00 00                callq   0x5611 <serial_test_tc_netkit_multi_links_target+0x18f1>
    5611: 89 85 18 fd ff ff             movl    %eax, -0x2e8(%rbp)
    5617: c7 85 1c fd ff ff 00 00 00 00 movl    $0x0, -0x2e4(%rbp)
    5621: 48 c7 85 20 fd ff ff 00 00 00 00      movq    $0x0, -0x2e0(%rbp)
    562c: c7 85 28 fd ff ff 08 00 00 00 movl    $0x8, -0x2d8(%rbp)
    5636: 48 8b 85 10 fd ff ff          movq    -0x2f0(%rbp), %rax
    563d: 48 89 45 98                   movq    %rax, -0x68(%rbp)
    5641: 48 8b 85 18 fd ff ff          movq    -0x2e8(%rbp), %rax
    5648: 48 89 45 a0                   movq    %rax, -0x60(%rbp)
    564c: 48 8b 85 20 fd ff ff          movq    -0x2e0(%rbp), %rax
    5653: 48 89 45 a8                   movq    %rax, -0x58(%rbp)
    5657: 48 8b 85 28 fd ff ff          movq    -0x2d8(%rbp), %rax
    565e: 48 89 45 b0                   movq    %rax, -0x50(%rbp)
    ;       link = bpf_program__attach_netkit(skel->progs.tc2, ifindex, &optl);

In the above code, a temporary buffer is zeroed and then has proper value assigned.
Finally, values in temporary buffer are copied to the original variable buffer,
hence tail padding is guaranteed to be 0.

Signed-off-by: Yonghong Song <yonghong.song@linux.dev>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Tested-by: Martin KaFai Lau <martin.lau@kernel.org>
Link: https://lore.kernel.org/bpf/20231107201511.2548645-1-yonghong.song@linux.dev
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

selftests/bpf: satisfy compiler by having explicit return in btf test

[ Upstream commit f4c7e887324f5776eef6e6e47a90e0ac8058a7a8 ]

Some compilers complain about get_pprint_mapv_size() not returning value
in some code paths. Fix with explicit return.

Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20231102033759.2541186-3-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/bpf: fix RELEASE=1 build for tc_opts

[ Upstream commit 2b62aa59d02ed281fa4fc218df3ca91b773e1e62 ]

Compiler complains about malloc(). We also don't need to dynamically
allocate anything, so make the life easier by using statically sized
buffer.

Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Link: https://lore.kernel.org/r/20231102033759.2541186-2-andrii@kernel.org
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rt2x00: restart beacon queue when hardware reset

[ Upstream commit a11d965a218f0cd95b13fe44d0bcd8a20ce134a8 ]

When a hardware reset is triggered, all registers are reset, so all
queues are forced to stop in hardware interface. However, mac80211
will not automatically stop the queue. If we don't manually stop the
beacon queue, the queue will be deadlocked and unable to start again.
This patch fixes the issue where Apple devices cannot connect to the
AP after calling ieee80211_restart_hw().

Signed-off-by: Shiji Yang <yangshiji66@outlook.com>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/TYAP286MB031530EB6D98DCE4DF20766CBCA4A@TYAP286MB0315.JPNP286.PROD.OUTLOOK.COM
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtw89: fix timeout calculation in rtw89_roc_end()

[ Upstream commit e416514e309f7e25e577fee45a65f246f67b2261 ]

Since 'rtw89_core_tx_kick_off_and_wait()' assumes timeout
(actually RTW89_ROC_TX_TIMEOUT) in milliseconds, I suppose
that RTW89_ROC_IDLE_TIMEOUT is in milliseconds as well. If
so, 'msecs_to_jiffies()' should be used in a call to
'ieee80211_queue_delayed_work()' from 'rtw89_roc_end()'.
Compile tested only.

Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
Acked-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20231024143137.30393-1-dmantipov@yandex.ru
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: avoid online resizing failures due to oversized flex bg

[ Upstream commit 5d1935ac02ca5aee364a449a35e2977ea84509b0 ]

When we online resize an ext4 filesystem with a oversized flexbg_size,

     mkfs.ext4 -F -G 67108864 $dev -b 4096 100M
     mount $dev $dir
     resize2fs $dev 16G

the following WARN_ON is triggered:
==================================================================
WARNING: CPU: 0 PID: 427 at mm/page_alloc.c:4402 __alloc_pages+0x411/0x550
Modules linked in: sg(E)
CPU: 0 PID: 427 Comm: resize2fs Tainted: G  E  6.6.0-rc5+ #314
RIP: 0010:__alloc_pages+0x411/0x550
Call Trace:
 <TASK>
 __kmalloc_large_node+0xa2/0x200
 __kmalloc+0x16e/0x290
 ext4_resize_fs+0x481/0xd80
 __ext4_ioctl+0x1616/0x1d90
 ext4_ioctl+0x12/0x20
 __x64_sys_ioctl+0xf0/0x150
 do_syscall_64+0x3b/0x90
==================================================================

This is because flexbg_size is too large and the size of the new_group_data
array to be allocated exceeds MAX_ORDER. Currently, the minimum value of
MAX_ORDER is 8, the minimum value of PAGE_SIZE is 4096, the corresponding
maximum number of groups that can be allocated is:

 (PAGE_SIZE << MAX_ORDER) / sizeof(struct ext4_new_group_data) ≈ 21845

And the value that is down-aligned to the power of 2 is 16384. Therefore,
this value is defined as MAX_RESIZE_BG, and the number of groups added
each time does not exceed this value during resizing, and is added multiple
times to complete the online resizing. The difference is that the metadata
in a flex_bg may be more dispersed.

Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20231023013057.2117948-4-libaokun1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: remove unnecessary check from alloc_flex_gd()

[ Upstream commit b099eb87de105cf07cad731ded6fb40b2675108b ]

In commit 967ac8af4475 ("ext4: fix potential integer overflow in
alloc_flex_gd()"), an overflow check is added to alloc_flex_gd() to
prevent the allocated memory from being smaller than expected due to
the overflow. However, after kmalloc() is replaced with kmalloc_array()
in commit 6da2ec56059c ("treewide: kmalloc() -> kmalloc_array()"), the
kmalloc_array() function has an overflow check, so the above problem
will not occur. Therefore, the extra check is removed.

Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20231023013057.2117948-3-libaokun1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: unify the type of flexbg_size to unsigned int

[ Upstream commit 658a52344fb139f9531e7543a6e0015b630feb38 ]

The maximum value of flexbg_size is 2^31, but the maximum value of int
is (2^31 - 1), so overflow may occur when the type of flexbg_size is
declared as int.

For example, when uninit_mask is initialized in ext4_alloc_group_tables(),
if flexbg_size == 2^31, the initialized uninit_mask is incorrect, and this
may causes set_flexbg_block_bitmap() to trigger a BUG_ON().

Therefore, the flexbg_size type is declared as unsigned int to avoid
overflow and memory waste.

Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20231023013057.2117948-2-libaokun1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: fix inconsistent between segment fstrim and full fstrim

[ Upstream commit 68da4c44b994aea797eb9821acb3a4a36015293e ]

Suppose we issue two FITRIM ioctls for ranges [0,15] and [16,31] with
mininum length of trimmed range set to 8 blocks. If we have say a range of
blocks 10-22 free, this range will not be trimmed because it straddles the
boundary of the two FITRIM ranges and neither part is big enough. This is a
bit surprising to some users that call FITRIM on smaller ranges of blocks
to limit impact on the system. Also XFS trims all free space extents that
overlap with the specified range so we are inconsistent among filesystems.
Let's change ext4_try_to_trim_range() to consider for trimming the whole
free space extent that straddles the end of specified range, not just the
part of it within the range.

Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20231216010919.1995851-1-yebin10@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ecryptfs: Reject casefold directory inodes

[ Upstream commit cd72c7ef5fed44272272a105b1da22810c91be69 ]

Even though it seems to be able to resolve some names of
case-insensitive directories, the lack of d_hash and d_compare means we
end up with a broken state in the d_cache.  Considering it was never a
goal to support these two together, and we are preparing to use
d_revalidate in case-insensitive filesystems, which would make the
combination even more broken, reject any attempt to get a casefolded
inode from ecryptfs.

Signed-off-by: Gabriel Krisman Bertazi <krisman@suse.de>
Reviewed-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

smb: client: fix hardlinking of reparse points

[ Upstream commit 5408990aa662bcfd6ba894734023a023a16e8729 ]

The client was sending an SMB2_CREATE request without setting
OPEN_REPARSE_POINT flag thus failing the entire hardlink operation.

Fix this by setting OPEN_REPARSE_POINT in create options for
SMB2_CREATE request when the source inode is a repase point.

Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

smb: client: fix renaming of reparse points

[ Upstream commit 7435d51b7ea2ab7801279c43ecd72063e9d5c92f ]

The client was sending an SMB2_CREATE request without setting
OPEN_REPARSE_POINT flag thus failing the entire rename operation.

