What it is: vhost net is a character device that can be used to reduce
the number of system calls involved in virtio networking.
Existing virtio net code is used in the guest without modification.
There's similarity with vringfd, with some differences and reduced scope
- uses eventfd for signalling
- structures can be moved around in memory at any time (good for
migration, bug work-arounds in userspace)
- write logging is supported (good for migration)
- support memory table and not just an offset (needed for kvm)
common virtio related code has been put in a separate file vhost.c and
can be made into a separate module if/when more backends appear. I used
Rusty's lguest.c as the source for developing this part : this supplied
me with witty comments I wouldn't be able to write myself.
What it is not: vhost net is not a bus, and not a generic new system
call. No assumptions are made on how guest performs hypercalls.
Userspace hypervisors are supported as well as kvm.
How it works: Basically, we connect virtio frontend (configured by
userspace) to a backend. The backend could be a network device, or a tap
device. Backend is also configured by userspace, including vlan/mac
etc.
Status: This works for me, and I haven't see any crashes.
Compared to userspace, people reported improved latency (as I save up to
4 system calls per packet), as well as better bandwidth and CPU
utilization.
Features that I plan to look at in the future:
- mergeable buffers
- zero copy
- scalability tuning: figure out the best threading model to use
Note on RCU usage (this is also documented in vhost.h, near
private_pointer which is the value protected by this variant of RCU):
what is happening is that the rcu_dereference() is being used in a
workqueue item. The role of rcu_read_lock() is taken on by the start of
execution of the workqueue item, of rcu_read_unlock() by the end of
execution of the workqueue item, and of synchronize_rcu() by
flush_workqueue()/flush_work(). In the future we might need to apply
some gcc attribute or sparse annotation to the function passed to
INIT_WORK(). Paul's ack below is for this RCU usage.
(Includes fixes by Alan Cox <alan@linux.intel.com>,
David L Stevens <dlstevens@us.ibm.com>,
Chris Wright <chrisw@redhat.com>)
Acked-by: Rusty Russell <rusty@rustcorp.com.au>
Acked-by: Arnd Bergmann <arnd@arndb.de>
Acked-by: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
We were shifting the Ks/Kp/N bits one bit too far on mtsrin. It took
me some time to figure that out, so I also put in some debugging and a
comment explaining the conversion.
This fixes current OpenBIOS boot on PPC64 KVM.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Avi Kivity <avi@redhat.com>
Currently userspace has no chance to find out which virtual address space we're
in and resolve addresses. While that is a big problem for migration, it's also
unpleasent when debugging, as gdb and the monitor don't work on virtual
addresses.
This patch exports enough of the MMU segment state to userspace to make
debugging work and thus also includes the groundwork for migration.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
The old BUILD_BUG_ON implementation didn't work with __builtin_constant_p().
Fixing that revealed this test had been inverted for a long time without
anybody noticing...
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
X86 CPUs need to have some magic happening to enable the virtualization
extensions on them. This magic can result in unpleasant results for
users, like blocking other VMMs from working (vmx) or using invalid TLB
entries (svm).
Currently KVM activates virtualization when the respective kernel module
is loaded. This blocks us from autoloading KVM modules without breaking
other VMMs.
To circumvent this problem at least a bit, this patch introduces on
demand activation of virtualization. This means, that instead
virtualization is enabled on creation of the first virtual machine
and disabled on destruction of the last one.
So using this, KVM can be easily autoloaded, while keeping other
hypervisors usable.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Following S390's good example we should use hrtimers for the decrementer too!
This patch converts the timer from the old mechanism to hrtimers.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
It looks like the variable "pc" is defined. At least the current code always
failed on me stating that "pc" is already defined somewhere else.
Let's use _pc instead, because that doesn't collide.
Is this the right approach? Does it break on 440 too? If not, why not?
