udp: move udp->accept_udp_{l4|fraglist} to udp->udp_flags

[ Upstream commit f5f52f0884a595ff99ab1a608643fe4025fca2d5 ]

These are read locklessly, move them to udp_flags to fix data-races.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Stable-dep-of: 70a36f571362 ("udp: annotate data-races around udp->encap_type")
Signed-off-by: Sasha Levin <sashal@kernel.org>

udp: add missing WRITE_ONCE() around up->encap_rcv

[ Upstream commit 6d5a12eb91224d707f8691dccb40a5719fe5466d ]

UDP_ENCAP_ESPINUDP_NON_IKE setsockopt() writes over up->encap_rcv
while other cpus read it.

Fixes: 067b207b281d ("[UDP]: Cleanup UDP encapsulation code")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

udp: move udp->gro_enabled to udp->udp_flags

[ Upstream commit e1dc0615c6b08ef36414f08c011965b8fb56198b ]

syzbot reported that udp->gro_enabled can be read locklessly.
Use one atomic bit from udp->udp_flags.

Fixes: e20cf8d3f1f7 ("udp: implement GRO for plain UDP sockets.")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

udp: move udp->no_check6_rx to udp->udp_flags

[ Upstream commit bcbc1b1de884647aa0318bf74eb7f293d72a1e40 ]

syzbot reported that udp->no_check6_rx can be read locklessly.
Use one atomic bit from udp->udp_flags.

Fixes: 1c19448c9ba6 ("net: Make enabling of zero UDP6 csums more restrictive")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

udp: move udp->no_check6_tx to udp->udp_flags

[ Upstream commit a0002127cd746fcaa182ad3386ef6931c37f3bda ]

syzbot reported that udp->no_check6_tx can be read locklessly.
Use one atomic bit from udp->udp_flags

Fixes: 1c19448c9ba6 ("net: Make enabling of zero UDP6 csums more restrictive")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

udp: introduce udp->udp_flags

[ Upstream commit 81b36803ac139827538ac5ce4028e750a3c53f53 ]

According to syzbot, it is time to use proper atomic flags
for various UDP flags.

Add udp_flags field, and convert udp->corkflag to first
bit in it.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Willem de Bruijn <willemb@google.com>
Signed-off-by: Paolo Abeni <pabeni@redhat.com>
Stable-dep-of: a0002127cd74 ("udp: move udp->no_check6_tx to udp->udp_flags")
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: cfg80211: fix kernel-doc for wiphy_delayed_work_flush()

[ Upstream commit 8c73d5248dcf112611654bcd32352dc330b02397 ]

Clearly, there's no space in the function name, not sure how
that could've happened. Put the underscore that it should be.

Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Fixes: 56cfb8ce1f7f ("wifi: cfg80211: add flush functions for wiphy work")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

bpf, x64: Fix tailcall infinite loop

[ Upstream commit 2b5dcb31a19a2e0acd869b12c9db9b2d696ef544 ]

From commit ebf7d1f508a73871 ("bpf, x64: rework pro/epilogue and tailcall
handling in JIT"), the tailcall on x64 works better than before.

From commit e411901c0b775a3a ("bpf: allow for tailcalls in BPF subprograms
for x64 JIT"), tailcall is able to run in BPF subprograms on x64.

From commit 5b92a28aae4dd0f8 ("bpf: Support attaching tracing BPF program
to other BPF programs"), BPF program is able to trace other BPF programs.

How about combining them all together?

1. FENTRY/FEXIT on a BPF subprogram.
2. A tailcall runs in the BPF subprogram.
3. The tailcall calls the subprogram's caller.

As a result, a tailcall infinite loop comes up. And the loop would halt
the machine.

As we know, in tail call context, the tail_call_cnt propagates by stack
and rax register between BPF subprograms. So do in trampolines.

Fixes: ebf7d1f508a7 ("bpf, x64: rework pro/epilogue and tailcall handling in JIT")
Fixes: e411901c0b77 ("bpf: allow for tailcalls in BPF subprograms for x64 JIT")
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Signed-off-by: Leon Hwang <hffilwlqm@gmail.com>
Link: https://lore.kernel.org/r/20230912150442.2009-3-hffilwlqm@gmail.com
Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/bpf: Correct map_fd to data_fd in tailcalls

[ Upstream commit 96daa9874211d5497aa70fa409b67afc29f0cb86 ]

Get and check data_fd. It should not check map_fd again.

Meanwhile, correct some 'return' to 'goto out'.

Thank the suggestion from Maciej in "bpf, x64: Fix tailcall infinite
loop"[0] discussions.

[0] https://lore.kernel.org/bpf/e496aef8-1f80-0f8e-dcdd-25a8c300319a@gmail.com/T/#m7d3b601066ba66400d436b7e7579b2df4a101033

Fixes: 79d49ba048ec ("bpf, testing: Add various tail call test cases")
Fixes: 3b0379111197 ("selftests/bpf: Add tailcall_bpf2bpf tests")
Fixes: 5e0b0a4c52d3 ("selftests/bpf: Test tail call counting with bpf2bpf and data on stack")
Signed-off-by: Leon Hwang <hffilwlqm@gmail.com>
Reviewed-by: Maciej Fijalkowski <maciej.fijalkowski@intel.com>
Link: https://lore.kernel.org/r/20230906154256.95461-1-hffilwlqm@gmail.com
Signed-off-by: Martin KaFai Lau <martin.lau@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

iavf: Fix promiscuous mode configuration flow messages

[ Upstream commit 221465de6bd8090ab61267f019866e8d2dd4ea3d ]

Currently when configuring promiscuous mode on the AVF we detect a
change in the netdev->flags. We use IFF_PROMISC and IFF_ALLMULTI to
determine whether or not we need to request/release promiscuous mode
and/or multicast promiscuous mode. The problem is that the AQ calls for
setting/clearing promiscuous/multicast mode are treated separately. This
leads to a case where we can trigger two promiscuous mode AQ calls in
a row with the incorrect state. To fix this make a few changes.

Use IAVF_FLAG_AQ_CONFIGURE_PROMISC_MODE instead of the previous
IAVF_FLAG_AQ_[REQUEST|RELEASE]_[PROMISC|ALLMULTI] flags.

In iavf_set_rx_mode() detect if there is a change in the
netdev->flags in comparison with adapter->flags and set the
IAVF_FLAG_AQ_CONFIGURE_PROMISC_MODE aq_required bit. Then in
iavf_process_aq_command() only check for IAVF_FLAG_CONFIGURE_PROMISC_MODE
and call iavf_set_promiscuous() if it's set.

In iavf_set_promiscuous() check again to see which (if any) promiscuous
mode bits have changed when comparing the netdev->flags with the
adapter->flags. Use this to set the flags which get sent to the PF
driver.

Add a spinlock that is used for updating current_netdev_promisc_flags
and only allows one promiscuous mode AQ at a time.

[1] Fixes the fact that we will only have one AQ call in the aq_required
queue at any one time.

[2] Streamlines the change in promiscuous mode to only set one AQ
required bit.

[3] This allows us to keep track of the current state of the flags and
also makes it so we can take the most recent netdev->flags promiscuous
mode state.

[4] This fixes the problem where a change in the netdev->flags can cause
IAVF_FLAG_AQ_CONFIGURE_PROMISC_MODE to be set in iavf_set_rx_mode(),
but cleared in iavf_set_promiscuous() before the change is ever made via
AQ call.

Fixes: 47d3483988f6 ("i40evf: Add driver support for promiscuous mode")
Signed-off-by: Brett Creeley <brett.creeley@intel.com>
Signed-off-by: Ahmed Zaki <ahmed.zaki@intel.com>
Tested-by: Rafal Romanowski <rafal.romanowski@intel.com>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

i40e: fix potential memory leaks in i40e_remove()

[ Upstream commit 5ca636d927a106780451d957734f02589b972e2b ]

Instead of freeing memory of a single VSI, make sure
the memory for all VSIs is cleared before releasing VSIs.
Add releasing of their resources in a loop with the iteration
number equal to the number of allocated VSIs.

Fixes: 41c445ff0f48 ("i40e: main driver core")
Signed-off-by: Andrii Staikov <andrii.staikov@intel.com>
Signed-off-by: Aleksandr Loktionov <aleksandr.loktionov@intel.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Tony Nguyen <anthony.l.nguyen@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: iwlwifi: don't use an uninitialized variable

[ Upstream commit c46fcc6e43d617252945e706f04e5f82a59f2b8e ]

Don't use variable err uninitialized.
The reason for removing the check instead of initializing it
in the beginning of the function is because that way
static checkers will be able to catch issues if we do something
wrong in the future.

Fixes: bf976c814c86 ("wifi: iwlwifi: mvm: implement link change ops")
Signed-off-by: Miri Korenblit <miriam.rachel.korenblit@intel.com>
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Link: https://lore.kernel.org/r/20230830112059.431b01bd8779.I31fc4ab35f551b85a10f974a6b18fc30191e9c35@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: iwlwifi: honor the enable_ini value

[ Upstream commit e0c1ca236e28e4263fba76d47a108ed95dcae33e ]

In case the user sets the enable_ini to some preset, we want to honor
the value.

Remove the ops to set the value of the module parameter is runtime, we
don't want to allow to modify the value in runtime since we configure
the firmware once at the beginning on its life.

Fixes: b49c2b252b58 ("iwlwifi: Configure FW debug preset via module param.")
Signed-off-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Link: https://lore.kernel.org/r/20230830112059.5734e0f374bb.I6698eda8ed2112378dd47ac5d62866ebe7a94f77@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mac80211: fix # of MSDU in A-MSDU calculation

[ Upstream commit 428e8976a15f849ad92b1c1e38dda2a684350ff7 ]

During my refactoring I wanted to get rid of the switch,
but replaced it with the wrong calculation. Fix that.

Fixes: 175ad2ec89fe ("wifi: mac80211: limit A-MSDU subframes for client too")
Reported-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Link: https://lore.kernel.org/r/20230827135854.51bf1b8b0adb.Iffbd337fdad2b86ae12f5a39c69fb82b517f7486@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: cfg80211: fix off-by-one in element defrag

[ Upstream commit 43125539fc69c6aa63d34b516939431391bddeac ]

If a fragment is the last element, it's erroneously not
accepted. Fix that.

Fixes: f837a653a097 ("wifi: cfg80211: add element defragmentation helper")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Gregory Greenman <gregory.greenman@intel.com>
Link: https://lore.kernel.org/r/20230827135854.adca9fbd3317.I6b2df45eb71513f3e48efd196ae3cddec362dc1c@changeid
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mac80211: fix RCU usage warning in mesh fast-xmit

[ Upstream commit 5ea82df1f50e42416d0a8a7c42d37cc1df1545fe ]

In mesh_fast_tx_flush_addr() we already hold the lock, so
don't need additional hashtable RCU protection. Use the
rhashtable_lookup_fast() variant to avoid RCU protection
warnings.

Fixes: d5edb9ae8d56 ("wifi: mac80211: mesh fast xmit support")
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mac80211: move sched-scan stop work to wiphy work

[ Upstream commit eadfb54756aea5610d8d0a467f66305f777c85dd ]

This also has the wiphy locked here then. We need to use
the _locked version of cfg80211_sched_scan_stopped() now,
which also fixes an old deadlock there.

Fixes: a05829a7222e ("cfg80211: avoid holding the RTNL when calling the driver")
Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mac80211: move offchannel works to wiphy work

[ Upstream commit 97c19e42b264e6b71a9ff9deea04c19f621805b9 ]

Make the offchannel works wiphy works to have the
wiphy locked for executing them.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Stable-dep-of: eadfb54756ae ("wifi: mac80211: move sched-scan stop work to wiphy work")
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mac80211: move scan work to wiphy work

[ Upstream commit 201712512cbbda360f62c222a4bab260350462a0 ]

Move the scan work to wiphy work, which also simplifies
the way we handle the work vs. the scan configuration.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Stable-dep-of: eadfb54756ae ("wifi: mac80211: move sched-scan stop work to wiphy work")
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: mac80211: move radar detect work to wiphy work

[ Upstream commit 228e4f931b0e630dacca8dd867ddd863aea53913 ]

Move the radar detect work to wiphy work in order
to lock the wiphy for it without doing it manually.

Reviewed-by: Emmanuel Grumbach <emmanuel.grumbach@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Stable-dep-of: eadfb54756ae ("wifi: mac80211: move sched-scan stop work to wiphy work")
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: cfg80211: add flush functions for wiphy work

[ Upstream commit 56cfb8ce1f7f6c4e5ca571a2ec0880e131cd0311 ]

There may be sometimes reasons to actually run the work
if it's pending, add flush functions for both regular and
delayed wiphy work that will do this.

Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Stable-dep-of: eadfb54756ae ("wifi: mac80211: move sched-scan stop work to wiphy work")
Signed-off-by: Sasha Levin <sashal@kernel.org>

wifi: ath12k: fix undefined behavior with __fls in dp

[ Upstream commit d48f55e773dcce8fcf9e587073452a4944011b11 ]

When max virtual ap interfaces are configured in all the bands
with ACS and hostapd restart is done every 60s,
a crash is observed at random times because of handling the
uninitialized peer fragments with fragment id of packet as 0.

