that helps the debugging of IRQ affinity logic bugs, and fix a memory leak.
Signed-off-by: Ingo Molnar <mingo@kernel.org>
-----BEGIN PGP SIGNATURE-----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=oKU9
-----END PGP SIGNATURE-----
Merge tag 'irq-urgent-2020-08-02' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull irq fixes from Ingo Molnar:
"Fix a recent IRQ affinities regression, add in a missing debugfs
printout that helps the debugging of IRQ affinity logic bugs, and fix
a memory leak"
* tag 'irq-urgent-2020-08-02' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
genirq/debugfs: Add missing irqchip flags
genirq/affinity: Make affinity setting if activated opt-in
irqdomain/treewide: Free firmware node after domain removal
- Removal of the tremendously unpopular read_barrier_depends() barrier,
which is a NOP on all architectures apart from Alpha, in favour of
allowing architectures to override READ_ONCE() and do whatever dance
they need to do to ensure address dependencies provide LOAD ->
LOAD/STORE ordering. This work also offers a potential solution if
compilers are shown to convert LOAD -> LOAD address dependencies into
control dependencies (e.g. under LTO), as weakly ordered architectures
will effectively be able to upgrade READ_ONCE() to smp_load_acquire().
The latter case is not used yet, but will be discussed further at LPC.
- Make the MSI/IOMMU input/output ID translation PCI agnostic, augment
the MSI/IOMMU ACPI/OF ID mapping APIs to accept an input ID
bus-specific parameter and apply the resulting changes to the device
ID space provided by the Freescale FSL bus.
- arm64 support for TLBI range operations and translation table level
hints (part of the ARMv8.4 architecture version).
- Time namespace support for arm64.
- Export the virtual and physical address sizes in vmcoreinfo for
makedumpfile and crash utilities.
- CPU feature handling cleanups and checks for programmer errors
(overlapping bit-fields).
- ACPI updates for arm64: disallow AML accesses to EFI code regions and
kernel memory.
- perf updates for arm64.
- Miscellaneous fixes and cleanups, most notably PLT counting
optimisation for module loading, recordmcount fix to ignore
relocations other than R_AARCH64_CALL26, CMA areas reserved for
gigantic pages on 16K and 64K configurations.
- Trivial typos, duplicate words.
-----BEGIN PGP SIGNATURE-----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=mh79
-----END PGP SIGNATURE-----
Merge tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux
Pull arm64 and cross-arch updates from Catalin Marinas:
"Here's a slightly wider-spread set of updates for 5.9.
Going outside the usual arch/arm64/ area is the removal of
read_barrier_depends() series from Will and the MSI/IOMMU ID
translation series from Lorenzo.
The notable arm64 updates include ARMv8.4 TLBI range operations and
translation level hint, time namespace support, and perf.
Summary:
- Removal of the tremendously unpopular read_barrier_depends()
barrier, which is a NOP on all architectures apart from Alpha, in
favour of allowing architectures to override READ_ONCE() and do
whatever dance they need to do to ensure address dependencies
provide LOAD -> LOAD/STORE ordering.
This work also offers a potential solution if compilers are shown
to convert LOAD -> LOAD address dependencies into control
dependencies (e.g. under LTO), as weakly ordered architectures will
effectively be able to upgrade READ_ONCE() to smp_load_acquire().
The latter case is not used yet, but will be discussed further at
LPC.
- Make the MSI/IOMMU input/output ID translation PCI agnostic,
augment the MSI/IOMMU ACPI/OF ID mapping APIs to accept an input ID
bus-specific parameter and apply the resulting changes to the
device ID space provided by the Freescale FSL bus.
- arm64 support for TLBI range operations and translation table level
hints (part of the ARMv8.4 architecture version).
- Time namespace support for arm64.
- Export the virtual and physical address sizes in vmcoreinfo for
makedumpfile and crash utilities.
- CPU feature handling cleanups and checks for programmer errors
(overlapping bit-fields).
- ACPI updates for arm64: disallow AML accesses to EFI code regions
and kernel memory.
- perf updates for arm64.
- Miscellaneous fixes and cleanups, most notably PLT counting
optimisation for module loading, recordmcount fix to ignore
relocations other than R_AARCH64_CALL26, CMA areas reserved for
gigantic pages on 16K and 64K configurations.
