openwrt/staging/blogic.git
8 years agoKVM: s390: vsie: initial support for nested virtualization
David Hildenbrand [Wed, 8 Jul 2015 11:19:48 +0000 (13:19 +0200)]
KVM: s390: vsie: initial support for nested virtualization

This patch adds basic support for nested virtualization on s390x, called
VSIE (virtual SIE) and allows it to be used by the guest if the necessary
facilities are supported by the hardware and enabled for the guest.

In order to make this work, we have to shadow the sie control block
provided by guest 2. In order to gain some performance, we have to
reuse the same shadow blocks as good as possible. For now, we allow
as many shadow blocks as we have VCPUs (that way, every VCPU can run the
VSIE concurrently).

We have to watch out for the prefix getting unmapped out of our shadow
gmap and properly get the VCPU out of VSIE in that case, to fault the
prefix pages back in. We use the PROG_REQUEST bit for that purpose.

This patch is based on an initial prototype by Tobias Elpelt.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agomm/page_ref: introduce page_ref_inc_return
David Hildenbrand [Mon, 20 Jun 2016 10:09:41 +0000 (12:09 +0200)]
mm/page_ref: introduce page_ref_inc_return

Let's introduce that helper.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390: introduce page_to_virt() and pfn_to_virt()
David Hildenbrand [Mon, 16 Nov 2015 09:45:03 +0000 (10:45 +0100)]
s390: introduce page_to_virt() and pfn_to_virt()

Acked-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: backup the currently enabled gmap when scheduled out
David Hildenbrand [Wed, 11 Mar 2015 15:47:33 +0000 (16:47 +0100)]
KVM: s390: backup the currently enabled gmap when scheduled out

Nested virtualization will have to enable own gmaps. Current code
would enable the wrong gmap whenever scheduled out and back in,
therefore resulting in the wrong gmap being enabled.

This patch reenables the last enabled gmap, therefore avoiding having to
touch vcpu->arch.gmap when enabling a different gmap.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: fast path for shadow gmaps in gmap notifier
David Hildenbrand [Mon, 27 Apr 2015 14:29:34 +0000 (16:29 +0200)]
KVM: s390: fast path for shadow gmaps in gmap notifier

The default kvm gmap notifier doesn't have to handle shadow gmaps.
So let's just directly exit in case we get notified about one.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: don't fault everything in read-write in gmap_pte_op_fixup()
David Hildenbrand [Mon, 13 Jun 2016 08:49:04 +0000 (10:49 +0200)]
s390/mm: don't fault everything in read-write in gmap_pte_op_fixup()

Let's not fault in everything in read-write but limit it to read-only
where possible.

When restricting access rights, we already have the required protection
level in our hands. When reading from guest 2 storage (gmap_read_table),
it is obviously PROT_READ. When shadowing a pte, the required protection
level is given via the guest 2 provided pte.

Based on an initial patch by Martin Schwidefsky.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: allow to check if a gmap shadow is valid
David Hildenbrand [Fri, 27 May 2016 16:57:33 +0000 (18:57 +0200)]
s390/mm: allow to check if a gmap shadow is valid

It will be very helpful to have a mechanism to check without any locks
if a given gmap shadow is still valid and matches the given properties.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: remember the int code for the last gmap fault
David Hildenbrand [Tue, 8 Mar 2016 11:31:52 +0000 (12:31 +0100)]
s390/mm: remember the int code for the last gmap fault

For nested virtualization, we want to know if we are handling a protection
exception, because these can directly be forwarded to the guest without
additional checks.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: limit number of real-space gmap shadows
David Hildenbrand [Mon, 2 May 2016 10:10:17 +0000 (12:10 +0200)]
s390/mm: limit number of real-space gmap shadows

We have no known user of real-space designation and only support it to
be architecture compliant.

Gmap shadows with real-space designation are never unshadowed
automatically, as there is nothing to protect for the top level table.

So let's simply limit the number of such shadows to one by removing
existing ones on creation of another one.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: support real-space for gmap shadows
David Hildenbrand [Mon, 18 Apr 2016 14:22:24 +0000 (16:22 +0200)]
s390/mm: support real-space for gmap shadows

We can easily support real-space designation just like EDAT1 and EDAT2.
So guest2 can provide for guest3 an asce with the real-space control being
set.

We simply have to allocate the biggest page table possible and fake all
levels.

There is no protection to consider. If we exceed guest memory, vsie code
will inject an addressing exception (via program intercept). In the future,
we could limit the fake table level to the gmap page table.

As the top level page table can never go away, such gmap shadows will never
get unshadowed, we'll have to come up with another way to limit the number
of kept gmap shadows.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: push rte protection down to shadow pte
David Hildenbrand [Mon, 18 Apr 2016 15:46:21 +0000 (17:46 +0200)]
s390/mm: push rte protection down to shadow pte

Just like we already do with ste protection, let's take rte protection
into account. This way, the host pte doesn't have to be mapped writable.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: support EDAT2 for gmap shadows
David Hildenbrand [Mon, 18 Apr 2016 11:42:05 +0000 (13:42 +0200)]
s390/mm: support EDAT2 for gmap shadows

If the guest is enabled for EDAT2, we can easily create shadows for
guest2 -> guest3 provided tables that make use of EDAT2.

If guest2 references a 2GB page, this memory looks consecutive for guest2,
but it does not have to be so for us. Therefore we have to create fake
segment and page tables.

This works just like EDAT1 support, so page tables are removed when the
parent table (r3t table entry) is changed.

We don't hve to care about:
- ACCF-Validity Control in RTTE
- Access-Control Bits in RTTE
- Fetch-Protection Bit in RTTE
- Common-Region Bit in RTTE

Just like for EDAT1, all bits might be dropped and there is no guaranteed
that they are active.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: support EDAT1 for gmap shadows
David Hildenbrand [Mon, 18 Apr 2016 11:24:52 +0000 (13:24 +0200)]
s390/mm: support EDAT1 for gmap shadows

If the guest is enabled for EDAT1, we can easily create shadows for
guest2 -> guest3 provided tables that make use of EDAT1.

If guest2 references a 1MB page, this memory looks consecutive for guest2,
but it might not be so for us. Therefore we have to create fake page tables.

We can easily add that to our existing infrastructure. The invalidation
mechanism will make sure that fake page tables are removed when the parent
table (sgt table entry) is changed.

As EDAT1 also introduced protection on all page table levels, we have to
also shadow these correctly.

