Currently our implementation of the H_ENTER hypercall, which inserts a
mapping in the hash page table assumes that only ordinary memory is ever
mapped, and only permits mapping attribute bits accordingly (WIMG==0010).

However, we intend to start adding emulated IO to the pseries platform
(and real IO with PCI passthrough on kvm) which means this simple test
will no longer suffice.

This patch extends the h_enter validation code to check if the given
address is a RAM address.  If it is it enforces WIMG==0010, otherwise
it assumes that it is an IO mapping and instead enforces WIMG=010x.
Signed-off-by: NDavid Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NAlexander Graf <agraf@suse.de>

The interrupt controller presented in the device tree for the pseries
machine is manipulated by the guest only through hypervisor calls.  It
has no real or emulated registers for the guest to access.

However, it currently has a bogus 'reg' property advertising a register
window.  Moreover, this property has an invalid format, being a 32-bit
zero, when the #address-cells property on the root bus indicates that it
needs a 64-bit address.  Since the guest never attempts to manipulate
the node directly, it works, but it is ugly and can cause warnings when
manipulating the device tree in other tools (such as future firmware
versions).

This patch, therefore, corrects the problem by entirely removing the
interrupt-controller node's 'reg' property.
Signed-off-by: NDavid Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NAlexander Graf <agraf@suse.de>

Future devices we will be adding to the pseries machine (e.g. PCI) will
need nodes in the device tree which explicitly reference the top-level
interrupt controller via interrupt-parent or interrupt-map properties.

In order to do this, the interrupt controller node needs an assigned
phandle.  This patch adds the appropriate property, in preparation.
Signed-off-by: NDavid Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NAlexander Graf <agraf@suse.de>

The implementation of the XICS interrupt controller contains several
(difficult to trigger) bugs due to the fact that we were not 100%
consistent with which irq numbering we used.  In most places, global
numbers were used as handled by the presentation layer, however a few
functions took "local" numberings, that is the source number within
the interrupt source controller which is offset from the global
number.  In most cases the function and its caller agreed on this, but
in a few cases it didn't.

This patch cleans this up by always using global numbering.
Translation to the local number is now always and only done when we
look up the individual interrupt source state structure.  This should
remove the existing bugs and with luck reduce the chances of
re-introducing such bugs.
Signed-off-by: NDavid Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NAlexander Graf <agraf@suse.de>

One of the things we can't fake on PPC is the timer speed. So
we need to extract the frequency information from the host and
put it back into the guest device tree.

Luckily, we already have functions for that from the non-pseries
targets, so all we need to do is to connect the dots and the guest
suddenly gets to know its real timer speeds.
Signed-off-by: NAlexander Graf <agraf@suse.de>

When running PR style KVM, we need to tell the kernel that we want
to run in PAPR mode now. This means that we need to pass some more
register information down and enable papr mode. We also need to align
the HTAB to htab_size boundary.

Using this patch, -M pseries works with kvm even on non-hv kvm
implementations, as long as the preceding kernel patches are in.
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

  - match on CONFIG_PSERIES

v2 -> v3:

  - remove HIOR pieces from PAPR patch (ABI breakage)

This patch updates the kvm kernel headers to the latest version.
Signed-off-by: NAlexander Graf <agraf@suse.de>

We only support -M pseries when certain prerequisites are met, such
as a PPC64 guest and libfdt. To only gather these requirements in
a single place, this patch introduces a new CONFIG_PSERIES variable
that gets set when all prerequisites are met.
Signed-off-by: NAlexander Graf <agraf@suse.de>

Now that we have everything in place, make the machine description
aware of the fact that we can now handle 15 virtual CPUs!
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

  - Max cpus is 15 because of MPIC

With this patch, we generate CPU nodes in the machine initialization, giving
us the freedom to generate as many nodes as we want and as the machine supports,
but only those.

This is a first step towards a much cleaner device tree generation
infrastructure, where we would not require precompiled dtb blobs anymore.
Signed-off-by: NAlexander Graf <agraf@suse.de>

We want to generate the CPU nodes in machine init code, so remove them from
the device tree definition that we precompile.
Signed-off-by: NAlexander Graf <agraf@suse.de>

We currently load a device tree blob and then just take its size x2 to
account for modifications we do inside. While this is nice and great,
it fails when we have a small device tree as blob and lots of nodes added
in machine init code.

