net: If sock is dead don't access sock's sk_wq in sk_stream_wait_memory

[ Upstream commit 3f8ef65af927db247418d4e1db49164d7a158fc5 ]

Fixes the below NULL pointer dereference:

  [...]
  [   14.471200] Call Trace:
  [   14.471562]  <TASK>
  [   14.471882]  lock_acquire+0x245/0x2e0
  [   14.472416]  ? remove_wait_queue+0x12/0x50
  [   14.473014]  ? _raw_spin_lock_irqsave+0x17/0x50
  [   14.473681]  _raw_spin_lock_irqsave+0x3d/0x50
  [   14.474318]  ? remove_wait_queue+0x12/0x50
  [   14.474907]  remove_wait_queue+0x12/0x50
  [   14.475480]  sk_stream_wait_memory+0x20d/0x340
  [   14.476127]  ? do_wait_intr_irq+0x80/0x80
  [   14.476704]  do_tcp_sendpages+0x287/0x600
  [   14.477283]  tcp_bpf_push+0xab/0x260
  [   14.477817]  tcp_bpf_sendmsg_redir+0x297/0x500
  [   14.478461]  ? __local_bh_enable_ip+0x77/0xe0
  [   14.479096]  tcp_bpf_send_verdict+0x105/0x470
  [   14.479729]  tcp_bpf_sendmsg+0x318/0x4f0
  [   14.480311]  sock_sendmsg+0x2d/0x40
  [   14.480822]  ____sys_sendmsg+0x1b4/0x1c0
  [   14.481390]  ? copy_msghdr_from_user+0x62/0x80
  [   14.482048]  ___sys_sendmsg+0x78/0xb0
  [   14.482580]  ? vmf_insert_pfn_prot+0x91/0x150
  [   14.483215]  ? __do_fault+0x2a/0x1a0
  [   14.483738]  ? do_fault+0x15e/0x5d0
  [   14.484246]  ? __handle_mm_fault+0x56b/0x1040
  [   14.484874]  ? lock_is_held_type+0xdf/0x130
  [   14.485474]  ? find_held_lock+0x2d/0x90
  [   14.486046]  ? __sys_sendmsg+0x41/0x70
  [   14.486587]  __sys_sendmsg+0x41/0x70
  [   14.487105]  ? intel_pmu_drain_pebs_core+0x350/0x350
  [   14.487822]  do_syscall_64+0x34/0x80
  [   14.488345]  entry_SYSCALL_64_after_hwframe+0x63/0xcd
  [...]

The test scenario has the following flow:

thread1                               thread2
-----------                           ---------------
 tcp_bpf_sendmsg
  tcp_bpf_send_verdict
   tcp_bpf_sendmsg_redir              sock_close
    tcp_bpf_push_locked                 __sock_release
     tcp_bpf_push                         //inet_release
      do_tcp_sendpages                    sock->ops->release
       sk_stream_wait_memory              // tcp_close
          sk_wait_event                      sk->sk_prot->close
           release_sock(__sk);
            ***
                                                lock_sock(sk);
                                                  __tcp_close
                                                    sock_orphan(sk)
                                                      sk->sk_wq  = NULL
                                                release_sock
            ****
           lock_sock(__sk);
          remove_wait_queue(sk_sleep(sk), &wait);
             sk_sleep(sk)
             //NULL pointer dereference
             &rcu_dereference_raw(sk->sk_wq)->wait

While waiting for memory in thread1, the socket is released with its wait
queue because thread2 has closed it. This caused by tcp_bpf_send_verdict
didn't increase the f_count of psock->sk_redir->sk_socket->file in thread1.

We should check if SOCK_DEAD flag is set on wakeup in sk_stream_wait_memory
before accessing the wait queue.

Suggested-by: Jakub Sitnicki <jakub@cloudflare.com>
Signed-off-by: Liu Jian <liujian56@huawei.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: John Fastabend <john.fastabend@gmail.com>
Cc: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/bpf/20220823133755.314697-2-liujian56@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rt2x00: correctly set BBP register 86 for MT7620

[ Upstream commit c9aada64fe6493461127f1522d7e2f01792d2424 ]

Instead of 0 set the correct value for BBP register 86 for MT7620.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/257267247ee4fa7ebc6a5d0c4948b3f8119c0d77.1663445157.git.daniel@makrotopia.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rt2x00: set SoC wmac clock register

[ Upstream commit cbde6ed406a51092d9e8a2df058f5f8490f27443 ]

Instead of using the default value 33 (pci), set US_CYC_CNT init based
on Programming guide:
If available, set chipset bus clock with fallback to cpu clock/3.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/3e275d259f476f597dab91a9c395015ef3fe3284.1663445157.git.daniel@makrotopia.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rt2x00: set VGC gain for both chains of MT7620

[ Upstream commit 0e09768c085709e10ece3b68f6ac921d3f6a9caa ]

Set bbp66 for all chains of the MT7620.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/29e161397e5c9d9399da0fe87d44458aa2b90a78.1663445157.git.daniel@makrotopia.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rt2x00: set correct TX_SW_CFG1 MAC register for MT7620

[ Upstream commit eeb50acf15762b61921f9df18663f839f387c054 ]

Set correct TX_SW_CFG1 MAC register as it is done also in v3 of the
vendor driver[1].

[1]: https://gitlab.com/dm38/padavan-ng/-/blob/master/trunk/proprietary/rt_wifi/rtpci/3.0.X.X/mt76x2/chips/rt6352.c#L531
Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Acked-by: Stanislaw Gruszka <stf_xl@wp.pl>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/4be38975ce600a34249e12d09a3cb758c6e71071.1663445157.git.daniel@makrotopia.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: rt2x00: don't run Rt5592 IQ calibration on MT7620

[ Upstream commit d3aad83d05aec0cfd7670cf0028f2ad4b81de92e ]

The function rt2800_iq_calibrate is intended for Rt5592 only.
Don't call it for MT7620 which has it's own calibration functions.

Reported-by: Serge Vasilugin <vasilugin@yandex.ru>
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/31a1c34ddbd296b82f38c18c9ae7339059215fdc.1663445157.git.daniel@makrotopia.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

can: bcm: check the result of can_send() in bcm_can_tx()

[ Upstream commit 3fd7bfd28cfd68ae80a2fe92ea1615722cc2ee6e ]

If can_send() fail, it should not update frames_abs counter
in bcm_can_tx(). Add the result check for can_send() in bcm_can_tx().

Suggested-by: Marc Kleine-Budde <mkl@pengutronix.de>
Suggested-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Ziyang Xuan <william.xuanziyang@huawei.com>
Link: https://lore.kernel.org/all/9851878e74d6d37aee2f1ee76d68361a46f89458.1663206163.git.william.xuanziyang@huawei.com
Acked-by: Oliver Hartkopp <socketcan@hartkopp.net>
Signed-off-by: Marc Kleine-Budde <mkl@pengutronix.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: hci_sysfs: Fix attempting to call device_add multiple times

[ Upstream commit 448a496f760664d3e2e79466aa1787e6abc922b5 ]

device_add shall not be called multiple times as stated in its
documentation:

 'Do not call this routine or device_register() more than once for
 any device structure'

Syzkaller reports a bug as follows [1]:
------------[ cut here ]------------
kernel BUG at lib/list_debug.c:33!
invalid opcode: 0000 [#1] PREEMPT SMP KASAN
[...]
Call Trace:
 <TASK>
 __list_add include/linux/list.h:69 [inline]
 list_add_tail include/linux/list.h:102 [inline]
 kobj_kset_join lib/kobject.c:164 [inline]
 kobject_add_internal+0x18f/0x8f0 lib/kobject.c:214
 kobject_add_varg lib/kobject.c:358 [inline]
 kobject_add+0x150/0x1c0 lib/kobject.c:410
 device_add+0x368/0x1e90 drivers/base/core.c:3452
 hci_conn_add_sysfs+0x9b/0x1b0 net/bluetooth/hci_sysfs.c:53
 hci_le_cis_estabilished_evt+0x57c/0xae0 net/bluetooth/hci_event.c:6799
 hci_le_meta_evt+0x2b8/0x510 net/bluetooth/hci_event.c:7110
 hci_event_func net/bluetooth/hci_event.c:7440 [inline]
 hci_event_packet+0x63d/0xfd0 net/bluetooth/hci_event.c:7495
 hci_rx_work+0xae7/0x1230 net/bluetooth/hci_core.c:4007
 process_one_work+0x991/0x1610 kernel/workqueue.c:2289
 worker_thread+0x665/0x1080 kernel/workqueue.c:2436
 kthread+0x2e4/0x3a0 kernel/kthread.c:376
 ret_from_fork+0x1f/0x30 arch/x86/entry/entry_64.S:306
 </TASK>

Link: https://syzkaller.appspot.com/bug?id=da3246e2d33afdb92d66bc166a0934c5b146404a
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Tested-by: Hawkins Jiawei <yin31149@gmail.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Bluetooth: L2CAP: initialize delayed works at l2cap_chan_create()

[ Upstream commit 2d2cb3066f2c90cd8ca540b36ba7a55e7f2406e0 ]

syzbot is reporting cancel_delayed_work() without INIT_DELAYED_WORK() at
l2cap_chan_del() [1], for CONF_NOT_COMPLETE flag (which meant to prevent
l2cap_chan_del() from calling cancel_delayed_work()) is cleared by timer
which fires before l2cap_chan_del() is called by closing file descriptor
created by socket(AF_BLUETOOTH, SOCK_STREAM, BTPROTO_L2CAP).

l2cap_bredr_sig_cmd(L2CAP_CONF_REQ) and l2cap_bredr_sig_cmd(L2CAP_CONF_RSP)
are calling l2cap_ertm_init(chan), and they call l2cap_chan_ready() (which
clears CONF_NOT_COMPLETE flag) only when l2cap_ertm_init(chan) succeeded.

l2cap_sock_init() does not call l2cap_ertm_init(chan), and it instead sets
CONF_NOT_COMPLETE flag by calling l2cap_chan_set_defaults(). However, when
connect() is requested, "command 0x0409 tx timeout" happens after 2 seconds
 from connect() request, and CONF_NOT_COMPLETE flag is cleared after 4
seconds from connect() request, for l2cap_conn_start() from
l2cap_info_timeout() callback scheduled by

  schedule_delayed_work(&conn->info_timer, L2CAP_INFO_TIMEOUT);

in l2cap_connect() is calling l2cap_chan_ready().

Fix this problem by initializing delayed works used by L2CAP_MODE_ERTM
mode as soon as l2cap_chan_create() allocates a channel, like I did in
commit be8597239379f0f5 ("Bluetooth: initialize skb_queue_head at
l2cap_chan_create()").

Link: https://syzkaller.appspot.com/bug?extid=83672956c7aa6af698b3
Reported-by: syzbot <syzbot+83672956c7aa6af698b3@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

regulator: core: Prevent integer underflow

[ Upstream commit 8d8e16592022c9650df8aedfe6552ed478d7135b ]

By using a ratio of delay to poll_enabled_time that is not integer
time_remaining underflows and does not exit the loop as expected.
As delay could be derived from DT and poll_enabled_time is defined
in the driver this can easily happen.

Use a signed iterator to make sure that the loop exits once
the remaining time is negative.

Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Link: https://lore.kernel.org/r/20220909125954.577669-1-patrick.rudolph@9elements.com
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: brcmfmac: fix use-after-free bug in brcmf_netdev_start_xmit()

[ Upstream commit 3f42faf6db431e04bf942d2ebe3ae88975723478 ]

> ret = brcmf_proto_tx_queue_data(drvr, ifp->ifidx, skb);

may be schedule, and then complete before the line

> ndev->stats.tx_bytes += skb->len;