We don't have to care about:
- ACCF-Validity Control in STE
- Access-Control Bits in STE
- Fetch-Protection Bit in STE
- Common-Segment Bit in STE

As all bits might be dropped and there is no guaranteed that they are
active ("unpredictable whether the CPU uses these bits", "may be used").
Without using EDAT1 in the shadow ourselfes (STE-format control == 0),
simply shadowing these bits would not be enough. They would be ignored.

Please note that we are using the "fake" flag to make this look consistent
with further changes (EDAT2, real-space designation support) and don't let
the shadow functions handle fc=1 stes.

In the future, with huge pages in the host, gmap_shadow_pgt() could simply
try to map a huge host page if "fake" is set to one and indicate via return
value that no lower fake tables / shadow ptes are required.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: prepare for EDAT1/EDAT2 support in gmap shadow
David Hildenbrand [Tue, 8 Mar 2016 11:17:40 +0000 (12:17 +0100)]
s390/mm: prepare for EDAT1/EDAT2 support in gmap shadow

In preparation for EDAT1/EDAT2 support for gmap shadows, we have to store
the requested edat level in the gmap shadow.

The edat level used during shadow translation is a property of the gmap
shadow. Depending on that level, the gmap shadow will look differently for
the same guest tables. We have to store it internally in order to support
it later.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: push ste protection down to shadow pte
David Hildenbrand [Mon, 18 Apr 2016 15:19:59 +0000 (17:19 +0200)]
s390/mm: push ste protection down to shadow pte

If a guest ste is read-only, it doesn't make sense to force the ptes in as
writable in the host. If the source page is read-only in the host, it won't
have to be made writable. Please note that if the source page is not
available, it will still be faulted in writable. This can be changed
internally later on.

If ste protection is removed, underlying shadow tables are also removed,
therefore this change does not affect the guest.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: take ipte_lock during shadow faults
David Hildenbrand [Wed, 27 Jan 2016 16:24:03 +0000 (17:24 +0100)]
s390/mm: take ipte_lock during shadow faults

Let's take the ipte_lock while working on guest 2 provided page table, just
like the other gaccess functions.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: protection exceptions are corrrectly shadowed
David Hildenbrand [Wed, 27 Jan 2016 16:18:41 +0000 (17:18 +0100)]
s390/mm: protection exceptions are corrrectly shadowed

As gmap shadows contains correct protection permissions, protection
exceptons can directly be forwarded to guest 3. If we would encounter
a protection exception while faulting, the next guest 3 run will
automatically handle that for us.

Keep the dat_protection logic in place, as it will be helpful later.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: take the mmap_sem in kvm_s390_shadow_fault()
David Hildenbrand [Tue, 2 Feb 2016 11:26:00 +0000 (12:26 +0100)]
s390/mm: take the mmap_sem in kvm_s390_shadow_fault()

Instead of doing it in the caller, let's just take the mmap_sem
in kvm_s390_shadow_fault(). By taking it as read, we allow parallel
faulting on shadow page tables, gmap shadow code is prepared for that.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: fix races on gmap_shadow creation
David Hildenbrand [Tue, 8 Mar 2016 11:30:46 +0000 (12:30 +0100)]
s390/mm: fix races on gmap_shadow creation

Before any thread is allowed to use a gmap_shadow, it has to be fully
initialized. However, for invalidation to work properly, we have to
register the new gmap_shadow before we protect the parent gmap table.

Because locking is tricky, and we have to avoid duplicate gmaps, let's
introduce an initialized field, that signalizes other threads if that
gmap_shadow can already be used or if they have to retry.

Let's properly return errors using ERR_PTR() instead of simply returning
NULL, so a caller can properly react on the error.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: avoid races on region/segment/page table shadowing
David Hildenbrand [Tue, 8 Mar 2016 11:23:38 +0000 (12:23 +0100)]
s390/mm: avoid races on region/segment/page table shadowing

We have to unlock sg->guest_table_lock in order to call
gmap_protect_rmap(). If we sleep just before that call, another VCPU
might pick up that shadowed page table (while it is not protected yet)
and use it.

In order to avoid these races, we have to introduce a third state -
"origin set but still invalid" for an entry. This way, we can avoid
another thread already using the entry before the table is fully protected.
As soon as everything is set up, we can clear the invalid bit - if we
had no race with the unshadowing code.

Suggested-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: shadow pages with real guest requested protection
David Hildenbrand [Tue, 8 Mar 2016 11:21:41 +0000 (12:21 +0100)]
s390/mm: shadow pages with real guest requested protection

We really want to avoid manually handling protection for nested
virtualization. By shadowing pages with the protection the guest asked us
for, the SIE can handle most protection-related actions for us (e.g.
special handling for MVPG) and we can directly forward protection
exceptions to the guest.

PTEs will now always be shadowed with the correct _PAGE_PROTECT flag.
Unshadowing will take care of any guest changes to the parent PTE and
any host changes to the host PTE. If the host PTE doesn't have the
fitting access rights or is not available, we have to fix it up.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: flush tlb of shadows in all situations
David Hildenbrand [Fri, 15 Apr 2016 10:45:45 +0000 (12:45 +0200)]
s390/mm: flush tlb of shadows in all situations

For now, the tlb of shadow gmap is only flushed when the parent is removed,
not when it is removed upfront. Therefore other shadow gmaps can reuse the
tables without the tlb getting flushed.

Fix this by simply flushing the tlb
1. Before the shadow tables are removed (analogouos to other unshadow functions)
2. When the gmap is freed and therefore the top level pages are freed.

Acked-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: add kvm shadow fault function
Martin Schwidefsky [Tue, 8 Mar 2016 11:16:35 +0000 (12:16 +0100)]
s390/mm: add kvm shadow fault function

This patch introduces function kvm_s390_shadow_fault() used to resolve a
fault on a shadow gmap. This function will do validity checking and
build up the shadow page table hierarchy in order to fault in the
requested page into the shadow page table structure.

If an exception occurs while shadowing, guest 2 has to be notified about
it using either an exception or a program interrupt intercept. If
concurrent unshadowing occurres, this function will simply return with
-EAGAIN and the caller has to retry.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: add shadow gmap support
Martin Schwidefsky [Tue, 8 Mar 2016 11:12:18 +0000 (12:12 +0100)]
s390/mm: add shadow gmap support

For a nested KVM guest the outer KVM host needs to create shadow
page tables for the nested guest. This patch adds the basic support
to the guest address space (gmap) code.

