platform/kernel/linux-rpi.git
18 months agoarm64: lds: move .got section out of .text
Fangrui Song [Tue, 2 May 2023 07:41:05 +0000 (07:41 +0000)]
arm64: lds: move .got section out of .text

Currently, the .got section is placed within the output section .text.
However, when .got is non-empty, the SHF_WRITE flag is set for .text
when linked by lld. GNU ld recognizes .text as a special section and
ignores the SHF_WRITE flag. By renaming .text, we can also get the
SHF_WRITE flag.

The kernel has performed R_AARCH64_RELATIVE resolving very early, and can
then assume that .got is read-only. Let's move .got to the vmlinux_rodata
pseudo-segment.

As Ard Biesheuvel notes:

"This matters to consumers of the vmlinux ELF representation of the
kernel image, such as syzkaller, which disregards writable PT_LOAD
segments when resolving code symbols. The kernel itself does not care
about this distinction, but given that the GOT contains data and not
code, it does not require executable permissions, and therefore does
not belong in .text to begin with."

Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Fangrui Song <maskray@google.com>
Link: https://lore.kernel.org/r/20230502074105.1541926-1-maskray@google.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: kernel: remove SHF_WRITE|SHF_EXECINSTR from .idmap.text
ndesaulniers@google.com [Fri, 28 Apr 2023 18:28:17 +0000 (11:28 -0700)]
arm64: kernel: remove SHF_WRITE|SHF_EXECINSTR from .idmap.text

commit d54170812ef1 ("arm64: fix .idmap.text assertion for large kernels")
modified some of the section assembler directives that declare
.idmap.text to be SHF_ALLOC instead of
SHF_ALLOC|SHF_WRITE|SHF_EXECINSTR.

This patch fixes up the remaining stragglers that were left behind.  Add
Fixes tag so that this doesn't precede related change in stable.

Fixes: d54170812ef1 ("arm64: fix .idmap.text assertion for large kernels")
Reported-by: Greg Thelen <gthelen@google.com>
Reviewed-by: Ard Biesheuvel <ardb@kernel.org>
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
Link: https://lore.kernel.org/r/20230428-awx-v2-1-b197ffa16edc@google.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: cpufeature: Fix pointer auth hwcaps
Kristina Martsenko [Fri, 28 Apr 2023 13:25:46 +0000 (14:25 +0100)]
arm64: cpufeature: Fix pointer auth hwcaps

The pointer auth hwcaps are not getting reported to userspace, as they
are missing the .matches field. Add the field back.

Fixes: 876e3c8efe79 ("arm64/cpufeature: Pull out helper for CPUID register definitions")
Signed-off-by: Kristina Martsenko <kristina.martsenko@arm.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230428132546.2513834-1-kristina.martsenko@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: Fix label placement in record_mmu_state()
Neeraj Upadhyay [Tue, 25 Apr 2023 09:57:00 +0000 (15:27 +0530)]
arm64: Fix label placement in record_mmu_state()

Fix label so that pre_disable_mmu_workaround() is called
before clearing sctlr_el1.M.

Fixes: 2ced0f30a426 ("arm64: head: Switch endianness before populating the ID map")
Signed-off-by: Neeraj Upadhyay <quic_neeraju@quicinc.com>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/20230425095700.22005-1-quic_neeraju@quicinc.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoMerge branch 'for-next/sysreg' into for-next/core
Will Deacon [Thu, 20 Apr 2023 17:03:07 +0000 (18:03 +0100)]
Merge branch 'for-next/sysreg' into for-next/core

* for-next/sysreg:
  arm64/sysreg: Convert HFGITR_EL2 to automatic generation
  arm64/idreg: Don't disable SME when disabling SVE
  arm64/sysreg: Update ID_AA64PFR1_EL1 for DDI0601 2022-12
  arm64/sysreg: Convert HFG[RW]TR_EL2 to automatic generation
  arm64/sysreg: allow *Enum blocks in SysregFields blocks

18 months agoMerge branch 'for-next/stacktrace' into for-next/core
Will Deacon [Thu, 20 Apr 2023 17:03:02 +0000 (18:03 +0100)]
Merge branch 'for-next/stacktrace' into for-next/core

* for-next/stacktrace:
  arm64: move PAC masks to <asm/pointer_auth.h>
  arm64: use XPACLRI to strip PAC
  arm64: avoid redundant PAC stripping in __builtin_return_address()
  arm64: stacktrace: always inline core stacktrace functions
  arm64: stacktrace: move dump functions to end of file
  arm64: stacktrace: recover return address for first entry

18 months agoMerge branch 'for-next/perf' into for-next/core
Will Deacon [Thu, 20 Apr 2023 17:02:56 +0000 (18:02 +0100)]
Merge branch 'for-next/perf' into for-next/core

* for-next/perf: (24 commits)
  KVM: arm64: Ensure CPU PMU probes before pKVM host de-privilege
  drivers/perf: hisi: add NULL check for name
  drivers/perf: hisi: Remove redundant initialized of pmu->name
  perf/arm-cmn: Fix port detection for CMN-700
  arm64: pmuv3: dynamically map PERF_COUNT_HW_BRANCH_INSTRUCTIONS
  perf/arm-cmn: Validate cycles events fully
  Revert "ARM: mach-virt: Select PMUv3 driver by default"
  drivers/perf: apple_m1: Add Apple M2 support
  dt-bindings: arm-pmu: Add PMU compatible strings for Apple M2 cores
  perf: arm_cspmu: Fix variable dereference warning
  perf/amlogic: Fix config1/config2 parsing issue
  drivers/perf: Use devm_platform_get_and_ioremap_resource()
  kbuild, drivers/perf: remove MODULE_LICENSE in non-modules
  perf: qcom: Use devm_platform_get_and_ioremap_resource()
  perf: arm: Use devm_platform_get_and_ioremap_resource()
  perf/arm-cmn: Move overlapping wp_combine field
  ARM: mach-virt: Select PMUv3 driver by default
  ARM: perf: Allow the use of the PMUv3 driver on 32bit ARM
  ARM: Make CONFIG_CPU_V7 valid for 32bit ARMv8 implementations
  perf: pmuv3: Change GENMASK to GENMASK_ULL
  ...

18 months agoKVM: arm64: Ensure CPU PMU probes before pKVM host de-privilege
Will Deacon [Thu, 20 Apr 2023 12:33:56 +0000 (13:33 +0100)]
KVM: arm64: Ensure CPU PMU probes before pKVM host de-privilege

Although pKVM supports CPU PMU emulation for non-protected guests since
722625c6f4c5 ("KVM: arm64: Reenable pmu in Protected Mode"), this relies
on the PMU driver probing before the host has de-privileged so that the
'kvm_arm_pmu_available' static key can still be enabled by patching the
hypervisor text.

As it happens, both of these events hang off device_initcall() but the
PMU consistently won the race until 7755cec63ade ("arm64: perf: Move
PMUv3 driver to drivers/perf"). Since then, the host will fail to boot
when pKVM is enabled:

  | hw perfevents: enabled with armv8_pmuv3_0 PMU driver, 7 counters available
  | kvm [1]: nVHE hyp BUG at: [<ffff8000090366e0>] __kvm_nvhe_handle_host_mem_abort+0x270/0x284!
  | kvm [1]: Cannot dump pKVM nVHE stacktrace: !CONFIG_PROTECTED_NVHE_STACKTRACE
  | kvm [1]: Hyp Offset: 0xfffea41fbdf70000
  | Kernel panic - not syncing: HYP panic:
  | PS:a00003c9 PC:0000dbe04b0c66e0 ESR:00000000f2000800
  | FAR:fffffbfffddfcf00 HPFAR:00000000010b0bf0 PAR:0000000000000000
  | VCPU:0000000000000000
  | CPU: 2 PID: 1 Comm: swapper/0 Not tainted 6.3.0-rc7-00083-g0bce6746d154 #1
  | Hardware name: QEMU QEMU Virtual Machine, BIOS 0.0.0 02/06/2015
  | Call trace:
  |  dump_backtrace+0xec/0x108
  |  show_stack+0x18/0x2c
  |  dump_stack_lvl+0x50/0x68
  |  dump_stack+0x18/0x24
  |  panic+0x13c/0x33c
  |  nvhe_hyp_panic_handler+0x10c/0x190
  |  aarch64_insn_patch_text_nosync+0x64/0xc8
  |  arch_jump_label_transform+0x4c/0x5c
  |  __jump_label_update+0x84/0xfc
  |  jump_label_update+0x100/0x134
  |  static_key_enable_cpuslocked+0x68/0xac
  |  static_key_enable+0x20/0x34
  |  kvm_host_pmu_init+0x88/0xa4
  |  armpmu_register+0xf0/0xf4
  |  arm_pmu_acpi_probe+0x2ec/0x368
  |  armv8_pmu_driver_init+0x38/0x44
  |  do_one_initcall+0xcc/0x240

Fix the race properly by deferring the de-privilege step to
device_initcall_sync(). This will also be needed in future when probing
IOMMU devices and allows us to separate the pKVM de-privilege logic from
the core hypervisor initialisation path.

Cc: Oliver Upton <oliver.upton@linux.dev>
Cc: Fuad Tabba <tabba@google.com>
Cc: Marc Zyngier <maz@kernel.org>
Fixes: 7755cec63ade ("arm64: perf: Move PMUv3 driver to drivers/perf")
Tested-by: Fuad Tabba <tabba@google.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20230420123356.2708-1-will@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoMerge branch 'for-next/mm' into for-next/core
Will Deacon [Thu, 20 Apr 2023 10:22:33 +0000 (11:22 +0100)]
Merge branch 'for-next/mm' into for-next/core

* for-next/mm:
  arm64: mm: always map fixmap at page granularity
  arm64: mm: move fixmap code to its own file
  arm64: add FIXADDR_TOT_{START,SIZE}
  Revert "Revert "arm64: dma: Drop cache invalidation from arch_dma_prep_coherent()""
  arm: uaccess: Remove memcpy_page_flushcache()
  mm,kfence: decouple kfence from page granularity mapping judgement

18 months agoMerge branch 'for-next/misc' into for-next/core
Will Deacon [Thu, 20 Apr 2023 10:22:09 +0000 (11:22 +0100)]
Merge branch 'for-next/misc' into for-next/core

* for-next/misc:
  arm64: kexec: include reboot.h
  arm64: delete dead code in this_cpu_set_vectors()
  arm64: kernel: Fix kernel warning when nokaslr is passed to commandline
  arm64: kgdb: Set PSTATE.SS to 1 to re-enable single-step
  arm64/sme: Fix some comments of ARM SME
  arm64/signal: Alloc tpidr2 sigframe after checking system_supports_tpidr2()
  arm64/signal: Use system_supports_tpidr2() to check TPIDR2
  arm64: compat: Remove defines now in asm-generic
  arm64: kexec: remove unnecessary (void*) conversions
  arm64: armv8_deprecated: remove unnecessary (void*) conversions
  firmware: arm_sdei: Fix sleep from invalid context BUG

18 months agoMerge branch 'for-next/kdump' into for-next/core
Will Deacon [Thu, 20 Apr 2023 10:22:04 +0000 (11:22 +0100)]
Merge branch 'for-next/kdump' into for-next/core

* for-next/kdump:
  arm64: kdump: defer the crashkernel reservation for platforms with no DMA memory zones
  arm64: kdump: do not map crashkernel region specifically
  arm64: kdump : take off the protection on crashkernel memory region

18 months agoMerge branch 'for-next/ftrace' into for-next/core
Will Deacon [Thu, 20 Apr 2023 10:21:56 +0000 (11:21 +0100)]
Merge branch 'for-next/ftrace' into for-next/core

* for-next/ftrace:
  arm64: ftrace: Simplify get_ftrace_plt
  arm64: ftrace: Add direct call support
  ftrace: selftest: remove broken trace_direct_tramp
  ftrace: Make DIRECT_CALLS work WITH_ARGS and !WITH_REGS
  ftrace: Store direct called addresses in their ops
  ftrace: Rename _ftrace_direct_multi APIs to _ftrace_direct APIs
  ftrace: Remove the legacy _ftrace_direct API
  ftrace: Replace uses of _ftrace_direct APIs with _ftrace_direct_multi
  ftrace: Let unregister_ftrace_direct_multi() call ftrace_free_filter()

