vfio/pci: Don't remove irqchip notifier if not registered

The kvm irqchip notifier is only registered if the device supports
INTx, however it's unconditionally removed.  If the assigned device
does not support INTx, this will cause QEMU to crash when unplugging
the device from the system.  Change it to conditionally remove the
notifier only if the notify hook is setup.

CC: Eduardo Habkost <ehabkost@redhat.com>
CC: David Gibson <david@gibson.dropbear.id.au>
CC: Alex Williamson <alex.williamson@redhat.com>
Cc: qemu-stable@nongnu.org # v4.2
Reported-by: yanghliu@redhat.com
Debugged-by: Eduardo Habkost <ehabkost@redhat.com>
Fixes: c5478fea27ac ("vfio/pci: Respond to KVM irqchip change notifier")
Bugzilla: https://bugzilla.redhat.com/show_bug.cgi?id=1782678
Signed-off-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Greg Kurz <groug@kaod.org>
Signed-off-by: Alex Williamson <alex.williamson@redhat.com>

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

* More uses of RCU_READ_LOCK_GUARD (Dave, myself)
* QOM doc improvments (Greg)
* Cleanups from the Meson conversion (Marc-André)
* Support for multiple -accel options (myself)
* Many x86 machine cleanup (Philippe, myself)
* tests/migration-test cleanup (Juan)
* PC machine removal and next round of deprecation (Thomas)
* kernel-doc integration (Peter, myself)

# gpg: Signature made Wed 18 Dec 2019 01:35:02 GMT
# gpg:                using RSA key BFFBD25F78C7AE83
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* remotes/bonzini/tags/for-upstream: (87 commits)
  vga: cleanup mapping of VRAM for non-PCI VGA
  hw/display: Remove "rombar" hack from vga-pci and vmware_vga
  hw/pci: Remove the "command_serr_enable" property
  hw/audio: Remove the "use_broken_id" hack from the AC97 device
  hw/i386: Remove the deprecated machines 0.12 up to 0.15
  hw/pci-host: Add Kconfig entry to select the IGD Passthrough Host Bridge
  hw/pci-host/i440fx: Extract the IGD passthrough host bridge device
  hw/pci-host/i440fx: Use definitions instead of magic values
  hw/pci-host/i440fx: Use size_t to iterate over ARRAY_SIZE()
  hw/pci-host/i440fx: Extract PCII440FXState to "hw/pci-host/i440fx.h"
  hw/pci-host/i440fx: Correct the header description
  Fix some comment spelling errors.
  target/i386: remove unused pci-assign codes
  WHPX: refactor load library
  migration: check length directly to make sure the range is aligned
  memory: include MemoryListener documentation and some missing function parameters
  docs: add memory API reference
  memory.h: Silence kernel-doc complaints
  docs: Create bitops.rst as example of kernel-docs
  bitops.h: Silence kernel-doc complaints
  ...

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

Merge remote-tracking branch 'remotes/armbru/tags/pull-error-2019-12-17-v2' into staging

Error reporting patches for 2019-12-17

# gpg: Signature made Wed 18 Dec 2019 07:45:24 GMT
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [full]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [full]
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* remotes/armbru/tags/pull-error-2019-12-17-v2: (35 commits)
  nbd: assert that Error** is not NULL in nbd_iter_channel_error
  hw/vfio/ap: drop local_err from vfio_ap_realize
  backends/cryptodev: drop local_err from cryptodev_backend_complete()
  include/qom/object.h: rename Error ** parameter to more common errp
  hw/usb: rename Error ** parameter to more common errp
  hw/tpm: rename Error ** parameter to more common errp
  hw/sd: drop extra whitespace in sdhci_sysbus_realize() header
  hw/s390x: rename Error ** parameter to more common errp
  monitor/qmp-cmds: rename Error ** parameter to more common errp
  qga: rename Error ** parameter to more common errp
  hw/i386/amd_iommu: rename Error ** parameter to more common errp
  block/snapshot: rename Error ** parameter to more common errp
  hw/core/qdev: cleanup Error ** variables
  9pfs: make Error **errp const where it is appropriate
  ppc: make Error **errp const where it is appropriate
  Revert "ppc: well form kvmppc_hint_smt_possible error hint helper"
  qdev-monitor: make Error **errp const where it is appropriate
  vnc: drop Error pointer indirection in vnc_client_io_error
  hmp: drop Error pointer indirection in hmp_handle_error
  error: make Error **errp const where it is appropriate
  ...