For each guest address space the inner KVM host creates, the first
outer KVM host needs to create shadow page tables. The address space
is identified by the ASCE loaded into the control register 1 at the
time the inner SIE instruction for the second nested KVM guest is
executed. The outer KVM host creates the shadow tables starting with
the table identified by the ASCE on a on-demand basis. The outer KVM
host will get repeated faults for all the shadow tables needed to
run the second KVM guest.

While a shadow page table for the second KVM guest is active the access
to the origin region, segment and page tables needs to be restricted
for the first KVM guest. For region and segment and page tables the first
KVM guest may read the memory, but write attempt has to lead to an
unshadow.  This is done using the page invalid and read-only bits in the
page table of the first KVM guest. If the first guest re-accesses one of
the origin pages of a shadow, it gets a fault and the affected parts of
the shadow page table hierarchy needs to be removed again.

PGSTE tables don't have to be shadowed, as all interpretation assist can't
deal with the invalid bits in the shadow pte being set differently than
the original ones provided by the first KVM guest.

Many bug fixes and improvements by David Hildenbrand.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: add reference counter to gmap structure
Martin Schwidefsky [Tue, 8 Mar 2016 10:55:04 +0000 (11:55 +0100)]
s390/mm: add reference counter to gmap structure

Let's use a reference counter mechanism to control the lifetime of
gmap structures. This will be needed for further changes related to
gmap shadows.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: extended gmap pte notifier
Martin Schwidefsky [Tue, 8 Mar 2016 10:54:42 +0000 (11:54 +0100)]
s390/mm: extended gmap pte notifier

The current gmap pte notifier forces a pte into to a read-write state.
If the pte is invalidated the gmap notifier is called to inform KVM
that the mapping will go away.

Extend this approach to allow read-write, read-only and no-access
as possible target states and call the pte notifier for any change
to the pte.

This mechanism is used to temporarily set specific access rights for
a pte without doing the heavy work of a true mprotect call.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: use RCU for gmap notifier list and the per-mm gmap list
Martin Schwidefsky [Tue, 8 Mar 2016 10:54:14 +0000 (11:54 +0100)]
s390/mm: use RCU for gmap notifier list and the per-mm gmap list

The gmap notifier list and the gmap list in the mm_struct change rarely.
Use RCU to optimize the reader of these lists.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/kvm: page table invalidation notifier
Martin Schwidefsky [Tue, 8 Mar 2016 10:52:54 +0000 (11:52 +0100)]
s390/kvm: page table invalidation notifier

Pass an address range to the page table invalidation notifier
for KVM. This allows to notify changes that affect a larger
virtual memory area, e.g. for 1MB pages.

Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agokvm: vmx: hook preemption timer support
Yunhong Jiang [Mon, 13 Jun 2016 21:19:59 +0000 (14:19 -0700)]
kvm: vmx: hook preemption timer support

Hook the VMX preemption timer to the "hv timer" functionality added
by the previous patch.  This includes: checking if the feature is
supported, if the feature is broken on the CPU, the hooks to
setup/clean the VMX preemption timer, arming the timer on vmentry
and handling the vmexit.

A module parameter states if the VMX preemption timer should be
utilized.

Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
[Move hv_deadline_tsc to struct vcpu_vmx, use -1 as the "unset" value.
 Put all VMX bits here.  Enable it by default #yolo. - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agokvm: vmx: rename vmx_pre/post_block to pi_pre/post_block
Yunhong Jiang [Mon, 13 Jun 2016 21:19:58 +0000 (14:19 -0700)]
kvm: vmx: rename vmx_pre/post_block to pi_pre/post_block

Prepare to switch from preemption timer to hrtimer in the
vmx_pre/post_block. Current functions are only for posted interrupt,
rename them accordingly.

Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: x86: support using the vmx preemption timer for tsc deadline timer
Yunhong Jiang [Mon, 13 Jun 2016 21:20:01 +0000 (14:20 -0700)]
KVM: x86: support using the vmx preemption timer for tsc deadline timer

The VMX preemption timer can be used to virtualize the TSC deadline timer.
The VMX preemption timer is armed when the vCPU is running, and a VMExit
will happen if the virtual TSC deadline timer expires.

When the vCPU thread is blocked because of HLT, KVM will switch to use
an hrtimer, and then go back to the VMX preemption timer when the vCPU
thread is unblocked.

This solution avoids the complex OS's hrtimer system, and the host
timer interrupt handling cost, replacing them with a little math
(for guest->host TSC and host TSC->preemption timer conversion)
and a cheaper VMexit.  This benefits latency for isolated pCPUs.

[A word about performance... Yunhong reported a 30% reduction in average
 latency from cyclictest.  I made a similar test with tscdeadline_latency
 from kvm-unit-tests, and measured

 - ~20 clock cycles loss (out of ~3200, so less than 1% but still
   statistically significant) in the worst case where the test halts
   just after programming the TSC deadline timer

 - ~800 clock cycles gain (25% reduction in latency) in the best case
   where the test busy waits.

 I removed the VMX bits from Yunhong's patch, to concentrate them in the
 next patch - Paolo]

Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agokvm: lapic: separate start_sw_tscdeadline from start_apic_timer
Yunhong Jiang [Mon, 13 Jun 2016 21:20:00 +0000 (14:20 -0700)]
kvm: lapic: separate start_sw_tscdeadline from start_apic_timer

The function to start the tsc deadline timer virtualization will be used
also by the pre_block hook when we use the preemption timer; change it
to a separate function. No logic changes.

Signed-off-by: Yunhong Jiang <yunhong.jiang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: s390: use kvm->created_vcpus
Paolo Bonzini [Mon, 13 Jun 2016 12:50:04 +0000 (14:50 +0200)]
KVM: s390: use kvm->created_vcpus

The new created_vcpus field avoids possible races between enabling
capabilities and creating VCPUs.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: remove kvm_vcpu_compatible
Paolo Bonzini [Mon, 13 Jun 2016 12:50:04 +0000 (14:50 +0200)]
KVM: remove kvm_vcpu_compatible

The new created_vcpus field makes it possible to avoid the race between
irqchip and VCPU creation in a much nicer way; just check under kvm->lock
whether a VCPU has already been created.