[   46.912801] ==================================================================
[   46.920552] BUG: KASAN: use-after-free in brcmf_netdev_start_xmit+0x718/0x8c8 [brcmfmac]
[   46.928673] Read of size 4 at addr ffffff803f5882e8 by task systemd-resolve/328
[   46.935991]
[   46.937514] CPU: 1 PID: 328 Comm: systemd-resolve Tainted: G           O      5.4.199-[REDACTED] #1
[   46.947255] Hardware name: [REDACTED]
[   46.954568] Call trace:
[   46.957037]  dump_backtrace+0x0/0x2b8
[   46.960719]  show_stack+0x24/0x30
[   46.964052]  dump_stack+0x128/0x194
[   46.967557]  print_address_description.isra.0+0x64/0x380
[   46.972877]  __kasan_report+0x1d4/0x240
[   46.976723]  kasan_report+0xc/0x18
[   46.980138]  __asan_report_load4_noabort+0x18/0x20
[   46.985027]  brcmf_netdev_start_xmit+0x718/0x8c8 [brcmfmac]
[   46.990613]  dev_hard_start_xmit+0x1bc/0xda0
[   46.994894]  sch_direct_xmit+0x198/0xd08
[   46.998827]  __qdisc_run+0x37c/0x1dc0
[   47.002500]  __dev_queue_xmit+0x1528/0x21f8
[   47.006692]  dev_queue_xmit+0x24/0x30
[   47.010366]  neigh_resolve_output+0x37c/0x678
[   47.014734]  ip_finish_output2+0x598/0x2458
[   47.018927]  __ip_finish_output+0x300/0x730
[   47.023118]  ip_output+0x2e0/0x430
[   47.026530]  ip_local_out+0x90/0x140
[   47.030117]  igmpv3_sendpack+0x14c/0x228
[   47.034049]  igmpv3_send_cr+0x384/0x6b8
[   47.037895]  igmp_ifc_timer_expire+0x4c/0x118
[   47.042262]  call_timer_fn+0x1cc/0xbe8
[   47.046021]  __run_timers+0x4d8/0xb28
[   47.049693]  run_timer_softirq+0x24/0x40
[   47.053626]  __do_softirq+0x2c0/0x117c
[   47.057387]  irq_exit+0x2dc/0x388
[   47.060715]  __handle_domain_irq+0xb4/0x158
[   47.064908]  gic_handle_irq+0x58/0xb0
[   47.068581]  el0_irq_naked+0x50/0x5c
[   47.072162]
[   47.073665] Allocated by task 328:
[   47.077083]  save_stack+0x24/0xb0
[   47.080410]  __kasan_kmalloc.isra.0+0xc0/0xe0
[   47.084776]  kasan_slab_alloc+0x14/0x20
[   47.088622]  kmem_cache_alloc+0x15c/0x468
[   47.092643]  __alloc_skb+0xa4/0x498
[   47.096142]  igmpv3_newpack+0x158/0xd78
[   47.099987]  add_grhead+0x210/0x288
[   47.103485]  add_grec+0x6b0/0xb70
[   47.106811]  igmpv3_send_cr+0x2e0/0x6b8
[   47.110657]  igmp_ifc_timer_expire+0x4c/0x118
[   47.115027]  call_timer_fn+0x1cc/0xbe8
[   47.118785]  __run_timers+0x4d8/0xb28
[   47.122457]  run_timer_softirq+0x24/0x40
[   47.126389]  __do_softirq+0x2c0/0x117c
[   47.130142]
[   47.131643] Freed by task 180:
[   47.134712]  save_stack+0x24/0xb0
[   47.138041]  __kasan_slab_free+0x108/0x180
[   47.142146]  kasan_slab_free+0x10/0x18
[   47.145904]  slab_free_freelist_hook+0xa4/0x1b0
[   47.150444]  kmem_cache_free+0x8c/0x528
[   47.154292]  kfree_skbmem+0x94/0x108
[   47.157880]  consume_skb+0x10c/0x5a8
[   47.161466]  __dev_kfree_skb_any+0x88/0xa0
[   47.165598]  brcmu_pkt_buf_free_skb+0x44/0x68 [brcmutil]
[   47.171023]  brcmf_txfinalize+0xec/0x190 [brcmfmac]
[   47.176016]  brcmf_proto_bcdc_txcomplete+0x1c0/0x210 [brcmfmac]
[   47.182056]  brcmf_sdio_sendfromq+0x8dc/0x1e80 [brcmfmac]
[   47.187568]  brcmf_sdio_dpc+0xb48/0x2108 [brcmfmac]
[   47.192529]  brcmf_sdio_dataworker+0xc8/0x238 [brcmfmac]
[   47.197859]  process_one_work+0x7fc/0x1a80
[   47.201965]  worker_thread+0x31c/0xc40
[   47.205726]  kthread+0x2d8/0x370
[   47.208967]  ret_from_fork+0x10/0x18
[   47.212546]
[   47.214051] The buggy address belongs to the object at ffffff803f588280
[   47.214051]  which belongs to the cache skbuff_head_cache of size 208
[   47.227086] The buggy address is located 104 bytes inside of
[   47.227086]  208-byte region [ffffff803f588280, ffffff803f588350)
[   47.238814] The buggy address belongs to the page:
[   47.243618] page:ffffffff00dd6200 refcount:1 mapcount:0 mapping:ffffff804b6bf800 index:0xffffff803f589900 compound_mapcount: 0
[   47.255007] flags: 0x10200(slab|head)
[   47.258689] raw: 0000000000010200 ffffffff00dfa980 0000000200000002 ffffff804b6bf800
[   47.266439] raw: ffffff803f589900 0000000080190018 00000001ffffffff 0000000000000000
[   47.274180] page dumped because: kasan: bad access detected
[   47.279752]
[   47.281251] Memory state around the buggy address:
[   47.286051]  ffffff803f588180: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[   47.293277]  ffffff803f588200: fb fb fc fc fc fc fc fc fc fc fc fc fc fc fc fc
[   47.300502] >ffffff803f588280: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb
[   47.307723]                                                           ^
[   47.314343]  ffffff803f588300: fb fb fb fb fb fb fb fb fb fb fc fc fc fc fc fc
[   47.321569]  ffffff803f588380: fc fc fc fc fc fc fc fc fb fb fb fb fb fb fb fb
[   47.328789] ==================================================================

Signed-off-by: Alexander Coffin <alex.coffin@matician.com>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220808174925.3922558-1-alex.coffin@matician.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

xfrm: Update ipcomp_scratches with NULL when freed

[ Upstream commit 8a04d2fc700f717104bfb95b0f6694e448a4537f ]

Currently if ipcomp_alloc_scratches() fails to allocate memory
ipcomp_scratches holds obsolete address. So when we try to free the
percpu scratches using ipcomp_free_scratches() it tries to vfree non
existent vm area. Described below:

static void * __percpu *ipcomp_alloc_scratches(void)
{
        ...
        scratches = alloc_percpu(void *);
        if (!scratches)
                return NULL;
ipcomp_scratches does not know about this allocation failure.
Therefore holding the old obsolete address.
        ...
}

So when we free,

static void ipcomp_free_scratches(void)
{
        ...
        scratches = ipcomp_scratches;
Assigning obsolete address from ipcomp_scratches

        if (!scratches)
                return;

        for_each_possible_cpu(i)
               vfree(*per_cpu_ptr(scratches, i));
Trying to free non existent page, causing warning: trying to vfree
existent vm area.
        ...
}

Fix this breakage by updating ipcomp_scrtches with NULL when scratches
is freed

Suggested-by: Herbert Xu <herbert@gondor.apana.org.au>
Reported-by: syzbot+5ec9bb042ddfe9644773@syzkaller.appspotmail.com
Tested-by: syzbot+5ec9bb042ddfe9644773@syzkaller.appspotmail.com
Signed-off-by: Khalid Masum <khalid.masum.92@gmail.com>
Acked-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Steffen Klassert <steffen.klassert@secunet.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath9k: avoid uninit memory read in ath9k_htc_rx_msg()

[ Upstream commit b383e8abed41cc6ff1a3b34de75df9397fa4878c ]

syzbot is reporting uninit value at ath9k_htc_rx_msg() [1], for
ioctl(USB_RAW_IOCTL_EP_WRITE) can call ath9k_hif_usb_rx_stream() with
pkt_len = 0 but ath9k_hif_usb_rx_stream() uses
__dev_alloc_skb(pkt_len + 32, GFP_ATOMIC) based on an assumption that
pkt_len is valid. As a result, ath9k_hif_usb_rx_stream() allocates skb
with uninitialized memory and ath9k_htc_rx_msg() is reading from
uninitialized memory.

Since bytes accessed by ath9k_htc_rx_msg() is not known until
ath9k_htc_rx_msg() is called, it would be difficult to check minimal valid
pkt_len at "if (pkt_len > 2 * MAX_RX_BUF_SIZE) {" line in
ath9k_hif_usb_rx_stream().

We have two choices. One is to workaround by adding __GFP_ZERO so that
ath9k_htc_rx_msg() sees 0 if pkt_len is invalid. The other is to let
ath9k_htc_rx_msg() validate pkt_len before accessing. This patch chose
the latter.

Note that I'm not sure threshold condition is correct, for I can't find
details on possible packet length used by this protocol.

Link: https://syzkaller.appspot.com/bug?extid=2ca247c2d60c7023de7f
Reported-by: syzbot <syzbot+2ca247c2d60c7023de7f@syzkaller.appspotmail.com>
Signed-off-by: Tetsuo Handa <penguin-kernel@I-love.SAKURA.ne.jp>
Acked-by: Toke Høiland-Jørgensen <toke@toke.dk>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/7acfa1be-4b5c-b2ce-de43-95b0593fb3e5@I-love.SAKURA.ne.jp
Signed-off-by: Sasha Levin <sashal@kernel.org>

tcp: annotate data-race around tcp_md5sig_pool_populated

[ Upstream commit aacd467c0a576e5e44d2de4205855dc0fe43f6fb ]

tcp_md5sig_pool_populated can be read while another thread
changes its value.

The race has no consequence because allocations
are protected with tcp_md5sig_mutex.

This patch adds READ_ONCE() and WRITE_ONCE() to document
the race and silence KCSAN.

Reported-by: Abhishek Shah <abhishek.shah@columbia.edu>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

openvswitch: Fix overreporting of drops in dropwatch

[ Upstream commit c21ab2afa2c64896a7f0e3cbc6845ec63dcfad2e ]

Currently queue_userspace_packet will call kfree_skb for all frames,
whether or not an error occurred. This can result in a single dropped
frame being reported as multiple drops in dropwatch. This functions
caller may also call kfree_skb in case of an error. This patch will
consume the skbs instead and allow caller's to use kfree_skb.

Signed-off-by: Mike Pattrick <mkp@redhat.com>
Link: https://bugzilla.redhat.com/show_bug.cgi?id=2109957
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

openvswitch: Fix double reporting of drops in dropwatch

[ Upstream commit 1100248a5c5ccd57059eb8d02ec077e839a23826 ]

Frames sent to userspace can be reported as dropped in
ovs_dp_process_packet, however, if they are dropped in the netlink code
then netlink_attachskb will report the same frame as dropped.

This patch checks for error codes which indicate that the frame has
already been freed.

Signed-off-by: Mike Pattrick <mkp@redhat.com>
Link: https://bugzilla.redhat.com/show_bug.cgi?id=2109946
Signed-off-by: David S. Miller <davem@davemloft.net>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpftool: Clear errno after libcap's checks

[ Upstream commit cea558855c39b7f1f02ff50dcf701ca6596bc964 ]

When bpftool is linked against libcap, the library runs a "constructor"
function to compute the number of capabilities of the running kernel
[0], at the beginning of the execution of the program. As part of this,
it performs multiple calls to prctl(). Some of these may fail, and set
errno to a non-zero value:

    # strace -e prctl ./bpftool version
    prctl(PR_CAPBSET_READ, CAP_MAC_OVERRIDE) = 1
    prctl(PR_CAPBSET_READ, 0x30 /* CAP_??? */) = -1 EINVAL (Invalid argument)
    prctl(PR_CAPBSET_READ, CAP_CHECKPOINT_RESTORE) = 1
    prctl(PR_CAPBSET_READ, 0x2c /* CAP_??? */) = -1 EINVAL (Invalid argument)
    prctl(PR_CAPBSET_READ, 0x2a /* CAP_??? */) = -1 EINVAL (Invalid argument)
    prctl(PR_CAPBSET_READ, 0x29 /* CAP_??? */) = -1 EINVAL (Invalid argument)
    ** fprintf added at the top of main(): we have errno == 1
    ./bpftool v7.0.0
    using libbpf v1.0
    features: libbfd, libbpf_strict, skeletons
    +++ exited with 0 +++

This has been addressed in libcap 2.63 [1], but until this version is
available everywhere, we can fix it on bpftool side.

Let's clean errno at the beginning of the main() function, to make sure
that these checks do not interfere with the batch mode, where we error
out if errno is set after a bpftool command.

  [0] https://git.kernel.org/pub/scm/libs/libcap/libcap.git/tree/libcap/cap_alloc.c?h=libcap-2.65#n20
  [1] https://git.kernel.org/pub/scm/libs/libcap/libcap.git/commit/?id=f25a1b7e69f7b33e6afb58b3e38f3450b7d2d9a0

Signed-off-by: Quentin Monnet <quentin@isovalent.com>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Link: https://lore.kernel.org/bpf/20220815162205.45043-1-quentin@isovalent.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: brcmfmac: fix invalid address access when enabling SCAN log level

[ Upstream commit aa666b68e73fc06d83c070d96180b9010cf5a960 ]

The variable i is changed when setting random MAC address and causes
invalid address access when printing the value of pi->reqs[i]->reqid.

We replace reqs index with ri to fix the issue.

[  136.726473] Unable to handle kernel access to user memory outside uaccess routines at virtual address 0000000000000000
[  136.737365] Mem abort info:
[  136.740172]   ESR = 0x96000004
[  136.743359]   Exception class = DABT (current EL), IL = 32 bits
[  136.749294]   SET = 0, FnV = 0
[  136.752481]   EA = 0, S1PTW = 0
[  136.755635] Data abort info:
[  136.758514]   ISV = 0, ISS = 0x00000004
[  136.762487]   CM = 0, WnR = 0
[  136.765522] user pgtable: 4k pages, 48-bit VAs, pgdp = 000000005c4e2577
[  136.772265] [0000000000000000] pgd=0000000000000000
[  136.777160] Internal error: Oops: 96000004 [#1] PREEMPT SMP
[  136.782732] Modules linked in: brcmfmac(O) brcmutil(O) cfg80211(O) compat(O)
[  136.789788] Process wificond (pid: 3175, stack limit = 0x00000000053048fb)
[  136.796664] CPU: 3 PID: 3175 Comm: wificond Tainted: G           O      4.19.42-00001-g531a5f5 #1
[  136.805532] Hardware name: Freescale i.MX8MQ EVK (DT)
[  136.810584] pstate: 60400005 (nZCv daif +PAN -UAO)
[  136.815429] pc : brcmf_pno_config_sched_scans+0x6cc/0xa80 [brcmfmac]
[  136.821811] lr : brcmf_pno_config_sched_scans+0x67c/0xa80 [brcmfmac]
[  136.828162] sp : ffff00000e9a3880
[  136.831475] x29: ffff00000e9a3890 x28: ffff800020543400
[  136.836786] x27: ffff8000b1008880 x26: ffff0000012bf6a0
[  136.842098] x25: ffff80002054345c x24: ffff800088d22400
[  136.847409] x23: ffff0000012bf638 x22: ffff0000012bf6d8
[  136.852721] x21: ffff8000aced8fc0 x20: ffff8000ac164400
[  136.858032] x19: ffff00000e9a3946 x18: 0000000000000000
[  136.863343] x17: 0000000000000000 x16: 0000000000000000
[  136.868655] x15: ffff0000093f3b37 x14: 0000000000000050
[  136.873966] x13: 0000000000003135 x12: 0000000000000000
[  136.879277] x11: 0000000000000000 x10: ffff000009a61888
[  136.884589] x9 : 000000000000000f x8 : 0000000000000008
[  136.889900] x7 : 303a32303d726464 x6 : ffff00000a1f957d
[  136.895211] x5 : 0000000000000000 x4 : ffff00000e9a3942
[  136.900523] x3 : 0000000000000000 x2 : ffff0000012cead8
[  136.905834] x1 : ffff0000012bf6d8 x0 : 0000000000000000
[  136.911146] Call trace:
[  136.913623]  brcmf_pno_config_sched_scans+0x6cc/0xa80 [brcmfmac]
[  136.919658]  brcmf_pno_start_sched_scan+0xa4/0x118 [brcmfmac]
[  136.925430]  brcmf_cfg80211_sched_scan_start+0x80/0xe0 [brcmfmac]
[  136.931636]  nl80211_start_sched_scan+0x140/0x308 [cfg80211]
[  136.937298]  genl_rcv_msg+0x358/0x3f4
[  136.940960]  netlink_rcv_skb+0xb4/0x118
[  136.944795]  genl_rcv+0x34/0x48
[  136.947935]  netlink_unicast+0x264/0x300
[  136.951856]  netlink_sendmsg+0x2e4/0x33c
[  136.955781]  __sys_sendto+0x120/0x19c

Signed-off-by: Wright Feng <wright.feng@cypress.com>
Signed-off-by: Chi-hsien Lin <chi-hsien.lin@cypress.com>
Signed-off-by: Ahmad Fatoum <a.fatoum@pengutronix.de>
Signed-off-by: Alvin Šipraga <alsi@bang-olufsen.dk>
Signed-off-by: Kalle Valo <kvalo@kernel.org>
Link: https://lore.kernel.org/r/20220722115632.620681-4-alvin@pqrs.dk
Signed-off-by: Sasha Levin <sashal@kernel.org>

NFSD: fix use-after-free on source server when doing inter-server copy

[ Upstream commit 019805fea91599b22dfa62ffb29c022f35abeb06 ]

Use-after-free occurred when the laundromat tried to free expired
cpntf_state entry on the s2s_cp_stateids list after inter-server
copy completed. The sc_cp_list that the expired copy state was
inserted on was already freed.