18 months agoMerge branch 'for-next/cpufeature' into for-next/core
Will Deacon [Thu, 20 Apr 2023 10:21:45 +0000 (11:21 +0100)]
Merge branch 'for-next/cpufeature' into for-next/core

* for-next/cpufeature:
  arm64/cpufeature: Use helper macro to specify ID register for capabilites
  arm64/cpufeature: Consistently use symbolic constants for min_field_value
  arm64/cpufeature: Pull out helper for CPUID register definitions

18 months agoMerge branch 'for-next/asm' into for-next/core
Will Deacon [Thu, 20 Apr 2023 10:21:39 +0000 (11:21 +0100)]
Merge branch 'for-next/asm' into for-next/core

* for-next/asm:
  arm64: uaccess: remove unnecessary earlyclobber
  arm64: uaccess: permit put_{user,kernel} to use zero register
  arm64: uaccess: permit __smp_store_release() to use zero register
  arm64: atomics: lse: improve cmpxchg implementation

18 months agoMerge branch 'for-next/acpi' into for-next/core
Will Deacon [Thu, 20 Apr 2023 10:21:31 +0000 (11:21 +0100)]
Merge branch 'for-next/acpi' into for-next/core

* for-next/acpi:
  ACPI: AGDI: Improve error reporting for problems during .remove()

18 months agoarm64: kexec: include reboot.h
Simon Horman [Tue, 18 Apr 2023 11:54:00 +0000 (13:54 +0200)]
arm64: kexec: include reboot.h

Include reboot.h in machine_kexec.c for declaration of
machine_crash_shutdown.

gcc-12 with W=1 reports:

 arch/arm64/kernel/machine_kexec.c:257:6: warning: no previous prototype for 'machine_crash_shutdown' [-Wmissing-prototypes]
   257 | void machine_crash_shutdown(struct pt_regs *regs)

No functional changes intended.
Compile tested only.

Signed-off-by: Simon Horman <horms@kernel.org>
Link: https://lore.kernel.org/r/20230418-arm64-kexec-include-reboot-v1-1-8453fd4fb3fb@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: delete dead code in this_cpu_set_vectors()
Dan Carpenter [Wed, 19 Apr 2023 07:58:43 +0000 (10:58 +0300)]
arm64: delete dead code in this_cpu_set_vectors()

The "slot" variable is an enum, and in this context it is an unsigned
int.  So the type means it can never be negative and also we never pass
invalid data to this function.  If something did pass invalid data then
this check would be insufficient protection.

Signed-off-by: Dan Carpenter <dan.carpenter@linaro.org>
Acked-by: Ard Biesheuvel <ardb@kernel.org>
Link: https://lore.kernel.org/r/73859c9e-dea0-4764-bf01-7ae694fa2e37@kili.mountain
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/cpufeature: Use helper macro to specify ID register for capabilites
Mark Brown [Wed, 12 Apr 2023 17:13:31 +0000 (18:13 +0100)]
arm64/cpufeature: Use helper macro to specify ID register for capabilites

When defining which value to look for in a system register field we
currently manually specify the register, field shift, width and sign and
the value to look for. This opens the potential for error with for example
the wrong field width or sign being specified, an enumeration value for
a different similarly named field or letting something be initialised to 0.

Since we now generate defines for all the ID registers we now have named
constants for all of these things generated from the system register
description, meaning that we can generate initialisation for all the fields
used in matching from a minimal specification of register, field and match
value. This is both shorter and eliminates or makes build failures several
potential errors.

No change in the generated binary.

Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230303-arm64-cpufeature-helpers-v2-3-4c8f28a6f203@kernel.org
[will: Drop explicit '.sign' assignment for BTI feature]
Signed-off-by: Will Deacon <will@kernel.org>
18 months agodrivers/perf: hisi: add NULL check for name
Junhao He [Mon, 3 Apr 2023 08:14:23 +0000 (16:14 +0800)]
drivers/perf: hisi: add NULL check for name

When allocations fails that can be NULL now.

If the name provided is NULL, then the initialization process of the PMU
type and dev will be skipped in function perf_pmu_register().
Consequently, the PMU will not be able to register into the kernel.
Moreover, in the case of unregister the PMU, the function device_del()
will need to handle NULL pointers, which potentially can cause issues.

So move this allocation above the cpuhp_state_add_instance() and directly
return if it does fail.

Signed-off-by: Junhao He <hejunhao3@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230403081423.62460-3-hejunhao3@huawei.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agodrivers/perf: hisi: Remove redundant initialized of pmu->name
Junhao He [Mon, 3 Apr 2023 08:14:22 +0000 (16:14 +0800)]
drivers/perf: hisi: Remove redundant initialized of pmu->name

"pmu->name" is initialized by perf_pmu_register() function, so remove
the redundant initialized in hisi_pmu_init().

Signed-off-by: Junhao He <hejunhao3@huawei.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/r/20230403081423.62460-2-hejunhao3@huawei.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/cpufeature: Consistently use symbolic constants for min_field_value
Mark Brown [Wed, 12 Apr 2023 17:13:30 +0000 (18:13 +0100)]
arm64/cpufeature: Consistently use symbolic constants for min_field_value

A number of the cpufeatures use raw numbers for the minimum field values
specified rather than symbolic constants. In preparation for the use of
helper macros replace all these with the appropriate constants.

No change in the generated binary.

Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230303-arm64-cpufeature-helpers-v2-2-4c8f28a6f203@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/cpufeature: Pull out helper for CPUID register definitions
Mark Brown [Wed, 12 Apr 2023 17:13:29 +0000 (18:13 +0100)]
arm64/cpufeature: Pull out helper for CPUID register definitions

We use the same structure to match hwcaps and CPU features so we can use
the same helper to generate the fields required. Pull the portion of the
current hwcaps helper that initialises the fields out into a separate
define placed earlier in the file so we can use it for cpufeatures.

No functional change.

Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230303-arm64-cpufeature-helpers-v2-1-4c8f28a6f203@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/sysreg: Convert HFGITR_EL2 to automatic generation
Mark Brown [Wed, 12 Apr 2023 16:26:43 +0000 (17:26 +0100)]
arm64/sysreg: Convert HFGITR_EL2 to automatic generation

Automatically generate the Hypervisor Fine-Grained Instruction Trap
Register as per DDI0601 2023-03, currently we only have a definition for
the register name not any of the contents.  No functional change.

Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230306-arm64-fgt-reg-gen-v5-1-516a89cb50f6@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoACPI: AGDI: Improve error reporting for problems during .remove()
Uwe Kleine-König [Fri, 14 Oct 2022 16:06:23 +0000 (18:06 +0200)]
ACPI: AGDI: Improve error reporting for problems during .remove()

Returning an error value in a platform driver's remove callback results in
a generic error message being emitted by the driver core, but otherwise it
doesn't make a difference. The device goes away anyhow.

So instead of triggering the generic platform error message, emit a more
helpful message if a problem occurs and return 0 to suppress the generic
message.

This patch is a preparation for making platform remove callbacks return
void.

Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Lorenzo Pieralisi <lpieralisi@kernel.org>
Link: https://lore.kernel.org/r/20221014160623.467195-1-u.kleine-koenig@pengutronix.de
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: kernel: Fix kernel warning when nokaslr is passed to commandline
Pavankumar Kondeti [Wed, 12 Apr 2023 04:32:58 +0000 (10:02 +0530)]
arm64: kernel: Fix kernel warning when nokaslr is passed to commandline

'Unknown kernel command line parameters "nokaslr", will be passed to
user space' message is noticed in the dmesg when nokaslr is passed to
the kernel commandline on ARM64 platform. This is because nokaslr param
is handled by early cpufeature detection infrastructure and the parameter
is never consumed by a kernel param handler. Fix this warning by
providing a dummy kernel param handler for nokaslr.

Signed-off-by: Pavankumar Kondeti <quic_pkondeti@quicinc.com>
Link: https://lore.kernel.org/r/20230412043258.397455-1-quic_pkondeti@quicinc.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoperf/arm-cmn: Fix port detection for CMN-700
Robin Murphy [Wed, 12 Apr 2023 10:29:40 +0000 (11:29 +0100)]
perf/arm-cmn: Fix port detection for CMN-700

When the "extra device ports" configuration was first added, the
additional mxp_device_port_connect_info registers were added around the
existing mxp_mesh_port_connect_info registers. What I missed about
CMN-700 is that it shuffled them around to remove this discontinuity.
As such, tweak the definitions and factor out a helper for reading these
registers so we can deal with this discrepancy easily, which does at
least allow nicely tidying up the callsites. With this we can then also
do the nice thing and skip accesses completely rather than relying on
RES0 behaviour where we know the extra registers aren't defined.

Fixes: 23760a014417 ("perf/arm-cmn: Add CMN-700 support")
Reported-by: Jing Zhang <renyu.zj@linux.alibaba.com>
Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Link: https://lore.kernel.org/r/71d129241d4d7923cde72a0e5b4c8d2f6084525f.1681295193.git.robin.murphy@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: kgdb: Set PSTATE.SS to 1 to re-enable single-step
Sumit Garg [Thu, 2 Feb 2023 07:31:48 +0000 (13:01 +0530)]
arm64: kgdb: Set PSTATE.SS to 1 to re-enable single-step

Currently only the first attempt to single-step has any effect. After
that all further stepping remains "stuck" at the same program counter
value.

Refer to the ARM Architecture Reference Manual (ARM DDI 0487E.a) D2.12,
PSTATE.SS=1 should be set at each step before transferring the PE to the
'Active-not-pending' state. The problem here is PSTATE.SS=1 is not set
since the second single-step.

After the first single-step, the PE transferes to the 'Inactive' state,
with PSTATE.SS=0 and MDSCR.SS=1, thus PSTATE.SS won't be set to 1 due to
kernel_active_single_step()=true. Then the PE transferes to the
'Active-pending' state when ERET and returns to the debugger by step
exception.

Before this patch:
==================
Entering kdb (current=0xffff3376039f0000, pid 1) on processor 0 due to Keyboard Entry
[0]kdb>

[0]kdb>
[0]kdb> bp write_sysrq_trigger
Instruction(i) BP #0 at 0xffffa45c13d09290 (write_sysrq_trigger)
    is enabled   addr at ffffa45c13d09290, hardtype=0 installed=0

[0]kdb> go
$ echo h > /proc/sysrq-trigger

Entering kdb (current=0xffff4f7e453f8000, pid 175) on processor 1 due to Breakpoint @ 0xffffad651a309290
[1]kdb> ss

Entering kdb (current=0xffff4f7e453f8000, pid 175) on processor 1 due to SS trap @ 0xffffad651a309294
[1]kdb> ss

Entering kdb (current=0xffff4f7e453f8000, pid 175) on processor 1 due to SS trap @ 0xffffad651a309294
[1]kdb>

After this patch:
=================
Entering kdb (current=0xffff6851c39f0000, pid 1) on processor 0 due to Keyboard Entry
[0]kdb> bp write_sysrq_trigger
Instruction(i) BP #0 at 0xffffc02d2dd09290 (write_sysrq_trigger)
    is enabled   addr at ffffc02d2dd09290, hardtype=0 installed=0

[0]kdb> go
$ echo h > /proc/sysrq-trigger

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to Breakpoint @ 0xffffc02d2dd09290
[1]kdb> ss

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to SS trap @ 0xffffc02d2dd09294
[1]kdb> ss

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to SS trap @ 0xffffc02d2dd09298
[1]kdb> ss

Entering kdb (current=0xffff6851c53c1840, pid 174) on processor 1 due to SS trap @ 0xffffc02d2dd0929c
[1]kdb>

Fixes: 44679a4f142b ("arm64: KGDB: Add step debugging support")
Co-developed-by: Wei Li <liwei391@huawei.com>
Signed-off-by: Wei Li <liwei391@huawei.com>
Signed-off-by: Sumit Garg <sumit.garg@linaro.org>
Tested-by: Douglas Anderson <dianders@chromium.org>
Acked-by: Daniel Thompson <daniel.thompson@linaro.org>
Tested-by: Daniel Thompson <daniel.thompson@linaro.org>
Link: https://lore.kernel.org/r/20230202073148.657746-3-sumit.garg@linaro.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: move PAC masks to <asm/pointer_auth.h>
Mark Rutland [Wed, 12 Apr 2023 16:01:34 +0000 (17:01 +0100)]
arm64: move PAC masks to <asm/pointer_auth.h>

Now that we use XPACLRI to strip PACs within the kernel, the
ptrauth_user_pac_mask() and ptrauth_kernel_pac_mask() definitions no
longer need to live in <asm/compiler.h>.

