cuda.c: fix T2 timer and enable its interrupt

Fix the counter loading logic and enable the T2 interrupt when the timer
expires. Otherwise MacOS 9 hangs on boot.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: rename get_counter() state variable from s to ti for consistency

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: refactor get_tb() so that the time can be passed in

This is in preparation for sharing the code between timers.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: add defines for CUDA registers

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: fix CUDA SR interrupt clearing

Make sure that we also clear the data and clock interrupts at the same time.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: implement dummy IIC access commands

These are used by MacOS 9 on boot. Here we return an error except for 4-byte
commands which write to the IIC bus in a similar manner to MOL.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: implement simple CUDA_GET_6805_ADDR command

This simply returns an empty response with no error status as implemented by
MOL to allow MacOS 9 boot to proceed further.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: fix CUDA_PACKET response packet format

According to comments in MOL, the response to a CUDA_PACKET should be one of
the following:

Reply: (CUDA_PACKET, status, cmd)
Error: (ERROR_PACKET, status, CUDA_PACKET, cmd)

Update cuda_receive_packet() accordingly to reflect this in order to make
MacOS 9 happy.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

cuda.c: fix CUDA ADB error packet format

According to MOL, ADB error packets should be of the form (type, status, cmd)
rather than just (type, status). This fixes ADB device detection under MacOS 9.

Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

PPC: mac99: Always add USB controller

The mac99 machines always have a USB controller. Usually not having one around
doesn't hurt quite as much, but Mac OS 9 really really wants one or it crashes
on bootup.

So always add OHCI to make it happy.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

PPC: Fix lswx bounds checks

The lswx instruction checks whether the desired string actually fits
into all defined registers. Unfortunately it does the calculation wrong,
resulting in illegal instruction traps for loads that really should fit.

Fix it up, making Mac OS happier.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

PPC: Allow Rc bit to be set on mtspr

According to the ISA setting the Rc bit on mtspr is undefined behavior.
Real 750 hardware simply ignores the bit and doesn't touch cr0 though.

Unfortunately, Mac OS 9 relies on this fact and executes a few mtspr
instructions (to set XER for example) with Rc set.

So let's handle the bit the same way hardware does and ignore it.

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Merge remote-tracking branch 'remotes/cohuck/tags/s390x-20151111' into staging

Hopefully last big batch of s390x patches, including:
- bugfixes for LE host and for pci translation
- MAINTAINERS update
- hugetlbfs enablement (kernel patches pending)
- boot from El Torito iso images on virtio-blk
  (boot from scsi pending)
- cleanup in the ipl device code

There's also a helper function for resetting busless devices in the
qdev core in there.

# gpg: Signature made Wed 11 Nov 2015 17:49:58 GMT using RSA key ID C6F02FAF
# gpg: Good signature from "Cornelia Huck <huckc@linux.vnet.ibm.com>"
# gpg:                 aka "Cornelia Huck <cornelia.huck@de.ibm.com>"

* remotes/cohuck/tags/s390x-20151111:
  s390: deprecate the non-ccw machine in 2.5
  s390x/ipl: switch error reporting to error_setg
  s390x/ipl: clean up qom definitions and turn into TYPE_DEVICE
  qdev: provide qdev_reset_all_fn()
  pc-bios/s390-ccw: rebuild image
  pc-bios/s390-ccw: El Torito 16-bit boot image size field workaround
  pc-bios/s390-ccw: El Torito s390x boot entry check
  pc-bios/s390-ccw: ISO-9660 El Torito boot implementation
  pc-bios/s390-ccw: Always adjust virtio sector count
  s390x/kvm: don't enable CMMA when hugetlbfs will be used
  s390x: switch to memory_region_allocate_system_memory
  MAINTAINERS: update virtio-ccw/s390 git tree
  MAINTAINERS: update s390 file patterns
  s390x/pci : fix up s390 pci iommu translation function
  s390x/css: sense data endianness

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/kevin/tags/for-upstream' into staging

Block layer patches

# gpg: Signature made Wed 11 Nov 2015 16:03:19 GMT using RSA key ID C88F2FD6
# gpg: Good signature from "Kevin Wolf <kwolf@redhat.com>"

* remotes/kevin/tags/for-upstream: (41 commits)
  iotests: Check for quorum support in test 139
  qcow2: Fix qcow2_get_cluster_offset() for zero clusters
  iotests: Add tests for the x-blockdev-del command
  block: Add 'x-blockdev-del' QMP command
  block: Add blk_get_refcnt()
  mirror: block all operations on the target image during the job
  qemu-iotests: fix -valgrind option for check
  qemu-iotests: fix cleanup of background processes
  qemu-io: Correct error messages
  qemu-io: Check for trailing chars
  qemu-io: fix cvtnum lval types
  block: test 'blockdev-snapshot' using a file BDS as the overlay
  block: Remove inner quotation marks in iotest 085
  block: Disallow snapshots if the overlay doesn't support backing files
  throttle: Use bs->throttle_state instead of bs->io_limits_enabled
  throttle: Check for pending requests in throttle_group_unregister_bs()
  qemu-img: add check for zero-length job len
  qcow2: avoid misaligned 64bit bswap
  qemu-iotests: Test the reopening of overlay_bs in 'block-commit'
  commit: reopen overlay_bs before base
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

