platform/kernel/linux-starfive.git
4 years agoKVM: SVM: document KVM_MEM_ENCRYPT_OP, let userspace detect if SEV is available
Paolo Bonzini [Fri, 20 Mar 2020 17:34:50 +0000 (13:34 -0400)]
KVM: SVM: document KVM_MEM_ENCRYPT_OP, let userspace detect if SEV is available

Userspace has no way to query if SEV has been disabled with the
sev module parameter of kvm-amd.ko.  Actually it has one, but it
is a hack: do ioctl(KVM_MEM_ENCRYPT_OP, NULL) and check if it
returns EFAULT.  Make it a little nicer by returning zero for
SEV enabled and NULL argument, and while at it document the
ioctl arguments.

Cc: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: remove bogus user-triggerable WARN_ON
Paolo Bonzini [Thu, 19 Mar 2020 17:41:06 +0000 (13:41 -0400)]
KVM: x86: remove bogus user-triggerable WARN_ON

The WARN_ON is essentially comparing a user-provided value with 0.  It is
trivial to trigger it just by passing garbage to KVM_SET_CLOCK.  Guests
can break if you do so, but the same applies to every KVM_SET_* ioctl.
So, if it hurts when you do like this, just do not do it.

Reported-by: syzbot+00be5da1d75f1cc95f6b@syzkaller.appspotmail.com
Fixes: 9446e6fce0ab ("KVM: x86: fix WARN_ON check of an unsigned less than zero")
Cc: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoMerge branch 'kvm-null-pointer-fix' into kvm-master
Paolo Bonzini [Sat, 14 Mar 2020 11:49:37 +0000 (12:49 +0100)]
Merge branch 'kvm-null-pointer-fix' into kvm-master

4 years agoKVM: nVMX: avoid NULL pointer dereference with incorrect EVMCS GPAs
Vitaly Kuznetsov [Mon, 9 Mar 2020 15:52:11 +0000 (16:52 +0100)]
KVM: nVMX: avoid NULL pointer dereference with incorrect EVMCS GPAs

When an EVMCS enabled L1 guest on KVM will tries doing enlightened VMEnter
with EVMCS GPA = 0 the host crashes because the

evmcs_gpa != vmx->nested.hv_evmcs_vmptr

condition in nested_vmx_handle_enlightened_vmptrld() will evaluate to
false (as nested.hv_evmcs_vmptr is zeroed after init). The crash will
happen on vmx->nested.hv_evmcs pointer dereference.

Another problematic EVMCS ptr value is '-1' but it only causes host crash
after nested_release_evmcs() invocation. The problem is exactly the same as
with '0', we mistakenly think that the EVMCS pointer hasn't changed and
thus nested.hv_evmcs_vmptr is valid.

Resolve the issue by adding an additional !vmx->nested.hv_evmcs
check to nested_vmx_handle_enlightened_vmptrld(), this way we will
always be trying kvm_vcpu_map() when nested.hv_evmcs is NULL
and this is supposed to catch all invalid EVMCS GPAs.

Also, initialize hv_evmcs_vmptr to '0' in nested_release_evmcs()
to be consistent with initialization where we don't currently
set hv_evmcs_vmptr to '-1'.

Cc: stable@vger.kernel.org
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoMerge tag 'kvm-s390-master-5.6-1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Paolo Bonzini [Sat, 14 Mar 2020 10:59:08 +0000 (11:59 +0100)]
Merge tag 'kvm-s390-master-5.6-1' of git://git./linux/kernel/git/kvms390/linux into kvm-master

KVM: s390: Fully do the CPU resets as intended

With 7de3f1423ff9 ("KVM: s390: Add new reset vcpu API") we clarified
the meaning of the reset ioctl to fully reset the CPU and not only the
parts that can not be handled by userspace. Turns out that we missed
some parts.

4 years agoKVM: x86: Initializing all kvm_lapic_irq fields in ioapic_write_indirect
Nitesh Narayan Lal [Fri, 13 Mar 2020 13:16:24 +0000 (09:16 -0400)]
KVM: x86: Initializing all kvm_lapic_irq fields in ioapic_write_indirect

Previously all fields of structure kvm_lapic_irq were not initialized
before it was passed to kvm_bitmap_or_dest_vcpus(). Which will cause
an issue when any of those fields are used for processing a request.
For example not initializing the msi_redir_hint field before passing
to the kvm_bitmap_or_dest_vcpus(), may lead to a misbehavior of
kvm_apic_map_get_dest_lapic(). This will specifically happen when the
kvm_lowest_prio_delivery() returns TRUE due to a non-zero garbage
value of msi_redir_hint, which should not happen as the request belongs
to APIC fixed delivery mode and we do not want to deliver the
interrupt only to the lowest priority candidate.

This patch initializes all the fields of kvm_lapic_irq based on the
values of ioapic redirect_entry object before passing it on to
kvm_bitmap_or_dest_vcpus().

Fixes: 7ee30bc132c6 ("KVM: x86: deliver KVM IOAPIC scan request to target vCPUs")
Signed-off-by: Nitesh Narayan Lal <nitesh@redhat.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
[Set level to false since the value doesn't really matter. Suggested
 by Vitaly Kuznetsov. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: VMX: Condition ENCLS-exiting enabling on CPU support for SGX1
Sean Christopherson [Thu, 12 Mar 2020 18:04:16 +0000 (11:04 -0700)]
KVM: VMX: Condition ENCLS-exiting enabling on CPU support for SGX1

Enable ENCLS-exiting (and thus set vmcs.ENCLS_EXITING_BITMAP) only if
the CPU supports SGX1.  Per Intel's SDM, all ENCLS leafs #UD if SGX1
is not supported[*], i.e. intercepting ENCLS to inject a #UD is
unnecessary.

Avoiding ENCLS-exiting even when it is reported as supported by the CPU
works around a reported issue where SGX is "hard" disabled after an S3
suspend/resume cycle, i.e. CPUID.0x7.SGX=0 and the VMCS field/control
are enumerated as unsupported.  While the root cause of the S3 issue is
unknown, it's definitely _not_ a KVM (or kernel) bug, i.e. this is a
workaround for what is most likely a hardware or firmware issue.  As a
bonus side effect, KVM saves a VMWRITE when first preparing vmcs01 and
vmcs02.

Note, SGX must be disabled in BIOS to take advantage of this workaround

[*] The additional ENCLS CPUID check on SGX1 exists so that SGX can be
    globally "soft" disabled post-reset, e.g. if #MC bits in MCi_CTL are
    cleared.  Soft disabled meaning disabling SGX without clearing the
    primary CPUID bit (in leaf 0x7) and without poking into non-SGX
    CPU paths, e.g. for the VMCS controls.

Fixes: 0b665d304028 ("KVM: vmx: Inject #UD for SGX ENCLS instruction in guest")
Reported-by: Toni Spets <toni.spets@iki.fi>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: s390: Also reset registers in sync regs for initial cpu reset
Christian Borntraeger [Tue, 3 Mar 2020 08:10:57 +0000 (03:10 -0500)]
KVM: s390: Also reset registers in sync regs for initial cpu reset

When we do the initial CPU reset we must not only clear the registers
in the internal data structures but also in kvm_run sync_regs. For
modern userspace sync_regs is the only place that it looks at.

Fixes: 7de3f1423ff9 ("KVM: s390: Add new reset vcpu API")
Acked-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
4 years agoKVM: fix Kconfig menu text for -Werror
Jason A. Donenfeld [Wed, 4 Mar 2020 10:42:21 +0000 (18:42 +0800)]
KVM: fix Kconfig menu text for -Werror

This was evidently copy and pasted from the i915 driver, but the text
wasn't updated.

Fixes: 4f337faf1c55 ("KVM: allow disabling -Werror")
Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: remove stale comment from struct x86_emulate_ctxt
Vitaly Kuznetsov [Tue, 3 Mar 2020 14:33:16 +0000 (15:33 +0100)]
KVM: x86: remove stale comment from struct x86_emulate_ctxt

Commit c44b4c6ab80e ("KVM: emulate: clean up initializations in
init_decode_cache") did some field shuffling and instead of
[opcode_len, _regs) started clearing [has_seg_override, modrm).
The comment about clearing fields altogether is not true anymore.