"__fls" would have an undefined behavior if the argument is passed
as "0". Hence, added changes to handle the same.

Tested-on: QCN9274 hw2.0 PCI WLAN.WBE.1.0.1-00029-QCAHKSWPL_SILICONZ-1

Fixes: d889913205cf ("wifi: ath12k: driver for Qualcomm Wi-Fi 7 devices")
Signed-off-by: Harshitha Prem <quic_hprem@quicinc.com>
Signed-off-by: Kalle Valo <quic_kvalo@quicinc.com>
Link: https://lore.kernel.org/r/20230821130343.29495-3-quic_hprem@quicinc.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

irqchip/sifive-plic: Fix syscore registration for multi-socket systems

[ Upstream commit f99b926f6543faeadba1b4524d8dc9c102489135 ]

Multi-socket systems have a separate PLIC in each socket, so __plic_init()
is invoked for each PLIC. __plic_init() registers syscore operations, which
obviously fails on the second invocation.

Move it into the already existing condition for installing the CPU hotplug
state so it is only invoked once when the first PLIC is initialized.

[ tglx: Massaged changelog ]

Fixes: e80f0b6a2cf3 ("irqchip/irq-sifive-plic: Add syscore callbacks for hibernation")
Signed-off-by: Anup Patel <apatel@ventanamicro.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20231025142820.390238-4-apatel@ventanamicro.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

genirq/matrix: Exclude managed interrupts in irq_matrix_allocated()

[ Upstream commit a0b0bad10587ae2948a7c36ca4ffc206007fbcf3 ]

When a CPU is about to be offlined, x86 validates that all active
interrupts which are targeted to this CPU can be migrated to the remaining
online CPUs. If not, the offline operation is aborted.

The validation uses irq_matrix_allocated() to retrieve the number of
vectors which are allocated on the outgoing CPU. The returned number of
allocated vectors includes also vectors which are associated to managed
interrupts.

That's overaccounting because managed interrupts are:

  - not migrated when the affinity mask of the interrupt targets only
    the outgoing CPU

  - migrated to another CPU, but in that case the vector is already
    pre-allocated on the potential target CPUs and must not be taken into
    account.

As a consequence the check whether the remaining online CPUs have enough
capacity for migrating the allocated vectors from the outgoing CPU might
fail incorrectly.

Let irq_matrix_allocated() return only the number of allocated non-managed
interrupts to make this validation check correct.

[ tglx: Amend changelog and fixup kernel-doc comment ]

Fixes: 2f75d9e1c905 ("genirq: Implement bitmap matrix allocator")
Reported-by: Wendy Wang <wendy.wang@intel.com>
Signed-off-by: Chen Yu <yu.c.chen@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20231020072522.557846-1-yu.c.chen@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

string: Adjust strtomem() logic to allow for smaller sources

[ Upstream commit 0e108725f6cc5b3be9e607f89c9fbcbb236367b7 ]

Arnd noticed we have a case where a shorter source string is being copied
into a destination byte array, but this results in a strnlen() call that
exceeds the size of the source. This is seen with -Wstringop-overread:

In file included from ../include/linux/uuid.h:11,
                 from ../include/linux/mod_devicetable.h:14,
                 from ../include/linux/cpufeature.h:12,
                 from ../arch/x86/coco/tdx/tdx.c:7:
../arch/x86/coco/tdx/tdx.c: In function 'tdx_panic.constprop':
../include/linux/string.h:284:9: error: 'strnlen' specified bound 64 exceeds source size 60 [-Werror=stringop-overread]
  284 |         memcpy_and_pad(dest, _dest_len, src, strnlen(src, _dest_len), pad); \
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../arch/x86/coco/tdx/tdx.c:124:9: note: in expansion of macro 'strtomem_pad'
  124 |         strtomem_pad(message.str, msg, '\0');
      |         ^~~~~~~~~~~~

Use the smaller of the two buffer sizes when calling strnlen(). When
src length is unknown (SIZE_MAX), it is adjusted to use dest length,
which is what the original code did.

Reported-by: Arnd Bergmann <arnd@arndb.de>
Fixes: dfbafa70bde2 ("string: Introduce strtomem() and strtomem_pad()")
Tested-by: Arnd Bergmann <arnd@arndb.de>
Cc: Andy Shevchenko <andy@kernel.org>
Cc: linux-hardening@vger.kernel.org
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

PCI/MSI: Provide stubs for IMS functions

[ Upstream commit 41efa431244f6498833ff8ee8dde28c4924c5479 ]

The IMS related functions (pci_create_ims_domain(), pci_ims_alloc_irq(),
and pci_ims_free_irq()) are not declared when CONFIG_PCI_MSI is disabled.

Provide definitions of these functions for use when callers are compiled
with CONFIG_PCI_MSI disabled.

Fixes: 0194425af0c8 ("PCI/MSI: Provide IMS (Interrupt Message Store) support")
Fixes: c9e5bea27383 ("PCI/MSI: Provide pci_ims_alloc/free_irq()")
Signed-off-by: Reinette Chatre <reinette.chatre@intel.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/14ff656899a3757453f8584c1109d7a9b98fa258.1697564731.git.reinette.chatre@intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

selftests/x86/lam: Zero out buffer for readlink()

[ Upstream commit 29060633411a02f6f2dd9d5245919385d69d81f0 ]

Zero out the buffer for readlink() since readlink() does not append a
terminating null byte to the buffer.  Also change the buffer length
passed to readlink() to 'PATH_MAX - 1' to ensure the resulting string
is always null terminated.

Fixes: 833c12ce0f430 ("selftests/x86/lam: Add inherit test cases for linear-address masking")
Signed-off-by: Binbin Wu <binbin.wu@linux.intel.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Link: https://lore.kernel.org/r/20231016062446.695-1-binbin.wu@linux.intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

perf: Optimize perf_cgroup_switch()

[ Upstream commit f06cc667f79909e9175460b167c277b7c64d3df0 ]

Namhyung reported that bd2756811766 ("perf: Rewrite core context handling")
regresses context switch overhead when perf-cgroup is in use together
with 'slow' PMUs like uncore.

Specifically, perf_cgroup_switch()'s perf_ctx_disable() /
ctx_sched_out() etc.. all iterate the full list of active PMUs for
that CPU, even if they don't have cgroup events.

Previously there was cgrp_cpuctx_list which linked the relevant PMUs
together, but that got lost in the rework. Instead of re-instruducing
a similar list, let the perf_event_pmu_context iteration skip those
that do not have cgroup events. This avoids growing multiple versions
of the perf_event_pmu_context iteration.

Measured performance (on a slightly different patch):

Before)

  $ taskset -c 0 ./perf bench sched pipe -l 10000 -G AAA,BBB
  # Running 'sched/pipe' benchmark:
  # Executed 10000 pipe operations between two processes

       Total time: 0.901 [sec]

        90.128700 usecs/op
            11095 ops/sec

After)

  $ taskset -c 0 ./perf bench sched pipe -l 10000 -G AAA,BBB
  # Running 'sched/pipe' benchmark:
  # Executed 10000 pipe operations between two processes

       Total time: 0.065 [sec]

         6.560100 usecs/op
           152436 ops/sec

Fixes: bd2756811766 ("perf: Rewrite core context handling")
Reported-by: Namhyung Kim <namhyung@kernel.org>
Debugged-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lkml.kernel.org/r/20231009210425.GC6307@noisy.programming.kicks-ass.net
Signed-off-by: Sasha Levin <sashal@kernel.org>

pstore/platform: Add check for kstrdup

[ Upstream commit a19d48f7c5d57c0f0405a7d4334d1d38fe9d3c1c ]

Add check for the return value of kstrdup() and return the error
if it fails in order to avoid NULL pointer dereference.

Fixes: 563ca40ddf40 ("pstore/platform: Switch pstore_info::name to const")
Signed-off-by: Jiasheng Jiang <jiasheng@iscas.ac.cn>
Link: https://lore.kernel.org/r/20230623022706.32125-1-jiasheng@iscas.ac.cn
Signed-off-by: Kees Cook <keescook@chromium.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/nmi: Fix out-of-order NMI nesting checks & false positive warning

[ Upstream commit f44075ecafb726830e63d33fbca29413149eeeb8 ]

The ->idt_seq and ->recv_jiffies variables added by:

  1a3ea611fc10 ("x86/nmi: Accumulate NMI-progress evidence in exc_nmi()")

... place the exit-time check of the bottom bit of ->idt_seq after the
this_cpu_dec_return() that re-enables NMI nesting.  This can result in
the following sequence of events on a given CPU in kernels built with
CONFIG_NMI_CHECK_CPU=y:

  o   An NMI arrives, and ->idt_seq is incremented to an odd number.
      In addition, nmi_state is set to NMI_EXECUTING==1.

  o   The NMI is processed.

  o   The this_cpu_dec_return(nmi_state) zeroes nmi_state and returns
      NMI_EXECUTING==1, thus opting out of the "goto nmi_restart".

  o   Another NMI arrives and ->idt_seq is incremented to an even
      number, triggering the warning.  But all is just fine, at least
      assuming we don't get so many closely spaced NMIs that the stack
      overflows or some such.

Experience on the fleet indicates that the MTBF of this false positive
is about 70 years.  Or, for those who are not quite that patient, the
MTBF appears to be about one per week per 4,000 systems.

Fix this false-positive warning by moving the "nmi_restart" label before
the initial ->idt_seq increment/check and moving the this_cpu_dec_return()
to follow the final ->idt_seq increment/check.  This way, all nested NMIs
that get past the NMI_NOT_RUNNING check get a clean ->idt_seq slate.
And if they don't get past that check, they will set nmi_state to
NMI_LATCHED, which will cause the this_cpu_dec_return(nmi_state)
to restart.

Fixes: 1a3ea611fc10 ("x86/nmi: Accumulate NMI-progress evidence in exc_nmi()")
Reported-by: Chris Mason <clm@fb.com>
Signed-off-by: Paul E. McKenney <paulmck@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Link: https://lore.kernel.org/r/0cbff831-6e3d-431c-9830-ee65ee7787ff@paulmck-laptop
Signed-off-by: Sasha Levin <sashal@kernel.org>

drivers/clocksource/timer-ti-dm: Don't call clk_get_rate() in stop function

[ Upstream commit 12590d4d0e331d3cb9e6b3494515cd61c8a6624e ]

clk_get_rate() might sleep, and that prevents dm-timer based PWM from being
used from atomic context.

Fix that by getting fclk rate in probe() and using a notifier in case rate
changes.

Fixes: af04aa856e93 ("ARM: OMAP: Move dmtimer driver out of plat-omap to drivers under clocksource")
Signed-off-by: Ivaylo Dimitrov <ivo.g.dimitrov.75@gmail.com>
Reviewed-by: Tony Lindgren <tony@atomide.com>
Signed-off-by: Daniel Lezcano <daniel.lezcano@linaro.org>
Link: https://lore.kernel.org/r/1696312220-11550-1-git-send-email-ivo.g.dimitrov.75@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

srcu: Fix callbacks acceleration mishandling

[ Upstream commit 4a8e65b0c348e42107c64381e692e282900be361 ]

SRCU callbacks acceleration might fail if the preceding callbacks
advance also fails. This can happen when the following steps are met:

1) The RCU_WAIT_TAIL segment has callbacks (say for gp_num 8) and the
   RCU_NEXT_READY_TAIL also has callbacks (say for gp_num 12).

2) The grace period for RCU_WAIT_TAIL is observed as started but not yet
   completed so rcu_seq_current() returns 4 + SRCU_STATE_SCAN1 = 5.

3) This value is passed to rcu_segcblist_advance() which can't move
   any segment forward and fails.

4) srcu_gp_start_if_needed() still proceeds with callback acceleration.
   But then the call to rcu_seq_snap() observes the grace period for the
   RCU_WAIT_TAIL segment (gp_num 8) as completed and the subsequent one
   for the RCU_NEXT_READY_TAIL segment as started
   (ie: 8 + SRCU_STATE_SCAN1 = 9) so it returns a snapshot of the
   next grace period, which is 16.

5) The value of 16 is passed to rcu_segcblist_accelerate() but the
   freshly enqueued callback in RCU_NEXT_TAIL can't move to
   RCU_NEXT_READY_TAIL which already has callbacks for a previous grace
   period (gp_num = 12). So acceleration fails.

6) Note in all these steps, srcu_invoke_callbacks() hadn't had a chance
   to run srcu_invoke_callbacks().

Then some very bad outcome may happen if the following happens:

7) Some other CPU races and starts the grace period number 16 before the
   CPU handling previous steps had a chance. Therefore srcu_gp_start()
   isn't called on the latter sdp to fix the acceleration leak from
   previous steps with a new pair of call to advance/accelerate.