When COPY completes, the Linux client normally sends LOCKU(lock_state x),
FREE_STATEID(lock_state x) and CLOSE(open_state y) to the source server.
The nfs4_put_stid call from nfsd4_free_stateid cleans up the copy state
from the s2s_cp_stateids list before freeing the lock state's stid.

However, sometimes the CLOSE was sent before the FREE_STATEID request.
When this happens, the nfsd4_close_open_stateid call from nfsd4_close
frees all lock states on its st_locks list without cleaning up the copy
state on the sc_cp_list list. When the time the FREE_STATEID arrives the
server returns BAD_STATEID since the lock state was freed. This causes
the use-after-free error to occur when the laundromat tries to free
the expired cpntf_state.

This patch adds a call to nfs4_free_cpntf_statelist in
nfsd4_close_open_stateid to clean up the copy state before calling
free_ol_stateid_reaplist to free the lock state's stid on the reaplist.

Signed-off-by: Dai Ngo <dai.ngo@oracle.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

NFSD: Return nfserr_serverfault if splice_ok but buf->pages have data

[ Upstream commit 06981d560606ac48d61e5f4fff6738b925c93173 ]

This was discussed with Chuck as part of this patch set. Returning
nfserr_resource was decided to not be the best error message here, and
he suggested changing to nfserr_serverfault instead.

Signed-off-by: Anna Schumaker <Anna.Schumaker@Netapp.com>
Link: https://lore.kernel.org/linux-nfs/20220907195259.926736-1-anna@kernel.org/T/#t
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/entry: Work around Clang __bdos() bug

[ Upstream commit 3e1730842f142add55dc658929221521a9ea62b6 ]

Clang produces a false positive when building with CONFIG_FORTIFY_SOURCE=y
and CONFIG_UBSAN_BOUNDS=y when operating on an array with a dynamic
offset. Work around this by using a direct assignment of an empty
instance. Avoids this warning:

../include/linux/fortify-string.h:309:4: warning: call to __write_overflow_field declared with 'warn
ing' attribute: detected write beyond size of field (1st parameter); maybe use struct_group()? [-Wat
tribute-warning]
                        __write_overflow_field(p_size_field, size);
                        ^

which was isolated to the memset() call in xen_load_idt().

Note that this looks very much like another bug that was worked around:
https://github.com/ClangBuiltLinux/linux/issues/1592

Cc: Juergen Gross <jgross@suse.com>
Cc: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: x86@kernel.org
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: xen-devel@lists.xenproject.org
Reviewed-by: Boris Ostrovsky <boris.ostrovsky@oracle.com>
Link: https://lore.kernel.org/lkml/41527d69-e8ab-3f86-ff37-6b298c01d5bc@oracle.com
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

thermal: intel_powerclamp: Use get_cpu() instead of smp_processor_id() to avoid crash

[ Upstream commit 68b99e94a4a2db6ba9b31fe0485e057b9354a640 ]

When CPU 0 is offline and intel_powerclamp is used to inject
idle, it generates kernel BUG:

BUG: using smp_processor_id() in preemptible [00000000] code: bash/15687
caller is debug_smp_processor_id+0x17/0x20
CPU: 4 PID: 15687 Comm: bash Not tainted 5.19.0-rc7+ #57
Call Trace:
<TASK>
dump_stack_lvl+0x49/0x63
dump_stack+0x10/0x16
check_preemption_disabled+0xdd/0xe0
debug_smp_processor_id+0x17/0x20
powerclamp_set_cur_state+0x7f/0xf9 [intel_powerclamp]
...
...

Here CPU 0 is the control CPU by default and changed to the current CPU,
if CPU 0 offlined. This check has to be performed under cpus_read_lock(),
hence the above warning.

Use get_cpu() instead of smp_processor_id() to avoid this BUG.

Suggested-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Srinivas Pandruvada <srinivas.pandruvada@linux.intel.com>
[ rjw: Subject edits ]
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

powercap: intel_rapl: fix UBSAN shift-out-of-bounds issue

[ Upstream commit 2d93540014387d1c73b9ccc4d7895320df66d01b ]

When value < time_unit, the parameter of ilog2() will be zero and
the return value is -1. u64(-1) is too large for shift exponent
and then will trigger shift-out-of-bounds:

shift exponent 18446744073709551615 is too large for 32-bit type 'int'
Call Trace:
 rapl_compute_time_window_core
 rapl_write_data_raw
 set_time_window
 store_constraint_time_window_us

Signed-off-by: Chao Qin <chao.qin@intel.com>
Acked-by: Zhang Rui <rui.zhang@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

MIPS: BCM47XX: Cast memcmp() of function to (void *)

[ Upstream commit 0dedcf6e3301836eb70cfa649052e7ce4fcd13ba ]

Clang is especially sensitive about argument type matching when using
__overloaded functions (like memcmp(), etc). Help it see that function
pointers are just "void *". Avoids this error:

arch/mips/bcm47xx/prom.c:89:8: error: no matching function for call to 'memcmp'
                   if (!memcmp(prom_init, prom_init + mem, 32))
                        ^~~~~~
include/linux/string.h:156:12: note: candidate function not viable: no known conversion from 'void (void)' to 'const void *' for 1st argument extern int memcmp(const void *,const void *,__kernel_size_t);

Cc: Hauke Mehrtens <hauke@hauke-m.de>
Cc: "Rafał Miłecki" <zajec5@gmail.com>
Cc: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Cc: linux-mips@vger.kernel.org
Cc: Nathan Chancellor <nathan@kernel.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Cc: llvm@lists.linux.dev
Reported-by: kernel test robot <lkp@intel.com>
Link: https://lore.kernel.org/lkml/202209080652.sz2d68e5-lkp@intel.com
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: video: Add Toshiba Satellite/Portege Z830 quirk

[ Upstream commit 574160b8548deff8b80b174f03201e94ab8431e2 ]

Toshiba Satellite Z830 needs the quirk video_disable_backlight_sysfs_if
for proper backlight control after suspend/resume cycles.

Toshiba Portege Z830 is simply the same laptop rebranded for certain
markets (I looked through the manual to other language sections to confirm
this) and thus also needs this quirk.

Thanks to Hans de Goede for suggesting this fix.

Link: https://www.spinics.net/lists/platform-driver-x86/msg34394.html
Suggested-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Arvid Norlander <lkml@vorpal.se>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Tested-by: Arvid Norlander <lkml@vorpal.se>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rcu-tasks: Convert RCU_LOCKDEP_WARN() to WARN_ONCE()

[ Upstream commit fcd53c8a4dfa38bafb89efdd0b0f718f3a03f884 ]

Kernels built with CONFIG_PROVE_RCU=y and CONFIG_DEBUG_LOCK_ALLOC=y
attempt to emit a warning when the synchronize_rcu_tasks_generic()
function is called during early boot while the rcu_scheduler_active
variable is RCU_SCHEDULER_INACTIVE.  However the warnings is not
actually be printed because the debug_lockdep_rcu_enabled() returns
false, exactly because the rcu_scheduler_active variable is still equal
to RCU_SCHEDULER_INACTIVE.

This commit therefore replaces RCU_LOCKDEP_WARN() with WARN_ONCE()
to force these warnings to actually be printed.

Signed-off-by: Zqiang <qiang1.zhang@intel.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

rcu: Back off upon fill_page_cache_func() allocation failure

[ Upstream commit 093590c16b447f53e66771c8579ae66c96f6ef61 ]

The fill_page_cache_func() function allocates couple of pages to store
kvfree_rcu_bulk_data structures. This is a lightweight (GFP_NORETRY)
allocation which can fail under memory pressure. The function will,
however keep retrying even when the previous attempt has failed.

This retrying is in theory correct, but in practice the allocation is
invoked from workqueue context, which means that if the memory reclaim
gets stuck, these retries can hog the worker for quite some time.
Although the workqueues subsystem automatically adjusts concurrency, such
adjustment is not guaranteed to happen until the worker context sleeps.
And the fill_page_cache_func() function's retry loop is not guaranteed
to sleep (see the should_reclaim_retry() function).

And we have seen this function cause workqueue lockups:

kernel: BUG: workqueue lockup - pool cpus=93 node=1 flags=0x1 nice=0 stuck for 32s!
[...]
kernel: pool 74: cpus=37 node=0 flags=0x1 nice=0 hung=32s workers=2 manager: 2146
kernel:   pwq 498: cpus=249 node=1 flags=0x1 nice=0 active=4/256 refcnt=5
kernel:     in-flight: 1917:fill_page_cache_func
kernel:     pending: dbs_work_handler, free_work, kfree_rcu_monitor

Originally, we thought that the root cause of this lockup was several
retries with direct reclaim, but this is not yet confirmed.  Furthermore,
we have seen similar lockups without any heavy memory pressure.  This
suggests that there are other factors contributing to these lockups.
However, it is not really clear that endless retries are desireable.

So let's make the fill_page_cache_func() function back off after
allocation failure.

Cc: Uladzislau Rezki (Sony) <urezki@gmail.com>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Frederic Weisbecker <frederic@kernel.org>
Cc: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Cc: Lai Jiangshan <jiangshanlai@gmail.com>
Cc: Joel Fernandes <joel@joelfernandes.org>
Signed-off-by: Michal Hocko <mhocko@suse.com>
Reviewed-by: Uladzislau Rezki (Sony) <urezki@gmail.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftest: tpm2: Add Client.__del__() to close /dev/tpm* handle

[ Upstream commit 2d869f0b458547386fbcd8cf3004b271b7347b7f ]

The following output can bee seen when the test is executed:

  test_flush_context (tpm2_tests.SpaceTest) ... \
    /usr/lib64/python3.6/unittest/case.py:605: ResourceWarning: \
    unclosed file <_io.FileIO name='/dev/tpmrm0' mode='rb+' closefd=True>

An instance of Client does not implicitly close /dev/tpm* handle, once it
gets destroyed. Close the file handle in the class destructor
Client.__del__().

Fixes: 6ea3dfe1e0732 ("selftests: add TPM 2.0 tests")
Cc: Shuah Khan <shuah@kernel.org>
Cc: linux-kselftest@vger.kernel.org
Cc: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Stefan Berger <stefanb@linux.ibm.com>
Reviewed-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Jarkko Sakkinen <jarkko@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: fix to account FS_CP_DATA_IO correctly

[ Upstream commit d80afefb17e01aa0c46a8eebc01882e0ebd8b0f6 ]

f2fs_inode_info.cp_task was introduced for FS_CP_DATA_IO accounting
since commit b0af6d491a6b ("f2fs: add app/fs io stat").

However, cp_task usage coverage has been increased due to below
commits:
commit 040d2bb318d1 ("f2fs: fix to avoid deadloop if data_flush is on")
commit 186857c5a14a ("f2fs: fix potential recursive call when enabling data_flush")

So that, if data_flush mountoption is on, when data flush was
triggered from background, the IO from data flush will be accounted
as checkpoint IO type incorrectly.

In order to fix this issue, this patch splits cp_task into two:
a) cp_task: used for IO accounting
b) wb_task: used to avoid deadlock

Fixes: 040d2bb318d1 ("f2fs: fix to avoid deadloop if data_flush is on")
Fixes: 186857c5a14a ("f2fs: fix potential recursive call when enabling data_flush")
Signed-off-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: fix to avoid REQ_TIME and CP_TIME collision

[ Upstream commit 493720a4854343b7c3fe100cda6a3a2c3f8d4b5d ]

Lei Li reported a issue: if foreground operations are frequent, background
checkpoint may be always skipped due to below check, result in losing more
data after sudden power-cut.

f2fs_balance_fs_bg()
...
if (!is_idle(sbi, REQ_TIME) &&
(!excess_dirty_nats(sbi) && !excess_dirty_nodes(sbi)))
return;

E.g:
cp_interval = 5 second
idle_interval = 2 second
foreground operation interval = 1 second (append 1 byte per second into file)

In such case, no matter when it calls f2fs_balance_fs_bg(), is_idle(, REQ_TIME)
returns false, result in skipping background checkpoint.

This patch changes as below to make trigger condition being more reasonable:
- trigger sync_fs() if dirty_{nats,nodes} and prefree segs exceeds threshold;
- skip triggering sync_fs() if there is any background inflight IO or there is
foreground operation recently and meanwhile cp_rwsem is being held by someone;

Reported-by: Lei Li <noctis.akm@gmail.com>
Signed-off-by: Chao Yu <yuchao0@huawei.com>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Stable-dep-of: d80afefb17e0 ("f2fs: fix to account FS_CP_DATA_IO correctly")
Signed-off-by: Sasha Levin <sashal@kernel.org>

f2fs: fix race condition on setting FI_NO_EXTENT flag

[ Upstream commit 07725adc55c0a414c10acb5c8c86cea34b95ddef ]

The following scenarios exist.
process A:               process B:
->f2fs_drop_extent_tree  ->f2fs_update_extent_cache_range
                          ->f2fs_update_extent_tree_range
                           ->write_lock
 ->set_inode_flag
                           ->is_inode_flag_set
                           ->__free_extent_tree // Shouldn't
                                                // have been
                                                // cleaned up
                                                // here
  ->write_lock

In this case, the "FI_NO_EXTENT" flag is set between
f2fs_update_extent_tree_range and is_inode_flag_set
by other process. it leads to clearing the whole exten
tree which should not have happened. And we fix it by
move the setting it to the range of write_lock.

Fixes:5f281fab9b9a3 ("f2fs: disable extent_cache for fcollapse/finsert inodes")
Signed-off-by: Zhang Qilong <zhangqilong3@huawei.com>
Reviewed-by: Chao Yu <chao@kernel.org>
Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI: APEI: do not add task_work to kernel thread to avoid memory leak

[ Upstream commit 415fed694fe11395df56e05022d6e7cee1d39dd3 ]

If an error is detected as a result of user-space process accessing a
corrupt memory location, the CPU may take an abort. Then the platform
firmware reports kernel via NMI like notifications, e.g. NOTIFY_SEA,
NOTIFY_SOFTWARE_DELEGATED, etc.