Fixes: c44b4c6ab80e ("KVM: emulate: clean up initializations in init_decode_cache")
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: clear stale x86_emulate_ctxt->intercept value
Vitaly Kuznetsov [Tue, 3 Mar 2020 14:33:15 +0000 (15:33 +0100)]
KVM: x86: clear stale x86_emulate_ctxt->intercept value

After commit 07721feee46b ("KVM: nVMX: Don't emulate instructions in guest
mode") Hyper-V guests on KVM stopped booting with:

 kvm_nested_vmexit:    rip fffff802987d6169 reason EPT_VIOLATION info1 181
    info2 0 int_info 0 int_info_err 0
 kvm_page_fault:       address febd0000 error_code 181
 kvm_emulate_insn:     0:fffff802987d6169: f3 a5
 kvm_emulate_insn:     0:fffff802987d6169: f3 a5 FAIL
 kvm_inj_exception:    #UD (0x0)

"f3 a5" is a "rep movsw" instruction, which should not be intercepted
at all.  Commit c44b4c6ab80e ("KVM: emulate: clean up initializations in
init_decode_cache") reduced the number of fields cleared by
init_decode_cache() claiming that they are being cleared elsewhere,
'intercept', however, is left uncleared if the instruction does not have
any of the "slow path" flags (NotImpl, Stack, Op3264, Sse, Mmx, CheckPerm,
NearBranch, No16 and of course Intercept itself).

Fixes: c44b4c6ab80e ("KVM: emulate: clean up initializations in init_decode_cache")
Fixes: 07721feee46b ("KVM: nVMX: Don't emulate instructions in guest mode")
Cc: stable@vger.kernel.org
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: SVM: Fix the svm vmexit code for WRMSR
Haiwei Li [Mon, 2 Mar 2020 12:19:28 +0000 (20:19 +0800)]
KVM: SVM: Fix the svm vmexit code for WRMSR

In svm, exit_code for MSR writes is not EXIT_REASON_MSR_WRITE which
belongs to vmx.

According to amd manual, SVM_EXIT_MSR(7ch) is the exit_code of VMEXIT_MSR
due to RDMSR or WRMSR access to protected MSR. Additionally, the processor
indicates in the VMCB's EXITINFO1 whether a RDMSR(EXITINFO1=0) or
WRMSR(EXITINFO1=1) was intercepted.

Signed-off-by: Haiwei Li <lihaiwei@tencent.com>
Fixes: 1e9e2622a149 ("KVM: VMX: FIXED+PHYSICAL mode single target IPI fastpath", 2019-11-21)
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: X86: Fix dereference null cpufreq policy
Wanpeng Li [Mon, 2 Mar 2020 07:15:36 +0000 (15:15 +0800)]
KVM: X86: Fix dereference null cpufreq policy

Naresh Kamboju reported:

   Linux version 5.6.0-rc4 (oe-user@oe-host) (gcc version
  (GCC)) #1 SMP Sun Mar 1 22:59:08 UTC 2020
   kvm: no hardware support
   BUG: kernel NULL pointer dereference, address: 000000000000028c
   #PF: supervisor read access in kernel mode
   #PF: error_code(0x0000) - not-present page
   PGD 0 P4D 0
   Oops: 0000 [#1] SMP NOPTI
   CPU: 0 PID: 1 Comm: swapper/0 Not tainted 5.6.0-rc4 #1
   Hardware name: QEMU Standard PC (i440FX + PIIX, 1996),
  04/01/2014
   RIP: 0010:kobject_put+0x12/0x1c0
   Call Trace:
    cpufreq_cpu_put+0x15/0x20
    kvm_arch_init+0x1f6/0x2b0
    kvm_init+0x31/0x290
    ? svm_check_processor_compat+0xd/0xd
    ? svm_check_processor_compat+0xd/0xd
    svm_init+0x21/0x23
    do_one_initcall+0x61/0x2f0
    ? rdinit_setup+0x30/0x30
    ? rcu_read_lock_sched_held+0x4f/0x80
    kernel_init_freeable+0x219/0x279
    ? rest_init+0x250/0x250
    kernel_init+0xe/0x110
    ret_from_fork+0x27/0x50
   Modules linked in:
   CR2: 000000000000028c
   ---[ end trace 239abf40c55c409b ]---
   RIP: 0010:kobject_put+0x12/0x1c0

cpufreq policy which is get by cpufreq_cpu_get() can be NULL if it is failure,
this patch takes care of it.

Fixes: aaec7c03de (KVM: x86: avoid useless copy of cpufreq policy)
Reported-by: Naresh Kamboju <naresh.kamboju@linaro.org>
Cc: Naresh Kamboju <naresh.kamboju@linaro.org>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: VMX: check descriptor table exits on instruction emulation
Oliver Upton [Sat, 29 Feb 2020 19:30:14 +0000 (11:30 -0800)]
KVM: VMX: check descriptor table exits on instruction emulation

KVM emulates UMIP on hardware that doesn't support it by setting the
'descriptor table exiting' VM-execution control and performing
instruction emulation. When running nested, this emulation is broken as
KVM refuses to emulate L2 instructions by default.

Correct this regression by allowing the emulation of descriptor table
instructions if L1 hasn't requested 'descriptor table exiting'.

Fixes: 07721feee46b ("KVM: nVMX: Don't emulate instructions in guest mode")
Reported-by: Jan Kiszka <jan.kiszka@web.de>
Cc: stable@vger.kernel.org
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Jim Mattson <jmattson@google.com>
Signed-off-by: Oliver Upton <oupton@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoMerge tag 'kvmarm-fixes-5.6-1' of git://git.kernel.org/pub/scm/linux/kernel/git/kvmar...
Paolo Bonzini [Fri, 28 Feb 2020 10:46:59 +0000 (11:46 +0100)]
Merge tag 'kvmarm-fixes-5.6-1' of git://git./linux/kernel/git/kvmarm/kvmarm into HEAD

KVM/arm fixes for 5.6, take #1

- Fix compilation on 32bit
- Move  VHE guest entry/exit into the VHE-specific entry code
- Make sure all functions called by the non-VHE HYP code is tagged as __always_inline

4 years agokvm: x86: Limit the number of "kvm: disabled by bios" messages
Erwan Velu [Thu, 27 Feb 2020 18:00:46 +0000 (19:00 +0100)]
kvm: x86: Limit the number of "kvm: disabled by bios" messages

In older version of systemd(219), at boot time, udevadm is called with :
/usr/bin/udevadm trigger --type=devices --action=add"

This program generates an echo "add" in /sys/devices/system/cpu/cpu<x>/uevent,
leading to the "kvm: disabled by bios" message in case of your Bios disabled
the virtualization extensions.

On a modern system running up to 256 CPU threads, this pollutes the Kernel logs.

This patch offers to ratelimit this message to avoid any userspace program triggering
this uevent printing this message too often.

This patch is only a workaround but greatly reduce the pollution without
breaking the current behavior of printing a message if some try to instantiate
KVM on a system that doesn't support it.

Note that recent versions of systemd (>239) do not have trigger this behavior.

This patch will be useful at least for some using older systemd with recent Kernels.

Signed-off-by: Erwan Velu <e.velu@criteo.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: avoid useless copy of cpufreq policy
Paolo Bonzini [Fri, 28 Feb 2020 09:49:10 +0000 (10:49 +0100)]
KVM: x86: avoid useless copy of cpufreq policy

struct cpufreq_policy is quite big and it is not a good idea
to allocate one on the stack.  Just use cpufreq_cpu_get and
cpufreq_cpu_put which is even simpler.

Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: allow disabling -Werror
Paolo Bonzini [Fri, 28 Feb 2020 09:42:31 +0000 (10:42 +0100)]
KVM: allow disabling -Werror

Restrict -Werror to well-tested configurations and allow disabling it
via Kconfig.

Reported-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: allow compiling as non-module with W=1
Valdis Klētnieks [Fri, 28 Feb 2020 02:49:52 +0000 (21:49 -0500)]
KVM: x86: allow compiling as non-module with W=1

Compile error with CONFIG_KVM_INTEL=y and W=1:

  CC      arch/x86/kvm/vmx/vmx.o
arch/x86/kvm/vmx/vmx.c:68:32: error: 'vmx_cpu_id' defined but not used [-Werror=unused-const-variable=]
   68 | static const struct x86_cpu_id vmx_cpu_id[] = {
      |                                ^~~~~~~~~~
cc1: all warnings being treated as errors

When building with =y, the MODULE_DEVICE_TABLE macro doesn't generate a
reference to the structure (or any code at all).  This makes W=1 compiles
unhappy.

