If a core is offline then enabling SMT should not online CPUs of
this core. By enabling SMT, what is intended is either changing the SMT
value from "off" to "on" or setting the SMT level (threads per core) from a
lower to higher value.

On PowerPC the ppc64_cpu utility can be used, among other things, to
perform the following functions:

ppc64_cpu --cores-on                # Get the number of online cores
ppc64_cpu --cores-on=X              # Put exactly X cores online
ppc64_cpu --offline-cores=X[,Y,...] # Put specified cores offline
ppc64_cpu --smt={on|off|value}      # Enable, disable or change SMT level

If the user has decided to offline certain cores, enabling SMT should
not online CPUs in those cores. This patch fixes the issue and changes
the behaviour as described, by introducing an arch specific function
topology_is_core_online(). It is currently implemented only for PowerPC.

Fixes: 73c58e7e1412 ("powerpc: Add HOTPLUG_SMT support")
Reported-by: Tyrel Datwyler <tyreld@linux.ibm.com>
Closes: https://groups.google.com/g/powerpc-utils-devel/c/wrwVzAAnRlI/m/5KJSoqP4BAAJ
Signed-off-by: Nysal Jan K.A <nysal@linux.ibm.com>
Reviewed-by: Shrikanth Hegde <sshegde@linux.ibm.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Michael Ellerman <mpe@ellerman.id.au>
Link: https://msgid.link/20240731030126.956210-2-nysal@linux.ibm.com

Pull x86 confidential computing updates from Borislav Petkov:
 "Unrelated x86/cc changes queued here to avoid ugly cross-merges and
  conflicts:

   - Carve out CPU hotplug function declarations into a separate header
     with the goal to be able to use the lockdep assertions in a more
     flexible manner

   - As a result, refactor cacheinfo code after carving out a function
     to return the cache ID associated with a given cache level

   - Cleanups

  Add support to be able to kexec TDX guests:

   - Expand ACPI MADT CPU offlining support

   - Add machinery to prepare CoCo guests memory before kexec-ing into a
     new kernel

   - Cleanup, readjust and massage related code"

* tag 'x86_cc_for_v6.11_rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (24 commits)
  ACPI: tables: Print MULTIPROC_WAKEUP when MADT is parsed
  x86/acpi: Add support for CPU offlining for ACPI MADT wakeup method
  x86/mm: Introduce kernel_ident_mapping_free()
  x86/smp: Add smp_ops.stop_this_cpu() callback
  x86/acpi: Do not attempt to bring up secondary CPUs in the kexec case
  x86/acpi: Rename fields in the acpi_madt_multiproc_wakeup structure
  x86/mm: Do not zap page table entries mapping unaccepted memory table during kdump
  x86/mm: Make e820__end_ram_pfn() cover E820_TYPE_ACPI ranges
  x86/tdx: Convert shared memory back to private on kexec
  x86/mm: Add callbacks to prepare encrypted memory for kexec
  x86/tdx: Account shared memory
  x86/mm: Return correct level from lookup_address() if pte is none
  x86/mm: Make x86_platform.guest.enc_status_change_*() return an error
  x86/kexec: Keep CR4.MCE set during kexec for TDX guest
  x86/relocate_kernel: Use named labels for less confusion
  cpu/hotplug, x86/acpi: Disable CPU offlining for ACPI MADT wakeup
  cpu/hotplug: Add support for declaring CPU offlining not supported
  x86/apic: Mark acpi_mp_wake_* variables as __ro_after_init
  x86/acpi: Extract ACPI MADT wakeup code into a separate file
  x86/kexec: Remove spurious unconditional JMP from from identity_mapped()
  ...

Pull arm64 updates from Catalin Marinas:
 "The biggest part is the virtual CPU hotplug that touches ACPI,
  irqchip. We also have some GICv3 optimisation for pseudo-NMIs that has
  been queued via the arm64 tree. Otherwise the usual perf updates,
  kselftest, various small cleanups.