We can then remove KVM_APIC_ARCHITECTURE too, because at this point the
symbol is only governing the default definition of kvm_vcpu_compatible.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: introduce kvm->created_vcpus
Paolo Bonzini [Mon, 13 Jun 2016 12:48:25 +0000 (14:48 +0200)]
KVM: introduce kvm->created_vcpus

The race between creating the irqchip and the first VCPU is
currently fixed by checking the presence of an irqchip before
updating kvm->online_vcpus, and undoing the whole VCPU creation
if someone created the irqchip in the meanwhile.

Instead, introduce a new field in struct kvm that will count VCPUs
under a mutex, without the atomic access and memory ordering that we
need elsewhere to protect the vcpus array.  This also plugs the race
and is more easily applicable in all similar circumstances.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agox86/kvm/svm: Simplify cpu_has_svm()
Borislav Petkov [Mon, 9 May 2016 09:53:06 +0000 (11:53 +0200)]
x86/kvm/svm: Simplify cpu_has_svm()

Use already cached CPUID information instead of querying CPUID again.

No functionality change.

Signed-off-by: Borislav Petkov <bp@suse.de>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: kvm@vger.kernel.org
Cc: x86@kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Use mipsregs.h defs for config registers
James Hogan [Wed, 15 Jun 2016 18:30:01 +0000 (19:30 +0100)]
MIPS: KVM: Use mipsregs.h defs for config registers

Convert MIPS KVM guest register state initialisation to use the standard
<asm/mipsregs.h> register field definitions for Config registers, and
drop the custom definitions in kvm_host.h which it was using before.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Report more accurate CP0_Config fields to guest
James Hogan [Wed, 15 Jun 2016 18:30:00 +0000 (19:30 +0100)]
MIPS: KVM: Report more accurate CP0_Config fields to guest

Initialise the guest's CP0_Config register with a few more bits of
information from the host. The BE bit should be set on big endian
machines, the VI bit should be set on machines with a virtually tagged
instruction cache, and the reported architecture revision should match
that of the host (since we won't support emulating pre-r6 instruction
encodings on r6 or vice versa).

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: Add define for Config.VI (virtual icache) bit
James Hogan [Wed, 15 Jun 2016 18:29:59 +0000 (19:29 +0100)]
MIPS: Add define for Config.VI (virtual icache) bit

The Config.VI bit specifies that the instruction cache is virtually
tagged, which is checked in c-r4k.c's probe_pcache(). Add a proper
definition for it in mipsregs.h and make use of it.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Use host CCA for TLB mappings
James Hogan [Wed, 15 Jun 2016 18:29:58 +0000 (19:29 +0100)]
MIPS: KVM: Use host CCA for TLB mappings

KVM TLB mappings for the guest were being created with a cache coherency
attribute (CCA) of 3, which is cached incoherent. Create them instead
with the default host CCA, which should be the correct one for coherency
on SMP systems.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Move commpage so 0x0 is unmapped
James Hogan [Wed, 15 Jun 2016 18:29:57 +0000 (19:29 +0100)]
MIPS: KVM: Move commpage so 0x0 is unmapped

The comm page which is mapped into the guest kernel address space at
0x0 has the unfortunate side effect of allowing guest kernel NULL
pointer dereferences to succeed. The only constraint on this address is
that it must be within 32KiB of 0x0, so that single lw/sw instructions
(which have 16-bit signed offset fields) can be used to access it, using
the zero register as a base.

So lets move the comm page as high as possible within that constraint so
that 0x0 can be left unmapped, at least for page sizes < 32KiB.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Add KScratch registers
James Hogan [Wed, 15 Jun 2016 18:29:56 +0000 (19:29 +0100)]
MIPS: KVM: Add KScratch registers

Allow up to 6 KVM guest KScratch registers to be enabled and accessed
via the KVM guest register API and from the guest itself (the fallback
reading and writing of commpage registers is sufficient for KScratch
registers to work as expected).

User mode can expose the registers by setting the appropriate bits of
the guest Config4.KScrExist field. KScratch registers that aren't usable
won't be writeable via the KVM Ioctl API.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Emulate RDHWR CPUNum register
James Hogan [Wed, 15 Jun 2016 18:29:55 +0000 (19:29 +0100)]
MIPS: KVM: Emulate RDHWR CPUNum register

Actually provide the VCPU number when emulating the RDHWR CPUNum
register, so that it will match the CPUNum field of CP0_EBase register,
rather than always returning 0.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Allow ULRI to restrict UserLocal register
James Hogan [Wed, 15 Jun 2016 18:29:54 +0000 (19:29 +0100)]
MIPS: KVM: Allow ULRI to restrict UserLocal register

The ULRI bit in Config3 specifies whether the UserLocal register is
implemented, but it is assumed to always be set. Now that the Config
registers can be modified by userland, allow Config3.ULRI to be cleared
and check ULRI before allowing the corresponding bit to be set in
HWREna.

In fact any HWREna bits corresponding to unimplemented RDHWR registers
should read as zero and be ignored on write, so we actually prevent
other unimplemented bits being set too.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Don't hardcode restored HWREna
James Hogan [Wed, 15 Jun 2016 18:29:53 +0000 (19:29 +0100)]
MIPS: KVM: Don't hardcode restored HWREna

KVM modifies CP0_HWREna during guest execution so it can trap and
emulate RDHWR instructions, however it always restores the hardcoded
value 0x2000000F. This assumes the presence of the UserLocal register,
and the absence of any implementation dependent or future HW registers.

Fix by exporting the value that traps.c write into CP0_HWREna, and
loading from there instead of hard coding.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: Clean up RDHWR handling
James Hogan [Wed, 15 Jun 2016 18:29:52 +0000 (19:29 +0100)]
MIPS: Clean up RDHWR handling

No preprocessor definitions are used in the handling of the registers
accessible with the RDHWR instruction, nor the corresponding bits in the
CP0 HWREna register.

Add definitions for both the register numbers (MIPS_HWR_*) and HWREna
bits (MIPS_HWRENA_*) in asm/mipsregs.h and make use of them in the
initialisation of HWREna and emulation of the RDHWR instruction.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: David Daney <david.daney@cavium.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: List FPU/MSA registers
James Hogan [Wed, 15 Jun 2016 18:29:51 +0000 (19:29 +0100)]
MIPS: KVM: List FPU/MSA registers

Make KVM_GET_REG_LIST list FPU & MSA registers. Specifically we list all
32 vector registers when MSA can be enabled, 32 single-precision FP
registers when FPU can be enabled, and either 16 or 32 double-precision
FP registers when FPU can be enabled depending on whether FR mode is
supported (which provides 32 doubles instead of 16 even doubles).

