In cases like IPI, we could be queueing an interrupt for a VCPU
that is already running and is not about to exit, because the
VCPU has entered the VM with the interrupt pending and would
not trap on EOI'ing that interrupt. This could result to delays
in interrupt deliveries or even loss of interrupts.
To guarantee prompt interrupt injection, here we have to try to
kick the VCPU.

Signed-off-by: Shih-Wei Li <shihwei@cs.columbia.edu>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

In our VGIC implementation we limit the number of SPIs to a number
that the userland application told us. Accordingly we limit the
allocation of memory for virtual IRQs to that number.
However in our MMIO dispatcher we didn't check if we ever access an
IRQ beyond that limit, leading to out-of-bound accesses.
Add a test against the number of allocated SPIs in check_region().
Adjust the VGIC_ADDR_TO_INT macro to avoid an actual division, which
is not implemented on ARM(32).

[maz: cleaned-up original patch]

Cc: stable@vger.kernel.org
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

The conversion done by commit 3706feac ("KVM: Remove deprecated
create_singlethread_workqueue") is broken.  It flushes a single work
item &irqfd->shutdown instead of all of them, and even worse if there
is no irqfd on the list then you get a NULL pointer dereference.
Revert the virt/kvm/eventfd.c part of that patch; to avoid the
deprecated function, just allocate our own workqueue---it does
not even have to be unbound---with alloc_workqueue.

Fixes: 3706feac
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

This patch unexports the low-level __get_user_pages() function.

Recent refactoring of the get_user_pages* functions allow flags to be
passed through get_user_pages() which eliminates the need for access to
this function from its one user, kvm.

We can see that the two calls to get_user_pages() which replace
__get_user_pages() in kvm_main.c are equivalent by examining their call
stacks:

  get_user_page_nowait():
    get_user_pages(start, 1, flags, page, NULL)
    __get_user_pages_locked(current, current->mm, start, 1, page, NULL, NULL,
			    false, flags | FOLL_TOUCH)
    __get_user_pages(current, current->mm, start, 1,
		     flags | FOLL_TOUCH | FOLL_GET, page, NULL, NULL)

  check_user_page_hwpoison():
    get_user_pages(addr, 1, flags, NULL, NULL)
    __get_user_pages_locked(current, current->mm, addr, 1, NULL, NULL, NULL,
			    false, flags | FOLL_TOUCH)
    __get_user_pages(current, current->mm, addr, 1, flags | FOLL_TOUCH, NULL,
		     NULL, NULL)

Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

This removes the redundant 'write' and 'force' parameters from
__get_user_pages_unlocked() to make the use of FOLL_FORCE explicit in
callers as use of this flag can result in surprising behaviour (and
hence bugs) within the mm subsystem.

Signed-off-by: Lorenzo Stoakes <lstoakes@gmail.com>
Acked-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Jan Kara <jack@suse.cz>
Acked-by: Michal Hocko <mhocko@suse.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

If the vgic hasn't been created and initialized, we shouldn't attempt to
look at its data structures or flush/sync anything to the GIC hardware.

This fixes an issue reported by Alexander Graf when using a userspace
irqchip.

Fixes: 0919e84c ("KVM: arm/arm64: vgic-new: Add IRQ sync/flush framework")
Cc: stable@vger.kernel.org
Reported-by: Alexander Graf <agraf@suse.de>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

If userspace creates a PMU for the VCPU, but doesn't create an in-kernel
irqchip, then we end up in a nasty path where we try to take an
uninitialized spinlock, which can lead to all sorts of breakages.

Luckily, QEMU always creates the VGIC before the PMU, so we can
establish this as ABI and check for the VGIC in the PMU init stage.
This can be relaxed at a later time if we want to support PMU with a
userspace irqchip.

Cc: stable@vger.kernel.org
Cc: Shannon Zhao <shannon.zhao@linaro.org>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

This patch allows to build and use vgic-v3 in 32-bit mode.

Unfortunately, it can not be split in several steps without extra
stubs to keep patches independent and bisectable.  For instance,
virt/kvm/arm/vgic/vgic-v3.c uses function from vgic-v3-sr.c, handling
access to GICv3 cpu interface from the guest requires vgic_v3.vgic_sre
to be already defined.