8) The grace period 16 completes and srcu_invoke_callbacks() is finally
   called. All the callbacks from previous grace periods (8 and 12) are
   correctly advanced and executed but callbacks in RCU_NEXT_READY_TAIL
   still remain. Then rcu_segcblist_accelerate() is called with a
   snaphot of 20.

9) Since nothing started the grace period number 20, callbacks stay
   unhandled.

This has been reported in real load:

[3144162.608392] INFO: task kworker/136:12:252684 blocked for more
than 122 seconds.
[3144162.615986]       Tainted: G           O  K   5.4.203-1-tlinux4-0011.1 #1
[3144162.623053] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
disables this message.
[3144162.631162] kworker/136:12  D    0 252684      2 0x90004000
[3144162.631189] Workqueue: kvm-irqfd-cleanup irqfd_shutdown [kvm]
[3144162.631192] Call Trace:
[3144162.631202]  __schedule+0x2ee/0x660
[3144162.631206]  schedule+0x33/0xa0
[3144162.631209]  schedule_timeout+0x1c4/0x340
[3144162.631214]  ? update_load_avg+0x82/0x660
[3144162.631217]  ? raw_spin_rq_lock_nested+0x1f/0x30
[3144162.631218]  wait_for_completion+0x119/0x180
[3144162.631220]  ? wake_up_q+0x80/0x80
[3144162.631224]  __synchronize_srcu.part.19+0x81/0xb0
[3144162.631226]  ? __bpf_trace_rcu_utilization+0x10/0x10
[3144162.631227]  synchronize_srcu+0x5f/0xc0
[3144162.631236]  irqfd_shutdown+0x3c/0xb0 [kvm]
[3144162.631239]  ? __schedule+0x2f6/0x660
[3144162.631243]  process_one_work+0x19a/0x3a0
[3144162.631244]  worker_thread+0x37/0x3a0
[3144162.631247]  kthread+0x117/0x140
[3144162.631247]  ? process_one_work+0x3a0/0x3a0
[3144162.631248]  ? __kthread_cancel_work+0x40/0x40
[3144162.631250]  ret_from_fork+0x1f/0x30

Fix this with taking the snapshot for acceleration _before_ the read
of the current grace period number.

The only side effect of this solution is that callbacks advancing happen
then _after_ the full barrier in rcu_seq_snap(). This is not a problem
because that barrier only cares about:

1) Ordering accesses of the update side before call_srcu() so they don't
   bleed.
2) See all the accesses prior to the grace period of the current gp_num

The only things callbacks advancing need to be ordered against are
carried by snp locking.

Reported-by: Yong He <alexyonghe@tencent.com>
Co-developed-by:: Yong He <alexyonghe@tencent.com>
Signed-off-by: Yong He <alexyonghe@tencent.com>
Co-developed-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Signed-off-by: Joel Fernandes (Google) <joel@joelfernandes.org>
Co-developed-by: Neeraj upadhyay <Neeraj.Upadhyay@amd.com>
Signed-off-by: Neeraj upadhyay <Neeraj.Upadhyay@amd.com>
Link: http://lore.kernel.org/CANZk6aR+CqZaqmMWrC2eRRPY12qAZnDZLwLnHZbNi=xXMB401g@mail.gmail.com
Fixes: da915ad5cf25 ("srcu: Parallelize callback handling")
Signed-off-by: Frederic Weisbecker <frederic@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/apic: Fake primary thread mask for XEN/PV

[ Upstream commit 965e05ff8af98c44f9937366715c512000373164 ]

The SMT control mechanism got added as speculation attack vector
mitigation. The implemented logic relies on the primary thread mask to
be set up properly.

This turns out to be an issue with XEN/PV guests because their CPU hotplug
mechanics do not enumerate APICs and therefore the mask is never correctly
populated.

This went unnoticed so far because by chance XEN/PV ends up with
smp_num_siblings == 2. So cpu_smt_control stays at its default value
CPU_SMT_ENABLED and the primary thread mask is never evaluated in the
context of CPU hotplug.

This stopped "working" with the upcoming overhaul of the topology
evaluation which legitimately provides a fake topology for XEN/PV. That
sets smp_num_siblings to 1, which causes the core CPU hot-plug core to
refuse to bring up the APs.

This happens because cpu_smt_control is set to CPU_SMT_NOT_SUPPORTED which
causes cpu_bootable() to evaluate the unpopulated primary thread mask with
the conclusion that all non-boot CPUs are not valid to be plugged.

The core code has already been made more robust against this kind of fail,
but the primary thread mask really wants to be populated to avoid other
issues all over the place.

Just fake the mask by pretending that all XEN/PV vCPUs are primary threads,
which is consistent because all of XEN/PVs topology is fake or non-existent.

Fixes: 6a4d2657e048 ("x86/smp: Provide topology_is_primary_thread()")
Fixes: f54d4434c281 ("x86/apic: Provide cpu_primary_thread mask")
Reported-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230814085112.210011520@linutronix.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

cpu/SMT: Make SMT control more robust against enumeration failures

[ Upstream commit d91bdd96b55cc3ce98d883a60f133713821b80a6 ]

The SMT control mechanism got added as speculation attack vector
mitigation. The implemented logic relies on the primary thread mask to
be set up properly.

This turns out to be an issue with XEN/PV guests because their CPU hotplug
mechanics do not enumerate APICs and therefore the mask is never correctly
populated.

This went unnoticed so far because by chance XEN/PV ends up with
smp_num_siblings == 2. So smt_hotplug_control stays at its default value
CPU_SMT_ENABLED and the primary thread mask is never evaluated in the
context of CPU hotplug.

This stopped "working" with the upcoming overhaul of the topology
evaluation which legitimately provides a fake topology for XEN/PV. That
sets smp_num_siblings to 1, which causes the core CPU hot-plug core to
refuse to bring up the APs.

This happens because smt_hotplug_control is set to CPU_SMT_NOT_SUPPORTED
which causes cpu_smt_allowed() to evaluate the unpopulated primary thread
mask with the conclusion that all non-boot CPUs are not valid to be
plugged.

Make cpu_smt_allowed() more robust and take CPU_SMT_NOT_SUPPORTED and
CPU_SMT_NOT_IMPLEMENTED into account. Rename it to cpu_bootable() while at
it as that makes it more clear what the function is about.

The primary mask issue on x86 XEN/PV needs to be addressed separately as
there are users outside of the CPU hotplug code too.

Fixes: 05736e4ac13c ("cpu/hotplug: Provide knobs to control SMT")
Reported-by: Juergen Gross <jgross@suse.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Juergen Gross <jgross@suse.com>
Tested-by: Sohil Mehta <sohil.mehta@intel.com>
Tested-by: Michael Kelley <mikelley@microsoft.com>
Tested-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: Zhang Rui <rui.zhang@intel.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230814085112.149440843@linutronix.de
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/boot: Fix incorrect startup_gdt_descr.size

[ Upstream commit 001470fed5959d01faecbd57fcf2f60294da0de1 ]

Since the size value is added to the base address to yield the last valid
byte address of the GDT, the current size value of startup_gdt_descr is
incorrect (too large by one), fix it.

[ mingo: This probably never mattered, because startup_gdt[] is only used
         in a very controlled fashion - but make it consistent nevertheless. ]

Fixes: 866b556efa12 ("x86/head/64: Install startup GDT")
Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Link: https://lore.kernel.org/r/20230807084547.217390-1-ytcoode@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/sev-es: Allow copy_from_kernel_nofault() in earlier boot

[ Upstream commit f79936545fb122856bd78b189d3c7ee59928c751 ]

Previously, if copy_from_kernel_nofault() was called before
boot_cpu_data.x86_virt_bits was set up, then it would trigger undefined
behavior due to a shift by 64.

This ended up causing boot failures in the latest version of ubuntu2204
in the gcp project when using SEV-SNP.

Specifically, this function is called during an early #VC handler which
is triggered by a CPUID to check if NX is implemented.

Fixes: 1aa9aa8ee517 ("x86/sev-es: Setup GHCB-based boot #VC handler")
Suggested-by: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Adam Dunlap <acdunlap@google.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Tested-by: Jacob Xu <jacobhxu@google.com>
Link: https://lore.kernel.org/r/20230912002703.3924521-2-acdunlap@google.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

cgroup/cpuset: Fix load balance state in update_partition_sd_lb()

[ Upstream commit 6fcdb0183bf024a70abccb0439321c25891c708d ]

Commit a86ce68078b2 ("cgroup/cpuset: Extract out CS_CPU_EXCLUSIVE
& CS_SCHED_LOAD_BALANCE handling") adds a new helper function
update_partition_sd_lb() to update the load balance state of the
cpuset. However the new load balance is determined by just looking at
whether the cpuset is a valid isolated partition root or not.  That is
not enough if the cpuset is not a valid partition root but its parent
is in the isolated state (load balance off). Update the function to
set the new state to be the same as its parent in this case like what
has been done in commit c8c926200c55 ("cgroup/cpuset: Inherit parent's
load balance state in v2").

Fixes: a86ce68078b2 ("cgroup/cpuset: Extract out CS_CPU_EXCLUSIVE & CS_SCHED_LOAD_BALANCE handling")
Signed-off-by: Waiman Long <longman@redhat.com>
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

ACPI/NUMA: Apply SRAT proximity domain to entire CFMWS window

[ Upstream commit 8f1004679987302b155f14b966ca6d4335814fcb ]

Commit fd49f99c1809 ("ACPI: NUMA: Add a node and memblk for each
CFMWS not in SRAT") did not account for the case where the BIOS
only partially describes a CFMWS Window in the SRAT. That means
the omitted address ranges, of a partially described CFMWS Window,
do not get assigned to a NUMA node.

Replace the call to phys_to_target_node() with numa_add_memblks().
Numa_add_memblks() searches an HPA range for existing memblk(s)
and extends those memblk(s) to fill the entire CFMWS Window.

Extending the existing memblks is a simple strategy that reuses
SRAT defined proximity domains from part of a window to fill out
the entire window, based on the knowledge* that all of a CFMWS
window is of a similar performance class.

*Note that this heuristic will evolve when CFMWS Windows present
a wider range of characteristics. The extension of the proximity
domain, implemented here, is likely a step in developing a more
sophisticated performance profile in the future.

There is no change in behavior when the SRAT does not describe
the CFMWS Window at all. In that case, a new NUMA node with a
single memblk covering the entire CFMWS Window is created.

Fixes: fd49f99c1809 ("ACPI: NUMA: Add a node and memblk for each CFMWS not in SRAT")
Reported-by: Derick Marks <derick.w.marks@intel.com>
Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Derick Marks <derick.w.marks@intel.com>
Link: https://lore.kernel.org/all/eaa0b7cffb0951a126223eef3cbe7b55b8300ad9.1689018477.git.alison.schofield%40intel.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/numa: Introduce numa_fill_memblks()

[ Upstream commit 8f012db27c9516be1a7aca93ea4a6ca9c75056c9 ]

numa_fill_memblks() fills in the gaps in numa_meminfo memblks
over an physical address range.

The ACPI driver will use numa_fill_memblks() to implement a new Linux
policy that prescribes extending proximity domains in a portion of a
CFMWS window to the entire window.

Dan Williams offered this explanation of the policy:
A CFWMS is an ACPI data structure that indicates *potential* locations
where CXL memory can be placed. It is the playground where the CXL
driver has free reign to establish regions. That space can be populated
by BIOS created regions, or driver created regions, after hotplug or
other reconfiguration.

When BIOS creates a region in a CXL Window it additionally describes
that subset of the Window range in the other typical ACPI tables SRAT,
SLIT, and HMAT. The rationale for BIOS not pre-describing the entire
CXL Window in SRAT, SLIT, and HMAT is that it can not predict the
future. I.e. there is nothing stopping higher or lower performance
devices being placed in the same Window. Compare that to ACPI memory
hotplug that just onlines additional capacity in the proximity domain
with little freedom for dynamic performance differentiation.

That leaves the OS with a choice, should unpopulated window capacity
match the proximity domain of an existing region, or should it allocate
a new one? This patch takes the simple position of minimizing proximity
domain proliferation by reusing any proximity domain intersection for
the entire Window. If the Window has no intersections then allocate a
new proximity domain. Note that SRAT, SLIT and HMAT information can be
enumerated dynamically in a standard way from device provided data.
Think of CXL as the end of ACPI needing to describe memory attributes,
CXL offers a standard discovery model for performance attributes, but
Linux still needs to interoperate with the old regime.

Reported-by: Derick Marks <derick.w.marks@intel.com>
Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Alison Schofield <alison.schofield@intel.com>
Signed-off-by: Dave Hansen <dave.hansen@linux.intel.com>
Reviewed-by: Dan Williams <dan.j.williams@intel.com>
Tested-by: Derick Marks <derick.w.marks@intel.com>
Link: https://lore.kernel.org/all/ef078a6f056ca974e5af85997013c0fda9e3326d.1689018477.git.alison.schofield%40intel.com
Stable-dep-of: 8f1004679987 ("ACPI/NUMA: Apply SRAT proximity domain to entire CFMWS window")
Signed-off-by: Sasha Levin <sashal@kernel.org>

futex: Don't include process MM in futex key on no-MMU

[ Upstream commit c73801ae4f22b390228ebf471d55668e824198b6 ]

On no-MMU, all futexes are treated as private because there is no need
to map a virtual address to physical to match the futex across
processes. This doesn't quite work though, because private futexes
include the current process's mm_struct as part of their key. This makes
it impossible for one process to wake up a shared futex being waited on
in another process.