Wrap both in a #ifdef to avoid the issue.

Signed-off-by: Valdis Kletnieks <valdis.kletnieks@vt.edu>
[Do the same for CONFIG_KVM_AMD. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: Pre-allocate 1 cpumask variable per cpu for both pv tlb and pv ipis
Wanpeng Li [Tue, 18 Feb 2020 01:08:24 +0000 (09:08 +0800)]
KVM: Pre-allocate 1 cpumask variable per cpu for both pv tlb and pv ipis

Nick Desaulniers Reported:

  When building with:
  $ make CC=clang arch/x86/ CFLAGS=-Wframe-larger-than=1000
  The following warning is observed:
  arch/x86/kernel/kvm.c:494:13: warning: stack frame size of 1064 bytes in
  function 'kvm_send_ipi_mask_allbutself' [-Wframe-larger-than=]
  static void kvm_send_ipi_mask_allbutself(const struct cpumask *mask, int
  vector)
              ^
  Debugging with:
  https://github.com/ClangBuiltLinux/frame-larger-than
  via:
  $ python3 frame_larger_than.py arch/x86/kernel/kvm.o \
    kvm_send_ipi_mask_allbutself
  points to the stack allocated `struct cpumask newmask` in
  `kvm_send_ipi_mask_allbutself`. The size of a `struct cpumask` is
  potentially large, as it's CONFIG_NR_CPUS divided by BITS_PER_LONG for
  the target architecture. CONFIG_NR_CPUS for X86_64 can be as high as
  8192, making a single instance of a `struct cpumask` 1024 B.

This patch fixes it by pre-allocate 1 cpumask variable per cpu and use it for
both pv tlb and pv ipis..

Reported-by: Nick Desaulniers <ndesaulniers@google.com>
Acked-by: Nick Desaulniers <ndesaulniers@google.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Nick Desaulniers <ndesaulniers@google.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: Introduce pv check helpers
Wanpeng Li [Tue, 18 Feb 2020 01:08:23 +0000 (09:08 +0800)]
KVM: Introduce pv check helpers

Introduce some pv check helpers for consistency.

Suggested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: let declaration of kvm_get_running_vcpus match implementation
Christian Borntraeger [Fri, 28 Feb 2020 08:49:41 +0000 (09:49 +0100)]
KVM: let declaration of kvm_get_running_vcpus match implementation

Sparse notices that declaration and implementation do not match:
arch/s390/kvm/../../../virt/kvm/kvm_main.c:4435:17: warning: incorrect type in return expression (different address spaces)
arch/s390/kvm/../../../virt/kvm/kvm_main.c:4435:17:    expected struct kvm_vcpu [noderef] <asn:3> **
arch/s390/kvm/../../../virt/kvm/kvm_main.c:4435:17:    got struct kvm_vcpu *[noderef] <asn:3> *

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: SVM: allocate AVIC data structures based on kvm_amd module parameter
Paolo Bonzini [Tue, 25 Feb 2020 07:54:26 +0000 (08:54 +0100)]
KVM: SVM: allocate AVIC data structures based on kvm_amd module parameter

Even if APICv is disabled at startup, the backing page and ir_list need
to be initialized in case they are needed later.  The only case in
which this can be skipped is for userspace irqchip, and that must be
done because avic_init_backing_page dereferences vcpu->arch.apic
(which is NULL for userspace irqchip).

Tested-by: rmuncrief@humanavance.com
Fixes: https://bugzilla.kernel.org/show_bug.cgi?id=206579
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: s390: rstify new ioctls in api.rst
Christian Borntraeger [Mon, 24 Feb 2020 10:15:59 +0000 (11:15 +0100)]
KVM: s390: rstify new ioctls in api.rst

We also need to rstify the new ioctls that we added in parallel to the
rstification of the kvm docs.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Check IO instruction VM-exit conditions
Oliver Upton [Tue, 4 Feb 2020 23:26:31 +0000 (15:26 -0800)]
KVM: nVMX: Check IO instruction VM-exit conditions

Consult the 'unconditional IO exiting' and 'use IO bitmaps' VM-execution
controls when checking instruction interception. If the 'use IO bitmaps'
VM-execution control is 1, check the instruction access against the IO
bitmaps to determine if the instruction causes a VM-exit.

Signed-off-by: Oliver Upton <oupton@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Refactor IO bitmap checks into helper function
Oliver Upton [Tue, 4 Feb 2020 23:26:30 +0000 (15:26 -0800)]
KVM: nVMX: Refactor IO bitmap checks into helper function

Checks against the IO bitmap are useful for both instruction emulation
and VM-exit reflection. Refactor the IO bitmap checks into a helper
function.

Signed-off-by: Oliver Upton <oupton@google.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Don't emulate instructions in guest mode
Paolo Bonzini [Tue, 4 Feb 2020 23:26:29 +0000 (15:26 -0800)]
KVM: nVMX: Don't emulate instructions in guest mode

vmx_check_intercept is not yet fully implemented. To avoid emulating
instructions disallowed by the L1 hypervisor, refuse to emulate
instructions by default.

Cc: stable@vger.kernel.org
[Made commit, added commit msg - Oliver]
Signed-off-by: Oliver Upton <oupton@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Emulate MTF when performing instruction emulation
Oliver Upton [Fri, 7 Feb 2020 10:36:07 +0000 (02:36 -0800)]
KVM: nVMX: Emulate MTF when performing instruction emulation

Since commit 5f3d45e7f282 ("kvm/x86: add support for
MONITOR_TRAP_FLAG"), KVM has allowed an L1 guest to use the monitor trap
flag processor-based execution control for its L2 guest. KVM simply
forwards any MTF VM-exits to the L1 guest, which works for normal
instruction execution.

However, when KVM needs to emulate an instruction on the behalf of an L2
guest, the monitor trap flag is not emulated. Add the necessary logic to
kvm_skip_emulated_instruction() to synthesize an MTF VM-exit to L1 upon
instruction emulation for L2.

Fixes: 5f3d45e7f282 ("kvm/x86: add support for MONITOR_TRAP_FLAG")
Signed-off-by: Oliver Upton <oupton@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: fix error handling in svm_hardware_setup
Li RongQing [Sun, 23 Feb 2020 08:13:12 +0000 (16:13 +0800)]
KVM: fix error handling in svm_hardware_setup

rename svm_hardware_unsetup as svm_hardware_teardown, move
it before svm_hardware_setup, and call it to free all memory
if fail to setup in svm_hardware_setup, otherwise memory will
be leaked

remove __exit attribute for it since it is called in __init
function

Signed-off-by: Li RongQing <lirongqing@baidu.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoarm64: Ask the compiler to __always_inline functions used by KVM at HYP
James Morse [Thu, 20 Feb 2020 16:58:39 +0000 (16:58 +0000)]
arm64: Ask the compiler to __always_inline functions used by KVM at HYP

KVM uses some of the static-inline helpers like icache_is_vipt() from
its HYP code. This assumes the function is inlined so that the code is
mapped to EL2. The compiler may decide not to inline these, and the
out-of-line version may not be in the __hyp_text section.

Add the additional __always_ hint to these static-inlines that are used
by KVM.

Signed-off-by: James Morse <james.morse@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Acked-by: Will Deacon <will@kernel.org>
Link: https://lore.kernel.org/r/20200220165839.256881-4-james.morse@arm.com
4 years agoKVM: arm64: Define our own swab32() to avoid a uapi static inline
James Morse [Thu, 20 Feb 2020 16:58:38 +0000 (16:58 +0000)]
KVM: arm64: Define our own swab32() to avoid a uapi static inline

KVM uses swab32() when mediating GIC MMIO accesses if the GICV is badly
aligned, and the host and guest differ in endianness.

arm64 doesn't provide a __arch_swab32(), so __fswab32() is always backed
by the macro implementation that the compiler reduces to a single
instruction. But the static-inline causes problems for KVM if the compiler
chooses not to inline this function, it may not be located in the
__hyp_text where __vgic_v2_perform_cpuif_access() needs it.