Move them to <asm/pointer_auth.h>, and ensure that this header is
included where they are used.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Amit Daniel Kachhap <amit.kachhap@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Kristina Martsenko <kristina.martsenko@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230412160134.306148-4-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: use XPACLRI to strip PAC
Mark Rutland [Wed, 12 Apr 2023 16:01:33 +0000 (17:01 +0100)]
arm64: use XPACLRI to strip PAC

Currently we strip the PAC from pointers using C code, which requires
generating bitmasks, and conditionally clearing/setting bits depending
on bit 55. We can do better by using XPACLRI directly.

When the logic was originally written to strip PACs from user pointers,
contemporary toolchains used for the kernel had assemblers which were
unaware of the PAC instructions. As stripping the PAC from userspace
pointers required unconditional clearing of a fixed set of bits (which
could be performed with a single instruction), it was simpler to
implement the masking in C than it was to make use of XPACI or XPACLRI.

When support for in-kernel pointer authentication was added, the
stripping logic was extended to cover TTBR1 pointers, requiring several
instructions to handle whether to clear/set bits dependent on bit 55 of
the pointer.

This patch simplifies the stripping of PACs by using XPACLRI directly,
as contemporary toolchains do within __builtin_return_address(). This
saves a number of instructions, especially where
__builtin_return_address() does not implicitly strip the PAC but is
heavily used (e.g. with tracepoints). As the kernel might be compiled
with an assembler without knowledge of XPACLRI, it is assembled using
the 'HINT #7' alias, which results in an identical opcode.

At the same time, I've split ptrauth_strip_insn_pac() into
ptrauth_strip_user_insn_pac() and ptrauth_strip_kernel_insn_pac()
helpers so that we can avoid unnecessary PAC stripping when pointer
authentication is not in use in userspace or kernel respectively.

The underlying xpaclri() macro uses inline assembly which clobbers x30.
The clobber causes the compiler to save/restore the original x30 value
in a frame record (protected with PACIASP and AUTIASP when in-kernel
authentication is enabled), so this does not provide a gadget to alter
the return address. Similarly this does not adversely affect unwinding
due to the presence of the frame record.

The ptrauth_user_pac_mask() and ptrauth_kernel_pac_mask() are exported
from the kernel in ptrace and core dumps, so these are retained. A
subsequent patch will move them out of <asm/compiler.h>.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Amit Daniel Kachhap <amit.kachhap@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Kristina Martsenko <kristina.martsenko@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230412160134.306148-3-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: avoid redundant PAC stripping in __builtin_return_address()
Mark Rutland [Wed, 12 Apr 2023 16:01:32 +0000 (17:01 +0100)]
arm64: avoid redundant PAC stripping in __builtin_return_address()

In old versions of GCC and Clang, __builtin_return_address() did not
strip the PAC. This was not the behaviour we desired, and so we wrapped
this with code to strip the PAC in commit:

  689eae42afd7a916 ("arm64: mask PAC bits of __builtin_return_address")

Since then, both GCC and Clang decided that __builtin_return_address()
*should* strip the PAC, and the existing behaviour was a bug.

GCC was fixed in 11.1.0, with those fixes backported to 10.2.0, 9.4.0,
8.5.0, but not earlier:

  https://gcc.gnu.org/bugzilla/show_bug.cgi?id=94891

Clang was fixed in 12.0.0, though this was not backported:

  https://reviews.llvm.org/D75044

When using a compiler whose __builtin_return_address() strips the PAC,
our wrapper to strip the PAC is redundant. Similarly, when pointer
authentication is not in use within the kernel pointers will not have a
PAC, and so there's no point stripping those pointers.

To avoid this redundant work, this patch updates the
__builtin_return_address() wrapper to only be used when in-kernel
pointer authentication is configured and the compiler's
__builtin_return_address() does not strip the PAC.

This is a cleanup/optimization, and not a fix that requires backporting.
Stripping a PAC should be an idempotent operation, and so redundantly
stripping the PAC is not harmful.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Amit Daniel Kachhap <amit.kachhap@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: James Morse <james.morse@arm.com>
Cc: Kristina Martsenko <kristina.martsenko@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230412160134.306148-2-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/sme: Fix some comments of ARM SME
Dongxu Sun [Fri, 17 Mar 2023 12:49:15 +0000 (20:49 +0800)]
arm64/sme: Fix some comments of ARM SME

When TIF_SME is clear, fpsimd_restore_current_state will disable
SME trap during ret_to_user, then SME access trap is impossible
in userspace, not SVE.

Besides, fix typo: alocated->allocated.

Signed-off-by: Dongxu Sun <sundongxu3@huawei.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230317124915.1263-5-sundongxu3@huawei.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/signal: Alloc tpidr2 sigframe after checking system_supports_tpidr2()
Dongxu Sun [Fri, 17 Mar 2023 12:49:13 +0000 (20:49 +0800)]
arm64/signal: Alloc tpidr2 sigframe after checking system_supports_tpidr2()

Move tpidr2 sigframe allocation from under the checking of
system_supports_sme() to the checking of system_supports_tpidr2().

Signed-off-by: Dongxu Sun <sundongxu3@huawei.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230317124915.1263-3-sundongxu3@huawei.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/signal: Use system_supports_tpidr2() to check TPIDR2
Dongxu Sun [Fri, 17 Mar 2023 12:49:12 +0000 (20:49 +0800)]
arm64/signal: Use system_supports_tpidr2() to check TPIDR2

Since commit a9d6915859501("arm64/sme: Implement support
for TPIDR2"), We introduced system_supports_tpidr2() for
TPIDR2 handling. Let's use the specific check instead.

No functional changes.

Signed-off-by: Dongxu Sun <sundongxu3@huawei.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230317124915.1263-2-sundongxu3@huawei.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64/idreg: Don't disable SME when disabling SVE
Mark Brown [Thu, 23 Mar 2023 22:06:32 +0000 (22:06 +0000)]
arm64/idreg: Don't disable SME when disabling SVE

SVE and SME are separate features which can be implemented without each
other but currently if the user specifies arm64.nosve then we disable SME
as well as SVE. There is already a separate override for SME so remove the
implicit disablement from the SVE override.

One usecase for this would be testing SME only support on a system which
implements both SVE and SME.

Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230315-arm64-override-sve-sme-v2-1-bab7593e842b@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: kdump: defer the crashkernel reservation for platforms with no DMA memory...
Baoquan He [Fri, 7 Apr 2023 01:15:07 +0000 (09:15 +0800)]
arm64: kdump: defer the crashkernel reservation for platforms with no DMA memory zones

In commit 031495635b46 ("arm64: Do not defer reserve_crashkernel() for
platforms with no DMA memory zones"), reserve_crashkernel() is called
much earlier in arm64_memblock_init() to avoid causing base apge
mapping on platforms with no DMA meomry zones.

With taking off protection on crashkernel memory region, no need to call
reserve_crashkernel() specially in advance. The deferred invocation of
reserve_crashkernel() in bootmem_init() can cover all cases. So revert
the whole commit now.

Signed-off-by: Baoquan He <bhe@redhat.com>
Reviewed-by: Zhen Lei <thunder.leizhen@huawei.com>
Link: https://lore.kernel.org/r/20230407011507.17572-4-bhe@redhat.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: kdump: do not map crashkernel region specifically
Baoquan He [Fri, 7 Apr 2023 01:15:06 +0000 (09:15 +0800)]
arm64: kdump: do not map crashkernel region specifically

After taking off the protection functions on crashkernel memory region,
there's no need to map crashkernel region with page granularity during
linear mapping.

With this change, the system can make use of block or section mapping
on linear region to largely improve perforcemence during system bootup
and running.

Signed-off-by: Baoquan He <bhe@redhat.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Zhen Lei <thunder.leizhen@huawei.com>
Link: https://lore.kernel.org/r/20230407011507.17572-3-bhe@redhat.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: kdump : take off the protection on crashkernel memory region
Baoquan He [Fri, 7 Apr 2023 01:15:05 +0000 (09:15 +0800)]
arm64: kdump : take off the protection on crashkernel memory region

Problem:
=======
On arm64, block and section mapping is supported to build page tables.
However, currently it enforces to take base page mapping for the whole
linear mapping if CONFIG_ZONE_DMA or CONFIG_ZONE_DMA32 is enabled and
crashkernel kernel parameter is set. This will cause longer time of the
linear mapping process during bootup and severe performance degradation
during running time.

Root cause:
==========
On arm64, crashkernel reservation relies on knowing the upper limit of
low memory zone because it needs to reserve memory in the zone so that
devices' DMA addressing in kdump kernel can be satisfied. However, the
upper limit of low memory on arm64 is variant. And the upper limit can
only be decided late till bootmem_init() is called [1].

And we need to map the crashkernel region with base page granularity when
doing linear mapping, because kdump needs to protect the crashkernel region
via set_memory_valid(,0) after kdump kernel loading. However, arm64 doesn't
support well on splitting the built block or section mapping due to some
cpu reststriction [2]. And unfortunately, the linear mapping is done before
bootmem_init().

To resolve the above conflict on arm64, the compromise is enforcing to
take base page mapping for the entire linear mapping if crashkernel is
set, and CONFIG_ZONE_DMA or CONFIG_ZONE_DMA32 is enabed. Hence
performance is sacrificed.

Solution:
=========
Comparing with the base page mapping for the whole linear region, it's
better to take off the protection on crashkernel memory region for the
time being because the anticipated stamping on crashkernel memory region
could only happen in a chance in one million, while the base page mapping
for the whole linear region is mitigating arm64 systems with crashkernel
set always.

[1]
https://lore.kernel.org/all/YrIIJkhKWSuAqkCx@arm.com/T/#u

[2]
https://lore.kernel.org/linux-arm-kernel/20190911182546.17094-1-nsaenzjulienne@suse.de/T/

Signed-off-by: Baoquan He <bhe@redhat.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Acked-by: Mike Rapoport (IBM) <rppt@kernel.org>
Reviewed-by: Zhen Lei <thunder.leizhen@huawei.com>
Link: https://lore.kernel.org/r/20230407011507.17572-2-bhe@redhat.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: compat: Remove defines now in asm-generic
Teo Couprie Diaz [Tue, 14 Mar 2023 14:00:38 +0000 (14:00 +0000)]
arm64: compat: Remove defines now in asm-generic

Some generic COMPAT definitions have been consolidated in
asm-generic/compat.h by commit 84a0c977ab98
("asm-generic: compat: Cleanup duplicate definitions")

Remove those that are already defined to the same value there from
arm64 asm/compat.h.

Signed-off-by: Teo Couprie Diaz <teo.coupriediaz@arm.com>
Reviewed-by: Guo Ren <guoren@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20230314140038.252908-1-teo.coupriediaz@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: mm: always map fixmap at page granularity
Mark Rutland [Thu, 6 Apr 2023 15:27:59 +0000 (16:27 +0100)]
arm64: mm: always map fixmap at page granularity

Today the fixmap code largely maps elements at PAGE_SIZE granularity,
but we special-case the FDT mapping such that it can be mapped with 2M
block mappings when 4K pages are in use. The original rationale for this
was simplicity, but it has some unfortunate side-effects, and
complicates portions of the fixmap code (i.e. is not so simple after
all).

