When we map a page that wasn't icache cleared before, do so when first
mapping it in KVM using the same information bits as the Linux mapping
logic. That way we are 100% sure that any page we map does not have stale
entries in the icache.
Signed-off-by: Alexander Graf <agraf@suse.de>
Added the decrementer auto-reload support. DECAR is readable
on e500v2/e500mc and later cpus.
Signed-off-by: Bharat Bhushan <bharat.bhushan@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
This adds a new ioctl to enable userspace to control the size of the guest
hashed page table (HPT) and to clear it out when resetting the guest.
The KVM_PPC_ALLOCATE_HTAB ioctl is a VM ioctl and takes as its parameter
a pointer to a u32 containing the desired order of the HPT (log base 2
of the size in bytes), which is updated on successful return to the
actual order of the HPT which was allocated.
There must be no vcpus running at the time of this ioctl. To enforce
this, we now keep a count of the number of vcpus running in
kvm->arch.vcpus_running.
If the ioctl is called when a HPT has already been allocated, we don't
reallocate the HPT but just clear it out. We first clear the
kvm->arch.rma_setup_done flag, which has two effects: (a) since we hold
the kvm->lock mutex, it will prevent any vcpus from starting to run until
we're done, and (b) it means that the first vcpu to run after we're done
will re-establish the VRMA if necessary.
If userspace doesn't call this ioctl before running the first vcpu, the
kernel will allocate a default-sized HPT at that point. We do it then
rather than when creating the VM, as the code did previously, so that
userspace has a chance to do the ioctl if it wants.
When allocating the HPT, we can allocate either from the kernel page
allocator, or from the preallocated pool. If userspace is asking for
a different size from the preallocated HPTs, we first try to allocate
using the kernel page allocator. Then we try to allocate from the
preallocated pool, and then if that fails, we try allocating decreasing
sizes from the kernel page allocator, down to the minimum size allowed
(256kB). Note that the kernel page allocator limits allocations to
1 << CONFIG_FORCE_MAX_ZONEORDER pages, which by default corresponds to
16MB (on 64-bit powerpc, at least).
Signed-off-by: Paul Mackerras <paulus@samba.org>
[agraf: fix module compilation]
Signed-off-by: Alexander Graf <agraf@suse.de>
There is nothing in the code for emulating TCE tables in the kernel
that prevents it from working on "PR" KVM... other than ifdef's and
location of the code.
This and moves the bulk of the code there to a new file called
book3s_64_vio.c.
This speeds things up a bit on my G5.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
[agraf: fix for hv kvm, 32bit, whitespace]
Signed-off-by: Alexander Graf <agraf@suse.de>
When emulating updating load/store instructions (lwzu, stwu, ...) we need to
write the effective address of the load/store into a register.
Currently, we write the physical address in there, which is very wrong. So
instead let's save off where the virtual fault was on MMIO and use that
information as value to put into the register.
While at it, also move the XOP variants of the above instructions to the new
scheme of using the already known vaddr instead of calculating it themselves.
Reported-by: Jörg Sommer <joerg@alea.gnuu.de>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This adds code to measure "stolen" time per virtual core in units of
timebase ticks, and to report the stolen time to the guest using the
dispatch trace log (DTL). The guest can register an area of memory
for the DTL for a given vcpu. The DTL is a ring buffer where KVM
fills in one entry every time it enters the guest for that vcpu.
Stolen time is measured as time when the virtual core is not running,
either because the vcore is not runnable (e.g. some of its vcpus are
executing elsewhere in the kernel or in userspace), or when the vcpu
thread that is running the vcore is preempted. This includes time
when all the vcpus are idle (i.e. have executed the H_CEDE hypercall),
which is OK because the guest accounts stolen time while idle as idle
time.
Each vcpu keeps a record of how much stolen time has been reported to
the guest for that vcpu so far. When we are about to enter the guest,
we create a new DTL entry (if the guest vcpu has a DTL) and report the
difference between total stolen time for the vcore and stolen time
reported so far for the vcpu as the "enqueue to dispatch" time in the
DTL entry.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
The PAPR API allows three sorts of per-virtual-processor areas to be
registered (VPA, SLB shadow buffer, and dispatch trace log), and
furthermore, these can be registered and unregistered for another
virtual CPU. Currently we just update the vcpu fields pointing to
these areas at the time of registration or unregistration. If this
is done on another vcpu, there is the possibility that the target vcpu
is using those fields at the time and could end up using a bogus
pointer and corrupting memory.
This fixes the race by making the target cpu itself do the update, so
we can be sure that the update happens at a time when the fields
aren't being used. Each area now has a struct kvmppc_vpa which is
used to manage these updates. There is also a spinlock which protects
access to all of the kvmppc_vpa structs, other than to the pinned_addr
fields. (We could have just taken the spinlock when using the vpa,
slb_shadow or dtl fields, but that would mean taking the spinlock on
every guest entry and exit.)