Create our own __kvm_swab32() macro that calls ___constant_swab32()
directly. This way we know it will always be inlined.

Signed-off-by: James Morse <james.morse@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200220165839.256881-3-james.morse@arm.com
4 years agoKVM: arm64: Ask the compiler to __always_inline functions used at HYP
James Morse [Thu, 20 Feb 2020 16:58:37 +0000 (16:58 +0000)]
KVM: arm64: Ask the compiler to __always_inline functions used at HYP

On non VHE CPUs, KVM's __hyp_text contains code run at EL2 while the rest
of the kernel runs at EL1. This code lives in its own section with start
and end markers so we can map it to EL2.

The compiler may decide not to inline static-inline functions from the
header file. It may also decide not to put these out-of-line functions
in the same section, meaning they aren't mapped when called at EL2.

Clang-9 does exactly this with __kern_hyp_va() and a few others when
x18 is reserved for the shadow call stack. Add the additional __always_
hint to all the static-inlines that are called from a hyp file.

Signed-off-by: James Morse <james.morse@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200220165839.256881-2-james.morse@arm.com
----
kvm_get_hyp_vector() pulls in all the regular per-cpu accessors
and this_cpu_has_cap(), fortunately its only called for VHE.

4 years agoKVM: SVM: Fix potential memory leak in svm_cpu_init()
Miaohe Lin [Sat, 4 Jan 2020 08:56:49 +0000 (16:56 +0800)]
KVM: SVM: Fix potential memory leak in svm_cpu_init()

When kmalloc memory for sd->sev_vmcbs failed, we forget to free the page
held by sd->save_area. Also get rid of the var r as '-ENOMEM' is actually
the only possible outcome here.

Reviewed-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: apic: avoid calculating pending eoi from an uninitialized val
Miaohe Lin [Fri, 21 Feb 2020 14:04:46 +0000 (22:04 +0800)]
KVM: apic: avoid calculating pending eoi from an uninitialized val

When pv_eoi_get_user() fails, 'val' may remain uninitialized and the return
value of pv_eoi_get_pending() becomes random. Fix the issue by initializing
the variable.

Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: clear PIN_BASED_POSTED_INTR from nested pinbased_ctls only when apicv...
Vitaly Kuznetsov [Thu, 20 Feb 2020 17:22:04 +0000 (18:22 +0100)]
KVM: nVMX: clear PIN_BASED_POSTED_INTR from nested pinbased_ctls only when apicv is globally disabled

When apicv is disabled on a vCPU (e.g. by enabling KVM_CAP_HYPERV_SYNIC*),
nothing happens to VMX MSRs on the already existing vCPUs, however, all new
ones are created with PIN_BASED_POSTED_INTR filtered out. This is very
confusing and results in the following picture inside the guest:

$ rdmsr -ax 0x48d
ff00000016
7f00000016
7f00000016
7f00000016

This is observed with QEMU and 4-vCPU guest: QEMU creates vCPU0, does
KVM_CAP_HYPERV_SYNIC2 and then creates the remaining three.

L1 hypervisor may only check CPU0's controls to find out what features
are available and it will be very confused later. Switch to setting
PIN_BASED_POSTED_INTR control based on global 'enable_apicv' setting.

Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: handle nested posted interrupts when apicv is disabled for L1
Vitaly Kuznetsov [Thu, 20 Feb 2020 17:22:05 +0000 (18:22 +0100)]
KVM: nVMX: handle nested posted interrupts when apicv is disabled for L1

Even when APICv is disabled for L1 it can (and, actually, is) still
available for L2, this means we need to always call
vmx_deliver_nested_posted_interrupt() when attempting an interrupt
delivery.

Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: x86: svm: Fix NULL pointer dereference when AVIC not enabled
Suravee Suthikulpanit [Fri, 21 Feb 2020 14:52:17 +0000 (08:52 -0600)]
kvm: x86: svm: Fix NULL pointer dereference when AVIC not enabled

Launching VM w/ AVIC disabled together with pass-through device
results in NULL pointer dereference bug with the following call trace.

    RIP: 0010:svm_refresh_apicv_exec_ctrl+0x17e/0x1a0 [kvm_amd]

    Call Trace:
     kvm_vcpu_update_apicv+0x44/0x60 [kvm]
     kvm_arch_vcpu_ioctl_run+0x3f4/0x1c80 [kvm]
     kvm_vcpu_ioctl+0x3d8/0x650 [kvm]
     do_vfs_ioctl+0xaa/0x660
     ? tomoyo_file_ioctl+0x19/0x20
     ksys_ioctl+0x67/0x90
     __x64_sys_ioctl+0x1a/0x20
     do_syscall_64+0x57/0x190
     entry_SYSCALL_64_after_hwframe+0x44/0xa9

Investigation shows that this is due to the uninitialized usage of
struct vapu_svm.ir_list in the svm_set_pi_irte_mode(), which is
called from svm_refresh_apicv_exec_ctrl().

The ir_list is initialized only if AVIC is enabled. So, fixes by
adding a check if AVIC is enabled in the svm_refresh_apicv_exec_ctrl().

Bugzilla: https://bugzilla.kernel.org/show_bug.cgi?id=206579
Fixes: 8937d762396d ("kvm: x86: svm: Add support to (de)activate posted interrupts.")
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Tested-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: VMX: Add VMX_FEATURE_USR_WAIT_PAUSE
Xiaoyao Li [Sun, 16 Feb 2020 10:48:57 +0000 (18:48 +0800)]
KVM: VMX: Add VMX_FEATURE_USR_WAIT_PAUSE

Commit 159348784ff0 ("x86/vmx: Introduce VMX_FEATURES_*") missed
bit 26 (enable user wait and pause) of Secondary Processor-based
VM-Execution Controls.

Add VMX_FEATURE_USR_WAIT_PAUSE flag so that it shows up in /proc/cpuinfo,
and use it to define SECONDARY_EXEC_ENABLE_USR_WAIT_PAUSE to make them
uniform.

Signed-off-by: Xiaoyao Li <xiaoyao.li@intel.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Hold KVM's srcu lock when syncing vmcs12->shadow
wanpeng li [Mon, 17 Feb 2020 10:37:43 +0000 (18:37 +0800)]
KVM: nVMX: Hold KVM's srcu lock when syncing vmcs12->shadow

For the duration of mapping eVMCS, it derefences ->memslots without holding
->srcu or ->slots_lock when accessing hv assist page. This patch fixes it by
moving nested_sync_vmcs12_to_shadow to prepare_guest_switch, where the SRCU
is already taken.

It can be reproduced by running kvm's evmcs_test selftest.

  =============================
  warning: suspicious rcu usage
  5.6.0-rc1+ #53 tainted: g        w ioe
  -----------------------------
  ./include/linux/kvm_host.h:623 suspicious rcu_dereference_check() usage!

  other info that might help us debug this:

   rcu_scheduler_active = 2, debug_locks = 1
  1 lock held by evmcs_test/8507:
   #0: ffff9ddd156d00d0 (&vcpu->mutex){+.+.}, at:
kvm_vcpu_ioctl+0x85/0x680 [kvm]

  stack backtrace:
  cpu: 6 pid: 8507 comm: evmcs_test tainted: g        w ioe     5.6.0-rc1+ #53
  hardware name: dell inc. optiplex 7040/0jctf8, bios 1.4.9 09/12/2016
  call trace:
   dump_stack+0x68/0x9b
   kvm_read_guest_cached+0x11d/0x150 [kvm]
   kvm_hv_get_assist_page+0x33/0x40 [kvm]
   nested_enlightened_vmentry+0x2c/0x60 [kvm_intel]
   nested_vmx_handle_enlightened_vmptrld.part.52+0x32/0x1c0 [kvm_intel]
   nested_sync_vmcs12_to_shadow+0x439/0x680 [kvm_intel]
   vmx_vcpu_run+0x67a/0xe60 [kvm_intel]
   vcpu_enter_guest+0x35e/0x1bc0 [kvm]
   kvm_arch_vcpu_ioctl_run+0x40b/0x670 [kvm]
   kvm_vcpu_ioctl+0x370/0x680 [kvm]
   ksys_ioctl+0x235/0x850
   __x64_sys_ioctl+0x16/0x20
   do_syscall_64+0x77/0x780
   entry_syscall_64_after_hwframe+0x49/0xbe

Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: don't notify userspace IOAPIC on edge-triggered interrupt EOI
Miaohe Lin [Fri, 14 Feb 2020 02:32:38 +0000 (10:32 +0800)]
KVM: x86: don't notify userspace IOAPIC on edge-triggered interrupt EOI

Commit 13db77347db1 ("KVM: x86: don't notify userspace IOAPIC on edge
EOI") said, edge-triggered interrupts don't set a bit in TMR, which means
that IOAPIC isn't notified on EOI. And var level indicates level-triggered
interrupt.
But commit 3159d36ad799 ("KVM: x86: use generic function for MSI parsing")
replace var level with irq.level by mistake. Fix it by changing irq.level
to irq.trig_mode.