Note, these registers may still be inaccessible depending on the current
FP mode of the guest.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Use raw_cpu_has_fpu in kvm_mips_guest_can_have_fpu()
James Hogan [Wed, 15 Jun 2016 18:29:50 +0000 (19:29 +0100)]
MIPS: KVM: Use raw_cpu_has_fpu in kvm_mips_guest_can_have_fpu()

We need to use kvm_mips_guest_can_have_fpu() when deciding which
registers to list with KVM_GET_REG_LIST, however it causes warnings with
preemption since it uses cpu_has_fpu. KVM is only really supported on
CPUs which have symmetric FPUs, so switch to raw_cpu_has_fpu to avoid
the warning.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Make KVM_GET_REG_LIST dynamic
James Hogan [Wed, 15 Jun 2016 18:29:49 +0000 (19:29 +0100)]
MIPS: KVM: Make KVM_GET_REG_LIST dynamic

Make the implementation of KVM_GET_REG_LIST more dynamic so that only
the subset of registers actually available can be exposed to user mode.
This is important for VZ where some of the guest register state may not
be possible to prevent the guest from accessing, therefore the user
process may need to be aware of the state even if it doesn't understand
what the state is for.

This also allows different MIPS KVM implementations to provide different
registers to one another, by way of new num_regs(vcpu) and
copy_reg_indices(vcpu, indices) callback functions, currently just
stubbed for trap & emulate.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Pass all unknown registers to callbacks
James Hogan [Wed, 15 Jun 2016 18:29:48 +0000 (19:29 +0100)]
MIPS: KVM: Pass all unknown registers to callbacks

Pass all unrecognised register IDs through to the set_one_reg() and
get_one_reg() callbacks, not just select ones. This allows
implementation specific registers to be more easily added without having
to modify arch/mips/kvm/mips.c.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Convert emulation to use asm/inst.h
James Hogan [Wed, 15 Jun 2016 18:29:47 +0000 (19:29 +0100)]
MIPS: KVM: Convert emulation to use asm/inst.h

Convert various MIPS KVM guest instruction emulation functions to decode
instructions (and encode translations) using the union mips_instruction
and related enumerations in asm/inst.h rather than #defines and
hardcoded values.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Factor writing of translated guest instructions
James Hogan [Wed, 15 Jun 2016 18:29:46 +0000 (19:29 +0100)]
MIPS: KVM: Factor writing of translated guest instructions

The code in kvm_mips_dyntrans.c to write a translated guest instruction
to guest memory depending on the segment is duplicated between each of
the functions. Additionally the cache op translation functions assume
the instruction is in the KSEG0/1 segment rather than KSEG2/3, which is
generally true but isn't guaranteed.

Factor that code into a new kvm_mips_trans_replace() which handles both
KSEG0/1 and KSEG2/3.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Fix translation of MFC0 ErrCtl
James Hogan [Wed, 15 Jun 2016 18:29:45 +0000 (19:29 +0100)]
MIPS: KVM: Fix translation of MFC0 ErrCtl

The MIPS KVM dynamic translation is meant to translate "MFC0 rt, ErrCtl"
instructions into "ADD rt, zero, zero" to zero the destination register,
however the rt register number was copied into rt of the ADD instruction
encoding, which is the 2nd source operand. This results in "ADD zero,
zero, rt" which is a no-op, so only the first execution of each such
MFC0 from ErrCtl will actually read 0.

Fix the shift to put the rt from the MFC0 encoding into the rd field of
the ADD.

Fixes: 50c8308538dc ("KVM/MIPS32: Binary patching of select privileged instructions.")
Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge tag 'kvm-s390-next-4.8-1' of git://git.kernel.org/pub/scm/linux/kernel/git...
Paolo Bonzini [Wed, 15 Jun 2016 07:21:46 +0000 (09:21 +0200)]
Merge tag 'kvm-s390-next-4.8-1' of git://git./linux/kernel/git/kvms390/linux into HEAD

KVM: s390: Features and fixes for 4.8 part1

Four bigger things:
1. The implementation of the STHYI opcode in the kernel. This is used
   in libraries like qclib [1] to provide enough information for a
   capacity and usage based software licence pricing. The STHYI content
   is defined by the related z/VM documentation [2]. Its data can be
   composed by accessing several other interfaces provided by LPAR or
   the machine. This information is partially sensitive or root-only
   so the kernel does the necessary filtering.
2. Preparation for nested virtualization (VSIE). KVM should query the
   proper sclp interfaces for the availability of some features before
   using it. In the past we have been sloppy and simply assumed that
   several features are available. With this we should be able to handle
   most cases of a missing feature.
3. CPU model interfaces extended by some additional features that are
   not covered by a facility bit in STFLE. For example all the crypto
   instructions of the coprocessor provide a query function. As reality
   tends to be more complex (e.g. export regulations might block some
   algorithms) we have to provide additional interfaces to query or
   set these non-stfle features.
4. Several fixes and changes detected and fixed when doing 1-3.

All features change base s390 code. All relevant patches have an ACK
from the s390 or component maintainers.

The next pull request for 4.8 (part2) will contain the implementation
of VSIE.

[1] http://www.ibm.com/developerworks/linux/linux390/qclib.html
[2] https://www.ibm.com/support/knowledgecenter/SSB27U_6.3.0/com.ibm.zvm.v630.hcpb4/hcpb4sth.htm

8 years agoKVM: x86: Fix typos
Andrea Gelmini [Sat, 21 May 2016 12:14:44 +0000 (14:14 +0200)]
KVM: x86: Fix typos

Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: S390: Fix typo
Andrea Gelmini [Sat, 21 May 2016 12:08:55 +0000 (14:08 +0200)]
KVM: S390: Fix typo

Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: ARM64: Fix typos
Andrea Gelmini [Sat, 21 May 2016 11:53:14 +0000 (13:53 +0200)]
KVM: ARM64: Fix typos

Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: ARM: Fix typos
Andrea Gelmini [Sat, 21 May 2016 11:48:35 +0000 (13:48 +0200)]
KVM: ARM: Fix typos

Signed-off-by: Andrea Gelmini <andrea.gelmini@gelma.net>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Print unknown load/store encodings
James Hogan [Tue, 14 Jun 2016 08:40:17 +0000 (09:40 +0100)]
MIPS: KVM: Print unknown load/store encodings