Fix this bug by excluding the mm_struct from the key. With
a single address space, the futex address is already a unique key.

Fixes: 784bdf3bb694 ("futex: Assume all mappings are private on !MMU systems")
Signed-off-by: Ben Wolsieffer <ben.wolsieffer@hefring.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Darren Hart <dvhart@infradead.org>
Cc: Davidlohr Bueso <dave@stgolabs.net>
Cc: André Almeida <andrealmeid@igalia.com>
Link: https://lore.kernel.org/r/20231019204548.1236437-2-ben.wolsieffer@hefring.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/srso: Fix unret validation dependencies

[ Upstream commit eeb9f34df065f42f0c9195b322ba6df420c9fc92 ]

CONFIG_CPU_SRSO isn't dependent on CONFIG_CPU_UNRET_ENTRY (AMD
Retbleed), so the two features are independently configurable.  Fix
several issues for the (presumably rare) case where CONFIG_CPU_SRSO is
enabled but CONFIG_CPU_UNRET_ENTRY isn't.

Fixes: fb3bd914b3ec ("x86/srso: Add a Speculative RAS Overflow mitigation")
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Acked-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/299fb7740174d0f2335e91c58af0e9c242b4bac1.1693889988.git.jpoimboe@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/srso: Fix vulnerability reporting for missing microcode

[ Upstream commit dc6306ad5b0dda040baf1fde3cfd458e6abfc4da ]

The SRSO default safe-ret mitigation is reported as "mitigated" even if
microcode hasn't been updated.  That's wrong because userspace may still
be vulnerable to SRSO attacks due to IBPB not flushing branch type
predictions.

Report the safe-ret + !microcode case as vulnerable.

Also report the microcode-only case as vulnerable as it leaves the
kernel open to attacks.

Fixes: fb3bd914b3ec ("x86/srso: Add a Speculative RAS Overflow mitigation")
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Acked-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/a8a14f97d1b0e03ec255c81637afdf4cf0ae9c99.1693889988.git.jpoimboe@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/srso: Print mitigation for retbleed IBPB case

[ Upstream commit de9f5f7b06a5b7adbfdd8016f011120a4e928add ]

When overriding the requested mitigation with IBPB due to retbleed=ibpb,
print the mitigation in the usual format instead of a custom error
message.

Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Acked-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/ec3af919e267773d896c240faf30bfc6a1fd6304.1693889988.git.jpoimboe@kernel.org
Stable-dep-of: dc6306ad5b0d ("x86/srso: Fix vulnerability reporting for missing microcode")
Signed-off-by: Sasha Levin <sashal@kernel.org>

x86/srso: Fix SBPB enablement for (possible) future fixed HW

[ Upstream commit 1d1142ac51307145dbb256ac3535a1d43a1c9800 ]

Make the SBPB check more robust against the (possible) case where future
HW has SRSO fixed but doesn't have the SRSO_NO bit set.

Fixes: 1b5277c0ea0b ("x86/srso: Add SRSO_NO support")
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Acked-by: Borislav Petkov (AMD) <bp@alien8.de>
Link: https://lore.kernel.org/r/cee5050db750b391c9f35f5334f8ff40e66c01b9.1693889988.git.jpoimboe@kernel.org
Signed-off-by: Sasha Levin <sashal@kernel.org>

writeback, cgroup: switch inodes with dirty timestamps to release dying cgwbs

[ Upstream commit 6654408a33e6297d8e1d2773409431d487399b95 ]

The cgwb cleanup routine will try to release the dying cgwb by switching
the attached inodes.  It fetches the attached inodes from wb->b_attached
list, omitting the fact that inodes only with dirty timestamps reside in
wb->b_dirty_time list, which is the case when lazytime is enabled.  This
causes enormous zombie memory cgroup when lazytime is enabled, as inodes
with dirty timestamps can not be switched to a live cgwb for a long time.

It is reasonable not to switch cgwb for inodes with dirty data, as
otherwise it may break the bandwidth restrictions.  However since the
writeback of inode metadata is not accounted for, let's also switch
inodes with dirty timestamps to avoid zombie memory and block cgroups
when laztytime is enabled.

Fixes: c22d70a162d3 ("writeback, cgroup: release dying cgwbs by switching attached inodes")
Reviewed-by: Jan Kara <jack@suse.cz>
Signed-off-by: Jingbo Xu <jefflexu@linux.alibaba.com>
Link: https://lore.kernel.org/r/20231014125511.102978-1-jefflexu@linux.alibaba.com
Acked-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

vfs: fix readahead(2) on block devices

[ Upstream commit 7116c0af4b8414b2f19fdb366eea213cbd9d91c2 ]

Readahead was factored to call generic_fadvise.  That refactor added an
S_ISREG restriction which broke readahead on block devices.

In addition to S_ISREG, this change checks S_ISBLK to fix block device
readahead.  There is no change in behavior with any file type besides block
devices in this change.

Fixes: 3d8f7615319b ("vfs: implement readahead(2) using POSIX_FADV_WILLNEED")
Signed-off-by: Reuben Hawkins <reubenhwk@gmail.com>
Link: https://lore.kernel.org/r/20231003015704.2415-1-reubenhwk@gmail.com
Reviewed-by: Amir Goldstein <amir73il@gmail.com>
Signed-off-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

nfsd: Handle EOPENSTALE correctly in the filecache

[ Upstream commit d59b3515ab021e010fdc58a8f445ea62dd2f7f4c ]

The nfsd_open code handles EOPENSTALE correctly, by retrying the call to
fh_verify() and __nfsd_open(). However the filecache just drops the
error on the floor, and immediately returns nfserr_stale to the caller.

This patch ensures that we propagate the EOPENSTALE code back to
nfsd_file_do_acquire, and that we handle it correctly.

Fixes: 65294c1f2c5e ("nfsd: add a new struct file caching facility to nfsd")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Reviewed-by: Jeff Layton <jlayton@kernel.org>
Message-Id: <20230911183027.11372-1-trond.myklebust@hammerspace.com>
Signed-off-by: Chuck Lever <chuck.lever@oracle.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

sched: Fix stop_one_cpu_nowait() vs hotplug

[ Upstream commit f0498d2a54e7966ce23cd7c7ff42c64fa0059b07 ]

Kuyo reported sporadic failures on a sched_setaffinity() vs CPU
hotplug stress-test -- notably affine_move_task() remains stuck in
wait_for_completion(), leading to a hung-task detector warning.

Specifically, it was reported that stop_one_cpu_nowait(.fn =
migration_cpu_stop) returns false -- this stopper is responsible for
the matching complete().

The race scenario is:

CPU0 CPU1

// doing _cpu_down()

  __set_cpus_allowed_ptr()
    task_rq_lock();
takedown_cpu()
  stop_machine_cpuslocked(take_cpu_down..)

<PREEMPT: cpu_stopper_thread()
  MULTI_STOP_PREPARE
  ...
    __set_cpus_allowed_ptr_locked()
      affine_move_task()
        task_rq_unlock();

  <PREEMPT: cpu_stopper_thread()\>
    ack_state()
  MULTI_STOP_RUN
    take_cpu_down()
      __cpu_disable();
      stop_machine_park();
stopper->enabled = false;
 />
   />
stop_one_cpu_nowait(.fn = migration_cpu_stop);
          if (stopper->enabled) // false!!!

That is, by doing stop_one_cpu_nowait() after dropping rq-lock, the
stopper thread gets a chance to preempt and allows the cpu-down for
the target CPU to complete.

OTOH, since stop_one_cpu_nowait() / cpu_stop_queue_work() needs to
issue a wakeup, it must not be ran under the scheduler locks.

Solve this apparent contradiction by keeping preemption disabled over
the unlock + queue_stopper combination:

preempt_disable();
task_rq_unlock(...);
if (!stop_pending)
  stop_one_cpu_nowait(...)
preempt_enable();

This respects the lock ordering contraints while still avoiding the
above race. That is, if we find the CPU is online under rq-lock, the
targeted stop_one_cpu_nowait() must succeed.

Apply this pattern to all similar stop_one_cpu_nowait() invocations.

Fixes: 6d337eab041d ("sched: Fix migrate_disable() vs set_cpus_allowed_ptr()")
Reported-by: "Kuyo Chang (張建文)" <Kuyo.Chang@mediatek.com>
Signed-off-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Tested-by: "Kuyo Chang (張建文)" <Kuyo.Chang@mediatek.com>
Link: https://lkml.kernel.org/r/20231010200442.GA16515@noisy.programming.kicks-ass.net
Signed-off-by: Sasha Levin <sashal@kernel.org>

objtool: Propagate early errors

[ Upstream commit e959c279d391c10b35ce300fb4b0fe3b98e86bd2 ]

If objtool runs into a problem that causes it to exit early, the overall
tool still returns a status code of 0, which causes the build to
continue as if nothing went wrong.

Note this only affects early errors, as later errors are still ignored
by check().

Fixes: b51277eb9775 ("objtool: Ditch subcommands")
Signed-off-by: Aaron Plattner <aplattner@nvidia.com>
Link: https://lore.kernel.org/r/cb6a28832d24b2ebfafd26da9abb95f874c83045.1696355111.git.aplattner@nvidia.com
Signed-off-by: Josh Poimboeuf <jpoimboe@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

sched/uclamp: Ignore (util == 0) optimization in feec() when p_util_max = 0

[ Upstream commit 23c9519def98ee0fa97ea5871535e9b136f522fc ]

find_energy_efficient_cpu() bails out early if effective util of the
task is 0 as the delta at this point will be zero and there's nothing
for EAS to do. When uclamp is being used, this could lead to wrong
decisions when uclamp_max is set to 0. In this case the task is capped
to performance point 0, but it is actually running and consuming energy
and we can benefit from EAS energy calculations.

Rework the condition so that it bails out when both util and uclamp_min
are 0.

We can do that without needing to use uclamp_task_util(); remove it.

Fixes: d81304bc6193 ("sched/uclamp: Cater for uclamp in find_energy_efficient_cpu()'s early exit condition")
Signed-off-by: Qais Yousef (Google) <qyousef@layalina.io>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230916232955.2099394-3-qyousef@layalina.io
Signed-off-by: Sasha Levin <sashal@kernel.org>

sched/uclamp: Set max_spare_cap_cpu even if max_spare_cap is 0

[ Upstream commit 6b00a40147653c8ea748e8f4396510f252763364 ]

When uclamp_max is being used, the util of the task could be higher than
the spare capacity of the CPU, but due to uclamp_max value we force-fit
it there.

The way the condition for checking for max_spare_cap in
find_energy_efficient_cpu() was constructed; it ignored any CPU that has
its spare_cap less than or _equal_ to max_spare_cap. Since we initialize
max_spare_cap to 0; this lead to never setting max_spare_cap_cpu and
hence ending up never performing compute_energy() for this cluster and
missing an opportunity for a better energy efficient placement to honour
uclamp_max setting.

max_spare_cap = 0;
cpu_cap = capacity_of(cpu) - cpu_util(p);  // 0 if cpu_util(p) is high

...

util_fits_cpu(...); // will return true if uclamp_max forces it to fit

...

// this logic will fail to update max_spare_cap_cpu if cpu_cap is 0
if (cpu_cap > max_spare_cap) {
max_spare_cap = cpu_cap;
max_spare_cap_cpu = cpu;
}

prev_spare_cap suffers from a similar problem.

Fix the logic by converting the variables into long and treating -1
value as 'not populated' instead of 0 which is a viable and correct
spare capacity value. We need to be careful signed comparison is used
when comparing with cpu_cap in one of the conditions.

Fixes: 1d42509e475c ("sched/fair: Make EAS wakeup placement consider uclamp restrictions")
Signed-off-by: Qais Yousef (Google) <qyousef@layalina.io>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Reviewed-by: Dietmar Eggemann <dietmar.eggemann@arm.com>
Acked-by: Peter Zijlstra (Intel) <peterz@infradead.org>
Link: https://lore.kernel.org/r/20230916232955.2099394-2-qyousef@layalina.io
Signed-off-by: Sasha Levin <sashal@kernel.org>

iov_iter, x86: Be consistent about the __user tag on copy_mc_to_user()

[ Upstream commit 066baf92bed934c9fb4bcee97a193f47aa63431c ]

copy_mc_to_user() has the destination marked __user on powerpc, but not on
x86; the latter results in a sparse warning in lib/iov_iter.c.

Fix this by applying the tag on x86 too.