The FDT can be up to 2M in size but is only required to have 8-byte
alignment, and so it may straddle a 2M boundary. Thus when using 2M
block mappings we may map up to 4M of memory surrounding the FDT. This
is unfortunate as most of that memory will be unrelated to the FDT, and
any pages which happen to share a 2M block with the FDT will by mapped
with Normal Write-Back Cacheable attributes, which might not be what we
want elsewhere (e.g. for carve-outs using Non-Cacheable attributes).

The logic to handle mapping the FDT with 2M blocks requires some special
cases in the fixmap code, and ties it to the early page table
configuration by virtue of the SWAPPER_TABLE_SHIFT and
SWAPPER_BLOCK_SIZE constants used to determine the granularity used to
map the FDT.

This patch simplifies the FDT logic and removes the unnecessary mappings
of surrounding pages by always mapping the FDT at page granularity as
with all other fixmap mappings. To do so we statically reserve multiple
PTE tables to cover the fixmap VA range. Since the FDT can be at most
2M, for 4K pages we only need to allocate a single additional PTE table,
and for 16K and 64K pages the existing single PTE table is sufficient.

The PTE table allocation scales with the number of slots reserved in the
fixmap, and so this also makes it easier to add more fixmap entries if
we require those in future.

Our VA layout means that the fixmap will always fall within a single PMD
table (and consequently, within a single PUD/P4D/PGD entry), which we
can verify at compile time with a static_assert(). With that assert a
number of runtime warnings become impossible, and are removed.

I've boot-tested this patch with both 4K and 64K pages.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Link: https://lore.kernel.org/r/20230406152759.4164229-4-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: mm: move fixmap code to its own file
Mark Rutland [Thu, 6 Apr 2023 15:27:58 +0000 (16:27 +0100)]
arm64: mm: move fixmap code to its own file

Over time, arm64's mm/mmu.c has become increasingly large and painful to
navigate. Move the fixmap code to its own file where it can be understood in
isolation.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Link: https://lore.kernel.org/r/20230406152759.4164229-3-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: add FIXADDR_TOT_{START,SIZE}
Mark Rutland [Thu, 6 Apr 2023 15:27:57 +0000 (16:27 +0100)]
arm64: add FIXADDR_TOT_{START,SIZE}

Currently arm64's FIXADDR_{START,SIZE} definitions only cover the
runtime fixmap slots (and not the boot-time fixmap slots), but the code
for creating the fixmap assumes that these definitions cover the entire
fixmap range. This means that the ptdump boundaries are reported in a
misleading way, missing the VA region of the runtime slots. In theory
this could also cause the fixmap creation to go wrong if the boot-time
fixmap slots end up spilling into a separate PMD entry, though luckily
this is not currently the case in any configuration.

While it seems like we could extend FIXADDR_{START,SIZE} to cover the
entire fixmap area, core code relies upon these *only* covering the
runtime slots. For example, fix_to_virt() and virt_to_fix() try to
reject manipulation of the boot-time slots based upon
FIXADDR_{START,SIZE}, while __fix_to_virt() and __virt_to_fix() can
handle any fixmap slot.

This patch follows the lead of x86 in commit:

  55f49fcb879fbeeb ("x86/mm: Fix overlap of i386 CPU_ENTRY_AREA with FIX_BTMAP")

... and add new FIXADDR_TOT_{START,SIZE} definitions which cover the
entire fixmap area, using these for the fixmap creation and ptdump code.

As the boot-time fixmap slots are now rejected by fix_to_virt(),
the early_fixmap_init() code is changed to consistently use
__fix_to_virt(), as it already does in a few cases.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Ard Biesheuvel <ardb@kernel.org>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Ryan Roberts <ryan.roberts@arm.com>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Ryan Roberts <ryan.roberts@arm.com>
Link: https://lore.kernel.org/r/20230406152759.4164229-2-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: stacktrace: always inline core stacktrace functions
Mark Rutland [Tue, 11 Apr 2023 16:29:43 +0000 (17:29 +0100)]
arm64: stacktrace: always inline core stacktrace functions

The arm64 stacktrace code can be used in kprobe context, and so cannot
be safely probed. Some (but not all) of the unwind functions are
annotated with `NOKPROBE_SYMBOL()` to ensure this, with others markes as
`__always_inline`, relying on the top-level unwind function being marked
as `noinstr`.

This patch has stacktrace.c consistently mark the internal stacktrace
functions as `__always_inline`, removing the need for NOKPROBE_SYMBOL()
as the top-level unwind function (arch_stack_walk()) is marked as
`noinstr`. This is more consistent and is a simpler pattern to follow
for future additions to stacktrace.c.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Kalesh Singh <kaleshsingh@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Madhavan T. Venkataraman <madvenka@linux.microsoft.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230411162943.203199-4-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: stacktrace: move dump functions to end of file
Mark Rutland [Tue, 11 Apr 2023 16:29:42 +0000 (17:29 +0100)]
arm64: stacktrace: move dump functions to end of file

For historical reasons, the backtrace dumping functions are placed in
the middle of stacktrace.c, despite using functions defined later. For
clarity, and to make subsequent refactoring easier, move the dumping
functions to the end of stacktrace.c

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Kalesh Singh <kaleshsingh@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Madhavan T. Venkataraman <madvenka@linux.microsoft.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230411162943.203199-3-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: stacktrace: recover return address for first entry
Mark Rutland [Tue, 11 Apr 2023 16:29:41 +0000 (17:29 +0100)]
arm64: stacktrace: recover return address for first entry

The function which calls the top-level backtracing function may have
been instrumented with ftrace and/or kprobes, and hence the first return
address may have been rewritten.

Factor out the existing fgraph / kretprobes address recovery, and use
this for the first address. As the comment for the fgraph case isn't all
that helpful, I've also dropped that.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Reviewed-by: Kalesh Singh <kaleshsingh@google.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Madhavan T. Venkataraman <madvenka@linux.microsoft.com>
Cc: Mark Brown <broonie@kernel.org>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230411162943.203199-2-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: ftrace: Simplify get_ftrace_plt
Florent Revest [Wed, 5 Apr 2023 18:02:47 +0000 (20:02 +0200)]
arm64: ftrace: Simplify get_ftrace_plt

Following recent refactorings, the get_ftrace_plt function only ever
gets called with addr = FTRACE_ADDR so its code can be simplified to
always return the ftrace trampoline plt.

Signed-off-by: Florent Revest <revest@chromium.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Link: https://lore.kernel.org/r/20230405180250.2046566-3-revest@chromium.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoarm64: ftrace: Add direct call support
Florent Revest [Wed, 5 Apr 2023 18:02:46 +0000 (20:02 +0200)]
arm64: ftrace: Add direct call support

This builds up on the CALL_OPS work which extends the ftrace patchsite
on arm64 with an ops pointer usable by the ftrace trampoline.

This ops pointer is valid at all time. Indeed, it is either pointing to
ftrace_list_ops or to the single ops which should be called from that
patchsite.

There are a few cases to distinguish:
- If a direct call ops is the only one tracing a function:
  - If the direct called trampoline is within the reach of a BL
    instruction
     -> the ftrace patchsite jumps to the trampoline
  - Else
     -> the ftrace patchsite jumps to the ftrace_caller trampoline which
        reads the ops pointer in the patchsite and jumps to the direct
        call address stored in the ops
- Else
  -> the ftrace patchsite jumps to the ftrace_caller trampoline and its
     ops literal points to ftrace_list_ops so it iterates over all
     registered ftrace ops, including the direct call ops and calls its
     call_direct_funcs handler which stores the direct called
     trampoline's address in the ftrace_regs and the ftrace_caller
     trampoline will return to that address instead of returning to the
     traced function

Signed-off-by: Florent Revest <revest@chromium.org>
Co-developed-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Link: https://lore.kernel.org/r/20230405180250.2046566-2-revest@chromium.org
Signed-off-by: Will Deacon <will@kernel.org>
18 months agoMerge tag 'trace-direct-v6.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git...
Will Deacon [Tue, 11 Apr 2023 17:04:05 +0000 (18:04 +0100)]
Merge tag 'trace-direct-v6.3-rc3' of git://git./linux/kernel/git/trace/linux-trace into for-next/ftrace

Pull in ftrace trampoline updates from Steve so that we can implement
support for direct calls for arm64 on top:

tracing: Direct trampoline updates

Updates to the direct trampoline to allow ARM64 to have direct
trampolines.

18 months agoarm64: pmuv3: dynamically map PERF_COUNT_HW_BRANCH_INSTRUCTIONS
Stephane Eranian [Tue, 11 Apr 2023 09:38:09 +0000 (11:38 +0200)]
arm64: pmuv3: dynamically map PERF_COUNT_HW_BRANCH_INSTRUCTIONS

The mapping of perf_events generic hardware events to actual PMU events on
ARM PMUv3 may not always be correct. This is in particular true for the
PERF_COUNT_HW_BRANCH_INSTRUCTIONS event. Although the mapping points to an
architected event, it may not always be available. This can be seen with a
simple:

$ perf stat -e branches sleep 0
 Performance counter stats for 'sleep 0':

   <not supported>      branches

       0.001401081 seconds time elapsed

Yet the hardware does have an event that could be used for branches.

Dynamically check for a supported hardware event which can be used for
PERF_COUNT_HW_BRANCH_INSTRUCTIONS at mapping time.

And with that:

$ perf stat -e branches sleep 0

 Performance counter stats for 'sleep 0':

           166,739      branches

       0.000832163 seconds time elapsed

Co-developed-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Stephane Eranian <eranian@google.com>
Co-developed-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Co-developed-by: Peter Newman <peternewman@google.com>
Signed-off-by: Peter Newman <peternewman@google.com>
Link: https://lore.kernel.org/all/YvunKCJHSXKz%2FkZB@FVFF77S0Q05N
Link: https://lore.kernel.org/r/20230411093809.657501-1-peternewman@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf/arm-cmn: Validate cycles events fully
Robin Murphy [Mon, 3 Apr 2023 11:49:05 +0000 (12:49 +0100)]
perf/arm-cmn: Validate cycles events fully

DTC cycle count events don't have anything to validate or initialise in
themselves, but we should not forget to still validate their whole group
context. Otherwise, we may fail to correctly reject a contrived group
containing an impossible number of cycles events.

Signed-off-by: Robin Murphy <robin.murphy@arm.com>
Link: https://lore.kernel.org/r/3124e8c276a1f513c1a415dc839ca4181b3c8bc8.1680522545.git.robin.murphy@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64/sysreg: Update ID_AA64PFR1_EL1 for DDI0601 2022-12
Mark Brown [Wed, 8 Mar 2023 15:20:42 +0000 (15:20 +0000)]
arm64/sysreg: Update ID_AA64PFR1_EL1 for DDI0601 2022-12

Version 2022-12 of DDI0601 has defined a number of new fields in
previously RES0 space in ID_AA64PFR1_EL1, update our definition to
include them.

No functional changes.

Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230308-arm64-aa64pfr1-2022-12-v1-1-f03c1ea39611@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64/sysreg: Convert HFG[RW]TR_EL2 to automatic generation
Mark Brown [Thu, 23 Mar 2023 20:44:53 +0000 (20:44 +0000)]
arm64/sysreg: Convert HFG[RW]TR_EL2 to automatic generation

Convert the fine grained traps read and write control registers to
automatic generation as per DDI0601 2022-12. No functional changes.

Reviewed-by: Joey Gouly <joey.gouly@arm.com>
Acked-by: Catalin Marinas <catalin.marinas@arm.com>
Signed-off-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230306-arm64-fgt-reg-gen-v3-1-decba93cbaab@kernel.org
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64/sysreg: allow *Enum blocks in SysregFields blocks
Mark Rutland [Mon, 6 Mar 2023 11:48:36 +0000 (11:48 +0000)]
arm64/sysreg: allow *Enum blocks in SysregFields blocks

We'd like to support Enum/SignedEnum/UnsignedEnum blocks within
SysregFields blocks, so that we can define enumerations for sets of
registers. This isn't currently supported by gen-sysreg.awk due to the
way we track the active block, which can't handle more than a single
layer of nesting, which imposes an awkward requirement that when ending
a block we know what the parent block is when calling change_block()

Make this nicer by using a stack of active blocks, with block_push() to
start a block, and block_pop() to end a block. Doing so means that we
only need to check the active block at the start of parsing a line: for
the start of a block we can check the parent is valid, and for the end
of a block we check that the active block is valid.