This also changes 'struct dtl' (which was undefined) to 'struct dtl_entry',
which is what the rest of the kernel uses.
Thanks to Michael Ellerman <michael@ellerman.id.au> for pointing out
the need to initialize vcpu->arch.vpa_update_lock.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
For BookE HV the guest visible MSR is shared->msr and is identical to
the MSR that is in use while the guest is running, because we can't trap
reads from/to MSR.
So shadow_msr is unused there. Indicate that with a comment.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
Chips such as e500mc that implement category E.HV in Power ISA 2.06
provide hardware virtualization features, including a new MSR mode for
guest state. The guest OS can perform many operations without trapping
into the hypervisor, including transitions to and from guest userspace.
Since we can use SRR1[GS] to reliably tell whether an exception came from
guest state, instead of messing around with IVPR, we use DO_KVM similarly
to book3s.
Current issues include:
- Machine checks from guest state are not routed to the host handler.
- The guest can cause a host oops by executing an emulated instruction
in a page that lacks read permission. Existing e500/4xx support has
the same problem.
Includes work by Ashish Kalra <Ashish.Kalra@freescale.com>,
Varun Sethi <Varun.Sethi@freescale.com>, and
Liu Yu <yu.liu@freescale.com>.
Signed-off-by: Scott Wood <scottwood@freescale.com>
[agraf: remove pt_regs usage]
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
The PID handling is e500v1/v2-specific, and is moved to e500.c.
The MMU sregs code and kvmppc_core_vcpu_translate will be shared with
e500mc, and is moved from e500.c to e500_tlb.c.
Partially based on patches from Liu Yu <yu.liu@freescale.com>.
Signed-off-by: Scott Wood <scottwood@freescale.com>
[agraf: fix bisectability]
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
On PowerPC, we sometimes use a waitqueue per core, not per thread,
so we can't always use the vcpu internal waitqueue.
This code has been generalized by Christoffer Dall recently, but
unfortunately broke compilation for PowerPC. At the time the helper
function is defined, struct kvm_vcpu is not declared yet, so we can't
dereference it.
This patch moves all logic into the generic inline function, at which
time we have all information necessary.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
The kvm_vcpu_kick function performs roughly the same funcitonality on
most all architectures, so we shouldn't have separate copies.
PowerPC keeps a pointer to interchanging waitqueues on the vcpu_arch
structure and to accomodate this special need a
__KVM_HAVE_ARCH_VCPU_GET_WQ define and accompanying function
kvm_arch_vcpu_wq have been defined. For all other architectures this
is a generic inline that just returns &vcpu->wq;
Acked-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Christoffer Dall <c.dall@virtualopensystems.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Some members of kvm_memory_slot are not used by every architecture.
This patch is the first step to make this difference clear by
introducing kvm_memory_slot::arch; lpage_info is moved into it.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
We're currently allocating 16MB of linear memory on demand when creating
a guest. That does work some times, but finding 16MB of linear memory
available in the system at runtime is definitely not a given.
So let's add another command line option similar to the RMA preallocator,
that we can use to keep a pool of page tables around. Now, when a guest
gets created it has a pretty low chance of receiving an OOM.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
We have code to allocate big chunks of linear memory on bootup for later use.
This code is currently used for RMA allocation, but can be useful beyond that
extent.
Make it generic so we can reuse it for other stuff later.
Signed-off-by: Alexander Graf <agraf@suse.de>
Acked-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Avi Kivity <avi@redhat.com>
We need the KVM_REG namespace for generic register settings now, so
let's rename the existing users to something different, enabling
us to reuse the namespace for more visible interfaces.
While at it, also move these private constants to a private header.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This allows both the guest and the host to use the referenced (R) and
changed (C) bits in the guest hashed page table. The guest has a view
of R and C that is maintained in the guest_rpte field of the revmap
entry for the HPTE, and the host has a view that is maintained in the
rmap entry for the associated gfn.
Both view are updated from the guest HPT. If a bit (R or C) is zero
in either view, it will be initially set to zero in the HPTE (or HPTEs),
until set to 1 by hardware. When an HPTE is removed for any reason,
the R and C bits from the HPTE are ORed into both views. We have to
be careful to read the R and C bits from the HPTE after invalidating
it, but before unlocking it, in case of any late updates by the hardware.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This adds the infrastructure to enable us to page out pages underneath
a Book3S HV guest, on processors that support virtualized partition
memory, that is, POWER7. Instead of pinning all the guest's pages,
we now look in the host userspace Linux page tables to find the
mapping for a given guest page. Then, if the userspace Linux PTE
gets invalidated, kvm_unmap_hva() gets called for that address, and
we replace all the guest HPTEs that refer to that page with absent
HPTEs, i.e. ones with the valid bit clear and the HPTE_V_ABSENT bit
set, which will cause an HDSI when the guest tries to access them.