For NMI like notifications, commit 7f17b4a121d0 ("ACPI: APEI: Kick the
memory_failure() queue for synchronous errors") keep track of whether
memory_failure() work was queued, and make task_work pending to flush out
the queue so that the work is processed before return to user-space.

The code use init_mm to check whether the error occurs in user space:

    if (current->mm != &init_mm)

The condition is always true, becase _nobody_ ever has "init_mm" as a real
VM any more.

In addition to abort, errors can also be signaled as asynchronous
exceptions, such as interrupt and SError. In such case, the interrupted
current process could be any kind of thread. When a kernel thread is
interrupted, the work ghes_kick_task_work deferred to task_work will never
be processed because entry_handler returns to call ret_to_kernel() instead
of ret_to_user(). Consequently, the estatus_node alloced from
ghes_estatus_pool in ghes_in_nmi_queue_one_entry() will not be freed.
After around 200 allocations in our platform, the ghes_estatus_pool will
run of memory and ghes_in_nmi_queue_one_entry() returns ENOMEM. As a
result, the event failed to be processed.

    sdei: event 805 on CPU 113 failed with error: -2

Finally, a lot of unhandled events may cause platform firmware to exceed
some threshold and reboot.

The condition should generally just do

    if (current->mm)

as described in active_mm.rst documentation.

Then if an asynchronous error is detected when a kernel thread is running,
(e.g. when detected by a background scrubber), do not add task_work to it
as the original patch intends to do.

Fixes: 7f17b4a121d0 ("ACPI: APEI: Kick the memory_failure() queue for synchronous errors")
Signed-off-by: Shuai Xue <xueshuai@linux.alibaba.com>
Reviewed-by: Tony Luck <tony.luck@intel.com>
Signed-off-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

thermal/drivers/qcom/tsens-v0_1: Fix MSM8939 fourth sensor hw_id

[ Upstream commit b0c883e900702f408d62cf92b0ef01303ed69be9 ]

Reading temperature from this sensor fails with 'Invalid argument'.

Looking at old vendor dts [1], its hw_id should be 3 instead of 4.
Change this hw_id accordingly.

[1] https://github.com/msm8916-mainline/android_kernel_qcom_msm8916/blob/master/arch/arm/boot/dts/qcom/msm8939-common.dtsi#L511

Fixes: 332bc8ebab2c ("thermal: qcom: tsens-v0_1: Add support for MSM8939")
Signed-off-by: Vincent Knecht <vincent.knecht@mailoo.org>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Reviewed-by: Bjorn Andersson <andersson@kernel.org>
Reviewed-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Link: https://lore.kernel.org/r/20220811105014.7194-1-vincent.knecht@mailoo.org
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: cavium - prevent integer overflow loading firmware

[ Upstream commit 2526d6bf27d15054bb0778b2f7bc6625fd934905 ]

The "code_length" value comes from the firmware file.  If your firmware
is untrusted realistically there is probably very little you can do to
protect yourself.  Still we try to limit the damage as much as possible.
Also Smatch marks any data read from the filesystem as untrusted and
prints warnings if it not capped correctly.

The "ntohl(ucode->code_length) * 2" multiplication can have an
integer overflow.

Fixes: 9e2c7d99941d ("crypto: cavium - Add Support for Octeon-tx CPT Engine")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: marvell/octeontx - prevent integer overflows

[ Upstream commit caca37cf6c749ff0303f68418cfe7b757a4e0697 ]

The "code_length" value comes from the firmware file.  If your firmware
is untrusted realistically there is probably very little you can do to
protect yourself.  Still we try to limit the damage as much as possible.
Also Smatch marks any data read from the filesystem as untrusted and
prints warnings if it not capped correctly.

The "code_length * 2" can overflow.  The round_up(ucode_size, 16) +
sizeof() expression can overflow too.  Prevent these overflows.

Fixes: d9110b0b01ff ("crypto: marvell - add support for OCTEON TX CPT engine")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

kbuild: rpm-pkg: fix breakage when V=1 is used

[ Upstream commit 2e07005f4813a9ff6e895787e0c2d1fea859b033 ]

Doing make V=1 binrpm-pkg results in:

 Executing(%install): /bin/sh -e /var/tmp/rpm-tmp.EgV6qJ
 + umask 022
 + cd .
 + /bin/rm -rf /home/scgl/rpmbuild/BUILDROOT/kernel-6.0.0_rc5+-1.s390x
 + /bin/mkdir -p /home/scgl/rpmbuild/BUILDROOT
 + /bin/mkdir /home/scgl/rpmbuild/BUILDROOT/kernel-6.0.0_rc5+-1.s390x
 + mkdir -p /home/scgl/rpmbuild/BUILDROOT/kernel-6.0.0_rc5+-1.s390x/boot
 + make -f ./Makefile image_name
 + cp test -e include/generated/autoconf.h -a -e include/config/auto.conf || ( \ echo >&2; \ echo >&2 " ERROR: Kernel configuration is invalid."; \ echo >&2 " include/generated/autoconf.h or include/config/auto.conf are missing.";\ echo >&2 " Run 'make oldconfig && make prepare' on kernel src to fix it."; \ echo >&2 ; \ /bin/false) arch/s390/boot/bzImage /home/scgl/rpmbuild/BUILDROOT/kernel-6.0.0_rc5+-1.s390x/boot/vmlinuz-6.0.0-rc5+
 cp: invalid option -- 'e'
 Try 'cp --help' for more information.
 error: Bad exit status from /var/tmp/rpm-tmp.EgV6qJ (%install)

Because the make call to get the image name is verbose and prints
additional information.

Fixes: 993bdde94547 ("kbuild: add image_name to no-sync-config-targets")
Signed-off-by: Janis Schoetterl-Glausch <scgl@linux.ibm.com>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

kbuild: remove the target in signal traps when interrupted

[ Upstream commit a7f3257da8a86b96fb9bf1bba40ae0bbd7f1885a ]

When receiving some signal, GNU Make automatically deletes the target if
it has already been changed by the interrupted recipe.

If the target is possibly incomplete due to interruption, it must be
deleted so that it will be remade from scratch on the next run of make.
Otherwise, the target would remain corrupted permanently because its
timestamp had already been updated.

Thanks to this behavior of Make, you can stop the build any time by
pressing Ctrl-C, and just run 'make' to resume it.

Kbuild also relies on this feature, but it is equivalently important
for any build systems that make decisions based on timestamps (if you
want to support Ctrl-C reliably).

However, this does not always work as claimed; Make immediately dies
with Ctrl-C if its stderr goes into a pipe.

  [Test Makefile]

    foo:
            echo hello > $@
            sleep 3
            echo world >> $@

  [Test Result]

    $ make                         # hit Ctrl-C
    echo hello > foo
    sleep 3
    ^Cmake: *** Deleting file 'foo'
    make: *** [Makefile:3: foo] Interrupt

    $ make 2>&1 | cat              # hit Ctrl-C
    echo hello > foo
    sleep 3
    ^C$                            # 'foo' is often left-over

The reason is because SIGINT is sent to the entire process group.
In this example, SIGINT kills 'cat', and 'make' writes the message to
the closed pipe, then dies with SIGPIPE before cleaning the target.

A typical bad scenario (as reported by [1], [2]) is to save build log
by using the 'tee' command:

    $ make 2>&1 | tee log

This can be problematic for any build systems based on Make, so I hope
it will be fixed in GNU Make. The maintainer of GNU Make stated this is
a long-standing issue and difficult to fix [3]. It has not been fixed
yet as of writing.

So, we cannot rely on Make cleaning the target. We can do it by
ourselves, in signal traps.

As far as I understand, Make takes care of SIGHUP, SIGINT, SIGQUIT, and
SITERM for the target removal. I added the traps for them, and also for
SIGPIPE just in case cmd_* rule prints something to stdout or stderr
(but I did not observe an actual case where SIGPIPE was triggered).

[Note 1]

The trap handler might be worth explaining.

    rm -f $@; trap - $(sig); kill -s $(sig) $$

This lets the shell kill itself by the signal it caught, so the parent
process can tell the child has exited on the signal. Generally, this is
a proper manner for handling signals, in case the calling program (like
Bash) may monitor WIFSIGNALED() and WTERMSIG() for WCE although this may
not be a big deal here because GNU Make handles SIGHUP, SIGINT, SIGQUIT
in WUE and SIGTERM in IUE.

  IUE - Immediate Unconditional Exit
  WUE - Wait and Unconditional Exit
  WCE - Wait and Cooperative Exit

For details, see "Proper handling of SIGINT/SIGQUIT" [4].

[Note 2]

Reverting 392885ee82d3 ("kbuild: let fixdep directly write to .*.cmd
files") would directly address [1], but it only saves if_changed_dep.
As reported in [2], all commands that use redirection can potentially
leave an empty (i.e. broken) target.

[Note 3]

Another (even safer) approach might be to always write to a temporary
file, and rename it to $@ at the end of the recipe.

   <command>  > $(tmp-target)
   mv $(tmp-target) $@

It would require a lot of Makefile changes, and result in ugly code,
so I did not take it.

[Note 4]

A little more thoughts about a pattern rule with multiple targets (or
a grouped target).

    %.x %.y: %.z
            <recipe>

When interrupted, GNU Make deletes both %.x and %.y, while this solution
only deletes $@. Probably, this is not a big deal. The next run of make
will execute the rule again to create $@ along with the other files.

[1]: https://lore.kernel.org/all/YLeot94yAaM4xbMY@gmail.com/
[2]: https://lore.kernel.org/all/20220510221333.2770571-1-robh@kernel.org/
[3]: https://lists.gnu.org/archive/html/help-make/2021-06/msg00001.html
[4]: https://www.cons.org/cracauer/sigint.html

Fixes: 392885ee82d3 ("kbuild: let fixdep directly write to .*.cmd files")
Reported-by: Ingo Molnar <mingo@kernel.org>
Reported-by: Rob Herring <robh@kernel.org>
Signed-off-by: Masahiro Yamada <masahiroy@kernel.org>
Tested-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Nicolas Schier <nicolas@fjasle.eu>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tracing: kprobe: Make gen test module work in arm and riscv

[ Upstream commit d8ef45d66c01425ff748e13ef7dd1da7a91cc93c ]

For now, this selftest module can only work in x86 because of the
kprobe cmd was fixed use of x86 registers.
This patch adapted to register names under arm and riscv, So that
this module can be worked on those platform.

Link: https://lkml.kernel.org/r/20220919125629.238242-3-zouyipeng@huawei.com
Cc: <linux-riscv@lists.infradead.org>
Cc: <mingo@redhat.com>
Cc: <paul.walmsley@sifive.com>
Cc: <palmer@dabbelt.com>
Cc: <aou@eecs.berkeley.edu>
Cc: <zanussi@kernel.org>
Cc: <liaochang1@huawei.com>
Cc: <chris.zjh@huawei.com>
Fixes: 64836248dda2 ("tracing: Add kprobe event command generation test module")
Signed-off-by: Yipeng Zou <zouyipeng@huawei.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tracing: kprobe: Fix kprobe event gen test module on exit

[ Upstream commit ac48e189527fae87253ef2bf58892e782fb36874 ]

Correct gen_kretprobe_test clr event para on module exit.
This will make it can't to delete.

Link: https://lkml.kernel.org/r/20220919125629.238242-2-zouyipeng@huawei.com
Cc: <linux-riscv@lists.infradead.org>
Cc: <mingo@redhat.com>
Cc: <paul.walmsley@sifive.com>
Cc: <palmer@dabbelt.com>
Cc: <aou@eecs.berkeley.edu>
Cc: <zanussi@kernel.org>
Cc: <liaochang1@huawei.com>
Cc: <chris.zjh@huawei.com>
Fixes: 64836248dda2 ("tracing: Add kprobe event command generation test module")
Signed-off-by: Yipeng Zou <zouyipeng@huawei.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iommu/iova: Fix module config properly

[ Upstream commit 4f58330fcc8482aa90674e1f40f601e82f18ed4a ]

IOMMU_IOVA is intended to be an optional library for users to select as
and when they desire. Since it can be a module now, this means that
built-in code which has chosen not to select it should not fail to link
if it happens to have selected as a module by someone else. Replace
IS_ENABLED() with IS_REACHABLE() to do the right thing.

CC: Thierry Reding <thierry.reding@gmail.com>
Reported-by: John Garry <john.garry@huawei.com>
Fixes: 15bbdec3931e ("iommu: Make the iova library a module")
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Thierry Reding <treding@nvidia.com>
Link: https://lore.kernel.org/r/548c2f683ca379aface59639a8f0cccc3a1ac050.1663069227.git.robin.murphy@arm.com
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: qat - fix DMA transfer direction

[ Upstream commit cf5bb835b7c8a5fee7f26455099cca7feb57f5e9 ]

When CONFIG_DMA_API_DEBUG is selected, while running the crypto self
test on the QAT crypto algorithms, the function add_dma_entry() reports
a warning similar to the one below, saying that overlapping mappings
are not supported. This occurs in tests where the input and the output
scatter list point to the same buffers (i.e. two different scatter lists
which point to the same chunks of memory).

The logic that implements the mapping uses the flag DMA_BIDIRECTIONAL
for both the input and the output scatter lists which leads to
overlapped write mappings. These are not supported by the DMA layer.

Fix by specifying the correct DMA transfer directions when mapping
buffers. For in-place operations where the input scatter list
matches the output scatter list, buffers are mapped once with
DMA_BIDIRECTIONAL, otherwise input buffers are mapped using the flag
DMA_TO_DEVICE and output buffers are mapped with DMA_FROM_DEVICE.
Overlapping a read mapping with a write mapping is a valid case in
dma-coherent devices like QAT.
The function that frees and unmaps the buffers, qat_alg_free_bufl()
has been changed accordingly to the changes to the mapping function.