s390: deprecate the non-ccw machine in 2.5

The non-ccw machine for s390 (s390-virtio) is not very well maintained
and caused several issues in the past:
- aliases like virtio-blk did not work for s390
- virtio refactoring failed due to long standing bugs (e.g.see
commit cb927b8a "s390-virtio: Accommodate guests using virtqueues too early")
- some features like memory hotplug will cause trouble due to virtio storage
  being above guest memory
- the boot loader bios no longer seems to work. the source code of that
  loader is also no longer maintained

2.4 changed the default to the ccw machine, let's deprecate the old
machine for 2.5.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Acked-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Message-Id: <1446811645-25565-1-git-send-email-borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

s390x/ipl: switch error reporting to error_setg

Now that we can report errors in the realize function, let's replace
the fprintf's and hw_error's with error_setg.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

s390x/ipl: clean up qom definitions and turn into TYPE_DEVICE

Let's move the qom definitions of the ipl device into ipl.h, replace
"s390-ipl" by a proper type define, turn it into a TYPE_DEVICE
and remove the unneeded class definition.

Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

qdev: provide qdev_reset_all_fn()

For TYPE_DEVICE, the dc->reset() function is not called on system resets
yet. Until that is changed, we have to manually register a reset handler.
Let's provide qdev_reset_all_fn(), that can directly be used - just like
the reset handler that is already available for qbus.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

pc-bios/s390-ccw: rebuild image

Contains:
  pc-bios/s390-ccw: Always adjust virtio sector count
  pc-bios/s390-ccw: ISO-9660 El Torito boot implementation
  pc-bios/s390-ccw: El Torito s390x boot entry check
  pc-bios/s390-ccw: El Torito 16-bit boot image size field workaround

Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

pc-bios/s390-ccw: El Torito 16-bit boot image size field workaround

Because of El Torito spec flaw boot image size needs to be verified.

Boot catalog entry size field has 16-bit width, and specifies size
in 512-byte units.

Thus, boot image size cannot exceed 32M.

We actually search for the file to get the file size.

This is done by scanning the ISO directory tree for the ISO block number
and reading the file size from the directory entry.

Signed-off-by: Maxim Samoylov <max7255@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

pc-bios/s390-ccw: El Torito s390x boot entry check

Boot entry is considered compatible if boot image is Linux kernel
with matching S390 Linux magic string.

Empty boot images with sector_count == 0 are considered broken.

Signed-off-by: Maxim Samoylov <max7255@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

pc-bios/s390-ccw: ISO-9660 El Torito boot implementation

This patch enables boot from media formatted according to
ISO-9660 and El Torito bootable CD specification.

We try to boot from device as ISO-9660 media when SCSI IPL failed.

The first boot catalog entry with bootable flag is used.

ISO-9660 media with default 2048-bytes sector size only is supported.

Signed-off-by: Maxim Samoylov <max7255@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

pc-bios/s390-ccw: Always adjust virtio sector count

Let's always adjust the sector number to be read using the current
virtio block size value.

This prepares for the implementation of IPL from ISO-9660 media.

Signed-off-by: Maxim Samoylov <max7255@linux.vnet.ibm.com>
Reviewed-by: David Hildenbrand <dahi@linux.vnet.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

s390x/kvm: don't enable CMMA when hugetlbfs will be used

On hugetlbfs CMMA will not be useful as every ESSA instruction will trap.
So don't offer CMMA to guests with a hugepages backing.

Signed-off-by: Dominik Dingel <dingel@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

s390x: switch to memory_region_allocate_system_memory

By replacing memory_region_init_ram with memory_region_allocate_system_memory
we gain goodies like mem-path backends. This will allow us to use hugetlbfs
once the kernel supports it.

Signed-off-by: Dominik Dingel <dingel@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

MAINTAINERS: update virtio-ccw/s390 git tree

Let's reference the git branch I actually use, and add Christian's
git tree.

Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

MAINTAINERS: update s390 file patterns

We were missing some files, and some files should get an additional
entry to add the people actually looking after the code.

Reported-by: Thomas Huth <thuth@redhat.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

s390x/pci : fix up s390 pci iommu translation function

On s390x, each pci device has its own iommu, which is only properly
setup in qemu once the mpcifc instruction used to register the
translation table has been intercepted. Therefore, for a pci device that
is not configured or has not been initialized, proper translation is
neither required nor possible. Moreover, we may not have a host bridge
device ready yet.

This was exposed by a recent vfio change that triggers iommu translation
during the initialization of the vfio pci device. Let's do an early exit
in that case.

Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Jens Freimann <jfrei@linux.vnet.ibm.com>
Signed-off-by: Yi Min Zhao <zyimin@linux.vnet.ibm.com>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>

s390x/css: sense data endianness

We keep the device's sense data in a byte array (following the
architecture), but the ecws are an array of 32 bit values. If we
just blindly copy the values, the sense data will change from
de-facto BE data to de-facto cpu-endian data, which means we end
up doing an incorrect conversion on LE hosts.

Let's just explicitly convert to cpu-endianness while assembling
the irb.

Reported-by: Andy Lutomirski <luto@kernel.org>
Tested-by: Andy Lutomirski <luto@kernel.org>
Signed-off-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Reviewed-by: Christian Borntraeger <borntraeger@de.ibm.com>

Merge remote-tracking branch 'mreitz/tags/pull-block-for-kevin-2015-11-11' into queue-block

Block patches from 2015-10-26 until 2015-11-11.

# gpg: Signature made Wed Nov 11 17:00:50 2015 CET using RSA key ID E838ACAD
# gpg: Good signature from "Max Reitz <mreitz@redhat.com>"

* mreitz/tags/pull-block-for-kevin-2015-11-11:
  iotests: Check for quorum support in test 139
  qcow2: Fix qcow2_get_cluster_offset() for zero clusters
  iotests: Add tests for the x-blockdev-del command
  block: Add 'x-blockdev-del' QMP command
  block: Add blk_get_refcnt()
  mirror: block all operations on the target image during the job
  qemu-iotests: fix -valgrind option for check
  qemu-iotests: fix cleanup of background processes

Signed-off-by: Kevin Wolf <kwolf@redhat.com>

iotests: Check for quorum support in test 139

The quorum driver is always built in, but it is disabled during
run-time if there's no SHA256 support available (see commit e94867e).

This patch skips the quorum test in iotest 139 in that case.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Message-id: 1447172891-20410-1-git-send-email-berto@igalia.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

qcow2: Fix qcow2_get_cluster_offset() for zero clusters

When searching for contiguous zero clusters, we only need to check the
cluster type. Before this patch, an increasing offset (L2E_OFFSET_MASK)
was expected, so that the function never returned more than a single
zero cluster in practice. This patch fixes it to actually return as many
contiguous zero clusters as it can.

Cc: qemu-stable@nongnu.org
Signed-off-by: Kevin Wolf <kwolf@redhat.com>
Message-id: 1446657384-5907-1-git-send-email-kwolf@redhat.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

iotests: Add tests for the x-blockdev-del command

Signed-off-by: Alberto Garcia <berto@igalia.com>
Message-id: 57c3b0d4d0c73ddadd19e5bded9492c359cc4568.1446475331.git.berto@igalia.com
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>

block: Add 'x-blockdev-del' QMP command

This command is still experimental, hence the name.

This is the companion to 'blockdev-add'. It allows deleting a
BlockBackend with its associated BlockDriverState tree, or a
BlockDriverState that is not attached to any backend.

In either case, the command fails if the reference count is greater
than 1 or the BlockDriverState has any parents.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 6cfc148c77aca1da942b094d811bfa3fcf7ac7bb.1446475331.git.berto@igalia.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

block: Add blk_get_refcnt()

This function returns the reference count of a given BlockBackend.
For convenience, it returns 0 if the BlockBackend pointer is NULL.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: dfdd8a17dbe3288842840636d2cfe5bb895abcb0.1446475331.git.berto@igalia.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

mirror: block all operations on the target image during the job

There's nothing preventing the target image from being used by other
operations during the 'drive-mirror' job, so we should block them all
until the job is done.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Message-id: 82b88fd04cde918a08a6f9a4ab85626d7fd7b502.1446475331.git.berto@igalia.com
Signed-off-by: Max Reitz <mreitz@redhat.com>

qemu-iotests: fix -valgrind option for check

Commit 934659c switched the iotests to run qemu-io from a bash subshell,
in order to catch segfaults.  This method is incompatible with the
current valgrind_qemu_io() bash function.

Move the valgrind usage into the exec subshell in _qemu_io_wrapper(),
while making sure the original return value is passed back to the
caller.

Update test output for tests 039, 061, and 137 as it looks for the
specific subshell command when the process is terminated.

Reported-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
Message-id: 0066fd85d26ca641a1c25135ff2479b7985701cf.1446232490.git.jcody@redhat.com
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Max Reitz <mreitz@redhat.com>

qemu-iotests: fix cleanup of background processes

Commit 934659c switched the iotests to run qemu and qemu-nbd from a bash
subshell, in order to catch segfaults.  Unfortunately, this means the
process PID cannot be captured via '$!'. We stopped killing qemu and
qemu-nbd processes, leaving a lot of orphaned, running qemu processes
after executing iotests.

Since the process is using exec in the subshell, the PID is the
same as the subshell PID.

Track these PIDs for cleanup using pidfiles in the $TEST_DIR. Only
track the qemu PID, however, if requested - not all usage requires
killing the process.