Finally, the page fault handler is extended to reinstantiate the
guest HPTE when the guest tries to access a page which has been paged
out.
Since we can't intercept the guest DSI and ISI interrupts on PPC970,
we still have to pin all the guest pages on PPC970. We have a new flag,
kvm->arch.using_mmu_notifiers, that indicates whether we can page
guest pages out. If it is not set, the MMU notifier callbacks do
nothing and everything operates as before.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This provides the low-level support for MMIO emulation in Book3S HV
guests. When the guest tries to map a page which is not covered by
any memslot, that page is taken to be an MMIO emulation page. Instead
of inserting a valid HPTE, we insert an HPTE that has the valid bit
clear but another hypervisor software-use bit set, which we call
HPTE_V_ABSENT, to indicate that this is an absent page. An
absent page is treated much like a valid page as far as guest hcalls
(H_ENTER, H_REMOVE, H_READ etc.) are concerned, except of course that
an absent HPTE doesn't need to be invalidated with tlbie since it
was never valid as far as the hardware is concerned.
When the guest accesses a page for which there is an absent HPTE, it
will take a hypervisor data storage interrupt (HDSI) since we now set
the VPM1 bit in the LPCR. Our HDSI handler for HPTE-not-present faults
looks up the hash table and if it finds an absent HPTE mapping the
requested virtual address, will switch to kernel mode and handle the
fault in kvmppc_book3s_hv_page_fault(), which at present just calls
kvmppc_hv_emulate_mmio() to set up the MMIO emulation.
This is based on an earlier patch by Benjamin Herrenschmidt, but since
heavily reworked.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This expands the reverse mapping array to contain two links for each
HPTE which are used to link together HPTEs that correspond to the
same guest logical page. Each circular list of HPTEs is pointed to
by the rmap array entry for the guest logical page, pointed to by
the relevant memslot. Links are 32-bit HPT entry indexes rather than
full 64-bit pointers, to save space. We use 3 of the remaining 32
bits in the rmap array entries as a lock bit, a referenced bit and
a present bit (the present bit is needed since HPTE index 0 is valid).
The bit lock for the rmap chain nests inside the HPTE lock bit.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This provides for the case where userspace maps an I/O device into the
address range of a memory slot using a VM_PFNMAP mapping. In that
case, we work out the pfn from vma->vm_pgoff, and record the cache
enable bits from vma->vm_page_prot in two low-order bits in the
slot_phys array entries. Then, in kvmppc_h_enter() we check that the
cache bits in the HPTE that the guest wants to insert match the cache
bits in the slot_phys array entry. However, we do allow the guest to
create what it thinks is a non-cacheable or write-through mapping to
memory that is actually cacheable, so that we can use normal system
memory as part of an emulated device later on. In that case the actual
HPTE we insert is a cacheable HPTE.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This relaxes the requirement that the guest memory be provided as
16MB huge pages, allowing it to be provided as normal memory, i.e.
in pages of PAGE_SIZE bytes (4k or 64k). To allow this, we index
the kvm->arch.slot_phys[] arrays with a small page index, even if
huge pages are being used, and use the low-order 5 bits of each
entry to store the order of the enclosing page with respect to
normal pages, i.e. log_2(enclosing_page_size / PAGE_SIZE).
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This removes the code from kvmppc_core_prepare_memory_region() that
looked up the VMA for the region being added and called hva_to_page
to get the pfns for the memory. We have no guarantee that there will
be anything mapped there at the time of the KVM_SET_USER_MEMORY_REGION
ioctl call; userspace can do that ioctl and then map memory into the
region later.
Instead we defer looking up the pfn for each memory page until it is
needed, which generally means when the guest does an H_ENTER hcall on
the page. Since we can't call get_user_pages in real mode, if we don't
already have the pfn for the page, kvmppc_h_enter() will return
H_TOO_HARD and we then call kvmppc_virtmode_h_enter() once we get back
to kernel context. That calls kvmppc_get_guest_page() to get the pfn
for the page, and then calls back to kvmppc_h_enter() to redo the HPTE
insertion.
When the first vcpu starts executing, we need to have the RMO or VRMA
region mapped so that the guest's real mode accesses will work. Thus
we now have a check in kvmppc_vcpu_run() to see if the RMO/VRMA is set
up and if not, call kvmppc_hv_setup_rma(). It checks if the memslot
starting at guest physical 0 now has RMO memory mapped there; if so it
sets it up for the guest, otherwise on POWER7 it sets up the VRMA.
The function that does that, kvmppc_map_vrma, is now a bit simpler,
as it calls kvmppc_virtmode_h_enter instead of creating the HPTE itself.
Since we are now potentially updating entries in the slot_phys[]
arrays from multiple vcpu threads, we now have a spinlock protecting
those updates to ensure that we don't lose track of any references
to pages.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This allocates an array for each memory slot that is added to store
the physical addresses of the pages in the slot. This array is
vmalloc'd and accessed in kvmppc_h_enter using real_vmalloc_addr().