  Core:

   - Virtual CPU hotplug support for arm64 ACPI systems

   - cpufeature infrastructure cleanups and making the FEAT_ECBHB ID
     bits visible to guests

   - CPU errata: expand the speculative SSBS workaround to more CPUs

   - GICv3, use compile-time PMR values: optimise the way regular IRQs
     are masked/unmasked when GICv3 pseudo-NMIs are used, removing the
     need for a static key in fast paths by using a priority value
     chosen dynamically at boot time

  ACPI:

   - 'acpi=nospcr' option to disable SPCR as default console for arm64

   - Move some ACPI code (cpuidle, FFH) to drivers/acpi/arm64/

  Perf updates:

   - Rework of the IMX PMU driver to enable support for I.MX95

   - Enable support for tertiary match groups in the CMN PMU driver

   - Initial refactoring of the CPU PMU code to prepare for the fixed
     instruction counter introduced by Arm v9.4

   - Add missing PMU driver MODULE_DESCRIPTION() strings

   - Hook up DT compatibles for recent CPU PMUs

  Kselftest updates:

   - Kernel mode NEON fp-stress

   - Cleanups, spelling mistakes

  Miscellaneous:

   - arm64 Documentation update with a minor clarification on TBI

   - Fix missing IPI statistics

   - Implement raw_smp_processor_id() using thread_info rather than a
     per-CPU variable (better code generation)

   - Make MTE checking of in-kernel asynchronous tag faults conditional
     on KASAN being enabled

   - Minor cleanups, typos"

* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (69 commits)
  selftests: arm64: tags: remove the result script
  selftests: arm64: tags_test: conform test to TAP output
  perf: add missing MODULE_DESCRIPTION() macros
  arm64: smp: Fix missing IPI statistics
  irqchip/gic-v3: Fix 'broken_rdists' unused warning when !SMP and !ACPI
  ACPI: Add acpi=nospcr to disable ACPI SPCR as default console on ARM64
  Documentation: arm64: Update memory.rst for TBI
  arm64/cpufeature: Replace custom macros with fields from ID_AA64PFR0_EL1
  KVM: arm64: Replace custom macros with fields from ID_AA64PFR0_EL1
  perf: arm_pmuv3: Include asm/arm_pmuv3.h from linux/perf/arm_pmuv3.h
  perf: arm_v6/7_pmu: Drop non-DT probe support
  perf/arm: Move 32-bit PMU drivers to drivers/perf/
  perf: arm_pmuv3: Drop unnecessary IS_ENABLED(CONFIG_ARM64) check
  perf: arm_pmuv3: Avoid assigning fixed cycle counter with threshold
  arm64: Kconfig: Fix dependencies to enable ACPI_HOTPLUG_CPU
  perf: imx_perf: add support for i.MX95 platform
  perf: imx_perf: fix counter start and config sequence
  perf: imx_perf: refactor driver for imx93
  perf: imx_perf: let the driver manage the counter usage rather the user
  perf: imx_perf: add macro definitions for parsing config attr
  ...

Pull CPU hotplug updates from Thomas Gleixner:
 "A small set of SMP/CPU hotplug updates:

   - Reverse the order of iteration when freezing secondary CPUs for
     hibernation.

     This avoids that drivers like the Intel uncore performance counter
     have to transfer the assignement of handling the per package uncore
     events for every CPU in a package, which is a considerable speedup
     on larger systems.

   - Add a missing destroy_work_on_stack() invocation in
     smp_call_on_cpu() to prevent debug objects to emit a false positive
     warning when the stack is freed.

   - Small cleanups in comments and a str_plural() conversion"

* tag 'smp-core-2024-07-14' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  smp: Add missing destroy_work_on_stack() call in smp_call_on_cpu()
  cpu/hotplug: Reverse order of iteration in freeze_secondary_cpus()
  smp: Use str_plural() to fix Coccinelle warnings
  cpu/hotplug: Fix typo in comment

The 'offline' file in sysfs shows all offline CPUs, including those
that aren't present. User-space is expected to remove not-present CPUs
from this list to learn which CPUs could be brought online.

CPUs can be present but not-enabled. These CPUs can't be brought online
until the firmware policy changes, which comes with an ACPI notification
that will register the CPUs.

With only the offline and present files, user-space is unable to
determine which CPUs it can try to bring online. Add a new CPU mask
that shows this based on all the registered CPUs.

Signed-off-by: James Morse <james.morse@arm.com>
Tested-by: Miguel Luis <miguel.luis@oracle.com>
Tested-by: Vishnu Pajjuri <vishnu@os.amperecomputing.com>
Tested-by: Jianyong Wu <jianyong.wu@arm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Russell King (Oracle) <rmk+kernel@armlinux.org.uk>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20240529133446.28446-20-Jonathan.Cameron@huawei.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>

After the rework of "Parallel CPU bringup", the cmdline "nosmp" and
"maxcpus=0" parameters are not working anymore. These parameters set
setup_max_cpus to zero and that's handed to bringup_nonboot_cpus().