Reported-by: John Snow <jsnow@redhat.com>
Signed-off-by: Jeff Cody <jcody@redhat.com>
Message-id: 9e4f958b3895b7259b98d845bb46f000ba362869.1446232490.git.jcody@redhat.com
[mreitz@redhat.com: Replaced '! -z "..."' by '-n "..."']
Signed-off-by: Max Reitz <mreitz@redhat.com>

qemu-io: Correct error messages

Reported-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-io: Check for trailing chars

Make sure there's not trailing garbage, e.g.
"64k-whatever-i-want-here"

Reported-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-io: fix cvtnum lval types

cvtnum() returns int64_t: we should not be storing this
result inside of an int.

In a few cases, we need an extra sprinkling of error handling
where we expect to pass this number on towards a function that
expects something smaller than int64_t.

Reported-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: John Snow <jsnow@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: test 'blockdev-snapshot' using a file BDS as the overlay

This test checks that it is not possible to create a snapshot if the
requested overlay node is a BDS which does not support backing images.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Remove inner quotation marks in iotest 085

This patch removes the inner quotation marks in all cases like this:

   cmd=" ... "${variable}" ... "

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Disallow snapshots if the overlay doesn't support backing files

This addresses scenarios like this one:

  { 'execute': 'blockdev-add', 'arguments':
    { 'options': { 'driver': 'qcow2',
                   'node-name': 'new0',
                   'file': { 'driver': 'file',
                             'filename': 'new.qcow2',
                             'node-name': 'file0' } } } }

  { 'execute': 'blockdev-snapshot', 'arguments':
    { 'node': 'virtio0',
      'overlay': 'file0' } }

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

throttle: Use bs->throttle_state instead of bs->io_limits_enabled

There are two ways to check for I/O limits in a BlockDriverState:

- bs->throttle_state: if this pointer is not NULL, it means that this
  BDS is member of a throttling group, its ThrottleTimers structure
  has been initialized and its I/O limits are ready to be applied.

- bs->io_limits_enabled: if true it means that the throttle_state
  pointer is valid _and_ the limits are currently enabled.

The latter is used in several places to check whether a BDS has I/O
limits configured, but what it really checks is whether requests
are being throttled or not. For example, io_limits_enabled can be
temporarily set to false in cases like bdrv_read_unthrottled() without
otherwise touching the throtting configuration of that BDS.

This patch replaces bs->io_limits_enabled with bs->throttle_state in
all cases where what we really want to check is the existence of I/O
limits, not whether they are currently enabled or not.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

throttle: Check for pending requests in throttle_group_unregister_bs()

throttle_group_unregister_bs() removes a BlockDriverState from its
throttling group and destroys the timers. This means that there must
be no pending throttled requests at that point (because it would be
impossible to complete them), so the caller has to drain them first.

At the moment throttle_group_unregister_bs() is only called from
bdrv_io_limits_disable(), which already takes care of draining the
requests, so there's nothing to worry about, but this patch makes
this invariant explicit in the documentation and adds the relevant
assertions.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-img: add check for zero-length job len

The mirror job doesn't update its total length until
it has already started running, so we should translate
a zero-length job-len as meaning 0%.

Otherwise, we may get divide-by-zero faults.

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qcow2: avoid misaligned 64bit bswap

If we create a buffer directly on the stack by using 12 bytes, there's
no guarantee the 64bit value we want to swap will be aligned, which
could cause errors with undefined behavior.

Spotted with clang -fsanitize=undefined and observed in iotests 15, 26,
44, 115 and 121.

Signed-off-by: John Snow <jsnow@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qemu-iotests: Test the reopening of overlay_bs in 'block-commit'

The 'block-commit' command needs the overlay image of 'top' to
be opened in read-write mode in order to update the backing file
string. If 'top' is not the active layer or its backing file then its
overlay needs to be reopened during the block job.

This is a test case for that scenario.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

commit: reopen overlay_bs before base

'block-commit' needs write access to two different nodes of the chain:

- 'base', because that's where the data is written to.
- the overlay of 'top', because it needs to update the backing file
  string to point to 'base' after the operation.

Both images have to be opened in read-write mode, and commit_start()
takes care of reopening them if necessary.

With the current implementation, however, when overlay_bs is reopened
in read-write mode it has the side effect of making 'base' read-only
again, eventually making 'block-commit' fail.

This needs to be fixed in bdrv_reopen(), but until we get to that it
can be worked around simply by swapping the order of base and
overlay_bs in the reopen queue.

In order to reproduce this bug, overlay_bs needs to be initially in
read-only mode. That is: the 'top' parameter of 'block-commit' cannot
be the active layer nor its immediate backing chain.

Cc: qemu-stable@nongnu.org
Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: add tests for the 'blockdev-snapshot' command

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: add a 'blockdev-snapshot' QMP command

One of the limitations of the 'blockdev-snapshot-sync' command is that
it does not allow passing BlockdevOptions to the newly created
snapshots, so they are always opened using the default values.

Extending the command to allow passing options is not a practical
solution because there is overlap between those options and some of
the existing parameters of the command.

This patch introduces a new 'blockdev-snapshot' command with a simpler
interface: it just takes two references to existing block devices that
will be used as the source and target for the snapshot.