This allows us to remove the ram_pginfo field from the kvm_arch
struct, and removes the 64GB guest RAM limit that we had.
We use the low-order bits of the array entries to store a flag
indicating that we have done get_page on the corresponding page,
and therefore need to call put_page when we are finished with the
page. Currently this is set for all pages except those in our
special RMO regions.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This adds an array that parallels the guest hashed page table (HPT),
that is, it has one entry per HPTE, used to store the guest's view
of the second doubleword of the corresponding HPTE. The first
doubleword in the HPTE is the same as the guest's idea of it, so we
don't need to store a copy, but the second doubleword in the HPTE has
the real page number rather than the guest's logical page number.
This allows us to remove the back_translate() and reverse_xlate()
functions.
This "reverse mapping" array is vmalloc'd, meaning that to access it
in real mode we have to walk the kernel's page tables explicitly.
That is done by the new real_vmalloc_addr() function. (In fact this
returns an address in the linear mapping, so the result is usable
both in real mode and in virtual mode.)
There are also some minor cleanups here: moving the definitions of
HPT_ORDER etc. to a header file and defining HPT_NPTE for HPT_NPTEG << 3.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
Decrementers are now properly driven by TCR/TSR, and the guest
has full read/write access to these registers.
The decrementer keeps ticking (and setting the TSR bit) regardless of
whether the interrupts are enabled with TCR.
The decrementer stops at zero, rather than going negative.
Decrementers (and FITs, once implemented) are delivered as
level-triggered interrupts -- dequeued when the TSR bit is cleared, not
on delivery.
Signed-off-by: Liu Yu <yu.liu@freescale.com>
[scottwood@freescale.com: significant changes]
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
This allows additional registers to be accessed by the guest
in PR-mode KVM without trapping.
SPRG4-7 are readable from userspace. On booke, KVM will sync
these registers when it enters the guest, so that accesses from
guest userspace will work. The guest kernel, OTOH, must consistently
use either the real registers or the shared area between exits. This
also applies to the already-paravirted SPRG3.
On non-booke, it's not clear to what extent SPRG4-7 are supported
(they're not architected for book3s, but exist on at least some classic
chips). They are copied in the get/set regs ioctls, but I do not see any
non-booke emulation. I also do not see any syncing with real registers
(in PR-mode) including the user-readable SPRG3. This patch should not
make that situation any worse.
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
With a KVM guest operating in SMT4 mode (i.e. 4 hardware threads per
core), whenever a CPU goes idle, we have to pull all the other
hardware threads in the core out of the guest, because the H_CEDE
hcall is handled in the kernel. This is inefficient.
This adds code to book3s_hv_rmhandlers.S to handle the H_CEDE hcall
in real mode. When a guest vcpu does an H_CEDE hcall, we now only
exit to the kernel if all the other vcpus in the same core are also
idle. Otherwise we mark this vcpu as napping, save state that could
be lost in nap mode (mainly GPRs and FPRs), and execute the nap
instruction. When the thread wakes up, because of a decrementer or
external interrupt, we come back in at kvm_start_guest (from the
system reset interrupt vector), find the `napping' flag set in the
paca, and go to the resume path.
This has some other ramifications. First, when starting a core, we
now start all the threads, both those that are immediately runnable and
those that are idle. This is so that we don't have to pull all the
threads out of the guest when an idle thread gets a decrementer interrupt
and wants to start running. In fact the idle threads will all start
with the H_CEDE hcall returning; being idle they will just do another
H_CEDE immediately and go to nap mode.
This required some changes to kvmppc_run_core() and kvmppc_run_vcpu().
These functions have been restructured to make them simpler and clearer.
We introduce a level of indirection in the wait queue that gets woken
when external and decrementer interrupts get generated for a vcpu, so
that we can have the 4 vcpus in a vcore using the same wait queue.
We need this because the 4 vcpus are being handled by one thread.
Secondly, when we need to exit from the guest to the kernel, we now
have to generate an IPI for any napping threads, because an HDEC
interrupt doesn't wake up a napping thread.
Thirdly, we now need to be able to handle virtual external interrupts
and decrementer interrupts becoming pending while a thread is napping,
and deliver those interrupts to the guest when the thread wakes.
This is done in kvmppc_cede_reentry, just before fast_guest_return.
Finally, since we are not using the generic kvm_vcpu_block for book3s_hv,
and hence not calling kvm_arch_vcpu_runnable, we can remove the #ifdef
from kvm_arch_vcpu_runnable.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
This simplifies the way that the book3s_pr makes the transition to
real mode when entering the guest. We now call kvmppc_entry_trampoline
(renamed from kvmppc_rmcall) in the base kernel using a normal function
call instead of doing an indirect call through a pointer in the vcpu.