When trying to emulate an unrecognised load or store instruction, print
the encoding to aid debug.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Dump guest tlbs if kvm_get_inst() fails
James Hogan [Tue, 14 Jun 2016 08:40:16 +0000 (09:40 +0100)]
MIPS: KVM: Dump guest tlbs if kvm_get_inst() fails

If kvm_get_inst() fails to find a guest TLB mapping for the guest PC
then dump the guest TLB entries. The contents of the guest TLB is likely
to be more interesting than the host TLB entries.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Trace guest register access emulation
James Hogan [Tue, 14 Jun 2016 08:40:15 +0000 (09:40 +0100)]
MIPS: KVM: Trace guest register access emulation

Trace emulation of guest access to various registers via
MFC0/MTC0/DMFC0/DMTC0 instructions (coprocessor 0) and the RDHWR
instruction (hardware registers exposed to userland), replacing some
existing kvm_debug calls. Trace events are much more practical for this
kind of debug output.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Add guest mode switch trace events
James Hogan [Tue, 14 Jun 2016 08:40:14 +0000 (09:40 +0100)]
MIPS: KVM: Add guest mode switch trace events

Add a few trace events for entering and coming out of guest mode, as well
as re-entering it from a guest exit exception.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Add kvm_asid_change trace event
James Hogan [Tue, 14 Jun 2016 08:40:13 +0000 (09:40 +0100)]
MIPS: KVM: Add kvm_asid_change trace event

Add a trace event for guest ASID changes, replacing the existing
kvm_debug call.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Clean up kvm_exit trace event
James Hogan [Tue, 14 Jun 2016 08:40:12 +0000 (09:40 +0100)]
MIPS: KVM: Clean up kvm_exit trace event

Clean up the MIPS kvm_exit trace event so that the exit reasons are
specified in a trace friendly way (via __print_symbolic), and so that
the exit reasons that derive straight from Cause.ExcCode values map
directly, allowing a single trace_kvm_exit() call to replace a bunch of
individual ones.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Add kvm_aux trace event
James Hogan [Tue, 14 Jun 2016 08:40:11 +0000 (09:40 +0100)]
MIPS: KVM: Add kvm_aux trace event

Add a MIPS specific trace event for auxiliary context operations
(notably FPU and MSA). Unfortunately the generic kvm_fpu trace event
isn't flexible enough to handle the range of interesting things that can
happen with FPU and MSA context.

The type of state being operated on is traced:
- FPU: Just the FPU registers.
- MSA: Just the upper half of the MSA vector registers (low half already
       loaded with FPU state).
- FPU & MSA: Full MSA vector state (includes FPU state).

As is the type of operation:
- Restore: State was enabled and restored.
- Save: State was saved and disabled.
- Enable: State was enabled (already loaded).
- Disable: State was disabled (kept loaded).
- Discard: State was discarded and disabled.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
[Fix remaining occurrence of "fpu_msa", change to "aux". - Paolo]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Generalise fpu_inuse for other state
James Hogan [Tue, 14 Jun 2016 08:40:10 +0000 (09:40 +0100)]
MIPS: KVM: Generalise fpu_inuse for other state

Rename fpu_inuse and the related definitions to aux_inuse so it can be
used for lazy context management of other auxiliary processor state too,
such as VZ guest timer, watchpoints and performance counters.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Use va in kvm_get_inst()
James Hogan [Thu, 9 Jun 2016 13:19:21 +0000 (14:19 +0100)]
MIPS: KVM: Use va in kvm_get_inst()

Like other functions, make use of a local unsigned long va, for the
virtual address of the PC. This reduces the amount of verbose casting of
the opc pointer to an unsigned long.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Combine handle_tlb_ld/st_miss
James Hogan [Thu, 9 Jun 2016 13:19:20 +0000 (14:19 +0100)]
MIPS: KVM: Combine handle_tlb_ld/st_miss

The handle_tlb_ld/st_miss handlers are logically equivalent and
textually almost identical, so combine their implementations into a
single kvm_trap_emul_handle_tlb_miss().

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Use MIPS_ENTRYLO_* defs from mipsregs.h
James Hogan [Thu, 9 Jun 2016 13:19:19 +0000 (14:19 +0100)]
MIPS: KVM: Use MIPS_ENTRYLO_* defs from mipsregs.h

Convert KVM to use the MIPS_ENTRYLO_* definitions from <asm/mipsregs.h>
rather than custom definitions in kvm_host.h

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Simplify TLB_* macros
James Hogan [Thu, 9 Jun 2016 13:19:18 +0000 (14:19 +0100)]
MIPS: KVM: Simplify TLB_* macros

Simplify some of the TLB_ macros making use of the arrayification of
tlb_lo. Basically we index the array by the bit of the virtual address
which determines whether the even or odd entry is used, instead of
having a conditional.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Arrayify struct kvm_mips_tlb::tlb_lo*
James Hogan [Thu, 9 Jun 2016 13:19:17 +0000 (14:19 +0100)]
MIPS: KVM: Arrayify struct kvm_mips_tlb::tlb_lo*

The values of the EntryLo0 and EntryLo1 registers for a TLB entry are
stored in separate members of struct kvm_mips_tlb called tlb_lo0 and
tlb_lo1 respectively. To allow future code which needs to manipulate
arbitrary EntryLo data in the TLB entry to be simpler and less
conditional, replace these members with an array of two elements.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Use dump_tlb_all() for kvm_mips_dump_host_tlbs()
James Hogan [Thu, 9 Jun 2016 13:19:16 +0000 (14:19 +0100)]
MIPS: KVM: Use dump_tlb_all() for kvm_mips_dump_host_tlbs()

KVM implements its own routine for dumping the host TLB entries, but we
already have dump_tlb_all() which does something very similar (although
it only prints out TLB entries which match the current ASID or are
global).

Make KVM use dump_tlb_all() along with dump_tlb_regs() to avoid the
duplication and inevitable bitrot, allowing TLB dumping enhancements
(e.g. for VZ and GuestIDs) to be made in a single place.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Clean up TLB management hazards
James Hogan [Thu, 9 Jun 2016 13:19:15 +0000 (14:19 +0100)]
MIPS: KVM: Clean up TLB management hazards

KVM's host TLB handling routines were using tlbw hazard barrier macros
around tlb_read(). Now that hazard barrier macros exist for tlbr, update
this case to use them.