Since the main difference between the two commands is that one of them
creates and opens the target image, while the other uses an already
opened one, the bulk of the implementation is shared.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: support passing 'backing': '' to 'blockdev-add'

Passing an empty string allows opening an image but not its backing
file. This was already described in the API documentation, only the
implementation was missing.

This is useful for creating snapshots using images opened with
blockdev-add, since they are not supposed to have a backing image
before the operation.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: rename BlockdevSnapshot to BlockdevSnapshotSync

We will introduce the 'blockdev-snapshot' command that will require
its own struct for the parameters, so we need to rename this one in
order to avoid name clashes.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: check for existing device IDs in external_snapshot_prepare()

The 'snapshot-node-name' parameter of blockdev-snapshot-sync allows
setting the node name of the image that is going to be created.

Before creating the image, external_snapshot_prepare() checks that the
name is not already being used. The check is however incomplete since
it only considers existing node names, but node names must not clash
with device IDs either because they share the same namespace.

If the user attempts to create a snapshot using the name of an
existing device for the 'snapshot-node-name' parameter the operation
will eventually fail, but only after the new image has been created.

This patch replaces bdrv_find_node() with bdrv_lookup_bs() to extend
the check to existing device IDs, and thus detect possible name
clashes before the new image is created.

Signed-off-by: Alberto Garcia <berto@igalia.com>
Reviewed-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Jeff Cody <jcody@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

iotests: Add test for change-related QMP commands

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

hmp: Add read-only-mode option to change command

Expose the new read-only-mode option of 'blockdev-change-medium' for the
'change' HMP command.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

blockdev: read-only-mode for blockdev-change-medium

Add an option to qmp_blockdev_change_medium() which allows changing the
read-only status of the block device whose medium is changed.

Some drives do not have a inherently fixed read-only status; for
instance, floppy disks can be set read-only or writable independently of
the drive. Some users may find it useful to be able to therefore change
the read-only status of a block device when changing the medium.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

hmp: Use blockdev-change-medium for change command

Use separate code paths for the two overloaded functions of the 'change'
HMP command, and invoke the 'blockdev-change-medium' QMP command if used
on a block device (by calling qmp_blockdev_change_medium()).

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

qmp: Introduce blockdev-change-medium

Introduce a new QMP command 'blockdev-change-medium' which is intended
to replace the 'change' command for block devices. The existing function
qmp_change_blockdev() is accordingly renamed to
qmp_blockdev_change_medium().

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Inquire tray state before tray-moved events

blk_dev_change_media_cb() is called for all potential tray movements;
however, it is possible to request closing the tray but nothing actually
happening (on a floppy disk drive without a medium).

Thus, the actual tray status should be inquired before sending a
tray-moved event (and an event should be sent whenever the status
changed).

Checking @load is now superfluous; it was necessary because it was
possible to change a medium without having explicitly opened the tray
and closed it again (or it might have been possible, at least). This is
no longer possible, though.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

blockdev: Implement change with basic operations

Implement 'change' on block devices by calling blockdev-open-tray,
blockdev-remove-medium, blockdev-insert-medium (a variation of that
which does not need a node-name) and blockdev-close-tray.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

blockdev: Implement eject with basic operations

Implement 'eject' by calling blockdev-open-tray and
blockdev-remove-medium.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

blockdev: Add blockdev-insert-medium

And a helper function for that, which directly takes a pointer to the
BDS to be inserted instead of its node-name (which will be used for
implementing 'change' using blockdev-insert-medium).

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

blockdev: Add blockdev-remove-medium

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

blockdev: Add blockdev-close-tray

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

blockdev: Add blockdev-open-tray

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Add functions for inheriting a BBRS

In order to open a BDS which inherits a BB's root state,
blk_get_open_flags_from_root_state() is used to inquire the flags to be
passed to bdrv_open(), and blk_apply_root_state() is used to apply the
remaining state after the BDS has been opened.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Make bdrv_states public

When inserting a BDS tree into a BB, we will need to add the root BDS to
this list. Since we will want to do that in the blockdev-insert-medium
implementation in blockdev.c, we will need access to it there.

This patch is not exactly elegant, but bdrv_states will be removed in
the future anyway because we no longer need it since we have BBs.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Reviewed-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Add blk_remove_bs()

This function removes the BlockDriverState associated with the given
BlockBackend from that BB and sets the BDS pointer in the BB to NULL.