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

nbd: assert that Error** is not NULL in nbd_iter_channel_error

All callers of nbd_iter_channel_error() pass the address of a
local_err variable, and only call this function if an error has
already occurred, using this function to propagate that error.
This is already implied by its name (local_err instead of the classic
errp), but it is worth additionally stressing this by adding an
assertion to make it part of the function contract.

The local_err parameter is not here to return information about
nbd_iter_channel_error failure. Instead it's assumed to be filled when
passed to the function. This is already stressed by its name
(local_err, instead of classic errp). Stress it additionally by
assertion.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20191205174635.18758-22-vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hw/vfio/ap: drop local_err from vfio_ap_realize

No reason for local_err here, use errp directly instead.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Message-Id: <20191205174635.18758-21-vsementsov@virtuozzo.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

backends/cryptodev: drop local_err from cryptodev_backend_complete()

No reason for local_err here, use errp directly instead.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Gonglei <arei.gonglei@huawei.com>
Message-Id: <20191205174635.18758-20-vsementsov@virtuozzo.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

include/qom/object.h: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20191205174635.18758-19-vsementsov@virtuozzo.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hw/usb: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20191205174635.18758-18-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hw/tpm: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Stefan Berger <stefanb@linux.ibm.com>
Message-Id: <20191205174635.18758-17-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hw/sd: drop extra whitespace in sdhci_sysbus_realize() header

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20191205174635.18758-16-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hw/s390x: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Message-Id: <20191205174635.18758-15-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

monitor/qmp-cmds: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20191205174635.18758-14-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qga: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20191205174635.18758-13-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hw/i386/amd_iommu: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20191205174635.18758-12-vsementsov@virtuozzo.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

block/snapshot: rename Error ** parameter to more common errp

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@Redhat.com>
Message-Id: <20191205174635.18758-11-vsementsov@virtuozzo.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hw/core/qdev: cleanup Error ** variables

Rename Error ** parameter in check_only_migratable to common errp.

In device_set_realized:

 - Move "if (local_err != NULL)" closer to error setters.

 - Drop 'Error **local_errp': it doesn't save any LoCs, but it's very
   unusual.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191205174635.18758-10-vsementsov@virtuozzo.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

9pfs: make Error **errp const where it is appropriate

Mostly, Error ** is for returning error from the function, so the
callee sets it. However error_append_security_model_hint and
error_append_socket_sockfd_hint get already filled errp
parameter. They don't change the pointer itself, only change the
internal state of referenced Error object. So we can make it Error
*const * errp, to stress the behavior. It will also help coccinelle
script (in future) to distinguish such cases from common errp usage.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Acked-by: Greg Kurz <groug@kaod.org>
Message-Id: <20191205174635.18758-9-vsementsov@virtuozzo.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Commit message replaced]
Signed-off-by: Markus Armbruster <armbru@redhat.com>

ppc: make Error **errp const where it is appropriate

Mostly, Error ** is for returning error from the function, so the
callee sets it. However kvmppc_hint_smt_possible gets already filled
errp parameter. It doesn't change the pointer itself, only change the
internal state of referenced Error object. So we can make it Error
*const * errp, to stress the behavior. It will also help coccinelle
script (in future) to distinguish such cases from common errp usage.

While there, rename the function to
kvmppc_error_append_smt_possible_hint().

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Greg Kurz <groug@kaod.org>
Message-Id: <20191205174635.18758-8-vsementsov@virtuozzo.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Commit message replaced]
Signed-off-by: Markus Armbruster <armbru@redhat.com>

Revert "ppc: well form kvmppc_hint_smt_possible error hint helper"

This reverts commit cdcca22aabafc0496894ce05c80097684832c7d9.

Commit cdcca22aaba is a superseded version of the next commit that
crept in by accident.  Revert it, so the final version applies.

Signed-off-by: Markus Armbruster <armbru@redhat.com>

qdev-monitor: make Error **errp const where it is appropriate

Mostly, Error ** is for returning error from the function, so the
callee sets it. However qbus_list_bus and qbus_list_dev get already
filled errp parameter. They don't change the pointer itself, only
change the internal state of referenced Error object. So we can make
it Error *const * errp, to stress the behavior. It will also help
coccinelle script (in future) to distinguish such cases from common
errp usage.