   DMA-API: 4xxx 0000:06:00.0: cacheline tracking EEXIST, overlapping mappings aren't supported
   WARNING: CPU: 53 PID: 4362 at kernel/dma/debug.c:570 add_dma_entry+0x1e9/0x270
   ...
   Call Trace:
   dma_map_page_attrs+0x82/0x2d0
   ? preempt_count_add+0x6a/0xa0
   qat_alg_sgl_to_bufl+0x45b/0x990 [intel_qat]
   qat_alg_aead_dec+0x71/0x250 [intel_qat]
   crypto_aead_decrypt+0x3d/0x70
   test_aead_vec_cfg+0x649/0x810
   ? number+0x310/0x3a0
   ? vsnprintf+0x2a3/0x550
   ? scnprintf+0x42/0x70
   ? valid_sg_divisions.constprop.0+0x86/0xa0
   ? test_aead_vec+0xdf/0x120
   test_aead_vec+0xdf/0x120
   alg_test_aead+0x185/0x400
   alg_test+0x3d8/0x500
   ? crypto_acomp_scomp_free_ctx+0x30/0x30
   ? __schedule+0x32a/0x12a0
   ? ttwu_queue_wakelist+0xbf/0x110
   ? _raw_spin_unlock_irqrestore+0x23/0x40
   ? try_to_wake_up+0x83/0x570
   ? _raw_spin_unlock_irqrestore+0x23/0x40
   ? __set_cpus_allowed_ptr_locked+0xea/0x1b0
   ? crypto_acomp_scomp_free_ctx+0x30/0x30
   cryptomgr_test+0x27/0x50
   kthread+0xe6/0x110
   ? kthread_complete_and_exit+0x20/0x20
   ret_from_fork+0x1f/0x30

Fixes: d370cec ("crypto: qat - Intel(R) QAT crypto interface")
Link: https://lore.kernel.org/linux-crypto/20220223080400.139367-1-gilad@benyossef.com/
Signed-off-by: Damian Muszynski <damian.muszynski@intel.com>
Signed-off-by: Giovanni Cabiddu <giovanni.cabiddu@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: qat - use pre-allocated buffers in datapath

[ Upstream commit e0831e7af4e03f2715de102e18e9179ec0a81562 ]

In order to do DMAs, the QAT device requires that the scatterlist
structures are mapped and translated into a format that the firmware can
understand. This is defined as the composition of a scatter gather list
(SGL) descriptor header, the struct qat_alg_buf_list, plus a variable
number of flat buffer descriptors, the struct qat_alg_buf.

The allocation and mapping of these data structures is done each time a
request is received from the skcipher and aead APIs.
In an OOM situation, this behaviour might lead to a dead-lock if an
allocation fails.

Based on the conversation in [1], increase the size of the aead and
skcipher request contexts to include an SGL descriptor that can handle
a maximum of 4 flat buffers.
If requests exceed 4 entries buffers, memory is allocated dynamically.

[1] https://lore.kernel.org/linux-crypto/20200722072932.GA27544@gondor.apana.org.au/

Cc: stable@vger.kernel.org
Fixes: d370cec32194 ("crypto: qat - Intel(R) QAT crypto interface")
Reported-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Giovanni Cabiddu <giovanni.cabiddu@intel.com>
Reviewed-by: Marco Chiappero <marco.chiappero@intel.com>
Reviewed-by: Wojciech Ziemba <wojciech.ziemba@intel.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Stable-dep-of: cf5bb835b7c8 ("crypto: qat - fix DMA transfer direction")
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: qat - fix use of 'dma_map_single'

[ Upstream commit 7cc05071f930a631040fea16a41f9d78771edc49 ]

DMA_TO_DEVICE synchronisation must be done after the last modification
of the memory region by the software and before it is handed off to
the device.

Signed-off-by: Hui Tang <tanghui20@huawei.com>
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Stable-dep-of: cf5bb835b7c8 ("crypto: qat - fix DMA transfer direction")
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: inside-secure - Change swab to swab32

[ Upstream commit 664593407e936b6438fbfaaf98876910fd31cf9a ]

The use of swab() is causing failures in 64-bit arch, as it
translates to __swab64() instead of the intended __swab32().
It eventually causes wrong results in xcbcmac & cmac algo.

Fixes: 78cf1c8bfcb8 ("crypto: inside-secure - Move ipad/opad into safexcel_context")
Signed-off-by: Peter Harliman Liem <pliem@maxlinear.com>
Acked-by: Antoine Tenart <atenart@kernel.org>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: ccp - Release dma channels before dmaengine unrgister

[ Upstream commit 68dbe80f5b510c66c800b9e8055235c5b07e37d1 ]

A warning is shown during shutdown,

__dma_async_device_channel_unregister called while 2 clients hold a reference
WARNING: CPU: 15 PID: 1 at drivers/dma/dmaengine.c:1110 __dma_async_device_channel_unregister+0xb7/0xc0

Call dma_release_channel for occupied channles before dma_async_device_unregister.

Fixes: 54cce8ecb925 ("crypto: ccp - ccp_dmaengine_unregister release dma channels")
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Koba Ko <koba.ko@canonical.com>
Acked-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: akcipher - default implementation for setting a private key

[ Upstream commit bc155c6c188c2f0c5749993b1405673d25a80389 ]

Changes from v1:
  * removed the default implementation from set_pub_key: it is assumed that
    an implementation must always have this callback defined as there are
    no use case for an algorithm, which doesn't need a public key

Many akcipher implementations (like ECDSA) support only signature
verifications, so they don't have all callbacks defined.

Commit 78a0324f4a53 ("crypto: akcipher - default implementations for
request callbacks") introduced default callbacks for sign/verify
operations, which just return an error code.

However, these are not enough, because before calling sign the caller would
likely call set_priv_key first on the instantiated transform (as the
in-kernel testmgr does). This function does not have a default stub, so the
kernel crashes, when trying to set a private key on an akcipher, which
doesn't support signature generation.

I've noticed this, when trying to add a KAT vector for ECDSA signature to
the testmgr.

With this patch the testmgr returns an error in dmesg (as it should)
instead of crashing the kernel NULL ptr dereference.

Fixes: 78a0324f4a53 ("crypto: akcipher - default implementations for request callbacks")
Signed-off-by: Ignat Korchagin <ignat@cloudflare.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iommu/omap: Fix buffer overflow in debugfs

[ Upstream commit 184233a5202786b20220acd2d04ddf909ef18f29 ]

There are two issues here:

1) The "len" variable needs to be checked before the very first write.
   Otherwise if omap2_iommu_dump_ctx() with "bytes" less than 32 it is a
   buffer overflow.
2) The snprintf() function returns the number of bytes that *would* have
   been copied if there were enough space.  But we want to know the
   number of bytes which were *actually* copied so use scnprintf()
   instead.

Fixes: bd4396f09a4a ("iommu/omap: Consolidate OMAP IOMMU modules")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Robin Murphy <robin.murphy@arm.com>
Reviewed-by: Laurent Pinchart <laurent.pinchart@ideasonboard.com>
Link: https://lore.kernel.org/r/YuvYh1JbE3v+abd5@kili
Signed-off-by: Joerg Roedel <jroedel@suse.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>

cgroup/cpuset: Enable update_tasks_cpumask() on top_cpuset

[ Upstream commit ec5fbdfb99d18482619ac42605cb80fbb56068ee ]

Previously, update_tasks_cpumask() is not supposed to be called with
top cpuset. With cpuset partition that takes CPUs away from the top
cpuset, adjusting the cpus_mask of the tasks in the top cpuset is
necessary. Percpu kthreads, however, are ignored.

Fixes: ee8dde0cd2ce ("cpuset: Add new v2 cpuset.sched.partition flag")
Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

hwrng: imx-rngc - Moving IRQ handler registering after imx_rngc_irq_mask_clear()

[ Upstream commit 10a2199caf437e893d9027d97700b3c6010048b7 ]

Issue:
While servicing interrupt, if the IRQ happens to be because of a SEED_DONE
due to a previous boot stage, you end up completing the completion
prematurely, hence causing kernel to crash while booting.

Fix:
Moving IRQ handler registering after imx_rngc_irq_mask_clear()

Fixes: 1d5449445bd0 (hwrng: mx-rngc - add a driver for Freescale RNGC)
Signed-off-by: Kshitiz Varshney <kshitiz.varshney@nxp.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: hisilicon/zip - fix mismatch in get/set sgl_sge_nr

[ Upstream commit d74f9340097a881869c4c22ca376654cc2516ecc ]

KASAN reported this Bug:

[17619.659757] BUG: KASAN: global-out-of-bounds in param_get_int+0x34/0x60
[17619.673193] Read of size 4 at addr fffff01332d7ed00 by task read_all/1507958
...
[17619.698934] The buggy address belongs to the variable:
[17619.708371]  sgl_sge_nr+0x0/0xffffffffffffa300 [hisi_zip]

There is a mismatch in hisi_zip when get/set the variable sgl_sge_nr.
The type of sgl_sge_nr is u16, and get/set sgl_sge_nr by
param_get/set_int.

Replacing param_get/set_int to param_get/set_ushort can fix this bug.

Fixes: f081fda293ffb ("crypto: hisilicon - add sgl_sge_nr module param for zip")
Signed-off-by: Ye Weihua <yeweihua4@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

crypto: sahara - don't sleep when in softirq

[ Upstream commit 108586eba094b318e6a831f977f4ddcc403a15da ]

Function of sahara_aes_crypt maybe could be called by function
of crypto_skcipher_encrypt during the rx softirq, so it is not
allowed to use mutex lock.

Fixes: c0c3c89ae347 ("crypto: sahara - replace tasklets with...")
Signed-off-by: Zhengchao Shao <shaozhengchao@huawei.com>
Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc: Fix SPE Power ISA properties for e500v1 platforms

[ Upstream commit 37b9345ce7f4ab17538ea62def6f6d430f091355 ]

Commit 2eb28006431c ("powerpc/e500v2: Add Power ISA properties to comply
with ePAPR 1.1") introduced new include file e500v2_power_isa.dtsi and
should have used it for all e500v2 platforms. But apparently it was used
also for e500v1 platforms mpc8540, mpc8541, mpc8555 and mpc8560.

e500v1 cores compared to e500v2 do not support double precision floating
point SPE instructions. Hence power-isa-sp.fd should not be set on e500v1
platforms, which is in e500v2_power_isa.dtsi include file.

Fix this issue by introducing a new e500v1_power_isa.dtsi include file and
use it in all e500v1 device tree files.

Fixes: 2eb28006431c ("powerpc/e500v2: Add Power ISA properties to comply with ePAPR 1.1")
Signed-off-by: Pali Rohár <pali@kernel.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220902212103.22534-1-pali@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/64s: Fix GENERIC_CPU build flags for PPC970 / G5

[ Upstream commit 58ec7f06b74e0d6e76c4110afce367c8b5f0837d ]

Big-endian GENERIC_CPU supports 970, but builds with -mcpu=power5.
POWER5 is ISA v2.02 whereas 970 is v2.01 plus Altivec. 2.02 added
the popcntb instruction which a compiler might use.

Use -mcpu=power4.

Fixes: 471d7ff8b51b ("powerpc/64s: Remove POWER4 support")
Signed-off-by: Nicholas Piggin <npiggin@gmail.com>
Reviewed-by: Segher Boessenkool <segher@kernel.crashing.org>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220921014103.587954-1-npiggin@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/hyperv: Fix 'struct hv_enlightened_vmcs' definition

[ Upstream commit ea9da788a61e47e7ab9cbad397453e51cd82ac0d ]

Section 1.9 of TLFS v6.0b says:

"All structures are padded in such a way that fields are aligned
naturally (that is, an 8-byte field is aligned to an offset of 8 bytes
and so on)".

'struct enlightened_vmcs' has a glitch:

...
        struct {
                u32                nested_flush_hypercall:1; /*   836: 0  4 */
                u32                msr_bitmap:1;         /*   836: 1  4 */
                u32                reserved:30;          /*   836: 2  4 */
        } hv_enlightenments_control;                     /*   836     4 */
        u32                        hv_vp_id;             /*   840     4 */
        u64                        hv_vm_id;             /*   844     8 */
        u64                        partition_assist_page; /*   852     8 */
...

And the observed values in 'partition_assist_page' make no sense at
all. Fix the layout by padding the structure properly.

Fixes: 68d1eb72ee99 ("x86/hyper-v: define struct hv_enlightened_vmcs and clean field bits")
Reviewed-by: Maxim Levitsky <mlevitsk@redhat.com>
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/r/20220830133737.1539624-2-vkuznets@redhat.com
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/powernv: add missing of_node_put() in opal_export_attrs()

[ Upstream commit 71a92e99c47900cc164620948b3863382cec4f1a ]

After using 'np' returned by of_find_node_by_path(), of_node_put()
need be called to decrease the refcount.

Fixes: 11fe909d2362 ("powerpc/powernv: Add OPAL exports attributes to sysfs")
Signed-off-by: Zheng Yongjun <zhengyongjun3@huawei.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220906141703.118192-1-zhengyongjun3@huawei.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/pci_dn: Add missing of_node_put()

[ Upstream commit 110a1fcb6c4d55144d8179983a475f17a1d6f832 ]

In pci_add_device_node_info(), use of_node_put() to drop the reference
to 'parent' returned by of_get_parent() to keep refcount balance.

Fixes: cca87d303c85 ("powerpc/pci: Refactor pci_dn")
Co-authored-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Liang He <windhl@126.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Tyrel Datwyler <tyreld@linux.ibm.com>
Link: https://lore.kernel.org/r/20220701131750.240170-1-windhl@126.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/sysdev/fsl_msi: Add missing of_node_put()

[ Upstream commit def435c04ee984a5f9ed2711b2bfe946936c6a21 ]

In fsl_setup_msi_irqs(), use of_node_put() to drop the reference
returned by of_parse_phandle().

Fixes: 895d603f945ba ("powerpc/fsl_msi: add support for the fsl, msi property in PCI nodes")
Co-authored-by: Miaoqian Lin <linmq006@gmail.com>
Signed-off-by: Liang He <windhl@126.com>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://lore.kernel.org/r/20220704145233.278539-1-windhl@126.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

powerpc/math_emu/efp: Include module.h

[ Upstream commit cfe0d370e0788625ce0df3239aad07a2506c1796 ]

When building with a recent version of clang, there are a couple of
errors around the call to module_init():

  arch/powerpc/math-emu/math_efp.c:927:1: error: type specifier missing, defaults to 'int'; ISO C99 and later do not support implicit int [-Wimplicit-int]
  module_init(spe_mathemu_init);
  ^
  int
  arch/powerpc/math-emu/math_efp.c:927:13: error: a parameter list without types is only allowed in a function definition
  module_init(spe_mathemu_init);
              ^
  2 errors generated.

module_init() is a macro, which is not getting expanded because module.h
is not included in this file. Add the include so that the macro can
expand properly, clearing up the build failure.