Signed-off-by: Max Reitz <mreitz@redhat.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

block: Don't call blk_bs() twice in bdrv_lookup_bs()

Signed-off-by: Alberto Garcia <berto@igalia.com>
Signed-off-by: Kevin Wolf <kwolf@redhat.com>

Merge remote-tracking branch 'remotes/dgibson/tags/ppc-next-20151111' into staging

ppc patch queue - 2015-11-11

Highlights:
  - Updated SLOF version for "pseries machine
  - Bugfix / cleanup for KVM hash page table allocation

# gpg: Signature made Wed 11 Nov 2015 02:30:51 GMT using RSA key ID 20D9B392
# gpg: Good signature from "David Gibson <david@gibson.dropbear.id.au>"
# gpg:                 aka "David Gibson (Red Hat) <dgibson@redhat.com>"
# gpg:                 aka "David Gibson (ozlabs.org) <dgibson@ozlabs.org>"
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: 75F4 6586 AE61 A66C C44E  87DC 6C38 CACA 20D9 B392

* remotes/dgibson/tags/ppc-next-20151111:
  spapr: Handle failure of KVM_PPC_ALLOCATE_HTAB ioctl
  ppc: Let kvmppc_reset_htab() return 0 for !CONFIG_KVM
  pseries: Update SLOF firmware image to qemu-slof-20151103
  ppc: Add/Re-introduce MMU model definitions needed by PR KVM

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

spapr: Handle failure of KVM_PPC_ALLOCATE_HTAB ioctl

KVM_PPC_ALLOCATE_HTAB ioctl can return -ENOMEM for KVM guests and QEMU
never handled this correctly. But this didn't cause any problems till
now as KVM_PPC_ALLOCATE_HTAB ioctl returned with smaller than requested
HTAB when enough contiguous memory wasn't available in the host.
After the proposed kernel change: https://patchwork.ozlabs.org/patch/530501/,
KVM_PPC_ALLOCATE_HTAB ioctl will not fallback to lower sized HTAB
allocation and will fail if requested HTAB size can't be met.

Check for such failures in QEMU and abort appropriately. This will
prevent guest kernel from hanging/freezing during early boot by doing
graceful exit when host is unable to allocate requested HTAB.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

ppc: Let kvmppc_reset_htab() return 0 for !CONFIG_KVM

The !CONFIG_KVM implementation of kvmppc_reset_htab() returns -1
by default. Change this to return 0 so that we fall back to user space
HTAB allocation for emulated guests.

This fixes the make check failures for ppc64 emulated target.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

pseries: Update SLOF firmware image to qemu-slof-20151103

The changes are:
1. supports recent binutils;
2. 64bit BARs behind PCI bridges supported;
3. Many fixes for USB keyboard support - keys, XHCI;
4. virtio-vga support.

This image was built with:
gcc version 4.8.3 20140911 (Red Hat 4.8.3-7) (GCC)
GNU ld version 2.23.2

The full changelog is:
  > version: update to 20151103
  > documentation: Add a clause about signing off
  > qemu/js2x/client: Support binutils >= 2.25.1
  > Fix special keys on USB
  > Fix function keys on USB
  > pci-scan: program 64-bit mem bar range in pci-bridge bar
  > Allow to build SLOF on Little Endian host
  > usb-xhci: add keyboard support
  > usb-xhci: ready the link trb early
  > usb-xhci: scan usb high speed ports
  > usb-xhci: bulk improve event handling loop
  > usb-xhci: return on allocation failure
  > usb-xhci: add delay in shutdown path
  > usb-xhci: event trbs does not need link trb
  > usb-hid: refactor usb key reading
  > takeover: Fix header includes
  > board-js2x: Add missing file dma-function.fs
  > vga: Add support for virtio-vga
  > qemu-vga: Use MMIO BAR instead of legacy IO ports
  > slof: Change call_c() function to a proper assembler function

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

ppc: Add/Re-introduce MMU model definitions needed by PR KVM

Commit aa4bb5875231 (ppc: Add mmu_model defines for arch 2.03 and 2.07)
removed the mmu_model definition POWERPC_MMU_2_06a which is needed by
PR KVM. Reintroduce it and also add POWERPC_MMU_2_07a.

This fixes QEMU crash (qemu: fatal: Unknown MMU model) during booting
of PR KVM guest.

Signed-off-by: Bharata B Rao <bharata@linux.vnet.ibm.com>
Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: David Gibson <david@gibson.dropbear.id.au>

Merge remote-tracking branch 'remotes/awilliam/tags/vfio-update-20151110.0' into staging

VFIO updates 2015-11-10

 - Make Windows happy with vfio-pci devices exposed on conventional
   PCI buses on q35 by hiding PCIe capability (Alex Williamson)
 - Convert to g_new() where appropriate (Markus Armbruster)

# gpg: Signature made Tue 10 Nov 2015 19:46:41 GMT using RSA key ID 3BB08B22
# gpg: Good signature from "Alex Williamson <alex.williamson@redhat.com>"
# gpg:                 aka "Alex Williamson <alex@shazbot.org>"
# gpg:                 aka "Alex Williamson <alwillia@redhat.com>"
# gpg:                 aka "Alex Williamson <alex.l.williamson@gmail.com>"

* remotes/awilliam/tags/vfio-update-20151110.0:
  vfio: Use g_new() & friends where that makes obvious sense
  vfio/pci: Hide device PCIe capability on non-express buses for PCIe VMs

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

vfio: Use g_new() & friends where that makes obvious sense

g_new(T, n) is neater than g_malloc(sizeof(T) * n).  It's also safer,
for two reasons.  One, it catches multiplication overflowing size_t.
Two, it returns T * rather than void *, which lets the compiler catch
more type errors.