While there, rename the functions to
qbus_error_append_bus_list_hint(), qbus_error_append_dev_list_hint().

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20191205174635.18758-7-vsementsov@virtuozzo.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Commit message replaced]
Signed-off-by: Markus Armbruster <armbru@redhat.com>

vnc: drop Error pointer indirection in vnc_client_io_error

We don't need Error **, as all callers pass local Error object, which
isn't used after the call, or NULL. Use Error * instead.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191205174635.18758-6-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

hmp: drop Error pointer indirection in hmp_handle_error

We don't need Error **, as all callers pass local Error object, which
isn't used after the call. Use Error * instead.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Acked-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191205174635.18758-5-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

error: make Error **errp const where it is appropriate

Mostly, Error ** is for returning error from the function, so the
callee sets it. However these three functions get already filled errp
parameter. They don't change the pointer itself, only change the
internal state of referenced Error object. So we can make it
Error *const * errp, to stress the behavior. It will also help
coccinelle script (in future) to distinguish such cases from common
errp usage.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Message-Id: <20191205174635.18758-4-vsementsov@virtuozzo.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
[Commit message typo fixed]
Signed-off-by: Markus Armbruster <armbru@redhat.com>

net/net: Clean up variable shadowing in net_client_init()

Variable int err in inner scope shadows Error *err in outer scope.

Signed-off-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20191205174635.18758-3-vsementsov@virtuozzo.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

tests-blockjob: Use error_free_or_abort()

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Message-Id: <20191204093625.14836-19-armbru@redhat.com>

hw/intc/s390: Simplify error handling in kvm_s390_flic_realize()

Cc: Halil Pasic <pasic@linux.ibm.com>
Cc: Cornelia Huck <cohuck@redhat.com>
Cc: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Acked-by: Halil Pasic <pasic@linux.ibm.com>
Message-Id: <20191204093625.14836-18-armbru@redhat.com>

error: Clean up unusual names of Error * variables

Local Error * variables are conventionally named @err or @local_err,
and Error ** parameters @errp.  Naming local variables like parameters
is confusing.  Clean that up.

Naming parameters like local variables is also confusing.  Left for
another day.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20191204093625.14836-17-armbru@redhat.com>

memory-device: Fix memory pre-plug error API violations

memory_device_get_free_addr() dereferences @errp when
memory_device_check_addable() fails.  That's wrong; see the big
comment in error.h.  Introduced in commit 1b6d6af21b "pc-dimm: factor
out capacity and slot checks into MemoryDevice".

No caller actually passes null.

Fix anyway: splice in a local Error *err, and error_propagate().

Cc: David Hildenbrand <david@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Message-Id: <20191204093625.14836-11-armbru@redhat.com>

qga: Fix guest-get-fsinfo error API violations

build_guest_fsinfo_for_virtual_device() dereferences @errp when
build_guest_fsinfo_for_device() fails.  That's wrong; see the big
comment in error.h.  Introduced in commit 46d4c5723e "qga: Add
guest-get-fsinfo command".

No caller actually passes null.

Fix anyway: splice in a local Error *err, and error_propagate().

Cc: Michael Roth <mdroth@linux.vnet.ibm.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191204093625.14836-10-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

hw/ipmi: Fix realize() error API violations

isa_ipmi_bt_realize(), ipmi_isa_realize(), pci_ipmi_bt_realize(), and
pci_ipmi_kcs_realize() dereference @errp when IPMIInterfaceClass
method init() fails.  That's wrong; see the big comment in error.h.
Introduced in commit 0719029c47 "ipmi: Add an ISA KCS low-level
interface", then imitated in commit a9b74079cb "ipmi: Add a BT
low-level interface" and commit 12f983c6aa "ipmi: Add PCI IPMI
interfaces".

No caller actually passes null.

Fix anyway: splice in a local Error *err, and error_propagate().

Cc: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191204093625.14836-9-armbru@redhat.com>

hw/core: Fix fit_load_fdt() error API violations

fit_load_fdt() passes @errp to fit_image_addr(), then recovers from
ENOENT failures.  Passing @errp is wrong, because it works only as
long as @errp is neither @error_fatal nor @error_abort.  Error
recovery dereferences @errp.  That's also wrong; see the big comment
in error.h.  Error recovery can leave *errp pointing to a freed
Error object.  Wrong, it must be null on success.  Messed up in
commit 3eb99edb48 "loader-fit: Wean off error_printf()".