Cc: stable@vger.kernel.org
Fixes: 3159d36ad799 ("KVM: x86: use generic function for MSI parsing")
Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm/emulate: fix a -Werror=cast-function-type
Qian Cai [Mon, 17 Feb 2020 16:48:26 +0000 (11:48 -0500)]
kvm/emulate: fix a -Werror=cast-function-type

arch/x86/kvm/emulate.c: In function 'x86_emulate_insn':
arch/x86/kvm/emulate.c:5686:22: error: cast between incompatible
function types from 'int (*)(struct x86_emulate_ctxt *)' to 'void
(*)(struct fastop *)' [-Werror=cast-function-type]
    rc = fastop(ctxt, (fastop_t)ctxt->execute);

Fix it by using an unnamed union of a (*execute) function pointer and a
(*fastop) function pointer.

Fixes: 3009afc6e39e ("KVM: x86: Use a typedef for fastop functions")
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Qian Cai <cai@lca.pw>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: fix incorrect comparison in trace event
Paolo Bonzini [Thu, 13 Feb 2020 17:24:48 +0000 (18:24 +0100)]
KVM: x86: fix incorrect comparison in trace event

The "u" field in the event has three states, -1/0/1.  Using u8 however means that
comparison with -1 will always fail, so change to signed char.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agokvm: arm/arm64: Fold VHE entry/exit work into kvm_vcpu_run_vhe()
Mark Rutland [Mon, 10 Feb 2020 11:47:57 +0000 (11:47 +0000)]
kvm: arm/arm64: Fold VHE entry/exit work into kvm_vcpu_run_vhe()

With VHE, running a vCPU always requires the sequence:

1. kvm_arm_vhe_guest_enter();
2. kvm_vcpu_run_vhe();
3. kvm_arm_vhe_guest_exit()

... and as we invoke this from the shared arm/arm64 KVM code, 32-bit arm
has to provide stubs for all three functions.

To simplify the common code, and make it easier to make further
modifications to the arm64-specific portions in the near future, let's
fold kvm_arm_vhe_guest_enter() and kvm_arm_vhe_guest_exit() into
kvm_vcpu_run_vhe().

The 32-bit stubs for kvm_arm_vhe_guest_enter() and
kvm_arm_vhe_guest_exit() are removed, as they are no longer used. The
32-bit stub for kvm_vcpu_run_vhe() is left as-is.

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

Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/20200210114757.2889-1-mark.rutland@arm.com
4 years agoKVM: nVMX: Fix some obsolete comments and grammar error
Miaohe Lin [Fri, 14 Feb 2020 02:44:05 +0000 (10:44 +0800)]
KVM: nVMX: Fix some obsolete comments and grammar error

Fix wrong variable names and grammar error in comment.

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: fix missing prototypes
Paolo Bonzini [Thu, 13 Feb 2020 17:22:55 +0000 (18:22 +0100)]
KVM: x86: fix missing prototypes

Reported with "make W=1" due to -Wmissing-prototypes.

Reported-by: Qian Cai <cai@lca.pw>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: enable -Werror
Paolo Bonzini [Wed, 12 Feb 2020 19:15:29 +0000 (20:15 +0100)]
KVM: x86: enable -Werror

Avoid more embarrassing mistakes.  At least those that the compiler
can catch.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: virt: guest-halt-polling.txt convert to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:03:06 +0000 (07:03 +0100)]
docs: virt: guest-halt-polling.txt convert to ReST

Due to some merge conflict, this file ended being alone under
Documentation/virtual.

The file itself is almost at ReST format. Just minor
adjustments are needed:

- Adjust title markup;
- Adjust a list identation;
- add a literal block markup;
- Add some blank lines.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: review-checklist.txt: rename to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:03:05 +0000 (07:03 +0100)]
docs: kvm: review-checklist.txt: rename to ReST

This file is already in ReST compatible format.
So, rename it and add to the kvm's index.rst.

While here, use the standard conversion for document titles.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert timekeeping.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:03:04 +0000 (07:03 +0100)]
docs: kvm: Convert timekeeping.txt to ReST format

- Use document title and chapter markups;
- Add markups for literal blocks;
- Add markups for tables;
- use :field: for field descriptions;
- Add blank lines and adjust indentation.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert s390-diag.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:03:03 +0000 (07:03 +0100)]
docs: kvm: Convert s390-diag.txt to ReST format

This file is almost in ReST format. Just one change was
needed:

    - Add markups for a literal block and change its indentation.

While here, use the standard markup for the document title.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert ppc-pv.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:03:02 +0000 (07:03 +0100)]
docs: kvm: Convert ppc-pv.txt to ReST format

- Use document title and chapter markups;
- Add markups for tables;
- Use list markups;
- Add markups for literal blocks;
- Add blank lines.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert nested-vmx.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:03:01 +0000 (07:03 +0100)]
docs: kvm: Convert nested-vmx.txt to ReST format

This file is almost in ReST format. Just a small set of
changes were needed:

    - Add markups for lists;
    - Add markups for a literal block;
    - Adjust whitespaces.

While here, use the standard markup for the document title.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert mmu.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:03:00 +0000 (07:03 +0100)]
docs: kvm: Convert mmu.txt to ReST format

- Use document title and chapter markups;
- Add markups for tables;
- Add markups for literal blocks;
- Add blank lines and adjust indentation.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert locking.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:59 +0000 (07:02 +0100)]
docs: kvm: Convert locking.txt to ReST format

- Use document title and chapter markups;
- Add markups for literal blocks;
- use :field: for field descriptions;
- Add blank lines and adjust indentation.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert hypercalls.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:58 +0000 (07:02 +0100)]
docs: kvm: Convert hypercalls.txt to ReST format

- Use document title and chapter markups;
- Convert tables;
- Add markups for literal blocks;
- use :field: for field descriptions;
- Add blank lines and adjust indentation

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: arm/psci.txt: convert to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:57 +0000 (07:02 +0100)]
docs: kvm: arm/psci.txt: convert to ReST

- Add a title for the document;
- Adjust whitespaces for it to be properly formatted after
  parsed.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert arm/hyp-abi.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:56 +0000 (07:02 +0100)]
docs: kvm: convert arm/hyp-abi.txt to ReST

- Add proper markups for titles;
- Adjust whitespaces and blank lines to match ReST
  needs;
- Mark literal blocks as such.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: Convert api.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:55 +0000 (07:02 +0100)]
docs: kvm: Convert api.txt to ReST format

convert api.txt document to ReST format while trying to keep
its format as close as possible with the authors intent, and
avoid adding uneeded markups.