Fixes: ec6347bb4339 ("x86, powerpc: Rename memcpy_mcsafe() to copy_mc_to_{user, kernel}()")
Signed-off-by: David Howells <dhowells@redhat.com>
Link: https://lore.kernel.org/r/20230925120309.1731676-3-dhowells@redhat.com
cc: Dan Williams <dan.j.williams@intel.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: "H. Peter Anvin" <hpa@zytor.com>
cc: Alexander Viro <viro@zeniv.linux.org.uk>
cc: Jens Axboe <axboe@kernel.dk>
cc: Christoph Hellwig <hch@lst.de>
cc: Christian Brauner <christian@brauner.io>
cc: Matthew Wilcox <willy@infradead.org>
cc: Linus Torvalds <torvalds@linux-foundation.org>
cc: David Laight <David.Laight@ACULAB.COM>
cc: x86@kernel.org
cc: linux-block@vger.kernel.org
cc: linux-fsdevel@vger.kernel.org
cc: linux-mm@kvack.org
Signed-off-by: Christian Brauner <brauner@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

sched/fair: Fix cfs_rq_is_decayed() on !SMP

[ Upstream commit c0490bc9bb62d9376f3dd4ec28e03ca0fef97152 ]

We don't need to maintain per-queue leaf_cfs_rq_list on !SMP, since
it's used for cfs_rq load tracking & balancing on SMP.

But sched debug interface uses it to print per-cfs_rq stats.

This patch fixes the !SMP version of cfs_rq_is_decayed(), so the
per-queue leaf_cfs_rq_list is also maintained correctly on !SMP,
to fix the warning in assert_list_leaf_cfs_rq().

Fixes: 0a00a354644e ("sched/fair: Delete useless condition in tg_unthrottle_up()")
Reported-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Signed-off-by: Chengming Zhou <zhouchengming@bytedance.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Tested-by: Leo Yu-Chi Liang <ycliang@andestech.com>
Reviewed-by: Vincent Guittot <vincent.guittot@linaro.org>
Closes: https://lore.kernel.org/all/ZN87UsqkWcFLDxea@swlinux02/
Link: https://lore.kernel.org/r/20230913132031.2242151-1-chengming.zhou@linux.dev
Signed-off-by: Sasha Levin <sashal@kernel.org>

sched/topology: Fix sched_numa_find_nth_cpu() in non-NUMA case

[ Upstream commit 8ab63d418d4339d996f80d02a00dbce0aa3ff972 ]

When CONFIG_NUMA is enabled, sched_numa_find_nth_cpu() searches for a
CPU in sched_domains_numa_masks. The masks includes only online CPUs,
so effectively offline CPUs are skipped.

When CONFIG_NUMA is disabled, the fallback function should be consistent.

Fixes: cd7f55359c90 ("sched: add sched_numa_find_nth_cpu()")
Signed-off-by: Yury Norov <yury.norov@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Mel Gorman <mgorman@suse.de>
Link: https://lore.kernel.org/r/20230819141239.287290-5-yury.norov@gmail.com
Signed-off-by: Sasha Levin <sashal@kernel.org>

sched/topology: Fix sched_numa_find_nth_cpu() in CPU-less case

[ Upstream commit 617f2c38cb5ce60226042081c09e2ee3a90d03f8 ]

When the node provided by user is CPU-less, corresponding record in
sched_domains_numa_masks is not set. Trying to dereference it in the
following code leads to kernel crash.

To avoid it, start searching from the nearest node with CPUs.

Fixes: cd7f55359c90 ("sched: add sched_numa_find_nth_cpu()")
Reported-by: Yicong Yang <yangyicong@hisilicon.com>
Reported-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Yury Norov <yury.norov@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Reviewed-by: Yicong Yang <yangyicong@hisilicon.com>
Cc: Mel Gorman <mgorman@suse.de>
Link: https://lore.kernel.org/r/20230819141239.287290-4-yury.norov@gmail.com
Closes: https://lore.kernel.org/lkml/CAAH8bW8C5humYnfpW3y5ypwx0E-09A3QxFE1JFzR66v+mO4XfA@mail.gmail.com/T/
Closes: https://lore.kernel.org/lkml/ZMHSNQfv39HN068m@yury-ThinkPad/T/#mf6431cb0b7f6f05193c41adeee444bc95bf2b1c4
Signed-off-by: Sasha Levin <sashal@kernel.org>

numa: Generalize numa_map_to_online_node()

[ Upstream commit b1f099b1cf51d553c510c6c8141c27d9ba7ea1fe ]

The function in fact searches the nearest node for a given one,
based on a N_ONLINE state. This is a common pattern to search
for a nearest node.

This patch converts numa_map_to_online_node() to numa_nearest_node()
so that others won't need to opencode the logic.

Signed-off-by: Yury Norov <yury.norov@gmail.com>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Mel Gorman <mgorman@suse.de>
Link: https://lore.kernel.org/r/20230819141239.287290-2-yury.norov@gmail.com
Stable-dep-of: 617f2c38cb5c ("sched/topology: Fix sched_numa_find_nth_cpu() in CPU-less case")
Signed-off-by: Sasha Levin <sashal@kernel.org>

hwmon: (nct6775) Fix incorrect variable reuse in fan_div calculation

commit 920057ad521dc8669e534736c2a12c14ec9fb2d7 upstream.

In the regmap conversion in commit 4ef2774511dc ("hwmon: (nct6775)
Convert register access to regmap API") I reused the 'reg' variable
for all three register reads in the fan speed calculation loop in
nct6775_update_device(), but failed to notice that the value from the
first one (data->REG_FAN[i]) is actually used in the call to
nct6775_select_fan_div() at the end of the loop body.  Since that
patch the register value passed to nct6775_select_fan_div() has been
(conditionally) incorrectly clobbered with the value of a different
register than intended, which has in at least some cases resulted in
fan speeds being adjusted down to zero.

Fix this by using dedicated temporaries for the two intermediate
register reads instead of 'reg'.

Signed-off-by: Zev Weiss <zev@bewilderbeest.net>
Fixes: 4ef2774511dc ("hwmon: (nct6775) Convert register access to regmap API")
Reported-by: Thomas Zajic <zlatko@gmx.at>
Tested-by: Thomas Zajic <zlatko@gmx.at>
Cc: stable@vger.kernel.org # v5.19+
Link: https://lore.kernel.org/r/20230929200822.964-2-zev@bewilderbeest.net
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Linux 6.6.1

Link: https://lore.kernel.org/r/20231106130257.903265688@linuxfoundation.org
Tested-by: Ronald Warsow <rwarsow@gmx.de>
Tested-by: SeongJae Park <sj@kernel.org>
Tested-by: Florian Fainelli <florian.fainelli@broadcom.com>
Tested-by: Allen Pais <apais@linux.microsoft.com>
Tested-by: Rudi Heitbaum <rudi@heitbaum.com>
Tested-by: Bagas Sanjaya <bagasdotme@gmail.com>
Tested-by: Ron Economos <re@w6rz.net>
Tested-by: Takeshi Ogasawara <takeshi.ogasawara@futuring-girl.com>
Tested-by: Shuah Khan <skhan@linuxfoundation.org>
Tested-by: Ricardo B. Marliere <ricardo@marliere.net>
Tested-by: Guenter Roeck <linux@roeck-us.net>
Tested-by: Linux Kernel Functional Testing <lkft@linaro.org>
Tested-by: Jon Hunter <jonathanh@nvidia.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ASoC: SOF: sof-pci-dev: Fix community key quirk detection

commit 7dd692217b861a8292ff8ac2c9d4458538fd6b96 upstream.

Some Chromebooks do not populate the product family DMI value resulting
in firmware load failures.

Add another quirk detection entry that looks for "Google" in the BIOS
version. Theoretically, PRODUCT_FAMILY could be replaced with
BIOS_VERSION, but it is left as a quirk to be conservative.

Cc: stable@vger.kernel.org
Signed-off-by: Mark Hasemeyer <markhas@chromium.org>
Acked-by: Curtis Malainey <cujomalainey@chromium.org>
Link: https://lore.kernel.org/r/20231020145953.v1.1.Iaf5702dc3f8af0fd2f81a22ba2da1a5e15b3604c@changeid
Signed-off-by: Mark Brown <broonie@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: hda: intel-dsp-config: Fix JSL Chromebook quirk detection

commit 7c05b44e1a50d9cbfc4f731dddc436a24ddc129a upstream.

Some Jasperlake Chromebooks overwrite the system vendor DMI value to the
name of the OEM that manufactured the device. This breaks Chromebook
quirk detection as it expects the system vendor to be "Google".

Add another quirk detection entry that looks for "Google" in the BIOS
version.

Cc: stable@vger.kernel.org
Signed-off-by: Mark Hasemeyer <markhas@chromium.org>
Reviewed-by: Pierre-Louis Bossart <pierre-louis.bossart@linux.intel.com>
Link: https://lore.kernel.org/r/20231018235944.1860717-1-markhas@chromium.org
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

serial: core: Fix runtime PM handling for pending tx

commit 6f699743aebf07538e506a46c5965eb8bdd2c716 upstream.

Richard reported that a serial port may end up sometimes with tx data
pending in the buffer for long periods of time.

Turns out we bail out early on any errors from pm_runtime_get(),
including -EINPROGRESS. To fix the issue, we need to ignore -EINPROGRESS
as we only care about the runtime PM usage count at this point. We check
for an active runtime PM state later on for tx.

Fixes: 84a9582fd203 ("serial: core: Start managing serial controllers to enable runtime PM")
Cc: stable <stable@kernel.org>
Reported-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Cc: Bruce Ashfield <bruce.ashfield@gmail.com>
Cc: Mikko Rapeli <mikko.rapeli@linaro.org>
Cc: Paul Gortmaker <paul.gortmaker@windriver.com>
Cc: Randy MacLeod <randy.macleod@windriver.com>
Signed-off-by: Tony Lindgren <tony@atomide.com>
Tested-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Link: https://lore.kernel.org/r/20231023074856.61896-1-tony@atomide.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

misc: pci_endpoint_test: Add deviceID for J721S2 PCIe EP device support

commit 8293703a492ae97c86af27c75b76e6239ec86483 upstream.

Add DEVICE_ID for J721S2 and enable support for endpoints configured
with this DEVICE_ID in the pci_endpoint_test driver.

Signed-off-by: Siddharth Vadapalli <s-vadapalli@ti.com>
Cc: stable <stable@kernel.org>
Reviewed-by: Manivannan Sadhasivam <manivannan.sadhasivam@linaro.org>
Link: https://lore.kernel.org/r/20231020120248.3168406-1-s-vadapalli@ti.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

dt-bindings: serial: rs485: Add rs485-rts-active-high

commit 0c01b20fb50ba63c03841aa83070dc59c3b1b02f upstream.

Add rs485-rts-active-high property, this is a legacy property
used by 8250_omap.

This fixes the following make dt_binding_check warning:

Documentation/devicetree/bindings/serial/8250_omap.yaml:
rs485-rts-active-high: missing type definition

Reported-by: Geert Uytterhoeven <geert@linux-m68k.org>
Closes: https://lore.kernel.org/all/CAMuHMdUkPiA=o_QLyuwsTYW7y1ksCjHAqyNSHFx2QZ-dP-HGsQ@mail.gmail.com/
Fixes: 403e97d6ab2c ("dt-bindings: serial: 8250_omap: add rs485-rts-active-high")
Cc: stable <stable@kernel.org>
Signed-off-by: Francesco Dolcini <francesco.dolcini@toradex.com>
Reviewed-by: Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>
Link: https://lore.kernel.org/r/20231019154834.41721-1-francesco@dolcini.it
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Add Brainboxes Oxford Semiconductor-based quirks

commit e4876dacaca46a1b09f9b417480924ab12019a5b upstream.

Some of the later revisions of the Brainboxes PX cards are based
on the Oxford Semiconductor chipset. Due to the chip's unique setup
these cards need to be initialised.
Previously these were tested against a reference card with the same broken
baudrate on another PC, cancelling out the effect. With this patch they
work and can transfer/receive find against an FTDI-based device.

Add all of the cards which require this setup to the quirks table.
Thanks to Maciej W. Rozycki for clarification on this chip.

Fixes: ef5a03a26c87 ("tty: 8250: Add support for Brainboxes PX cards.")
Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB7899D222A4AB2A4E8C57108FC4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Add support for Intashield IX cards

commit 62d2ec2ded278c7512d91ca7bf8eb9bac46baf90 upstream.

Add support for the IX-100, IX-200 and IX-400 serial cards.

Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB7899614E5837E82A03272A4BC4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Add support for additional Brainboxes PX cards

commit 9604884e592cd04ead024c9737c67a77f175cab9 upstream.

Add support for some more of the Brainboxes PX (PCIe) range
of serial cards, namely
PX-275/PX-279, PX-475 (serial port, not LPT), PX-820,
PX-803/PX-857 (additional ID).

Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB78996BEC353FB346FC35444BC4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Fix up PX-803/PX-857

commit ee61337b934c99c2611e0a945d592019b2e00c82 upstream.