So for now, just make it 20k bigger than it was before. We maybe want to
be more clever about this later.
Signed-off-by: NAlexander Graf <agraf@suse.de>

When we screw up and issue an FDT command that doesn't work, we really need to
know immediately and usually can't continue to create the machine. To make sure
we don't need to add error checking in all device tree modification code users,
we can just add the fail checks to the qemu abstract functions.
Signed-off-by: NAlexander Graf <agraf@suse.de>

We want to be able to create subnodes in our device tree, so export it through
the qemu device tree abstraction framework.
Signed-off-by: NAlexander Graf <agraf@suse.de>

The guest OS wants to know where the guest spins, so let's tell him while
updating the CPU nodes with the frequencies anyways.
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

  - use new spin table address

CPUs that are not the boot CPU need to run in spinning code to check if they
should run off to execute and if so where to jump to. This usually happens
by leaving secondary CPUs looping and checking if some variable in memory
changed.

In an environment like Qemu however we can be more clever. We can just export
the spin table the primary CPU modifies as MMIO region that would event based
wake up the respective secondary CPUs. That saves us quite some cycles while
the secondary CPUs are not up yet.

So this patch adds a PV device that simply exports the spinning table into the
guest and thus allows the primary CPU to wake up secondary ones.
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

  - change into MMIO scheme
  - map the secondary NIP instead of 0 1:1
  - only map 64MB for TLB, same as u-boot
  - prepare code for 64-bit spinnings

v2 -> v3:

  - remove r6
  - set MAS2_M
  - map EA 0
  - use second TLB1 entry

v3 -> v4:

  - change to memoryops

v4 -> v5:

  - fix endianness bugs

v5 -> v6:

  - add header

We should only keep CPU nodes in the device tree around that we really have
virtual CPUs for. So remove all superfluous entries that we just keep there
in case someone wants to create a lot of vCPUs.
Signed-off-by: NAlexander Graf <agraf@suse.de>

Now that we can so nicely find out the host's frequencies, we should also
make sure that we get them into all virtual CPUs' device tree nodes.
Signed-off-by: NAlexander Graf <agraf@suse.de>

We have a bunch of helper functions that don't have any stubs for them in case
we don't have CONFIG_KVM enabled. That didn't bite us so far, because gcc can
optimize them out pretty well, but we should really provide them.
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

   - use uint64_t for clockfreq

We just got rid of the last user of kvmppc_read_host_property, so we
can now safely remove it.
Signed-off-by: NAlexander Graf <agraf@suse.de>

Now that we have nice and shiny APIs to read out the host's clock and timebase
frequencies, let's use them in the bamboo code as well!
Signed-off-by: NAlexander Graf <agraf@suse.de>

We don't need mpc8544_copy_soc_cell anymore, since we're explicitly reading
host values and writing guest values respectively.
Signed-off-by: NAlexander Graf <agraf@suse.de>

Now that we have generic KVM functions to read out the host tb and clock
frequencies, let's use them in the e500 code!
Signed-off-by: NAlexander Graf <agraf@suse.de>

We need to find out the host's clock-frequency when running on KVM, so
let's export a respective function.
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

  - enable 64bit values

We have some code in generic kvm_ppc.c that is only used by 440. Move to
the 440 specific device code.
Signed-off-by: NAlexander Graf <agraf@suse.de>

We have a qemu internal abstraction layer on FDT. While I'm not fully convinced
we need it at all, it's missing the nop_node functionality that we now need
on e500. So let's add it and think about the general future of that API later.
Signed-off-by: NAlexander Graf <agraf@suse.de>

Now that we can generate multiple envs for all our virtual CPUs, we
also need to tell the MPIC that we have multiple CPUs connected and
connect them all to the respective virtual interrupt lines.
Signed-off-by: NAlexander Graf <agraf@suse.de>