No caller actually passes such values, or uses *errp on success.

Fix anyway: splice in a local Error *err, and error_propagate().

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191204093625.14836-8-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

hw/acpi: Fix legacy CPU plug error API violations

legacy_acpi_cpu_plug_cb() dereferences @errp when
acpi_set_cpu_present_bit() fails.  That's wrong; see the big comment
in error.h.  Introduced in commit cc43364de7 "acpi/cpu-hotplug:
introduce helper function to keep bit setting in one place".

No caller actually passes null, and acpi_set_cpu_present_bit() can't
actually fail.

Fix anyway: drop acpi_set_cpu_present_bit()'s @errp parameter.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20191204093625.14836-7-armbru@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>

exec: Fix file_ram_alloc() error API violations

When os_mem_prealloc() fails, file_ram_alloc() calls qemu_ram_munmap()
and returns null.  Except it doesn't when its @errp argument is null,
because it checks for failure with (errp && *errp).  Introduced in
commit 056b68af77 "fix qemu exit on memory hotplug when allocation
fails at prealloc time".

No caller actually passes null.

Fix anyway: splice in a local Error *err, and error_propagate().

Cc: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20191204093625.14836-6-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>

tests: Clean up initialization of Error *err variables

Declaring a local Error *err without initializer looks suspicious.
Fuse the declaration with the initialization to avoid that.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191204093625.14836-5-armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

io: Fix Error usage in a comment <example>

Cc: "Daniel P. Berrangé" <berrange@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191204093625.14836-4-armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

crypto: Fix typo in QCryptoTLSSession's <example> comment

Cc: "Daniel P. Berrangé" <berrange@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191204093625.14836-3-armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

crypto: Fix certificate file error handling crash bug

qcrypto_tls_creds_load_cert() passes uninitialized GError *gerr by
reference to g_file_get_contents().  When g_file_get_contents() fails,
it'll try to set a GError.  Unless @gerr is null by dumb luck, this
logs a ERROR_OVERWRITTEN_WARNING warning message and leaves @gerr
unchanged.  qcrypto_tls_creds_load_cert() then dereferences the
uninitialized @gerr.

Fix by initializing @gerr properly.

Fixes: 9a2fd4347c40321f5cbb4ab4220e759fcbf87d03
Cc: "Daniel P. Berrangé" <berrange@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191204093625.14836-2-armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@virtuozzo.com>

error: Fix -msg timestamp default

-msg parameter "timestamp" defaults to "off" if you don't specify msg,
and to "on" if you do.  Messed up right in commit 5e2ac51917 "add
timestamp to error_report()".  Mostly harmless, because "timestamp" is
the only parameter, so "if you do" is "-msg ''", which nobody does.

Change the default to "off" no matter what.

While there, rename enable_timestamp_msg to error_with_timestamp, and
polish documentation.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20191010081508.8978-1-armbru@redhat.com>

vga: cleanup mapping of VRAM for non-PCI VGA

vga_init_vbe is now used only from ISA VGA cards.  Since the alias is
not needed anymore, remove it (effectively reverting commit 8294a64d7f,
"vga: fix vram double-mapping with -vga std and -M pc-0.12", 2012-05-29)
and the now unused vbe_mapped field of VGACommonState.  The function now
consists of a single memory_region_add_subregion call, so we can inline
it; this avoids incorrect usage from PCI cards.

Suggested-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Based-on: <05af415a-5058-98b4-4a12-9d093a30b1e3@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/display: Remove "rombar" hack from vga-pci and vmware_vga

Now that the old pc-0.x machine types have been removed, we do not need
the old "rombar" hacks anymore.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209125248.5849-5-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/pci: Remove the "command_serr_enable" property

Now that the old pc-0.x machine types have been removed, this config
knob is not required anymore.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209125248.5849-4-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/audio: Remove the "use_broken_id" hack from the AC97 device

Now that the old pc-0.x machine types are gone, we do not need
the "use_broken_id" hack anymore.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209125248.5849-3-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/i386: Remove the deprecated machines 0.12 up to 0.15

These machines can't be used reliably for migration anymore, quoting
https://lists.gnu.org/archive/html/qemu-devel/2018-12/msg04516.html :

"
 due to the introduction of the memory API, the firmware is not
 migrated correctly from source to destination.  On QEMU <1.3 the
 0xf0000-0xfffff area is basically a copy of the higher
 0xffff0000-0xffffffff area, while on more recent versions it is
 initialized with zeroes and the firmware copies from 0xffff0000 to
 0xf0000.  When you migrate from old to new QEMU, after reboot there's
 nothing at 0xf0000 and bugs ensue.
"

The pc-0.x machines have been marked as deprecated since QEMU v4.0, so
it is time to remove them now.