The PX-803/PX-857 are variants of each other, add a note.
Additionally fix up the port counts for the card (2, not 1).

Fixes: ef5a03a26c87 ("tty: 8250: Add support for Brainboxes PX cards.")
Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB789978C8ED872FB4B014E132C4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Fix port count of PX-257

commit d0ff5b24c2f112f29dea4c38b3bac9597b1be9ba upstream.

The port count of the PX-257 Rev3 is actually 2, not 4.

Fixes: ef5a03a26c87 ("tty: 8250: Add support for Brainboxes PX cards.")
Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB7899C804D9F04E727B5A0E8FC4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Add support for Intashield IS-100

commit 4d994e3cf1b541ff32dfb03fbbc60eea68f9645b upstream.

Add support for the Intashield IS-100 1 port serial card.

Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB7899A0E0CDAA505AF5A874CDC4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Add support for Brainboxes UP cards

commit 2c6fec1e1532f15350be7e14ba6b88a39d289fe4 upstream.

Add support for the Brainboxes UP (powered PCI) range of
cards, namely UP-189, UP-200, UP-869 and UP-880.

Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB7899B5B59FF3D8587E88C117C4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Add support for additional Brainboxes UC cards

commit c563db486db7d245c0e2f319443417ae8e692f7f upstream.

Add device IDs for some more Brainboxes UC cards, namely
UC-235/UC-246, UC-253/UC-734, UC-302, UC-313, UC-346, UC-357,
UC-607 and UC-836.

Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB789969998A6C3FAFCD95C85DC4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: 8250: Remove UC-257 and UC-431

commit 33092fb3af51deb80849e90a17bada44bbcde6b3 upstream.

The UC-257 is a serial + LPT card, so remove it from this driver.
A patch has been submitted to add it to parport_serial instead.

Additionaly, the UC-431 does not use this card ID, only the UC-420
does. The 431 is a 3-port card and there is no generic 3-port configuration
available, so remove reference to it from this driver.

Fixes: 152d1afa834c ("tty: Add support for Brainboxes UC cards.")
Cc: stable@vger.kernel.org
Signed-off-by: Cameron Williams <cang1@live.co.uk>
Link: https://lore.kernel.org/r/DU0PR02MB78995ADF7394C74AD4CF3357C4DBA@DU0PR02MB7899.eurprd02.prod.outlook.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tty: n_gsm: fix race condition in status line change on dead connections

commit 3a75b205de43365f80a33b98ec9289785da56243 upstream.

gsm_cleanup_mux() cleans up the gsm by closing all DLCIs, stopping all
timers, removing the virtual tty devices and clearing the data queues.
This procedure, however, may cause subsequent changes of the virtual modem
status lines of a DLCI. More data is being added the outgoing data queue
and the deleted kick timer is restarted to handle this. At this point many
resources have already been removed by the cleanup procedure. Thus, a
kernel panic occurs.

Fix this by proving in gsm_modem_update() that the cleanup procedure has
not been started and the mux is still alive.

Note that writing to a virtual tty is already protected by checks against
the DLCI specific connection state.

Fixes: c568f7086c6e ("tty: n_gsm: fix missing timer to handle stalled links")
Cc: stable <stable@kernel.org>
Signed-off-by: Daniel Starke <daniel.starke@siemens.com>
Link: https://lore.kernel.org/r/20231026055844.3127-1-daniel.starke@siemens.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

Bluetooth: hci_bcm4377: Mark bcm4378/bcm4387 as BROKEN_LE_CODED

commit 41e9cdea9c4ab6606ca462ff4ec901a82d022c05 upstream.

bcm4378 and bcm4387 claim to support LE Coded PHY but fail to pair
(reliably) with BLE devices if it is enabled.
On bcm4378 pairing usually succeeds after 2-3 tries. On bcm4387
pairing appears to be completely broken.

Cc: stable@vger.kernel.org # 6.4.y+
Link: https://discussion.fedoraproject.org/t/mx-master-3-bluetooth-mouse-doesnt-connect/87072/33
Link: https://github.com/AsahiLinux/linux/issues/177
Fixes: 288c90224eec ("Bluetooth: Enable all supported LE PHY by default")
Signed-off-by: Janne Grunau <j@jannau.net>
Reviewed-by: Eric Curtin <ecurtin@redhat.com>
Reviewed-by: Neal Gompa <neal@gompa.dev>
Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: raw-gadget: properly handle interrupted requests

commit e8033bde451eddfb9b1bbd6e2d848c1b5c277222 upstream.

Currently, if a USB request that was queued by Raw Gadget is interrupted
(via a signal), wait_for_completion_interruptible returns -ERESTARTSYS.
Raw Gadget then attempts to propagate this value to userspace as a return
value from its ioctls. However, when -ERESTARTSYS is returned by a syscall
handler, the kernel internally restarts the syscall.

This doesn't allow userspace applications to interrupt requests queued by
Raw Gadget (which is required when the emulated device is asked to switch
altsettings). It also violates the implied interface of Raw Gadget that a
single ioctl must only queue a single USB request.

Instead, make Raw Gadget do what GadgetFS does: check whether the request
was interrupted (dequeued with status == -ECONNRESET) and report -EINTR to
userspace.

Fixes: f2c2e717642c ("usb: gadget: add raw-gadget interface")
Cc: stable <stable@kernel.org>
Signed-off-by: Andrey Konovalov <andreyknvl@gmail.com>
Link: https://lore.kernel.org/r/0db45b1d7cc466e3d4d1ab353f61d63c977fbbc5.1698350424.git.andreyknvl@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: typec: tcpm: Fix NULL pointer dereference in tcpm_pd_svdm()

commit 4987daf86c152ff882d51572d154ad12e4ff3a4b upstream.

It is possible that typec_register_partner() returns ERR_PTR on failure.
When port->partner is an error, a NULL pointer dereference may occur as
shown below.

[91222.095236][  T319] typec port0: failed to register partner (-17)
...
[91225.061491][  T319] Unable to handle kernel NULL pointer dereference
at virtual address 000000000000039f
[91225.274642][  T319] pc : tcpm_pd_data_request+0x310/0x13fc
[91225.274646][  T319] lr : tcpm_pd_data_request+0x298/0x13fc
[91225.308067][  T319] Call trace:
[91225.308070][  T319]  tcpm_pd_data_request+0x310/0x13fc
[91225.308073][  T319]  tcpm_pd_rx_handler+0x100/0x9e8
[91225.355900][  T319]  kthread_worker_fn+0x178/0x58c
[91225.355902][  T319]  kthread+0x150/0x200
[91225.355905][  T319]  ret_from_fork+0x10/0x30

Add a check for port->partner to avoid dereferencing a NULL pointer.

Fixes: 5e1d4c49fbc8 ("usb: typec: tcpm: Determine common SVDM Version")
Cc: stable@vger.kernel.org
Signed-off-by: Jimmy Hu <hhhuuu@google.com>
Link: https://lore.kernel.org/r/20231020012132.100960-1-hhhuuu@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: typec: tcpm: Add additional checks for contaminant

commit 1a4a2df07c1f087704c24282cebe882268e38146 upstream.

When transitioning from SNK_DEBOUNCED to unattached, its worthwhile to
check for contaminant to mitigate wakeups.

```
[81334.219571] Start toggling
[81334.228220] CC1: 0 -> 0, CC2: 0 -> 0 [state TOGGLING, polarity 0, disconnected]
[81334.305147] CC1: 0 -> 0, CC2: 0 -> 3 [state TOGGLING, polarity 0, connected]
[81334.305162] state change TOGGLING -> SNK_ATTACH_WAIT [rev3 NONE_AMS]
[81334.305187] pending state change SNK_ATTACH_WAIT -> SNK_DEBOUNCED @ 170 ms [rev3 NONE_AMS]
[81334.475515] state change SNK_ATTACH_WAIT -> SNK_DEBOUNCED [delayed 170 ms]
[81334.486480] CC1: 0 -> 0, CC2: 3 -> 0 [state SNK_DEBOUNCED, polarity 0, disconnected]
[81334.486495] state change SNK_DEBOUNCED -> SNK_DEBOUNCED [rev3 NONE_AMS]
[81334.486515] pending state change SNK_DEBOUNCED -> SNK_UNATTACHED @ 20 ms [rev3 NONE_AMS]
[81334.506621] state change SNK_DEBOUNCED -> SNK_UNATTACHED [delayed 20 ms]
[81334.506640] Start toggling
[81334.516972] CC1: 0 -> 0, CC2: 0 -> 0 [state TOGGLING, polarity 0, disconnected]
[81334.592759] CC1: 0 -> 0, CC2: 0 -> 3 [state TOGGLING, polarity 0, connected]
[81334.592773] state change TOGGLING -> SNK_ATTACH_WAIT [rev3 NONE_AMS]
[81334.592792] pending state change SNK_ATTACH_WAIT -> SNK_DEBOUNCED @ 170 ms [rev3 NONE_AMS]
[81334.762940] state change SNK_ATTACH_WAIT -> SNK_DEBOUNCED [delayed 170 ms]
[81334.773557] CC1: 0 -> 0, CC2: 3 -> 0 [state SNK_DEBOUNCED, polarity 0, disconnected]
[81334.773570] state change SNK_DEBOUNCED -> SNK_DEBOUNCED [rev3 NONE_AMS]
[81334.773588] pending state change SNK_DEBOUNCED -> SNK_UNATTACHED @ 20 ms [rev3 NONE_AMS]
[81334.793672] state change SNK_DEBOUNCED -> SNK_UNATTACHED [delayed 20 ms]
[81334.793681] Start toggling
[81334.801840] CC1: 0 -> 0, CC2: 0 -> 0 [state TOGGLING, polarity 0, disconnected]
[81334.878655] CC1: 0 -> 0, CC2: 0 -> 3 [state TOGGLING, polarity 0, connected]
[81334.878672] state change TOGGLING -> SNK_ATTACH_WAIT [rev3 NONE_AMS]
[81334.878696] pending state change SNK_ATTACH_WAIT -> SNK_DEBOUNCED @ 170 ms [rev3 NONE_AMS]
[81335.048968] state change SNK_ATTACH_WAIT -> SNK_DEBOUNCED [delayed 170 ms]
[81335.060684] CC1: 0 -> 0, CC2: 3 -> 0 [state SNK_DEBOUNCED, polarity 0, disconnected]
[81335.060754] state change SNK_DEBOUNCED -> SNK_DEBOUNCED [rev3 NONE_AMS]
[81335.060775] pending state change SNK_DEBOUNCED -> SNK_UNATTACHED @ 20 ms [rev3 NONE_AMS]
[81335.080884] state change SNK_DEBOUNCED -> SNK_UNATTACHED [delayed 20 ms]
[81335.080900] Start toggling
```

Cc: stable@vger.kernel.org
Fixes: 599f008c257d ("usb: typec: tcpm: Add callbacks to mitigate wakeups due to contaminant")
Signed-off-by: Badhri Jagan Sridharan <badhri@google.com>
Acked-by: Heikki Krogerus <heikki.krogerus@linux.intel.com>
Link: https://lore.kernel.org/r/20231015053108.2349570-1-badhri@google.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

usb: storage: set 1.50 as the lower bcdDevice for older "Super Top" compatibility

commit 0e3139e6543b241b3e65956a55c712333bef48ac upstream.

Change lower bcdDevice value for "Super Top USB 2.0  SATA BRIDGE" to match
1.50. I have such an older device with bcdDevice=1.50 and it will not work
otherwise.

Cc: stable@vger.kernel.org
Signed-off-by: Liha Sikanen <lihasika@gmail.com>
Link: https://lore.kernel.org/r/ccf7d12a-8362-4916-b3e0-f4150f54affd@gmail.com
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

PCI: Prevent xHCI driver from claiming AMD VanGogh USB3 DRD device

commit 7e6f3b6d2c352b5fde37ce3fed83bdf6172eebd4 upstream.

The AMD VanGogh SoC contains a DesignWare USB3 Dual-Role Device that can be
operated as either a USB Host or a USB Device, similar to on the AMD Nolan
platform.

be6646bfbaec ("PCI: Prevent xHCI driver from claiming AMD Nolan USB3 DRD
device") added a quirk to let the dwc3 driver claim the Nolan device since
it provides more specific support.

Extend that quirk to include the VanGogh SoC USB3 device.

Link: https://lore.kernel.org/r/20230927202212.2388216-1-vi@endrift.com
Signed-off-by: Vicki Pfau <vi@endrift.com>
[bhelgaas: include be6646bfbaec reference, add stable tag]
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Cc: stable@vger.kernel.org # v3.19+
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

ALSA: usb-audio: add quirk flag to enable native DSD for McIntosh devices

commit 99248c8902f505ec064cf2b0f74629016f2f4c82 upstream.

McIntosh devices supporting native DSD require the feature to be
explicitly exposed. Add a flag that fixes an issue where DSD audio was
defaulting to DSD over PCM instead of delivering raw DSD data.