Fix this by setting OPEN_REPARSE_POINT in create options for
SMB2_CREATE request when the source inode is a repase point.

Signed-off-by: Paulo Alcantara (SUSE) <pc@manguebit.com>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ext4: treat end of range as exclusive in ext4_zero_range()

[ Upstream commit 92573369144f40397e8514440afdf59f24905b40 ]

The call to filemap_write_and_wait_range() assumes the range passed to be
inclusive, so fix the call to make sure we follow that.

Signed-off-by: Ojaswin Mujoo <ojaswin@linux.ibm.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/e503107a7c73a2b68dec645c5ad798c437717c45.1698856309.git.ojaswin@linux.ibm.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

SUNRPC: Fix a suspicious RCU usage warning

[ Upstream commit 31b62908693c90d4d07db597e685d9f25a120073 ]

I received the following warning while running cthon against an ontap
server running pNFS:

[   57.202521] =============================
[   57.202522] WARNING: suspicious RCU usage
[   57.202523] 6.7.0-rc3-g2cc14f52aeb7 #41492 Not tainted
[   57.202525] -----------------------------
[   57.202525] net/sunrpc/xprtmultipath.c:349 RCU-list traversed in non-reader section!!
[   57.202527]
               other info that might help us debug this:

[   57.202528]
               rcu_scheduler_active = 2, debug_locks = 1
[   57.202529] no locks held by test5/3567.
[   57.202530]
               stack backtrace:
[   57.202532] CPU: 0 PID: 3567 Comm: test5 Not tainted 6.7.0-rc3-g2cc14f52aeb7 #41492 5b09971b4965c0aceba19f3eea324a4a806e227e
[   57.202534] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS unknown 2/2/2022
[   57.202536] Call Trace:
[   57.202537]  <TASK>
[   57.202540]  dump_stack_lvl+0x77/0xb0
[   57.202551]  lockdep_rcu_suspicious+0x154/0x1a0
[   57.202556]  rpc_xprt_switch_has_addr+0x17c/0x190 [sunrpc ebe02571b9a8ceebf7d98e71675af20c19bdb1f6]
[   57.202596]  rpc_clnt_setup_test_and_add_xprt+0x50/0x180 [sunrpc ebe02571b9a8ceebf7d98e71675af20c19bdb1f6]
[   57.202621]  ? rpc_clnt_add_xprt+0x254/0x300 [sunrpc ebe02571b9a8ceebf7d98e71675af20c19bdb1f6]
[   57.202646]  rpc_clnt_add_xprt+0x27a/0x300 [sunrpc ebe02571b9a8ceebf7d98e71675af20c19bdb1f6]
[   57.202671]  ? __pfx_rpc_clnt_setup_test_and_add_xprt+0x10/0x10 [sunrpc ebe02571b9a8ceebf7d98e71675af20c19bdb1f6]
[   57.202696]  nfs4_pnfs_ds_connect+0x345/0x760 [nfsv4 c716d88496ded0ea6d289bbea684fa996f9b57a9]
[   57.202728]  ? __pfx_nfs4_test_session_trunk+0x10/0x10 [nfsv4 c716d88496ded0ea6d289bbea684fa996f9b57a9]
[   57.202754]  nfs4_fl_prepare_ds+0x75/0xc0 [nfs_layout_nfsv41_files e3a4187f18ae8a27b630f9feae6831b584a9360a]
[   57.202760]  filelayout_write_pagelist+0x4a/0x200 [nfs_layout_nfsv41_files e3a4187f18ae8a27b630f9feae6831b584a9360a]
[   57.202765]  pnfs_generic_pg_writepages+0xbe/0x230 [nfsv4 c716d88496ded0ea6d289bbea684fa996f9b57a9]
[   57.202788]  __nfs_pageio_add_request+0x3fd/0x520 [nfs 6c976fa593a7c2976f5a0aeb4965514a828e6902]
[   57.202813]  nfs_pageio_add_request+0x18b/0x390 [nfs 6c976fa593a7c2976f5a0aeb4965514a828e6902]
[   57.202831]  nfs_do_writepage+0x116/0x1e0 [nfs 6c976fa593a7c2976f5a0aeb4965514a828e6902]
[   57.202849]  nfs_writepages_callback+0x13/0x30 [nfs 6c976fa593a7c2976f5a0aeb4965514a828e6902]
[   57.202866]  write_cache_pages+0x265/0x450
[   57.202870]  ? __pfx_nfs_writepages_callback+0x10/0x10 [nfs 6c976fa593a7c2976f5a0aeb4965514a828e6902]
[   57.202891]  nfs_writepages+0x141/0x230 [nfs 6c976fa593a7c2976f5a0aeb4965514a828e6902]
[   57.202913]  do_writepages+0xd2/0x230
[   57.202917]  ? filemap_fdatawrite_wbc+0x5c/0x80
[   57.202921]  filemap_fdatawrite_wbc+0x67/0x80
[   57.202924]  filemap_write_and_wait_range+0xd9/0x170
[   57.202930]  nfs_wb_all+0x49/0x180 [nfs 6c976fa593a7c2976f5a0aeb4965514a828e6902]
[   57.202947]  nfs4_file_flush+0x72/0xb0 [nfsv4 c716d88496ded0ea6d289bbea684fa996f9b57a9]
[   57.202969]  __se_sys_close+0x46/0xd0
[   57.202972]  do_syscall_64+0x68/0x100
[   57.202975]  ? do_syscall_64+0x77/0x100
[   57.202976]  ? do_syscall_64+0x77/0x100
[   57.202979]  entry_SYSCALL_64_after_hwframe+0x6e/0x76
[   57.202982] RIP: 0033:0x7fe2b12e4a94
[   57.202985] Code: 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 80 3d d5 18 0e 00 00 74 13 b8 03 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 44 c3 0f 1f 00 48 83 ec 18 89 7c 24 0c e8 c3
[   57.202987] RSP: 002b:00007ffe857ddb38 EFLAGS: 00000202 ORIG_RAX: 0000000000000003
[   57.202989] RAX: ffffffffffffffda RBX: 00007ffe857dfd68 RCX: 00007fe2b12e4a94
[   57.202991] RDX: 0000000000002000 RSI: 00007ffe857ddc40 RDI: 0000000000000003
[   57.202992] RBP: 00007ffe857dfc50 R08: 7fffffffffffffff R09: 0000000065650f49
[   57.202993] R10: 00007fe2b11f8300 R11: 0000000000000202 R12: 0000000000000000
[   57.202994] R13: 00007ffe857dfd80 R14: 00007fe2b1445000 R15: 0000000000000000
[   57.202999]  </TASK>

The problem seems to be that two out of three callers aren't taking the
rcu_read_lock() before calling the list_for_each_entry_rcu() function in
rpc_xprt_switch_has_addr(). I fix this by having
rpc_xprt_switch_has_addr() unconditionaly take the rcu_read_lock(),
which is okay to do recursively in the case that the lock has already
been taken by a caller.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

sysctl: Fix out of bounds access for empty sysctl registers

[ Upstream commit 315552310c7de92baea4e570967066569937a843 ]

When registering tables to the sysctl subsystem there is a check to see
if header is a permanently empty directory (used for mounts). This check
evaluates the first element of the ctl_table. This results in an out of
bounds evaluation when registering empty directories.

The function register_sysctl_mount_point now passes a ctl_table of size
1 instead of size 0. It now relies solely on the type to identify
a permanently empty register.

Make sure that the ctl_table has at least one element before testing for
permanent emptiness.

Signed-off-by: Joel Granados <j.granados@samsung.com>
Reported-by: kernel test robot <oliver.sang@intel.com>
Closes: https://lore.kernel.org/oe-lkp/202311201431.57aae8f3-oliver.sang@intel.com
Signed-off-by: Luis Chamberlain <mcgrof@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

KVM: s390: fix setting of fpc register

[ Upstream commit b988b1bb0053c0dcd26187d29ef07566a565cf55 ]

kvm_arch_vcpu_ioctl_set_fpu() allows to set the floating point control
(fpc) register of a guest cpu. The new value is tested for validity by
temporarily loading it into the fpc register.

This may lead to corruption of the fpc register of the host process:
if an interrupt happens while the value is temporarily loaded into the fpc
register, and within interrupt context floating point or vector registers
are used, the current fp/vx registers are saved with save_fpu_regs()
assuming they belong to user space and will be loaded into fp/vx registers
when returning to user space.

test_fp_ctl() restores the original user space / host process fpc register
value, however it will be discarded, when returning to user space.

In result the host process will incorrectly continue to run with the value
that was supposed to be used for a guest cpu.

Fix this by simply removing the test. There is another test right before
the SIE context is entered which will handles invalid values.