This commit only touches allocations with size arguments of the form
sizeof(T).  Same Coccinelle semantic patch as in commit b45c03f.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>

vfio/pci: Hide device PCIe capability on non-express buses for PCIe VMs

When we have a PCIe VM, such as Q35, guests start to care more about
valid configurations of devices relative to the VM view of the PCI
topology.  Windows will error with a Code 10 for an assigned device if
a PCIe capability is found for a device on a conventional bus.  We
also have the possibility of IOMMUs, like VT-d, where the where the
guest may be acutely aware of valid express capabilities on physical
hardware.

Some devices, like tg3 are adversely affected by this due to driver
dependencies on the PCIe capability.  The only solution for such
devices is to attach them to an express capable bus in the VM.

Signed-off-by: Alex Williamson <alex.williamson@redhat.com>

Merge remote-tracking branch 'remotes/juanquintela/tags/migration/20151110' into staging

migration/next for 20151110

# gpg: Signature made Tue 10 Nov 2015 14:23:26 GMT using RSA key ID 5872D723
# gpg: Good signature from "Juan Quintela <quintela@redhat.com>"
# gpg:                 aka "Juan Quintela <quintela@trasno.org>"

* remotes/juanquintela/tags/migration/20151110: (57 commits)
  migration: qemu_savevm_state_cleanup becomes mandatory operation
  Inhibit ballooning during postcopy
  Disable mlock around incoming postcopy
  End of migration for postcopy
  Postcopy: Mark nohugepage before discard
  postcopy: Wire up loadvm_postcopy_handle_ commands
  Start up a postcopy/listener thread ready for incoming page data
  Postcopy; Handle userfault requests
  Round up RAMBlock sizes to host page sizes
  Host page!=target page: Cleanup bitmaps
  Don't iterate on precopy-only devices during postcopy
  Don't sync dirty bitmaps in postcopy
  postcopy: Check order of received target pages
  Postcopy: Use helpers to map pages during migration
  postcopy_ram.c: place_page and helpers
  Page request: Consume pages off the post-copy queue
  Page request: Process incoming page request
  Page request: Add MIG_RP_MSG_REQ_PAGES reverse command
  Postcopy: End of iteration
  Postcopy: Postcopy startup in migration thread
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

migration: qemu_savevm_state_cleanup becomes mandatory operation

since commit
    commit 94f5a43704129ca4995aa3385303c5ae225bde42
    Author: Liang Li <liang.z.li@intel.com>
    Date:   Mon Nov 2 15:37:00 2015 +0800

    migration: defer migration_end & blk_mig_cleanup

when actual .cleanup callbacks calling was removed from complete operations.

The patch fixes regression introduced by the commit above results in
100% reliable assert for virtio-scsi VM with iothreads enabled during
'virsh create-snapshot' operation:
    assert(i != mr->ioeventfd_nb);
    memory_region_del_eventfd
    virtio_pci_set_host_notifier_internal
    virtio_pci_set_host_notifier
    virtio_scsi_dataplane_start
    virtio_scsi_handle_cmd
    virtio_queue_notify_vq
    virtio_queue_host_notifier_read
    aio_dispatch

Signed-off-by: Denis V. Lunev <den@openvz.org>
Reviewed-by: Liang Li <liang.z.li@intel.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
CC: Paolo Bonzini <pbonzini@redhat.com>
CC: Juan Quintela <quintela@redhat.com>
CC: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Inhibit ballooning during postcopy

Postcopy detects accesses to pages that haven't been transferred yet
using userfaultfd, and it causes exceptions on pages that are 'not
present'.
Ballooning also causes pages to be marked as 'not present' when the
guest inflates the balloon.
Potentially a balloon could be inflated to discard pages that are
currently inflight during postcopy and that may be arriving at about
the same time.

To avoid this confusion, disable ballooning during postcopy.

When disabled we drop balloon requests from the guest.  Since ballooning
is generally initiated by the host, the management system should avoid
initiating any balloon instructions to the guest during migration,
although it's not possible to know how long it would take a guest to
process a request made prior to the start of migration.
Guest initiated ballooning will not know if it's really freed a page
of host memory or not.

Queueing the requests until after migration would be nice, but is
non-trivial, since the set of inflate/deflate requests have to
be compared with the state of the page to know what the final
outcome is allowed to be.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Disable mlock around incoming postcopy

Userfault doesn't work with mlock; mlock is designed to nail down pages
so they don't move, userfault is designed to tell you when they're not
there.

munlock the pages we userfault protect before postcopy.
mlock everything again at the end if mlock is enabled.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

End of migration for postcopy

Tweak the end of migration cleanup; we don't want to close stuff down
at the end of the main stream, since the postcopy is still sending pages
on the other thread.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Postcopy: Mark nohugepage before discard

Prior to servicing userfault requests we must ensure we've not got
huge pages in the area that might include non-transferred memory,
since a hugepage could incorrectly mark the whole huge page as present.