- Trivial typos, duplicate words"
Link: http://lkml.kernel.org/r/20200710165203.31284-1-will@kernel.org
Link: http://lkml.kernel.org/r/20200619082013.13661-1-lorenzo.pieralisi@arm.com
* tag 'arm64-upstream' of git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux: (82 commits)
arm64: use IRQ_STACK_SIZE instead of THREAD_SIZE for irq stack
arm64/mm: save memory access in check_and_switch_context() fast switch path
arm64: sigcontext.h: delete duplicated word
arm64: ptrace.h: delete duplicated word
arm64: pgtable-hwdef.h: delete duplicated words
bus: fsl-mc: Add ACPI support for fsl-mc
bus/fsl-mc: Refactor the MSI domain creation in the DPRC driver
of/irq: Make of_msi_map_rid() PCI bus agnostic
of/irq: make of_msi_map_get_device_domain() bus agnostic
dt-bindings: arm: fsl: Add msi-map device-tree binding for fsl-mc bus
of/device: Add input id to of_dma_configure()
of/iommu: Make of_map_rid() PCI agnostic
ACPI/IORT: Add an input ID to acpi_dma_configure()
ACPI/IORT: Remove useless PCI bus walk
ACPI/IORT: Make iort_msi_map_rid() PCI agnostic
ACPI/IORT: Make iort_get_device_domain IRQ domain agnostic
ACPI/IORT: Make iort_match_node_callback walk the ACPI namespace for NC
arm64: enable time namespace support
arm64/vdso: Restrict splitting VVAR VMA
arm64/vdso: Handle faults on timens page
...
Add ACPI support in the fsl-mc driver. Driver parses MC DSDT table to
extract memory and other resources.
Interrupt (GIC ITS) information is extracted from the MADT table
by drivers/irqchip/irq-gic-v3-its-fsl-mc-msi.c.
IORT table is parsed to configure DMA.
Signed-off-by: Makarand Pawagi <makarand.pawagi@nxp.com>
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Signed-off-by: Laurentiu Tudor <laurentiu.tudor@nxp.com>
Link: https://lore.kernel.org/r/20200619082013.13661-13-lorenzo.pieralisi@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
The DPRC driver is not taking into account the msi-map property
and assumes that the icid is the same as the stream ID. Although
this assumption is correct, generalize the code to include a
translation between icid and streamID.
Furthermore do not just copy the MSI domain from parent (for child
containers), but use the information provided by the msi-map property.
If the msi-map property is missing from the device tree retain the old
behaviour for backward compatibility ie the child DPRC objects
inherit the MSI domain from the parent.
Signed-off-by: Diana Craciun <diana.craciun@oss.nxp.com>
Acked-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200619082013.13661-12-lorenzo.pieralisi@arm.com
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
John reported that on a RK3288 system the perf per CPU interrupts are all
affine to CPU0 and provided the analysis:
"It looks like what happens is that because the interrupts are not per-CPU
in the hardware, armpmu_request_irq() calls irq_force_affinity() while
the interrupt is deactivated and then request_irq() with IRQF_PERCPU |
IRQF_NOBALANCING.
Now when irq_startup() runs with IRQ_STARTUP_NORMAL, it calls
irq_setup_affinity() which returns early because IRQF_PERCPU and
IRQF_NOBALANCING are set, leaving the interrupt on its original CPU."
This was broken by the recent commit which blocked interrupt affinity
setting in hardware before activation of the interrupt. While this works in
general, it does not work for this particular case. As contrary to the
initial analysis not all interrupt chip drivers implement an activate
callback, the safe cure is to make the deferred interrupt affinity setting
at activation time opt-in.
Implement the necessary core logic and make the two irqchip implementations
for which this is required opt-in. In hindsight this would have been the
right thing to do, but ...