This results in a change of behaviour: invalid values will now be accepted
instead of that the ioctl fails with -EINVAL. This seems to be acceptable,
given that this interface is most likely not used anymore, and this is in
addition the same behaviour implemented with the memory mapped interface
(replace invalid values with zero) - see sync_regs() in kvm-s390.c.

Reviewed-by: Christian Borntraeger <borntraeger@linux.ibm.com>
Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

s390/ptrace: handle setting of fpc register correctly

[ Upstream commit 8b13601d19c541158a6e18b278c00ba69ae37829 ]

If the content of the floating point control (fpc) register of a traced
process is modified with the ptrace interface the new value is tested for
validity by temporarily loading it into the fpc register.

This may lead to corruption of the fpc register of the tracing process:
if an interrupt happens while the value is temporarily loaded into the
fpc register, and within interrupt context floating point or vector
registers are used, the current fp/vx registers are saved with
save_fpu_regs() assuming they belong to user space and will be loaded into
fp/vx registers when returning to user space.

test_fp_ctl() restores the original user space fpc register value, however
it will be discarded, when returning to user space.

In result the tracer will incorrectly continue to run with the value that
was supposed to be used for the traced process.

Fix this by saving fpu register contents with save_fpu_regs() before using
test_fp_ctl().

Reviewed-by: Claudio Imbrenda <imbrenda@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

s390/vfio-ap: fix sysfs status attribute for AP queue devices

[ Upstream commit a0d8f4eeb7c4ffaee21702bcc91a09b3988c5b7a ]

The 'status' attribute for AP queue devices bound to the vfio_ap device
driver displays incorrect status when the mediated device is attached to a
guest, but the queue device is not passed through. In the current
implementation, the status displayed is 'in_use' which is not correct; it
should be 'assigned'. This can happen if one of the queue devices
associated with a given adapter is not bound to the vfio_ap device driver.
For example:

Queues listed in /sys/bus/ap/drivers/vfio_ap:
14.0005
14.0006
14.000d
16.0006
16.000d

Queues listed in /sys/devices/vfio_ap/matrix/$UUID/matrix
14.0005
14.0006
14.000d
16.0005
16.0006
16.000d

Queues listed in /sys/devices/vfio_ap/matrix/$UUID/guest_matrix
14.0005
14.0006
14.000d

The reason no queues for adapter 0x16 are listed in the guest_matrix is
because queue 16.0005 is not bound to the vfio_ap device driver, so no
queue associated with the adapter is passed through to the guest;
therefore, each queue device for adapter 0x16 should display 'assigned'
instead of 'in_use', because those queues are not in use by a guest, but
only assigned to the mediated device.

Let's check the AP configuration for the guest to determine whether a
queue device is passed through before displaying a status of 'in_use'.

Signed-off-by: Tony Krowiak <akrowiak@linux.ibm.com>
Acked-by: Halil Pasic <pasic@linux.ibm.com>
Acked-by: Harald Freudenberger <freude@linux.ibm.com>
Link: https://lore.kernel.org/r/20231108201135.351419-1-akrowiak@linux.ibm.com
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arch: consolidate arch_irq_work_raise prototypes

[ Upstream commit 64bac5ea17d527872121adddfee869c7a0618f8f ]

The prototype was hidden in an #ifdef on x86, which causes a warning:

kernel/irq_work.c:72:13: error: no previous prototype for 'arch_irq_work_raise' [-Werror=missing-prototypes]

Some architectures have a working prototype, while others don't.
Fix this by providing it in only one place that is always visible.

Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
Acked-by: Guo Ren <guoren@kernel.org>
Signed-off-by: Arnd Bergmann <arnd@arndb.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

s390/boot: always align vmalloc area on segment boundary

[ Upstream commit 65f8780e2d70257200547b5a7654974aa7c37ce1 ]

The size of vmalloc area depends from various factors
on boot and could be set to:

1. Default size as determined by VMALLOC_DEFAULT_SIZE macro;
2. One half of the virtual address space not occupied by
   modules and fixed mappings;
3. The size provided by user with vmalloc= kernel command
   line parameter;

In cases [1] and [2] the vmalloc area base address is aligned
on Region3 table type boundary, while in case [3] in might get
aligned on page boundary.

Limit the waste of page tables and always align vmalloc area
size and base address on segment boundary.

Acked-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

jfs: fix array-index-out-of-bounds in diNewExt

[ Upstream commit 49f9637aafa6e63ba686c13cb8549bf5e6920402 ]

[Syz report]
UBSAN: array-index-out-of-bounds in fs/jfs/jfs_imap.c:2360:2
index -878706688 is out of range for type 'struct iagctl[128]'
CPU: 1 PID: 5065 Comm: syz-executor282 Not tainted 6.7.0-rc4-syzkaller-00009-gbee0e7762ad2 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 11/10/2023
Call Trace:
 <TASK>
 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0x1e7/0x2d0 lib/dump_stack.c:106
 ubsan_epilogue lib/ubsan.c:217 [inline]
 __ubsan_handle_out_of_bounds+0x11c/0x150 lib/ubsan.c:348
 diNewExt+0x3cf3/0x4000 fs/jfs/jfs_imap.c:2360
 diAllocExt fs/jfs/jfs_imap.c:1949 [inline]
 diAllocAG+0xbe8/0x1e50 fs/jfs/jfs_imap.c:1666
 diAlloc+0x1d3/0x1760 fs/jfs/jfs_imap.c:1587
 ialloc+0x8f/0x900 fs/jfs/jfs_inode.c:56
 jfs_mkdir+0x1c5/0xb90 fs/jfs/namei.c:225
 vfs_mkdir+0x2f1/0x4b0 fs/namei.c:4106
 do_mkdirat+0x264/0x3a0 fs/namei.c:4129
 __do_sys_mkdir fs/namei.c:4149 [inline]
 __se_sys_mkdir fs/namei.c:4147 [inline]
 __x64_sys_mkdir+0x6e/0x80 fs/namei.c:4147
 do_syscall_x64 arch/x86/entry/common.c:51 [inline]
 do_syscall_64+0x45/0x110 arch/x86/entry/common.c:82
 entry_SYSCALL_64_after_hwframe+0x63/0x6b
RIP: 0033:0x7fcb7e6a0b57
Code: ff ff 77 07 31 c0 c3 0f 1f 40 00 48 c7 c2 b8 ff ff ff f7 d8 64 89 02 b8 ff ff ff ff c3 66 0f 1f 44 00 00 b8 53 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007ffd83023038 EFLAGS: 00000286 ORIG_RAX: 0000000000000053
RAX: ffffffffffffffda RBX: 00000000ffffffff RCX: 00007fcb7e6a0b57
RDX: 00000000000a1020 RSI: 00000000000001ff RDI: 0000000020000140
RBP: 0000000020000140 R08: 0000000000000000 R09: 0000000000000000
R10: 0000000000000000 R11: 0000000000000286 R12: 00007ffd830230d0
R13: 0000000000000000 R14: 0000000000000000 R15: 0000000000000000

[Analysis]
When the agstart is too large, it can cause agno overflow.

[Fix]
After obtaining agno, if the value is invalid, exit the subsequent process.

Reported-and-tested-by: syzbot+553d90297e6d2f50dbc7@syzkaller.appspotmail.com
Signed-off-by: Edward Adam Davis <eadavis@qq.com>
Modified the test from agno > MAXAG to agno >= MAXAG based on linux-next
report by kernel test robot (Dan Carpenter).

Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rxrpc_find_service_conn_rcu: fix the usage of read_seqbegin_or_lock()

[ Upstream commit bad1a11c0f061aa073bab785389fe04f19ba02e1 ]

rxrpc_find_service_conn_rcu() should make the "seq" counter odd on the
second pass, otherwise read_seqbegin_or_lock() never takes the lock.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
Link: https://lore.kernel.org/r/20231117164846.GA10410@redhat.com/
Signed-off-by: Sasha Levin <sashal@kernel.org>

afs: fix the usage of read_seqbegin_or_lock() in afs_find_server*()

[ Upstream commit 1702e0654ca9a7bcd7c7619c8a5004db58945b71 ]

David Howells says:

 (5) afs_find_server().

     There could be a lot of servers in the list and each server can have
     multiple addresses, so I think this would be better with an exclusive
     second pass.

     The server list isn't likely to change all that often, but when it does
     change, there's a good chance several servers are going to be
     added/removed one after the other.  Further, this is only going to be
     used for incoming cache management/callback requests from the server,
     which hopefully aren't going to happen too often - but it is remotely
     drivable.

 (6) afs_find_server_by_uuid().

     Similarly to (5), there could be a lot of servers to search through, but
     they are in a tree not a flat list, so it should be faster to process.
     Again, it's not likely to change that often and, again, when it does
     change it's likely to involve multiple changes.  This can be driven
     remotely by an incoming cache management request but is mostly going to
     be driven by setting up or reconfiguring a volume's server list -
     something that also isn't likely to happen often.