It is how support has been done:

* handle SGI requests from the guest

* report configured SRE on access to GICv3 cpu interface from the guest

* required vgic-v3 macros are provided via uapi.h

* static keys are used to select GIC backend

* to make vgic-v3 build KVM_ARM_VGIC_V3 guard is removed along with
  the static inlines

Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

We have couple of 64-bit registers defined in GICv3 architecture, so
unsigned long accesses to these registers will only access a single
32-bit part of that regitser. On the other hand these registers can't
be accessed as 64-bit with a single instruction like ldrd/strd or
ldmia/stmia if we run a 32-bit host because KVM does not support
access to MMIO space done by these instructions.

It means that a 32-bit guest accesses these registers in 32-bit
chunks, so the only thing we need to do is to ensure that
extract_bytes() always takes 64-bit data.

Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

Well, this patch is looking ahead of time, but we'll get following
compiler warnings as soon as we introduce vgic-v3 to 32-bit world

  CC      arch/arm/kvm/../../../virt/kvm/arm/vgic/vgic-mmio-v3.o
arch/arm/kvm/../../../virt/kvm/arm/vgic/vgic-mmio-v3.c: In function 'vgic_mmio_read_v3r_typer':
arch/arm/kvm/../../../virt/kvm/arm/vgic/vgic-mmio-v3.c:184:35: warning: left shift count >= width of type [-Wshift-count-overflow]
  value = (mpidr & GENMASK(23, 0)) << 32;
                                   ^
In file included from ./include/linux/kernel.h:10:0,
                 from ./include/asm-generic/bug.h:13,
                 from ./arch/arm/include/asm/bug.h:59,
                 from ./include/linux/bug.h:4,
                 from ./include/linux/io.h:23,
                 from ./arch/arm/include/asm/arch_gicv3.h:23,
                 from ./include/linux/irqchip/arm-gic-v3.h:411,
                 from arch/arm/kvm/../../../virt/kvm/arm/vgic/vgic-mmio-v3.c:14:
arch/arm/kvm/../../../virt/kvm/arm/vgic/vgic-mmio-v3.c: In function 'vgic_v3_dispatch_sgi':
./include/linux/bitops.h:6:24: warning: left shift count >= width of type [-Wshift-count-overflow]
 #define BIT(nr)   (1UL << (nr))
                        ^
arch/arm/kvm/../../../virt/kvm/arm/vgic/vgic-mmio-v3.c:614:20: note: in expansion of macro 'BIT'
  broadcast = reg & BIT(ICC_SGI1R_IRQ_ROUTING_MODE_BIT);
                    ^
Let's fix them now.

Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

By now ITS code guarded with KVM_ARM_VGIC_V3 config option which was
introduced to hide everything specific to vgic-v3 from 32-bit world.
We are going to support vgic-v3 in 32-bit world and KVM_ARM_VGIC_V3
will gone, but we don't have support for ITS there yet and we need to
continue keeping ITS away.
Introduce the new config option to prevent ITS code being build in
32-bit mode when support for vgic-v3 is done.

Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

So we can reuse the code under arch/arm

Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

Currently GIC backend is selected via alternative framework and this
is fine. We are going to introduce vgic-v3 to 32-bit world and there
we don't have patching framework in hand, so we can either check
support for GICv3 every time we need to choose which backend to use or
try to optimise it by using static keys. The later looks quite
promising because we can share logic involved in selecting GIC backend
between architectures if both uses static keys.

This patch moves arm64 from alternative to static keys framework for
selecting GIC backend. For that we embed static key into vgic_global
and enable the key during vgic initialisation based on what has
already been exposed by the host GIC driver.

Acked-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Vladimir Murzin <vladimir.murzin@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

This commit adds the ability for archs to export
per-vcpu information via a new per-vcpu dir in
the VM's debugfs directory.

If kvm_arch_has_vcpu_debugfs() returns true, then KVM
will create a vcpu dir for each vCPU in the VM's
debugfs directory. Then kvm_arch_create_vcpu_debugfs()
is responsible for populating each vcpu directory
with arch specific entries.

The per-vcpu path in debugfs will look like:

/sys/kernel/debug/kvm/29162-10/vcpu0
/sys/kernel/debug/kvm/29162-10/vcpu1

This is all arch specific for now because the only
user of this interface (x86) wants to export x86-specific
per-vcpu information to user-space.

Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

This make it possible to call kvm_destroy_vm_debugfs() from
kvm_create_vm_debugfs() in error conditions.

Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Luiz Capitulino <lcapitulino@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Remove two unnecessary labels now that kvm_timer_hyp_init is not
creating its own workqueue anymore.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

If, when proxying a GICV access at EL2, we detect that the guest is
doing something silly, report an EL1 SError instead ofgnoring the
access.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

So far, we've been disabling KVM on systems where the GICV region couldn't
be safely given to a guest. Now that we're able to handle this access
safely by emulating it in HYP, we can enable this feature when we detect
an unsafe configuration.

Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

Now that we have the necessary infrastructure to handle MMIO accesses
in HYP, perform the GICV access on behalf of the guest. This requires
checking that the access is strictly 32bit, properly aligned, and
falls within the expected range.

When all condition are satisfied, we perform the access and tell
the rest of the HYP code that the instruction has been correctly
emulated.

Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

In order to efficiently perform the GICV access on behalf of the
guest, we need to be able to avoid going back all the way to
the host kernel.

For this, we introduce a new hook in the world switch code,
conveniently placed just after populating the fault info.
At that point, we only have saved/restored the GP registers,
and we can quickly perform all the required checks (data abort,
translation fault, valid faulting syndrome, not an external
abort, not a PTW).

Coming back from the emulation code, we need to skip the emulated
instruction. This involves an additional bit of save/restore in
order to be able to access the guest's PC (and possibly CPSR if
this is a 32bit guest).

At this stage, no emulation code is provided.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

As we plan to do some emulation at HYP, let's make kvm_skip_instr32
as part of the hyp_text section. This doesn't preclude the kernel
from using it.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

Add the bit of glue and const-ification that is required to use
the code inherited from the arm64 port, and move over to it.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

It would make some sense to share the conditional execution code
between 32 and 64bit. In order to achieve this, let's move that
code to virt/kvm/arm/aarch32.c. While we're at it, drop a
superfluous BUG_ON() that wasn't that useful.

Following patches will migrate the 32bit port to that code base.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

As kvm_set_routing_entry() was changing prototype between 4.7 and 4.8,
an ugly hack was put in place in order to survive both building in
-next and the merge window.

Now that everything has been merged, let's dump the compatibility
hack for good.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

Just a rename so we can implement a v3-specific function later.

We take the chance to get rid of the V2/V3 ops comments as well.

No functional change.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>

As we are about to deal with multiple data types and situations where
the vgic should not be initialized when doing userspace accesses on the
register attributes, factor out the functionality of
vgic_attr_regs_access into smaller bits which can be reused by a new
function later.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>

vms and vcpus have statistics associated with them which can be viewed
within the debugfs. Currently it is assumed within the vcpu_stat_get() and
vm_stat_get() functions that all of these statistics are represented as
u32s, however the next patch adds some u64 vcpu statistics.

Change all vcpu statistics to u64 and modify vcpu_stat_get() accordingly.
Since vcpu statistics are per vcpu, they will only be updated by a single
vcpu at a time so this shouldn't present a problem on 32-bit machines
which can't atomically increment 64-bit numbers. However vm statistics
could potentially be updated by multiple vcpus from that vm at a time.
To avoid the overhead of atomics make all vm statistics ulong such that
they are 64-bit on 64-bit systems where they can be atomically incremented
and are 32-bit on 32-bit systems which may not be able to atomically
increment 64-bit numbers. Modify vm_stat_get() to expect ulongs.

Signed-off-by: Suraj Jitindar Singh <sjitindarsingh@gmail.com>
Reviewed-by: David Matlack <dmatlack@google.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Paul Mackerras <paulus@ozlabs.org>

The workqueue "irqfd_cleanup_wq" queues a single work item
&irqfd->shutdown and hence doesn't require ordering. It is a host-wide
workqueue for issuing deferred shutdown requests aggregated from all
vm* instances. It is not being used on a memory reclaim path.
Hence, it has been converted to use system_wq.
The work item has been flushed in kvm_irqfd_release().

The workqueue "wqueue" queues a single work item &timer->expired
and hence doesn't require ordering. Also, it is not being used on
a memory reclaim path. Hence, it has been converted to use system_wq.

System workqueues have been able to handle high level of concurrency
for a long time now and hence it's not required to have a singlethreaded
workqueue just to gain concurrency. Unlike a dedicated per-cpu workqueue
created with create_singlethread_workqueue(), system_wq allows multiple
work items to overlap executions even on the same CPU; however, a
per-cpu workqueue doesn't have any CPU locality or global ordering
guarantee unless the target CPU is explicitly specified and thus the
increase of local concurrency shouldn't make any difference.