Also fix various other unnecessary hazard barriers in this code.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Restore host EBase from ebase variable
James Hogan [Thu, 9 Jun 2016 13:19:14 +0000 (14:19 +0100)]
MIPS: KVM: Restore host EBase from ebase variable

The host kernel's exception vector base address is currently saved in
the VCPU structure at creation time, and restored on a guest exit.
However it doesn't change and can already be easily accessed from the
'ebase' variable (arch/mips/kernel/traps.c), so drop the host_ebase
member of kvm_vcpu_arch, export the 'ebase' variable to modules and load
from there instead.

This does result in a single extra instruction (lui) on the guest exit
path, but simplifies the code a bit and removes the redundant storage of
the host exception base address.

Credit for the idea goes to Cavium's VZ KVM implementation.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Drop unused hpa0/hpa1 args from function
James Hogan [Thu, 9 Jun 2016 13:19:13 +0000 (14:19 +0100)]
MIPS: KVM: Drop unused hpa0/hpa1 args from function

The function kvm_mips_handle_mapped_seg_tlb_fault() has two completely
unused pointer arguments, hpa0 and hpa1, for which all users always pass
NULL.

Drop these two arguments and update the callers.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Simplify even/odd TLB handling
James Hogan [Thu, 9 Jun 2016 13:19:12 +0000 (14:19 +0100)]
MIPS: KVM: Simplify even/odd TLB handling

When handling TLB faults in the guest KSeg0 region, a pair of physical
addresses are read from the guest physical address map. However that
process is rather convoluted with an if/then/else statement. Simplify it
to just clear the lowest bit for the even entry and set the lowest bit
for the odd entry.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Don't indirect KVM functions
James Hogan [Thu, 9 Jun 2016 13:19:11 +0000 (14:19 +0100)]
MIPS: KVM: Don't indirect KVM functions

Several KVM module functions are indirected so that they can be accessed
from tlb.c which is statically built into the kernel. This is no longer
necessary as the relevant bits of code have moved into mmu.c which is
part of the KVM module, so drop the indirections.

Note: is_error_pfn() is defined inline in kvm_host.h, so didn't actually
require the KVM module to be loaded for it to work anyway.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Move non-TLB handling code out of tlb.c
James Hogan [Thu, 9 Jun 2016 13:19:10 +0000 (14:19 +0100)]
MIPS: KVM: Move non-TLB handling code out of tlb.c

Various functions in tlb.c perform higher level MMU handling, but don't
strictly need to be statically built into the kernel as they don't
directly manipulate TLB entries. Move these functions out into a
separate mmu.c which will be built into the KVM kernel module. This
allows them to directly reference KVM functions in the KVM kernel module
in future.

Module exports of these functions have been removed, since they aren't
needed outside of KVM.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Make various Cause variables 32-bit
James Hogan [Thu, 9 Jun 2016 13:19:09 +0000 (14:19 +0100)]
MIPS: KVM: Make various Cause variables 32-bit

The CP0 Cause register is passed around in KVM quite a bit, often as an
unsigned long, even though it is always 32-bits long.

Resize it to u32 throughout MIPS KVM.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Convert code to kernel sized types
James Hogan [Thu, 9 Jun 2016 13:19:08 +0000 (14:19 +0100)]
MIPS: KVM: Convert code to kernel sized types

Convert the MIPS KVM C code to use standard kernel sized types (e.g.
u32) instead of inttypes.h style ones (e.g. uint32_t) or other types as
appropriate.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Convert headers to kernel sized types
James Hogan [Thu, 9 Jun 2016 13:19:07 +0000 (14:19 +0100)]
MIPS: KVM: Convert headers to kernel sized types

Convert the MIPS kvm_host.h structs, function declaration prototypes and
associated definition prototypes to use standard kernel sized types
(e.g. u32) instead of inttypes.h style ones (e.g. uint32_t).

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Drop unused kvm_mips_sync_icache()
James Hogan [Thu, 9 Jun 2016 13:19:06 +0000 (14:19 +0100)]
MIPS: KVM: Drop unused kvm_mips_sync_icache()

The function kvm_mips_sync_icache() is unused, so lets remove it.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Drop unused host_cp0_entryhi
James Hogan [Thu, 9 Jun 2016 13:19:05 +0000 (14:19 +0100)]
MIPS: KVM: Drop unused host_cp0_entryhi

The host EntryHi in the KVM VCPU context is virtually unused. It gets
stored on exceptions, but only ever used in a kvm_debug() when a TLB
miss occurs.

Drop it entirely, removing that information from the kvm_debug output.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Drop unused guest_inst from kvm_vcpu_arch
James Hogan [Thu, 9 Jun 2016 13:19:04 +0000 (14:19 +0100)]
MIPS: KVM: Drop unused guest_inst from kvm_vcpu_arch

The MIPS kvm_vcpu_arch::guest_inst isn't used, so drop it from the
struct and drop its asm-offsets definition.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: linux-mips@linux-mips.org
Cc: kvm@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMerge branch 'kvm-mips-fixes' into HEAD
Paolo Bonzini [Tue, 14 Jun 2016 09:00:16 +0000 (11:00 +0200)]
Merge branch 'kvm-mips-fixes' into HEAD

Merge MIPS patches destined to both 4.7 and kvm/next, to avoid
unnecessary conflicts.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Fix CACHE triggered exception emulation
James Hogan [Thu, 9 Jun 2016 09:50:46 +0000 (10:50 +0100)]
MIPS: KVM: Fix CACHE triggered exception emulation

When emulating TLB miss / invalid exceptions during CACHE instruction
emulation, be sure to set up the correct PC and host_cp0_badvaddr state
for the kvm_mips_emlulate_tlb*_ld() function to pick up for guest EPC
and BadVAddr.

PC needs to be rewound otherwise the guest EPC will end up pointing at
the next instruction after the faulting CACHE instruction.

host_cp0_badvaddr must be set because guest CACHE instructions trap with
a Coprocessor Unusable exception, which doesn't update the host BadVAddr
as a TLB exception would.

This doesn't tend to get hit when dynamic translation of emulated
instructions is enabled, since only the first execution of each CACHE
instruction actually goes through this code path, with subsequent
executions hitting the SYNCI instruction that it gets replaced with.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Don't unwind PC when emulating CACHE
James Hogan [Thu, 9 Jun 2016 09:50:45 +0000 (10:50 +0100)]
MIPS: KVM: Don't unwind PC when emulating CACHE

When a CACHE instruction is emulated by kvm_mips_emulate_cache(), the PC
is first updated to point to the next instruction, and afterwards it
falls through the "dont_update_pc" label, which rewinds the PC back to
its original address.