If kvm is a module, the module loader takes care of generating a
trampoline as it does for other calls to functions outside the module.
kvmppc_entry_trampoline then disables interrupts and jumps to
kvmppc_handler_trampoline_enter in real mode using an rfi[d].
That then uses the link register as the address to return to
(potentially in module space) when the guest exits.
This also simplifies the way that we call the Linux interrupt handler
when we exit the guest due to an external, decrementer or performance
monitor interrupt. Instead of turning on the MMU, then deciding that
we need to call the Linux handler and turning the MMU back off again,
we now go straight to the handler at the point where we would turn the
MMU on. The handler will then return to the virtual-mode code
(potentially in the module).
Along the way, this moves the setting and clearing of the HID5 DCBZ32
bit into real-mode interrupts-off code, and also makes sure that
we clear the MSR[RI] bit before loading values into SRR0/1.
The net result is that we no longer need any code addresses to be
stored in vcpu->arch.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
There are multiple features in PowerPC KVM that can now be enabled
depending on the user's wishes. Some of the combinations don't make
sense or don't work though.
So this patch adds a way to check if the executing environment would
actually be able to run the guest properly. It also adds sanity
checks if PVR is set (should always be true given the current code
flow), if PAPR is only used with book3s_64 where it works and that
HV KVM is only used in PAPR mode.
Signed-off-by: Alexander Graf <agraf@suse.de>
When running a PAPR guest, some things change. The privilege level drops
from hypervisor to supervisor, SDR1 gets treated differently and we interpret
hypercalls. For bisectability sake, add the flag now, but only enable it when
all the support code is there.
Signed-off-by: Alexander Graf <agraf@suse.de>
This adds support for running KVM guests in supervisor mode on those
PPC970 processors that have a usable hypervisor mode. Unfortunately,
Apple G5 machines have supervisor mode disabled (MSR[HV] is forced to
1), but the YDL PowerStation does have a usable hypervisor mode.
There are several differences between the PPC970 and POWER7 in how
guests are managed. These differences are accommodated using the
CPU_FTR_ARCH_201 (PPC970) and CPU_FTR_ARCH_206 (POWER7) CPU feature
bits. Notably, on PPC970:
* The LPCR, LPID or RMOR registers don't exist, and the functions of
those registers are provided by bits in HID4 and one bit in HID0.
* External interrupts can be directed to the hypervisor, but unlike
POWER7 they are masked by MSR[EE] in non-hypervisor modes and use
SRR0/1 not HSRR0/1.
* There is no virtual RMA (VRMA) mode; the guest must use an RMO
(real mode offset) area.
* The TLB entries are not tagged with the LPID, so it is necessary to
flush the whole TLB on partition switch. Furthermore, when switching
partitions we have to ensure that no other CPU is executing the tlbie
or tlbsync instructions in either the old or the new partition,
otherwise undefined behaviour can occur.
* The PMU has 8 counters (PMC registers) rather than 6.
* The DSCR, PURR, SPURR, AMR, AMOR, UAMOR registers don't exist.
* The SLB has 64 entries rather than 32.
* There is no mediated external interrupt facility, so if we switch to
a guest that has a virtual external interrupt pending but the guest
has MSR[EE] = 0, we have to arrange to have an interrupt pending for
it so that we can get control back once it re-enables interrupts. We
do that by sending ourselves an IPI with smp_send_reschedule after
hard-disabling interrupts.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
This adds infrastructure which will be needed to allow book3s_hv KVM to
run on older POWER processors, including PPC970, which don't support
the Virtual Real Mode Area (VRMA) facility, but only the Real Mode
Offset (RMO) facility. These processors require a physically
contiguous, aligned area of memory for each guest. When the guest does
an access in real mode (MMU off), the address is compared against a
limit value, and if it is lower, the address is ORed with an offset
value (from the Real Mode Offset Register (RMOR)) and the result becomes
the real address for the access. The size of the RMA has to be one of
a set of supported values, which usually includes 64MB, 128MB, 256MB
and some larger powers of 2.
Since we are unlikely to be able to allocate 64MB or more of physically
contiguous memory after the kernel has been running for a while, we
allocate a pool of RMAs at boot time using the bootmem allocator. The
size and number of the RMAs can be set using the kvm_rma_size=xx and
kvm_rma_count=xx kernel command line options.
KVM exports a new capability, KVM_CAP_PPC_RMA, to signal the availability
of the pool of preallocated RMAs. The capability value is 1 if the
processor can use an RMA but doesn't require one (because it supports
the VRMA facility), or 2 if the processor requires an RMA for each guest.
This adds a new ioctl, KVM_ALLOCATE_RMA, which allocates an RMA from the
pool and returns a file descriptor which can be used to map the RMA. It
also returns the size of the RMA in the argument structure.
Having an RMA means we will get multiple KMV_SET_USER_MEMORY_REGION
ioctl calls from userspace. To cope with this, we now preallocate the
kvm->arch.ram_pginfo array when the VM is created with a size sufficient
for up to 64GB of guest memory. Subsequently we will get rid of this
array and use memory associated with each memslot instead.