- Use document title and chapter markups;
- Convert tables;
- Add markups for literal blocks;
- use :field: for field descriptions;
- Add blank lines and adjust indentation

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/xive.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:54 +0000 (07:02 +0100)]
docs: kvm: convert devices/xive.txt to ReST

- Use title markups;
- adjust indentation and add blank lines as needed;
- adjust tables to match ReST accepted formats;
- mark code blocks as such.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/xics.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:53 +0000 (07:02 +0100)]
docs: kvm: convert devices/xics.txt to ReST

- Use title markups;
- adjust indentation and add blank lines as needed;
- adjust tables to match ReST accepted formats;
- use :field: markups.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/vm.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:52 +0000 (07:02 +0100)]
docs: kvm: convert devices/vm.txt to ReST

- Use title markups;
- adjust indentation and add blank lines as needed;
- use :field: markups;
- Use cross-references;
- mark code blocks as such.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/vfio.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:51 +0000 (07:02 +0100)]
docs: kvm: convert devices/vfio.txt to ReST

- Use standard title markup;
- adjust lists;
- mark code blocks as such.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/vcpu.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:50 +0000 (07:02 +0100)]
docs: kvm: convert devices/vcpu.txt to ReST

- Use title markups;
- adjust indentation and add blank lines as needed;
- adjust tables to match ReST accepted formats;
- use :field: markups;
- mark code blocks as such.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/s390_flic.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:49 +0000 (07:02 +0100)]
docs: kvm: convert devices/s390_flic.txt to ReST

- Use standard markup for document title;
- Adjust indentation and add blank lines as needed;
- use the notes markup;
- mark code blocks as such.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/mpic.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:48 +0000 (07:02 +0100)]
docs: kvm: convert devices/mpic.txt to ReST

This document is almost in ReST format. The only thing
needed is to mark a list as such and to add an extra
whitespace.

Yet, let's also use the standard document title markup,
as it makes easier if anyone wants later to add sessions
to it.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: convert devices/arm-vgit.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:47 +0000 (07:02 +0100)]
docs: kvm: convert devices/arm-vgit.txt to ReST

- Use title markups;
- change indent to match ReST syntax;
- use proper table markups;
- use literal block markups.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: devices/arm-vgit-v3.txt to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:46 +0000 (07:02 +0100)]
docs: kvm: devices/arm-vgit-v3.txt to ReST

- Use title markups;
- change indent to match ReST syntax;
- use proper table markups;
- use literal block markups.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: devices/arm-vgic-its.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:45 +0000 (07:02 +0100)]
docs: kvm: devices/arm-vgic-its.txt to ReST format

- Fix document title to match ReST format
- Convert the table to be properly recognized
- use proper markups for literal blocks
- Some indentation fixes to match ReST

While here, add an index for kvm devices.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: virt: Convert msr.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:44 +0000 (07:02 +0100)]
docs: virt: Convert msr.txt to ReST format

- Use document title markup;
- Convert tables;
- Add blank lines and adjust indentation.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: virt: convert halt-polling.txt to ReST format
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:43 +0000 (07:02 +0100)]
docs: virt: convert halt-polling.txt to ReST format

- Fix document title to match ReST format
- Convert the table to be properly recognized
- Some indentation fixes to match ReST syntax.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: virt: user_mode_linux.rst: fix URL references
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:42 +0000 (07:02 +0100)]
docs: virt: user_mode_linux.rst: fix URL references

Several URLs are pointing to outdated places.

Update the references for the URLs whose contents still exists,
removing the others.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: virt: user_mode_linux.rst: update compiling instructions
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:41 +0000 (07:02 +0100)]
docs: virt: user_mode_linux.rst: update compiling instructions

Instead of pointing for a pre-2.4 and a seaparate patch,
update it to match current upstream, as UML was merged
a long time ago.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: virt: convert UML documentation to ReST
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:40 +0000 (07:02 +0100)]
docs: virt: convert UML documentation to ReST

Despite being an old document, it contains lots of information
that could still be useful.

The document has a nice style with makes easy to convert to
ReST. So, let's convert it to ReST.

This patch does:

- Use proper markups for titles;
- Mark and proper indent literal blocks;
- don't use an 'o' character for lists;
- other minor changes required for the doc to be parsed.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agodocs: kvm: add arm/pvtime.rst to index.rst
Mauro Carvalho Chehab [Mon, 10 Feb 2020 06:02:39 +0000 (07:02 +0100)]
docs: kvm: add arm/pvtime.rst to index.rst

Add this file to a new kvm/arm index.rst, in order for it to
be shown as part of the virt book.

Signed-off-by: Mauro Carvalho Chehab <mchehab+huawei@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: fix WARN_ON check of an unsigned less than zero
Paolo Bonzini [Wed, 12 Feb 2020 12:27:10 +0000 (13:27 +0100)]
KVM: x86: fix WARN_ON check of an unsigned less than zero

The check cpu->hv_clock.system_time < 0 is redundant since system_time
is a u64 and hence can never be less than zero.  But what was actually
meant is to check that the result is positive, since kernel_ns and
v->kvm->arch.kvmclock_offset are both s64.

Reported-by: Colin King <colin.king@canonical.com>
Suggested-by: Sean Christopherson <sean.j.christopherson@intel.com>
Addresses-Coverity: ("Macro compares unsigned to 0")
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoselftests: KVM: Remove unused x86_register enum
Eric Auger [Fri, 7 Feb 2020 14:27:13 +0000 (15:27 +0100)]
selftests: KVM: Remove unused x86_register enum

x86_register enum is not used, let's remove it.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
Suggested-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86/mmu: Fix struct guest_walker arrays for 5-level paging
Sean Christopherson [Fri, 7 Feb 2020 17:37:42 +0000 (09:37 -0800)]
KVM: x86/mmu: Fix struct guest_walker arrays for 5-level paging

Define PT_MAX_FULL_LEVELS as PT64_ROOT_MAX_LEVEL, i.e. 5, to fix shadow
paging for 5-level guest page tables.  PT_MAX_FULL_LEVELS is used to
size the arrays that track guest pages table information, i.e. using a
"max levels" of 4 causes KVM to access garbage beyond the end of an
array when querying state for level 5 entries.  E.g. FNAME(gpte_changed)
will read garbage and most likely return %true for a level 5 entry,
soft-hanging the guest because FNAME(fetch) will restart the guest
instead of creating SPTEs because it thinks the guest PTE has changed.

Note, KVM doesn't yet support 5-level nested EPT, so PT_MAX_FULL_LEVELS
gets to stay "4" for the PTTYPE_EPT case.

Fixes: 855feb673640 ("KVM: MMU: Add 5 level EPT & Shadow page table support.")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Use correct root level for nested EPT shadow page tables
Sean Christopherson [Fri, 7 Feb 2020 17:37:41 +0000 (09:37 -0800)]
KVM: nVMX: Use correct root level for nested EPT shadow page tables

Hardcode the EPT page-walk level for L2 to be 4 levels, as KVM's MMU
currently also hardcodes the page walk level for nested EPT to be 4
levels.  The L2 guest is all but guaranteed to soft hang on its first
instruction when L1 is using EPT, as KVM will construct 4-level page
tables and then tell hardware to use 5-level page tables.

Fixes: 855feb673640 ("KVM: MMU: Add 5 level EPT & Shadow page table support.")
Cc: stable@vger.kernel.org
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Fix some comment typos and coding style
Miaohe Lin [Fri, 7 Feb 2020 15:22:45 +0000 (23:22 +0800)]
KVM: nVMX: Fix some comment typos and coding style

Fix some typos in the comments. Also fix coding style.
[Sean Christopherson rewrites the comment of write_fault_to_shadow_pgtable
field in struct kvm_vcpu_arch.]

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86/mmu: Avoid retpoline on ->page_fault() with TDP
Sean Christopherson [Thu, 6 Feb 2020 22:14:34 +0000 (14:14 -0800)]
KVM: x86/mmu: Avoid retpoline on ->page_fault() with TDP

Wrap calls to ->page_fault() with a small shim to directly invoke the
TDP fault handler when the kernel is using retpolines and TDP is being
used.  Single out the TDP fault handler and annotate the TDP path as
likely to coerce the compiler into preferring it over the indirect
function call.

Rename tdp_page_fault() to kvm_tdp_page_fault(), as it's exposed outside
of mmu.c to allow inlining the shim.

Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: apic: reuse smp_wmb() in kvm_make_request()
Miaohe Lin [Fri, 7 Feb 2020 15:22:07 +0000 (23:22 +0800)]
KVM: apic: reuse smp_wmb() in kvm_make_request()

kvm_make_request() provides smp_wmb() so pending_events changes are
guaranteed to be visible.

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: remove duplicated KVM_REQ_EVENT request
Miaohe Lin [Fri, 7 Feb 2020 15:27:51 +0000 (23:27 +0800)]
KVM: x86: remove duplicated KVM_REQ_EVENT request

The KVM_REQ_EVENT request is already made in kvm_set_rflags(). We should
not make it again.