Make the "seq" counter odd on the 2nd pass, otherwise read_seqbegin_or_lock()
never takes the lock.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
Link: https://lore.kernel.org/r/20231130115614.GA21581@redhat.com/
Signed-off-by: Sasha Levin <sashal@kernel.org>

afs: fix the usage of read_seqbegin_or_lock() in afs_lookup_volume_rcu()

[ Upstream commit 4121b4337146b64560d1e46ebec77196d9287802 ]

David Howells says:

 (2) afs_lookup_volume_rcu().

     There can be a lot of volumes known by a system.  A thousand would
     require a 10-step walk and this is drivable by remote operation, so I
     think this should probably take a lock on the second pass too.

Make the "seq" counter odd on the 2nd pass, otherwise read_seqbegin_or_lock()
never takes the lock.

Signed-off-by: Oleg Nesterov <oleg@redhat.com>
Signed-off-by: David Howells <dhowells@redhat.com>
cc: Marc Dionne <marc.dionne@auristor.com>
cc: linux-afs@lists.infradead.org
Link: https://lore.kernel.org/r/20231130115606.GA21571@redhat.com/
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: stm32/crc32 - fix parsing list of devices

[ Upstream commit 0eaef675b94c746900dcea7f6c41b9a103ed5d53 ]

smatch warnings:
drivers/crypto/stm32/stm32-crc32.c:108 stm32_crc_get_next_crc() warn:
can 'crc' even be NULL?

Use list_first_entry_or_null instead of list_first_entry to retrieve
the first device registered.
The function list_first_entry always return a non NULL pointer even if
the list is empty. Hence checking if the pointer returned is NULL does
not tell if the list is empty or not.

Reported-by: kernel test robot <lkp@intel.com>
Closes: https://lore.kernel.org/r/202311281111.ou2oUL2i-lkp@intel.com/
Reported-by: Dan Carpenter <error27@gmail.com>
Closes: https://lore.kernel.org/r/202311281111.ou2oUL2i-lkp@intel.com/
Signed-off-by: Thomas Bourgoin <thomas.bourgoin@foss.st.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

erofs: fix ztailpacking for subpage compressed blocks

[ Upstream commit e5aba911dee5e20fa82efbe13e0af8f38ea459e7 ]

`pageofs_in` should be the compressed data offset of the page rather
than of the block.

Acked-by: Chao Yu <chao@kernel.org>
Reviewed-by: Yue Hu <huyue2@coolpad.com>
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Link: https://lore.kernel.org/r/20231214161337.753049-1-hsiangkao@linux.alibaba.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: octeontx2 - Fix cptvf driver cleanup

[ Upstream commit c480a421a4faf693c38e60b0fe6e554c9a3fee02 ]

This patch fixes following cleanup issues:
 - Missing instruction queue free on cleanup. This
   will lead to memory leak.
 - lfs->lfs_num is set to zero before cleanup, which
   will lead to improper cleanup.

Signed-off-by: Bharat Bhushan <bbhushan2@marvell.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: starfive - Fix dev_err_probe return error

[ Upstream commit 8517c34e87025b3f74f3c07813d493828f369598 ]

Current dev_err_probe will return 0 instead of proper error code if
driver failed to get irq number. Fix the return code.

Signed-off-by: Jia Jie Ho <jiajie.ho@starfivetech.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

erofs: fix up compacted indexes for block size < 4096

[ Upstream commit 8d2517aaeea3ab8651bb517bca8f3c8664d318ea ]

Previously, the block size always equaled to PAGE_SIZE, therefore
`lclusterbits` couldn't be less than 12.

Since sub-page compressed blocks are now considered, `lobits` for
a lcluster in each pack cannot always be `lclusterbits` as before.
Otherwise, there is no enough room for the special value
`Z_EROFS_LI_D0_CBLKCNT`.

To support smaller block sizes, `lobits` for each compacted lcluster is
now calculated as:
   lobits = max(lclusterbits, ilog2(Z_EROFS_LI_D0_CBLKCNT) + 1)

Reviewed-by: Yue Hu <huyue2@coolpad.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Gao Xiang <hsiangkao@linux.alibaba.com>
Link: https://lore.kernel.org/r/20231206091057.87027-4-hsiangkao@linux.alibaba.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

pstore/ram: Fix crash when setting number of cpus to an odd number

[ Upstream commit d49270a04623ce3c0afddbf3e984cb245aa48e9c ]

When the number of cpu cores is adjusted to 7 or other odd numbers,
the zone size will become an odd number.
The address of the zone will become:
    addr of zone0 = BASE
    addr of zone1 = BASE + zone_size
    addr of zone2 = BASE + zone_size*2
    ...
The address of zone1/3/5/7 will be mapped to non-alignment va.
Eventually crashes will occur when accessing these va.

So, use ALIGN_DOWN() to make sure the zone size is even
to avoid this bug.

Signed-off-by: Weichen Chen <weichen.chen@mediatek.com>
Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com>
Tested-by: "Guilherme G. Piccoli" <gpiccoli@igalia.com>
Link: https://lore.kernel.org/r/20230224023632.6840-1-weichen.chen@mediatek.com
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: p10-aes-gcm - Avoid -Wstringop-overflow warnings

[ Upstream commit aaa03fdb56c781db4a4831dd5d6ec8817918c726 ]

The compiler doesn't know that `32` is an offset into the Hash table:

 56 struct Hash_ctx {
 57         u8 H[16];       /* subkey */
 58         u8 Htable[256]; /* Xi, Hash table(offset 32) */
 59 };

So, it legitimately complains about a potential out-of-bounds issue
if `256 bytes` are accessed in `htable` (this implies going
`32 bytes` beyond the boundaries of `Htable`):

arch/powerpc/crypto/aes-gcm-p10-glue.c: In function 'gcmp10_init':
arch/powerpc/crypto/aes-gcm-p10-glue.c:120:9: error: 'gcm_init_htable' accessing 256 bytes in a region of size 224 [-Werror=stringop-overflow=]
  120 |         gcm_init_htable(hash->Htable+32, hash->H);
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
arch/powerpc/crypto/aes-gcm-p10-glue.c:120:9: note: referencing argument 1 of type 'unsigned char[256]'
arch/powerpc/crypto/aes-gcm-p10-glue.c:120:9: note: referencing argument 2 of type 'unsigned char[16]'
arch/powerpc/crypto/aes-gcm-p10-glue.c:40:17: note: in a call to function 'gcm_init_htable'
   40 | asmlinkage void gcm_init_htable(unsigned char htable[256], unsigned char Xi[16]);
      |                 ^~~~~~~~~~~~~~~

Address this by avoiding specifying the size of `htable` in the function
prototype; and just for consistency, do the same for parameter `Xi`.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Closes: https://lore.kernel.org/linux-next/20231121131903.68a37932@canb.auug.org.au/
Signed-off-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

hwrng: starfive - Fix dev_err_probe return error

[ Upstream commit 2d37b3649c412b3bcecfea932cb677f7a5775b15 ]

Current dev_err_probe will return 0 instead of proper error code if
driver failed to get irq number. Fix the return err code.

Signed-off-by: Jia Jie Ho <jiajie.ho@starfivetech.com>
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <error27@gmail.com>
Closes: https://lore.kernel.org/r/202311160649.3GhKCfhd-lkp@intel.com/
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

jfs: fix uaf in jfs_evict_inode

[ Upstream commit e0e1958f4c365e380b17ccb35617345b31ef7bf3 ]

When the execution of diMount(ipimap) fails, the object ipimap that has been
released may be accessed in diFreeSpecial(). Asynchronous ipimap release occurs
when rcu_core() calls jfs_free_node().

Therefore, when diMount(ipimap) fails, sbi->ipimap should not be initialized as
ipimap.

Reported-and-tested-by: syzbot+01cf2dbcbe2022454388@syzkaller.appspotmail.com
Signed-off-by: Edward Adam Davis <eadavis@qq.com>
Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

jfs: fix array-index-out-of-bounds in dbAdjTree

[ Upstream commit 74ecdda68242b174920fe7c6133a856fb7d8559b ]

Currently there is a bound check missing in the dbAdjTree while
accessing the dmt_stree. To add the required check added the bool is_ctl
which is required to determine the size as suggest in the following
commit.
https://lore.kernel.org/linux-kernel-mentees/f9475918-2186-49b8-b801-6f0f9e75f4fa@oracle.com/

Reported-by: syzbot+39ba34a099ac2e9bd3cb@syzkaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=39ba34a099ac2e9bd3cb
Signed-off-by: Manas Ghandat <ghandatmanas@gmail.com>
Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

jfs: fix slab-out-of-bounds Read in dtSearch

[ Upstream commit fa5492ee89463a7590a1449358002ff7ef63529f ]

Currently while searching for current page in the sorted entry table
of the page there is a out of bound access. Added a bound check to fix
the error.

