Re-factor code into a helper function for setting lapic parameters when
activate/deactivate APICv, and export the function for subsequent usage.
Signed-off-by: NSuravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Now that we are mapping kvm_steal_time from the guest directly we
don't need keep a copy of it in kvm_vcpu_arch.st. The same is true
for the stime field.

This is part of CVE-2019-3016.
Signed-off-by: NBoris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: NJoao Martins <joao.m.martins@oracle.com>
Cc: stable@vger.kernel.org
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

There is a potential race in record_steal_time() between setting
host-local vcpu->arch.st.steal.preempted to zero (i.e. clearing
KVM_VCPU_PREEMPTED) and propagating this value to the guest with
kvm_write_guest_cached(). Between those two events the guest may
still see KVM_VCPU_PREEMPTED in its copy of kvm_steal_time, set
KVM_VCPU_FLUSH_TLB and assume that hypervisor will do the right
thing. Which it won't.

Instad of copying, we should map kvm_steal_time and that will
guarantee atomicity of accesses to @preempted.

This is part of CVE-2019-3016.
Signed-off-by: NBoris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: NJoao Martins <joao.m.martins@oracle.com>
Cc: stable@vger.kernel.org
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

__kvm_map_gfn()'s call to gfn_to_pfn_memslot() is
* relatively expensive
* in certain cases (such as when done from atomic context) cannot be called

Stashing gfn-to-pfn mapping should help with both cases.

This is part of CVE-2019-3016.
Signed-off-by: NBoris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: NJoao Martins <joao.m.martins@oracle.com>
Cc: stable@vger.kernel.org
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

kvm_steal_time_set_preempted() may accidentally clear KVM_VCPU_FLUSH_TLB
bit if it is called more than once while VCPU is preempted.

This is part of CVE-2019-3016.

(This bug was also independently discovered by Jim Mattson
<jmattson@google.com>)
Signed-off-by: NBoris Ostrovsky <boris.ostrovsky@oracle.com>
Reviewed-by: NJoao Martins <joao.m.martins@oracle.com>
Cc: stable@vger.kernel.org
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

'struct timex' is one of the last users of 'struct timeval' and is
only referenced in one place in the kernel any more, to convert the
user space timex into the kernel-internal version on sparc64, with a
different tv_usec member type.

As a preparation for hiding the time_t definition and everything
using that in the kernel, change the implementation once more
to only convert the timeval member, and then enclose the
struct definition in an #ifdef.
Signed-off-by: NArnd Bergmann <arnd@arndb.de>
Reviewed-by: NJulian Calaby <julian.calaby@gmail.com>
Acked-by: NThomas Gleixner <tglx@linutronix.de>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Implement primitives necessary for the 4th level folding, add walks of p4d
level where appropriate and replace 5level-fixup.h with pgtable-nop4d.h.
Signed-off-by: NMike Rapoport <rppt@linux.ibm.com>
Acked-by: NDavid S. Miller <davem@davemloft.net>
Signed-off-by: NDavid S. Miller <davem@davemloft.net>

Fixes: 3a167bea ("kvm: powerpc: Add kvmppc_ops callback")
Signed-off-by: NDavid Michael <fedora.dm0@gmail.com>
Signed-off-by: NPaul Mackerras <paulus@ozlabs.org>

When migrate_vma_setup() fails in kvmppc_svm_page_out(),
release kvm->arch.uvmem_lock before returning.

Fixes: ca9f4942 ("KVM: PPC: Book3S HV: Support for running secure guests")
Signed-off-by: NBharata B Rao <bharata@linux.ibm.com>
Reviewed-by: NKamalesh Babulal <kamalesh@linux.vnet.ibm.com>
Signed-off-by: NPaul Mackerras <paulus@ozlabs.org>

There's an OF helper called of_dma_is_coherent(), which checks if a
device has a "dma-coherent" property to see if the device is coherent
for DMA.

But on some platforms devices are coherent by default, and on some
platforms it's not possible to update existing device trees to add the
"dma-coherent" property.