The code there does a decrement before checking for zero, which brings it
into the negative space and brings up all CPUs.

Add a zero check at the beginning of the function to prevent this.

[ tglx: Massaged change log ]

Fixes: 18415f33e2ac4ab382 ("cpu/hotplug: Allow "parallel" bringup up to CPUHP_BP_KICK_AP_STATE")
Fixes: 06c6796e0304234da6 ("cpu/hotplug: Fix off by one in cpuhp_bringup_mask()")
Signed-off-by: Huacai Chen <chenhuacai@loongson.cn>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20240618081336.3996825-1-chenhuacai@loongson.cn

ACPI MADT doesn't allow to offline a CPU after it has been woken up.

Currently, CPU hotplug is prevented based on the confidential computing
attribute which is set for Intel TDX. But TDX is not the only possible user of
the wake up method. Any platform that uses ACPI MADT wakeup method cannot
offline CPU.

Disable CPU offlining on ACPI MADT wakeup enumeration.

This has no visible effects for users: currently, TDX guest is the only platform
that uses the ACPI MADT wakeup method.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Rafael J. Wysocki <rafael.j.wysocki@intel.com>
Tested-by: Tao Liu <ltao@redhat.com>
Link: https://lore.kernel.org/r/20240614095904.1345461-5-kirill.shutemov@linux.intel.com

The ACPI MADT mailbox wakeup method doesn't allow to offline a CPU after
it has been woken up.

Currently, offlining is prevented based on the confidential computing attribute
which is set for Intel TDX. But TDX is not the only possible user of the wake up
method. The MADT wakeup can be implemented outside of a confidential computing
environment. Offline support is a property of the wakeup method, not the CoCo
implementation.

Introduce cpu_hotplug_disable_offlining() that can be called to indicate that
CPU offlining should be disabled.

This function is going to replace CC_ATTR_HOTPLUG_DISABLED for ACPI MADT wakeup
method.

Signed-off-by: Kirill A. Shutemov <kirill.shutemov@linux.intel.com>
Signed-off-by: Borislav Petkov (AMD) <bp@alien8.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Tao Liu <ltao@redhat.com>
Link: https://lore.kernel.org/r/20240614095904.1345461-4-kirill.shutemov@linux.intel.com

Whenever CPU hotplug state callbacks are registered, the startup callback
is invoked on CPUs that have already reached the provided state in order of
ascending CPU IDs.

In freeze_secondary_cpus() the teardown of CPUs happens in the same are
invoked in the same order. This is known to make a difference is the
current implementation of these callbacks in arch/x86/events/intel/uncore.c:

 - uncore_event_cpu_online() designates the first CPU it is invoked for
   on each package as the uncore event collector for that package

 - uncore_event_cpu_offline() if the CPU being offlined is the event
   collector for its package, transfers that responsibility over to
   the next (by ascending CPU id) one in the same package

With the current order of CPU teardowns in freeze_secondary_cpus(), the
latter ends up doing the ownership transfer work on every single CPU.  That
work involves a synchronize_rcu() call, ultimately unnecessarily degrading
the performance of CPU offlining.

To address this make freeze_secondary_cpus() iterate through the CPUs in
reverse order, so that the teardown happens in order of descending CPU IDs.

[ tglx: Massage change log ]

Signed-off-by: Stanislav Spassov <stanspas@amazon.de>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Link: https://lore.kernel.org/r/20240524160449.48594-1-stanspas@amazon.de

Commit 4205e4786d0b ("cpu/hotplug: Provide dynamic range for prepare
stage") added a dynamic range for the prepare states, but did not handle
the assignment of the dynstate variable in __cpuhp_setup_state_cpuslocked().

This causes the corresponding startup callback not to be invoked when
calling __cpuhp_setup_state_cpuslocked() with the CPUHP_BP_PREPARE_DYN
parameter, even though it should be.

Currently, the users of __cpuhp_setup_state_cpuslocked(), for one reason or
another, have not triggered this bug.

Fixes: 4205e4786d0b ("cpu/hotplug: Provide dynamic range for prepare stage")
Signed-off-by: Yuntao Wang <ytcoode@gmail.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20240515134554.427071-1-ytcoode@gmail.com