i2c: mlxbf: support lock mechanism

[ Upstream commit 86067ccfa1424a26491542d6f6d7546d40b61a10 ]

Linux is not the only entity using the BlueField I2C busses so
support a lock mechanism provided by hardware to avoid issues
when multiple entities are trying to access the same bus.

The lock is acquired whenever written explicitely or the lock
register is read. So make sure it is always released at the end
of a successful or failed transaction.

Fixes: b5b5b32081cd206b (i2c: mlxbf: I2C SMBus driver for Mellanox BlueField SoC)
Reviewed-by: Khalil Blaiech <kblaiech@nvidia.com>
Signed-off-by: Asmaa Mnebhi <asmaa@nvidia.com>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

skmsg: Schedule psock work if the cached skb exists on the psock

[ Upstream commit bec217197b412d74168c6a42fc0f76d0cc9cad00 ]

In sk_psock_backlog function, for ingress direction skb, if no new data
packet arrives after the skb is cached, the cached skb does not have a
chance to be added to the receive queue of psock. As a result, the cached
skb cannot be received by the upper-layer application. Fix this by reschedule
the psock work to dispose the cached skb in sk_msg_recvmsg function.

Fixes: 604326b41a6f ("bpf, sockmap: convert to generic sk_msg interface")
Signed-off-by: Liu Jian <liujian56@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20220907071311.60534-1-liujian56@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi/omap100k:Fix PM disable depth imbalance in omap1_spi100k_probe

[ Upstream commit 29f65f2171c85a9633daa380df14009a365f42f2 ]

The pm_runtime_enable will increase power disable depth. Thus
a pairing decrement is needed on the error handling path to
keep it balanced according to context.

Fixes:db91841b58f9a ("spi/omap100k: Convert to runtime PM")

Signed-off-by: Zhang Qilong <zhangqilong3@huawei.com>
Link: https://lore.kernel.org/r/20220924121310.78331-4-zhangqilong3@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: dw: Fix PM disable depth imbalance in dw_spi_bt1_probe

[ Upstream commit 618d815fc93477b1675878f3c04ff32657cc18b4 ]

The pm_runtime_enable will increase power disable depth. Thus
a pairing decrement is needed on the error handling path to
keep it balanced according to context.

Fixes:abf00907538e2 ("spi: dw: Add Baikal-T1 SPI Controller glue driver")

Signed-off-by: Zhang Qilong <zhangqilong3@huawei.com>
Link: https://lore.kernel.org/r/20220924121310.78331-3-zhangqilong3@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/cpu: Include the header of init_ia32_feat_ctl()'s prototype

[ Upstream commit 30ea703a38ef76ca119673cd8bdd05c6e068e2ac ]

Include the header containing the prototype of init_ia32_feat_ctl(),
solving the following warning:

  $ make W=1 arch/x86/kernel/cpu/feat_ctl.o
  arch/x86/kernel/cpu/feat_ctl.c:112:6: warning: no previous prototype for ‘init_ia32_feat_ctl’ [-Wmissing-prototypes]
    112 | void init_ia32_feat_ctl(struct cpuinfo_x86 *c)

This warning appeared after commit

  5d5103595e9e5 ("x86/cpu: Reinitialize IA32_FEAT_CTL MSR on BSP during wakeup")

had moved the function init_ia32_feat_ctl()'s prototype from
arch/x86/kernel/cpu/cpu.h to arch/x86/include/asm/cpu.h.

Note that, before the commit mentioned above, the header include "cpu.h"
(arch/x86/kernel/cpu/cpu.h) was added by commit

  0e79ad863df43 ("x86/cpu: Fix a -Wmissing-prototypes warning for init_ia32_feat_ctl()")

solely to fix init_ia32_feat_ctl()'s missing prototype. So, the header
include "cpu.h" is no longer necessary.

  [ bp: Massage commit message. ]

Fixes: 5d5103595e9e5 ("x86/cpu: Reinitialize IA32_FEAT_CTL MSR on BSP during wakeup")
Signed-off-by: Luciano Leão <lucianorsleao@gmail.com>
Signed-off-by: Borislav Petkov <bp@suse.de>
Reviewed-by: Nícolas F. R. A. Prado <n@nfraprado.net>
Link: https://lore.kernel.org/r/20220922200053.1357470-1-lucianorsleao@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/microcode/AMD: Track patch allocation size explicitly

[ Upstream commit 712f210a457d9c32414df246a72781550bc23ef6 ]

In preparation for reducing the use of ksize(), record the actual
allocation size for later memcpy(). This avoids copying extra
(uninitialized!) bytes into the patch buffer when the requested
allocation size isn't exactly the size of a kmalloc bucket.
Additionally, fix potential future issues where runtime bounds checking
will notice that the buffer was allocated to a smaller value than
returned by ksize().

Fixes: 757885e94a22 ("x86, microcode, amd: Early microcode patch loading support for AMD")
Suggested-by: Daniel Micay <danielmicay@gmail.com>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Borislav Petkov <bp@suse.de>
Link: https://lore.kernel.org/lkml/CA+DvKQ+bp7Y7gmaVhacjv9uF6Ar-o4tet872h4Q8RPYPJjcJQA@mail.gmail.com/
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath11k: fix number of VHT beamformee spatial streams

[ Upstream commit 55b5ee3357d7bb98ee578cf9b84a652e7a1bc199 ]

The number of spatial streams used when acting as a beamformee in VHT
mode are reported by the firmware as 7 (8 sts - 1) both in IPQ6018 and
IPQ8074 which respectively have 2 and 4 sts each. So the firmware should
report 1 (2 - 1) and 3 (4 - 1).

Fix this by checking that the number of VHT beamformee sts reported by
the firmware is not greater than the number of receiving antennas - 1.
The fix is based on the same approach used in this same function for
sanitizing the number of sounding dimensions reported by the firmware.

Without this change, acting as a beamformee in VHT mode is not working
properly.

Tested-on: IPQ6018 hw1.0 AHB WLAN.HK.2.5.0.1-01208-QCAHKSWPL_SILICONZ-1
Tested-on: IPQ8074 hw2.0 AHB WLAN.HK.2.5.0.1-01208-QCAHKSWPL_SILICONZ-1

Fixes: d5c65159f289 ("ath11k: driver for Qualcomm IEEE 802.11ax devices")
Signed-off-by: Jesus Fernandez Manzano <jesus.manzano@galgus.net>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220616173947.21901-1-jesus.manzano@galgus.net
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: conntrack: revisit the gc initial rescheduling bias

[ Upstream commit 2aa192757005f130b2dd3547dda6e462e761199f ]

The previous commit changed the way the rescheduling delay is computed
which has a side effect: the bias is now represented as much as the
other entries in the rescheduling delay which makes the logic to kick in
only with very large sets, as the initial interval is very large
(INT_MAX).

Revisit the GC initial bias to allow more frequent GC for smaller sets
while still avoiding wakeups when a machine is mostly idle. We're moving
from a large initial value to pretending we have 100 entries expiring at
the upper bound. This way only a few entries having a small timeout
won't impact much the rescheduling delay and non-idle machines will have
enough entries to lower the delay when needed. This also improves
readability as the initial bias is now linked to what is computed
instead of being an arbitrary large value.

Fixes: 2cfadb761d3d ("netfilter: conntrack: revisit gc autotuning")
Suggested-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Antoine Tenart <atenart@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

netfilter: conntrack: fix the gc rescheduling delay

[ Upstream commit 95eabdd207024312876d0ebed90b4c977e050e85 ]

Commit 2cfadb761d3d ("netfilter: conntrack: revisit gc autotuning")
changed the eviction rescheduling to the use average expiry of scanned
entries (within 1-60s) by doing:

  for (...) {
      expires = clamp(nf_ct_expires(tmp), ...);
      next_run += expires;
      next_run /= 2;
  }

The issue is the above will make the average ('next_run' here) more
dependent on the last expiration values than the firsts (for sets > 2).
Depending on the expiration values used to compute the average, the
result can be quite different than what's expected. To fix this we can
do the following:

  for (...) {
      expires = clamp(nf_ct_expires(tmp), ...);
      next_run += (expires - next_run) / ++count;
  }

Fixes: 2cfadb761d3d ("netfilter: conntrack: revisit gc autotuning")
Cc: Florian Westphal <fw@strlen.de>
Signed-off-by: Antoine Tenart <atenart@kernel.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: hci_{ldisc,serdev}: check percpu_init_rwsem() failure

[ Upstream commit 3124d320c22f3f4388d9ac5c8f37eaad0cefd6b1 ]

syzbot is reporting NULL pointer dereference at hci_uart_tty_close() [1],
for rcu_sync_enter() is called without rcu_sync_init() due to
hci_uart_tty_open() ignoring percpu_init_rwsem() failure.

While we are at it, fix that hci_uart_register_device() ignores
percpu_init_rwsem() failure and hci_uart_unregister_device() does not
call percpu_free_rwsem().

Link: https://syzkaller.appspot.com/bug?extid=576dfca25381fb6fbc5f
Reported-by: syzbot <syzbot+576dfca25381fb6fbc5f@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Fixes: 67d2f8781b9f00d1 ("Bluetooth: hci_ldisc: Allow sleeping while proto locks are held.")
Fixes: d73e172816652772 ("Bluetooth: hci_serdev: Init hci_uart proto_lock to avoid oops")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Ensure correct locking around vulnerable function find_vpid()

[ Upstream commit 83c10cc362d91c0d8d25e60779ee52fdbbf3894d ]

The documentation for find_vpid() clearly states:

  "Must be called with the tasklist_lock or rcu_read_lock() held."

Presently we do neither for find_vpid() instance in bpf_task_fd_query().
Add proper rcu_read_lock/unlock() to fix the issue.

Fixes: 41bdc4b40ed6f ("bpf: introduce bpf subcommand BPF_TASK_FD_QUERY")
Signed-off-by: Lee Jones <lee@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Yonghong Song <yhs@fb.com>
Link: https://lore.kernel.org/bpf/20220912133855.1218900-1-lee@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

net: fs_enet: Fix wrong check in do_pd_setup

[ Upstream commit ec3f06b542a960806a81345042e4eee3f8c5dec4 ]

Should check of_iomap return value 'fep->fec.fecp' instead of 'fep->fcc.fccp'

Fixes: 976de6a8c304 ("fs_enet: Be an of_platform device when CONFIG_PPC_CPM_NEW_BINDING is set.")
Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: RFCOMM: Fix possible deadlock on socket shutdown/release

[ Upstream commit 812e92b824c1db16c9519f8624d48a9901a0d38f ]

Due to change to switch to use lock_sock inside rfcomm_sk_state_change
the socket shutdown/release procedure can cause a deadlock:

    rfcomm_sock_shutdown():
      lock_sock();
      __rfcomm_sock_close():
        rfcomm_dlc_close():
          __rfcomm_dlc_close():
            rfcomm_dlc_lock();
            rfcomm_sk_state_change():
              lock_sock();

To fix this when the call __rfcomm_sock_close is now done without
holding the lock_sock since rfcomm_dlc_lock exists to protect
the dlc data there is no need to use lock_sock in that code path.

Link: https://lore.kernel.org/all/CAD+dNTsbuU4w+Y_P7o+VEN7BYCAbZuwZx2+tH+OTzCdcZF82YA@mail.gmail.com/
Fixes: b7ce436a5d79 ("Bluetooth: switch to lock_sock in RFCOMM")
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mt76: mt7915: do not check state before configuring implicit beamform

[ Upstream commit d2b5bb6dfab29fe32bedefaade88dcd182c03a00 ]

Do not need to check running state before configuring implicit Tx
beamform. It is okay to configure implicit Tx beamform in run time.
Noted that the existing connected stations will be applied for new
configuration only if they reconnected to the interface.