And while we're at it, mark the remaining pc-1.x machine types
as deprecated now, too, so that we finally only have "pc-i440fx"
and "pc-q35" machine types left (apart from the non-versioned
"isapc" and "microvm") once we remove them in a couple of releases.

Reviewed-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209125248.5849-2-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/pci-host: Add Kconfig entry to select the IGD Passthrough Host Bridge

Add the XEN_IGD_PASSTHROUGH Kconfig option.

Xen build has that option selected by default. Non-Xen builds now
have to select this feature manually.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20191209095002.32194-7-philmd@redhat.com>
Acked-by: Paul Durrant <paul@xen.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/pci-host/i440fx: Extract the IGD passthrough host bridge device

We can use a i440FX without the IGD passthrough host bridge.
Extract it into a new file, 'hw/pci-host/xen_igd_pt.c'.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20191209095002.32194-6-philmd@redhat.com>
Acked-by: Paul Durrant <paul@xen.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/pci-host/i440fx: Use definitions instead of magic values

Use definitions from "hw/pci/pci_regs.h".
This also helps when using git-grep.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209095002.32194-5-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/pci-host/i440fx: Use size_t to iterate over ARRAY_SIZE()

We don't enforce the -Wsign-conversion CPPFLAG, but it doesn't hurt
to avoid this warning:

  warning: implicit conversion changes signedness: 'int' to 'size_t' (aka 'unsigned long') [-Wsign-conversion]

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209095002.32194-4-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/pci-host/i440fx: Extract PCII440FXState to "hw/pci-host/i440fx.h"

Make the PCII440FXState structure public, so it can be used out of
this source file. This will allow us to extract the IGD Passthrough
Host Bridge, which is a children of the TYPE_I440FX_PCI_DEVICE.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209095002.32194-3-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/pci-host/i440fx: Correct the header description

Missed during the refactor in commits 14a026dd58 and 0f25d865a,
this file is now only about the i440FX chipset.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20191209095002.32194-2-philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Fix some comment spelling errors.

Signed-off-by: Cameron Esfahani <dirty@apple.com>
Message-Id: <086c197db928384b8697edfa64755e2cb46c8100.1575685843.git.dirty@apple.com>
Reviewed-by: Stefan Weil <sw@weilnetz.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386: remove unused pci-assign codes

Legacy PCI device assignment has been already removed in commit ab37bfc7d641
("pci-assign: Remove"), but some codes remain unused.

CC: qemu-trivial@nongnu.org
Signed-off-by: Eiichi Tsukata <devel@etsukata.com>
Message-Id: <20191209072932.313056-1-devel@etsukata.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

WHPX: refactor load library

This refactors the load library of WHV libraries to make it more
modular. It makes a helper routine that can be called on demand.
This allows future expansion of load library/functions to support
functionality that is dependent on some feature being available.

Signed-off-by: Sunil Muthuswamy <sunilmut@microsoft.com>
Message-Id: <MW2PR2101MB1116578040BE1F0C1B662318C0760@MW2PR2101MB1116.namprd21.prod.outlook.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

migration: check length directly to make sure the range is aligned

Since the start addr is already checked, to make sure the range is
aligned, checking the length is enough.

Signed-off-by: Wei Yang <richardw.yang@linux.intel.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Message-Id: <20190712032704.7826-1-richardw.yang@linux.intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

memory: include MemoryListener documentation and some missing function parameters

These cover the remaining warnings from kernel-doc.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Message-Id: <20191029162248.13383-7-pbonzini@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

docs: add memory API reference

Add kernel-doc directive to parse and include doc comments from
include/exec/memory.h.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

memory.h: Silence kernel-doc complaints

Fix a few instances where kernel-doc complains about doc comments
in memory.h.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

docs: Create bitops.rst as example of kernel-docs

Create a bitops.rst which is just a container for the
kernel-doc comments in qemu/bitops.h.