Fixes: ac6f120369ff ("powerpc/85xx: Workaroudn e500 CPU erratum A005")
[chleroy: added fixes tag]
Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Nathan Chancellor <nathan@kernel.org>
Signed-off-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Reviewed-by: Christophe Leroy <christophe.leroy@csgroup.eu>
Link: https://lore.kernel.org/r/8403854a4c187459b2f4da3537f51227b70b9223.1662134272.git.christophe.leroy@csgroup.eu
Signed-off-by: Sasha Levin <sashal@kernel.org>

mailbox: bcm-ferxrm-mailbox: Fix error check for dma_map_sg

[ Upstream commit 6b207ce8a96a71e966831e3a13c38143ba9a73c1 ]

dma_map_sg return 0 on error, fix the error check, and return -EIO
to caller.

Fixes: dbc049eee730 ("mailbox: Add driver for Broadcom FlexRM ring manager")
Signed-off-by: Jack Wang <jinpu.wang@ionos.com>
Signed-off-by: Jassi Brar <jaswinder.singh@linaro.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: ast2600: BCLK comes from EPLL

[ Upstream commit b8c1dc9c00b252b3be853720a71b05ed451ddd9f ]

This correction was made in the u-boot SDK recently. There are no
in-tree users of this clock so the impact is minimal.

Fixes: d3d04f6c330a ("clk: Add support for AST2600 SoC")
Link: https://github.com/AspeedTech-BMC/u-boot/commit/8ad54a5ae15f27fea5e894cc2539a20d90019717
Signed-off-by: Joel Stanley <joel@jms.id.au>
Link: https://lore.kernel.org/r/20220421040426.171256-1-joel@jms.id.au
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: ti: dra7-atl: Fix reference leak in of_dra7_atl_clk_probe

[ Upstream commit 9c59a01caba26ec06fefd6ca1f22d5fd1de57d63 ]

pm_runtime_get_sync() will increment pm usage counter.
Forgetting to putting operation will result in reference leak.
Add missing pm_runtime_put_sync in some error paths.

Fixes: 9ac33b0ce81f ("CLK: TI: Driver for DRA7 ATL (Audio Tracking Logic)")
Signed-off-by: Miaoqian Lin <linmq006@gmail.com>
Link: https://lore.kernel.org/r/20220602030838.52057-1-linmq006@gmail.com
Reviewed-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: bcm2835: fix bcm2835_clock_rate_from_divisor declaration

[ Upstream commit 0b919a3728691c172312dee99ba654055ccd8c84 ]

The return value of bcm2835_clock_rate_from_divisor is always unsigned
and also all caller expect this. So fix the declaration accordingly.

Fixes: 41691b8862e2 ("clk: bcm2835: Add support for programming the audio domain clocks")
Signed-off-by: Stefan Wahren <stefan.wahren@i2se.com>
Link: https://lore.kernel.org/r/20220904141037.38816-1-stefan.wahren@i2se.com
Reviewed-by: Ivan T. Ivanov <iivanov@suse.de>
Reviewed-by: Florian Fainelli <f.fainelli@gmail.com>
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: baikal-t1: Add SATA internal ref clock buffer

[ Upstream commit 081a9b7c74eae4e12b2cb1b86720f836a8f29247 ]

It turns out the internal SATA reference clock signal will stay
unavailable for the SATA interface consumer until the buffer on it's way
is ungated. So aside with having the actual clock divider enabled we need
to ungate a buffer placed on the signal way to the SATA controller (most
likely some rudiment from the initial SoC release). Seeing the switch flag
is placed in the same register as the SATA-ref clock divider at a
non-standard ffset, let's implement it as a separate clock controller with
the set-rate propagation to the parental clock divider wrapper. As such
we'll be able to disable/enable and still change the original clock source
rate.

Fixes: 353afa3a8d2e ("clk: Add Baikal-T1 CCU Dividers driver")
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20220929225402.9696-5-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: baikal-t1: Add shared xGMAC ref/ptp clocks internal parent

[ Upstream commit e2eef312762e0b5a5a70d29fe59a245c0a3cffa0 ]

Baikal-T1 CCU reference manual says that both xGMAC reference and xGMAC
PTP clocks are generated by two different wrappers with the same constant
divider thus each producing a 156.25 MHz signal. But for some reason both
of these clock sources are gated by a single switch-flag in the CCU
registers space - CCU_SYS_XGMAC_BASE.BIT(0). In order to make the clocks
handled independently we need to define a shared parental gate so the base
clock signal would be switched off only if both of the child-clocks are
disabled.

Note the ID is intentionally set to -2 since we are going to add a one
more internal clock identifier in the next commit.

Fixes: 353afa3a8d2e ("clk: Add Baikal-T1 CCU Dividers driver")
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20220929225402.9696-4-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: baikal-t1: Fix invalid xGMAC PTP clock divider

[ Upstream commit 3c742088686ce922704aec5b11d09bcc5a396589 ]

Most likely due to copy-paste mistake the divider has been set to 10 while
according to the SoC reference manual it's supposed to be 8 thus having
PTP clock frequency of 156.25 MHz.

Fixes: 353afa3a8d2e ("clk: Add Baikal-T1 CCU Dividers driver")
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Link: https://lore.kernel.org/r/20220929225402.9696-3-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: vc5: Fix 5P49V6901 outputs disabling when enabling FOD

[ Upstream commit c388cc804016cf0f65afdc2362b120aa594ff3e6 ]

We have discovered random glitches during the system boot up procedure.
The problem investigation led us to the weird outcomes: when none of the
Renesas 5P49V6901 ports are explicitly enabled by the kernel driver, the
glitches disappeared. It was a mystery since the SoC external clock
domains were fed with different 5P49V6901 outputs. The driver code didn't
seem like bogus either. We almost despaired to find out a root cause when
the solution has been found for a more modern revision of the chip. It
turned out the 5P49V6901 clock generator stopped its output for a short
period of time during the VC5_OUT_DIV_CONTROL register writing. The same
problem was found for the 5P49V6965 revision of the chip and was
successfully fixed in commit fc336ae622df ("clk: vc5: fix output disabling
when enabling a FOD") by enabling the "bypass_sync" flag hidden inside
"Unused Factory Reserved Register". Even though the 5P49V6901 registers
description and programming guide doesn't provide any intel regarding that
flag, setting it up anyway in the officially unused register completely
eliminated the denoted glitches. Thus let's activate the functionality
submitted in commit fc336ae622df ("clk: vc5: fix output disabling when
enabling a FOD") for the Renesas 5P49V6901 chip too in order to remove the
ports implicit inter-dependency.

Fixes: dbf6b16f5683 ("clk: vc5: Add support for IDT VersaClock 5P49V6901")
Signed-off-by: Serge Semin <Sergey.Semin@baikalelectronics.ru>
Reviewed-by: Luca Ceresoli <luca@lucaceresoli.net>
Link: https://lore.kernel.org/r/20220929225402.9696-2-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

spmi: pmic-arb: correct duplicate APID to PPID mapping logic

[ Upstream commit 1f1693118c2476cb1666ad357edcf3cf48bf9b16 ]

Correct the way that duplicate PPID mappings are handled for PMIC
arbiter v5.  The final APID mapped to a given PPID should be the
one which has write owner = APPS EE, if it exists, or if not
that, then the first APID mapped to the PPID, if it exists.

Fixes: 40f318f0ed67 ("spmi: pmic-arb: add support for HW version 5")
Signed-off-by: David Collins <collinsd@codeaurora.org>
Signed-off-by: Fenglin Wu <quic_fenglinw@quicinc.com>
Link: https://lore.kernel.org/r/1655004286-11493-7-git-send-email-quic_fenglinw@quicinc.com
Signed-off-by: Stephen Boyd <sboyd@kernel.org>
Link: https://lore.kernel.org/r/20220930005019.2663064-8-sboyd@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

dmaengine: ioat: stop mod_timer from resurrecting deleted timer in __cleanup()

[ Upstream commit 898ec89dbb55b8294695ad71694a0684e62b2a73 ]

User reports observing timer event report channel halted but no error
observed in CHANERR register. The driver finished self-test and released
channel resources. Debug shows that __cleanup() can call
mod_timer() after the timer has been deleted and thus resurrect the
timer. While harmless, it causes suprious error message to be emitted.
Use mod_timer_pending() call to prevent deleted timer from being
resurrected.

Fixes: 3372de5813e4 ("dmaengine: ioatdma: removal of dma_v3.c and relevant ioat3 references")
Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Link: https://lore.kernel.org/r/166360672197.3851724.17040290563764838369.stgit@djiang5-desk3.ch.intel.com
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: mediatek: mt8183: mfgcfg: Propagate rate changes to parent

[ Upstream commit 9f94f545f258b15bfa6357eb62e1e307b712851e ]

The only clock in the MT8183 MFGCFG block feeds the GPU. Propagate its
rate change requests to its parent, so that DVFS for the GPU can work
properly.

Fixes: acddfc2c261b ("clk: mediatek: Add MT8183 clock support")
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Reviewed-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Signed-off-by: AngeloGioacchino Del Regno <angelogioacchino.delregno@collabora.com>
Link: https://lore.kernel.org/r/20220927101128.44758-3-angelogioacchino.delregno@collabora.com
Signed-off-by: Chen-Yu Tsai <wenst@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mfd: sm501: Add check for platform_driver_register()

[ Upstream commit 8325a6c24ad78b8c1acc3c42b098ee24105d68e5 ]

As platform_driver_register() can return error numbers,
it should be better to check platform_driver_register()
and deal with the exception.

Fixes: b6d6454fdb66 ("[PATCH] mfd: SM501 core driver")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Signed-off-by: Lee Jones <lee@kernel.org>
Link: https://lore.kernel.org/r/20220913091112.1739138-1-jiasheng@iscas.ac.cn
Signed-off-by: Sasha Levin <sashal@kernel.org>

mfd: fsl-imx25: Fix check for platform_get_irq() errors

[ Upstream commit 75db7907355ca5e2ff606e9dd3e86b6c3a455fe2 ]

The mx25_tsadc_remove() function assumes all non-zero returns are success
but the platform_get_irq() function returns negative on error and
positive non-zero values on success.  It never returns zero, but if it
did then treat that as a success.

Fixes: 18f773937968 ("mfd: fsl-imx25: Clean up irq settings during removal")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Reviewed-by: Martin Kaiser <martin@kaiser.cx>
Signed-off-by: Lee Jones <lee@kernel.org>
Link: https://lore.kernel.org/r/YvTfkbVQWYKMKS/t@kili
Signed-off-by: Sasha Levin <sashal@kernel.org>

mfd: lp8788: Fix an error handling path in lp8788_irq_init() and lp8788_irq_init()

[ Upstream commit 557244f6284f30613f2d61f14b579303165876c3 ]

In lp8788_irq_init(), if an error occurs after a successful
irq_domain_add_linear() call, it must be undone by a corresponding
irq_domain_remove() call.

irq_domain_remove() should also be called in lp8788_irq_exit() for the same
reason.

Fixes: eea6b7cc53aa ("mfd: Add lp8788 mfd driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Lee Jones <lee@kernel.org>
Link: https://lore.kernel.org/r/bcd5a72c9c1c383dd6324680116426e32737655a.1659261275.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Sasha Levin <sashal@kernel.org>

mfd: lp8788: Fix an error handling path in lp8788_probe()

[ Upstream commit becfdcd75126b20b8ec10066c5e85b34f8994ad5 ]

Should an error occurs in mfd_add_devices(), some resources need to be
released, as already done in the .remove() function.

Add an error handling path and a lp8788_irq_exit() call to undo a previous
lp8788_irq_init().

Fixes: eea6b7cc53aa ("mfd: Add lp8788 mfd driver")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Lee Jones <lee@kernel.org>
Link: https://lore.kernel.org/r/18398722da9df9490722d853e4797350189ae79b.1659261275.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Sasha Levin <sashal@kernel.org>

mfd: fsl-imx25: Fix an error handling path in mx25_tsadc_setup_irq()

[ Upstream commit 3fa9e4cfb55da512ebfd57336fde468830719298 ]

If devm_of_platform_populate() fails, some resources need to be
released.

Introduce a mx25_tsadc_unset_irq() function that undoes
mx25_tsadc_setup_irq() and call it both from the new error handling path
of the probe and in the remove function.

Fixes: a55196eff6d6 ("mfd: fsl-imx25: Use devm_of_platform_populate()")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Lee Jones <lee@kernel.org>
Link: https://lore.kernel.org/r/d404e04828fc06bcfddf81f9f3e9b4babbe35415.1659269156.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Sasha Levin <sashal@kernel.org>

mfd: intel_soc_pmic: Fix an error handling path in intel_soc_pmic_i2c_probe()

[ Upstream commit 48749cabba109397b4e7dd556e85718ec0ec114d ]

The commit in Fixes: has added a pwm_add_table() call in the probe() and
a pwm_remove_table() call in the remove(), but forget to update the error
handling path of the probe.

Add the missing pwm_remove_table() call.

Fixes: a3aa9a93df9f ("mfd: intel_soc_pmic_core: ADD PWM lookup table for CRC PMIC based PWM")
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Reviewed-by: Hans de Goede <hdegoede@redhat.com>
Signed-off-by: Lee Jones <lee@kernel.org>
Link: https://lore.kernel.org/r/20220801114211.36267-1-andriy.shevchenko@linux.intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

fsi: core: Check error number after calling ida_simple_get

[ Upstream commit 35af9fb49bc5c6d61ef70b501c3a56fe161cce3e ]

If allocation fails, the ida_simple_get() will return error number.
So master->idx could be error number and be used in dev_set_name().
Therefore, it should be better to check it and return error if fails,
like the ida_simple_get() in __fsi_get_new_minor().

Fixes: 09aecfab93b8 ("drivers/fsi: Add fsi master definition")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Reviewed-by: Eddie James <eajames@linux.ibm.com>
Link: https://lore.kernel.org/r/20220111073411.614138-1-jiasheng@iscas.ac.cn
Signed-off-by: Joel Stanley <joel@jms.id.au>
Signed-off-by: Sasha Levin <sashal@kernel.org>

clk: qcom: apss-ipq6018: mark apcs_alias0_core_clk as critical

[ Upstream commit 86e78995c93ee182433f965babfccd48417d4dcf ]

While fixing up the driver I noticed that my IPQ8074 board was hanging
after CPUFreq switched the frequency during boot, WDT would eventually
reset it.

So mark apcs_alias0_core_clk as critical since its the clock feeding the
CPU cluster and must never be disabled.

Fixes: 5e77b4ef1b19 ("clk: qcom: Add ipq6018 apss clock controller")
Signed-off-by: Robert Marko <robimarko@gmail.com>
Reviewed-by: Dmitry Baryshkov <dmitry.baryshkov@linaro.org>
Signed-off-by: Bjorn Andersson <andersson@kernel.org>
Link: https://lore.kernel.org/r/20220818220628.339366-3-robimarko@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: iscsi: iscsi_tcp: Fix null-ptr-deref while calling getpeername()

[ Upstream commit 57569c37f0add1b6489e1a1563c71519daf732cf ]

Fix a NULL pointer crash that occurs when we are freeing the socket at the
same time we access it via sysfs.