Signed-off-by: Bhaktipriya Shridhar <bhaktipriya96@gmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Similarily to f005bd7e ("clocksource/arm_arch_timer: Force
per-CPU interrupt to be level-triggered"), make sure we can
survive an interrupt that has been misconfigured as edge-triggered
by forcing it to be level-triggered (active low is assumed, but
the GIC doesn't really care whether this is high or low).

Hopefully, the amount of shouting in the kernel log will convince
the user to do something about their firmware.

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

When a guest wants to map a device-ID/event-ID combination that is
already mapped, we may end up in a situation where an LPI is never
"put", thus never being freed.
Since the GICv3 spec says that mapping an already mapped LPI is
UNPREDICTABLE, lets just bail out early in this situation to avoid
any potential leaks.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

When userspace provides the doorbell address for an MSI to be
injected into the guest, we find a KVM device which feels responsible.
Lets check that this device is really an emulated ITS before we make
real use of the container_of-ed pointer.

  [ Moved NULL-pointer check to caller of static function
    - Christoffer ]

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

Currently we register an ITS device upon userland issuing the CTLR_INIT
ioctl to mark initialization of the ITS as done.
This deviates from the initialization sequence of the existing GIC
devices and does not play well with the way QEMU handles things.
To be more in line with what we are used to, register the ITS(es) just
before the first VCPU is about to run, so in the map_resources() call.
This involves iterating through the list of KVM devices and map each
ITS that we find.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Tested-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

There are two problems with the current implementation of the MMIO
handlers for the propbaser and pendbaser:

First, the write to the value itself is not guaranteed to be an atomic
64-bit write so two concurrent writes to the structure field could be
intermixed.

Second, because we do a read-modify-update operation without any
synchronization, if we have two 32-bit accesses to separate parts of the
register, we can loose one of them.

By using the atomic cmpxchg64 we should cover both issues above.

Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

KVM devices were manipulating list data structures without any form of
synchronization, and some implementations of the create operations also
suffered from a lack of synchronization.

Now when we've split the xics create operation into create and init, we
can hold the kvm->lock mutex while calling the create operation and when
manipulating the devices list.

The error path in the generic code gets slightly ugly because we have to
take the mutex again and delete the device from the list, but holding
the mutex during anon_inode_getfd or releasing/locking the mutex in the
common non-error path seemed wrong.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

As we are about to hold the kvm->lock during the create operation on KVM
devices, we should move the call to xics_debugfs_init into its own
function, since holding a mutex over extended amounts of time might not
be a good idea.

Introduce an init operation on the kvm_device_ops struct which cannot
fail and call this, if configured, after the device has been created.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

Right now the following sequence of events can happen:

  1. Thread X calls vgic_put_irq
  2. Thread Y calls vgic_add_lpi
  3. Thread Y gets lpi_list_lock
  4. Thread X drops the ref count to 0 and blocks on lpi_list_lock
  5. Thread Y finds the irq via the lpi_list_lock, raises the ref
     count to 1, and release the lpi_list_lock.
  6. Thread X proceeds and frees the irq.

Avoid this by holding the spinlock around the kref_put.

Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

During low memory conditions, we could be dereferencing a NULL pointer
when vgic_add_lpi fails to allocate memory.

Consider for example this call sequence:

  vgic_its_cmd_handle_mapi
      itte->irq = vgic_add_lpi(kvm, lpi_nr);
          update_lpi_config(kvm, itte->irq, NULL);
              ret = kvm_read_guest(kvm, propbase + irq->intid
	                                             ^^^^
						     kaboom?

Instead, return an error pointer from vgic_add_lpi and check the return
value from its single caller.

Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

According to the KVM API documentation a successful MSI injection
should return a value > 0 on success.
Return possible errors in vgic_its_trigger_msi() and report a
successful injection back to userland, while also reporting the
case where the MSI could not be delivered due to the guest not
having the LPI mapped, for instance.

Signed-off-by: Andre Przywara <andre.przywara@arm.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Christoffer Dall <christoffer.dall@linaro.org>
Signed-off-by: Christoffer Dall <christoffer.dall@linaro.org>

kvm_set_routing_entry is changing in -next, and causes things to
explode. Add a temporary workaround that should be dropped when
we hit 4.8-rc1

Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>

Up to now, only irqchip routing entries could be set. This patch
adds the capability to insert MSI routing entries.

For ARM64, let's also increase KVM_MAX_IRQ_ROUTES to 4096: this
include SPI irqchip routes plus MSI routes. In the future this
might be extended.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Andre Przywara <andre.przywara@arm.com>
Signed-off-by: Marc Zyngier <marc.zyngier@arm.com>