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Now we have everything in place to be able to build KVM, so let's add it
as config option and in the Makefile.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
To be able to keep KVM as module, we need to export the SLB trampoline
addresses to the module, so it knows where to jump to.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Little opcodes behave differently on desktop and embedded PowerPC cores.
In order to reflect those differences, let's add some #ifdef code to emulate.c.
We could probably also handle them in the core specific emulation files, but I
would prefer to reuse as much code as possible.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We support setting the DEC to a certain value right now. Doing that basically
triggers the CPU local timer.
But there's also an mfdec command that enabled the OS to read the decrementor.
This is required at least by all desktop and server PowerPC Linux kernels. It
can't really hurt to allow embedded ones to do it as well though.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
There are generic parts of PowerPC that can be shared across all
implementations and specific parts that only apply to BookE or desktop PPCs.
This patch adds emulation for desktop specific opcodes that don't apply
to BookE CPUs.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This patch adds an implementation for a G3/G4 MMU, so we can run G3 and
G4 guests in KVM on Book3s_64.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
To be able to run a guest, we also need to implement a guest MMU.
This patch adds MMU handling for Book3s_64 guests.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
We designed the Book3S port of KVM as modular as possible. Most
of the code could be easily used on a Book3S_32 host as well.
The main difference between 32 and 64 bit cores is the MMU. To keep
things well separated, we treat the book3s_64 MMU as one possible compile
option.
This patch adds all the MMU helpers the rest of the code needs in
order to modify the host's MMU, like setting PTEs and segments.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This adds the book3s core handling file. Here everything that is generic to
desktop PowerPC cores is handled, including interrupt injections, MSR settings,
etc.
It basically takes over the same role as booke.c for embedded PowerPCs.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Getting from host state to the guest is only half the story. We also need
to return to our host context and handle whatever happened to get us out of
the guest.
On PowerPC every guest exit is an interrupt. So all we need to do is trap
the host's interrupt handlers and get into our #VMEXIT code to handle it.
PowerPCs also have a register that can add an offset to the interrupt handlers'
adresses which is what the booke KVM code uses. Unfortunately that is a
hypervisor ressource and we also want to be able to run KVM when we're running
in an LPAR. So we have to hook into the Linux interrupt handlers.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This is the really low level of guest entry/exit code.
Book3s_64 has an SLB, which stores all ESID -> VSID mappings we're
currently aware of.
The segments in the guest differ from the ones on the host, so we need
to switch the SLB to tell the MMU that we're in a new context.
So we store a shadow of the guest's SLB in the PACA, switch to that on
entry and only restore bolted entries on exit, leaving the rest to the
Linux SLB fault handler.
That way we get a really clean way of switching the SLB.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
This is the of entry / exit code. In order to switch between host and guest
context, we need to switch register state and call the exit code handler on
exit.
This assembly file does exactly that. To finally enter the guest it calls
into book3s_64_slb.S. On exit it gets jumped at from book3s_64_slb.S too.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
PowerPC code handles dirty logging in the generic parts atm. While this
is great for "return -ENOTSUPP", we need to be rather target specific
when actually implementing it.
So let's split it to implementation specific code, so we can implement
it for book3s.
Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/benh/powerpc: (134 commits)
powerpc/nvram: Enable use Generic NVRAM driver for different size chips
powerpc/iseries: Fix oops reading from /proc/iSeries/mf/*/cmdline
powerpc/ps3: Workaround for flash memory I/O error
powerpc/booke: Don't set DABR on 64-bit BookE, use DAC1 instead
powerpc/perf_counters: Reduce stack usage of power_check_constraints
powerpc: Fix bug where perf_counters breaks oprofile
powerpc/85xx: Fix SMP compile error and allow NULL for smp_ops
powerpc/irq: Improve nanodoc
powerpc: Fix some late PowerMac G5 with PCIe ATI graphics
powerpc/fsl-booke: Use HW PTE format if CONFIG_PTE_64BIT
powerpc/book3e: Add missing page sizes
powerpc/pseries: Fix to handle slb resize across migration
powerpc/powermac: Thermal control turns system off too eagerly
powerpc/pci: Merge ppc32 and ppc64 versions of phb_scan()
powerpc/405ex: support cuImage via included dtb
powerpc/405ex: provide necessary fixup function to support cuImage
powerpc/40x: Add support for the ESTeem 195E (PPC405EP) SBC
powerpc/44x: Add Eiger AMCC (AppliedMicro) PPC460SX evaluation board support.