The problem is that:

 1. iscsi_sw_tcp_conn_get_param() and iscsi_sw_tcp_host_get_param() take
    the frwd_lock and do sock_hold() then drop the frwd_lock. sock_hold()
    does a get on the "struct sock".

 2. iscsi_sw_tcp_release_conn() does sockfd_put() which does the last put
    on the "struct socket" and that does __sock_release() which sets the
    sock->ops to NULL.

 3. iscsi_sw_tcp_conn_get_param() and iscsi_sw_tcp_host_get_param() then
    call kernel_getpeername() which accesses the NULL sock->ops.

Above we do a get on the "struct sock", but we needed a get on the "struct
socket". Originally, we just held the frwd_lock the entire time but in
commit bcf3a2953d36 ("scsi: iscsi: iscsi_tcp: Avoid holding spinlock while
calling getpeername()") we switched to refcount based because the network
layer changed and started taking a mutex in that path, so we could no
longer hold the frwd_lock.

Instead of trying to maintain multiple refcounts, this just has us use a
mutex for accessing the socket in the interface code paths.

Link: https://lore.kernel.org/r/20220907221700.10302-1-michael.christie@oracle.com
Fixes: bcf3a2953d36 ("scsi: iscsi: iscsi_tcp: Avoid holding spinlock while calling getpeername()")
Signed-off-by: Mike Christie <michael.christie@oracle.com>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

scsi: libsas: Fix use-after-free bug in smp_execute_task_sg()

[ Upstream commit 46ba53c30666717cb06c2b3c5d896301cd00d0c0 ]

When executing SMP task failed, the smp_execute_task_sg() calls del_timer()
to delete "slow_task->timer". However, if the timer handler
sas_task_internal_timedout() is running, the del_timer() in
smp_execute_task_sg() will not stop it and a UAF will happen. The process
is shown below:

      (thread 1)               |        (thread 2)
smp_execute_task_sg()          | sas_task_internal_timedout()
 ...                           |
 del_timer()                   |
 ...                           |  ...
 sas_free_task(task)           |
  kfree(task->slow_task) //FREE|
                               |  task->slow_task->... //USE

Fix by calling del_timer_sync() in smp_execute_task_sg(), which makes sure
the timer handler have finished before the "task->slow_task" is
deallocated.

Link: https://lore.kernel.org/r/20220920144213.10536-1-duoming@zju.edu.cn
Fixes: 2908d778ab3e ("[SCSI] aic94xx: new driver")
Reviewed-by: Jason Yan <yanaijie@huawei.com>
Signed-off-by: Duoming Zhou <duoming@zju.edu.cn>
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

serial: 8250: Fix restoring termios speed after suspend

[ Upstream commit 379a33786d489ab81885ff0b3935cfeb36137fea ]

Since commit edc6afc54968 ("tty: switch to ktermios and new framework")
termios speed is no longer stored only in c_cflag member but also in new
additional c_ispeed and c_ospeed members. If BOTHER flag is set in c_cflag
then termios speed is stored only in these new members.

Since commit 027b57170bf8 ("serial: core: Fix initializing and restoring
termios speed") termios speed is available also in struct console.

So properly restore also c_ispeed and c_ospeed members after suspend to fix
restoring termios speed which is not represented by Bnnn constant.

Fixes: 4516d50aabed ("serial: 8250: Use canary to restart console after suspend")
Signed-off-by: Pali Rohár <pali@kernel.org>
Link: https://lore.kernel.org/r/20220924104324.4035-1-pali@kernel.org
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

firmware: google: Test spinlock on panic path to avoid lockups

[ Upstream commit 3e081438b8e639cc76ef1a5ce0c1bd8a154082c7 ]

Currently the gsmi driver registers a panic notifier as well as
reboot and die notifiers. The callbacks registered are called in
atomic and very limited context - for instance, panic disables
preemption and local IRQs, also all secondary CPUs (not executing
the panic path) are shutdown.

With that said, taking a spinlock in this scenario is a dangerous
invitation for lockup scenarios. So, fix that by checking if the
spinlock is free to acquire in the panic notifier callback - if not,
bail-out and avoid a potential hang.

Fixes: 74c5b31c6618 ("driver: Google EFI SMI")
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: David Gow <davidgow@google.com>
Cc: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Cc: Julius Werner <jwerner@chromium.org>
Cc: Petr Mladek <pmladek@suse.com>
Reviewed-by: Evan Green <evgreen@chromium.org>
Signed-off-by: Guilherme G. Piccoli <gpiccoli@igalia.com>
Link: https://lore.kernel.org/r/20220909200755.189679-1-gpiccoli@igalia.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

staging: vt6655: fix some erroneous memory clean-up loops

[ Upstream commit 2a2db520e3ca5aafba7c211abfd397666c9b5f9d ]

In some initialization functions of this driver, memory is allocated with
'i' acting as an index variable and increasing from 0. The commit in
"Fixes" introduces some clean-up codes in case of allocation failure,
which free memory in reverse order with 'i' decreasing to 0. However,
there are some problems:
  - The case i=0 is left out. Thus memory is leaked.
  - In case memory allocation fails right from the start, the memory
    freeing loops will start with i=-1 and invalid memory locations will
    be accessed.

One of these loops has been fixed in commit c8ff91535880 ("staging:
vt6655: fix potential memory leak"). Fix the remaining erroneous loops.

Link: https://lore.kernel.org/linux-staging/Yx9H1zSpxmNqx6Xc@kadam/
Fixes: 5341ee0adb17 ("staging: vt6655: check for memory allocation failures")
Reported-by: Dan Carpenter <dan.carpenter@oracle.com>
Tested-by: Philipp Hortmann <philipp.g.hortmann@gmail.com>
Signed-off-by: Nam Cao <namcaov@gmail.com>
Link: https://lore.kernel.org/r/20220912170429.29852-1-namcaov@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

phy: qualcomm: call clk_disable_unprepare in the error handling

[ Upstream commit c3966ced8eb8dc53b6c8d7f97d32cc8a2107d83e ]

Smatch reports the following error:

drivers/phy/qualcomm/phy-qcom-usb-hsic.c:82 qcom_usb_hsic_phy_power_on()
warn: 'uphy->cal_clk' from clk_prepare_enable() not released on lines:
58.
drivers/phy/qualcomm/phy-qcom-usb-hsic.c:82 qcom_usb_hsic_phy_power_on()
warn: 'uphy->cal_sleep_clk' from clk_prepare_enable() not released on
lines: 58.
drivers/phy/qualcomm/phy-qcom-usb-hsic.c:82 qcom_usb_hsic_phy_power_on()
warn: 'uphy->phy_clk' from clk_prepare_enable() not released on lines:
58.

Fix this by calling proper clk_disable_unprepare calls.

Fixes: 0b56e9a7e835 ("phy: Group vendor specific phy drivers")
Signed-off-by: Dongliang Mu <mudongliangabcd@gmail.com>
Reviewed-by: Neil Armstrong <neil.armstrong@linaro.org>
Link: https://lore.kernel.org/r/20220914051334.69282-1-dzm91@hust.edu.cn
Signed-off-by: Vinod Koul <vkoul@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

tty: serial: fsl_lpuart: disable dma rx/tx use flags in lpuart_dma_shutdown

[ Upstream commit 316ae95c175a7d770d1bfe4c011192712f57aa4a ]

lpuart_dma_shutdown tears down lpuart dma, but lpuart_flush_buffer can
still occur which in turn tries to access dma apis if lpuart_dma_tx_use
flag is true. At this point since dma is torn down, these dma apis can
abort. Set lpuart_dma_tx_use and the corresponding rx flag
lpuart_dma_rx_use to false in lpuart_dma_shutdown so that dmas are not
accessed after they are relinquished.

Otherwise, when try to kill btattach, kernel may panic. This patch may
fix this issue.
root@imx8ulpevk:~# btattach -B /dev/ttyLP2 -S 115200
^C[   90.182296] Internal error: synchronous external abort: 96000210 [#1] PREEMPT SMP
[   90.189806] Modules linked in: moal(O) mlan(O)
[   90.194258] CPU: 0 PID: 503 Comm: btattach Tainted: G           O      5.15.32-06136-g34eecdf2f9e4 #37
[   90.203554] Hardware name: NXP i.MX8ULP 9X9 EVK (DT)
[   90.208513] pstate: 600000c5 (nZCv daIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
[   90.215470] pc : fsl_edma3_disable_request+0x8/0x60
[   90.220358] lr : fsl_edma3_terminate_all+0x34/0x20c
[   90.225237] sp : ffff800013f0bac0
[   90.228548] x29: ffff800013f0bac0 x28: 0000000000000001 x27: ffff000008404800
[   90.235681] x26: ffff000008404960 x25: ffff000008404a08 x24: ffff000008404a00
[   90.242813] x23: ffff000008404a60 x22: 0000000000000002 x21: 0000000000000000
[   90.249946] x20: ffff800013f0baf8 x19: ffff00000559c800 x18: 0000000000000000
[   90.257078] x17: 0000000000000000 x16: 0000000000000000 x15: 0000000000000000
[   90.264211] x14: 0000000000000003 x13: 0000000000000000 x12: 0000000000000040
[   90.271344] x11: ffff00000600c248 x10: ffff800013f0bb10 x9 : ffff000057bcb090
[   90.278477] x8 : fffffc0000241a08 x7 : ffff00000534ee00 x6 : ffff000008404804
[   90.285609] x5 : 0000000000000000 x4 : 0000000000000000 x3 : ffff0000055b3480
[   90.292742] x2 : ffff8000135c0000 x1 : ffff00000534ee00 x0 : ffff00000559c800
[   90.299876] Call trace:
[   90.302321]  fsl_edma3_disable_request+0x8/0x60
[   90.306851]  lpuart_flush_buffer+0x40/0x160
[   90.311037]  uart_flush_buffer+0x88/0x120
[   90.315050]  tty_driver_flush_buffer+0x20/0x30
[   90.319496]  hci_uart_flush+0x44/0x90
[   90.323162]  +0x34/0x12c
[   90.327253]  tty_ldisc_close+0x38/0x70
[   90.331005]  tty_ldisc_release+0xa8/0x190
[   90.335018]  tty_release_struct+0x24/0x8c
[   90.339022]  tty_release+0x3ec/0x4c0
[   90.342593]  __fput+0x70/0x234
[   90.345652]  ____fput+0x14/0x20
[   90.348790]  task_work_run+0x84/0x17c
[   90.352455]  do_exit+0x310/0x96c
[   90.355688]  do_group_exit+0x3c/0xa0
[   90.359259]  __arm64_sys_exit_group+0x1c/0x20
[   90.363609]  invoke_syscall+0x48/0x114
[   90.367362]  el0_svc_common.constprop.0+0xd4/0xfc
[   90.372068]  do_el0_svc+0x2c/0x94
[   90.375379]  el0_svc+0x28/0x80
[   90.378438]  el0t_64_sync_handler+0xa8/0x130
[   90.382711]  el0t_64_sync+0x1a0/0x1a4
[   90.386376] Code: 17ffffda d503201f d503233f f9409802 (b9400041)
[   90.392467] ---[ end trace 2f60524b4a43f1f6 ]---
[   90.397073] note: btattach[503] exited with preempt_count 1
[   90.402636] Fixing recursive fault but reboot is needed!

Fixes: 6250cc30c4c4 ("tty: serial: fsl_lpuart: Use scatter/gather DMA for Tx")
Reviewed-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Signed-off-by: Thara Gopinath <tgopinath@microsoft.com>
Signed-off-by: Sherry Sun <sherry.sun@nxp.com>
Link: https://lore.kernel.org/r/20220920111703.1532-1-sherry.sun@nxp.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

serial: 8250: Toggle IER bits on only after irq has been set up

[ Upstream commit 039d4926379b1d1c17b51cf21c500a5eed86899e ]

Invoking TIOCVHANGUP on 8250_mid port on Ice Lake-D and then reopening
the port triggers these faults during serial8250_do_startup():

  DMAR: DRHD: handling fault status reg 3
  DMAR: [DMA Write NO_PASID] Request device [00:1a.0] fault addr 0x0 [fault reason 0x05] PTE Write access is not set

If the IRQ hasn't been set up yet, the UART will have zeroes in its MSI
address/data registers. Disabling the IRQ at the interrupt controller
won't stop the UART from performing a DMA write to the address programmed
in its MSI address register (zero) when it wants to signal an interrupt.

The UARTs (in Ice Lake-D) implement PCI 2.1 style MSI without masking
capability, so there is no way to mask the interrupt at the source PCI
function level, except disabling the MSI capability entirely, but that
would cause it to fall back to INTx# assertion, and the PCI specification
prohibits disabling the MSI capability as a way to mask a function's
interrupt service request.

The MSI address register is zeroed by the hangup as the irq is freed.
The interrupt is signalled during serial8250_do_startup() performing a
THRE test that temporarily toggles THRI in IER. The THRE test currently
occurs before UART's irq (and MSI address) is properly set up.

Refactor serial8250_do_startup() such that irq is set up before the
THRE test. The current irq setup code is intermixed with the timer
setup code. As THRE test must be performed prior to the timer setup,
extract it into own function and call it only after the THRE test.

The ->setup_timer() needs to be part of the struct uart_8250_ops in
order to not create circular dependency between 8250 and 8250_base
modules.

Fixes: 40b36daad0ac ("[PATCH] 8250 UART backup timer")
Reported-by: Lennert Buytenhek <buytenh@arista.com>
Tested-by: Lennert Buytenhek <buytenh@arista.com>
Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Ilpo Järvinen <ilpo.jarvinen@linux.intel.com>
Link: https://lore.kernel.org/r/20220922070005.2965-1-ilpo.jarvinen@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

serial: 8250: Add an empty line and remove some useless {}

[ Upstream commit 26f7591632d74f637f346f5d642d8ebe6b433fc9 ]

This fixes the following checkpatch.pl warnings:
   WARNING: Missing a blank line after declarations
   WARNING: braces {} are not necessary for any arm of this statement

Reviewed-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
Signed-off-by: Christophe JAILLET <christophe.jaillet@wanadoo.fr>
Link: https://lore.kernel.org/r/257ffd691b4a062ad017333c9430d69da6dbd29a.1619594713.git.christophe.jaillet@wanadoo.fr
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Stable-dep-of: 039d4926379b ("serial: 8250: Toggle IER bits on only after irq has been set up")
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers: serial: jsm: fix some leaks in probe

[ Upstream commit 1d5859ef229e381f4db38dce8ed58e4bf862006b ]

This error path needs to unwind instead of just returning directly.