Dave:
Set return code to -EIO

Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Closes: https://lore.kernel.org/r/202310241724.Ed02yUz9-lkp@intel.com/
Signed-off-by: Manas Ghandat <ghandatmanas@gmail.com>
Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

UBSAN: array-index-out-of-bounds in dtSplitRoot

[ Upstream commit 27e56f59bab5ddafbcfe69ad7a4a6ea1279c1b16 ]

Syzkaller reported the following issue:

oop0: detected capacity change from 0 to 32768

UBSAN: array-index-out-of-bounds in fs/jfs/jfs_dtree.c:1971:9
index -2 is out of range for type 'struct dtslot [128]'
CPU: 0 PID: 3613 Comm: syz-executor270 Not tainted 6.0.0-syzkaller-09423-g493ffd6605b2 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 09/22/2022
Call Trace:
 <TASK>
 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0x1b1/0x28e lib/dump_stack.c:106
 ubsan_epilogue lib/ubsan.c:151 [inline]
 __ubsan_handle_out_of_bounds+0xdb/0x130 lib/ubsan.c:283
 dtSplitRoot+0x8d8/0x1900 fs/jfs/jfs_dtree.c:1971
 dtSplitUp fs/jfs/jfs_dtree.c:985 [inline]
 dtInsert+0x1189/0x6b80 fs/jfs/jfs_dtree.c:863
 jfs_mkdir+0x757/0xb00 fs/jfs/namei.c:270
 vfs_mkdir+0x3b3/0x590 fs/namei.c:4013
 do_mkdirat+0x279/0x550 fs/namei.c:4038
 __do_sys_mkdirat fs/namei.c:4053 [inline]
 __se_sys_mkdirat fs/namei.c:4051 [inline]
 __x64_sys_mkdirat+0x85/0x90 fs/namei.c:4051
 do_syscall_x64 arch/x86/entry/common.c:50 [inline]
 do_syscall_64+0x3d/0xb0 arch/x86/entry/common.c:80
 entry_SYSCALL_64_after_hwframe+0x63/0xcd
RIP: 0033:0x7fcdc0113fd9
Code: ff ff c3 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 40 00 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 c0 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007ffeb8bc67d8 EFLAGS: 00000246 ORIG_RAX: 0000000000000102
RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007fcdc0113fd9
RDX: 0000000000000000 RSI: 0000000020000340 RDI: 0000000000000003
RBP: 00007fcdc00d37a0 R08: 0000000000000000 R09: 00007fcdc00d37a0
R10: 00005555559a72c0 R11: 0000000000000246 R12: 00000000f8008000
R13: 0000000000000000 R14: 00083878000000f8 R15: 0000000000000000
 </TASK>

The issue is caused when the value of fsi becomes less than -1.
The check to break the loop when fsi value becomes -1 is present
but syzbot was able to produce value less than -1 which cause the error.
This patch simply add the change for the values less than 0.

The patch is tested via syzbot.

Reported-and-tested-by: syzbot+d4b1df2e9d4ded6488ec@syzkaller.appspotmail.com
Link: https://syzkaller.appspot.com/bug?extid=d4b1df2e9d4ded6488ec
Signed-off-by: Osama Muhammad <osmtendev@gmail.com>
Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

FS:JFS:UBSAN:array-index-out-of-bounds in dbAdjTree

[ Upstream commit 9862ec7ac1cbc6eb5ee4a045b5d5b8edbb2f7e68 ]

Syzkaller reported the following issue:

UBSAN: array-index-out-of-bounds in fs/jfs/jfs_dmap.c:2867:6
index 196694 is out of range for type 's8[1365]' (aka 'signed char[1365]')
CPU: 1 PID: 109 Comm: jfsCommit Not tainted 6.6.0-rc3-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/04/2023
Call Trace:
 <TASK>
 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0x1e7/0x2d0 lib/dump_stack.c:106
 ubsan_epilogue lib/ubsan.c:217 [inline]
 __ubsan_handle_out_of_bounds+0x11c/0x150 lib/ubsan.c:348
 dbAdjTree+0x474/0x4f0 fs/jfs/jfs_dmap.c:2867
 dbJoin+0x210/0x2d0 fs/jfs/jfs_dmap.c:2834
 dbFreeBits+0x4eb/0xda0 fs/jfs/jfs_dmap.c:2331
 dbFreeDmap fs/jfs/jfs_dmap.c:2080 [inline]
 dbFree+0x343/0x650 fs/jfs/jfs_dmap.c:402
 txFreeMap+0x798/0xd50 fs/jfs/jfs_txnmgr.c:2534
 txUpdateMap+0x342/0x9e0
 txLazyCommit fs/jfs/jfs_txnmgr.c:2664 [inline]
 jfs_lazycommit+0x47a/0xb70 fs/jfs/jfs_txnmgr.c:2732
 kthread+0x2d3/0x370 kernel/kthread.c:388
 ret_from_fork+0x48/0x80 arch/x86/kernel/process.c:147
 ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:304
 </TASK>
================================================================================
Kernel panic - not syncing: UBSAN: panic_on_warn set ...
CPU: 1 PID: 109 Comm: jfsCommit Not tainted 6.6.0-rc3-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 08/04/2023
Call Trace:
 <TASK>
 __dump_stack lib/dump_stack.c:88 [inline]
 dump_stack_lvl+0x1e7/0x2d0 lib/dump_stack.c:106
 panic+0x30f/0x770 kernel/panic.c:340
 check_panic_on_warn+0x82/0xa0 kernel/panic.c:236
 ubsan_epilogue lib/ubsan.c:223 [inline]
 __ubsan_handle_out_of_bounds+0x13c/0x150 lib/ubsan.c:348
 dbAdjTree+0x474/0x4f0 fs/jfs/jfs_dmap.c:2867
 dbJoin+0x210/0x2d0 fs/jfs/jfs_dmap.c:2834
 dbFreeBits+0x4eb/0xda0 fs/jfs/jfs_dmap.c:2331
 dbFreeDmap fs/jfs/jfs_dmap.c:2080 [inline]
 dbFree+0x343/0x650 fs/jfs/jfs_dmap.c:402
 txFreeMap+0x798/0xd50 fs/jfs/jfs_txnmgr.c:2534
 txUpdateMap+0x342/0x9e0
 txLazyCommit fs/jfs/jfs_txnmgr.c:2664 [inline]
 jfs_lazycommit+0x47a/0xb70 fs/jfs/jfs_txnmgr.c:2732
 kthread+0x2d3/0x370 kernel/kthread.c:388
 ret_from_fork+0x48/0x80 arch/x86/kernel/process.c:147
 ret_from_fork_asm+0x11/0x20 arch/x86/entry/entry_64.S:304
 </TASK>
Kernel Offset: disabled
Rebooting in 86400 seconds..

The issue is caused when the value of lp becomes greater than
CTLTREESIZE which is the max size of stree. Adding a simple check
solves this issue.

Dave:
As the function returns a void, good error handling
would require a more intrusive code reorganization, so I modified
Osama's patch at use WARN_ON_ONCE for lack of a cleaner option.

The patch is tested via syzbot.

Reported-by: syzbot+39ba34a099ac2e9bd3cb@syzkaller.appspotmail.com
Link: https://syzkaller.appspot.com/bug?extid=39ba34a099ac2e9bd3cb
Signed-off-by: Osama Muhammad <osmtendev@gmail.com>
Signed-off-by: Dave Kleikamp <dave.kleikamp@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

thermal: core: Fix thermal zone suspend-resume synchronization

[ Upstream commit 4e814173a8c4f432fd068b1c796f0416328c9d99 ]

There are 3 synchronization issues with thermal zone suspend-resume
during system-wide transitions:

 1. The resume code runs in a PM notifier which is invoked after user
    space has been thawed, so it can run concurrently with user space
    which can trigger a thermal zone device removal.  If that happens,
    the thermal zone resume code may use a stale pointer to the next
    list element and crash, because it does not hold thermal_list_lock
    while walking thermal_tz_list.

 2. The thermal zone resume code calls thermal_zone_device_init()
    outside the zone lock, so user space or an update triggered by
    the platform firmware may see an inconsistent state of a
    thermal zone leading to unexpected behavior.

 3. Clearing the in_suspend global variable in thermal_pm_notify()
    allows __thermal_zone_device_update() to continue for all thermal
    zones and it may as well run before the thermal_tz_list walk (or
    at any point during the list walk for that matter) and attempt to
    operate on a thermal zone that has not been resumed yet.  It may
    also race destructively with thermal_zone_device_init().

To address these issues, add thermal_list_lock locking to
thermal_pm_notify(), especially arount the thermal_tz_list,
make it call thermal_zone_device_init() back-to-back with
__thermal_zone_device_update() under the zone lock and replace
in_suspend with per-zone bool "suspend" indicators set and unset
under the given zone's lock.