Fixes: 6d6dc980e07d ("mt76: mt7915: add implicit Tx beamforming support")
Signed-off-by: Howard Hsu <howard-yh.hsu@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mt76: mt7615: add mt7615_mutex_acquire/release in mt7615_sta_set_decap_offload

[ Upstream commit 765c69d477a44c088e5d19e7758dfa4db418e3ba ]

Similar to mt7921 driver, introduce mt7615_mutex_acquire/release in
mt7615_sta_set_decap_offload in order to avoid sending mcu commands
while the device is in low-power state.

Fixes: d4b98c63d7a77 ("mt76: mt7615: add support for rx decapsulation offload")
Signed-off-by: Lorenzo Bianconi <lorenzo@kernel.org>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mt76: sdio: fix transmitting packet hangs

[ Upstream commit 250b1827205846ff346a76044955cb79d4963f70 ]

Fix transmitting packets hangs with continuing to pull the pending packet
from mac80211 queues when receiving Tx status notification from the device.

Fixes: aac5104bf631 ("mt76: sdio: do not run mt76_txq_schedule directly")
Acked-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: YN Chen <yn.chen@mediatek.com>
Signed-off-by: Felix Fietkau <nbd@nbd.name>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtl8xxxu: Remove copy-paste leftover in gen2_update_rate_mask

[ Upstream commit d5350756c03cdf18696295c6b11d7acc4dbf825c ]

It looks like a leftover from copying rtl8xxxu_update_rate_mask,
which is used with the gen1 chips.

It wasn't causing any problems for my RTL8188FU test device, but it's
clearly a mistake, so remove it.

Fixes: f653e69009c6 ("rtl8xxxu: Implement basic 8723b specific update_rate_mask() function")
Signed-off-by: Bitterblue Smith <rtl8821cerfe2@gmail.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/d5544fe8-9798-28f1-54bd-6839a1974b10@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtl8xxxu: gen2: Fix mistake in path B IQ calibration

[ Upstream commit e963a19c64ac0d2f8785d36a27391abd91ac77aa ]

Found by comparing with the vendor driver. Currently this affects
only the RTL8192EU, which is the only gen2 chip with 2 TX paths
supported by this driver. It's unclear what kind of effect the
mistake had in practice, since I don't have any RTL8192EU devices
to test it.

Fixes: e1547c535ede ("rtl8xxxu: First stab at adding IQK calibration for 8723bu parts")
Signed-off-by: Bitterblue Smith <rtl8821cerfe2@gmail.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/30a59f3a-cfa9-8379-7af0-78a8f4c77cfd@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: btf: fix truncated last_member_type_id in btf_struct_resolve

[ Upstream commit a37a32583e282d8d815e22add29bc1e91e19951a ]

When trying to finish resolving a struct member, btf_struct_resolve
saves the member type id in a u16 temporary variable. This truncates
the 32 bit type id value if it exceeds UINT16_MAX.

As a result, structs that have members with type ids > UINT16_MAX and
which need resolution will fail with a message like this:

    [67414] STRUCT ff_device size=120 vlen=12
        effect_owners type_id=67434 bits_offset=960 Member exceeds struct_size

Fix this by changing the type of last_member_type_id to u32.

Fixes: a0791f0df7d2 ("bpf: fix BTF limits")
Reviewed-by: Stanislav Fomichev <sdf@google.com>
Signed-off-by: Lorenz Bauer <oss@lmb.io>
Link: https://lore.kernel.org/r/20220910110120.339242-1-oss@lmb.io
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: meson-spicc: do not rely on busy flag in pow2 clk ops

[ Upstream commit 36acf80fc0c4b5ebe6fa010b524d442ee7f08fd3 ]

Since [1], controller's busy flag isn't set anymore when the
__spi_transfer_message_noqueue() is used instead of the
__spi_pump_transfer_message() logic for spi_sync transfers.

Since the pow2 clock ops were limited to only be available when a
transfer is ongoing (between prepare_transfer_hardware and
unprepare_transfer_hardware callbacks), the only way to track this
down is to check for the controller cur_msg.

[1] ae7d2346dc89 ("spi: Don't use the message queue if possible in spi_sync")

Fixes: 09992025dacd ("spi: meson-spicc: add local pow2 clock ops to preserve rate between messages")
Fixes: ae7d2346dc89 ("spi: Don't use the message queue if possible in spi_sync")
Reported-by: Markus Schneider-Pargmann <msp@baylibre.com>
Signed-off-by: Neil Armstrong <narmstrong@baylibre.com>
Tested-by: Markus Schneider-Pargmann <msp@baylibre.com>
Link: https://lore.kernel.org/r/20220908121803.919943-1-narmstrong@baylibre.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtl8xxxu: Fix skb misuse in TX queue selection

[ Upstream commit edd5747aa12ed61a5ecbfa58d3908623fddbf1e8 ]

rtl8xxxu_queue_select() selects the wrong TX queues because it's
reading memory from the wrong address. It expects to find ieee80211_hdr
at skb->data, but that's not the case after skb_push(). Move the call
to rtl8xxxu_queue_select() before the call to skb_push().

Fixes: 26f1fad29ad9 ("New driver: rtl8xxxu (mac80211)")
Signed-off-by: Bitterblue Smith <rtl8821cerfe2@gmail.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/7fa4819a-4f20-b2af-b7a6-8ee01ac49295@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: qup: add missing clk_disable_unprepare on error in spi_qup_pm_resume_runtime()

[ Upstream commit 494a22765ce479c9f8ad181c5d24cffda9f534bb ]

Add the missing clk_disable_unprepare() before return
from spi_qup_pm_resume_runtime() in the error handling case.

Fixes: dae1a7700b34 (“spi: qup: Handle clocks in pm_runtime suspend and resume”)
Signed-off-by: Xu Qiang <xuqiang36@huawei.com>
Link: https://lore.kernel.org/r/20220825065324.68446-2-xuqiang36@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: qup: add missing clk_disable_unprepare on error in spi_qup_resume()

[ Upstream commit 70034320fdc597b8f58b4a43bb547f17c4c5557a ]

Add the missing clk_disable_unprepare() before return
from spi_qup_resume() in the error handling case.

Fixes: 64ff247a978f (“spi: Add Qualcomm QUP SPI controller support”)
Signed-off-by: Xu Qiang <xuqiang36@huawei.com>
Link: https://lore.kernel.org/r/20220825065324.68446-1-xuqiang36@huawei.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/xsk: Avoid use-after-free on ctx

[ Upstream commit af515a5587b8f45f19e11657746e0c89411b0380 ]

The put lowers the reference count to 0 and frees ctx, reading it
afterwards is invalid. Move the put after the uses and determine the
last use by the reference count being 1.

Fixes: 39e940d4abfa ("selftests/xsk: Destroy BPF resources only when ctx refcount drops to 0")
Signed-off-by: Ian Rogers <irogers@google.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Magnus Karlsson <magnus.karlsson@intel.com>
Link: https://lore.kernel.org/bpf/20220901202645.1463552-1-irogers@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtw88: add missing destroy_workqueue() on error path in rtw_core_init()

[ Upstream commit b0ea758b30bbdf7c4323c78b7c50c05d2e1224d5 ]

Add the missing destroy_workqueue() before return from rtw_core_init()
in error path.

Fixes: fe101716c7c9 ("rtw88: replace tx tasklet with work queue")
Signed-off-by: Yang Yingliang <yangyingliang@huawei.com>
Reviewed-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220826023817.3908255-1-yangyingliang@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtl8xxxu: tighten bounds checking in rtl8xxxu_read_efuse()

[ Upstream commit 620d5eaeb9059636864bda83ca1c68c20ede34a5 ]

There some bounds checking to ensure that "map_addr" is not out of
bounds before the start of the loop.  But the checking needs to be
done as we iterate through the loop because "map_addr" gets larger as
we iterate.

Fixes: 26f1fad29ad9 ("New driver: rtl8xxxu (mac80211)")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Jes Sorensen <Jes.Sorensen@gmail.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/Yv8eGLdBslLAk3Ct@kili
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: btusb: mediatek: fix WMT failure during runtime suspend

[ Upstream commit fd3f106677bac70437dc12e76c827294ed495a44 ]

WMT cmd/event doesn't follow up the generic HCI cmd/event handling, it
needs constantly polling control pipe until the host received the WMT
event, thus, we should require to specifically acquire PM counter on the
USB to prevent the interface from entering auto suspended while WMT
cmd/event in progress.

Fixes: a1c49c434e15 ("Bluetooth: btusb: Add protocol support for MediaTek MT7668U USB devices")
Co-developed-by: Jing Cai <jing.cai@mediatek.com>
Signed-off-by: Jing Cai <jing.cai@mediatek.com>
Signed-off-by: Sean Wang <sean.wang@mediatek.com>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Use this_cpu_{inc|dec|inc_return} for bpf_task_storage_busy

[ Upstream commit 197827a05e13808c60f52632e9887eede63f1c16 ]

Now migrate_disable() does not disable preemption and under some
architectures (e.g. arm64) __this_cpu_{inc|dec|inc_return} are neither
preemption-safe nor IRQ-safe, so for fully preemptible kernel concurrent
lookups or updates on the same task local storage and on the same CPU
may make bpf_task_storage_busy be imbalanced, and
bpf_task_storage_trylock() on the specific cpu will always fail.

Fixing it by using this_cpu_{inc|dec|inc_return} when manipulating
bpf_task_storage_busy.

Fixes: bc235cdb423a ("bpf: Prevent deadlock from recursive bpf_task_storage_[get|delete]")
Signed-off-by: Hou Tao <houtao1@huawei.com>
Acked-by: Alexei Starovoitov <ast@kernel.org>
Link: https://lore.kernel.org/r/20220901061938.3789460-2-houtao@huaweicloud.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Propagate error from htab_lock_bucket() to userspace

[ Upstream commit 66a7a92e4d0d091e79148a4c6ec15d1da65f4280 ]

In __htab_map_lookup_and_delete_batch() if htab_lock_bucket() returns
-EBUSY, it will go to next bucket. Going to next bucket may not only
skip the elements in current bucket silently, but also incur
out-of-bound memory access or expose kernel memory to userspace if
current bucket_cnt is greater than bucket_size or zero.

Fixing it by stopping batch operation and returning -EBUSY when
htab_lock_bucket() fails, and the application can retry or skip the busy
batch as needed.

Fixes: 20b6cc34ea74 ("bpf: Avoid hashtab deadlock with map_locked")
Reported-by: Hao Sun <sunhao.th@gmail.com>
Signed-off-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/r/20220831042629.130006-3-houtao@huaweicloud.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Disable preemption when increasing per-cpu map_locked

[ Upstream commit 2775da21628738ce073a3a6a806adcbaada0f091 ]

Per-cpu htab->map_locked is used to prohibit the concurrent accesses
from both NMI and non-NMI contexts. But since commit 74d862b682f5
("sched: Make migrate_disable/enable() independent of RT"),
migrate_disable() is also preemptible under CONFIG_PREEMPT case, so now
map_locked also disallows concurrent updates from normal contexts
(e.g. userspace processes) unexpectedly as shown below:

process A                      process B

htab_map_update_elem()
  htab_lock_bucket()
    migrate_disable()
    /* return 1 */
    __this_cpu_inc_return()
    /* preempted by B */

                               htab_map_update_elem()
                                 /* the same bucket as A */
                                 htab_lock_bucket()
                                   migrate_disable()
                                   /* return 2, so lock fails */
                                   __this_cpu_inc_return()
                                   return -EBUSY

A fix that seems feasible is using in_nmi() in htab_lock_bucket() and
only checking the value of map_locked for nmi context. But it will
re-introduce dead-lock on bucket lock if htab_lock_bucket() is re-entered
through non-tracing program (e.g. fentry program).

One cannot use preempt_disable() to fix this issue as htab_use_raw_lock
being false causes the bucket lock to be a spin lock which can sleep and
does not work with preempt_disable().

Therefore, use migrate_disable() when using the spinlock instead of
preempt_disable() and defer fixing concurrent updates to when the kernel
has its own BPF memory allocator.