Signed-off-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoselftests: KVM: SVM: Add vmcall test
Eric Auger [Thu, 6 Feb 2020 10:47:10 +0000 (11:47 +0100)]
selftests: KVM: SVM: Add vmcall test

L2 guest calls vmcall and L1 checks the exit status does
correspond.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Tested-by: Wei Huang <wei.huang2@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoselftests: KVM: AMD Nested test infrastructure
Eric Auger [Thu, 6 Feb 2020 10:47:09 +0000 (11:47 +0100)]
selftests: KVM: AMD Nested test infrastructure

Add the basic infrastructure needed to test AMD nested SVM.
This is largely copied from the KVM unit test infrastructure.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoselftests: KVM: Replace get_{gdt,idt}_base() by get_{gdt,idt}()
Eric Auger [Thu, 6 Feb 2020 10:47:08 +0000 (11:47 +0100)]
selftests: KVM: Replace get_{gdt,idt}_base() by get_{gdt,idt}()

get_gdt_base() and get_idt_base() only return the base address
of the descriptor tables. Soon we will need to get the size as well.
Change the prototype of those functions so that they return
the whole desc_ptr struct instead of the address field.

Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Vitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: Miaohe Lin <linmiaohe@huawei.com>
Reviewed-by: Wei Huang <wei.huang2@amd.com>
Reviewed-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: Deliver exception payload on KVM_GET_VCPU_EVENTS
Oliver Upton [Fri, 7 Feb 2020 10:36:06 +0000 (02:36 -0800)]
KVM: x86: Deliver exception payload on KVM_GET_VCPU_EVENTS

KVM allows the deferral of exception payloads when a vCPU is in guest
mode to allow the L1 hypervisor to intercept certain events (#PF, #DB)
before register state has been modified. However, this behavior is
incompatible with the KVM_{GET,SET}_VCPU_EVENTS ABI, as userspace
expects register state to have been immediately modified. Userspace may
opt-in for the payload deferral behavior with the
KVM_CAP_EXCEPTION_PAYLOAD per-VM capability. As such,
kvm_multiple_exception() will immediately manipulate guest registers if
the capability hasn't been requested.

Since the deferral is only necessary if a userspace ioctl were to be
serviced at the same as a payload bearing exception is recognized, this
behavior can be relaxed. Instead, opportunistically defer the payload
from kvm_multiple_exception() and deliver the payload before completing
a KVM_GET_VCPU_EVENTS ioctl.

Signed-off-by: Oliver Upton <oupton@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: nVMX: Handle pending #DB when injecting INIT VM-exit
Oliver Upton [Fri, 7 Feb 2020 10:36:05 +0000 (02:36 -0800)]
KVM: nVMX: Handle pending #DB when injecting INIT VM-exit

SDM 27.3.4 states that the 'pending debug exceptions' VMCS field will
be populated if a VM-exit caused by an INIT signal takes priority over a
debug-trap. Emulate this behavior when synthesizing an INIT signal
VM-exit into L1.

Fixes: 4b9852f4f389 ("KVM: x86: Fix INIT signal handling in various CPU states")
Signed-off-by: Oliver Upton <oupton@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: Mask off reserved bit from #DB exception payload
Oliver Upton [Fri, 7 Feb 2020 10:36:04 +0000 (02:36 -0800)]
KVM: x86: Mask off reserved bit from #DB exception payload

KVM defines the #DB payload as compatible with the 'pending debug
exceptions' field under VMX, not DR6. Mask off bit 12 when applying the
payload to DR6, as it is reserved on DR6 but not the 'pending debug
exceptions' field.

Fixes: f10c729ff965 ("kvm: vmx: Defer setting of DR6 until #DB delivery")
Signed-off-by: Oliver Upton <oupton@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: Disable preemption in kvm_get_running_vcpu()
Marc Zyngier [Fri, 7 Feb 2020 16:34:10 +0000 (16:34 +0000)]
KVM: Disable preemption in kvm_get_running_vcpu()

Accessing a per-cpu variable only makes sense when preemption is
disabled (and the kernel does check this when the right debug options
are switched on).

For kvm_get_running_vcpu(), it is fine to return the value after
re-enabling preemption, as the preempt notifiers will make sure that
this is kept consistent across task migration (the comment above the
function hints at it, but lacks the crucial preemption management).

While we're at it, move the comment from the ARM code, which explains
why the whole thing works.

Fixes: 7495e22bb165 ("KVM: Move running VCPU from ARM to common code").
Cc: Paolo Bonzini <pbonzini@redhat.com>
Reported-by: Zenghui Yu <yuzenghui@huawei.com>
Tested-by: Zenghui Yu <yuzenghui@huawei.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Link: https://lore.kernel.org/r/318984f6-bc36-33a3-abc6-bf2295974b06@huawei.com
Message-id: <20200207163410.31276-1-maz@kernel.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoKVM: x86: do not reset microcode version on INIT or RESET
Paolo Bonzini [Tue, 11 Feb 2020 17:40:58 +0000 (18:40 +0100)]
KVM: x86: do not reset microcode version on INIT or RESET

Do not initialize the microcode version at RESET or INIT, only on vCPU
creation.   Microcode updates are not lost during INIT, and exact
behavior across a warm RESET is not specified by the architecture.

Since we do not support a microcode update directly from the hypervisor,
but only as a result of userspace setting the microcode version MSR,
it's simpler for userspace if we do nothing in KVM and let userspace
emulate behavior for RESET as it sees fit.

Userspace can tie the fix to the availability of MSR_IA32_UCODE_REV in
the list of emulated MSRs.

Reported-by: Alex Williamson <alex.williamson@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
4 years agoLinux 5.6-rc1
Linus Torvalds [Mon, 10 Feb 2020 00:08:48 +0000 (16:08 -0800)]
Linux 5.6-rc1

4 years agoMerge tag 'kbuild-v5.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy...
Linus Torvalds [Mon, 10 Feb 2020 00:05:50 +0000 (16:05 -0800)]
Merge tag 'kbuild-v5.6-2' of git://git./linux/kernel/git/masahiroy/linux-kbuild

Pull more Kbuild updates from Masahiro Yamada:

 - fix randconfig to generate a sane .config

 - rename hostprogs-y / always to hostprogs / always-y, which are more
   natual syntax.

 - optimize scripts/kallsyms

 - fix yes2modconfig and mod2yesconfig

 - make multiple directory targets ('make foo/ bar/') work

* tag 'kbuild-v5.6-2' of git://git.kernel.org/pub/scm/linux/kernel/git/masahiroy/linux-kbuild:
  kbuild: make multiple directory targets work
  kconfig: Invalidate all symbols after changing to y or m.
  kallsyms: fix type of kallsyms_token_table[]
  scripts/kallsyms: change table to store (strcut sym_entry *)
  scripts/kallsyms: rename local variables in read_symbol()
  kbuild: rename hostprogs-y/always to hostprogs/always-y
  kbuild: fix the document to use extra-y for vmlinux.lds
  kconfig: fix broken dependency in randconfig-generated .config

4 years agoMerge tag 'zonefs-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal...
Linus Torvalds [Sun, 9 Feb 2020 23:51:46 +0000 (15:51 -0800)]
Merge tag 'zonefs-5.6-rc1' of git://git./linux/kernel/git/dlemoal/zonefs

Pull new zonefs file system from Damien Le Moal:
 "Zonefs is a very simple file system exposing each zone of a zoned
  block device as a file.

  Unlike a regular file system with native zoned block device support
  (e.g. f2fs or the on-going btrfs effort), zonefs does not hide the
  sequential write constraint of zoned block devices to the user. As a
  result, zonefs is not a POSIX compliant file system. Its goal is to
  simplify the implementation of zoned block devices support in
  applications by replacing raw block device file accesses with a richer
  file based API, avoiding relying on direct block device file ioctls
  which may be more obscure to developers.

  One example of this approach is the implementation of LSM
  (log-structured merge) tree structures (such as used in RocksDB and
  LevelDB) on zoned block devices by allowing SSTables to be stored in a
  zone file similarly to a regular file system rather than as a range of
  sectors of a zoned device. The introduction of the higher level
  construct "one file is one zone" can help reducing the amount of
  changes needed in the application while at the same time allowing the
  use of zoned block devices with various programming languages other
  than C.