Link: https://lore.kernel.org/linux-pm/20231218162348.69101-1-bo.ye@mediatek.com/
Reported-by: Bo Ye <bo.ye@mediatek.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: APEI: set memory failure flags as MF_ACTION_REQUIRED on synchronous events

[ Upstream commit a70297d2213253853e95f5b49651f924990c6d3b ]

There are two major types of uncorrected recoverable (UCR) errors :

 - Synchronous error: The error is detected and raised at the point of
   the consumption in the execution flow, e.g. when a CPU tries to
   access a poisoned cache line. The CPU will take a synchronous error
   exception such as Synchronous External Abort (SEA) on Arm64 and
   Machine Check Exception (MCE) on X86. OS requires to take action (for
   example, offline failure page/kill failure thread) to recover this
   uncorrectable error.

 - Asynchronous error: The error is detected out of processor execution
   context, e.g. when an error is detected by a background scrubber.
   Some data in the memory are corrupted. But the data have not been
   consumed. OS is optional to take action to recover this uncorrectable
   error.

When APEI firmware first is enabled, a platform may describe one error
source for the handling of synchronous errors (e.g. MCE or SEA notification
), or for handling asynchronous errors (e.g. SCI or External Interrupt
notification). In other words, we can distinguish synchronous errors by
APEI notification. For synchronous errors, kernel will kill the current
process which accessing the poisoned page by sending SIGBUS with
BUS_MCEERR_AR. In addition, for asynchronous errors, kernel will notify the
process who owns the poisoned page by sending SIGBUS with BUS_MCEERR_AO in
early kill mode. However, the GHES driver always sets mf_flags to 0 so that
all synchronous errors are handled as asynchronous errors in memory failure.

To this end, set memory failure flags as MF_ACTION_REQUIRED on synchronous
events.

Signed-off-by: Shuai Xue <xueshuai@linux.alibaba.com>
Tested-by: Ma Wupeng <mawupeng1@huawei.com>
Reviewed-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Xiaofei Tan <tanxiaofei@huawei.com>
Reviewed-by: Baolin Wang <baolin.wang@linux.alibaba.com>
Reviewed-by: James Morse <james.morse@arm.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PM / devfreq: Synchronize devfreq_monitor_[start/stop]

[ Upstream commit aed5ed595960c6d301dcd4ed31aeaa7a8054c0c6 ]

There is a chance if a frequent switch of the governor
done in a loop result in timer list corruption where
timer cancel being done from two place one from
cancel_delayed_work_sync() and followed by expire_timers()
can be seen from the traces[1].

while true
do
        echo "simple_ondemand" > /sys/class/devfreq/1d84000.ufshc/governor
        echo "performance" > /sys/class/devfreq/1d84000.ufshc/governor
done

It looks to be issue with devfreq driver where
device_monitor_[start/stop] need to synchronized so that
delayed work should get corrupted while it is either
being queued or running or being cancelled.

Let's use polling flag and devfreq lock to synchronize the
queueing the timer instance twice and work data being
corrupted.

[1]
...
..
<idle>-0    [003]   9436.209662:  timer_cancel   timer=0xffffff80444f0428
<idle>-0    [003]   9436.209664:  timer_expire_entry   timer=0xffffff80444f0428  now=0x10022da1c  function=__typeid__ZTSFvP10timer_listE_global_addr  baseclk=0x10022da1c
<idle>-0    [003]   9436.209718:  timer_expire_exit   timer=0xffffff80444f0428
kworker/u16:6-14217    [003]   9436.209863:  timer_start   timer=0xffffff80444f0428  function=__typeid__ZTSFvP10timer_listE_global_addr  expires=0x10022da2b  now=0x10022da1c  flags=182452227
vendor.xxxyyy.ha-1593    [004]   9436.209888:  timer_cancel   timer=0xffffff80444f0428
vendor.xxxyyy.ha-1593    [004]   9436.216390:  timer_init   timer=0xffffff80444f0428
vendor.xxxyyy.ha-1593    [004]   9436.216392:  timer_start   timer=0xffffff80444f0428  function=__typeid__ZTSFvP10timer_listE_global_addr  expires=0x10022da2c  now=0x10022da1d  flags=186646532
vendor.xxxyyy.ha-1593    [005]   9436.220992:  timer_cancel   timer=0xffffff80444f0428
xxxyyyTraceManag-7795    [004]   9436.261641:  timer_cancel   timer=0xffffff80444f0428

[2]

 9436.261653][    C4] Unable to handle kernel paging request at virtual address dead00000000012a
[ 9436.261664][    C4] Mem abort info:
[ 9436.261666][    C4]   ESR = 0x96000044
[ 9436.261669][    C4]   EC = 0x25: DABT (current EL), IL = 32 bits
[ 9436.261671][    C4]   SET = 0, FnV = 0
[ 9436.261673][    C4]   EA = 0, S1PTW = 0
[ 9436.261675][    C4] Data abort info:
[ 9436.261677][    C4]   ISV = 0, ISS = 0x00000044
[ 9436.261680][    C4]   CM = 0, WnR = 1
[ 9436.261682][    C4] [dead00000000012a] address between user and kernel address ranges
[ 9436.261685][    C4] Internal error: Oops: 96000044 [#1] PREEMPT SMP
[ 9436.261701][    C4] Skip md ftrace buffer dump for: 0x3a982d0
...

[ 9436.262138][    C4] CPU: 4 PID: 7795 Comm: TraceManag Tainted: G S      W  O      5.10.149-android12-9-o-g17f915d29d0c #1
[ 9436.262141][    C4] Hardware name: Qualcomm Technologies, Inc.  (DT)
[ 9436.262144][    C4] pstate: 22400085 (nzCv daIf +PAN -UAO +TCO BTYPE=--)
[ 9436.262161][    C4] pc : expire_timers+0x9c/0x438
[ 9436.262164][    C4] lr : expire_timers+0x2a4/0x438
[ 9436.262168][    C4] sp : ffffffc010023dd0
[ 9436.262171][    C4] x29: ffffffc010023df0 x28: ffffffd0636fdc18
[ 9436.262178][    C4] x27: ffffffd063569dd0 x26: ffffffd063536008
[ 9436.262182][    C4] x25: 0000000000000001 x24: ffffff88f7c69280
[ 9436.262185][    C4] x23: 00000000000000e0 x22: dead000000000122
[ 9436.262188][    C4] x21: 000000010022da29 x20: ffffff8af72b4e80
[ 9436.262191][    C4] x19: ffffffc010023e50 x18: ffffffc010025038
[ 9436.262195][    C4] x17: 0000000000000240 x16: 0000000000000201
[ 9436.262199][    C4] x15: ffffffffffffffff x14: ffffff889f3c3100
[ 9436.262203][    C4] x13: ffffff889f3c3100 x12: 00000000049f56b8
[ 9436.262207][    C4] x11: 00000000049f56b8 x10: 00000000ffffffff
[ 9436.262212][    C4] x9 : ffffffc010023e50 x8 : dead000000000122
[ 9436.262216][    C4] x7 : ffffffffffffffff x6 : ffffffc0100239d8
[ 9436.262220][    C4] x5 : 0000000000000000 x4 : 0000000000000101
[ 9436.262223][    C4] x3 : 0000000000000080 x2 : ffffff889edc155c
[ 9436.262227][    C4] x1 : ffffff8001005200 x0 : ffffff80444f0428
[ 9436.262232][    C4] Call trace:
[ 9436.262236][    C4]  expire_timers+0x9c/0x438
[ 9436.262240][    C4]  __run_timers+0x1f0/0x330
[ 9436.262245][    C4]  run_timer_softirq+0x28/0x58
[ 9436.262255][    C4]  efi_header_end+0x168/0x5ec
[ 9436.262265][    C4]  __irq_exit_rcu+0x108/0x124
[ 9436.262274][    C4]  __handle_domain_irq+0x118/0x1e4
[ 9436.262282][    C4]  gic_handle_irq.30369+0x6c/0x2bc
[ 9436.262286][    C4]  el0_irq_naked+0x60/0x6c

Link: https://lore.kernel.org/all/1700860318-4025-1-git-send-email-quic_mojha@quicinc.com/
Reported-by: Joyyoung Huang <huangzaiyang@oppo.com>
Acked-by: MyungJoo Ham <myungjoo.ham@samsung.com>
Signed-off-by: Mukesh Ojha <quic_mojha@quicinc.com>
Signed-off-by: Chanwoo Choi <cw00.choi@samsung.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

kunit: tool: fix parsing of test attributes

[ Upstream commit 8ae27bc7fff4ef467a7964821a6cedb34a05d3b2 ]

Add parsing of attributes as diagnostic data. Fixes issue with test plan
being parsed incorrectly as diagnostic data when located after
suite-level attributes.

Note that if there does not exist a test plan line, the diagnostic lines
between the suite header and the first result will be saved in the suite
log rather than the first test case log.

