linux.git/arch/arm/kvm, branch v4.1.29

KVM: arm/arm64: Stop leaking vcpu pid references

2016-07-11T02:27:12+00:00

[ Upstream commit 591d215afcc2f94e8e2c69a63c924c044677eb31 ]

kvm provides kvm_vcpu_uninit(), which amongst other things, releases the
last reference to the struct pid of the task that was last running the vcpu.

On arm64 built with CONFIG_DEBUG_KMEMLEAK, starting a guest with kvmtool,
then killing it with SIGKILL results (after some considerable time) in:
> cat /sys/kernel/debug/kmemleak
> unreferenced object 0xffff80007d5ea080 (size 128):
>  comm "lkvm", pid 2025, jiffies 4294942645 (age 1107.776s)
>  hex dump (first 32 bytes):
>    01 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>    00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00  ................
>  backtrace:
>    [] create_object+0xfc/0x278
>    [] kmemleak_alloc+0x34/0x70
>    [] kmem_cache_alloc+0x16c/0x1d8
>    [] alloc_pid+0x34/0x4d0
>    [] copy_process.isra.6+0x79c/0x1338
>    [] _do_fork+0x74/0x320
>    [] SyS_clone+0x18/0x20
>    [] el0_svc_naked+0x24/0x28
>    [] 0xffffffffffffffff

On x86 kvm_vcpu_uninit() is called on the path from kvm_arch_destroy_vm(),
on arm no equivalent call is made. Add the call to kvm_arch_vcpu_free().

Signed-off-by: James Morse 
Fixes: 749cf76c5a36 ("KVM: ARM: Initial skeleton to compile KVM support")
Cc:  # 3.10+
Acked-by: Marc Zyngier 
Signed-off-by: Christoffer Dall 
Signed-off-by: Sasha Levin

arm/arm64: KVM: Enforce Break-Before-Make on Stage-2 page tables

2016-06-03T23:15:37+00:00

[ Upstream commit d4b9e0790aa764c0b01e18d4e8d33e93ba36d51f ]

The ARM architecture mandates that when changing a page table entry
from a valid entry to another valid entry, an invalid entry is first
written, TLB invalidated, and only then the new entry being written.

The current code doesn't respect this, directly writing the new
entry and only then invalidating TLBs. Let's fix it up.

Cc: 
Reported-by: Christoffer Dall 
Signed-off-by: Marc Zyngier 
Signed-off-by: Christoffer Dall 
Signed-off-by: Sasha Levin

arm/arm64: KVM: Fix ioctl error handling

2016-03-09T18:15:17+00:00

[ Upstream commit 4cad67fca3fc952d6f2ed9e799621f07666a560f ]

Calling return copy_to_user(...) in an ioctl will not
do the right thing if there's a pagefault:
copy_to_user returns the number of bytes not copied
in this case.

Fix up kvm to do
	return copy_to_user(...)) ?  -EFAULT : 0;

everywhere.

Cc: stable@vger.kernel.org
Acked-by: Christoffer Dall 
Signed-off-by: Michael S. Tsirkin 
Signed-off-by: Marc Zyngier 
Signed-off-by: Sasha Levin

ARM/arm64: KVM: correct PTE uncachedness check

2016-01-31T19:23:39+00:00

commit 0de58f852875a0f0dcfb120bb8433e4e73c7803b upstream.

Commit e6fab5442345 ("ARM/arm64: KVM: test properly for a PTE's
uncachedness") modified the logic to test whether a HYP or stage-2
mapping needs flushing, from [incorrectly] interpreting the page table
attributes to [incorrectly] checking whether the PFN that backs the
mapping is covered by host system RAM. The PFN number is part of the
output of the translation, not the input, so we have to use pte_pfn()
on the contents of the PTE, not __phys_to_pfn() on the HYP virtual
address or stage-2 intermediate physical address.

Fixes: e6fab5442345 ("ARM/arm64: KVM: test properly for a PTE's uncachedness")
Tested-by: Pavel Fedin 
Signed-off-by: Ard Biesheuvel 
Signed-off-by: Marc Zyngier 
Signed-off-by: Greg Kroah-Hartman

ARM/arm64: KVM: test properly for a PTE's uncachedness

2016-01-31T19:23:39+00:00

commit e6fab54423450d699a09ec2b899473a541f61971 upstream.

The open coded tests for checking whether a PTE maps a page as
uncached use a flawed '(pte_val(xxx) & CONST) != CONST' pattern,
which is not guaranteed to work since the type of a mapping is
not a set of mutually exclusive bits

For HYP mappings, the type is an index into the MAIR table (i.e, the
index itself does not contain any information whatsoever about the
type of the mapping), and for stage-2 mappings it is a bit field where
normal memory and device types are defined as follows:

    #define MT_S2_NORMAL            0xf
    #define MT_S2_DEVICE_nGnRE      0x1

I.e., masking *and* comparing with the latter matches on the former,
and we have been getting lucky merely because the S2 device mappings
also have the PTE_UXN bit set, or we would misidentify memory mappings
as device mappings.

Since the unmap_range() code path (which contains one instance of the
flawed test) is used both for HYP mappings and stage-2 mappings, and
considering the difference between the two, it is non-trivial to fix
this by rewriting the tests in place, as it would involve passing
down the type of mapping through all the functions.