So add a Kconfig symbol to allow arch code to tell
of_dma_is_coherent() that devices are coherent by default, regardless
of the presence of the property.

Select that symbol on powerpc when NOT_COHERENT_CACHE is not set, ie.
when the system has a coherent cache.

Fixes: 92ea637e ("of: introduce of_dma_is_coherent() helper")
Cc: stable@vger.kernel.org # v3.16+
Reported-by: NChristian Zigotzky <chzigotzky@xenosoft.de>
Tested-by: NChristian Zigotzky <chzigotzky@xenosoft.de>
Signed-off-by: NMichael Ellerman <mpe@ellerman.id.au>
Reviewed-by: NUlf Hansson <ulf.hansson@linaro.org>
Signed-off-by: NRob Herring <robh@kernel.org>

Add a typedef to for the fastop function prototype to make the code more
readable.

No functional change intended.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Reviewed-by: NMiaohe Lin <linmiaohe@huawei.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

It also helps eliminate some duplicated code.
Signed-off-by: NMiaohe Lin <linmiaohe@huawei.com>
Reviewed-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

The function now has a single caller, so there is no point
in keeping it separate.
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Walk the host page tables to identify hugepage mappings for ZONE_DEVICE
pfns, i.e. DAX pages.  Explicitly query kvm_is_zone_device_pfn() when
deciding whether or not to bother walking the host page tables, as DAX
pages do not set up the head/tail infrastructure, i.e. will return false
for PageCompound() even when using huge pages.

Zap ZONE_DEVICE sptes when disabling dirty logging, e.g. if live
migration fails, to allow KVM to rebuild large pages for DAX-based
mappings.  Presumably DAX favors large pages, and worst case scenario is
a minor performance hit as KVM will need to re-fault all DAX-based
pages.
Suggested-by: NBarret Rhoden <brho@google.com>
Cc: David Hildenbrand <david@redhat.com>
Cc: Dan Williams <dan.j.williams@intel.com>
Cc: Jason Zeng <jason.zeng@intel.com>
Cc: Dave Jiang <dave.jiang@intel.com>
Cc: Liran Alon <liran.alon@oracle.com>
Cc: linux-nvdimm <linux-nvdimm@lists.01.org>
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Remove the late "lpage is disallowed" check from set_spte() now that the
initial check is performed after acquiring mmu_lock.  Fold the guts of
the remaining helper, __mmu_gfn_lpage_is_disallowed(), into
kvm_mmu_hugepage_adjust() to eliminate the unnecessary slot !NULL check.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Fold max_mapping_level() into kvm_mmu_hugepage_adjust() now that HugeTLB
mappings are handled in kvm_mmu_hugepage_adjust(), i.e. there isn't a
need to pre-calculate the max mapping level.  Co-locating all hugepage
checks eliminates a memslot lookup, at the cost of performing the
__mmu_gfn_lpage_is_disallowed() checks while holding mmu_lock.

The latency of lpage_is_disallowed() is likely negligible relative to
the rest of the code run while holding mmu_lock, and can be offset to
some extent by eliminating the mmu_gfn_lpage_is_disallowed() check in
set_spte() in a future patch.  Eliminating the check in set_spte() is
made possible by performing the initial lpage_is_disallowed() checks
while holding mmu_lock.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Zap any compound page, e.g. THP or HugeTLB pages, when zapping sptes
that can potentially be converted to huge sptes after disabling dirty
logging on the associated memslot.  Note, this approach could result in
false positives, e.g. if a random compound page is mapped into the
guest, but mapping non-huge compound pages into the guest is far from
the norm, and toggling dirty logging is not a frequent operation.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Remove logic to retrieve the original gfn now that HugeTLB mappings are
are identified in FNAME(fetch), i.e. FNAME(page_fault) no longer adjusts
the level or gfn.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Remove KVM's HugeTLB specific logic and instead rely on walking the host
page tables (already done for THP) to identify HugeTLB mappings.
Eliminating the HugeTLB-only logic avoids taking mmap_sem and calling
find_vma() for all hugepage compatible page faults, and simplifies KVM's
page fault code by consolidating all hugepage adjustments into a common
helper.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Remove fast_page_fault()'s optimization to stop the shadow walk if the
iterator level drops below the intended map level.  The intended map
level is only acccurate for HugeTLB mappings (THP mappings are detected
after fast_page_fault()), i.e. it's not required for correctness, and
a future patch will also move HugeTLB mapping detection to after
fast_page_fault().
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Explicitly walk the host page tables to identify THP mappings instead
of relying solely on the metadata in struct page.  This sets the stage
for using a common method of identifying huge mappings regardless of the
underlying implementation (HugeTLB vs THB vs DAX), and hopefully avoids
the pitfalls of relying on metadata to identify THP mappings, e.g. see
commit 169226f7 ("mm: thp: handle page cache THP correctly in
PageTransCompoundMap") and the need for KVM to explicitly check for a
THP compound page.  KVM will also naturally work with 1gb THP pages, if
they are ever supported.