Fixes: baedb87d1b ("genirq/affinity: Handle affinity setting on inactive interrupts correctly")
Reported-by: John Keeping <john@metanate.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Marc Zyngier <maz@kernel.org>
Acked-by: Marc Zyngier <maz@kernel.org>
Cc: stable@vger.kernel.org
Link: https://lkml.kernel.org/r/87blk4tzgm.fsf@nanos.tec.linutronix.de
-----BEGIN PGP SIGNATURE-----
iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAl8DWosUHHBib256aW5p
QHJlZGhhdC5jb20ACgkQv/vSX3jHroO8cAf/UskNg8qoLGG17rQwhFpmigSllbiJ
TAyi3tpb1Y0Z2MfYeGkeiEb1L34bS28Cxl929DoqI3hrXy1wDCmsHPB5c3URXrzd
aswvr7pJtQV9iH1ykaS2woFJnOUovMFsFYMhj46yUPoAvdKOZKvuqcduxbogYHFw
YeRhS+1lGfiP2A0j3O/nnNJ0wq+FxKO46G3CgWeqG75+FSL6y/tl0bZJUMKKajQZ
GNaOv/CYCHAfUdvgy0ZitRD8lV8yxng3dYGjm+a52Kmn2ZWiFlxNrnxzHySk16Rn
Lq6MfFOqgrYpoZv7SnsFYnRE05U5bEFQ8BGr22fImQ+ktKDgq+9gv6cKwA==
=+DN/
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull kvm fixes from Paolo Bonzini:
"Bugfixes and a one-liner patch to silence a sparse warning"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
KVM: arm64: Stop clobbering x0 for HVC_SOFT_RESTART
KVM: arm64: PMU: Fix per-CPU access in preemptible context
KVM: VMX: Use KVM_POSSIBLE_CR*_GUEST_BITS to initialize guest/host masks
KVM: x86: Mark CR4.TSD as being possibly owned by the guest
KVM: x86: Inject #GP if guest attempts to toggle CR4.LA57 in 64-bit mode
kvm: use more precise cast and do not drop __user
KVM: x86: bit 8 of non-leaf PDPEs is not reserved
KVM: X86: Fix async pf caused null-ptr-deref
KVM: arm64: vgic-v4: Plug race between non-residency and v4.1 doorbell
KVM: arm64: pvtime: Ensure task delay accounting is enabled
KVM: arm64: Fix kvm_reset_vcpu() return code being incorrect with SVE
KVM: arm64: Annotate hyp NMI-related functions as __always_inline
KVM: s390: reduce number of IO pins to 1
When making a vPE non-resident because it has hit a blocking WFI,
the doorbell can fire at any time after the write to the RD.
Crucially, it can fire right between the write to GICR_VPENDBASER
and the write to the pending_last field in the its_vpe structure.
This means that we would overwrite pending_last with stale data,
and potentially not wakeup until some unrelated event (such as
a timer interrupt) puts the vPE back on the CPU.
GICv4 isn't affected by this as we actively mask the doorbell on
entering the guest, while GICv4.1 automatically manages doorbell
delivery without any hypervisor-driven masking.
Use the vpe_lock to synchronize such update, which solves the
problem altogether.
Fixes: ae699ad348 ("irqchip/gic-v4.1: Move doorbell management to the GICv4 abstraction layer")
Reported-by: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
The GIC driver uses a RMW sequence to update the affinity, and
relies on the gic_lock_irqsave/gic_unlock_irqrestore sequences
to update it atomically.
But these sequences only expand into anything meaningful if
the BL_SWITCHER option is selected, which almost never happens.
It also turns out that using a RMW and locks is just as silly,
as the GIC distributor supports byte accesses for the GICD_TARGETRn
registers, which when used make the update atomic by definition.
Drop the terminally broken code and replace it by a byte write.
Fixes: 04c8b0f82c ("irqchip/gic: Make locking a BL_SWITCHER only feature")
Cc: stable@vger.kernel.org
Signed-off-by: Marc Zyngier <maz@kernel.org>
Anup originally re-spun his patch set to include this fix, but it was a bit too
late for my PR so I've split it out.
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200611175302.253540-1-palmer@dabbelt.com
readx_poll_timeout() can sleep if @sleep_us is specified by the caller,
and is therefore unsafe to be used inside the atomic context, which is
this case when we use it to poll the GICR_VPENDBASER.Dirty bit in
irq_set_vcpu_affinity() callback.
Let's convert to its atomic version instead which helps to get the v4.1
board back to life!
Fixes: 96806229ca ("irqchip/gic-v4.1: Add support for VPENDBASER's Dirty+Valid signaling")
Signed-off-by: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200605052345.1494-1-yuzenghui@huawei.com
Instead of directly calling RISC-V timer interrupt handler from
RISC-V local interrupt conntroller driver, this patch implements
RISC-V timer interrupt as a per-CPU interrupt using per-CPU APIs
of Linux IRQ subsystem.