This structure makes the block parsing simpler and makes it easy to
permit a block to live under several potential parents (e.g. by
permitting Enum to start when the active block is Sysreg or
SysregFields). It also permits further nesting, if we need that in
future.

To aid debugging, the stack of active blocks is reported for fatal
errors, and an error is raised if the file is terminated without ending
the active block. For clarity I've renamed the top-level element from
"None" to "Root".

The Fields element it intended only for use within Sysreg blocks, and
does not make sense within SysregFields blocks, and so remains forbidden
within a SysregFields block.

I've verified using sha1sum that this patch does not change the
current generated contents of <asm/sysreg-defs.h>.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Anshuman Khandual <anshuman.khandual@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Mark Brown <broonie@kernel.org>
Cc: Will Deacon <will@kernel.org>
Reviewed-by: Anshuman Khandual <anshuman.khandual@arm.com>
Reviewed-by: Mark Brown <broonie@kernel.org>
Link: https://lore.kernel.org/r/20230306114836.2575432-1-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoRevert "Revert "arm64: dma: Drop cache invalidation from arch_dma_prep_coherent()""
Will Deacon [Thu, 30 Mar 2023 15:23:03 +0000 (16:23 +0100)]
Revert "Revert "arm64: dma: Drop cache invalidation from arch_dma_prep_coherent()""

This reverts commit b7d9aae404841d9999b7476170867ae441e948d2.

With the Qualcomm remoteproc driver now modified to use a carveout
memory region in 57f72170a2b2 ("remoteproc: qcom_q6v5_mss: Use a
carveout to authenticate modem headers"), we can reinstate c44094eee32f
("arm64: dma: Drop cache invalidation from arch_dma_prep_coherent()")
which relaxes the arm64 implementation of arch_dma_prep_coherent() to
perform only a data cache clean operation, rather than a
clean-and-invalidate.

Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: uaccess: remove unnecessary earlyclobber
Mark Rutland [Tue, 14 Mar 2023 15:37:00 +0000 (15:37 +0000)]
arm64: uaccess: remove unnecessary earlyclobber

Currently the asm constraints for __get_mem_asm() mark the value
register as an earlyclobber operand. This means that the compiler can't
reuse the same register for both the address and value, even when the
value is not subsequently used.

There's no need for the value register to be marked as earlyclobber, as
it's only written to after the address register is consumed, even when
the access faults.

Remove the unnecessary earlyclobber.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230314153700.787701-5-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: uaccess: permit put_{user,kernel} to use zero register
Mark Rutland [Tue, 14 Mar 2023 15:36:59 +0000 (15:36 +0000)]
arm64: uaccess: permit put_{user,kernel} to use zero register

Currently the asm constraints for __put_mem_asm() require that the value
is placed in a "real" GPR (i.e. one other than [XW]ZR or SP). This means
that for cases such as:

__put_user(0, addr)

... the compiler has to move '0' into "real" GPR, e.g.

mov xN, #0
sttr xN, [<addr>]

This is unfortunate, as using the zero register would require fewer
instructions and save a "real" GPR for other usage, allowing the
compiler to generate:

sttr xzr, [<addr>]

Modify the asm constaints for __put_mem_asm() to permit the use of the
zero register for the value.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230314153700.787701-4-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: uaccess: permit __smp_store_release() to use zero register
Mark Rutland [Tue, 14 Mar 2023 15:36:58 +0000 (15:36 +0000)]
arm64: uaccess: permit __smp_store_release() to use zero register

Currently the asm constraints for __smp_store_release() require that the
value is placed in a "real" GPR (i.e. one other than [XW]ZR or SP).
This means that for cases such as:

    __smp_store_release(ptr, 0)

... the compiler has to move '0' into "real" GPR, e.g.

    mov     xN, #0
    stlr    xN, [<addr>]

This is unfortunate, as using the zero register would require fewer
instructions and save a "real" GPR for other usage, allowing the
compiler to generate:

    stlr    xzr, [<addr>]

Modify the asm constaints for __smp_store_release() to permit the use of
the zero register for the value.

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230314153700.787701-3-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: atomics: lse: improve cmpxchg implementation
Mark Rutland [Tue, 14 Mar 2023 15:36:57 +0000 (15:36 +0000)]
arm64: atomics: lse: improve cmpxchg implementation

For historical reasons, the LSE implementation of cmpxchg*() hard-codes
the GPRs to use, and shuffles registers around with MOVs. This is no
longer necessary, and can be simplified.

When the LSE cmpxchg implementation was added in commit:

  c342f78217e822d2 ("arm64: cmpxchg: patch in lse instructions when supported by the CPU")

... the LL/SC implementation of cmpxchg() would be placed out-of-line,
and the in-line assembly for cmpxchg would default to:

NOP
BL <ll_sc_cmpxchg*_implementation>
NOP

The LL/SC implementation of each cmpxchg() function accepted arguments
as per AAPCS64 rules, to it was necessary to place the pointer in x0,
the older value in X1, and the new value in x2, and acquire the return
value from x0. The LL/SC implementation required a temporary register
(e.g. for the STXR status value). As the LL/SC implementation preserved
the old value, the LSE implementation does likewise.

Since commit:

  addfc38672c73efd ("arm64: atomics: avoid out-of-line ll/sc atomics")

... the LSE and LL/SC implementations of cmpxchg are inlined as separate
asm blocks, with another branch choosing between thw two. Due to this,
it is no longer necessary for the LSE implementation to match the
register constraints of the LL/SC implementation. This was partially
dealt with by removing the hard-coded use of x30 in commit:

  3337cb5aea594e40 ("arm64: avoid using hard-coded registers for LSE atomics")

... but we didn't clean up the hard-coding of x0, x1, and x2.

This patch simplifies the LSE implementation of cmpxchg, removing the
register shuffling and directly clobbering the 'old' argument. This
gives the compiler greater freedom for register allocation, and avoids
redundant work.

The new constraints permit 'old' (Rs) and 'new' (Rt) to be allocated to
the same register when the initial values of the two are the same, e.g.
resulting in:

CAS X0, X0, [X1]

This is safe as Rs is only written back after the initial values of Rs
and Rt are consumed, and there are no UNPREDICTABLE behaviours to avoid
when Rs == Rt.

The new constraints also permit 'new' to be allocated to the zero
register, avoiding a MOV in a few cases. The same cannot be done for
'old' as it is both an input and output, and any caller of cmpxchg()
should care about the output value. Note that for CAS* the use of the
zero register never affects the ordering (while for SWP* the use of the
zero regsiter for the 'old' value drops any ACQUIRE semantic).

Compared to v6.2-rc4, a defconfig vmlinux is ~116KiB smaller, though the
resulting Image is the same size due to internal alignment and padding:

  [mark@lakrids:~/src/linux]% ls -al vmlinux-*
  -rwxr-xr-x 1 mark mark 137269304 Jan 16 11:59 vmlinux-after
  -rwxr-xr-x 1 mark mark 137387936 Jan 16 10:54 vmlinux-before
  [mark@lakrids:~/src/linux]% ls -al Image-*
  -rw-r--r-- 1 mark mark 38711808 Jan 16 11:59 Image-after
  -rw-r--r-- 1 mark mark 38711808 Jan 16 10:54 Image-before

This patch does not touch cmpxchg_double*() as that requires contiguous
register pairs, and separate patches will replace it with cmpxchg128*().

There should be no functional change as a result of this patch.

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Robin Murphy <robin.murphy@arm.com>
Cc: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20230314153700.787701-2-mark.rutland@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoRevert "ARM: mach-virt: Select PMUv3 driver by default"
Will Deacon [Tue, 28 Mar 2023 17:52:14 +0000 (18:52 +0100)]
Revert "ARM: mach-virt: Select PMUv3 driver by default"

This reverts commit 3b16f6268e660f15aed0bb97aefe87e893eb8882.

Selecting a Kconfig option that has its own set of dependencies tends to
end badly, and in this case 'randconfig' builds blew up on 32-bit ARM
where ARM_PMUV3 was being selecting with HW_PERF_EVENTS=n:

  |    drivers/perf/arm_pmuv3.c:68:5: error: use of undeclared identifier 'DTLB'
  |            [C(DTLB)][C(OP_READ)][C(RESULT_ACCESS)] = ARMV8_PMUV3_PERFCTR_L1D_TLB,
  |               ^
  |    fatal error: too many errors emitted, stopping now [-ferror-limit=]
  |    20 errors generated.
  |
  | Kconfig warnings: (for reference only)
  |    WARNING: unmet direct dependencies detected for ARM_PMUV3
  |    Depends on [n]: PERF_EVENTS [=y] && HW_PERF_EVENTS [=n] && (ARM [=y] && CPU_V7 [=y] || ARM64)
  |    Selected by [y]:
  |    - ARCH_VIRT [=y] && ARCH_MULTI_V7 [=y] && PERF_EVENTS [=y]

As suggested by Marc, just drop the 'select' clause altogether by
reverting the patch which introduced it.

Link: https://lore.kernel.org/r/202303281539.zzI4vpw1-lkp@intel.com
Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Arnd Bergmann <arnd@arndb.de>
Suggested-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: kexec: remove unnecessary (void*) conversions
Yu Zhe [Fri, 3 Mar 2023 02:57:15 +0000 (10:57 +0800)]
arm64: kexec: remove unnecessary (void*) conversions

Pointer variables of void * type do not require type cast.

Signed-off-by: Yu Zhe <yuzhe@nfschina.com>
Link: https://lore.kernel.org/r/20230303025715.32570-1-yuzhe@nfschina.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: armv8_deprecated: remove unnecessary (void*) conversions
Yu Zhe [Fri, 3 Mar 2023 02:50:47 +0000 (10:50 +0800)]
arm64: armv8_deprecated: remove unnecessary (void*) conversions

Pointer variables of void * type do not require type cast.

Signed-off-by: Yu Zhe <yuzhe@nfschina.com>
Link: https://lore.kernel.org/r/20230303025047.19717-1-yuzhe@nfschina.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agofirmware: arm_sdei: Fix sleep from invalid context BUG
Pierre Gondois [Thu, 16 Feb 2023 08:49:19 +0000 (09:49 +0100)]
firmware: arm_sdei: Fix sleep from invalid context BUG

Running a preempt-rt (v6.2-rc3-rt1) based kernel on an Ampere Altra
triggers:

  BUG: sleeping function called from invalid context at kernel/locking/spinlock_rt.c:46
  in_atomic(): 0, irqs_disabled(): 128, non_block: 0, pid: 24, name: cpuhp/0
  preempt_count: 0, expected: 0
  RCU nest depth: 0, expected: 0
  3 locks held by cpuhp/0/24:
    #0: ffffda30217c70d0 (cpu_hotplug_lock){++++}-{0:0}, at: cpuhp_thread_fun+0x5c/0x248
    #1: ffffda30217c7120 (cpuhp_state-up){+.+.}-{0:0}, at: cpuhp_thread_fun+0x5c/0x248
    #2: ffffda3021c711f0 (sdei_list_lock){....}-{3:3}, at: sdei_cpuhp_up+0x3c/0x130
  irq event stamp: 36
  hardirqs last  enabled at (35): [<ffffda301e85b7bc>] finish_task_switch+0xb4/0x2b0
  hardirqs last disabled at (36): [<ffffda301e812fec>] cpuhp_thread_fun+0x21c/0x248
  softirqs last  enabled at (0): [<ffffda301e80b184>] copy_process+0x63c/0x1ac0
  softirqs last disabled at (0): [<0000000000000000>] 0x0
  CPU: 0 PID: 24 Comm: cpuhp/0 Not tainted 5.19.0-rc3-rt5-[...]
  Hardware name: WIWYNN Mt.Jade Server [...]
  Call trace:
    dump_backtrace+0x114/0x120
    show_stack+0x20/0x70
    dump_stack_lvl+0x9c/0xd8
    dump_stack+0x18/0x34
    __might_resched+0x188/0x228
    rt_spin_lock+0x70/0x120
    sdei_cpuhp_up+0x3c/0x130
    cpuhp_invoke_callback+0x250/0xf08
    cpuhp_thread_fun+0x120/0x248
    smpboot_thread_fn+0x280/0x320
    kthread+0x130/0x140
    ret_from_fork+0x10/0x20

sdei_cpuhp_up() is called in the STARTING hotplug section,
which runs with interrupts disabled. Use a CPUHP_AP_ONLINE_DYN entry
instead to execute the cpuhp cb later, with preemption enabled.