Signed-off-by: Max McCarthy <mmccarthy@mcintoshlabs.com>
Cc: <stable@vger.kernel.org>
Link: https://lore.kernel.org/r/BL0PR13MB4433226005162D186A8DFF4AD6DFA@BL0PR13MB4433.namprd13.prod.outlook.com
Signed-off-by: Takashi Iwai <tiwai@suse.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

eventfs: Use simple_recursive_removal() to clean up dentries

commit 407c6726ca71b33330d2d6345d9ea7ebc02575e9 upstream

Looking at how dentry is removed via the tracefs system, I found that
eventfs does not do everything that it did under tracefs. The tracefs
removal of a dentry calls simple_recursive_removal() that does a lot more
than a simple d_invalidate().

As it should be a requirement that any eventfs_inode that has a dentry, so
does its parent. When removing a eventfs_inode, if it has a dentry, a call
to simple_recursive_removal() on that dentry should clean up all the
dentries underneath it.

Add WARN_ON_ONCE() to check for the parent having a dentry if any children
do.

Link: https://lore.kernel.org/all/20231101022553.GE1957730@ZenIV/
Link: https://lkml.kernel.org/r/20231101172650.552471568@goodmis.org
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Fixes: 5bdcd5f5331a2 ("eventfs: Implement removal of meta data from eventfs")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

eventfs: Delete eventfs_inode when the last dentry is freed

commit 020010fbfa202aa528a52743eba4ab0da3400a4e upstream

There exists a race between holding a reference of an eventfs_inode dentry
and the freeing of the eventfs_inode. If user space has a dentry held long
enough, it may still be able to access the dentry's eventfs_inode after it
has been freed.

To prevent this, have he eventfs_inode freed via the last dput() (or via
RCU if the eventfs_inode does not have a dentry).

This means reintroducing the eventfs_inode del_list field at a temporary
place to put the eventfs_inode. It needs to mark it as freed (via the
list) but also must invalidate the dentry immediately as the return from
eventfs_remove_dir() expects that they are. But the dentry invalidation
must not be called under the eventfs_mutex, so it must be done after the
eventfs_inode is marked as free (put on a deletion list).

Link: https://lkml.kernel.org/r/20231101172650.123479767@goodmis.org
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ajay Kaher <akaher@vmware.com>
Fixes: 5bdcd5f5331a2 ("eventfs: Implement removal of meta data from eventfs")
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

eventfs: Save ownership and mode

commit 28e12c09f5aa081b2d13d1340e3610070b6c624d upstream

Now that inodes and dentries are created on the fly, they are also
reclaimed on memory pressure. Since the ownership and file mode are saved
in the inode, if they are freed, any changes to the ownership and mode
will be lost.

To counter this, if the user changes the permissions or ownership, save
them, and when creating the inodes again, restore those changes.

Link: https://lkml.kernel.org/r/20231101172649.691841445@goodmis.org
Cc: stable@vger.kernel.org
Cc: Ajay Kaher <akaher@vmware.com>
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Fixes: 63940449555e7 ("eventfs: Implement eventfs lookup, read, open functions")
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

eventfs: Remove "is_freed" union with rcu head

commit f2f496370afcbc5227d7002da28c74b91fed12ff upstream

The eventfs_inode->is_freed was a union with the rcu_head with the
assumption that when it was on the srcu list the head would contain a
pointer which would make "is_freed" true. But that was a wrong assumption
as the rcu head is a single link list where the last element is NULL.

Instead, split the nr_entries integer so that "is_freed" is one bit and
the nr_entries is the next 31 bits. As there shouldn't be more than 10
(currently there's at most 5 to 7 depending on the config), this should
not be a problem.

Link: https://lkml.kernel.org/r/20231101172649.049758712@goodmis.org
Cc: stable@vger.kernel.org
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Cc: Ajay Kaher <akaher@vmware.com>
Fixes: 63940449555e7 ("eventfs: Implement eventfs lookup, read, open functions")
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

tracing: Have trace_event_file have ref counters

commit bb32500fb9b78215e4ef6ee8b4345c5f5d7eafb4 upstream

The following can crash the kernel:

 # cd /sys/kernel/tracing
 # echo 'p:sched schedule' > kprobe_events
 # exec 5>>events/kprobes/sched/enable
 # > kprobe_events
 # exec 5>&-

The above commands:

 1. Change directory to the tracefs directory
 2. Create a kprobe event (doesn't matter what one)
 3. Open bash file descriptor 5 on the enable file of the kprobe event
 4. Delete the kprobe event (removes the files too)
 5. Close the bash file descriptor 5

The above causes a crash!

 BUG: kernel NULL pointer dereference, address: 0000000000000028
 #PF: supervisor read access in kernel mode
 #PF: error_code(0x0000) - not-present page
 PGD 0 P4D 0
 Oops: 0000 [#1] PREEMPT SMP PTI
 CPU: 6 PID: 877 Comm: bash Not tainted 6.5.0-rc4-test-00008-g2c6b6b1029d4-dirty #186
 Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
 RIP: 0010:tracing_release_file_tr+0xc/0x50

What happens here is that the kprobe event creates a trace_event_file
"file" descriptor that represents the file in tracefs to the event. It
maintains state of the event (is it enabled for the given instance?).
Opening the "enable" file gets a reference to the event "file" descriptor
via the open file descriptor. When the kprobe event is deleted, the file is
also deleted from the tracefs system which also frees the event "file"
descriptor.

But as the tracefs file is still opened by user space, it will not be
totally removed until the final dput() is called on it. But this is not
true with the event "file" descriptor that is already freed. If the user
does a write to or simply closes the file descriptor it will reference the
event "file" descriptor that was just freed, causing a use-after-free bug.

To solve this, add a ref count to the event "file" descriptor as well as a
new flag called "FREED". The "file" will not be freed until the last
reference is released. But the FREE flag will be set when the event is
removed to prevent any more modifications to that event from happening,
even if there's still a reference to the event "file" descriptor.

Link: https://lore.kernel.org/linux-trace-kernel/20231031000031.1e705592@gandalf.local.home/
Link: https://lore.kernel.org/linux-trace-kernel/20231031122453.7a48b923@gandalf.local.home
Cc: stable@vger.kernel.org
Cc: Mark Rutland <mark.rutland@arm.com>
Fixes: f5ca233e2e66d ("tracing: Increase trace array ref count on enable and filter files")
Reported-by: Beau Belgrave <beaub@linux.microsoft.com>
Tested-by: Beau Belgrave <beaub@linux.microsoft.com>
Reviewed-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>

perf evlist: Avoid frequency mode for the dummy event

[ Upstream commit f9cdeb58a9cf46c09b56f5f661ea8da24b6458c3 ]

Dummy events are created with an attribute where the period and freq
are zero. evsel__config will then see the uninitialized values and
initialize them in evsel__default_freq_period. As fequency mode is
used by default the dummy event would be set to use frequency
mode. However, this has no effect on the dummy event but does cause
unnecessary timers/interrupts. Avoid this overhead by setting the
period to 1 for dummy events.

evlist__add_aux_dummy calls evlist__add_dummy then sets freq=0 and
period=1. This isn't necessary after this change and so the setting is
removed.

From Stephane:

The dummy event is not counting anything. It is used to collect mmap
records and avoid a race condition during the synthesize mmap phase of
perf record. As such, it should not cause any overhead during active
profiling. Yet, it did. Because of a bug the dummy event was
programmed as a sampling event in frequency mode. Events in that mode
incur more kernel overheads because on timer tick, the kernel has to
look at the number of samples for each event and potentially adjust
the sampling period to achieve the desired frequency. The dummy event
was therefore adding a frequency event to task and ctx contexts we may
otherwise not have any, e.g.,

  perf record -a -e cpu/event=0x3c,period=10000000/.

On each timer tick the perf_adjust_freq_unthr_context() is invoked and
if ctx->nr_freq is non-zero, then the kernel will loop over ALL the
events of the context looking for frequency mode ones. In doing, so it
locks the context, and enable/disable the PMU of each hw event. If all
the events of the context are in period mode, the kernel will have to
traverse the list for nothing incurring overhead. The overhead is
multiplied by a very large factor when this happens in a guest kernel.
There is no need for the dummy event to be in frequency mode, it does
not count anything and therefore should not cause extra overhead for
no reason.

Fixes: 5bae0250237f ("perf evlist: Introduce perf_evlist__new_dummy constructor")
Reported-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Ian Rogers <irogers@google.com>
Acked-by: Adrian Hunter <adrian.hunter@intel.com>
Cc: Yang Jihong <yangjihong1@huawei.com>
Cc: Kan Liang <kan.liang@linux.intel.com>
Link: https://lore.kernel.org/r/20230916035640.1074422-1-irogers@google.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>

power: supply: core: Use blocking_notifier_call_chain to avoid RCU complaint

[ Upstream commit bbaa6ffa5b6c9609d3b3c431c389b407eea5441f ]