Signed-off-by: Anup Patel <anup.patel@wdc.com>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
The RISC-V per-HART local interrupt controller manages software
interrupts, timer interrupts, external interrupts (which are routed
via the platform level interrupt controller) and other per-HART
local interrupts.
We add a driver for the RISC-V local interrupt controller, which
eventually replaces the RISC-V architecture code, allowing for a
better split between arch code and drivers.
The driver is compliant with RISC-V Hart-Level Interrupt Controller
DT bindings located at:
Documentation/devicetree/bindings/interrupt-controller/riscv,cpu-intc.txt
Co-developed-by: Palmer Dabbelt <palmer@dabbelt.com>
Signed-off-by: Anup Patel <anup.patel@wdc.com>
[Palmer: Cleaned up warnings]
Signed-off-by: Palmer Dabbelt <palmer@dabbelt.com>
The plic_find_hart_id() can be useful to other interrupt controller
drivers (such as RISC-V local interrupt driver) so we rename this
function to riscv_of_parent_hartid() and place it in arch directory
along with riscv_of_processor_hartid().
Signed-off-by: Anup Patel <anup.patel@wdc.com>
Reviewed-by: Atish Patra <atish.patra@wdc.com>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Acked-by: Palmer Dabbelt <palmerdabbelt@google.com>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Palmer Dabbelt <palmerdabbelt@google.com>
This commit:
818e915fbac5: ("irqchip: Add Loongson HyperTransport Vector support")
Added a MIPS-only driver, but turned on compilation on all other architectures as well:
config LOONGSON_HTVEC
bool "Loongson3 HyperTransport Interrupt Vector Controller"
depends on MACH_LOONGSON64 || COMPILE_TEST
But this driver was never build tested on any other architecture than MIPS:
drivers/irqchip/irq-loongson-htvec.c: In function ‘htvec_irq_dispatch’:
drivers/irqchip/irq-loongson-htvec.c:59:3: error: implicit declaration of function ‘spurious_interrupt’; did you mean ‘smp_reboot_interrupt’? [-Werror=implicit-function-declaration]
Because spurious_interrupt() only exists on MIPS.
So make it MIPS-only.
Signed-off-by: Ingo Molnar <mingo@kernel.org>
Cc: Jiaxun Yang <jiaxun.yang@flygoat.com>
Cc: Marc Zyngier <maz@kernel.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
This controller appears on Loongson LS7A family of PCH to transform
interrupts from PCI MSI into HyperTransport vectorized interrrupts
and send them to procrssor's HT vector controller.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200528152757.1028711-6-jiaxun.yang@flygoat.com
This controller appears on Loongson LS7A family of PCH to transform
interrupts from devices into HyperTransport vectorized interrrupts
and send them to procrssor's HT vector controller.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200528152757.1028711-4-jiaxun.yang@flygoat.com
This controller appears on Loongson-3 chips for receiving interrupt
vectors from PCH's PIC and PCH's PCIe MSI interrupts.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200528152757.1028711-2-jiaxun.yang@flygoat.com
For multiple PLIC instances, the plic_init() is called once for each
PLIC instance. Due to this we have two issues:
1. cpuhp_setup_state() is called multiple times
2. plic_starting_cpu() can crash for boot CPU if cpuhp_setup_state()
is called before boot CPU PLIC handler is available.
Address both issues by only initializing the HP notifiers when
the boot CPU setup is complete.
Fixes: f1ad1133b1 ("irqchip/sifive-plic: Add support for multiple PLICs")
Signed-off-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Acked-by: Palmer Dabbelt <palmerdabbelt@google.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20200518091441.94843-3-anup.patel@wdc.com
For multiple PLIC instances, each PLIC can only target a subset of
CPUs which is represented by "lmask" in the "struct plic_priv".
Currently, the default irq affinity for each PLIC interrupt is all
online CPUs which is illegal value for default irq affinity when we
have multiple PLIC instances. To fix this, we now set "lmask" as the
default irq affinity in for each interrupt in plic_irqdomain_map().
Fixes: f1ad1133b1 ("irqchip/sifive-plic: Add support for multiple PLICs")
Signed-off-by: Anup Patel <anup.patel@wdc.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Acked-by: Palmer Dabbelt <palmerdabbelt@google.com>
Cc: stable@vger.kernel.org
Link: https://lore.kernel.org/r/20200518091441.94843-2-anup.patel@wdc.com
(E)PPIs are per-CPU interrupts, so we want each CPU to go and enable them
via enable_percpu_irq(); this also means we want IRQ_NOAUTOEN for them as
the autoenable would lead to calling irq_enable() instead of the more
appropriate irq_percpu_enable().