Fixes: 74d862b682f5 ("sched: Make migrate_disable/enable() independent of RT")
Reviewed-by: Hao Luo <haoluo@google.com>
Signed-off-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/r/20220831042629.130006-2-houtao@huaweicloud.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

xsk: Fix backpressure mechanism on Tx

[ Upstream commit c00c4461689e15ac2cc3b9a595a54e4d8afd3d77 ]

Commit d678cbd2f867 ("xsk: Fix handling of invalid descriptors in XSK TX
batching API") fixed batch API usage against set of descriptors with
invalid ones but introduced a problem when AF_XDP SW rings are smaller
than HW ones. Mismatch of reported Tx'ed frames between HW generator and
user space app was observed. It turned out that backpressure mechanism
became a bottleneck when the amount of produced descriptors to CQ is
lower than what we grabbed from XSK Tx ring.

Say that 512 entries had been taken from XSK Tx ring but we had only 490
free entries in CQ. Then callsite (ZC driver) will produce only 490
entries onto HW Tx ring but 512 entries will be released from Tx ring
and this is what will be seen by the user space.

In order to fix this case, mix XSK Tx/CQ ring interractions by moving
around internal functions and changing call order:

*  pull out xskq_prod_nb_free() from xskq_prod_reserve_addr_batch()
   up to xsk_tx_peek_release_desc_batch();
** move xskq_cons_release_n() into xskq_cons_read_desc_batch()

After doing so, algorithm can be described as follows:

1. lookup Tx entries
2. use value from 1. to reserve space in CQ (*)
3. Read from Tx ring as much descriptors as value from 2
 3a. release descriptors from XSK Tx ring (**)
4. Finally produce addresses to CQ

Fixes: d678cbd2f867 ("xsk: Fix handling of invalid descriptors in XSK TX batching API")
Signed-off-by: Magnus Karlsson <magnus.karlsson@intel.com>
Signed-off-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20220830121705.8618-1-maciej.fijalkowski@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/resctrl: Fix to restore to original value when re-enabling hardware prefetch register

[ Upstream commit 499c8bb4693d1c8d8f3d6dd38e5bdde3ff5bd906 ]

The current pseudo_lock.c code overwrites the value of the
MSR_MISC_FEATURE_CONTROL to 0 even if the original value is not 0.
Therefore, modify it to save and restore the original values.

Fixes: 018961ae5579 ("x86/intel_rdt: Pseudo-lock region creation/removal core")
Fixes: 443810fe6160 ("x86/intel_rdt: Create debugfs files for pseudo-locking testing")
Fixes: 8a2fc0e1bc0c ("x86/intel_rdt: More precise L2 hit/miss measurements")
Signed-off-by: Kohei Tarumizu <tarumizu.kohei@fujitsu.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Acked-by: Reinette Chatre <reinette.chatre@intel.com>
Link: https://lkml.kernel.org/r/eb660f3c2010b79a792c573c02d01e8e841206ad.1661358182.git.reinette.chatre@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

spi: mt7621: Fix an error message in mt7621_spi_probe()

[ Upstream commit 2b2bf6b7faa9010fae10dc7de76627a3fdb525b3 ]

'status' is known to be 0 at this point. The expected error code is
PTR_ERR(clk).

Switch to dev_err_probe() in order to display the expected error code (in a
human readable way).
This also filters -EPROBE_DEFER cases, should it happen.

Fixes: 1ab7f2a43558 ("staging: mt7621-spi: add mt7621 support")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Matthias Brugger <matthias.bgg@gmail.com>
Link: https://lore.kernel.org/r/928f3fb507d53ba0774df27cea0bbba4b055993b.1661599671.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpftool: Fix a wrong type cast in btf_dumper_int

[ Upstream commit 7184aef9c0f7a81db8fd18d183ee42481d89bf35 ]

When `data` points to a boolean value, casting it to `int *` is problematic
and could lead to a wrong value being passed to `jsonw_bool`. Change the
cast to `bool *` instead.

Fixes: b12d6ec09730 ("bpf: btf: add btf print functionality")
Signed-off-by: Lam Thai <lamthai@arista.com>
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Reviewed-by: Quentin Monnet <quentin@isovalent.com>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Link: https://lore.kernel.org/bpf/20220824225859.9038-1-lamthai@arista.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mac80211: allow bw change during channel switch in mesh

[ Upstream commit 6b75f133fe05c36c52d691ff21545d5757fff721 ]

From 'IEEE Std 802.11-2020 section 11.8.8.4.1':
  The mesh channel switch may be triggered by the need to avoid
  interference to a detected radar signal, or to reassign mesh STA
  channels to ensure the MBSS connectivity.

  A 20/40 MHz MBSS may be changed to a 20 MHz MBSS and a 20 MHz
  MBSS may be changed to a 20/40 MHz MBSS.

Since the standard allows the change of bandwidth during
the channel switch in mesh, remove the bandwidth check present in
ieee80211_set_csa_beacon.

Fixes: c6da674aff94 ("{nl,cfg,mac}80211: enable the triggering of CSA frame in mesh")
Signed-off-by: Hari Chandrakanthan <quic_haric@quicinc.com>
Link: https://lore.kernel.org/r/1658903549-21218-1-git-send-email-quic_haric@quicinc.com
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf: Fix reference state management for synchronous callbacks

[ Upstream commit 9d9d00ac29d0ef7ce426964de46fa6b380357d0a ]

Currently, verifier verifies callback functions (sync and async) as if
they will be executed once, (i.e. it explores execution state as if the
function was being called once). The next insn to explore is set to
start of subprog and the exit from nested frame is handled using
curframe > 0 and prepare_func_exit. In case of async callback it uses a
customized variant of push_stack simulating a kind of branch to set up
custom state and execution context for the async callback.

While this approach is simple and works when callback really will be
executed only once, it is unsafe for all of our current helpers which
are for_each style, i.e. they execute the callback multiple times.

A callback releasing acquired references of the caller may do so
multiple times, but currently verifier sees it as one call inside the
frame, which then returns to caller. Hence, it thinks it released some
reference that the cb e.g. got access through callback_ctx (register
filled inside cb from spilled typed register on stack).

Similarly, it may see that an acquire call is unpaired inside the
callback, so the caller will copy the reference state of callback and
then will have to release the register with new ref_obj_ids. But again,
the callback may execute multiple times, but the verifier will only
account for acquired references for a single symbolic execution of the
callback, which will cause leaks.

Note that for async callback case, things are different. While currently
we have bpf_timer_set_callback which only executes it once, even for
multiple executions it would be safe, as reference state is NULL and
check_reference_leak would force program to release state before
BPF_EXIT. The state is also unaffected by analysis for the caller frame.
Hence async callback is safe.

Since we want the reference state to be accessible, e.g. for pointers
loaded from stack through callback_ctx's PTR_TO_STACK, we still have to
copy caller's reference_state to callback's bpf_func_state, but we
enforce that whatever references it adds to that reference_state has
been released before it hits BPF_EXIT. This requires introducing a new
callback_ref member in the reference state to distinguish between caller
vs callee references. Hence, check_reference_leak now errors out if it
sees we are in callback_fn and we have not released callback_ref refs.
Since there can be multiple nested callbacks, like frame 0 -> cb1 -> cb2
etc. we need to also distinguish between whether this particular ref
belongs to this callback frame or parent, and only error for our own, so
we store state->frameno (which is always non-zero for callbacks).

In short, callbacks can read parent reference_state, but cannot mutate
it, to be able to use pointers acquired by the caller. They must only
undo their changes (by releasing their own acquired_refs before
BPF_EXIT) on top of caller reference_state before returning (at which
point the caller and callback state will match anyway, so no need to
copy it back to caller).

Fixes: 69c087ba6225 ("bpf: Add bpf_for_each_map_elem() helper")
Signed-off-by: Kumar Kartikeya Dwivedi <memxor@gmail.com>
Link: https://lore.kernel.org/r/20220823013125.24938-1-memxor@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

leds: lm3601x: Don't use mutex after it was destroyed

[ Upstream commit 32f7eed0c763a9b89f6b357ec54b48398fc7b99e ]

The mutex might still be in use until the devm cleanup callback
devm_led_classdev_flash_release() is called. This only happens some time
after lm3601x_remove() completed.

Fixes: e63a744871a3 ("leds: lm3601x: Convert class registration to device managed")
Acked-by: Pavel Machek <pavel@ucw.cz>
Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Signed-off-by: Wolfram Sang <wsa@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath10k: add peer map clean up for peer delete in ath10k_sta_state()

[ Upstream commit f020d9570a04df0762a2ac5c50cf1d8c511c9164 ]

When peer delete failed in a disconnect operation, use-after-free
detected by KFENCE in below log. It is because for each vdev_id and
address, it has only one struct ath10k_peer, it is allocated in
ath10k_peer_map_event(). When connected to an AP, it has more than
one HTT_T2H_MSG_TYPE_PEER_MAP reported from firmware, then the
array peer_map of struct ath10k will be set muti-elements to the
same ath10k_peer in ath10k_peer_map_event(). When peer delete failed
in ath10k_sta_state(), the ath10k_peer will be free for the 1st peer
id in array peer_map of struct ath10k, and then use-after-free happened
for the 2nd peer id because they map to the same ath10k_peer.

And clean up all peers in array peer_map for the ath10k_peer, then
user-after-free disappeared

peer map event log:
[  306.911021] wlan0: authenticate with b0:2a:43:e6:75:0e
[  306.957187] ath10k_pci 0000:01:00.0: mac vdev 0 peer create b0:2a:43:e6:75:0e (new sta) sta 1 / 32 peer 1 / 33
[  306.957395] ath10k_pci 0000:01:00.0: htt peer map vdev 0 peer b0:2a:43:e6:75:0e id 246
[  306.957404] ath10k_pci 0000:01:00.0: htt peer map vdev 0 peer b0:2a:43:e6:75:0e id 198
[  306.986924] ath10k_pci 0000:01:00.0: htt peer map vdev 0 peer b0:2a:43:e6:75:0e id 166

peer unmap event log:
[  435.715691] wlan0: deauthenticating from b0:2a:43:e6:75:0e by local choice (Reason: 3=DEAUTH_LEAVING)
[  435.716802] ath10k_pci 0000:01:00.0: mac vdev 0 peer delete b0:2a:43:e6:75:0e sta ffff990e0e9c2b50 (sta gone)
[  435.717177] ath10k_pci 0000:01:00.0: htt peer unmap vdev 0 peer b0:2a:43:e6:75:0e id 246
[  435.717186] ath10k_pci 0000:01:00.0: htt peer unmap vdev 0 peer b0:2a:43:e6:75:0e id 198
[  435.717193] ath10k_pci 0000:01:00.0: htt peer unmap vdev 0 peer b0:2a:43:e6:75:0e id 166