Walking the tables for THP mappings is likely marginally slower than
querying metadata, but a future patch will reuse the walk to identify
HugeTLB mappings, at which point eliminating the existing VMA lookup for
HugeTLB will make this a net positive.

Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: Barret Rhoden <brho@google.com>
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Refactor transparent_hugepage_adjust() in preparation for walking the
host page tables to identify hugepage mappings, initially for THP pages,
and eventualy for HugeTLB and DAX-backed pages as well.  The latter
cases support 1gb pages, i.e. the adjustment logic needs access to the
max allowed level.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Add a helper, lookup_address_in_mm(), to traverse the page tables of a
given mm struct.  KVM will use the helper to retrieve the host mapping
level, e.g. 4k vs. 2mb vs. 1gb, of a compound (or DAX-backed) page
without having to resort to implementation specific metadata.  E.g. KVM
currently uses different logic for HugeTLB vs. THP, and would add a
third variant for DAX-backed files.

Cc: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Reviewed-by: NThomas Gleixner <tglx@linutronix.de>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Use kvm_vcpu_gfn_to_hva() when retrieving the host page size so that the
correct set of memslots is used when handling x86 page faults in SMM.

Fixes: 54bf36aa ("KVM: x86: use vcpu-specific functions to read/write/translate GFNs")
Cc: stable@vger.kernel.org
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Add a helper, is_transparent_hugepage(), to explicitly check whether a
compound page is a THP and use it when populating KVM's secondary MMU.
The explicit check fixes a bug where a remapped compound page, e.g. for
an XDP Rx socket, is mapped into a KVM guest and is mistaken for a THP,
which results in KVM incorrectly creating a huge page in its secondary
MMU.

Fixes: 936a5fe6 ("thp: kvm mmu transparent hugepage support")
Reported-by: syzbot+c9d1fb51ac9d0d10c39d@syzkaller.appspotmail.com
Cc: Andrea Arcangeli <aarcange@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Limit KVM's mapping level for HugeTLB based on its calculated max_level.
The max_level check prior to invoking host_mapping_level() only filters
out the case where KVM cannot create a 2mb mapping, it doesn't handle
the scenario where KVM can create a 2mb but not 1gb mapping, and the
host is using a 1gb HugeTLB mapping.

Fixes: 2f57b705 ("KVM: x86/mmu: Persist gfn_lpage_is_disallowed() to max_level")
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

kvm_vector_hashing_enabled() is just called in kvm.ko module.
Signed-off-by: NPeng Hao <richard.peng@oppo.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

vmx_set_segment() clears segment cache unconditionally, so we should not
clear it again by calling vmx_segment_cache_clear().
Signed-off-by: NMiaohe Lin <linmiaohe@huawei.com>
Reviewed-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

These two conditions are in conflict, adding 'else' to reduce checking.
Signed-off-by: NHaiwei Li <lihaiwei@tencent.com>
Reviewed-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: NMiaohe Lin <linmiaohe@huawei.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

According to section "Checks on Guest Control Registers, Debug Registers, and
and MSRs" in Intel SDM vol 3C, the following checks are performed on vmentry
of nested guests:

    If the "load debug controls" VM-entry control is 1, bits 63:32 in the DR7
    field must be 0.