We mark the area as non-huge page (nhp) just before we perform
discards; the discard code now tells us to discard any areas
that haven't been sent (as well as any that are redirtied);
any already formed transparent-huge-pages get fragmented
by this discard process if they cotnain any discards.

Transparent huge pages that have been entirely transferred
and don't contain any discards are not broken by this mechanism;
they stay as huge pages.

By starting postcopy after a full precopy pass, many of the pages
then stay as huge pages; this is important for maintaining performance
after the end of the migration.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

postcopy: Wire up loadvm_postcopy_handle_ commands

Wire up more of the handlers for the commands on the destination side,
in particular loadvm_postcopy_handle_run now has enough to start the
guest running.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Start up a postcopy/listener thread ready for incoming page data

The loading of a device state (during postcopy) may access guest
memory that's still on the source machine and thus might need
a page fill; split off a separate thread that handles the incoming
page data so that the original incoming migration code can finish
off the device data.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Postcopy; Handle userfault requests

userfaultfd is a Linux syscall that gives an fd that receives a stream
of notifications of accesses to pages registered with it and allows
the program to acknowledge those stalls and tell the accessing
thread to carry on.

We convert the requests from the kernel into messages back to the
source asking for the pages.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Round up RAMBlock sizes to host page sizes

RAMBlocks that are not a multiple of host pages in length
cause problems for postcopy (I've seen an ACPI table on aarch64
be 5k in length - i.e. 5x target-page), so round RAMBlock sizes
up to a host-page.

This potentially breaks migration compatibility due to changes
in RAMBlock sizes; however:
   1) x86 and s390 I think always have host=target page size
   2) When I've tried on Power the block sizes already seem aligned.
   3) I don't think there's anything else that maintains per-version
      machine-types for compatibility.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Host page!=target page: Cleanup bitmaps

Prior to the start of postcopy, ensure that everything that will
be transferred later is a whole host-page in size.

This is accomplished by discarding partially transferred host pages
and marking any that are partially dirty as fully dirty.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Don't iterate on precopy-only devices during postcopy

During the postcopy phase we must not call the iterate method on
precopy-only devices, since they may have done some cleanup during
the _complete call at the end of the precopy phase.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Don't sync dirty bitmaps in postcopy

Once we're in postcopy the source processors are stopped and memory
shouldn't change any more, so there's no need to look at the dirty
map.

There are two notes to this:
  1) If we do resync and a page had changed then the page would get
     sent again, which the destination wouldn't allow (since it might
     have also modified the page)
  2) Before disabling this I'd seen very rare cases where a page had been
     marked dirtied although the memory contents are apparently identical

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

postcopy: Check order of received target pages

Ensure that target pages received within a host page are in order.
This shouldn't trigger, but in the cases where the sender goes
wrong and sends stuff out of order it produces a corruption that's
really nasty to debug.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Postcopy: Use helpers to map pages during migration

In postcopy, the destination guest is running at the same time
as it's receiving pages; as we receive new pages we must put
them into the guests address space atomically to avoid a running
CPU accessing a partially written page.

Use the helpers in postcopy-ram.c to map these pages.

qemu_get_buffer_in_place is used to avoid a copy out of qemu_file
in the case that postcopy is going to do a copy anyway.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

postcopy_ram.c: place_page and helpers

postcopy_place_page (etc) provide a way for postcopy to place a page
into guests memory atomically (using the copy ioctl on the ufd).

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Page request: Consume pages off the post-copy queue

When transmitting RAM pages, consume pages that have been queued by
MIG_RPCOMM_REQPAGE commands and send them ahead of normal page scanning.

Note:
  a) After a queued page the linear walk carries on from after the
unqueued page; there is a reasonable chance that the destination
was about to ask for other closeby pages anyway.

  b) We have to be careful of any assumptions that the page walking
code makes, in particular it does some short cuts on its first linear
walk that break as soon as we do a queued page.

  c) We have to be careful to not break up host-page size chunks, since
this makes it harder to place the pages on the destination.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Page request: Process incoming page request

On receiving MIG_RPCOMM_REQ_PAGES look up the address and
queue the page.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Page request: Add MIG_RP_MSG_REQ_PAGES reverse command

Add MIG_RP_MSG_REQ_PAGES command on Return path for the postcopy
destination to request a page from the source.

Two versions exist:
   MIG_RP_MSG_REQ_PAGES_ID that includes a RAMBlock name and start/len
   MIG_RP_MSG_REQ_PAGES that just has start/len for use with the same
                        RAMBlock as a previous MIG_RP_MSG_REQ_PAGES_ID

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Postcopy: End of iteration

The end of migration in postcopy is a bit different since some of
the things normally done at the end of migration have already been
done on the transition to postcopy.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>

Postcopy: Postcopy startup in migration thread

Rework the migration thread to setup and start postcopy.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Amit Shah <amit.shah@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Signed-off-by: Juan Quintela <quintela@redhat.com>