use-after-free log:
[21705.888627] wlan0: deauthenticating from d0:76:8f:82:be:75 by local choice (Reason: 3=DEAUTH_LEAVING)
[21713.799910] ath10k_pci 0000:01:00.0: failed to delete peer d0:76:8f:82:be:75 for vdev 0: -110
[21713.799925] ath10k_pci 0000:01:00.0: found sta peer d0:76:8f:82:be:75 (ptr 0000000000000000 id 102) entry on vdev 0 after it was supposedly removed
[21713.799968] ==================================================================
[21713.799991] BUG: KFENCE: use-after-free read in ath10k_sta_state+0x265/0xb8a [ath10k_core]
[21713.799991]
[21713.799997] Use-after-free read at 0x00000000abe1c75e (in kfence-#69):
[21713.800010]  ath10k_sta_state+0x265/0xb8a [ath10k_core]
[21713.800041]  drv_sta_state+0x115/0x677 [mac80211]
[21713.800059]  __sta_info_destroy_part2+0xb1/0x133 [mac80211]
[21713.800076]  __sta_info_flush+0x11d/0x162 [mac80211]
[21713.800093]  ieee80211_set_disassoc+0x12d/0x2f4 [mac80211]
[21713.800110]  ieee80211_mgd_deauth+0x26c/0x29b [mac80211]
[21713.800137]  cfg80211_mlme_deauth+0x13f/0x1bb [cfg80211]
[21713.800153]  nl80211_deauthenticate+0xf8/0x121 [cfg80211]
[21713.800161]  genl_rcv_msg+0x38e/0x3be
[21713.800166]  netlink_rcv_skb+0x89/0xf7
[21713.800171]  genl_rcv+0x28/0x36
[21713.800176]  netlink_unicast+0x179/0x24b
[21713.800181]  netlink_sendmsg+0x3a0/0x40e
[21713.800187]  sock_sendmsg+0x72/0x76
[21713.800192]  ____sys_sendmsg+0x16d/0x1e3
[21713.800196]  ___sys_sendmsg+0x95/0xd1
[21713.800200]  __sys_sendmsg+0x85/0xbf
[21713.800205]  do_syscall_64+0x43/0x55
[21713.800210]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
[21713.800213]
[21713.800219] kfence-#69: 0x000000009149b0d5-0x000000004c0697fb, size=1064, cache=kmalloc-2k
[21713.800219]
[21713.800224] allocated by task 13 on cpu 0 at 21705.501373s:
[21713.800241]  ath10k_peer_map_event+0x7e/0x154 [ath10k_core]
[21713.800254]  ath10k_htt_t2h_msg_handler+0x586/0x1039 [ath10k_core]
[21713.800265]  ath10k_htt_htc_t2h_msg_handler+0x12/0x28 [ath10k_core]
[21713.800277]  ath10k_htc_rx_completion_handler+0x14c/0x1b5 [ath10k_core]
[21713.800283]  ath10k_pci_process_rx_cb+0x195/0x1df [ath10k_pci]
[21713.800294]  ath10k_ce_per_engine_service+0x55/0x74 [ath10k_core]
[21713.800305]  ath10k_ce_per_engine_service_any+0x76/0x84 [ath10k_core]
[21713.800310]  ath10k_pci_napi_poll+0x49/0x144 [ath10k_pci]
[21713.800316]  net_rx_action+0xdc/0x361
[21713.800320]  __do_softirq+0x163/0x29a
[21713.800325]  asm_call_irq_on_stack+0x12/0x20
[21713.800331]  do_softirq_own_stack+0x3c/0x48
[21713.800337]  __irq_exit_rcu+0x9b/0x9d
[21713.800342]  common_interrupt+0xc9/0x14d
[21713.800346]  asm_common_interrupt+0x1e/0x40
[21713.800351]  ksoftirqd_should_run+0x5/0x16
[21713.800357]  smpboot_thread_fn+0x148/0x211
[21713.800362]  kthread+0x150/0x15f
[21713.800367]  ret_from_fork+0x22/0x30
[21713.800370]
[21713.800374] freed by task 708 on cpu 1 at 21713.799953s:
[21713.800498]  ath10k_sta_state+0x2c6/0xb8a [ath10k_core]
[21713.800515]  drv_sta_state+0x115/0x677 [mac80211]
[21713.800532]  __sta_info_destroy_part2+0xb1/0x133 [mac80211]
[21713.800548]  __sta_info_flush+0x11d/0x162 [mac80211]
[21713.800565]  ieee80211_set_disassoc+0x12d/0x2f4 [mac80211]
[21713.800581]  ieee80211_mgd_deauth+0x26c/0x29b [mac80211]
[21713.800598]  cfg80211_mlme_deauth+0x13f/0x1bb [cfg80211]
[21713.800614]  nl80211_deauthenticate+0xf8/0x121 [cfg80211]
[21713.800619]  genl_rcv_msg+0x38e/0x3be
[21713.800623]  netlink_rcv_skb+0x89/0xf7
[21713.800628]  genl_rcv+0x28/0x36
[21713.800632]  netlink_unicast+0x179/0x24b
[21713.800637]  netlink_sendmsg+0x3a0/0x40e
[21713.800642]  sock_sendmsg+0x72/0x76
[21713.800646]  ____sys_sendmsg+0x16d/0x1e3
[21713.800651]  ___sys_sendmsg+0x95/0xd1
[21713.800655]  __sys_sendmsg+0x85/0xbf
[21713.800659]  do_syscall_64+0x43/0x55
[21713.800663]  entry_SYSCALL_64_after_hwframe+0x44/0xa9

Tested-on: QCA6174 hw3.2 PCI WLAN.RM.4.4.1-00288-QCARMSWPZ-1

Fixes: d0eeafad1189 ("ath10k: Clean up peer when sta goes away.")
Signed-off-by: Wen Gong <quic_wgong@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20220801141930.16794-1-quic_wgong@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rtlwifi: 8192de: correct checking of IQK reload

[ Upstream commit 93fbc1ebd978cf408ef5765e9c1630fce9a8621b ]

Since IQK could spend time, we make a cache of IQK result matrix that looks
like iqk_matrix[channel_idx].val[x][y], and we can reload the matrix if we
have made a cache. To determine a cache is made, we check
iqk_matrix[channel_idx].val[0][0].

The initial commit 7274a8c22980 ("rtlwifi: rtl8192de: Merge phy routines")
make a mistake that checks incorrect iqk_matrix[channel_idx].val[0] that
is always true, and this mistake is found by commit ee3db469dd31
("wifi: rtlwifi: remove always-true condition pointed out by GCC 12"), so
I recall the vendor driver to find fix and apply the correctness.

Fixes: 7274a8c22980 ("rtlwifi: rtl8192de: Merge phy routines")
Signed-off-by: Ping-Ke Shih <pkshih@realtek.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220801113345.42016-1-pkshih@realtek.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

NFSD: Fix handling of oversized NFSv4 COMPOUND requests

[ Upstream commit 7518a3dc5ea249d4112156ce71b8b184eb786151 ]

If an NFS server returns NFS4ERR_RESOURCE on the first operation in
an NFSv4 COMPOUND, there's no way for a client to know where the
problem is and then simplify the compound to make forward progress.

So instead, make NFSD process as many operations in an oversized
COMPOUND as it can and then return NFS4ERR_RESOURCE on the first
operation it did not process.

pynfs NFSv4.0 COMP6 exercises this case, but checks only for the
COMPOUND status code, not whether the server has processed any
of the operations.

pynfs NFSv4.1 SEQ6 and SEQ7 exercise the NFSv4.1 case, which detects
too many operations per COMPOUND by checking against the limits
negotiated when the session was created.

Suggested-by: Bruce Fields <bfields@fieldses.org>
Fixes: 0078117c6d91 ("nfsd: return RESOURCE not GARBAGE_ARGS on too many ops")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

NFSD: Protect against send buffer overflow in NFSv2 READDIR

[ Upstream commit 00b4492686e0497fdb924a9d4c8f6f99377e176c ]

Restore the previous limit on the @count argument to prevent a
buffer overflow attack.

Fixes: 53b1119a6e50 ("NFSD: Fix READDIR buffer overflow")
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

SUNRPC: Fix svcxdr_init_encode's buflen calculation

[ Upstream commit 1242a87da0d8cd2a428e96ca68e7ea899b0f4624 ]

Commit 2825a7f90753 ("nfsd4: allow encoding across page boundaries")
added an explicit computation of the remaining length in the rq_res
XDR buffer.

The computation appears to suffer from an "off-by-one" bug. Because
buflen is too large by one page, XDR encoding can run off the end of
the send buffer by eventually trying to use the struct page address
in rq_page_end, which always contains NULL.

Fixes: bddfdbcddbe2 ("NFSD: Extract the svcxdr_init_encode() helper")
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

SUNRPC: Fix svcxdr_init_decode's end-of-buffer calculation

[ Upstream commit 90bfc37b5ab91c1a6165e3e5cfc49bf04571b762 ]

Ensure that stream-based argument decoding can't go past the actual
end of the receive buffer. xdr_init_decode's calculation of the
value of xdr->end over-estimates the end of the buffer because the
Linux kernel RPC server code does not remove the size of the RPC
header from rqstp->rq_arg before calling the upper layer's
dispatcher.

The server-side still uses the svc_getnl() macros to decode the
RPC call header. These macros reduce the length of the head iov
but do not update the total length of the message in the buffer
(buf->len).

A proper fix for this would be to replace the use of svc_getnl() and
friends in the RPC header decoder, but that would be a large and
invasive change that would be difficult to backport.

Fixes: 5191955d6fc6 ("SUNRPC: Prepare for xdr_stream-style decoding on the server-side")
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: Fix a memory leak in an error handling path

[ Upstream commit fd1ef88049de09bc70d60b549992524cfc0e66ff ]

If this memdup_user() call fails, the memory allocated in a previous call
a few lines above should be freed. Otherwise it leaks.

Fixes: 6ee95d1c8991 ("nfsd: add support for upcall version 2")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

objtool: Preserve special st_shndx indexes in elf_update_symbol

[ Upstream commit 5141d3a06b2da1731ac82091298b766a1f95d3d8 ]

elf_update_symbol fails to preserve the special st_shndx values
between [SHN_LORESERVE, SHN_HIRESERVE], which results in it
converting SHN_ABS entries into SHN_UNDEF, for example. Explicitly
check for the special indexes and ensure these symbols are not
marked undefined.

Fixes: ead165fa1042 ("objtool: Fix symbol creation")
Signed-off-by: Sami Tolvanen <samitolvanen@google.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Signed-off-by: Kees Cook <keescook@chromium.org>
Link: https://lore.kernel.org/r/20220908215504.3686827-17-samitolvanen@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: 9247/1: mm: set readonly for MT_MEMORY_RO with ARM_LPAE

[ Upstream commit 14ca1a4690750bb54e1049e49f3140ef48958a6e ]

MT_MEMORY_RO is introduced by commit 598f0a99fa8a ("ARM: 9210/1:
Mark the FDT_FIXED sections as shareable"), which is a readonly
memory type for FDT area, but there are some different between
ARM_LPAE and non-ARM_LPAE, we need to setup PMD_SECT_AP2 and
L_PMD_SECT_RDONLY for MT_MEMORY_RO when ARM_LAPE enabled.

non-ARM_LPAE 0xff800000-0xffa00000           2M PGD KERNEL      ro NX SHD
ARM_LPAE 0xff800000-0xffc00000           4M PMD RW NX SHD
ARM_LPAE+fix 0xff800000-0xffc00000           4M PMD ro NX SHD

Fixes: 598f0a99fa8a ("ARM: 9210/1: Mark the FDT_FIXED sections as shareable")
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ARM: 9244/1: dump: Fix wrong pg_level in walk_pmd()

[ Upstream commit 2ccd19b3ffac07cc7e75a2bd1ed779728bb67197 ]

After ARM supports p4d page tables, the pg_level for note_page()
in walk_pmd() should be 4, not 3, fix it.

Fixes: 84e6ffb2c49c ("arm: add support for folded p4d page tables")
Signed-off-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Signed-off-by: Sasha Levin <sashal@kernel.org>

MIPS: SGI-IP27: Fix platform-device leak in bridge_platform_create()

[ Upstream commit 11bec9cba4de06b3c0e9e4041453c2caaa1cbec1 ]

In error case in bridge_platform_create after calling
platform_device_add()/platform_device_add_data()/
platform_device_add_resources(), release the failed
'pdev' or it will be leak, call platform_device_put()
to fix this problem.

Besides, 'pdev' is divided into 'pdev_wd' and 'pdev_bd',
use platform_device_unregister() to release sgi_w1
resources when xtalk-bridge registration fails.

Fixes: 5dc76a96e95a ("MIPS: PCI: use information from 1-wire PROM for IOC3 detection")
Signed-off-by: Lin Yujun <linyujun809@huawei.com>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

MIPS: SGI-IP27: Free some unused memory

[ Upstream commit 33d7085682b4aa212ebfadbc21da81dfefaaac16 ]

platform_device_add_data() duplicates the memory it is passed. So we can
free some memory to save a few bytes that would remain unused otherwise.

Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Stable-dep-of: 11bec9cba4de ("MIPS: SGI-IP27: Fix platform-device leak in bridge_platform_create()")
Signed-off-by: Sasha Levin <sashal@kernel.org>

sh: machvec: Use char[] for section boundaries

[ Upstream commit c5783af354688b24abd359f7086c282ec74de993 ]

As done for other sections, define the extern as a character array,
which relaxes many of the compiler-time object size checks, which would
otherwise assume it's a single long. Solves the following build error:

arch/sh/kernel/machvec.c: error: array subscript 'struct sh_machine_vector[0]' is partly outside array bounds of 'long int[1]' [-Werror=array-bounds]:  => 105:33

Cc: Yoshinori Sato <ysato@users.sourceforge.jp>
Cc: Rich Felker <dalias@libc.org>
Cc: linux-sh@vger.kernel.org
Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Link: https://lore.kernel.org/lkml/alpine.DEB.2.22.394.2209050944290.964530@ramsan.of.borg/
Fixes: 9655ad03af2d ("sh: Fixup machvec support.")
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>
Reviewed-by: Gustavo A. R. Silva <gustavoars@kernel.org>
Acked-by: Rich Felker <dalias@libc.org>
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

thermal: cpufreq_cooling: Check the policy first in cpufreq_cooling_register()

[ Upstream commit cff895277c8558221ba180aefe26799dcb4eec86 ]

Since the policy needs to be accessed first when obtaining cpu devices,
first check whether the policy is legal before this.

Fixes: 5130802ddbb1 ("thermal: cpu_cooling: Switch to QoS requests for freq limits")
Signed-off-by: Xuewen Yan <xuewen.yan@unisoc.com>
Acked-by: Viresh Kumar <viresh.kumar@linaro.org>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ntfs3: rework xattr handlers and switch to POSIX ACL VFS helpers

[ Upstream commit a26aa12384158116c0d80d50e0bdc7b3323551e2 ]

The xattr code in ntfs3 is currently a bit confused. For example, it
defines a POSIX ACL i_op->set_acl() method but instead of relying on the
generic POSIX ACL VFS helpers it defines its own set of xattr helpers
with the consequence that i_op->set_acl() is currently dead code.

Switch ntfs3 to rely on the VFS POSIX ACL xattr handlers. Also remove
i_op->{g,s}et_acl() methods from symlink inode operations. Symlinks
don't support xattrs.

This is a preliminary change for the following patches which move
handling idmapped mounts directly in posix_acl_xattr_set().

This survives POSIX ACL xfstests.

Fixes: be71b5cba2e6 ("fs/ntfs3: Add attrib operations")
Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Reviewed-by: Seth Forshee (DigitalOcean) <sforshee@kernel.org>>
Signed-off-by: Sasha Levin <sashal@kernel.org>

userfaultfd: open userfaultfds with O_RDONLY

[ Upstream commit abec3d015fdfb7c63105c7e1c956188bf381aa55 ]

Since userfaultfd doesn't implement a write operation, it is more
appropriate to open it read-only.

When userfaultfds are opened read-write like it is now, and such fd is
passed from one process to another, SELinux will check both read and
write permissions for the target process, even though it can't actually
do any write operation on the fd later.

Inspired by the following bug report, which has hit the SELinux scenario
described above:
https://bugzilla.redhat.com/show_bug.cgi?id=1974559

Reported-by: Robert O'Callahan <roc@ocallahan.org>
Fixes: 86039bd3b4e6 ("userfaultfd: add new syscall to provide memory externalization")
Signed-off-by: Ondrej Mosnacek <omosnace@redhat.com>
Acked-by: Peter Xu <peterx@redhat.com>
Acked-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ima: fix blocking of security.ima xattrs of unsupported algorithms

[ Upstream commit 5926586f291b53cb8a0c9631fc19489be1186e2d ]

Limit validating the hash algorithm to just security.ima xattr, not
the security.evm xattr or any of the protected EVM security xattrs,
nor posix acls.

Fixes: 50f742dd9147 ("IMA: block writes of the security.ima xattr with unsupported algorithms")
Reported-by: Christian Brauner <brauner@kernel.org>
Acked-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Signed-off-by: Mimi Zohar <zohar@linux.ibm.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selinux: use "grep -E" instead of "egrep"

commit c969bb8dbaf2f3628927eae73e7c579a74cf1b6e upstream.

The latest version of grep claims that egrep is now obsolete so the build
now contains warnings that look like:
egrep: warning: egrep is obsolescent; using grep -E
fix this by using "grep -E" instead.

Cc: Paul Moore <paul@paul-moore.com>
Cc: Stephen Smalley <stephen.smalley.work@gmail.com>
Cc: Eric Paris <eparis@parisplace.org>
Cc: selinux@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
[PM: tweak to remove vdso reference, cleanup subj line]
Signed-off-by: Paul Moore <paul@paul-moore.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

smb3: must initialize two ACL struct fields to zero

commit f09bd695af3b8ab46fc24e5d6954a24104c38387 upstream.

Coverity spotted that we were not initalizing Stbz1 and Stbz2 to
zero in create_sd_buf.

Addresses-Coverity: 1513848 ("Uninitialized scalar variable")
Cc: <stable@vger.kernel.org>
Reviewed-by: Paulo Alcantara (SUSE) <pc@cjr.nz>
Signed-off-by: Steve French <stfrench@microsoft.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/amd/display: Fix vblank refcount in vrr transition

commit 8799c0be89ebb99a16098bdf618f49f817bef76a upstream.

manage_dm_interrupts disable/enable vblank using drm_crtc_vblank_off/on
which causes drm_crtc_vblank_get in vrr_transition to fail, and later
when drm_crtc_vblank_put is called the refcount on vblank will be messed
up. Therefore move the call to after manage_dm_interrupts.

Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1247
Bug: https://gitlab.freedesktop.org/drm/amd/-/issues/1380

Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Reviewed-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Yunxiang Li <Yunxiang.Li@amd.com>
Signed-off-by: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Cc: stable@vger.kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/i915: Fix watermark calculations for gen12+ CCS+CC modifier

commit 070a2855900de17b1e11a0dc35af9794e80f1a28 upstream.

Take the gen12+ CCS+CC modifier into account when calculating the
watermarks. Othwerwise we'll calculate the watermarks thinking this
Y-tiled modifier is linear.

The rc_surface part is actually a nop since that is not used
for any glk+ platform.

Cc: stable@vger.kernel.org
Fixes: d1e2775e9b96 ("drm/i915/tgl: Add Clear Color support for TGL Render Decompression")
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221003111544.8007-4-ville.syrjala@linux.intel.com
(cherry picked from commit a627455bbe50a111475d7a42beb58fa64bd96c83)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/i915: Fix watermark calculations for gen12+ MC CCS modifier

commit 484b2b9281000274ef7c5cb0a9ebc5da6f5c281c upstream.

Take the gen12+ MC CCS modifier into account when calculating the
watermarks. Othwerwise we'll calculate the watermarks thinking this
Y-tiled modifier is linear.

The rc_surface part is actually a nop since that is not used
for any glk+ platform.

v2: Split RC CCS vs. MC CCS to separate patches

Cc: stable@vger.kernel.org
Fixes: 2dfbf9d2873a ("drm/i915/tgl: Gen-12 display can decompress surfaces compressed by the media engine")
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221003111544.8007-3-ville.syrjala@linux.intel.com
(cherry picked from commit 91c9651425fe955b1387f3637607dda005f3f710)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/i915: Fix watermark calculations for gen12+ RC CCS modifier

commit c56453a00f19ccddee302f5f9fe96b80e0b47fd3 upstream.

Take the gen12+ RC CCS modifier into account when calculating the
watermarks. Othwerwise we'll calculate the watermarks thinking this
Y-tiled modifier is linear.

The rc_surface part is actually a nop since that is not used
for any glk+ platform.

v2: Split RC CCS vs. MC CCS to separate patches

Cc: stable@vger.kernel.org
Fixes: b3e57bccd68a ("drm/i915/tgl: Gen-12 render decompression")
Reviewed-by: Juha-Pekka Heikkila <juhapekka.heikkila@gmail.com>
Signed-off-by: Ville Syrjälä <ville.syrjala@linux.intel.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20221003111544.8007-2-ville.syrjala@linux.intel.com
(cherry picked from commit a89a96a586114f67598c6391c75678b4dba5c2da)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/nouveau: fix a use-after-free in nouveau_gem_prime_import_sg_table()

commit 540dfd188ea2940582841c1c220bd035a7db0e51 upstream.

nouveau_bo_init() is backed by ttm_bo_init() and ferries its return code
back to the caller. On failures, ttm will call nouveau_bo_del_ttm() and
free the memory.Thus, when nouveau_bo_init() returns an error, the gem
object has already been released. Then the call to nouveau_bo_ref() will
use the freed "nvbo->bo" and lead to a use-after-free bug.

We should delete the call to nouveau_bo_ref() to avoid the use-after-free.

Signed-off-by: Jianglei Nie <niejianglei2021@163.com>
Reviewed-by: Lyude Paul <lyude@redhat.com>
Signed-off-by: Lyude Paul <lyude@redhat.com>
Fixes: 019cbd4a4feb ("drm/nouveau: Initialize GEM object before TTM object")
Cc: Thierry Reding <treding@nvidia.com>
Cc: <stable@vger.kernel.org> # v5.4+
Link: https://patchwork.freedesktop.org/patch/msgid/20220705132546.2247677-1-niejianglei2021@163.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

drm/nouveau/kms/nv140-: Disable interlacing

commit 8ba9249396bef37cb68be9e8dee7847f1737db9d upstream.

As it turns out: while Nvidia does actually have interlacing knobs on their
GPU still pretty much no current GPUs since Volta actually support it.
Trying interlacing on these GPUs will result in NVDisplay being quite
unhappy like so:

nouveau 0000:1f:00.0: disp: chid 0 stat 00004802 reason 4 [INVALID_ARG] mthd 2008 data 00000001 code 00080000
nouveau 0000:1f:00.0: disp: chid 0 stat 10005080 reason 5 [INVALID_STATE] mthd 0200 data 00000001 code 00000001

So let's fix this by following the same behavior Nvidia's driver does and
disable interlacing entirely.

Signed-off-by: Lyude Paul <lyude@redhat.com>
Cc: stable@vger.kernel.org
Reviewed-by: Karol Herbst <kherbst@redhat.com>
Link: https://patchwork.freedesktop.org/patch/msgid/20220816180436.156310-1-lyude@redhat.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

staging: greybus: audio_helper: remove unused and wrong debugfs usage

commit d517cdeb904ddc0cbebcc959d43596426cac40b0 upstream.

In the greybus audio_helper code, the debugfs file for the dapm has the
potential to be removed and memory will be leaked.  There is also the
very real potential for this code to remove ALL debugfs entries from the
system, and it seems like this is what will really happen if this code
ever runs.  This all is very wrong as the greybus audio driver did not
create this debugfs file, the sound core did and controls the lifespan
of it.

So remove all of the debugfs logic from the audio_helper code as there's
no way it could be correct.  If this really is needed, it can come back
with a fixup for the incorrect usage of the debugfs_lookup() call which
is what caused this to be noticed at all.

Cc: Johan Hovold <johan@kernel.org>
Cc: Alex Elder <elder@kernel.org>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: stable <stable@kernel.org>
Link: https://lore.kernel.org/r/20220902143715.320500-1-gregkh@linuxfoundation.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

KVM: VMX: Drop bits 31:16 when shoving exception error code into VMCS

commit eba9799b5a6efe2993cf92529608e4aa8163d73b upstream.