This works when dynamic translation of emulated instructions is enabled,
since the CACHE instruction is replaced with a SYNCI which works without
trapping, however when dynamic translation is disabled the guest hangs
on CACHE instructions as they always trap and are never stepped over.

Roughly swap the meanings of the "done" and "dont_update_pc" to match
kvm_mips_emulate_CP0(), so that "done" will roll back the PC on failure,
and "dont_update_pc" won't change PC at all (for the sake of exceptions
that have already modified the PC).

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Include bit 31 in segment matches
James Hogan [Thu, 9 Jun 2016 09:50:44 +0000 (10:50 +0100)]
MIPS: KVM: Include bit 31 in segment matches

When faulting guest addresses are matched against guest segments with
the KVM_GUEST_KSEGX() macro, change the mask to 0xe0000000 so as to
include bit 31.

This is mainly for safety's sake, as it prevents a rogue BadVAddr in the
host kseg2/kseg3 segments (e.g. 0xC*******) after a TLB exception from
matching the guest kseg0 segment (e.g. 0x4*******), triggering an
internal KVM error instead of allowing the corresponding guest kseg0
page to be mapped into the host vmalloc space.

Such a rogue BadVAddr was observed to happen with the host MIPS kernel
running under QEMU with KVM built as a module, due to a not entirely
transparent optimisation in the QEMU TLB handling. This has already been
worked around properly in a previous commit.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoMIPS: KVM: Fix modular KVM under QEMU
James Hogan [Thu, 9 Jun 2016 09:50:43 +0000 (10:50 +0100)]
MIPS: KVM: Fix modular KVM under QEMU

Copy __kvm_mips_vcpu_run() into unmapped memory, so that we can never
get a TLB refill exception in it when KVM is built as a module.

This was observed to happen with the host MIPS kernel running under
QEMU, due to a not entirely transparent optimisation in the QEMU TLB
handling where TLB entries replaced with TLBWR are copied to a separate
part of the TLB array. Code in those pages continue to be executable,
but those mappings persist only until the next ASID switch, even if they
are marked global.

An ASID switch happens in __kvm_mips_vcpu_run() at exception level after
switching to the guest exception base. Subsequent TLB mapped kernel
instructions just prior to switching to the guest trigger a TLB refill
exception, which enters the guest exception handlers without updating
EPC. This appears as a guest triggered TLB refill on a host kernel
mapped (host KSeg2) address, which is not handled correctly as user
(guest) mode accesses to kernel (host) segments always generate address
error exceptions.

Signed-off-by: James Hogan <james.hogan@imgtec.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Ralf Baechle <ralf@linux-mips.org>
Cc: kvm@vger.kernel.org
Cc: linux-mips@linux-mips.org
Cc: <stable@vger.kernel.org> # 3.10.x-
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
8 years agoKVM: s390: handle missing storage-key facility
David Hildenbrand [Tue, 10 May 2016 07:50:21 +0000 (09:50 +0200)]
KVM: s390: handle missing storage-key facility

Without the storage-key facility, SIE won't interpret SSKE, ISKE and
RRBE for us. So let's add proper interception handlers that will be called
if lazy sske cannot be enabled.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: trace and count all skey intercepts
David Hildenbrand [Tue, 10 May 2016 07:40:09 +0000 (09:40 +0200)]
KVM: s390: trace and count all skey intercepts

Let's trace and count all skey handling operations, even if lazy skey
handling was already activated. Also, don't enable lazy skey handling if
anything went wrong while enabling skey handling for the SIE.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/sclp: detect storage-key facility
David Hildenbrand [Tue, 24 Nov 2015 11:56:43 +0000 (12:56 +0100)]
s390/sclp: detect storage-key facility

Let's correctly detect that facility.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: pfmf: handle address overflows
David Hildenbrand [Thu, 12 May 2016 12:07:05 +0000 (14:07 +0200)]
KVM: s390: pfmf: handle address overflows

In theory, end could always end up being < start, if overflowing to 0.
Although very unlikely for now, let's just fix it.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: pfmf: support conditional-sske facility
David Hildenbrand [Tue, 10 May 2016 07:43:11 +0000 (09:43 +0200)]
KVM: s390: pfmf: support conditional-sske facility

We already indicate that facility but don't implement it in our pfmf
interception handler. Let's add a new storage key handling function for
conditionally setting the guest storage key.

As we will reuse this function later on, let's directly implement returning
the old key via parameter and indicating if any change happened via rc.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: pfmf: take care of amode when setting reg2
David Hildenbrand [Wed, 13 Apr 2016 13:47:21 +0000 (15:47 +0200)]
KVM: s390: pfmf: take care of amode when setting reg2

Depending on the addressing mode, we must not overwrite bit 0-31 of the
register. In addition, 24 bit and 31 bit have to set certain bits to 0,
which is guaranteed by converting the end address to an effective
address.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: pfmf: MR and MC are ignored without CSSKE
David Hildenbrand [Wed, 13 Apr 2016 10:09:58 +0000 (12:09 +0200)]
KVM: s390: pfmf: MR and MC are ignored without CSSKE

These two bits are simply ignored when the conditional-SSKE facility is
not installed.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: pfmf: fix end address calculation
David Hildenbrand [Wed, 13 Apr 2016 08:09:47 +0000 (10:09 +0200)]
KVM: s390: pfmf: fix end address calculation

The current calculation is wrong if absolute != real address. Let's just
calculate the start address for 4k frames upfront. Otherwise, the
calculated end address will be wrong, resulting in wrong memory
location/storage keys getting touched.

To keep low-address protection working (using the effective address),
we have to move the check.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agoKVM: s390: storage keys fit into a char
David Hildenbrand [Mon, 9 May 2016 11:08:07 +0000 (13:08 +0200)]
KVM: s390: storage keys fit into a char

No need to convert the storage key into an unsigned long, the target
function expects a char as argument.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: return key via pointer in get_guest_storage_key
David Hildenbrand [Mon, 9 May 2016 09:22:34 +0000 (11:22 +0200)]
s390/mm: return key via pointer in get_guest_storage_key

Let's just split returning the key and reporting errors. This makes calling
code easier and avoids bugs as happened already.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
8 years agos390/mm: simplify get_guest_storage_key
David Hildenbrand [Mon, 9 May 2016 09:15:32 +0000 (11:15 +0200)]
s390/mm: simplify get_guest_storage_key

We can safe a few LOC and make that function easier to understand
by rewriting existing code.

Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>