Signed-off-by: Rae Moar <rmoar@google.com>
Reviewed-by: David Gow <davidgow@google.com>
Signed-off-by: Shuah Khan <skhan@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: NUMA: Fix the logic of getting the fake_pxm value

[ Upstream commit e3f577830ce216b0ca21d4750cbbd64cfc21efff ]

The for loop does not iterate over the last element of the node_to_pxm_map
array. This could lead to a conflict between the final fake_pxm value and
the existing pxm values. That is, the final fake_pxm value can not be
guaranteed to be an unused pxm value.

While at it, fix up white space in slit_valid().

Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
[ rjw: Changelog edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/nolibc: fix testcase status alignment

[ Upstream commit 07f679b50252dc9e3d0c19aca5801f82c230c527 ]

Center-align all possible status reports.
Before OK and FAIL were center-aligned in relation to each other but
SKIPPED and FAILED would be left-aligned.

Before:

7 environ_addr = <0x7fffef3e7c50>                                [OK]
8 environ_envp = <0x7fffef3e7c58>                               [FAIL]
9 environ_auxv                                                  [SKIPPED]
10 environ_total                                                [SKIPPED]
11 environ_HOME = <0x7fffef3e99bd>                               [OK]
12 auxv_addr                                                    [SKIPPED]
13 auxv_AT_UID = 1000                                            [OK]

After:

7 environ_addr = <0x7ffff13b00a0>                                 [OK]
8 environ_envp = <0x7ffff13b00a8>                                [FAIL]
9 environ_auxv                                                  [SKIPPED]
10 environ_total                                                [SKIPPED]
11 environ_HOME = <0x7ffff13b19bd>                                [OK]
12 auxv_addr                                                    [SKIPPED]
13 auxv_AT_UID = 1000                                             [OK]

Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: extlog: fix NULL pointer dereference check

[ Upstream commit 72d9b9747e78979510e9aafdd32eb99c7aa30dd1 ]

The gcc plugin -fanalyzer [1] tries to detect various
patterns of incorrect behaviour.  The tool reports:

drivers/acpi/acpi_extlog.c: In function ‘extlog_exit’:
drivers/acpi/acpi_extlog.c:307:12: warning: check of ‘extlog_l1_addr’ for NULL after already dereferencing it [-Wanalyzer-deref-before-check]
    |
    |  306 |         ((struct extlog_l1_head *)extlog_l1_addr)->flags &= ~FLAG_OS_OPTIN;
    |      |         ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~
    |      |                                                  |
    |      |                                                  (1) pointer ‘extlog_l1_addr’ is dereferenced here
    |  307 |         if (extlog_l1_addr)
    |      |            ~
    |      |            |
    |      |            (2) pointer ‘extlog_l1_addr’ is checked for NULL here but it was already dereferenced at (1)
    |

Fix the NULL pointer dereference check in extlog_exit().

Link: https://gcc.gnu.org/onlinedocs/gcc-10.1.0/gcc/Static-Analyzer-Options.html
Signed-off-by: Prarit Bhargava <prarit@redhat.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PNP: ACPI: fix fortify warning

[ Upstream commit ba3f5058db437d919f8468db50483dd9028ff688 ]

When compiling with gcc version 14.0.0 20231126 (experimental)
and CONFIG_FORTIFY_SOURCE=y, I've noticed the following:

In file included from ./include/linux/string.h:295,
                 from ./include/linux/bitmap.h:12,
                 from ./include/linux/cpumask.h:12,
                 from ./arch/x86/include/asm/paravirt.h:17,
                 from ./arch/x86/include/asm/cpuid.h:62,
                 from ./arch/x86/include/asm/processor.h:19,
                 from ./arch/x86/include/asm/cpufeature.h:5,
                 from ./arch/x86/include/asm/thread_info.h:53,
                 from ./include/linux/thread_info.h:60,
                 from ./arch/x86/include/asm/preempt.h:9,
                 from ./include/linux/preempt.h:79,
                 from ./include/linux/spinlock.h:56,
                 from ./include/linux/mmzone.h:8,
                 from ./include/linux/gfp.h:7,
                 from ./include/linux/slab.h:16,
                 from ./include/linux/resource_ext.h:11,
                 from ./include/linux/acpi.h:13,
                 from drivers/pnp/pnpacpi/rsparser.c:11:
In function 'fortify_memcpy_chk',
    inlined from 'pnpacpi_parse_allocated_vendor' at drivers/pnp/pnpacpi/rsparser.c:158:3,
    inlined from 'pnpacpi_allocated_resource' at drivers/pnp/pnpacpi/rsparser.c:249:3:
./include/linux/fortify-string.h:588:25: warning: call to '__read_overflow2_field'
declared with attribute warning: detected read beyond size of field (2nd parameter);
maybe use struct_group()? [-Wattribute-warning]
  588 |                         __read_overflow2_field(q_size_field, size);
      |                         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

According to the comments in include/linux/fortify-string.h, 'memcpy()',
'memmove()' and 'memset()' must not be used beyond individual struct
members to ensure that the compiler can enforce protection against
buffer overflows, and, IIUC, this also applies to partial copies from
the particular member ('vendor->byte_data' in this case). So it should
be better (and safer) to do both copies at once (and 'byte_data' of
'struct acpi_resource_vendor_typed' seems to be a good candidate for
'__counted_by(byte_length)' as well).

Signed-off-by: Dmitry Antipov <dmantipov@yandex.ru>
Reviewed-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: video: Add quirk for the Colorful X15 AT 23 Laptop

[ Upstream commit 143176a46bdd3bfbe9ba2462bf94458e80d65ebf ]

The Colorful X15 AT 23 ACPI video-bus device report spurious
ACPI_VIDEO_NOTIFY_CYCLE events resulting in spurious KEY_SWITCHVIDEOMODE
events being reported to userspace (and causing trouble there) when
an external screen plugged in.

Add a quirk setting the report_key_events mask to
REPORT_BRIGHTNESS_KEY_EVENTS so that the ACPI_VIDEO_NOTIFY_CYCLE
events will be ignored, while still reporting brightness up/down
hotkey-presses to userspace normally.

Signed-off-by: Yuluo Qiu <qyl27@outlook.com>
Co-developed-by: Celeste Liu <CoelacanthusHex@gmail.com>
Signed-off-by: Celeste Liu <CoelacanthusHex@gmail.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

audit: Send netlink ACK before setting connection in auditd_set

[ Upstream commit 022732e3d846e197539712e51ecada90ded0572a ]

When auditd_set sets the auditd_conn pointer, audit messages can
immediately be put on the socket by other kernel threads. If the backlog
is large or the rate is high, this can immediately fill the socket
buffer. If the audit daemon requested an ACK for this operation, a full
socket buffer causes the ACK to get dropped, also setting ENOBUFS on the
socket.

To avoid this race and ensure ACKs get through, fast-track the ACK in
this specific case to ensure it is sent before auditd_conn is set.

Signed-off-by: Chris Riches <chris.riches@nutanix.com>
[PM: fix some tab vs space damage]
Signed-off-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

regulator: core: Only increment use_count when enable_count changes

[ Upstream commit 7993d3a9c34f609c02171e115fd12c10e2105ff4 ]

The use_count of a regulator should only be incremented when the
enable_count changes from 0 to 1. Similarly, the use_count should
only be decremented when the enable_count changes from 1 to 0.

In the previous implementation, use_count was sometimes decremented
to 0 when some consumer called unbalanced disable,
leading to unexpected disable even the regulator is enabled by
other consumers. With this change, the use_count accurately reflects
the number of users which the regulator is enabled.

This should make things more robust in the case where a consumer does
leak references.

Signed-off-by: Rui Zhang <zr.zhang@vivo.com>
Link: https://lore.kernel.org/r/20231103074231.8031-1-zr.zhang@vivo.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

debugobjects: Stop accessing objects after releasing hash bucket lock

[ Upstream commit 9bb6362652f3f4d74a87d572a91ee1b38e673ef6 ]

After release of the hashbucket lock the tracking object can be modified or
freed by a concurrent thread.  Using it in such a case is error prone, even
for printing the object state:

    1. T1 tries to deactivate destroyed object, debugobjects detects it,
       hash bucket lock is released.

    2. T2 preempts T1 and frees the tracking object.

    3. The freed tracking object is allocated and initialized for a
       different to be tracked kernel object.

    4. T1 resumes and reports error for wrong kernel object.

Create a local copy of the tracking object before releasing the hash bucket
lock and use the local copy for reporting and fixups to prevent this.

Signed-off-by: Andrzej Hajda <andrzej.hajda@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20231025-debugobjects_fix-v3-1-2bc3bf7084c2@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf/core: Fix narrow startup race when creating the perf nr_addr_filters sysfs file

[ Upstream commit 652ffc2104ec1f69dd4a46313888c33527145ccf ]

Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/2023061204-decal-flyable-6090@gregkh
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/mce: Mark fatal MCE's page as poison to avoid panic in the kdump kernel

[ Upstream commit 9f3b130048bfa2e44a8cfb1b616f826d9d5d8188 ]

Memory errors don't happen very often, especially fatal ones. However,
in large-scale scenarios such as data centers, that probability
increases with the amount of machines present.