SDEI originally got its own cpuhp slot to allow interacting
with perf. It got superseded by pNMI and this early slot is not
relevant anymore. [1]

Some SDEI calls (e.g. SDEI_1_0_FN_SDEI_PE_MASK) take actions on the
calling CPU. It is checked that preemption is disabled for them.
_ONLINE cpuhp cb are executed in the 'per CPU hotplug thread'.
Preemption is enabled in those threads, but their cpumask is limited
to 1 CPU.
Move 'WARN_ON_ONCE(preemptible())' statements so that SDEI cpuhp cb
don't trigger them.

Also add a check for the SDEI_1_0_FN_SDEI_PRIVATE_RESET SDEI call
which acts on the calling CPU.

[1]:
https://lore.kernel.org/all/5813b8c5-ae3e-87fd-fccc-94c9cd08816d@arm.com/

Suggested-by: James Morse <james.morse@arm.com>
Signed-off-by: Pierre Gondois <pierre.gondois@arm.com>
Reviewed-by: James Morse <james.morse@arm.com>
Link: https://lore.kernel.org/r/20230216084920.144064-1-pierre.gondois@arm.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm: uaccess: Remove memcpy_page_flushcache()
Ira Weiny [Wed, 15 Mar 2023 23:20:56 +0000 (16:20 -0700)]
arm: uaccess: Remove memcpy_page_flushcache()

Commit 21b56c847753 ("iov_iter: get rid of separate bvec and xarray
callbacks") removed the calls to memcpy_page_flushcache().

Remove the unnecessary memcpy_page_flushcache() call.

Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: "Dan Williams" <dan.j.williams@intel.com>
Cc: Catalin Marinas <catalin.marinas@arm.com>
Cc: Will Deacon <will@kernel.org>
Cc: linux-arm-kernel@lists.infradead.org
Signed-off-by: Ira Weiny <ira.weiny@intel.com>
Link: https://lore.kernel.org/r/20221230-kmap-x86-v1-3-15f1ecccab50@intel.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agomm,kfence: decouple kfence from page granularity mapping judgement
Zhenhua Huang [Fri, 17 Mar 2023 15:29:34 +0000 (23:29 +0800)]
mm,kfence: decouple kfence from page granularity mapping judgement

Kfence only needs its pool to be mapped as page granularity, if it is
inited early. Previous judgement was a bit over protected. From [1], Mark
suggested to "just map the KFENCE region a page granularity". So I
decouple it from judgement and do page granularity mapping for kfence
pool only. Need to be noticed that late init of kfence pool still requires
page granularity mapping.

Page granularity mapping in theory cost more(2M per 1GB) memory on arm64
platform. Like what I've tested on QEMU(emulated 1GB RAM) with
gki_defconfig, also turning off rodata protection:
Before:
[root@liebao ]# cat /proc/meminfo
MemTotal:         999484 kB
After:
[root@liebao ]# cat /proc/meminfo
MemTotal:        1001480 kB

To implement this, also relocate the kfence pool allocation before the
linear mapping setting up, arm64_kfence_alloc_pool is to allocate phys
addr, __kfence_pool is to be set after linear mapping set up.

LINK: [1] https://lore.kernel.org/linux-arm-kernel/Y+IsdrvDNILA59UN@FVFF77S0Q05N/
Suggested-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Zhenhua Huang <quic_zhenhuah@quicinc.com>
Reviewed-by: Kefeng Wang <wangkefeng.wang@huawei.com>
Reviewed-by: Marco Elver <elver@google.com>
Link: https://lore.kernel.org/r/1679066974-690-1-git-send-email-quic_zhenhuah@quicinc.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agodrivers/perf: apple_m1: Add Apple M2 support
Janne Grunau [Tue, 14 Feb 2023 10:38:02 +0000 (11:38 +0100)]
drivers/perf: apple_m1: Add Apple M2 support

The PMU itself is compatible with the one found on M1. We still know
next to nothing about the counters so keep using CPU uarch specific
compatibles/PMU names.

Signed-off-by: Janne Grunau <j@jannau.net>
Acked-by: Mark Rutland <mark.rutland@arm.com.
Reviewed-by: Hector Martin <marcan@marcan.st>
Link: https://lore.kernel.org/r/20230214-apple_m2_pmu-v1-2-9c9213ab9b63@jannau.net
Signed-off-by: Will Deacon <will@kernel.org>
19 months agodt-bindings: arm-pmu: Add PMU compatible strings for Apple M2 cores
Janne Grunau [Tue, 14 Feb 2023 10:38:01 +0000 (11:38 +0100)]
dt-bindings: arm-pmu: Add PMU compatible strings for Apple M2 cores

The PMUs on the Apple M2 cores avalanche and blizzard CPU are compatible
with M1 ones. As on M1 we don't know exactly what the counters count so
use a distinct compatible for each micro-architecture.
Apple's PMU counter description omits a counter for M2 so there
is some variation on the interpretation of the counters.

Signed-off-by: Janne Grunau <j@jannau.net>
Acked-by: Rob Herring <robh@kernel.org>
Reviewed-by: Hector Martin <marcan@marcan.st>
Link: https://lore.kernel.org/r/20230214-apple_m2_pmu-v1-1-9c9213ab9b63@jannau.net
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf: arm_cspmu: Fix variable dereference warning
Besar Wicaksono [Thu, 2 Mar 2023 20:57:01 +0000 (14:57 -0600)]
perf: arm_cspmu: Fix variable dereference warning

Fix warning message from smatch tool:
  | smatch warnings:
  | drivers/perf/arm_cspmu/arm_cspmu.c:1075 arm_cspmu_find_cpu_container()
  |    warn: variable dereferenced before check 'cpu_dev' (see line 1073)

Reported-by: kernel test robot <lkp@intel.com>
Reported-by: Dan Carpenter <error27@gmail.com>
Link: https://lore.kernel.org/r/202302191227.kc0V8fM7-lkp@intel.com/
Reviewed-by: Suzuki K Poulose <suzuki.poulose@arm.com>
Signed-off-by: Besar Wicaksono <bwicaksono@nvidia.com>
Acked-by: Randy Dunlap <rdunlap@infradead.org>
Link: https://lore.kernel.org/r/20230302205701.35323-1-bwicaksono@nvidia.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf/amlogic: Fix config1/config2 parsing issue
Jiucheng Xu [Thu, 9 Feb 2023 11:54:01 +0000 (19:54 +0800)]
perf/amlogic: Fix config1/config2 parsing issue

The 3th argument of for_each_set_bit is incorrect, fix them.

Fixes: 2016e2113d35 ("perf/amlogic: Add support for Amlogic meson G12 SoC DDR PMU driver")
Signed-off-by: Jiucheng Xu <jiucheng.xu@amlogic.com>
Link: https://lore.kernel.org/r/20230209115403.521868-1-jiucheng.xu@amlogic.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agodrivers/perf: Use devm_platform_get_and_ioremap_resource()
Yang Li [Thu, 16 Feb 2023 06:34:03 +0000 (14:34 +0800)]
drivers/perf: Use devm_platform_get_and_ioremap_resource()

Convert platform_get_resource(), devm_ioremap_resource() to a single
call to devm_platform_get_and_ioremap_resource(), as this is exactly
what this function does.

Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Reviewed-by: Shuai Xue <xueshuai@linux.alibaba.com>
Link: https://lore.kernel.org/r/20230216063403.9753-1-yang.lee@linux.alibaba.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agokbuild, drivers/perf: remove MODULE_LICENSE in non-modules
Nick Alcock [Fri, 17 Feb 2023 14:10:43 +0000 (14:10 +0000)]
kbuild, drivers/perf: remove MODULE_LICENSE in non-modules

Since commit 8b41fc4454e ("kbuild: create modules.builtin without
Makefile.modbuiltin or tristate.conf"), MODULE_LICENSE declarations
are used to identify modules. As a consequence, uses of the macro
in non-modules will cause modprobe to misidentify their containing
object file as a module when it is not (false positives), and modprobe
might succeed rather than failing with a suitable error message.

So remove it in the files in this commit, none of which can be built as
modules.

Signed-off-by: Nick Alcock <nick.alcock@oracle.com>
Suggested-by: Luis Chamberlain <mcgrof@kernel.org>
Cc: Luis Chamberlain <mcgrof@kernel.org>
Cc: linux-modules@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Cc: Hitomi Hasegawa <hasegawa-hitomi@fujitsu.com>
Cc: Will Deacon <will@kernel.org>
Cc: Mark Rutland <mark.rutland@arm.com>
Cc: linux-arm-kernel@lists.infradead.org
Link: https://lore.kernel.org/r/20230217141059.392471-9-nick.alcock@oracle.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf: qcom: Use devm_platform_get_and_ioremap_resource()
Yang Li [Wed, 15 Mar 2023 02:31:08 +0000 (10:31 +0800)]
perf: qcom: Use devm_platform_get_and_ioremap_resource()

According to commit 890cc39a8799 ("drivers: provide
devm_platform_get_and_ioremap_resource()"), convert
platform_get_resource(), devm_ioremap_resource() to a single
call to devm_platform_get_and_ioremap_resource(), as this is exactly
what this function does.

Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Link: https://lore.kernel.org/r/20230315023108.36953-1-yang.lee@linux.alibaba.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf: arm: Use devm_platform_get_and_ioremap_resource()
Yang Li [Wed, 15 Mar 2023 02:30:17 +0000 (10:30 +0800)]
perf: arm: Use devm_platform_get_and_ioremap_resource()

According to commit 890cc39a8799 ("drivers: provide
devm_platform_get_and_ioremap_resource()"), convert
platform_get_resource(), devm_ioremap_resource() to a single
call to devm_platform_get_and_ioremap_resource(), as this is exactly
what this function does.

Signed-off-by: Yang Li <yang.lee@linux.alibaba.com>
Link: https://lore.kernel.org/r/20230315023017.35789-1-yang.lee@linux.alibaba.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf/arm-cmn: Move overlapping wp_combine field
Ilkka Koskinen [Wed, 1 Mar 2023 17:55:40 +0000 (09:55 -0800)]
perf/arm-cmn: Move overlapping wp_combine field

As eventid field was expanded to support new mesh versions, it started to
overlap with wp_combine field. Move wp_combine to fix the issue.

Fixes: 23760a014417 ("perf/arm-cmn: Add CMN-700 support")
Signed-off-by: Ilkka Koskinen <ilkka@os.amperecomputing.com>
Link: https://lore.kernel.org/r/20230301175540.19891-1-ilkka@os.amperecomputing.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoARM: mach-virt: Select PMUv3 driver by default
Marc Zyngier [Fri, 17 Mar 2023 19:50:27 +0000 (15:50 -0400)]
ARM: mach-virt: Select PMUv3 driver by default

Since 32bit guests are not unlikely to run on an ARMv8 host,
let's select the PMUv3 driver, which allows the PMU to be used
on such systems.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-9-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoARM: perf: Allow the use of the PMUv3 driver on 32bit ARM
Marc Zyngier [Fri, 17 Mar 2023 19:50:26 +0000 (15:50 -0400)]
ARM: perf: Allow the use of the PMUv3 driver on 32bit ARM

The only thing stopping the PMUv3 driver from compiling on 32bit
is the lack of defined system registers names and the handful of
required helpers.