In KVM, GUEST_DR7 is set prior to the vmcs02 VM-entry by kvm_set_dr() and the
latter synthesizes a #GP if any bit in the high dword in the former is set.
Hence this field needs to be checked in software.
Signed-off-by: NKrish Sadhukhan <krish.sadhukhan@oracle.com>
Reviewed-by: NKarl Heubaum <karl.heubaum@oracle.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

For ring-based dirty log tracking, it will be more efficient to account
writes during schedule-out or schedule-in to the currently running VCPU.
We would like to do it even if the write doesn't use the current VCPU's
address space, as is the case for cached writes (see commit 4e335d9e,
"Revert "KVM: Support vCPU-based gfn->hva cache"", 2017-05-02).

Therefore, add a mechanism to track the currently-loaded kvm_vcpu struct.
There is already something similar in KVM/ARM; one important difference
is that kvm_arch_vcpu_{load,put} have two callers in virt/kvm/kvm_main.c:
we have to update both the architecture-independent vcpu_{load,put} and
the preempt notifiers.

Another change made in the process is to allow using kvm_get_running_vcpu()
in preemptible code.  This is allowed because preempt notifiers ensure
that the value does not change even after the VCPU thread is migrated.
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
Reviewed-by: NPaolo Bonzini <pbonzini@redhat.com>
Signed-off-by: NPeter Xu <peterx@redhat.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

The helper x86_set_memory_region() is only used in vmx_set_tss_addr()
and kvm_arch_destroy_vm().  Push the lock upper in both cases.  With
that, drop x86_set_memory_region().

This prepares to allow __x86_set_memory_region() to return a HVA
mapped, because the HVA will need to be protected by the lock too even
after __x86_set_memory_region() returns.
Signed-off-by: NPeter Xu <peterx@redhat.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

We've already got the slots_lock, so we should be safe.
Signed-off-by: NPeter Xu <peterx@redhat.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

In nested_enable_evmcs() evmcs_already_enabled check doesn't really do
anything: controls are already sanitized and we return '0' regardless.
Just drop the check.
Signed-off-by: NVitaly Kuznetsov <vkuznets@redhat.com>
Reviewed-by: NLiran Alon <liran.alon@oracle.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Remove the CONFIG_X86_64 condition from the low level non-canonical
helpers to effectively enable non-canonical checks on 32-bit KVM.
Non-canonical checks are performed by hardware if the CPU *supports*
64-bit mode, whether or not the CPU is actually in 64-bit mode is
irrelevant.

For the most part, skipping non-canonical checks on 32-bit KVM is ok-ish
because 32-bit KVM always (hopefully) drops bits 63:32 of whatever value
it's checking before propagating it to hardware, and architecturally,
the expected behavior for the guest is a bit of a grey area since the
vCPU itself doesn't support 64-bit mode.  I.e. a 32-bit KVM guest can
observe the missed checks in several paths, e.g. INVVPID and VM-Enter,
but it's debatable whether or not the missed checks constitute a bug
because technically the vCPU doesn't support 64-bit mode.

The primary motivation for enabling the non-canonical checks is defense
in depth.  As mentioned above, a guest can trigger a missed check via
INVVPID or VM-Enter.  INVVPID is straightforward as it takes a 64-bit
virtual address as part of its 128-bit INVVPID descriptor and fails if
the address is non-canonical, even if INVVPID is executed in 32-bit PM.
Nested VM-Enter is a bit more convoluted as it requires the guest to
write natural width VMCS fields via memory accesses and then VMPTRLD the
VMCS, but it's still possible.  In both cases, KVM is saved from a true
bug only because its flows that propagate values to hardware (correctly)
take "unsigned long" parameters and so drop bits 63:32 of the bad value.