This moves most of the code that translates the user addresses into
host pfns (page frame numbers) out of kvmppc_prepare_vrma up one level
to kvmppc_core_prepare_memory_region. Also, instead of having to look
up the VMA for each page in order to check the page size, we now check
that the pages we get are compound pages of 16MB. However, if we are
adding memory that is mapped to an RMA, we don't bother with calling
get_user_pages_fast and instead just offset from the base pfn for the
RMA.
Typically the RMA gets added after vcpus are created, which makes it
inconvenient to have the LPCR (logical partition control register) value
in the vcpu->arch struct, since the LPCR controls whether the processor
uses RMA or VRMA for the guest. This moves the LPCR value into the
kvm->arch struct and arranges for the MER (mediated external request)
bit, which is the only bit that varies between vcpus, to be set in
assembly code when going into the guest if there is a pending external
interrupt request.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
This lifts the restriction that book3s_hv guests can only run one
hardware thread per core, and allows them to use up to 4 threads
per core on POWER7. The host still has to run single-threaded.
This capability is advertised to qemu through a new KVM_CAP_PPC_SMT
capability. The return value of the ioctl querying this capability
is the number of vcpus per virtual CPU core (vcore), currently 4.
To use this, the host kernel should be booted with all threads
active, and then all the secondary threads should be offlined.
This will put the secondary threads into nap mode. KVM will then
wake them from nap mode and use them for running guest code (while
they are still offline). To wake the secondary threads, we send
them an IPI using a new xics_wake_cpu() function, implemented in
arch/powerpc/sysdev/xics/icp-native.c. In other words, at this stage
we assume that the platform has a XICS interrupt controller and
we are using icp-native.c to drive it. Since the woken thread will
need to acknowledge and clear the IPI, we also export the base
physical address of the XICS registers using kvmppc_set_xics_phys()
for use in the low-level KVM book3s code.
When a vcpu is created, it is assigned to a virtual CPU core.
The vcore number is obtained by dividing the vcpu number by the
number of threads per core in the host. This number is exported
to userspace via the KVM_CAP_PPC_SMT capability. If qemu wishes
to run the guest in single-threaded mode, it should make all vcpu
numbers be multiples of the number of threads per core.
We distinguish three states of a vcpu: runnable (i.e., ready to execute
the guest), blocked (that is, idle), and busy in host. We currently
implement a policy that the vcore can run only when all its threads
are runnable or blocked. This way, if a vcpu needs to execute elsewhere
in the kernel or in qemu, it can do so without being starved of CPU
by the other vcpus.
When a vcore starts to run, it executes in the context of one of the
vcpu threads. The other vcpu threads all go to sleep and stay asleep
until something happens requiring the vcpu thread to return to qemu,
or to wake up to run the vcore (this can happen when another vcpu
thread goes from busy in host state to blocked).
It can happen that a vcpu goes from blocked to runnable state (e.g.
because of an interrupt), and the vcore it belongs to is already
running. In that case it can start to run immediately as long as
the none of the vcpus in the vcore have started to exit the guest.
We send the next free thread in the vcore an IPI to get it to start
to execute the guest. It synchronizes with the other threads via
the vcore->entry_exit_count field to make sure that it doesn't go
into the guest if the other vcpus are exiting by the time that it
is ready to actually enter the guest.
Note that there is no fixed relationship between the hardware thread
number and the vcpu number. Hardware threads are assigned to vcpus
as they become runnable, so we will always use the lower-numbered
hardware threads in preference to higher-numbered threads if not all
the vcpus in the vcore are runnable, regardless of which vcpus are
runnable.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
This improves I/O performance for guests using the PAPR
paravirtualization interface by making the H_PUT_TCE hcall faster, by
implementing it in real mode. H_PUT_TCE is used for updating virtual
IOMMU tables, and is used both for virtual I/O and for real I/O in the
PAPR interface.
Since this moves the IOMMU tables into the kernel, we define a new
KVM_CREATE_SPAPR_TCE ioctl to allow qemu to create the tables. The
ioctl returns a file descriptor which can be used to mmap the newly
created table. The qemu driver models use them in the same way as
userspace managed tables, but they can be updated directly by the
guest with a real-mode H_PUT_TCE implementation, reducing the number
of host/guest context switches during guest IO.
There are certain circumstances where it is useful for userland qemu
to write to the TCE table even if the kernel H_PUT_TCE path is used
most of the time. Specifically, allowing this will avoid awkwardness
when we need to reset the table. More importantly, we will in the
future need to write the table in order to restore its state after a
checkpoint resume or migration.