Deliberately truncate the exception error code when shoving it into the
VMCS (VM-Entry field for vmcs01 and vmcs02, VM-Exit field for vmcs12).
Intel CPUs are incapable of handling 32-bit error codes and will never
generate an error code with bits 31:16, but userspace can provide an
arbitrary error code via KVM_SET_VCPU_EVENTS.  Failure to drop the bits
on exception injection results in failed VM-Entry, as VMX disallows
setting bits 31:16.  Setting the bits on VM-Exit would at best confuse
L1, and at worse induce a nested VM-Entry failure, e.g. if L1 decided to
reinject the exception back into L2.

Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Link: https://lore.kernel.org/r/20220830231614.3580124-3-seanjc@google.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

KVM: nVMX: Don't propagate vmcs12's PERF_GLOBAL_CTRL settings to vmcs02

commit def9d705c05eab3fdedeb10ad67907513b12038e upstream.

Don't propagate vmcs12's VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL to vmcs02.
KVM doesn't disallow L1 from using VM_ENTRY_LOAD_IA32_PERF_GLOBAL_CTRL
even when KVM itself doesn't use the control, e.g. due to the various
CPU errata that where the MSR can be corrupted on VM-Exit.

Preserve KVM's (vmcs01) setting to hopefully avoid having to toggle the
bit in vmcs02 at a later point.  E.g. if KVM is loading PERF_GLOBAL_CTRL
when running L1, then odds are good KVM will also load the MSR when
running L2.

Fixes: 8bf00a529967 ("KVM: VMX: add support for switching of PERF_GLOBAL_CTRL")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Link: https://lore.kernel.org/r/20220830133737.1539624-18-vkuznets@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

KVM: nVMX: Unconditionally purge queued/injected events on nested "exit"

commit d953540430c5af57f5de97ea9e36253908204027 upstream.

Drop pending exceptions and events queued for re-injection when leaving
nested guest mode, even if the "exit" is due to VM-Fail, SMI, or forced
by host userspace.  Failure to purge events could result in an event
belonging to L2 being injected into L1.

This _should_ never happen for VM-Fail as all events should be blocked by
nested_run_pending, but it's possible if KVM, not the L1 hypervisor, is
the source of VM-Fail when running vmcs02.

SMI is a nop (barring unknown bugs) as recognition of SMI and thus entry
to SMM is blocked by pending exceptions and re-injected events.

Forced exit is definitely buggy, but has likely gone unnoticed because
userspace probably follows the forced exit with KVM_SET_VCPU_EVENTS (or
some other ioctl() that purges the queue).

Fixes: 4f350c6dbcb9 ("kvm: nVMX: Handle deferred early VMLAUNCH/VMRESUME failure properly")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Link: https://lore.kernel.org/r/20220830231614.3580124-2-seanjc@google.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

KVM: x86/emulator: Fix handing of POP SS to correctly set interruptibility

commit 6aa5c47c351b22c21205c87977c84809cd015fcf upstream.

The emulator checks the wrong variable while setting the CPU
interruptibility state, the target segment is embedded in the instruction
opcode, not the ModR/M register.  Fix the condition.

Signed-off-by: Michal Luczaj <mhal@rbox.co>
Fixes: a5457e7bcf9a ("KVM: emulate: POP SS triggers a MOV SS shadow too")
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/all/20220821215900.1419215-1-mhal@rbox.co
Signed-off-by: Sean Christopherson <seanjc@google.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

blk-wbt: call rq_qos_add() after wb_normal is initialized

commit 8c5035dfbb9475b67c82b3fdb7351236525bf52b upstream.

Our test found a problem that wbt inflight counter is negative, which
will cause io hang(noted that this problem doesn't exist in mainline):

t1: device create t2: issue io
add_disk
 blk_register_queue
  wbt_enable_default
   wbt_init
    rq_qos_add
    // wb_normal is still 0
/*
 * in mainline, disk can't be opened before
 * bdev_add(), however, in old kernels, disk
 * can be opened before blk_register_queue().
 */
blkdev_issue_flush
                        // disk size is 0, however, it's not checked
                         submit_bio_wait
                          submit_bio
                           blk_mq_submit_bio
                            rq_qos_throttle
                             wbt_wait
      bio_to_wbt_flags
                               rwb_enabled
       // wb_normal is 0, inflight is not increased

    wbt_queue_depth_changed(&rwb->rqos);
     wbt_update_limits
     // wb_normal is initialized
                            rq_qos_track
                             wbt_track
                              rq->wbt_flags |= bio_to_wbt_flags(rwb, bio);
      // wb_normal is not 0，wbt_flags will be set
t3: io completion
blk_mq_free_request
 rq_qos_done
  wbt_done
   wbt_is_tracked
   // return true
   __wbt_done
    wbt_rqw_done
     atomic_dec_return(&rqw->inflight);
     // inflight is decreased

commit 8235b5c1e8c1 ("block: call bdev_add later in device_add_disk") can
avoid this problem, however it's better to fix this problem in wbt:

1) Lower kernel can't backport this patch due to lots of refactor.
2) Root cause is that wbt call rq_qos_add() before wb_normal is
initialized.

Fixes: e34cbd307477 ("blk-wbt: add general throttling mechanism")
Cc: <stable@vger.kernel.org>
Signed-off-by: Yu Kuai <yukuai3@huawei.com>
Link: https://lore.kernel.org/r/20220913105749.3086243-1-yukuai1@huaweicloud.com
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

media: cedrus: Fix endless loop in cedrus_h265_skip_bits()

commit 91db7a3fc7fe670cf1770a398a43bb4a1f776bf1 upstream.

The busy status bit may never de-assert if number of programmed skip
bits is incorrect, resulting in a kernel hang because the bit is polled
endlessly in the code. Fix it by adding timeout for the bit-polling.
This problem is reproducible by setting the data_bit_offset field of
the HEVC slice params to a wrong value by userspace.

Cc: stable@vger.kernel.org
Fixes: 7678c5462680 (media: cedrus: Fix decoding for some HEVC videos)
Reported-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

media: cedrus: Set the platform driver data earlier

commit 708938f8495147fe2e77a9a3e1015d8e6899323e upstream.

The cedrus_hw_resume() crashes with NULL deference on driver probe if
runtime PM is disabled because it uses platform data that hasn't been
set up yet. Fix this by setting the platform data earlier during probe.

Cc: stable@vger.kernel.org
Fixes: 50e761516f2b (media: platform: Add Cedrus VPU decoder driver)
Signed-off-by: Dmitry Osipenko <dmitry.osipenko@collabora.com>
Signed-off-by: Nicolas Dufresne <nicolas.dufresne@collabora.com>
Reviewed-by: Samuel Holland <samuel@sholland.org>
Acked-by: Paul Kocialkowski <paul.kocialkowski@bootlin.com>
Signed-off-by: Hans Verkuil <hverkuil-cisco@xs4all.nl>
Signed-off-by: Mauro Carvalho Chehab <mchehab@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

efi: libstub: drop pointless get_memory_map() call

commit d80ca810f096ff66f451e7a3ed2f0cd9ef1ff519 upstream.

Currently, the non-x86 stub code calls get_memory_map() redundantly,
given that the data it returns is never used anywhere. So drop the call.

Cc: <stable@vger.kernel.org> # v4.14+
Fixes: 24d7c494ce46 ("efi/arm-stub: Round up FDT allocation to mapping size")
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

thunderbolt: Explicitly enable lane adapter hotplug events at startup

commit 5d2569cb4a65c373896ec0217febdf88739ed295 upstream.

Software that has run before the USB4 CM in Linux runs may have disabled
hotplug events for a given lane adapter.

Other CMs such as that one distributed with Windows 11 will enable hotplug
events. Do the same thing in the Linux CM which fixes hotplug events on
"AMD Pink Sardine".

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

tracing: Fix reading strings from synthetic events

commit 0934ae9977c27133449b6dd8c6213970e7eece38 upstream.

The follow commands caused a crash:

  # cd /sys/kernel/tracing
  # echo 's:open char file[]' > dynamic_events
  # echo 'hist:keys=common_pid:file=filename:onchange($file).trace(open,$file)' > events/syscalls/sys_enter_openat/trigger'
  # echo 1 > events/synthetic/open/enable

BOOM!

The problem is that the synthetic event field "char file[]" will read
the value given to it as a string without any memory checks to make sure
the address is valid. The above example will pass in the user space
address and the sythetic event code will happily call strlen() on it
and then strscpy() where either one will cause an oops when accessing
user space addresses.

Use the helper functions from trace_kprobe and trace_eprobe that can
read strings safely (and actually succeed when the address is from user
space and the memory is mapped in).

Now the above can show:

     packagekitd-1721    [000] ...2.   104.597170: open: file=/usr/lib/rpm/fileattrs/cmake.attr
    in:imjournal-978     [006] ...2.   104.599642: open: file=/var/lib/rsyslog/imjournal.state.tmp
     packagekitd-1721    [000] ...2.   104.626308: open: file=/usr/lib/rpm/fileattrs/debuginfo.attr

Link: https://lkml.kernel.org/r/20221012104534.826549315@goodmis.org
Cc: stable@vger.kernel.org
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Tom Zanussi <zanussi@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Tom Zanussi <zanussi@kernel.org>
Fixes: bd82631d7ccdc ("tracing: Add support for dynamic strings to synthetic events")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Add "(fault)" name injection to kernel probes

commit 2e9906f84fc7c99388bb7123ade167250d50f1c0 upstream.

Have the specific functions for kernel probes that read strings to inject
the "(fault)" name directly. trace_probes.c does this too (for uprobes)
but as the code to read strings are going to be used by synthetic events
(and perhaps other utilities), it simplifies the code by making sure those
other uses do not need to implement the "(fault)" name injection as well.

Link: https://lkml.kernel.org/r/20221012104534.644803645@goodmis.org
Cc: stable@vger.kernel.org
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Tom Zanussi <zanussi@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Tom Zanussi <zanussi@kernel.org>
Fixes: bd82631d7ccdc ("tracing: Add support for dynamic strings to synthetic events")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Move duplicate code of trace_kprobe/eprobe.c into header

commit f1d3cbfaafc10464550c6d3a125f4fc802bbaed5 upstream.

The functions:

  fetch_store_strlen_user()
  fetch_store_strlen()
  fetch_store_string_user()
  fetch_store_string()

are identical in both trace_kprobe.c and trace_eprobe.c. Move them into
a new header file trace_probe_kernel.h to share it. This code will later
be used by the synthetic events as well.

Marked for stable as a fix for a crash in synthetic events requires it.

Link: https://lkml.kernel.org/r/20221012104534.467668078@goodmis.org
Cc: stable@vger.kernel.org
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Tom Zanussi <zanussi@kernel.org>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Reviewed-by: Tom Zanussi <zanussi@kernel.org>
Fixes: bd82631d7ccdc ("tracing: Add support for dynamic strings to synthetic events")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Add ioctl() to force ring buffer waiters to wake up

commit 01b2a52171735c6eea80ee2f355f32bea6c41418 upstream.

If a process is waiting on the ring buffer for data, there currently isn't
a clean way to force it to wake up. Add an ioctl call that will force any
tasks that are waiting on the trace_pipe_raw file to wake up.

Link: https://lkml.kernel.org/r/20220929095029.117f913f@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: e30f53aad2202 ("tracing: Do not busy wait in buffer splice")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Wake up waiters when tracing is disabled

commit 2b0fd9a59b7990c161fa1cb7b79edb22847c87c2 upstream.

When tracing is disabled, there's no reason that waiters should stay
waiting, wake them up, otherwise tasks get stuck when they should be
flushing the buffers.

Cc: stable@vger.kernel.org
Fixes: e30f53aad2202 ("tracing: Do not busy wait in buffer splice")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Wake up ring buffer waiters on closing of the file

commit f3ddb74ad0790030c9592229fb14d8c451f4e9a8 upstream.

When the file that represents the ring buffer is closed, there may be
waiters waiting on more input from the ring buffer. Call
ring_buffer_wake_waiters() to wake up any waiters when the file is
closed.