Explicitly performing the non-canonical checks makes it less likely that
a bad value will be propagated to hardware, e.g. in the INVVPID case,
if __invvpid() didn't implicitly drop bits 63:32 then KVM would BUG() on
the resulting unexpected INVVPID failure due to hardware rejecting the
non-canonical address.

The only downside to enabling the non-canonical checks is that it adds a
relatively small amount of overhead, but the affected flows are not hot
paths, i.e. the overhead is negligible.

Note, KVM technically could gate the non-canonical checks on 32-bit KVM
with static_cpu_has(X86_FEATURE_LM), but on bare metal that's an even
bigger waste of code for everyone except the 0.00000000000001% of the
population running on Yonah, and nested 32-bit on 64-bit already fudges
things with respect to 64-bit CPU behavior.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
[Also do so in nested_vmx_check_host_state as reported by Krish. - Paolo]
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Writes to MSR_CORE_PERF_GLOBAL_CONTROL should never fail if the VM-exit
and VM-entry controls are exposed to L1. Promote the checks to perform a
full WARN if kvm_set_msr() fails and remove the now unused macro
SET_MSR_OR_WARN().
Suggested-by: NSean Christopherson <sean.j.christopherson@intel.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: NOliver Upton <oupton@google.com>
Reviewed-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Remove an unused struct x86_emulate_ctxt * param from low level helpers
used to access guest FPU state.  The unused param was left behind by
commit 6ab0b9fe ("x86,kvm: remove KVM emulator get_fpu / put_fpu").

No functional change intended.
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Reload the current thread's FPU state, which contains the guest's FPU
state, to the CPU registers if necessary during vcpu_enter_guest().
TIF_NEED_FPU_LOAD can be set any time control is transferred out of KVM,
e.g. if I/O is triggered during a KVM call to get_user_pages() or if a
softirq occurs while KVM is scheduled in.

Moving the handling of TIF_NEED_FPU_LOAD from vcpu_enter_guest() to
kvm_arch_vcpu_load(), effectively kvm_sched_in(), papered over a bug
where kvm_put_guest_fpu() failed to account for TIF_NEED_FPU_LOAD.  The
easiest way to the kvm_put_guest_fpu() bug was to run with involuntary
preemption enable, thus handling TIF_NEED_FPU_LOAD during kvm_sched_in()
made the bug go away.  But, removing the handling in vcpu_enter_guest()
exposed KVM to the rare case of a softirq triggering kernel_fpu_begin()
between vcpu_load() and vcpu_enter_guest().

Now that kvm_{load,put}_guest_fpu() correctly handle TIF_NEED_FPU_LOAD,
revert the commit to both restore the vcpu_enter_guest() behavior and
eliminate the superfluous switch_fpu_return() in kvm_arch_vcpu_load().

Note, leaving the handling in kvm_arch_vcpu_load() isn't wrong per se,
but it is unnecessary, and most critically, makes it extremely difficult
to find bugs such as the kvm_put_guest_fpu() issue due to shrinking the
window where a softirq can corrupt state.

A sample trace triggered by warning if TIF_NEED_FPU_LOAD is set while
vcpu state is loaded:

 <IRQ>
  gcmaes_crypt_by_sg.constprop.12+0x26e/0x660
  ? 0xffffffffc024547d
  ? __qdisc_run+0x83/0x510
  ? __dev_queue_xmit+0x45e/0x990
  ? ip_finish_output2+0x1a8/0x570
  ? fib4_rule_action+0x61/0x70
  ? fib4_rule_action+0x70/0x70
  ? fib_rules_lookup+0x13f/0x1c0
  ? helper_rfc4106_decrypt+0x82/0xa0
  ? crypto_aead_decrypt+0x40/0x70
  ? crypto_aead_decrypt+0x40/0x70
  ? crypto_aead_decrypt+0x40/0x70
  ? esp_output_tail+0x8f4/0xa5a [esp4]
  ? skb_ext_add+0xd3/0x170
  ? xfrm_input+0x7a6/0x12c0
  ? xfrm4_rcv_encap+0xae/0xd0
  ? xfrm4_transport_finish+0x200/0x200
  ? udp_queue_rcv_one_skb+0x1ba/0x460
  ? udp_unicast_rcv_skb.isra.63+0x72/0x90
  ? __udp4_lib_rcv+0x51b/0xb00
  ? ip_protocol_deliver_rcu+0xd2/0x1c0
  ? ip_local_deliver_finish+0x44/0x50
  ? ip_local_deliver+0xe0/0xf0
  ? ip_protocol_deliver_rcu+0x1c0/0x1c0
  ? ip_rcv+0xbc/0xd0
  ? ip_rcv_finish_core.isra.19+0x380/0x380
  ? __netif_receive_skb_one_core+0x7e/0x90
  ? netif_receive_skb_internal+0x3d/0xb0
  ? napi_gro_receive+0xed/0x150
  ? 0xffffffffc0243c77
  ? net_rx_action+0x149/0x3b0
  ? __do_softirq+0xe4/0x2f8
  ? handle_irq_event_percpu+0x6a/0x80
  ? irq_exit+0xe6/0xf0
  ? do_IRQ+0x7f/0xd0
  ? common_interrupt+0xf/0xf
  </IRQ>
  ? irq_entries_start+0x20/0x660
  ? vmx_get_interrupt_shadow+0x2f0/0x710 [kvm_intel]
  ? kvm_set_msr_common+0xfc7/0x2380 [kvm]
  ? recalibrate_cpu_khz+0x10/0x10
  ? ktime_get+0x3a/0xa0
  ? kvm_arch_vcpu_ioctl_run+0x107/0x560 [kvm]
  ? kvm_init+0x6bf/0xd00 [kvm]
  ? __seccomp_filter+0x7a/0x680
  ? do_vfs_ioctl+0xa4/0x630
  ? security_file_ioctl+0x32/0x50
  ? ksys_ioctl+0x60/0x90
  ? __x64_sys_ioctl+0x16/0x20
  ? do_syscall_64+0x5f/0x1a0
  ? entry_SYSCALL_64_after_hwframe+0x44/0xa9
---[ end trace 9564a1ccad733a90 ]---

This reverts commit e7517324.

Fixes: e7517324 ("KVM: X86: Fix fpu state crash in kvm guest")
Reported-by: NDerek Yerger <derek@djy.llc>
Reported-by: kernel@najdan.com
Cc: Wanpeng Li <wanpengli@tencent.com>
Cc: Thomas Lambertz <mail@thomaslambertz.de>
Cc: Rik van Riel <riel@surriel.com>
Cc: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: stable@vger.kernel.org
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Lock the FPU regs and reload the current thread's FPU state, which holds
the guest's FPU state, to the CPU registers if necessary prior to
accessing guest FPU state as part of emulation.  kernel_fpu_begin() can
be called from softirq context, therefore KVM must ensure softirqs are
disabled (locking the FPU regs disables softirqs) when touching CPU FPU
state.

Note, for all intents and purposes this reverts commit 6ab0b9fe
("x86,kvm: remove KVM emulator get_fpu / put_fpu"), but at the time it
was applied, removing get/put_fpu() was correct.  The re-introduction
of {get,put}_fpu() is necessitated by the deferring of FPU state load.

Fixes: 5f409e20 ("x86/fpu: Defer FPU state load until return to userspace")
Cc: stable@vger.kernel.org
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>

Handle TIF_NEED_FPU_LOAD similar to how fpu__copy() handles the flag
when duplicating FPU state to a new task struct.  TIF_NEED_FPU_LOAD can
be set any time control is transferred out of KVM, be it voluntarily,
e.g. if I/O is triggered during a KVM call to get_user_pages, or
involuntarily, e.g. if softirq runs after an IRQ occurs.  Therefore,
KVM must account for TIF_NEED_FPU_LOAD whenever it is (potentially)
accessing CPU FPU state.

Fixes: 5f409e20 ("x86/fpu: Defer FPU state load until return to userspace")
Cc: stable@vger.kernel.org
Signed-off-by: NSean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>