This is easily solved by providing the sysreg accessors and updating
the Kconfig entry.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Co-developed-by: Zaid Al-Bassam <zalbassam@google.com>
Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-8-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoARM: Make CONFIG_CPU_V7 valid for 32bit ARMv8 implementations
Marc Zyngier [Fri, 17 Mar 2023 19:50:25 +0000 (15:50 -0400)]
ARM: Make CONFIG_CPU_V7 valid for 32bit ARMv8 implementations

ARMv8 is a superset of ARMv7, and all the ARMv8 features are
discoverable with a set of ID registers. It means that we can
use CPU_V7 to guard ARMv8 features at compile time.

This commit simply amends the CPU_V7 configuration symbol comment
to reflect that CPU_V7 also covers ARMv8.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-7-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf: pmuv3: Change GENMASK to GENMASK_ULL
Zaid Al-Bassam [Fri, 17 Mar 2023 19:50:24 +0000 (15:50 -0400)]
perf: pmuv3: Change GENMASK to GENMASK_ULL

GENMASK macro uses "unsigned long" (32-bit wide on arm and 64-bit
on arm64), This causes build issues when enabling PMUv3 on arm as
it tries to access bits > 31. This patch switches the GENMASK to
GENMASK_ULL, which uses "unsigned long long" (64-bit on both arm
and arm64).

Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-6-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf: pmuv3: Move inclusion of kvm_host.h to the arch-specific helper
Zaid Al-Bassam [Fri, 17 Mar 2023 19:50:23 +0000 (15:50 -0400)]
perf: pmuv3: Move inclusion of kvm_host.h to the arch-specific helper

KVM host support is available only on arm64.

By moving the inclusion of kvm_host.h to an arm64-specific file,
the 32bit architecture will be able to implement dummy helpers.

Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-5-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoperf: pmuv3: Abstract PMU version checks
Zaid Al-Bassam [Fri, 17 Mar 2023 19:50:22 +0000 (15:50 -0400)]
perf: pmuv3: Abstract PMU version checks

The current PMU version definitions are available for arm64 only,
As we want to add PMUv3 support to arm (32-bit), abstracts
these definitions by using arch-specific helpers.

Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-4-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: perf: Abstract system register accesses away
Marc Zyngier [Fri, 17 Mar 2023 19:50:21 +0000 (15:50 -0400)]
arm64: perf: Abstract system register accesses away

As we want to enable 32bit support, we need to distanciate the
PMUv3 driver from the AArch64 system register names.

This patch moves all system register accesses to an architecture
specific include file, allowing the 32bit counterpart to be
slotted in at a later time.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Co-developed-by: Zaid Al-Bassam <zalbassam@google.com>
Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-3-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoarm64: perf: Move PMUv3 driver to drivers/perf
Marc Zyngier [Fri, 17 Mar 2023 19:50:20 +0000 (15:50 -0400)]
arm64: perf: Move PMUv3 driver to drivers/perf

Having the ARM PMUv3 driver sitting in arch/arm64/kernel is getting
in the way of being able to use perf on ARMv8 cores running a 32bit
kernel, such as 32bit KVM guests.

This patch moves it into drivers/perf/arm_pmuv3.c, with an include
file in include/linux/perf/arm_pmuv3.h. The only thing left in
arch/arm64 is some mundane perf stuff.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Zaid Al-Bassam <zalbassam@google.com>
Tested-by: Florian Fainelli <f.fainelli@gmail.com>
Link: https://lore.kernel.org/r/20230317195027.3746949-2-zalbassam@google.com
Signed-off-by: Will Deacon <will@kernel.org>
19 months agoftrace: selftest: remove broken trace_direct_tramp
Mark Rutland [Tue, 21 Mar 2023 14:04:24 +0000 (15:04 +0100)]
ftrace: selftest: remove broken trace_direct_tramp

The ftrace selftest code has a trace_direct_tramp() function which it
uses as a direct call trampoline. This happens to work on x86, since the
direct call's return address is in the usual place, and can be returned
to via a RET, but in general the calling convention for direct calls is
different from regular function calls, and requires a trampoline written
in assembly.

On s390, regular function calls place the return address in %r14, and an
ftrace patch-site in an instrumented function places the trampoline's
return address (which is within the instrumented function) in %r0,
preserving the original %r14 value in-place. As a regular C function
will return to the address in %r14, using a C function as the trampoline
results in the trampoline returning to the caller of the instrumented
function, skipping the body of the instrumented function.

Note that the s390 issue is not detcted by the ftrace selftest code, as
the instrumented function is trivial, and returning back into the caller
happens to be equivalent.

On arm64, regular function calls place the return address in x30, and
an ftrace patch-site in an instrumented function saves this into r9
and places the trampoline's return address (within the instrumented
function) in x30. A regular C function will return to the address in
x30, but will not restore x9 into x30. Consequently, using a C function
as the trampoline results in returning to the trampoline's return
address having corrupted x30, such that when the instrumented function
returns, it will return back into itself.

To avoid future issues in this area, remove the trace_direct_tramp()
function, and require that each architecture with direct calls provides
a stub trampoline, named ftrace_stub_direct_tramp. This can be written
to handle the architecture's trampoline calling convention, and in
future could be used elsewhere (e.g. in the ftrace ops sample, to
measure the overhead of direct calls), so we may as well always build it
in.

Link: https://lkml.kernel.org/r/20230321140424.345218-8-revest@chromium.org
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Cc: Li Huafei <lihuafei1@huawei.com>
Cc: Xu Kuohai <xukuohai@huawei.com>
Signed-off-by: Florent Revest <revest@chromium.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoftrace: Make DIRECT_CALLS work WITH_ARGS and !WITH_REGS
Florent Revest [Tue, 21 Mar 2023 14:04:23 +0000 (15:04 +0100)]
ftrace: Make DIRECT_CALLS work WITH_ARGS and !WITH_REGS

Direct called trampolines can be called in two ways:
- either from the ftrace callsite. In this case, they do not access any
  struct ftrace_regs nor pt_regs
- Or, if a ftrace ops is also attached, from the end of a ftrace
  trampoline. In this case, the call_direct_funcs ops is in charge of
  setting the direct call trampoline's address in a struct ftrace_regs

Since:

commit 9705bc709604 ("ftrace: pass fregs to arch_ftrace_set_direct_caller()")

The later case no longer requires a full pt_regs. It only needs a struct
ftrace_regs so DIRECT_CALLS can work with both WITH_ARGS or WITH_REGS.
With architectures like arm64 already abandoning WITH_REGS in favor of
WITH_ARGS, it's important to have DIRECT_CALLS work WITH_ARGS only.

Link: https://lkml.kernel.org/r/20230321140424.345218-7-revest@chromium.org
Signed-off-by: Florent Revest <revest@chromium.org>
Co-developed-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoftrace: Store direct called addresses in their ops
Florent Revest [Tue, 21 Mar 2023 14:04:22 +0000 (15:04 +0100)]
ftrace: Store direct called addresses in their ops

All direct calls are now registered using the register_ftrace_direct API
so each ops can jump to only one direct-called trampoline.

By storing the direct called trampoline address directly in the ops we
can save one hashmap lookup in the direct call ops and implement arm64
direct calls on top of call ops.

Link: https://lkml.kernel.org/r/20230321140424.345218-6-revest@chromium.org
Signed-off-by: Florent Revest <revest@chromium.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoftrace: Rename _ftrace_direct_multi APIs to _ftrace_direct APIs
Florent Revest [Tue, 21 Mar 2023 14:04:21 +0000 (15:04 +0100)]
ftrace: Rename _ftrace_direct_multi APIs to _ftrace_direct APIs

Now that the original _ftrace_direct APIs are gone, the "_multi"
suffixes only add confusion.

Link: https://lkml.kernel.org/r/20230321140424.345218-5-revest@chromium.org
Signed-off-by: Florent Revest <revest@chromium.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoftrace: Remove the legacy _ftrace_direct API
Florent Revest [Tue, 21 Mar 2023 14:04:20 +0000 (15:04 +0100)]
ftrace: Remove the legacy _ftrace_direct API

This API relies on a single global ops, used for all direct calls
registered with it. However, to implement arm64 direct calls, we need
each ops to point to a single direct call trampoline.

Link: https://lkml.kernel.org/r/20230321140424.345218-4-revest@chromium.org
Signed-off-by: Florent Revest <revest@chromium.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoftrace: Replace uses of _ftrace_direct APIs with _ftrace_direct_multi
Florent Revest [Tue, 21 Mar 2023 14:04:19 +0000 (15:04 +0100)]
ftrace: Replace uses of _ftrace_direct APIs with _ftrace_direct_multi

The _multi API requires that users keep their own ops but can enforce
that an op is only associated to one direct call.

Link: https://lkml.kernel.org/r/20230321140424.345218-3-revest@chromium.org
Signed-off-by: Florent Revest <revest@chromium.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Tested-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoftrace: Let unregister_ftrace_direct_multi() call ftrace_free_filter()
Florent Revest [Tue, 21 Mar 2023 14:04:18 +0000 (15:04 +0100)]
ftrace: Let unregister_ftrace_direct_multi() call ftrace_free_filter()

A common pattern when using the ftrace_direct_multi API is to unregister
the ops and also immediately free its filter. We've noticed it's very
easy for users to miss calling ftrace_free_filter().

This adds a "free_filters" argument to unregister_ftrace_direct_multi()
to both remind the user they should free filters and also to make their
life easier.

Link: https://lkml.kernel.org/r/20230321140424.345218-2-revest@chromium.org
Suggested-by: Steven Rostedt <rostedt@goodmis.org>
Signed-off-by: Florent Revest <revest@chromium.org>
Acked-by: Mark Rutland <mark.rutland@arm.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoLinux 6.3-rc3
Linus Torvalds [Sun, 19 Mar 2023 20:27:55 +0000 (13:27 -0700)]
Linux 6.3-rc3

19 months agoMerge tag 'trace-v6.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace...
Linus Torvalds [Sun, 19 Mar 2023 17:46:02 +0000 (10:46 -0700)]
Merge tag 'trace-v6.3-rc2' of git://git./linux/kernel/git/trace/linux-trace

Pull tracing fixes from Steven Rostedt:

 - Fix setting affinity of hwlat threads in containers

   Using sched_set_affinity() has unwanted side effects when being
   called within a container. Use set_cpus_allowed_ptr() instead

 - Fix per cpu thread management of the hwlat tracer:
    - Do not start per_cpu threads if one is already running for the CPU
    - When starting per_cpu threads, do not clear the kthread variable
      as it may already be set to running per cpu threads

 - Fix return value for test_gen_kprobe_cmd()

   On error the return value was overwritten by being set to the result
   of the call from kprobe_event_delete(), which would likely succeed,
   and thus have the function return success

 - Fix splice() reads from the trace file that was broken by commit
   36e2c7421f02 ("fs: don't allow splice read/write without explicit
   ops")

 - Remove obsolete and confusing comment in ring_buffer.c

   The original design of the ring buffer used struct page flags for
   tricks to optimize, which was shortly removed due to them being
   tricks. But a comment for those tricks remained

 - Set local functions and variables to static

* tag 'trace-v6.3-rc2' of git://git.kernel.org/pub/scm/linux/kernel/git/trace/linux-trace:
  tracing/hwlat: Replace sched_setaffinity with set_cpus_allowed_ptr
  ring-buffer: remove obsolete comment for free_buffer_page()
  tracing: Make splice_read available again
  ftrace: Set direct_ops storage-class-specifier to static
  trace/hwlat: Do not start per-cpu thread if it is already running
  trace/hwlat: Do not wipe the contents of per-cpu thread data
  tracing/osnoise: set several trace_osnoise.c variables storage-class-specifier to static
  tracing: Fix wrong return in kprobe_event_gen_test.c

19 months agotracing/hwlat: Replace sched_setaffinity with set_cpus_allowed_ptr
Costa Shulyupin [Thu, 16 Mar 2023 14:45:35 +0000 (16:45 +0200)]
tracing/hwlat: Replace sched_setaffinity with set_cpus_allowed_ptr

There is a problem with the behavior of hwlat in a container,
resulting in incorrect output. A warning message is generated:
"cpumask changed while in round-robin mode, switching to mode none",
and the tracing_cpumask is ignored. This issue arises because
the kernel thread, hwlatd, is not a part of the container, and
the function sched_setaffinity is unable to locate it using its PID.
Additionally, the task_struct of hwlatd is already known.
Ultimately, the function set_cpus_allowed_ptr achieves
the same outcome as sched_setaffinity, but employs task_struct
instead of PID.