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
This adds the infrastructure for handling PAPR hcalls in the kernel,
either early in the guest exit path while we are still in real mode,
or later once the MMU has been turned back on and we are in the full
kernel context. The advantage of handling hcalls in real mode if
possible is that we avoid two partition switches -- and this will
become more important when we support SMT4 guests, since a partition
switch means we have to pull all of the threads in the core out of
the guest. The disadvantage is that we can only access the kernel
linear mapping, not anything vmalloced or ioremapped, since the MMU
is off.
This also adds code to handle the following hcalls in real mode:
H_ENTER Add an HPTE to the hashed page table
H_REMOVE Remove an HPTE from the hashed page table
H_READ Read HPTEs from the hashed page table
H_PROTECT Change the protection bits in an HPTE
H_BULK_REMOVE Remove up to 4 HPTEs from the hashed page table
H_SET_DABR Set the data address breakpoint register
Plus code to handle the following hcalls in the kernel:
H_CEDE Idle the vcpu until an interrupt or H_PROD hcall arrives
H_PROD Wake up a ceded vcpu
H_REGISTER_VPA Register a virtual processor area (VPA)
The code that runs in real mode has to be in the base kernel, not in
the module, if KVM is compiled as a module. The real-mode code can
only access the kernel linear mapping, not vmalloc or ioremap space.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
This adds support for KVM running on 64-bit Book 3S processors,
specifically POWER7, in hypervisor mode. Using hypervisor mode means
that the guest can use the processor's supervisor mode. That means
that the guest can execute privileged instructions and access privileged
registers itself without trapping to the host. This gives excellent
performance, but does mean that KVM cannot emulate a processor
architecture other than the one that the hardware implements.
This code assumes that the guest is running paravirtualized using the
PAPR (Power Architecture Platform Requirements) interface, which is the
interface that IBM's PowerVM hypervisor uses. That means that existing
Linux distributions that run on IBM pSeries machines will also run
under KVM without modification. In order to communicate the PAPR
hypercalls to qemu, this adds a new KVM_EXIT_PAPR_HCALL exit code
to include/linux/kvm.h.
Currently the choice between book3s_hv support and book3s_pr support
(i.e. the existing code, which runs the guest in user mode) has to be
made at kernel configuration time, so a given kernel binary can only
do one or the other.
This new book3s_hv code doesn't support MMIO emulation at present.
Since we are running paravirtualized guests, this isn't a serious
restriction.
With the guest running in supervisor mode, most exceptions go straight
to the guest. We will never get data or instruction storage or segment
interrupts, alignment interrupts, decrementer interrupts, program
interrupts, single-step interrupts, etc., coming to the hypervisor from
the guest. Therefore this introduces a new KVMTEST_NONHV macro for the
exception entry path so that we don't have to do the KVM test on entry
to those exception handlers.
We do however get hypervisor decrementer, hypervisor data storage,
hypervisor instruction storage, and hypervisor emulation assist
interrupts, so we have to handle those.
In hypervisor mode, real-mode accesses can access all of RAM, not just
a limited amount. Therefore we put all the guest state in the vcpu.arch
and use the shadow_vcpu in the PACA only for temporary scratch space.
We allocate the vcpu with kzalloc rather than vzalloc, and we don't use
anything in the kvmppc_vcpu_book3s struct, so we don't allocate it.
We don't have a shared page with the guest, but we still need a
kvm_vcpu_arch_shared struct to store the values of various registers,
so we include one in the vcpu_arch struct.
The POWER7 processor has a restriction that all threads in a core have
to be in the same partition. MMU-on kernel code counts as a partition
(partition 0), so we have to do a partition switch on every entry to and
exit from the guest. At present we require the host and guest to run
in single-thread mode because of this hardware restriction.
This code allocates a hashed page table for the guest and initializes
it with HPTEs for the guest's Virtual Real Memory Area (VRMA). We
require that the guest memory is allocated using 16MB huge pages, in
order to simplify the low-level memory management. This also means that
we can get away without tracking paging activity in the host for now,
since huge pages can't be paged or swapped.
This also adds a few new exports needed by the book3s_hv code.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
This moves the slb field, which represents the state of the emulated
SLB, from the kvmppc_vcpu_book3s struct to the kvm_vcpu_arch, and the
hpte_hash_[v]pte[_long] fields from kvm_vcpu_arch to kvmppc_vcpu_book3s.
This is in accord with the principle that the kvm_vcpu_arch struct
represents the state of the emulated CPU, and the kvmppc_vcpu_book3s
struct holds the auxiliary data structures used in the emulation.
Signed-off-by: Paul Mackerras <paulus@samba.org>
Signed-off-by: Alexander Graf <agraf@suse.de>
Dynamically assign host PIDs to guest PIDs, splitting each guest PID into
multiple host (shadow) PIDs based on kernel/user and MSR[IS/DS]. Use
both PID0 and PID1 so that the shadow PIDs for the right mode can be
selected, that correspond both to guest TID = zero and guest TID = guest
PID.