However, since HYP mappings and stage-2 mappings both deal with host
physical addresses, we can simply check whether the mapping is backed
by memory that is managed by the host kernel, and only perform the
D-cache maintenance if this is the case.

Signed-off-by: Ard Biesheuvel 
Tested-by: Pavel Fedin 
Reviewed-by: Christoffer Dall 
Signed-off-by: Christoffer Dall 
Signed-off-by: Greg Kroah-Hartman

KVM: arm: use GIC support unconditionally

2015-11-09T22:33:34+00:00

commit 4a5d69b73948d0e03cd38d77dc11edb2e707165f upstream.

The vgic code on ARM is built for all configurations that enable KVM,
but the parent_data field that it references is only present when
CONFIG_IRQ_DOMAIN_HIERARCHY is set:

virt/kvm/arm/vgic.c: In function 'kvm_vgic_map_phys_irq':
virt/kvm/arm/vgic.c:1781:13: error: 'struct irq_data' has no member named 'parent_data'

This flag is implied by the GIC driver, and indeed the VGIC code only
makes sense if a GIC is present. This changes the CONFIG_KVM symbol
to always select GIC, which avoids the issue.

Fixes: 662d9715840 ("arm/arm64: KVM: Kill CONFIG_KVM_ARM_{VGIC,TIMER}")
Signed-off-by: Arnd Bergmann 
Acked-by: Marc Zyngier 
Signed-off-by: Christoffer Dall 
Signed-off-by: Greg Kroah-Hartman

arm: KVM: Disable virtual timer even if the guest is not using it

2015-10-22T21:43:11+00:00

commit 688bc577ac42ae3d07c889a1f0a72f0b23763d58 upstream.

When running a guest with the architected timer disabled (with QEMU and
the kernel_irqchip=off option, for example), it is important to make
sure the timer gets turned off. Otherwise, the guest may try to
enable it anyway, leading to a screaming HW interrupt.

The fix is to unconditionally turn off the virtual timer on guest
exit.

Signed-off-by: Marc Zyngier 
Signed-off-by: Greg Kroah-Hartman

arm: KVM: Fix incorrect device to IPA mapping

2015-10-22T21:43:10+00:00

commit ca09f02f122b2ecb0f5ddfc5fd47b29ed657d4fd upstream.

A critical bug has been found in device memory stage1 translation for
VMs with more then 4GB of address space. Once vm_pgoff size is smaller
then pa (which is true for LPAE case, u32 and u64 respectively) some
more significant bits of pa may be lost as a shift operation is performed
on u32 and later cast onto u64.

Example: vm_pgoff(u32)=0x00210030, PAGE_SHIFT=12
        expected pa(u64):   0x0000002010030000
        produced pa(u64):   0x0000000010030000

The fix is to change the order of operations (casting first onto phys_addr_t
and then shifting).

Reviewed-by: Marc Zyngier 
[maz: fixed changelog and patch formatting]
Signed-off-by: Marek Majtyka 
Signed-off-by: Marc Zyngier 
Signed-off-by: Greg Kroah-Hartman

arm/arm64: KVM: vgic: Check for !irqchip_in_kernel() when mapping resources

2015-09-29T17:26:00+00:00

commit c2f58514cfb374d5368c9da945f1765cd48eb0da upstream.

Until b26e5fdac43c ("arm/arm64: KVM: introduce per-VM ops"),
kvm_vgic_map_resources() used to include a check on irqchip_in_kernel(),
and vgic_v2_map_resources() still has it.

But now vm_ops are not initialized until we call kvm_vgic_create().
Therefore kvm_vgic_map_resources() can being called without a VGIC,
and we die because vm_ops.map_resources is NULL.

Fixing this restores QEMU's kernel-irqchip=off option to a working state,
allowing to use GIC emulation in userspace.

Fixes: b26e5fdac43c ("arm/arm64: KVM: introduce per-VM ops")
Signed-off-by: Pavel Fedin 
[maz: reworked commit message]
Signed-off-by: Marc Zyngier 
Signed-off-by: Greg Kroah-Hartman

ARM: kvm: psci: fix handling of unimplemented functions

2015-07-10T16:49:34+00:00

commit e2d997366dc5b6c9d14035867f73957f93e7578c upstream.

According to the PSCI specification and the SMC/HVC calling
convention, PSCI function_ids that are not implemented must
return NOT_SUPPORTED as return value.

Current KVM implementation takes an unhandled PSCI function_id
as an error and injects an undefined instruction into the guest
if PSCI implementation is called with a function_id that is not
handled by the resident PSCI version (ie it is not implemented),
which is not the behaviour expected by a guest when calling a
PSCI function_id that is not implemented.

This patch fixes this issue by returning NOT_SUPPORTED whenever
the kvm PSCI call is executed for a function_id that is not
implemented by the PSCI kvm layer.

Cc: Christoffer Dall 
Acked-by: Sudeep Holla 
Signed-off-by: Lorenzo Pieralisi 
Signed-off-by: Marc Zyngier 
Signed-off-by: Greg Kroah-Hartman