Test case:

  # cd /sys/kernel/tracing
  # echo 0 > tracing_on
  # echo round-robin > hwlat_detector/mode
  # echo hwlat > current_tracer
  # unshare --fork --pid bash -c 'echo 1 > tracing_on'
  # dmesg -c

Actual behavior:

[573502.809060] hwlat_detector: cpumask changed while in round-robin mode, switching to mode none

Link: https://lore.kernel.org/linux-trace-kernel/20230316144535.1004952-1-costa.shul@redhat.com
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Fixes: 0330f7aa8ee63 ("tracing: Have hwlat trace migrate across tracing_cpumask CPUs")
Signed-off-by: Costa Shulyupin <costa.shul@redhat.com>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoring-buffer: remove obsolete comment for free_buffer_page()
Vlastimil Babka [Wed, 15 Mar 2023 14:24:46 +0000 (15:24 +0100)]
ring-buffer: remove obsolete comment for free_buffer_page()

The comment refers to mm/slob.c which is being removed. It comes from
commit ed56829cb319 ("ring_buffer: reset buffer page when freeing") and
according to Steven the borrowed code was a page mapcount and mapping
reset, which was later removed by commit e4c2ce82ca27 ("ring_buffer:
allocate buffer page pointer"). Thus the comment is not accurate anyway,
remove it.

Link: https://lore.kernel.org/linux-trace-kernel/20230315142446.27040-1-vbabka@suse.cz
Cc: Masami Hiramatsu <mhiramat@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Reported-by: Mike Rapoport <mike.rapoport@gmail.com>
Suggested-by: Steven Rostedt (Google) <rostedt@goodmis.org>
Fixes: e4c2ce82ca27 ("ring_buffer: allocate buffer page pointer")
Signed-off-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Mukesh Ojha <quic_mojha@quicinc.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agotracing: Make splice_read available again
Sung-hun Kim [Tue, 14 Mar 2023 01:37:07 +0000 (10:37 +0900)]
tracing: Make splice_read available again

Since the commit 36e2c7421f02 ("fs: don't allow splice read/write
without explicit ops") is applied to the kernel, splice() and
sendfile() calls on the trace file (/sys/kernel/debug/tracing
/trace) return EINVAL.

This patch restores these system calls by initializing splice_read
in file_operations of the trace file. This patch only enables such
functionalities for the read case.

Link: https://lore.kernel.org/linux-trace-kernel/20230314013707.28814-1-sfoon.kim@samsung.com
Cc: stable@vger.kernel.org
Fixes: 36e2c7421f02 ("fs: don't allow splice read/write without explicit ops")
Signed-off-by: Sung-hun Kim <sfoon.kim@samsung.com>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agoMerge tag 'tty-6.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty
Linus Torvalds [Sun, 19 Mar 2023 17:09:58 +0000 (10:09 -0700)]
Merge tag 'tty-6.3-rc3' of git://git./linux/kernel/git/gregkh/tty

Pull tty/serial driver fixes from Greg KH:
 "Here are some small tty and serial driver fixes for 6.3-rc3 to resolve
  some reported issues.

  They include:

   - 8250 driver Kconfig issue pointed out by you that showed up in -rc1

   - qcom-geni serial driver fixes

   - various 8250 driver fixes for reported problems

   - fsl_lpuart driver fixes

   - serdev fix for regression in -rc1

   - vt.c bugfix

  All have been in linux-next for over a week with no reported problems"

* tag 'tty-6.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/tty:
  tty: vt: protect KD_FONT_OP_GET_TALL from unbound access
  serial: qcom-geni: drop bogus uart_write_wakeup()
  serial: qcom-geni: fix mapping of empty DMA buffer
  serial: qcom-geni: fix DMA mapping leak on shutdown
  serial: qcom-geni: fix console shutdown hang
  serdev: Set fwnode for serdev devices
  tty: serial: fsl_lpuart: fix race on RX DMA shutdown
  serial: 8250_pci1xxxx: Disable SERIAL_8250_PCI1XXXX config by default
  serial: 8250_fsl: fix handle_irq locking
  serial: 8250_em: Fix UART port type
  serial: 8250: ASPEED_VUART: select REGMAP instead of depending on it
  tty: serial: fsl_lpuart: skip waiting for transmission complete when UARTCTRL_SBK is asserted
  Revert "tty: serial: fsl_lpuart: adjust SERIAL_FSL_LPUART_CONSOLE config dependency"

19 months agoMerge tag 'char-misc-6.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh...
Linus Torvalds [Sun, 19 Mar 2023 17:04:58 +0000 (10:04 -0700)]
Merge tag 'char-misc-6.3-rc3' of git://git./linux/kernel/git/gregkh/char-misc

Pull char/misc driver fixes from Greg KH:
 "Here are a few small char/misc/other driver subsystem patches to
  resolve reported problems for 6.3-rc3.

  Included in here are:

   - Interconnect driver fixes for reported problems

   - Memory driver fixes for reported problems

   - nvmem core fix

   - firmware driver fix for reported problem

  All of these have been in linux-next for a while with no reported
  issues"

* tag 'char-misc-6.3-rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/gregkh/char-misc: (23 commits)
  memory: tegra30-emc: fix interconnect registration race
  memory: tegra20-emc: fix interconnect registration race
  memory: tegra124-emc: fix interconnect registration race
  memory: tegra: fix interconnect registration race
  interconnect: exynos: drop redundant link destroy
  interconnect: exynos: fix registration race
  interconnect: exynos: fix node leak in probe PM QoS error path
  interconnect: qcom: msm8974: fix registration race
  interconnect: qcom: rpmh: fix registration race
  interconnect: qcom: rpmh: fix probe child-node error handling
  interconnect: qcom: rpm: fix registration race
  nvmem: core: return -ENOENT if nvmem cell is not found
  firmware: xilinx: don't make a sleepable memory allocation from an atomic context
  interconnect: qcom: rpm: fix probe child-node error handling
  interconnect: qcom: osm-l3: fix registration race
  interconnect: imx: fix registration race
  interconnect: fix provider registration API
  interconnect: fix icc_provider_del() error handling
  interconnect: fix mem leak when freeing nodes
  interconnect: qcom: qcm2290: Fix MASTER_SNOC_BIMC_NRT
  ...

19 months agoMerge tag 'ras_urgent_for_v6.3_rc3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 Mar 2023 16:57:53 +0000 (09:57 -0700)]
Merge tag 'ras_urgent_for_v6.3_rc3' of git://git./linux/kernel/git/tip/tip

Pull RAS fix from Borislav Petkov:

 - Flush out logged errors immediately after MCA banks configuration
   changes over sysfs have been done instead of waiting until something
   else triggers the workqueue later - another error or the polling
   interval cycle is reached

* tag 'ras_urgent_for_v6.3_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mce: Make sure logged MCEs are processed after sysfs update

19 months agoMerge tag 'perf_urgent_for_v6.3_rc3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 Mar 2023 16:47:55 +0000 (09:47 -0700)]
Merge tag 'perf_urgent_for_v6.3_rc3' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Borislav Petkov:

 - Check whether sibling events have been deactivated before adding them
   to groups

 - Update the proper event time tracking variable depending on the event
   type

 - Fix a memory overwrite issue due to using the wrong function argument
   when outputting perf events

* tag 'perf_urgent_for_v6.3_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf: Fix check before add_event_to_groups() in perf_group_detach()
  perf: fix perf_event_context->time
  perf/core: Fix perf_output_begin parameter is incorrectly invoked in perf_event_bpf_output

19 months agoMerge tag 'x86_urgent_for_v6.3_rc3' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 19 Mar 2023 16:43:41 +0000 (09:43 -0700)]
Merge tag 'x86_urgent_for_v6.3_rc3' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Borislav Petkov:
 "There's a little bit more 'movement' in there for my taste but it
  needs to happen and should make the code better after it.

   - Check cmdline_find_option()'s return value before further
     processing

   - Clear temporary storage in the resctrl code to prevent access to an
     unexistent MSR

   - Add a simple throttling mechanism to protect the hypervisor from
     potentially malicious SEV guests issuing requests in rapid
     succession.

     In order to not jeopardize the sanity of everyone involved in
     maintaining this code, the request issuing side has received a
     cleanup, split in more or less trivial, small and digestible
     pieces. Otherwise, the code was threatening to become an
     unmaintainable mess.

     Therefore, that cleanup is marked indirectly also for stable so
     that there's no differences between the upstream code and the
     stable variant when it comes down to backporting more there"

* tag 'x86_urgent_for_v6.3_rc3' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/mm: Fix use of uninitialized buffer in sme_enable()
  x86/resctrl: Clear staged_config[] before and after it is used
  virt/coco/sev-guest: Add throttling awareness
  virt/coco/sev-guest: Convert the sw_exit_info_2 checking to a switch-case
  virt/coco/sev-guest: Do some code style cleanups
  virt/coco/sev-guest: Carve out the request issuing logic into a helper
  virt/coco/sev-guest: Remove the disable_vmpck label in handle_guest_request()
  virt/coco/sev-guest: Simplify extended guest request handling
  virt/coco/sev-guest: Check SEV_SNP attribute at probe time

19 months agoMerge tag 'ext4_for_linus_urgent' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 19 Mar 2023 16:38:26 +0000 (09:38 -0700)]
Merge tag 'ext4_for_linus_urgent' of git://git./linux/kernel/git/tytso/ext4

Pull ext4 fix from Ted Ts'o:
 "Fix a double unlock bug on an error path in ext4, found by smatch and
  syzkaller"

* tag 'ext4_for_linus_urgent' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  ext4: fix possible double unlock when moving a directory

19 months agoftrace: Set direct_ops storage-class-specifier to static
Tom Rix [Sat, 11 Mar 2023 13:51:13 +0000 (08:51 -0500)]
ftrace: Set direct_ops storage-class-specifier to static

smatch reports this warning
kernel/trace/ftrace.c:2594:19: warning:
  symbol 'direct_ops' was not declared. Should it be static?

The variable direct_ops is only used in ftrace.c, so it should be static

Link: https://lore.kernel.org/linux-trace-kernel/20230311135113.711824-1-trix@redhat.com
Signed-off-by: Tom Rix <trix@redhat.com>
Acked-by: Masami Hiramatsu (Google) <mhiramat@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>
19 months agotrace/hwlat: Do not start per-cpu thread if it is already running
Tero Kristo [Fri, 10 Mar 2023 10:04:51 +0000 (12:04 +0200)]
trace/hwlat: Do not start per-cpu thread if it is already running

The hwlatd tracer will end up starting multiple per-cpu threads with
the following script:

    #!/bin/sh
    cd /sys/kernel/debug/tracing
    echo 0 > tracing_on
    echo hwlat > current_tracer
    echo per-cpu > hwlat_detector/mode
    echo 100000 > hwlat_detector/width
    echo 200000 > hwlat_detector/window
    echo 1 > tracing_on

To fix the issue, check if the hwlatd thread for the cpu is already
running, before starting a new one. Along with the previous patch, this
avoids running multiple instances of the same CPU thread on the system.

Link: https://lore.kernel.org/all/20230302113654.2984709-1-tero.kristo@linux.intel.com/
Link: https://lkml.kernel.org/r/20230310100451.3948583-3-tero.kristo@linux.intel.com
Cc: stable@vger.kernel.org
Fixes: f46b16520a087 ("trace/hwlat: Implement the per-cpu mode")
Signed-off-by: Tero Kristo <tero.kristo@linux.intel.com>
Acked-by: Daniel Bristot de Oliveira <bristot@kernel.org>
Signed-off-by: Steven Rostedt (Google) <rostedt@goodmis.org>