Link: https://lkml.kernel.org/r/20220927231825.182416969@goodmis.org
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: e30f53aad2202 ("tracing: Do not busy wait in buffer splice")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Disable interrupt or preemption before acquiring arch_spinlock_t

commit c0a581d7126c0bbc96163276f585fd7b4e4d8d0e upstream.

It was found that some tracing functions in kernel/trace/trace.c acquire
an arch_spinlock_t with preemption and irqs enabled. An example is the
tracing_saved_cmdlines_size_read() function which intermittently causes
a "BUG: using smp_processor_id() in preemptible" warning when the LTP
read_all_proc test is run.

That can be problematic in case preemption happens after acquiring the
lock. Add the necessary preemption or interrupt disabling code in the
appropriate places before acquiring an arch_spinlock_t.

The convention here is to disable preemption for trace_cmdline_lock and
interupt for max_lock.

Link: https://lkml.kernel.org/r/20220922145622.1744826-1-longman@redhat.com
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Will Deacon <will@kernel.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: stable@vger.kernel.org
Fixes: a35873a0993b ("tracing: Add conditional snapshot")
Fixes: 939c7a4f04fc ("tracing: Introduce saved_cmdlines_size file")
Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ring-buffer: Fix race between reset page and reading page

commit a0fcaaed0c46cf9399d3a2d6e0c87ddb3df0e044 upstream.

The ring buffer is broken up into sub buffers (currently of page size).
Each sub buffer has a pointer to its "tail" (the last event written to the
sub buffer). When a new event is requested, the tail is locally
incremented to cover the size of the new event. This is done in a way that
there is no need for locking.

If the tail goes past the end of the sub buffer, the process of moving to
the next sub buffer takes place. After setting the current sub buffer to
the next one, the previous one that had the tail go passed the end of the
sub buffer needs to be reset back to the original tail location (before
the new event was requested) and the rest of the sub buffer needs to be
"padded".

The race happens when a reader takes control of the sub buffer. As readers
do a "swap" of sub buffers from the ring buffer to get exclusive access to
the sub buffer, it replaces the "head" sub buffer with an empty sub buffer
that goes back into the writable portion of the ring buffer. This swap can
happen as soon as the writer moves to the next sub buffer and before it
updates the last sub buffer with padding.

Because the sub buffer can be released to the reader while the writer is
still updating the padding, it is possible for the reader to see the event
that goes past the end of the sub buffer. This can cause obvious issues.

To fix this, add a few memory barriers so that the reader definitely sees
the updates to the sub buffer, and also waits until the writer has put
back the "tail" of the sub buffer back to the last event that was written
on it.

To be paranoid, it will only spin for 1 second, otherwise it will
warn and shutdown the ring buffer code. 1 second should be enough as
the writer does have preemption disabled. If the writer doesn't move
within 1 second (with preemption disabled) something is horribly
wrong. No interrupt should last 1 second!

Link: https://lore.kernel.org/all/20220830120854.7545-1-jiazi.li@transsion.com/
Link: https://bugzilla.kernel.org/show_bug.cgi?id=216369
Link: https://lkml.kernel.org/r/20220929104909.0650a36c@gandalf.local.home
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: stable@vger.kernel.org
Fixes: c7b0930857e22 ("ring-buffer: prevent adding write in discarded area")
Reported-by: Jiazi.Li <jiazi.li@transsion.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ring-buffer: Add ring_buffer_wake_waiters()

commit 7e9fbbb1b776d8d7969551565bc246f74ec53b27 upstream.

On closing of a file that represents a ring buffer or flushing the file,
there may be waiters on the ring buffer that needs to be woken up and exit
the ring_buffer_wait() function.

Add ring_buffer_wake_waiters() to wake up the waiters on the ring buffer
and allow them to exit the wait loop.

Link: https://lkml.kernel.org/r/20220928133938.28dc2c27@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: 15693458c4bc0 ("tracing/ring-buffer: Move poll wake ups into ring buffer code")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ring-buffer: Check pending waiters when doing wake ups as well

commit ec0bbc5ec5664dcee344f79373852117dc672c86 upstream.

The wake up waiters only checks the "wakeup_full" variable and not the
"full_waiters_pending". The full_waiters_pending is set when a waiter is
added to the wait queue. The wakeup_full is only set when an event is
triggered, and it clears the full_waiters_pending to avoid multiple calls
to irq_work_queue().

The irq_work callback really needs to check both wakeup_full as well as
full_waiters_pending such that this code can be used to wake up waiters
when a file is closed that represents the ring buffer and the waiters need
to be woken up.

Link: https://lkml.kernel.org/r/20220927231824.209460321@goodmis.org
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: 15693458c4bc0 ("tracing/ring-buffer: Move poll wake ups into ring buffer code")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ring-buffer: Have the shortest_full queue be the shortest not longest

commit 3b19d614b61b93a131f463817e08219c9ce1fee3 upstream.

The logic to know when the shortest waiters on the ring buffer should be
woken up or not has uses a less than instead of a greater than compare,
which causes the shortest_full to actually be the longest.

Link: https://lkml.kernel.org/r/20220927231823.718039222@goodmis.org
Cc: stable@vger.kernel.org
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: 2c2b0a78b3739 ("ring-buffer: Add percentage of ring buffer full to wake up reader")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ring-buffer: Allow splice to read previous partially read pages

commit fa8f4a89736b654125fb254b0db753ac68a5fced upstream.

If a page is partially read, and then the splice system call is run
against the ring buffer, it will always fail to read, no matter how much
is in the ring buffer. That's because the code path for a partial read of
the page does will fail if the "full" flag is set.

The splice system call wants full pages, so if the read of the ring buffer
is not yet full, it should return zero, and the splice will block. But if
a previous read was done, where the beginning has been consumed, it should
still be given to the splice caller if the rest of the page has been
written to.

This caused the splice command to never consume data in this scenario, and
let the ring buffer just fill up and lose events.

Link: https://lkml.kernel.org/r/20220927144317.46be6b80@gandalf.local.home
Cc: stable@vger.kernel.org
Fixes: 8789a9e7df6bf ("ring-buffer: read page interface")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ftrace: Properly unset FTRACE_HASH_FL_MOD

commit 0ce0638edf5ec83343302b884fa208179580700a upstream.

When executing following commands like what document said, but the log
"#### all functions enabled ####" was not shown as expect:
  1. Set a 'mod' filter:
    $ echo 'write*:mod:ext3' > /sys/kernel/tracing/set_ftrace_filter
  2. Invert above filter:
    $ echo '!write*:mod:ext3' >> /sys/kernel/tracing/set_ftrace_filter
  3. Read the file:
    $ cat /sys/kernel/tracing/set_ftrace_filter

By some debugging, I found that flag FTRACE_HASH_FL_MOD was not unset
after inversion like above step 2 and then result of ftrace_hash_empty()
is incorrect.

Link: https://lkml.kernel.org/r/20220926152008.2239274-1-zhengyejian1@huawei.com
Cc: <mingo@redhat.com>
Cc: stable@vger.kernel.org
Fixes: 8c08f0d5c6fb ("ftrace: Have cached module filters be an active filter")
Signed-off-by: Zheng Yejian <zhengyejian1@huawei.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

livepatch: fix race between fork and KLP transition

commit 747f7a2901174c9afa805dddfb7b24db6f65e985 upstream.

The KLP transition code depends on the TIF_PATCH_PENDING and
the task->patch_state to stay in sync. On a normal (forward)
transition, TIF_PATCH_PENDING will be set on every task in
the system, while on a reverse transition (after a failed
forward one) first TIF_PATCH_PENDING will be cleared from
every task, followed by it being set on tasks that need to
be transitioned back to the original code.

However, the fork code copies over the TIF_PATCH_PENDING flag
from the parent to the child early on, in dup_task_struct and
setup_thread_stack. Much later, klp_copy_process will set
child->patch_state to match that of the parent.

However, the parent's patch_state may have been changed by KLP loading
or unloading since it was initially copied over into the child.

This results in the KLP code occasionally hitting this warning in
klp_complete_transition:

        for_each_process_thread(g, task) {
                WARN_ON_ONCE(test_tsk_thread_flag(task, TIF_PATCH_PENDING));
                task->patch_state = KLP_UNDEFINED;
        }

Set, or clear, the TIF_PATCH_PENDING flag in the child task
depending on whether or not it is needed at the time
klp_copy_process is called, at a point in copy_process where the
tasklist_lock is held exclusively, preventing races with the KLP
code.

The KLP code does have a few places where the state is changed
without the tasklist_lock held, but those should not cause
problems because klp_update_patch_state(current) cannot be
called while the current task is in the middle of fork,
klp_check_and_switch_task() which is called under the pi_lock,
which prevents rescheduling, and manipulation of the patch
state of idle tasks, which do not fork.

This should prevent this warning from triggering again in the
future, and close the race for both normal and reverse transitions.

Signed-off-by: Rik van Riel <riel@surriel.com>
Reported-by: Breno Leitao <leitao@debian.org>
Reviewed-by: Petr Mladek <pmladek@suse.com>
Acked-by: Josh Poimboeuf <jpoimboe@kernel.org>
Fixes: d83a7cb375ee ("livepatch: change to a per-task consistency model")
Cc: stable@kernel.org
Signed-off-by: Petr Mladek <pmladek@suse.com>
Link: https://lore.kernel.org/r/20220808150019.03d6a67b@imladris.surriel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: update 'state->fc_regions_size' after successful memory allocation

commit 27cd49780381c6ccbf248798e5e8fd076200ffba upstream.

To avoid to 'state->fc_regions_size' mismatch with 'state->fc_regions'
when fail to reallocate 'fc_reqions',only update 'state->fc_regions_size'
after 'state->fc_regions' is allocated successfully.

Cc: stable@kernel.org
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220921064040.3693255-4-yebin10@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: fix potential memory leak in ext4_fc_record_regions()

commit 7069d105c1f15c442b68af43f7fde784f3126739 upstream.

As krealloc may return NULL, in this case 'state->fc_regions' may not be
freed by krealloc, but 'state->fc_regions' already set NULL. Then will
lead to 'state->fc_regions' memory leak.

Cc: stable@kernel.org
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220921064040.3693255-3-yebin10@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: fix potential memory leak in ext4_fc_record_modified_inode()

commit 9305721a309fa1bd7c194e0d4a2335bf3b29dca4 upstream.

As krealloc may return NULL, in this case 'state->fc_modified_inodes'
may not be freed by krealloc, but 'state->fc_modified_inodes' already
set NULL. Then will lead to 'state->fc_modified_inodes' memory leak.

Cc: stable@kernel.org
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220921064040.3693255-2-yebin10@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: fix miss release buffer head in ext4_fc_write_inode

commit ccbf8eeb39f2ff00b54726a2b20b35d788c4ecb5 upstream.

In 'ext4_fc_write_inode' function first call 'ext4_get_inode_loc' get 'iloc',
after use it miss release 'iloc.bh'.
So just release 'iloc.bh' before 'ext4_fc_write_inode' return.

Cc: stable@kernel.org
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220914100859.1415196-1-yebin10@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: fix dir corruption when ext4_dx_add_entry() fails

commit 7177dd009c7c04290891e9a534cd47d1b620bd04 upstream.