powerpc/44x: Update Arches defconfig
powerpc/44x: Update Arches dts
...
Fix up conflicts in drivers/char/agp/uninorth-agp.c
Remove kvm_cpu_has_interrupt() and kvm_arch_interrupt_allowed() from
interface between general code and arch code. kvm_arch_vcpu_runnable()
checks for interrupts instead.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Fixes a couple of warnings like this one:
WARNING: arch/powerpc/kvm/kvm-440.o(.text+0x1e8c): Section mismatch in reference from the function kvmppc_44x_exit() to the function .exit.text:kvmppc_booke_exit()
The function kvmppc_44x_exit() references a function in an exit section.
Often the function kvmppc_booke_exit() has valid usage outside the exit section
and the fix is to remove the __exit annotation of kvmppc_booke_exit.
Also add some __init annotations on obvious routines.
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
Signed-off-by: Avi Kivity <avi@redhat.com>
The kernel uses SPRG registers for various purposes, typically in
low level assembly code as scratch registers or to hold per-cpu
global infos such as the PACA or the current thread_info pointer.
We want to be able to easily shuffle the usage of those registers
as some implementations have specific constraints realted to some
of them, for example, some have userspace readable aliases, etc..
and the current choice isn't always the best.
This patch should not change any code generation, and replaces the
usage of SPRN_SPRGn everywhere in the kernel with a named replacement
and adds documentation next to the definition of the names as to
what those are used for on each processor family.
The only parts that still use the original numbers are bits of KVM
or suspend/resume code that just blindly needs to save/restore all
the SPRGs.
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Add the option to build the code under arch/powerpc with -Werror.
The intention is to make it harder for people to inadvertantly introduce
warnings in the arch/powerpc code. It needs to be configurable so that
if a warning is introduced, people can easily work around it while it's
being fixed.
The option is a negative, ie. don't enable -Werror, so that it will be
turned on for allyes and allmodconfig builds.
The default is n, in the hope that developers will build with -Werror,
that will probably lead to some build breaks, I am prepared to be flamed.
It's not enabled for math-emu, which is a steaming pile of warnings.
Signed-off-by: Michael Ellerman <michael@ellerman.id.au>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
kvm_vcpu_block() unhalts vpu on an interrupt/timer without checking
if interrupt window is actually opened.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
TLB entry should enable memory coherence in SMP.
And like commit 631fba9dd3aca519355322cef035730609e91593,
remove guard attribute to enable the prefetch of guest memory.
Signed-off-by: Liu Yu <yu.liu@freescale.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Should clear and then update the next victim area here.
Guest kernel only read TLB1 when startup kernel,
this bug result in an extra 4K TLB1 mapping in guest from 0x0 to 0x0.
As the problem has no impact to bootup a guest,
we didn't notice it before.
Signed-off-by: Liu Yu <yu.liu@freescale.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Two KVM archs support irqchips and two don't. Add a Kconfig item to
make selecting between the two models easier.
Signed-off-by: Avi Kivity <avi@redhat.com>
After the rewrite of KVM's debug support, this code doesn't even build any
more.
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Although BOOKE_MAX_INTERRUPT has the right value, the meaning is not match.
Signed-off-by: Liu Yu <yu.liu@freescale.com>
Signed-off-by: Hollis Blanchard <hollisb@us.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
David S. Miller