Calling irq_set_percpu_devid() is enough to get just that since it trickles
down to irq_set_percpu_devid_flags(), which gives us IRQ_NOAUTOEN (and a
few others). Setting IRQ_NOAUTOEN *again* right after this call is just
redundant, so don't do it.
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200521223500.834-1-valentin.schneider@arm.com
When mapping a LPI, the ITS driver picks the first possible
affinity, which is in most cases CPU0, assuming that if
that's not suitable, someone will come and set the affinity
to something more interesting.
It apparently isn't the case, and people complain of poor
performance when many interrupts are glued to the same CPU.
So let's place the interrupts by finding the "least loaded"
CPU (that is, the one that has the fewer LPIs mapped to it).
So called 'managed' interrupts are an interesting case where
the affinity is actually dictated by the kernel itself, and
we should honor this.
Reported-by: John Garry <john.garry@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Tested-by: John Garry <john.garry@huawei.com>
Link: https://lore.kernel.org/r/1575642904-58295-1-git-send-email-john.garry@huawei.com
Link: https://lore.kernel.org/r/20200515165752.121296-3-maz@kernel.org
In order to improve the distribution of LPIs among CPUs, let start by
tracking the number of LPIs assigned to CPUs, both for managed and
non-managed interrupts (as separate counters).
Signed-off-by: Marc Zyngier <maz@kernel.org>
Tested-by: John Garry <john.garry@huawei.com>
Link: https://lore.kernel.org/r/20200515165752.121296-2-maz@kernel.org
A PLIC may not be connected to all the cores. In that case, nr_contexts
may be less than num_possible_cpus. This requirement is only valid a single
PLIC is the only interrupt controller for the whole system.
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Signed-off-by: "Wesley W. Terpstra" <wesley@sifive.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Palmer Dabbelt <palmerdabbelt@google.com>
Acked-by: Palmer Dabbelt <palmerdabbelt@google.com>
Link: https://lore.kernel.org/r/20200512172636.96299-1-atish.patra@wdc.com
[Atish: Modified the commit text]
With an SMP configuration, gic_smp_init() calls set_smp_cross_call().
set_smp_cross_call() is marked with "__init".
So gic_smp_init() should also be marked with "__init".
gic_smp_init() is only called from gic_init_bases().
gic_init_bases() is also marked with "__init";
So marking gic_smp_init() with "__init" is fine.
Signed-off-by: Ingo Rohloff <ingo.rohloff@lauterbach.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200422112857.4300-1-ingo.rohloff@lauterbach.com
Fix the following sparse warning:
drivers/irqchip/irq-bcm7038-l1.c:419:12: warning: symbol
'bcm7038_l1_of_init' was not declared. Should it be static?
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200417074036.46594-1-yanaijie@huawei.com
Fix the following sparse warning:
drivers/irqchip/irq-mvebu-icu.c:69:1: warning: symbol 'legacy_bindings'
was not declared. Should it be static?