Fixes: 03a8482c17dd ("drivers: serial: jsm: Enable support for Digi Classic adapters")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Link: https://lore.kernel.org/r/YyxFh1+lOeZ9WfKO@kili
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

usb: gadget: function: fix dangling pnp_string in f_printer.c

[ Upstream commit 24b7ba2f88e04800b54d462f376512e8c41b8a3c ]

When opts->pnp_string is changed with configfs, new memory is allocated for
the string. It does not, however, update dev->pnp_string, even though the
memory is freed. When rquesting the string, the host then gets old or
corrupted data rather than the new string. The ieee 1284 id string should
be allowed to change while the device is connected.

The bug was introduced in commit fdc01cc286be ("usb: gadget: printer:
Remove pnp_string static buffer"), which changed opts->pnp_string from a
char[] to a char*.
This patch changes dev->pnp_string from a char* to a char** pointing to
opts->pnp_string.

Fixes: fdc01cc286be ("usb: gadget: printer: Remove pnp_string static buffer")
Signed-off-by: Albert Briscoe <albertsbriscoe@gmail.com>
Link: https://lore.kernel.org/r/20220911223753.20417-1-albertsbriscoe@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

xhci: Don't show warning for reinit on known broken suspend

[ Upstream commit 484d6f7aa3283d082c87654b7fe7a7f725423dfb ]

commit 8b328f8002bc ("xhci: re-initialize the HC during resume if HCE was
set") introduced a new warning message when the host controller error
was set and re-initializing.

This is expected behavior on some designs which already set
`xhci->broken_suspend` so the new warning is alarming to some users.

Modify the code to only show the warning if this was a surprising behavior
to the XHCI driver.

Link: https://bugzilla.kernel.org/show_bug.cgi?id=216470
Fixes: 8b328f8002bc ("xhci: re-initialize the HC during resume if HCE was set")
Reported-by: Artem S. Tashkinov <aros@gmx.com>
Signed-off-by: Mario Limonciello <mario.limonciello@amd.com>
Signed-off-by: Mathias Nyman <mathias.nyman@linux.intel.com>
Link: https://lore.kernel.org/r/20220921123450.671459-4-mathias.nyman@linux.intel.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

IB: Set IOVA/LENGTH on IB_MR in core/uverbs layers

[ Upstream commit 241f9a27e0fc0eaf23e3d52c8450f10648cd11f1 ]

Set 'iova' and 'length' on ib_mr in ib_uverbs and ib_core layers to let all
drivers have the members filled. Also, this commit removes redundancy in
the respective drivers.

Previously, commit 04c0a5fcfcf65 ("IB/uverbs: Set IOVA on IB MR in uverbs
layer") changed to set 'iova', but seems to have missed 'length' and the
ib_core layer at that time.

Fixes: 04c0a5fcfcf65 ("IB/uverbs: Set IOVA on IB MR in uverbs layer")
Signed-off-by: Daisuke Matsuda <matsuda-daisuke@fujitsu.com>
Link: https://lore.kernel.org/r/20220921080844.1616883-1-matsuda-daisuke@fujitsu.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

RDMA/cm: Use SLID in the work completion as the DLID in responder side

[ Upstream commit b7d95040c13f61a4a6a859c5355faf583eff9658 ]

The responder should always use WC's SLID as the dlid, to follow the
IB SPEC section "13.5.4.2 COMMON RESPONSE ACTIONS":
A responder always takes the following actions in constructing a
response packet:
- The SLID of the received packet is used as the DLID in the response
  packet.

Fixes: ac3a949fb2ff ("IB/CM: Set appropriate slid and dlid when handling CM request")
Signed-off-by: Mark Zhang <markzhang@nvidia.com>
Reviewed-by: Mark Bloch <mbloch@nvidia.com>
Link: https://lore.kernel.org/r/cd17c240231e059d2fc07c17dfe555d548b917eb.1662631201.git.leonro@nvidia.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

md/raid5: Ensure stripe_fill happens on non-read IO with journal

[ Upstream commit e2eed85bc75138a9eeb63863d20f8904ac42a577 ]

When doing degrade/recover tests using the journal a kernel BUG
is hit at drivers/md/raid5.c:4381 in handle_parity_checks5():

  BUG_ON(!test_bit(R5_UPTODATE, &dev->flags));

This was found to occur because handle_stripe_fill() was skipped
for stripes in the journal due to a condition in that function.
Thus blocks were not fetched and R5_UPTODATE was not set when
the code reached handle_parity_checks5().

To fix this, don't skip handle_stripe_fill() unless the stripe is
for read.

Fixes: 07e83364845e ("md/r5cache: shift complex rmw from read path to write path")
Link: https://lore.kernel.org/linux-raid/e05c4239-41a9-d2f7-3cfa-4aa9d2cea8c1@deltatee.com/
Suggested-by: Song Liu <song@kernel.org>
Signed-off-by: Logan Gunthorpe <logang@deltatee.com>
Signed-off-by: Song Liu <song@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

md: Replace snprintf with scnprintf

[ Upstream commit 1727fd5015d8f93474148f94e34cda5aa6ad4a43 ]

Current code produces a warning as shown below when total characters
in the constituent block device names plus the slashes exceeds 200.
snprintf() returns the number of characters generated from the given
input, which could cause the expression “200 – len” to wrap around
to a large positive number. Fix this by using scnprintf() instead,
which returns the actual number of characters written into the buffer.

[ 1513.267938] ------------[ cut here ]------------
[ 1513.267943] WARNING: CPU: 15 PID: 37247 at <snip>/lib/vsprintf.c:2509 vsnprintf+0x2c8/0x510
[ 1513.267944] Modules linked in:  <snip>
[ 1513.267969] CPU: 15 PID: 37247 Comm: mdadm Not tainted 5.4.0-1085-azure #90~18.04.1-Ubuntu
[ 1513.267969] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS Hyper-V UEFI Release v4.1 05/09/2022
[ 1513.267971] RIP: 0010:vsnprintf+0x2c8/0x510
<-snip->
[ 1513.267982] Call Trace:
[ 1513.267986]  snprintf+0x45/0x70
[ 1513.267990]  ? disk_name+0x71/0xa0
[ 1513.267993]  dump_zones+0x114/0x240 [raid0]
[ 1513.267996]  ? _cond_resched+0x19/0x40
[ 1513.267998]  raid0_run+0x19e/0x270 [raid0]
[ 1513.268000]  md_run+0x5e0/0xc50
[ 1513.268003]  ? security_capable+0x3f/0x60
[ 1513.268005]  do_md_run+0x19/0x110
[ 1513.268006]  md_ioctl+0x195e/0x1f90
[ 1513.268007]  blkdev_ioctl+0x91f/0x9f0
[ 1513.268010]  block_ioctl+0x3d/0x50
[ 1513.268012]  do_vfs_ioctl+0xa9/0x640
[ 1513.268014]  ? __fput+0x162/0x260
[ 1513.268016]  ksys_ioctl+0x75/0x80
[ 1513.268017]  __x64_sys_ioctl+0x1a/0x20
[ 1513.268019]  do_syscall_64+0x5e/0x200
[ 1513.268021]  entry_SYSCALL_64_after_hwframe+0x44/0xa9

Fixes: 766038846e875 ("md/raid0: replace printk() with pr_*()")
Reviewed-by: Michael Kelley <mikelley@microsoft.com>
Acked-by: Guoqing Jiang <guoqing.jiang@linux.dev>
Signed-off-by: Saurabh Sengar <ssengar@linux.microsoft.com>
Signed-off-by: Song Liu <song@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

mtd: rawnand: meson: fix bit map use in meson_nfc_ecc_correct()

[ Upstream commit 3e4ad3212cf22687410b1e8f4e68feec50646113 ]

The meson_nfc_ecc_correct() function accidentally does a right shift
instead of a left shift so it only works for BIT(0).  Also use
BIT_ULL() because "correct_bitmap" is a u64 and we want to avoid
shift wrapping bugs.

Fixes: 8fae856c5350 ("mtd: rawnand: meson: add support for Amlogic NAND flash controller")
Signed-off-by: Dan Carpenter <dan.carpenter@oracle.com>
Acked-by: Liang Yang <liang.yang@amlogic.com>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/YuI2zF1hP65+LE7r@kili
Signed-off-by: Sasha Levin <sashal@kernel.org>

ata: fix ata_id_has_dipm()

[ Upstream commit 630624cb1b5826d753ac8e01a0e42de43d66dedf ]

ACS-5 section
7.13.6.36 Word 78: Serial ATA features supported
states that:

If word 76 is not 0000h or FFFFh, word 78 reports the features supported
by the device. If this word is not supported, the word shall be cleared
to zero.

(This text also exists in really old ACS standards, e.g. ACS-3.)

The problem with ata_id_has_dipm() is that the while it performs a
check against 0 and 0xffff, it performs the check against
ATA_ID_FEATURE_SUPP (word 78), the same word where the feature bit
is stored.

Fix this by performing the check against ATA_ID_SATA_CAPABILITY
(word 76), like required by the spec. The feature bit check itself
is of course still performed against ATA_ID_FEATURE_SUPP (word 78).

Additionally, move the macro to the other ATA_ID_FEATURE_SUPP macros
(which already have this check), thus making it more likely that the
next ATA_ID_FEATURE_SUPP macro that is added will include this check.

Fixes: ca77329fb713 ("[libata] Link power management infrastructure")
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ata: fix ata_id_has_ncq_autosense()

[ Upstream commit a5fb6bf853148974dbde092ec1bde553bea5e49f ]

ACS-5 section
7.13.6.36 Word 78: Serial ATA features supported
states that:

If word 76 is not 0000h or FFFFh, word 78 reports the features supported
by the device. If this word is not supported, the word shall be cleared
to zero.

(This text also exists in really old ACS standards, e.g. ACS-3.)

Additionally, move the macro to the other ATA_ID_FEATURE_SUPP macros
(which already have this check), thus making it more likely that the
next ATA_ID_FEATURE_SUPP macro that is added will include this check.

Fixes: 5b01e4b9efa0 ("libata: Implement NCQ autosense")
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ata: fix ata_id_has_devslp()

[ Upstream commit 9c6e09a434e1317e09b78b3b69cd384022ec9a03 ]

ACS-5 section
7.13.6.36 Word 78: Serial ATA features supported
states that:

If word 76 is not 0000h or FFFFh, word 78 reports the features supported
by the device. If this word is not supported, the word shall be cleared
to zero.

(This text also exists in really old ACS standards, e.g. ACS-3.)

Additionally, move the macro to the other ATA_ID_FEATURE_SUPP macros
(which already have this check), thus making it more likely that the
next ATA_ID_FEATURE_SUPP macro that is added will include this check.

Fixes: 65fe1f0f66a5 ("ahci: implement aggressive SATA device sleep support")
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ata: fix ata_id_sense_reporting_enabled() and ata_id_has_sense_reporting()

[ Upstream commit 690aa8c3ae308bc696ec8b1b357b995193927083 ]

ACS-5 section
7.13.6.41 Words 85..87, 120: Commands and feature sets supported or enabled
states that:

If bit 15 of word 86 is set to one, bit 14 of word 119 is set to one,
and bit 15 of word 119 is cleared to zero, then word 119 is valid.

If bit 15 of word 86 is set to one, bit 14 of word 120 is set to one,
and bit 15 of word 120 is cleared to zero, then word 120 is valid.

(This text also exists in really old ACS standards, e.g. ACS-3.)

Currently, ata_id_sense_reporting_enabled() and
ata_id_has_sense_reporting() both check bit 15 of word 86,
but neither of them check that bit 14 of word 119 is set to one,
or that bit 15 of word 119 is cleared to zero.

Additionally, make ata_id_sense_reporting_enabled() return false
if !ata_id_has_sense_reporting(), similar to how e.g.
ata_id_flush_ext_enabled() returns false if !ata_id_has_flush_ext().

Fixes: e87fd28cf9a2 ("libata: Implement support for sense data reporting")
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Damien Le Moal <damien.lemoal@opensource.wdc.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

RDMA/siw: Always consume all skbuf data in sk_data_ready() upcall.

[ Upstream commit 754209850df8367c954ac1de7671c7430b1f342c ]

For header and trailer/padding processing, siw did not consume new
skb data until minimum amount present to fill current header or trailer
structure, including potential payload padding. Not consuming any
data during upcall may cause a receive stall, since tcp_read_sock()
is not upcalling again if no new data arrive.
A NFSoRDMA client got stuck at RDMA Write reception of unaligned
payload, if the current skb did contain only the expected 3 padding
bytes, but not the 4 bytes CRC trailer. Expecting 4 more bytes already
arrived in another skb, and not consuming those 3 bytes in the current
upcall left the Write incomplete, waiting for the CRC forever.

Fixes: 8b6a361b8c48 ("rdma/siw: receive path")
Reported-by: Olga Kornievskaia <kolga@netapp.com>
Tested-by: Olga Kornievskaia <kolga@netapp.com>
Signed-off-by: Bernard Metzler <bmt@zurich.ibm.com>
Link: https://lore.kernel.org/r/20220920081202.223629-1-bmt@zurich.ibm.com
Signed-off-by: Leon Romanovsky <leon@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

mtd: rawnand: fsl_elbc: Fix none ECC mode

[ Upstream commit 049e43b9fd8fd2966940485da163d67e96ee3fea ]

Commit f6424c22aa36 ("mtd: rawnand: fsl_elbc: Make SW ECC work") added
support for specifying ECC mode via DTS and skipping autodetection.

But it broke explicit specification of HW ECC mode in DTS as correct
settings for HW ECC mode are applied only when NONE mode or nothing was
specified in DTS file.

Also it started aliasing NONE mode to be same as when ECC mode was not
specified and disallowed usage of ON_DIE mode.

Fix all these issues. Use autodetection of ECC mode only in case when mode
was really not specified in DTS file by checking that ecc value is invalid.
Set HW ECC settings either when HW ECC was specified in DTS or it was
autodetected. And do not fail when ON_DIE mode is set.

Fixes: f6424c22aa36 ("mtd: rawnand: fsl_elbc: Make SW ECC work")
Signed-off-by: Pali Rohár <pali@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Reviewed-by: Marek Behún <kabel@kernel.org>
Signed-off-by: Miquel Raynal <miquel.raynal@bootlin.com>
Link: https://lore.kernel.org/linux-mtd/20220707184328.3845-1-pali@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>