AMD PMF driver can cause the following warning:
[  196.159546] ------------[ cut here ]------------
[  196.159556] Voluntary context switch within RCU read-side critical section!
[  196.159571] WARNING: CPU: 0 PID: 9 at kernel/rcu/tree_plugin.h:320 rcu_note_context_switch+0x43d/0x560
[  196.159604] Modules linked in: nvme_fabrics ccm rfcomm snd_hda_scodec_cs35l41_spi cmac algif_hash algif_skcipher af_alg bnep joydev btusb btrtl uvcvideo btintel btbcm videobuf2_vmalloc intel_rapl_msr btmtk videobuf2_memops uvc videobuf2_v4l2 intel_rapl_common binfmt_misc hid_sensor_als snd_sof_amd_vangogh hid_sensor_trigger bluetooth industrialio_triggered_buffer videodev snd_sof_amd_rembrandt hid_sensor_iio_common amdgpu ecdh_generic kfifo_buf videobuf2_common hp_wmi kvm_amd sparse_keymap snd_sof_amd_renoir wmi_bmof industrialio ecc mc nls_iso8859_1 kvm snd_sof_amd_acp irqbypass snd_sof_xtensa_dsp crct10dif_pclmul crc32_pclmul mt7921e snd_sof_pci snd_ctl_led polyval_clmulni mt7921_common polyval_generic snd_sof ghash_clmulni_intel mt792x_lib mt76_connac_lib sha512_ssse3 snd_sof_utils aesni_intel snd_hda_codec_realtek crypto_simd mt76 snd_hda_codec_generic cryptd snd_soc_core snd_hda_codec_hdmi rapl ledtrig_audio input_leds snd_compress i2c_algo_bit drm_ttm_helper mac80211 snd_pci_ps hid_multitouch ttm drm_exec
[  196.159970]  drm_suballoc_helper snd_rpl_pci_acp6x amdxcp drm_buddy snd_hda_intel snd_acp_pci snd_hda_scodec_cs35l41_i2c serio_raw gpu_sched snd_hda_scodec_cs35l41 snd_acp_legacy_common snd_intel_dspcfg snd_hda_cs_dsp_ctls snd_hda_codec libarc4 drm_display_helper snd_pci_acp6x cs_dsp snd_hwdep snd_soc_cs35l41_lib video k10temp snd_pci_acp5x thunderbolt snd_hda_core drm_kms_helper cfg80211 snd_seq snd_rn_pci_acp3x snd_pcm snd_acp_config cec snd_soc_acpi snd_seq_device rc_core ccp snd_pci_acp3x snd_timer snd soundcore wmi amd_pmf platform_profile amd_pmc mac_hid serial_multi_instantiate wireless_hotkey hid_sensor_hub sch_fq_codel msr parport_pc ppdev lp parport efi_pstore ip_tables x_tables autofs4 btrfs blake2b_generic raid10 raid456 async_raid6_recov async_memcpy async_pq async_xor async_tx libcrc32c xor raid6_pq raid1 raid0 multipath linear dm_mirror dm_region_hash dm_log cdc_ether usbnet r8152 mii hid_generic nvme i2c_hid_acpi i2c_hid nvme_core i2c_piix4 xhci_pci amd_sfh drm xhci_pci_renesas nvme_common hid
[  196.160382] CPU: 0 PID: 9 Comm: kworker/0:1 Not tainted 6.6.0-rc1 #4
[  196.160397] Hardware name: HP HP EliteBook 845 14 inch G10 Notebook PC/8B6E, BIOS V82 Ver. 01.02.00 08/24/2023
[  196.160405] Workqueue: events power_supply_changed_work
[  196.160426] RIP: 0010:rcu_note_context_switch+0x43d/0x560
[  196.160440] Code: 00 48 89 be 40 08 00 00 48 89 86 48 08 00 00 48 89 10 e9 63 fe ff ff 48 c7 c7 10 e7 b0 9e c6 05 e8 d8 20 02 01 e8 13 0f f3 ff <0f> 0b e9 27 fc ff ff a9 ff ff ff 7f 0f 84 cf fc ff ff 65 48 8b 3c
[  196.160450] RSP: 0018:ffffc900001878f0 EFLAGS: 00010046
[  196.160462] RAX: 0000000000000000 RBX: ffff88885e834040 RCX: 0000000000000000
[  196.160470] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000000
[  196.160476] RBP: ffffc90000187910 R08: 0000000000000000 R09: 0000000000000000
[  196.160482] R10: 0000000000000000 R11: 0000000000000000 R12: 0000000000000000
[  196.160488] R13: 0000000000000000 R14: ffff888100990000 R15: ffff888100990000
[  196.160495] FS:  0000000000000000(0000) GS:ffff88885e800000(0000) knlGS:0000000000000000
[  196.160504] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  196.160512] CR2: 000055cb053c8246 CR3: 000000013443a000 CR4: 0000000000750ef0
[  196.160520] PKRU: 55555554
[  196.160526] Call Trace:
[  196.160532]  <TASK>
[  196.160548]  ? show_regs+0x72/0x90
[  196.160570]  ? rcu_note_context_switch+0x43d/0x560
[  196.160580]  ? __warn+0x8d/0x160
[  196.160600]  ? rcu_note_context_switch+0x43d/0x560
[  196.160613]  ? report_bug+0x1bb/0x1d0
[  196.160637]  ? handle_bug+0x46/0x90
[  196.160658]  ? exc_invalid_op+0x19/0x80
[  196.160675]  ? asm_exc_invalid_op+0x1b/0x20
[  196.160709]  ? rcu_note_context_switch+0x43d/0x560
[  196.160727]  __schedule+0xb9/0x15f0
[  196.160746]  ? srso_alias_return_thunk+0x5/0x7f
[  196.160765]  ? srso_alias_return_thunk+0x5/0x7f
[  196.160778]  ? acpi_ns_search_one_scope+0xbe/0x270
[  196.160806]  schedule+0x68/0x110
[  196.160820]  schedule_timeout+0x151/0x160
[  196.160829]  ? srso_alias_return_thunk+0x5/0x7f
[  196.160842]  ? srso_alias_return_thunk+0x5/0x7f
[  196.160855]  ? acpi_ns_lookup+0x3c5/0xa90
[  196.160878]  __down_common+0xff/0x220
[  196.160905]  __down_timeout+0x16/0x30
[  196.160920]  down_timeout+0x64/0x70
[  196.160938]  acpi_os_wait_semaphore+0x85/0x200
[  196.160959]  acpi_ut_acquire_mutex+0x9e/0x280
[  196.160979]  acpi_ex_enter_interpreter+0x2d/0xb0
[  196.160992]  acpi_ns_evaluate+0x2f0/0x5f0
[  196.161005]  acpi_evaluate_object+0x172/0x490
[  196.161018]  ? acpi_os_signal_semaphore+0x8a/0xd0
[  196.161038]  acpi_evaluate_integer+0x52/0xe0
[  196.161055]  ? kfree+0x79/0x120
[  196.161071]  ? srso_alias_return_thunk+0x5/0x7f
[  196.161089]  acpi_ac_get_state.part.0+0x27/0x80
[  196.161110]  get_ac_property+0x5c/0x70
[  196.161127]  ? __pfx___power_supply_is_system_supplied+0x10/0x10
[  196.161146]  __power_supply_is_system_supplied+0x44/0xb0
[  196.161166]  class_for_each_device+0x124/0x160
[  196.161184]  ? acpi_ac_get_state.part.0+0x27/0x80
[  196.161203]  ? srso_alias_return_thunk+0x5/0x7f
[  196.161223]  power_supply_is_system_supplied+0x3c/0x70
[  196.161243]  amd_pmf_get_power_source+0xe/0x20 [amd_pmf]
[  196.161276]  amd_pmf_power_slider_update_event+0x49/0x90 [amd_pmf]
[  196.161310]  amd_pmf_pwr_src_notify_call+0xe7/0x100 [amd_pmf]
[  196.161340]  notifier_call_chain+0x5f/0xe0
[  196.161362]  atomic_notifier_call_chain+0x33/0x60
[  196.161378]  power_supply_changed_work+0x84/0x110
[  196.161394]  process_one_work+0x178/0x360
[  196.161412]  ? __pfx_worker_thread+0x10/0x10
[  196.161424]  worker_thread+0x307/0x430
[  196.161440]  ? __pfx_worker_thread+0x10/0x10
[  196.161451]  kthread+0xf4/0x130
[  196.161467]  ? __pfx_kthread+0x10/0x10
[  196.161486]  ret_from_fork+0x43/0x70
[  196.161502]  ? __pfx_kthread+0x10/0x10
[  196.161518]  ret_from_fork_asm+0x1b/0x30
[  196.161558]  </TASK>
[  196.161562] ---[ end trace 0000000000000000 ]---

Since there's no guarantee that all the callbacks can work in atomic
context, switch to use blocking_notifier_call_chain to relax the
constraint.

Signed-off-by: Kai-Heng Feng <kai.heng.feng@canonical.com>
Reported-by: Allen Zhong <allen@atr.me>
Fixes: 4c71ae414474 ("platform/x86/amd/pmf: Add support SPS PMF feature")
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=217571
Reviewed-by: Mario Limonciello <mario.limonciello@amd.com>
Link: https://lore.kernel.org/r/20230913033233.602986-1-kai.heng.feng@canonical.com
Signed-off-by: Sebastian Reichel <sebastian.reichel@collabora.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

drm/amd/display: Don't use fsleep for PSR exit waits

[ Upstream commit 79df45dc4bfb13d9bd3a75338b9d9dab948be3d6 ]

[Why]
These functions can be called from high IRQ levels and the OS will hang
if it tries to use a usleep_highres or a msleep.

[How]
Replace the fsleep with a udelay.

Reviewed-by: Aric Cyr <aric.cyr@amd.com>
Acked-by: Tom Chung <chiahsuan.chung@amd.com>
Signed-off-by: Nicholas Kazlauskas <nicholas.kazlauskas@amd.com>
Tested-by: Daniel Wheeler <daniel.wheeler@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
Signed-off-by: Sasha Levin <sashal@kernel.org>

Linux 6.6

Merge tag 'x86-urgent-2023-10-28' of git://git./linux/kernel/git/tip/tip

Pull misc x86 fixes from Ingo Molnar:

 - Fix a possible CPU hotplug deadlock bug caused by the new TSC
   synchronization code

 - Fix a legacy PIC discovery bug that results in device troubles on
   affected systems, such as non-working keybards, etc

 - Add a new Intel CPU model number to <asm/intel-family.h>

* tag 'x86-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/tsc: Defer marking TSC unstable to a worker
  x86/i8259: Skip probing when ACPI/MADT advertises PCAT compatibility
  x86/cpu: Add model number for Intel Arrow Lake mobile processor

Merge tag 'irq-urgent-2023-10-28' of git://git./linux/kernel/git/tip/tip

Pull irq fix from Ingo Molnar:
 "Restore unintentionally lost quirk settings in the GIC irqchip driver,
  which broke certain devices"

* tag 'irq-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  irqchip/gic-v3-its: Don't override quirk settings with default values

Merge tag 'perf-urgent-2023-10-28' of git://git./linux/kernel/git/tip/tip

Pull perf event fix from Ingo Molnar:
 "Fix a potential NULL dereference bug"

* tag 'perf-urgent-2023-10-28' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf/core: Fix potential NULL deref

Merge tag 'probes-fixes-v6.6-rc7' of git://git./linux/kernel/git/trace/linux-trace

Pull probes fixes from Masami Hiramatsu:

 - tracing/kprobes: Fix kernel-doc warnings for the variable length
   arguments

 - tracing/kprobes: Fix to count the symbols in modules even if the
   module name is not specified so that user can probe the symbols in
   the modules without module name

* tag 'probes-fixes-v6.6-rc7' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing/kprobes: Fix symbol counting logic by looking at modules as well
  tracing/kprobes: Fix the description of variable length arguments

Merge tag 'dma-mapping-6.6-2023-10-28' of git://git.infradead.org/users/hch/dma-mapping

Pull dma-mapping fix from Christoph Hellwig:

 - reduce the initialy dynamic swiotlb size to remove an annoying but
   harmless warning from the page allocator (Petr Tesarik)

* tag 'dma-mapping-6.6-2023-10-28' of git://git.infradead.org/users/hch/dma-mapping:
  swiotlb: do not try to allocate a TLB bigger than MAX_ORDER pages

Merge tag 'char-misc-6.6-final' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are some very small driver fixes for 6.6-final that have shown up
  in the past two weeks. Included in here are:

   - tiny fastrpc bugfixes for reported errors

   - nvmem register fixes

   - iio driver fixes for some reported problems

   - fpga test fix

   - MAINTAINERS file update for fpga

  All of these have been in linux-next this week with no reported
  problems"

* tag 'char-misc-6.6-final' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc:
  fpga: Fix memory leak for fpga_region_test_class_find()
  fpga: m10bmc-sec: Change contact for secure update driver
  fpga: disable KUnit test suites when module support is enabled
  iio: afe: rescale: Accept only offset channels
  nvmem: imx: correct nregs for i.MX6ULL
  nvmem: imx: correct nregs for i.MX6UL
  nvmem: imx: correct nregs for i.MX6SLL
  misc: fastrpc: Unmap only if buffer is unmapped from DSP
  misc: fastrpc: Clean buffers on remote invocation failures
  misc: fastrpc: Free DMA handles for RPC calls with no arguments
  misc: fastrpc: Reset metadata buffer to avoid incorrect free
  iio: exynos-adc: request second interupt only when touchscreen mode is used
  iio: adc: xilinx-xadc: Correct temperature offset/scale for UltraScale
  iio: adc: xilinx-xadc: Don't clobber preset voltage/temperature thresholds
  dt-bindings: iio: add missing reset-gpios constrain

Merge tag 'i2c-for-6.6-rc8' of git://git./linux/kernel/git/wsa/linux

Pull i2c fixes from Wolfram Sang:
 "Bugfixes for Axxia when it is a target and for PEC handling of
  stm32f7.

  Plus, fix an OF node leak pattern in the mux subsystem"

* tag 'i2c-for-6.6-rc8' of git://git.kernel.org/pub/scm/linux/kernel/git/wsa/linux:
  i2c: stm32f7: Fix PEC handling in case of SMBUS transfers
  i2c: muxes: i2c-mux-gpmux: Use of_get_i2c_adapter_by_node()
  i2c: muxes: i2c-demux-pinctrl: Use of_get_i2c_adapter_by_node()
  i2c: muxes: i2c-mux-pinctrl: Use of_get_i2c_adapter_by_node()
  i2c: aspeed: Fix i2c bus hang in slave read

Merge tag 'clk-fixes-for-linus' of git://git./linux/kernel/git/clk/linux

Pull clk fixes from Stephen Boyd:
 "Three fixes, one for the clk framework and two for clk drivers:

   - Avoid an oops in possible_parent_show() by checking for no parent
     properly when a DT index based lookup is used

   - Handle errors returned from divider_ro_round_rate() in
     clk_stm32_composite_determine_rate()

   - Fix clk_ops::determine_rate() implementation of socfpga's
     gateclk_ops that was ruining uart output because the divider
     was forgotten about"

* tag 'clk-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/clk/linux:
  clk: stm32: Fix a signedness issue in clk_stm32_composite_determine_rate()
  clk: Sanitize possible_parent_show to Handle Return Value of of_clk_get_parent_name
  clk: socfpga: gate: Account for the divider in determine_rate

Merge tag 'pull-fixes' of git://git./linux/kernel/git/viro/vfs

Pull misc filesystem fixes from Al Viro:
 "Assorted fixes all over the place: literally nothing in common, could
  have been three separate pull requests.

  All are simple regression fixes, but not for anything from this cycle"

* tag 'pull-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  ceph_wait_on_conflict_unlink(): grab reference before dropping ->d_lock
  io_uring: kiocb_done() should *not* trust ->ki_pos if ->{read,write}_iter() failed
  sparc32: fix a braino in fault handling in csum_and_copy_..._user()

tracing/kprobes: Fix symbol counting logic by looking at modules as well

Recent changes to count number of matching symbols when creating
a kprobe event failed to take into account kernel modules. As such, it
breaks kprobes on kernel module symbols, by assuming there is no match.

Fix this my calling module_kallsyms_on_each_symbol() in addition to
kallsyms_on_each_match_symbol() to perform a proper counting.

Link: https://lore.kernel.org/all/20231027233126.2073148-1-andrii@kernel.org/
Cc: Francis Laniel <flaniel@linux.microsoft.com>
Cc: stable@vger.kernel.org
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Fixes: b022f0c7e404 ("tracing/kprobes: Return EADDRNOTAVAIL when func matches several symbols")
Signed-off-by: Andrii Nakryiko <andrii@kernel.org>
Acked-by: Song Liu <song@kernel.org>
Signed-off-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>

ceph_wait_on_conflict_unlink(): grab reference before dropping ->d_lock

Use of dget() after we'd dropped ->d_lock is too late - dentry might
be gone by that point.

Reviewed-by: Jeff Layton <jlayton@kernel.org>
Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>