Reported-by: Hulk Robot <hulkci@huawei.com>
Signed-off-by: Jason Yan <yanaijie@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200417074046.46771-1-yanaijie@huawei.com
Running a lockedp-enabled kernel on a vim3l board (Amlogic SM1)
leads to the following splat:
[ 13.557138] WARNING: HARDIRQ-safe -> HARDIRQ-unsafe lock order detected
[ 13.587485] ip/456 [HC0[0]:SC0[0]:HE0:SE1] is trying to acquire:
[ 13.625922] ffff000059908cf0 (&irq_desc_lock_class){-.-.}-{2:2}, at: __setup_irq+0xf8/0x8d8
[ 13.632273] which would create a new lock dependency:
[ 13.637272] (&irq_desc_lock_class){-.-.}-{2:2} -> (&ctl->lock){+.+.}-{2:2}
[ 13.644209]
[ 13.644209] but this new dependency connects a HARDIRQ-irq-safe lock:
[ 13.654122] (&irq_desc_lock_class){-.-.}-{2:2}
[ 13.654125]
[ 13.654125] ... which became HARDIRQ-irq-safe at:
[ 13.664759] lock_acquire+0xec/0x368
[ 13.666926] _raw_spin_lock+0x60/0x88
[ 13.669979] handle_fasteoi_irq+0x30/0x178
[ 13.674082] generic_handle_irq+0x38/0x50
[ 13.678098] __handle_domain_irq+0x6c/0xc8
[ 13.682209] gic_handle_irq+0x5c/0xb0
[ 13.685872] el1_irq+0xd0/0x180
[ 13.689010] arch_cpu_idle+0x40/0x220
[ 13.692732] default_idle_call+0x54/0x60
[ 13.696677] do_idle+0x23c/0x2e8
[ 13.699903] cpu_startup_entry+0x30/0x50
[ 13.703852] rest_init+0x1e0/0x2b4
[ 13.707301] arch_call_rest_init+0x18/0x24
[ 13.711449] start_kernel+0x4ec/0x51c
[ 13.715167]
[ 13.715167] to a HARDIRQ-irq-unsafe lock:
[ 13.722426] (&ctl->lock){+.+.}-{2:2}
[ 13.722430]
[ 13.722430] ... which became HARDIRQ-irq-unsafe at:
[ 13.732319] ...
[ 13.732324] lock_acquire+0xec/0x368
[ 13.735985] _raw_spin_lock+0x60/0x88
[ 13.739452] meson_gpio_irq_domain_alloc+0xcc/0x290
[ 13.744392] irq_domain_alloc_irqs_hierarchy+0x24/0x60
[ 13.749586] __irq_domain_alloc_irqs+0x160/0x2f0
[ 13.754254] irq_create_fwspec_mapping+0x118/0x320
[ 13.759073] irq_create_of_mapping+0x78/0xa0
[ 13.763360] of_irq_get+0x6c/0x80
[ 13.766701] of_mdiobus_register_phy+0x10c/0x238 [of_mdio]
[ 13.772227] of_mdiobus_register+0x158/0x380 [of_mdio]
[ 13.777388] mdio_mux_init+0x180/0x2e8 [mdio_mux]
[ 13.782128] g12a_mdio_mux_probe+0x290/0x398 [mdio_mux_meson_g12a]
[ 13.788349] platform_drv_probe+0x5c/0xb0
[ 13.792379] really_probe+0xe4/0x448
[ 13.795979] driver_probe_device+0xe8/0x140
[ 13.800189] __device_attach_driver+0x94/0x120
[ 13.804639] bus_for_each_drv+0x84/0xd8
[ 13.808474] __device_attach+0xe4/0x168
[ 13.812361] device_initial_probe+0x1c/0x28
[ 13.816592] bus_probe_device+0xa4/0xb0
[ 13.820430] deferred_probe_work_func+0xa8/0x100
[ 13.825064] process_one_work+0x264/0x688
[ 13.829088] worker_thread+0x4c/0x458
[ 13.832768] kthread+0x154/0x158
[ 13.836018] ret_from_fork+0x10/0x18
[ 13.839612]
[ 13.839612] other info that might help us debug this:
[ 13.839612]
[ 13.850354] Possible interrupt unsafe locking scenario:
[ 13.850354]
[ 13.855720] CPU0 CPU1
[ 13.858774] ---- ----
[ 13.863242] lock(&ctl->lock);
[ 13.866330] local_irq_disable();
[ 13.872233] lock(&irq_desc_lock_class);
[ 13.878705] lock(&ctl->lock);
[ 13.884297] <Interrupt>
[ 13.886857] lock(&irq_desc_lock_class);
[ 13.891014]
[ 13.891014] *** DEADLOCK ***
The issue can occur when CPU1 is doing something like irq_set_type()
and CPU0 performing an interrupt allocation, for example. Taking
an interrupt (like the one being reconfigured) would lead to a deadlock.
A solution to this is:
- Reorder the locking so that meson_gpio_irq_update_bits takes the lock
itself at all times, instead of relying on the caller to lock or not,
hence making the RMW sequence atomic,
- Rework the critical section in meson_gpio_irq_request_channel to only
cover the allocation itself, and let the gpio_irq_sel_pin callback
deal with its own locking if required,
- Take the private spin-lock with interrupts disabled at all times
Reviewed-by: Jerome Brunet <jbrunet@baylibre.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
As per the PLIC specification, maximum priority threshold value is 0x7
not 0xF. Even though it doesn't cause any error in qemu/hifive unleashed,
there may be some implementation which checks the upper bound resulting in
an illegal access.