When a fatal machine check happens, mce_panic() is called based on the
severity grading of that error. The page containing the error is not
marked as poison.

However, when kexec is enabled, tools like makedumpfile understand when
pages are marked as poison and do not touch them so as not to cause
a fatal machine check exception again while dumping the previous
kernel's memory.

Therefore, mark the page containing the error as poisoned so that the
kexec'ed kernel can avoid accessing the page.

  [ bp: Rewrite commit message and comment. ]

Co-developed-by: Youquan Song <youquan.song@intel.com>
Signed-off-by: Youquan Song <youquan.song@intel.com>
Signed-off-by: Zhiquan Li <zhiquan1.li@intel.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Naoya Horiguchi <naoya.horiguchi@nec.com>
Link: https://lore.kernel.org/r/20231014051754.3759099-1-zhiquan1.li@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/lib: Validate size for vector operations

[ Upstream commit 8f9abaa6d7de0a70fc68acaedce290c1f96e2e59 ]

Some of the fp/vmx code in sstep.c assume a certain maximum size for the
instructions being emulated. The size of those operations however is
determined separately in analyse_instr().

Add a check to validate the assumption on the maximum size of the
operations, so as to prevent any unintended kernel stack corruption.

Signed-off-by: Naveen N Rao <naveen@kernel.org>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Build-tested-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231123071705.397625-1-naveen@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc: pmd_move_must_withdraw() is only needed for CONFIG_TRANSPARENT_HUGEPAGE

[ Upstream commit 0d555b57ee660d8a871781c0eebf006e855e918d ]

The linux-next build of powerpc64 allnoconfig fails with:

  arch/powerpc/mm/book3s64/pgtable.c:557:5: error: no previous prototype for 'pmd_move_must_withdraw'
    557 | int pmd_move_must_withdraw(struct spinlock *new_pmd_ptl,
        |     ^~~~~~~~~~~~~~~~~~~~~~

Caused by commit:

  c6345dfa6e3e ("Makefile.extrawarn: turn on missing-prototypes globally")

Fix it by moving the function definition under
CONFIG_TRANSPARENT_HUGEPAGE like the prototype. The function is only
called when CONFIG_TRANSPARENT_HUGEPAGE=y.

Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
[mpe: Flesh out change log from linux-next patch]
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231127132809.45c2b398@canb.auug.org.au
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/boot: Ignore NMIs during very early boot

[ Upstream commit 78a509fba9c9b1fcb77f95b7c6be30da3d24823a ]

When there are two racing NMIs on x86, the first NMI invokes NMI handler and
the 2nd NMI is latched until IRET is executed.

If panic on NMI and panic kexec are enabled, the first NMI triggers
panic and starts booting the next kernel via kexec. Note that the 2nd
NMI is still latched. During the early boot of the next kernel, once
an IRET is executed as a result of a page fault, then the 2nd NMI is
unlatched and invokes the NMI handler.

However, NMI handler is not set up at the early stage of boot, which
results in a boot failure.

Avoid such problems by setting up a NOP handler for early NMIs.

[ mingo: Refined the changelog. ]

Signed-off-by: Jun'ichi Nomura <junichi.nomura@nec.com>
Signed-off-by: Derek Barbosa <debarbos@redhat.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Kees Cook <keescook@chromium.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Paul E. McKenney <paulmck@kernel.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/64s: Fix CONFIG_NUMA=n build due to create_section_mapping()

[ Upstream commit ede66cd22441820cbd399936bf84fdc4294bc7fa ]

With CONFIG_NUMA=n the build fails with:

  arch/powerpc/mm/book3s64/pgtable.c:275:15: error: no previous prototype for ‘create_section_mapping’ [-Werror=missing-prototypes]
  275 | int __meminit create_section_mapping(unsigned long start, unsigned long end,
      |               ^~~~~~~~~~~~~~~~~~~~~~

That happens because the prototype for create_section_mapping() is in
asm/mmzone.h, but asm/mmzone.h is only included by linux/mmzone.h
when CONFIG_NUMA=y.

In fact the prototype is only needed by arch/powerpc/mm code, so move
the prototype into arch/powerpc/mm/mmu_decl.h, which also fixes the
build error.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231129131919.2528517-5-mpe@ellerman.id.au
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/mm: Fix build failures due to arch_reserved_kernel_pages()

[ Upstream commit d8c3f243d4db24675b653f0568bb65dae34e6455 ]

With NUMA=n and FA_DUMP=y or PRESERVE_FA_DUMP=y the build fails with:

  arch/powerpc/kernel/fadump.c:1739:22: error: no previous prototype for ‘arch_reserved_kernel_pages’ [-Werror=missing-prototypes]
  1739 | unsigned long __init arch_reserved_kernel_pages(void)
       |                      ^~~~~~~~~~~~~~~~~~~~~~~~~~

The prototype for arch_reserved_kernel_pages() is in include/linux/mm.h,
but it's guarded by __HAVE_ARCH_RESERVED_KERNEL_PAGES. The powerpc
headers define __HAVE_ARCH_RESERVED_KERNEL_PAGES in asm/mmzone.h, which
is not included into the generic headers when NUMA=n.

Move the definition of __HAVE_ARCH_RESERVED_KERNEL_PAGES into asm/mmu.h
which is included regardless of NUMA=n.

Additionally the ifdef around __HAVE_ARCH_RESERVED_KERNEL_PAGES needs to
also check for CONFIG_PRESERVE_FA_DUMP.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231130114433.3053544-1-mpe@ellerman.id.au
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc: Fix build error due to is_valid_bugaddr()

[ Upstream commit f8d3555355653848082c351fa90775214fb8a4fa ]

With CONFIG_GENERIC_BUG=n the build fails with:

  arch/powerpc/kernel/traps.c:1442:5: error: no previous prototype for ‘is_valid_bugaddr’ [-Werror=missing-prototypes]
  1442 | int is_valid_bugaddr(unsigned long addr)
       |     ^~~~~~~~~~~~~~~~

The prototype is only defined, and the function is only needed, when
CONFIG_GENERIC_BUG=y, so move the implementation under that.

Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231130114433.3053544-2-mpe@ellerman.id.au
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers/perf: pmuv3: don't expose SW_INCR event in sysfs

[ Upstream commit ca6f537e459e2da4b331fe8928d1a0b0f9301f42 ]

The SW_INCR event is somewhat unusual, and depends on the specific HW
counter that it is programmed into. When programmed into PMEVCNTR<n>,
SW_INCR will count any writes to PMSWINC_EL0 with bit n set, ignoring
writes to SW_INCR with bit n clear.

Event rotation means that there's no fixed relationship between
perf_events and HW counters, so this isn't all that useful.

Further, we program PMUSERENR.{SW,EN}=={0,0}, which causes EL0 writes to
PMSWINC_EL0 to be trapped and handled as UNDEFINED, resulting in a
SIGILL to userspace.

Given that, it's not a good idea to expose SW_INCR in sysfs. Hide it as
we did for CHAIN back in commit:

  4ba2578fa7b55701 ("arm64: perf: don't expose CHAIN event in sysfs")

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20231204115847.2993026-1-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

arm64: irq: set the correct node for VMAP stack

[ Upstream commit 75b5e0bf90bffaca4b1f19114065dc59f5cc161f ]

In current code, init_irq_stacks() will call cpu_to_node().
The cpu_to_node() depends on percpu "numa_node" which is initialized in:
     arch_call_rest_init() --> rest_init() -- kernel_init()
--> kernel_init_freeable() --> smp_prepare_cpus()

But init_irq_stacks() is called in init_IRQ() which is before
arch_call_rest_init().

So in init_irq_stacks(), the cpu_to_node() does not work, it
always return 0. In NUMA, it makes the node 1 cpu accesses the IRQ stack which
is in the node 0.

This patch fixes it by:
  1.) export the early_cpu_to_node(), and use it in the init_irq_stacks().
  2.) change init_irq_stacks() to __init function.

Reviewed-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Huang Shijie <shijie@os.amperecomputing.com>
Link: https://lore.kernel.org/r/20231124031513.81548-1-shijie@os.amperecomputing.com
Signed-off-by: Will Deacon <will@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/mm: Fix null-pointer dereference in pgtable_cache_add

[ Upstream commit f46c8a75263f97bda13c739ba1c90aced0d3b071 ]

kasprintf() returns a pointer to dynamically allocated memory
which can be NULL upon failure. Ensure the allocation was successful
by checking the pointer validity.

Suggested-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Suggested-by: Michael Ellerman <mpe@ellerman.id.au>
Signed-off-by: Kunwu Chan <chentao@kylinos.cn>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20231204023223.2447523-1-chentao@kylinos.cn
Signed-off-by: Sasha Levin <sashal@kernel.org>