When creating a VM, we should go through smp_cpus and create a virtual CPU for
every CPU the user requested. This patch adds support for that and moves some
code around to make that more convenient.
Signed-off-by: NAlexander Graf <agraf@suse.de>

The MPIC emulation is now capable of handling up to 32 CPUs. Reflect that in
the code exporting the numbers out and fix an integer overflow while at it.
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

  - Max cpus is 15 due to cINT routing
  - Report nb_cpus not MAX_CPUS in MPIC capabilities

The bit definitions for critical interrupt routing are in PowerPC order
(most significant bit is 0), while we end up shifting it with normal bit
order. Turn the numbers around so we actually end up fetching the
right ones.
Signed-off-by: NAlexander Graf <agraf@suse.de>

The IPI dispatch registers are write only according to every MPIC
spec I have found. So instead of pretending you could read back something
from them, better not handle them at all.
Reported-by: NElie Richa <richa@adacore.com>
Signed-off-by: NAlexander Graf <agraf@suse.de>

We use the IDE register with IPIs as a mask to keep track which processors
have already acknowledged the respective interrupt. So we need to initialize
it to 0 to make sure that it doesn't accidently fire an IPI on CPU0 when the
first IPI is triggered.
Reported-by: NElie Richa <richa@adacore.com>
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v2 -> v3:

  - fix IDE IPI reset

The current IPI support in the MPIC code is incomplete and doesn't work. This
code adds proper support for IPIs in MPIC by using the IDE register to remember
which CPUs IPIs are still outstanding to. New triggers through the IPI trigger
register only add to the list of CPUs we want to IPI.
Signed-off-by: NAlexander Graf <agraf@suse.de>

---

v1 -> v2:

  - Use MAX_IPI instead of hardcoded 4
Signed-off-by: NAlexander Graf <agraf@suse.de>

The MPIC exports a page for each CPU that it controls. To support more than
one CPU, we need to also reserve the MMIO space according to the amount of
CPUs we want to support.
Signed-off-by: NAlexander Graf <agraf@suse.de>

The MPIC exports a register set for each CPU connected to it. They can all
be accessed through specific registers or using a shadow page that is mapped
differently depending on which CPU accesses it.

This patch implements the shadow map, making it possible for guests to access
the CPU local registers using the same address on each CPU.
Signed-off-by: NAlexander Graf <agraf@suse.de>

The MPIC has some funny feature where it maps different registers to an MMIO
region depending which CPU accesses them.

To be able to reflect that, we need to make OpenPIC be compiled in the target
code, so it can access cpu_single_env.
Signed-off-by: NAlexander Graf <agraf@suse.de>

This also lets the user see the irq in "info qtree".
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NAlexander Graf <agraf@suse.de>

Restructure common properties for sPAPR devices so that IRQ definitions
can be added in one place.
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NAlexander Graf <agraf@suse.de>

Right now the spapr devices cannot be instantiated with -device,
because the IRQs need to be passed to the spapr_*_create functions.
Do this instead in the bus's init wrapper.

This is particularly important with the conversion from scsi-disk
to scsi-{cd,hd} that Markus made.  After his patches, if you
specify a scsi-cd device attached to an if=none drive, the default
VSCSI controller will not be created and, without qdevification,
you will not be able to add yours.

NOTE from agraf: added small compile fix
Signed-off-by: NPaolo Bonzini <pbonzini@redhat.com>
Cc: Alexander Graf <agraf@suse.de>
Cc: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: NAlexander Graf <agraf@suse.de>

QED's metadata caching strategy allows two parallel requests to race for
metadata lookup.  The first one to complete will populate the metadata
cache and the second one will drop the data it just read in favor of the
cached data.

There is a use-after-free in qed_read_l2_table_cb() and
qed_commit_l2_update() where l2_table->offset was used after the
l2_table may have been freed due to a metadata lookup race.  Fix this by
keeping the l2_offset in a local variable and not reaching into the
possibly freed l2_table.
Reported-by: NAmit Shah <amit.shah@redhat.com>
Signed-off-by: NStefan Hajnoczi <stefanha@linux.vnet.ibm.com>
Signed-off-by: NAnthony Liguori <aliguori@us.ibm.com>