Fixes: ccbe80bad5 ("irqchip/sifive-plic: Enable/Disable external interrupts upon cpu online/offline")
Signed-off-by: Atish Patra <atish.patra@wdc.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200403014609.71831-1-atish.patra@wdc.com
The ti_sci_inta_irq_handler() does not take into account INTA IRQs state
(masked/unmasked) as it uses INTA_STATUS_CLEAR_j register to get INTA IRQs
status, which provides raw status value.
This causes hard IRQ handlers to be called or threaded handlers to be
scheduled many times even if corresponding INTA IRQ is masked.
Above, first of all, affects the LEVEL interrupts processing and causes
unexpected behavior up the system stack or crash.
Fix it by using the Interrupt Masked Status INTA_STATUSM_j register which
provides masked INTA IRQs status.
Fixes: 9f1463b86c ("irqchip/ti-sci-inta: Add support for Interrupt Aggregator driver")
Signed-off-by: Grygorii Strashko <grygorii.strashko@ti.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Lokesh Vutla <lokeshvutla@ti.com>
Link: https://lore.kernel.org/r/20200408191532.31252-1-grygorii.strashko@ti.com
Cc: stable@vger.kernel.org
Using irq_domain_free_irqs_common() on the irqdomain free path will
leave the MSI descriptor unfreed when platform devices get removed.
Properly free it by MSI domain free function.
Fixes: 9650c60ebf ("irqchip/mbigen: Create irq domain for each mbigen device")
Signed-off-by: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200408114352.1604-1-yuzenghui@huawei.com
Although the vSGIs are not directly visible to the host, they still
get moved around by the CPU hotplug, for example. This results in
the kernel moaning on the console, such as:
genirq: irq_chip GICv4.1-sgi did not update eff. affinity mask of irq 38
Updating the effective affinity on set_affinity() fixes it.
Reviewed-by: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
When a vPE is made resident, the GIC starts parsing the virtual pending
table to deliver pending interrupts. This takes place asynchronously,
and can at times take a long while. Long enough that the vcpu enters
the guest and hits WFI before any interrupt has been signaled yet.
The vcpu then exits, blocks, and now gets a doorbell. Rince, repeat.
In order to avoid the above, a (optional on GICv4, mandatory on v4.1)
feature allows the GIC to feedback to the hypervisor whether it is
done parsing the VPT by clearing the GICR_VPENDBASER.Dirty bit.
The hypervisor can then wait until the GIC is ready before actually
running the vPE.
Plug the detection code as well as polling on vPE schedule. While
at it, tidy-up the kernel message that displays the GICv4 optional
features.
Reviewed-by: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
driver which affected the PPC users.
-----BEGIN PGP SIGNATURE-----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=pZAC
-----END PGP SIGNATURE-----
Merge tag 'irq-urgent-2020-04-05' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull irq fixes from Thomas Gleixner:
"Two reverts addressing regressions of the Xilinx interrupt controller
driver which affected the PPC users"
* tag 'irq-urgent-2020-04-05' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
Revert "irqchip/xilinx: Enable generic irq multi handler"
Revert "irqchip/xilinx: Do not call irq_set_default_host()"
Treewide:
- Cleanup of setup_irq() which is not longer required because the
memory allocator is available early. Most cleanup changes come
through the various maintainer trees, so the final removal of
setup_irq() is postponed towards the end of the merge window.
Core:
- Protection against unsafe invocation of interrupt handlers and unsafe
interrupt injection including a fixup of the offending PCI/AER error
injection mechanism.
Invoking interrupt handlers from arbitrary contexts, i.e. outside of
an actual interrupt, can cause inconsistent state on the fragile
x86 interrupt affinity changing hardware trainwreck.
Drivers:
- Second wave of support for the new ARM GICv4.1
- Multi-instance support for Xilinx and PLIC interrupt controllers
- CPU-Hotplug support for PLIC
- The obligatory new driver for X1000 TCU
- Enhancements, cleanups and fixes all over the place
-----BEGIN PGP SIGNATURE-----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=DZ/G
-----END PGP SIGNATURE-----
Merge tag 'irq-core-2020-03-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip
Pull irq updates from Thomas Gleixner:
"Updates for the interrupt subsystem:
Treewide:
- Cleanup of setup_irq() which is not longer required because the
memory allocator is available early.