  Zonefs IO management implementation uses the new iomap generic code.
  Zonefs has been successfully tested using a functional test suite
  (available with zonefs userland format tool on github) and a prototype
  implementation of LevelDB on top of zonefs"

* tag 'zonefs-5.6-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/dlemoal/zonefs:
  zonefs: Add documentation
  fs: New zonefs file system

4 years agoirqchip/gic-v4.1: Avoid 64bit division for the sake of 32bit ARM
Marc Zyngier [Sun, 9 Feb 2020 22:48:50 +0000 (22:48 +0000)]
irqchip/gic-v4.1: Avoid 64bit division for the sake of 32bit ARM

In order to allow the GICv4 code to link properly on 32bit ARM,
make sure we don't use 64bit divisions when it isn't strictly
necessary.

Fixes: 4e6437f12d6e ("irqchip/gic-v4.1: Ensure L2 vPE table is allocated at RD level")
Reported-by: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Zenghui Yu <yuzenghui@huawei.com>
Signed-off-by: Marc Zyngier <maz@kernel.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
4 years agoMerge tag '5.6-rc-smb3-plugfest-patches' of git://git.samba.org/sfrench/cifs-2.6
Linus Torvalds [Sun, 9 Feb 2020 21:27:17 +0000 (13:27 -0800)]
Merge tag '5.6-rc-smb3-plugfest-patches' of git://git.samba.org/sfrench/cifs-2.6

Pull cifs fixes from Steve French:
 "13 cifs/smb3 patches, most from testing at the SMB3 plugfest this week:

   - Important fix for multichannel and for modefromsid mounts.

   - Two reconnect fixes

   - Addition of SMB3 change notify support

   - Backup tools fix

   - A few additional minor debug improvements (tracepoints and
     additional logging found useful during testing this week)"

* tag '5.6-rc-smb3-plugfest-patches' of git://git.samba.org/sfrench/cifs-2.6:
  smb3: Add defines for new information level, FileIdInformation
  smb3: print warning once if posix context returned on open
  smb3: add one more dynamic tracepoint missing from strict fsync path
  cifs: fix mode bits from dir listing when mounted with modefromsid
  cifs: fix channel signing
  cifs: add SMB3 change notification support
  cifs: make multichannel warning more visible
  cifs: fix soft mounts hanging in the reconnect code
  cifs: Add tracepoints for errors on flush or fsync
  cifs: log warning message (once) if out of disk space
  cifs: fail i/o on soft mounts if sessionsetup errors out
  smb3: fix problem with null cifs super block with previous patch
  SMB3: Backup intent flag missing from some more ops

4 years agoMerge branch 'work.vboxsf' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs
Linus Torvalds [Sun, 9 Feb 2020 20:41:00 +0000 (12:41 -0800)]
Merge branch 'work.vboxsf' of git://git./linux/kernel/git/viro/vfs

Pull vboxfs from Al Viro:
 "This is the VirtualBox guest shared folder support by Hans de Goede,
  with fixups for fs_parse folded in to avoid bisection hazards from
  those API changes..."

* 'work.vboxsf' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs:
  fs: Add VirtualBox guest shared folder (vboxsf) support

4 years agoMerge tag 'x86-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 9 Feb 2020 20:11:12 +0000 (12:11 -0800)]
Merge tag 'x86-urgent-2020-02-09' of git://git./linux/kernel/git/tip/tip

Pull x86 fixes from Thomas Gleixner:
 "A set of fixes for X86:

   - Ensure that the PIT is set up when the local APIC is disable or
     configured in legacy mode. This is caused by an ordering issue
     introduced in the recent changes which skip PIT initialization when
     the TSC and APIC frequencies are already known.

   - Handle malformed SRAT tables during early ACPI parsing which caused
     an infinite loop anda boot hang.

   - Fix a long standing race in the affinity setting code which affects
     PCI devices with non-maskable MSI interrupts. The problem is caused
     by the non-atomic writes of the MSI address (destination APIC id)
     and data (vector) fields which the device uses to construct the MSI
     message. The non-atomic writes are mandated by PCI.

     If both fields change and the device raises an interrupt after
     writing address and before writing data, then the MSI block
     constructs a inconsistent message which causes interrupts to be
     lost and subsequent malfunction of the device.

     The fix is to redirect the interrupt to the new vector on the
     current CPU first and then switch it over to the new target CPU.
     This allows to observe an eventually raised interrupt in the
     transitional stage (old CPU, new vector) to be observed in the APIC
     IRR and retriggered on the new target CPU and the new vector.

     The potential spurious interrupts caused by this are harmless and
     can in the worst case expose a buggy driver (all handlers have to
     be able to deal with spurious interrupts as they can and do happen
     for various reasons).

   - Add the missing suspend/resume mechanism for the HYPERV hypercall
     page which prevents resume hibernation on HYPERV guests. This
     change got lost before the merge window.

   - Mask the IOAPIC before disabling the local APIC to prevent
     potentially stale IOAPIC remote IRR bits which cause stale
     interrupt lines after resume"

* tag 'x86-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/apic: Mask IOAPIC entries when disabling the local APIC
  x86/hyperv: Suspend/resume the hypercall page for hibernation
  x86/apic/msi: Plug non-maskable MSI affinity race
  x86/boot: Handle malformed SRAT tables during early ACPI parsing
  x86/timer: Don't skip PIT setup when APIC is disabled or in legacy mode

4 years agoMerge tag 'smp-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 9 Feb 2020 20:09:43 +0000 (12:09 -0800)]
Merge tag 'smp-urgent-2020-02-09' of git://git./linux/kernel/git/tip/tip

Pull SMP fixes from Thomas Gleixner:
 "Two fixes for the SMP related functionality:

   - Make the UP version of smp_call_function_single() match SMP
     semantics when called for a not available CPU. Instead of emitting
     a warning and assuming that the function call target is CPU0,
     return a proper error code like the SMP version does.

   - Remove a superfluous check in smp_call_function_many_cond()"

* tag 'smp-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  smp/up: Make smp_call_function_single() match SMP semantics
  smp: Remove superfluous cond_func check in smp_call_function_many_cond()

4 years agoMerge tag 'perf-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git...
Linus Torvalds [Sun, 9 Feb 2020 20:04:09 +0000 (12:04 -0800)]
Merge tag 'perf-urgent-2020-02-09' of git://git./linux/kernel/git/tip/tip

Pull perf fixes from Thomas Gleixner:
 "A set of fixes and improvements for the perf subsystem:

  Kernel fixes:

   - Install cgroup events to the correct CPU context to prevent a
     potential list double add

   - Prevent an integer underflow in the perf mlock accounting

   - Add a missing prototype for arch_perf_update_userpage()

  Tooling:

   - Add a missing unlock in the error path of maps__insert() in perf
     maps.

   - Fix the build with the latest libbfd

   - Fix the perf parser so it does not delete parse event terms, which
     caused a regression for using perf with the ARM CoreSight as the
     sink configuration was missing due to the deletion.

   - Fix the double free in the perf CPU map merging test case

   - Add the missing ustring support for the perf probe command"

* tag 'perf-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  perf maps: Add missing unlock to maps__insert() error case
  perf probe: Add ustring support for perf probe command
  perf: Make perf able to build with latest libbfd
  perf test: Fix test case Merge cpu map
  perf parse: Copy string to perf_evsel_config_term
  perf parse: Refactor 'struct perf_evsel_config_term'
  kernel/events: Add a missing prototype for arch_perf_update_userpage()
  perf/cgroups: Install cgroup events to correct cpuctx
  perf/core: Fix mlock accounting in perf_mmap()

4 years agoMerge tag 'timers-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Sun, 9 Feb 2020 20:00:12 +0000 (12:00 -0800)]
Merge tag 'timers-urgent-2020-02-09' of git://git./linux/kernel/git/tip/tip

Pull timer fixes from Thomas Gleixner:
 "Two small fixes for the time(r) subsystem:

   - Handle a subtle race between the clocksource watchdog and a
     concurrent clocksource watchdog stop/start sequence correctly to
     prevent a timer double add bug.

   - Fix the file path for the core time namespace file"

* tag 'timers-urgent-2020-02-09' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  clocksource: Prevent double add_timer_on() for watchdog_timer
  MAINTAINERS: Correct path to time namespace source file