Following process may lead to fs corruption:
1. ext4_create(dir/foo)
 ext4_add_nondir
  ext4_add_entry
   ext4_dx_add_entry
     a. add_dirent_to_buf
      ext4_mark_inode_dirty
      ext4_handle_dirty_metadata   // dir inode bh is recorded into journal
     b. ext4_append    // dx_get_count(entries) == dx_get_limit(entries)
       ext4_bread(EXT4_GET_BLOCKS_CREATE)
        ext4_getblk
         ext4_map_blocks
          ext4_ext_map_blocks
            ext4_mb_new_blocks
             dquot_alloc_block
              dquot_alloc_space_nodirty
               inode_add_bytes    // update dir's i_blocks
            ext4_ext_insert_extent
     ext4_ext_dirty  // record extent bh into journal
              ext4_handle_dirty_metadata(bh)
      // record new block into journal
       inode->i_size += inode->i_sb->s_blocksize   // new size(in mem)
     c. ext4_handle_dirty_dx_node(bh2)
// record dir's new block(dx_node) into journal
     d. ext4_handle_dirty_dx_node((frame - 1)->bh)
     e. ext4_handle_dirty_dx_node(frame->bh)
     f. do_split    // ret err!
     g. add_dirent_to_buf
 ext4_mark_inode_dirty(dir)  // update raw_inode on disk(skipped)
2. fsck -a /dev/sdb
 drop last block(dx_node) which beyonds dir's i_size.
  /dev/sdb: recovering journal
  /dev/sdb contains a file system with errors, check forced.
  /dev/sdb: Inode 12, end of extent exceeds allowed value
(logical block 128, physical block 3938, len 1)
3. fsck -fn /dev/sdb
 dx_node->entry[i].blk > dir->i_size
  Pass 2: Checking directory structure
  Problem in HTREE directory inode 12 (/dir): bad block number 128.
  Clear HTree index? no
  Problem in HTREE directory inode 12: block #3 has invalid depth (2)
  Problem in HTREE directory inode 12: block #3 has bad max hash
  Problem in HTREE directory inode 12: block #3 not referenced

Fix it by marking inode dirty directly inside ext4_append().
Fetch a reproducer in [Link].

Link: https://bugzilla.kernel.org/show_bug.cgi?id=216466
Cc: stable@vger.kernel.org
Signed-off-by: Zhihao Cheng <chengzhihao1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220911045204.516460-1-chengzhihao1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: place buffer head allocation before handle start

commit d1052d236eddf6aa851434db1897b942e8db9921 upstream.

In our product environment, we encounter some jbd hung waiting handles to
stop while several writters were doing memory reclaim for buffer head
allocation in delay alloc write path. Ext4 do buffer head allocation with
holding transaction handle which may be blocked too long if the reclaim
works not so smooth. According to our bcc trace, the reclaim time in
buffer head allocation can reach 258s and the jbd transaction commit also
take almost the same time meanwhile. Except for these extreme cases,
we often see several seconds delays for cgroup memory reclaim on our
servers. This is more likely to happen considering docker environment.

One thing to note, the allocation of buffer heads is as often as page
allocation or more often when blocksize less than page size. Just like
page cache allocation, we should also place the buffer head allocation
before startting the handle.

Cc: stable@kernel.org
Signed-off-by: Jinke Han <hanjinke.666@bytedance.com>
Link: https://lore.kernel.org/r/20220903012429.22555-1-hanjinke.666@bytedance.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: ext4_read_bh_lock() should submit IO if the buffer isn't uptodate

commit 0b73284c564d3ae4feef4bc920292f004acf4980 upstream.

Recently we notice that ext4 filesystem would occasionally fail to read
metadata from disk and report error message, but the disk and block
layer looks fine. After analyse, we lockon commit 88dbcbb3a484
("blkdev: avoid migration stalls for blkdev pages"). It provide a
migration method for the bdev, we could move page that has buffers
without extra users now, but it lock the buffers on the page, which
breaks the fragile metadata read operation on ext4 filesystem,
ext4_read_bh_lock() was copied from ll_rw_block(), it depends on the
assumption of that locked buffer means it is under IO. So it just
trylock the buffer and skip submit IO if it lock failed, after
wait_on_buffer() we conclude IO error because the buffer is not
uptodate.

This issue could be easily reproduced by add some delay just after
buffer_migrate_lock_buffers() in __buffer_migrate_folio() and do
fsstress on ext4 filesystem.

  EXT4-fs error (device pmem1): __ext4_find_entry:1658: inode #73193:
  comm fsstress: reading directory lblock 0
  EXT4-fs error (device pmem1): __ext4_find_entry:1658: inode #75334:
  comm fsstress: reading directory lblock 0

Fix it by removing the trylock logic in ext4_read_bh_lock(), just lock
the buffer and submit IO if it's not uptodate, and also leave over
readahead helper.

Cc: stable@kernel.org
Signed-off-by: Zhang Yi <yi.zhang@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220831074629.3755110-1-yi.zhang@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: don't increase iversion counter for ea_inodes

commit 50f094a5580e6297bf10a807d16f0ee23fa576cf upstream.

ea_inodes are using i_version for storing part of the reference count so
we really need to leave it alone.

The problem can be reproduced by xfstest ext4/026 when iversion is
enabled. Fix it by not calling inode_inc_iversion() for EXT4_EA_INODE_FL
inodes in ext4_mark_iloc_dirty().

Cc: stable@kernel.org
Signed-off-by: Lukas Czerner <lczerner@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Reviewed-by: Christian Brauner (Microsoft) <brauner@kernel.org>
Link: https://lore.kernel.org/r/20220824160349.39664-1-lczerner@redhat.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: fix check for block being out of directory size

commit 61a1d87a324ad5e3ed27c6699dfc93218fcf3201 upstream.

The check in __ext4_read_dirblock() for block being outside of directory
size was wrong because it compared block number against directory size
in bytes. Fix it.

Fixes: 65f8ea4cd57d ("ext4: check if directory block is within i_size")
CVE: CVE-2022-1184
CC: stable@vger.kernel.org
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Lukas Czerner <lczerner@redhat.com>
Link: https://lore.kernel.org/r/20220822114832.1482-1-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: make ext4_lazyinit_thread freezable

commit 3b575495ab8dbb4dbe85b4ac7f991693c3668ff5 upstream.

ext4_lazyinit_thread is not set freezable. Hence when the thread calls
try_to_freeze it doesn't freeze during suspend and continues to send
requests to the storage during suspend, resulting in suspend failures.

Cc: stable@kernel.org
Signed-off-by: Lalith Rajendran <lalithkraj@google.com>
Link: https://lore.kernel.org/r/20220818214049.1519544-1-lalithkraj@google.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: fix null-ptr-deref in ext4_write_info

commit f9c1f248607d5546075d3f731e7607d5571f2b60 upstream.

I caught a null-ptr-deref bug as follows:
==================================================================
KASAN: null-ptr-deref in range [0x0000000000000068-0x000000000000006f]
CPU: 1 PID: 1589 Comm: umount Not tainted 5.10.0-02219-dirty #339
RIP: 0010:ext4_write_info+0x53/0x1b0
[...]
Call Trace:
 dquot_writeback_dquots+0x341/0x9a0
 ext4_sync_fs+0x19e/0x800
 __sync_filesystem+0x83/0x100
 sync_filesystem+0x89/0xf0
 generic_shutdown_super+0x79/0x3e0
 kill_block_super+0xa1/0x110
 deactivate_locked_super+0xac/0x130
 deactivate_super+0xb6/0xd0
 cleanup_mnt+0x289/0x400
 __cleanup_mnt+0x16/0x20
 task_work_run+0x11c/0x1c0
 exit_to_user_mode_prepare+0x203/0x210
 syscall_exit_to_user_mode+0x5b/0x3a0
 do_syscall_64+0x59/0x70
 entry_SYSCALL_64_after_hwframe+0x44/0xa9
 ==================================================================

Above issue may happen as follows:
-------------------------------------
exit_to_user_mode_prepare
 task_work_run
  __cleanup_mnt
   cleanup_mnt
    deactivate_super
     deactivate_locked_super
      kill_block_super
       generic_shutdown_super
        shrink_dcache_for_umount
         dentry = sb->s_root
         sb->s_root = NULL              <--- Here set NULL
        sync_filesystem
         __sync_filesystem
          sb->s_op->sync_fs > ext4_sync_fs
           dquot_writeback_dquots
            sb->dq_op->write_info > ext4_write_info
             ext4_journal_start(d_inode(sb->s_root), EXT4_HT_QUOTA, 2)
              d_inode(sb->s_root)
               s_root->d_inode          <--- Null pointer dereference

To solve this problem, we use ext4_journal_start_sb directly
to avoid s_root being used.

Cc: stable@kernel.org
Signed-off-by: Baokun Li <libaokun1@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220805123947.565152-1-libaokun1@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ext4: avoid crash when inline data creation follows DIO write

commit 4bb26f2885ac6930984ee451b952c5a6042f2c0e upstream.

When inode is created and written to using direct IO, there is nothing
to clear the EXT4_STATE_MAY_INLINE_DATA flag. Thus when inode gets
truncated later to say 1 byte and written using normal write, we will
try to store the data as inline data. This confuses the code later
because the inode now has both normal block and inline data allocated
and the confusion manifests for example as:

kernel BUG at fs/ext4/inode.c:2721!
invalid opcode: 0000 [#1] PREEMPT SMP KASAN
CPU: 0 PID: 359 Comm: repro Not tainted 5.19.0-rc8-00001-g31ba1e3b8305-dirty #15
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 1.16.0-1.fc36 04/01/2014
RIP: 0010:ext4_writepages+0x363d/0x3660
RSP: 0018:ffffc90000ccf260 EFLAGS: 00010293
RAX: ffffffff81e1abcd RBX: 0000008000000000 RCX: ffff88810842a180
RDX: 0000000000000000 RSI: 0000008000000000 RDI: 0000000000000000
RBP: ffffc90000ccf650 R08: ffffffff81e17d58 R09: ffffed10222c680b
R10: dfffe910222c680c R11: 1ffff110222c680a R12: ffff888111634128
R13: ffffc90000ccf880 R14: 0000008410000000 R15: 0000000000000001
FS:  00007f72635d2640(0000) GS:ffff88811b000000(0000) knlGS:0000000000000000
CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
CR2: 0000565243379180 CR3: 000000010aa74000 CR4: 0000000000150eb0
Call Trace:
 <TASK>
 do_writepages+0x397/0x640
 filemap_fdatawrite_wbc+0x151/0x1b0
 file_write_and_wait_range+0x1c9/0x2b0
 ext4_sync_file+0x19e/0xa00
 vfs_fsync_range+0x17b/0x190
 ext4_buffered_write_iter+0x488/0x530
 ext4_file_write_iter+0x449/0x1b90
 vfs_write+0xbcd/0xf40
 ksys_write+0x198/0x2c0
 __x64_sys_write+0x7b/0x90
 do_syscall_64+0x3d/0x90
 entry_SYSCALL_64_after_hwframe+0x63/0xcd
 </TASK>

Fix the problem by clearing EXT4_STATE_MAY_INLINE_DATA when we are doing
direct IO write to a file.

Cc: stable@kernel.org
Reported-by: Tadeusz Struk <tadeusz.struk@linaro.org>
Reported-by: syzbot+bd13648a53ed6933ca49@syzkaller.appspotmail.com
Link: https://syzkaller.appspot.com/bug?id=a1e89d09bbbcbd5c4cb45db230ee28c822953984
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Lukas Czerner <lczerner@redhat.com>
Tested-by: Tadeusz Struk<tadeusz.struk@linaro.org>
Link: https://lore.kernel.org/r/20220727155753.13969-1-jack@suse.cz
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

jbd2: add miss release buffer head in fc_do_one_pass()

commit dfff66f30f66b9524b661f311bbed8ff3d2ca49f upstream.

In fc_do_one_pass() miss release buffer head after use which will lead
to reference count leak.

Cc: stable@kernel.org
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220917093805.1782845-1-yebin10@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

jbd2: fix potential use-after-free in jbd2_fc_wait_bufs

commit 243d1a5d505d0b0460c9af0ad56ed4a56ef0bebd upstream.

In 'jbd2_fc_wait_bufs' use 'bh' after put buffer head reference count
which may lead to use-after-free.
So judge buffer if uptodate before put buffer head reference count.

Cc: stable@kernel.org
Signed-off-by: Ye Bin <yebin10@huawei.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Link: https://lore.kernel.org/r/20220914100812.1414768-3-yebin10@huawei.com
Signed-off-by: Theodore Ts'o <tytso@mit.edu>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>