Most cleanup changes come through the various maintainer trees, so
the final removal of setup_irq() is postponed towards the end of
the merge window.
Core:
- Protection against unsafe invocation of interrupt handlers and
unsafe interrupt injection including a fixup of the offending
PCI/AER error injection mechanism.
Invoking interrupt handlers from arbitrary contexts, i.e. outside
of an actual interrupt, can cause inconsistent state on the
fragile x86 interrupt affinity changing hardware trainwreck.
Drivers:
- Second wave of support for the new ARM GICv4.1
- Multi-instance support for Xilinx and PLIC interrupt controllers
- CPU-Hotplug support for PLIC
- The obligatory new driver for X1000 TCU
- Enhancements, cleanups and fixes all over the place"
* tag 'irq-core-2020-03-30' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip: (58 commits)
unicore32: Replace setup_irq() by request_irq()
sh: Replace setup_irq() by request_irq()
hexagon: Replace setup_irq() by request_irq()
c6x: Replace setup_irq() by request_irq()
alpha: Replace setup_irq() by request_irq()
irqchip/gic-v4.1: Eagerly vmap vPEs
irqchip/gic-v4.1: Add VSGI property setup
irqchip/gic-v4.1: Add VSGI allocation/teardown
irqchip/gic-v4.1: Move doorbell management to the GICv4 abstraction layer
irqchip/gic-v4.1: Plumb set_vcpu_affinity SGI callbacks
irqchip/gic-v4.1: Plumb get/set_irqchip_state SGI callbacks
irqchip/gic-v4.1: Plumb mask/unmask SGI callbacks
irqchip/gic-v4.1: Add initial SGI configuration
irqchip/gic-v4.1: Plumb skeletal VSGI irqchip
irqchip/stm32: Retrigger both in eoi and unmask callbacks
irqchip/gic-v3: Move irq_domain_update_bus_token to after checking for NULL domain
irqchip/xilinx: Do not call irq_set_default_host()
irqchip/xilinx: Enable generic irq multi handler
irqchip/xilinx: Fill error code when irq domain registration fails
irqchip/xilinx: Add support for multiple instances
...
This controller appeared on Loongson-3 family of chips to receive
interrupts from PCH PIC.
It is a I8259 with optimized interrupt polling flow. We can poll
interrupt number from HT vector directly but still have to follow
standard I8259 routines to mask, unmask and EOI.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Co-developed-by: Huacai Chen <chenhc@lemote.com>
Signed-off-by: Huacai Chen <chenhc@lemote.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
The 1.0 version of that controller has a bug that status bit
of LPC IRQ sometimes doesn't get set correctly.
So we can always blame LPC IRQ when spurious interrupt happens
at the parent interrupt line which LPC IRQ supposed to route
to.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Co-developed-by: Huacai Chen <chenhc@lemote.com>
Signed-off-by: Huacai Chen <chenhc@lemote.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
This controller appeared on Loongson family of chips as the primary
package interrupt source.
Signed-off-by: Jiaxun Yang <jiaxun.yang@flygoat.com>
Co-developed-by: Huacai Chen <chenhc@lemote.com>
Signed-off-by: Huacai Chen <chenhc@lemote.com>
Reviewed-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Thomas Bogendoerfer <tsbogend@alpha.franken.de>
Now that we have HW-accelerated SGIs being delivered to VPEs, it
becomes required to map the VPEs on all ITSs instead of relying
on the lazy approach that we would use when using the ITS-list
mechanism.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Zenghui Yu <yuzenghui@huawei.com>
Link: https://lore.kernel.org/r/20200304203330.4967-17-maz@kernel.org
In order to hide some of the differences between v4.0 and v4.1, move
the doorbell management out of the KVM code, and into the GICv4-specific
layer. This allows the calling code to ask for the doorbell when blocking,
and otherwise to leave the doorbell permanently disabled.
This matches the v4.1 code perfectly, and only results in a minor
refactoring of the v4.0 code.
Signed-off-by: Marc Zyngier <maz@kernel.org>
Reviewed-by: Zenghui Yu <yuzenghui@huawei.com>
Link: https://lore.kernel.org/r/20200304203330.4967-14-maz@kernel.org
Linus Torvalds