This allows us to significantly reduce the frequency of needing to
invalidate the entire TLB. When the guest mode or PID changes, we just
update the host PID0/PID1. And since the allocation of shadow PIDs is
global, multiple guests can share the TLB without conflict.
Note that KVM does not yet support the guest setting PID1 or PID2 to
a value other than zero. This will need to be fixed for nested KVM
to work. Until then, we enforce the requirement for guest PID1/PID2
to stay zero by failing the emulation if the guest tries to set them
to something else.
Signed-off-by: Liu Yu <yu.liu@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
This is done lazily. The SPE save will be done only if the guest has
used SPE since the last preemption or heavyweight exit. Restore will be
done only on demand, when enabling MSR_SPE in the shadow MSR, in response
to an SPE fault or mtmsr emulation.
For SPEFSCR, Linux already switches it on context switch (non-lazily), so
the only remaining bit is to save it between qemu and the guest.
Signed-off-by: Liu Yu <yu.liu@freescale.com>
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Keep the guest MSR and the guest-mode true MSR separate, rather than
modifying the guest MSR on each guest entry to produce a true MSR.
Any bits which should be modified based on guest MSR must be explicitly
propagated from vcpu->arch.shared->msr to vcpu->arch.shadow_msr in
kvmppc_set_msr().
While we're modifying the guest entry code, reorder a few instructions
to bury some load latencies.
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Linux doesn't use USPRG0 (now renamed VRSAVE in the architecture, even
when Altivec isn't involved), but a guest might.
Signed-off-by: Scott Wood <scottwood@freescale.com>
Signed-off-by: Alexander Graf <agraf@suse.de>
Following dump is observed on host when clearing the exit timing counters
[root@p1021mds kvm]# echo -n 'c' > vm1200_vcpu0_timing
INFO: task echo:1276 blocked for more than 120 seconds.
"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
echo D 0ff5bf94 0 1276 1190 0x00000000
Call Trace:
[c2157e40] [c0007908] __switch_to+0x9c/0xc4
[c2157e50] [c040293c] schedule+0x1b4/0x3bc
[c2157e90] [c04032dc] __mutex_lock_slowpath+0x74/0xc0
[c2157ec0] [c00369e4] kvmppc_init_timing_stats+0x20/0xb8
[c2157ed0] [c0036b00] kvmppc_exit_timing_write+0x84/0x98
[c2157ef0] [c00b9f90] vfs_write+0xc0/0x16c
[c2157f10] [c00ba284] sys_write+0x4c/0x90
[c2157f40] [c000e320] ret_from_syscall+0x0/0x3c
The vcpu->mutex is used by kvm_ioctl_* (KVM_RUN etc) and same was
used when clearing the stats (in kvmppc_init_timing_stats()). What happens
is that when the guest is idle then it held the vcpu->mutx. While the
exiting timing process waits for guest to release the vcpu->mutex and
a hang state is reached.
Now using seprate lock for exit timing stats.
Signed-off-by: Bharat Bhushan <Bharat.Bhushan@freescale.com>
Acked-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
On Book3s_32 the tlbie instruction flushed effective addresses by the mask
0x0ffff000. This is pretty hard to reflect with a hash that hashes ~0xfff, so
to speed up that target we should also keep a special hash around for it.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
So far we've been running all code without locking of any sort. This wasn't
really an issue because I didn't see any parallel access to the shadow MMU
code coming.
But then I started to implement dirty bitmapping to MOL which has the video
code in its own thread, so suddenly we had the dirty bitmap code run in
parallel to the shadow mmu code. And with that came trouble.
So I went ahead and made the MMU modifying functions as parallelizable as
I could think of. I hope I didn't screw up too much RCU logic :-). If you
know your way around RCU and locking and what needs to be done when, please
take a look at this patch.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
We will be introducing a method to project the shared page in guest context.
As soon as we're talking about this coupling, the shared page is colled magic
page.
This patch introduces simple defines, so the follow-up patches are easier to
read.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
On PowerPC it's very normal to not support all of the physical RAM in real mode.
To check if we're matching on the shared page or not, we need to know the limits
so we can restrain ourselves to that range.
So let's make it a define instead of open-coding it. And while at it, let's also
increase it.
Signed-off-by: Alexander Graf <agraf@suse.de>
v2 -> v3:
- RMO -> PAM (non-magic page)
Signed-off-by: Avi Kivity <avi@redhat.com>
When in kernel mode there are 4 additional registers available that are
simple data storage. Instead of exiting to the hypervisor to read and
write those, we can just share them with the guest using the page.
This patch converts all users of the current field to the shared page.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
The SRR0 and SRR1 registers contain cached values of the PC and MSR
respectively. They get written to by the hypervisor when an interrupt
occurs or directly by the kernel. They are also used to tell the rfi(d)
instruction where to jump to.
Because it only gets touched on defined events that, it's very simple to
share with the guest. Hypervisor and guest both have full r/w access.
This patch converts all users of the current field to the shared page.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