This is mostly a test of the kernel-doc extension machinery.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20190521122519.12573-7-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

bitops.h: Silence kernel-doc complaints

Fix the problems with kernel-doc/sphinx syntax in the
doc comments for the shuffle and unshuffle functions:
 * mismatch between comment and prototype for argument name
 * the inline bit patterns need to be marked up so they
   are processed properly and rendered as monospace

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20190521122519.12573-6-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Makefile: disable Sphinx nitpicking

Turn off Sphinx nitpicking as a temporary (?) measure so
sphinx builds complete even with warnings about missing
references.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20190521122519.12573-11-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

docs/conf.py: Enable use of kerneldoc sphinx extension

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20190521122519.12573-4-peter.maydell@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

docs: tweak kernel-doc for QEMU coding standards

Surprisingly, QEMU does have a pretty consistent doc comment style and
it is not very different from the Linux kernel's.  Of the documentation
"sigils", only "#" separates the QEMU doc comment style from Linux's,
and it has 200+ instances vs. 6 for the kernel's '&struct foo' (all in
accel/tcg/translate-all.c), so it's clear that the two standards are
different in this respect.  In addition, our structs are typedefed and
recognized by CamelCase names.

Adjust kernel-doc's parser for these two aspects of the QEMU coding
standards.  The patch has been valid, with hardly any change, for over
two years, so it should not be an issue to keep kernel-doc in sync with
the Linux copy.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

docs: import Linux kernel-doc script and extension

Import Linux's kernel-doc script as of commit 15e2544ed38a1e, as well
as the Sphinx extension to call kernel-doc according to the arguments
and parameters given to a reStructuredText directive.

The kernel-doc extension accepts a filename, which is relative to
the QEMU source tree root.  The extension also notifies Sphinx about the
document dependency on the file, causing the document to be rebuilt when
the file has been changed.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

build: rename CONFIG_LIBCAP to CONFIG_LIBCAP_NG

Since we are actually testing for the newer capng library, rename the
symbol to match.

Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

colo: fix return without releasing RCU

Use WITH_RCU_READ_LOCK_GUARD to avoid exiting colo_init_ram_cache
without releasing RCU.

Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

memory: use RCU_READ_LOCK_GUARD

Cc: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

qsp: Use WITH_RCU_READ_LOCK_GUARD

The automatic rcu read lock maintenance works quite
nicely in this case where it previously relied on a comment to
delimit the lifetime and now has a block.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hyperv: Use auto rcu_read macros

Use RCU_READ_LOCK_GUARD and WITH_RCU_READ_LOCK_GUARD
to replace the manual rcu_read_(un)lock calls.

Signed-off-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/i386/pc: Extract the port92 device

This device is only used by the PC machines. The pc.c file is
already big enough, with 2255 lines. By removing 113 lines of
it, we reduced it by 5%. It is now a bit easier to navigate
the file.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/i386/pc: Inline port92_init()

This one-line function is not very helpful, so remove it
by inlining the call to qdev_connect_gpio_out_named().

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/i386/pc: Use TYPE_PORT92 instead of hardcoded string

By using the TYPE_* definitions for devices, we can:
- quickly find where devices are used with 'git-grep'
- easily rename a device (one-line change).

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/isa/isa-bus: cleanup irq functions

The irq number is unsigned; we reject negative values.  But -1
is used for the isairq array, which is declared unsigned!  And
since we have a definition for the number of ISA IRQs, use it.

Based on a patch by Philippe Mathieu-Daudé.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/i386: Simplify ioapic_init_gsi()

All callers of ioapic_init_gsi() provide a parent. We want new
uses to follow the same good practice and provide the parent
name, so do not make this optional: assert the parent name is
provided, and simplify the code.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/i386: De-duplicate gsi_handler() to remove kvm_pc_gsi_handler()

Both gsi_handler() and kvm_pc_gsi_handler() have the same content,
except one comment. Move the comment, and de-duplicate the code.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

pc: stubify x86 iommu

Allow building microvm without x86-iommu.c.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

acpi: move PC stubs out of stubs/

This is a small cleanup that lets microvm build entirely without
include/hw/i386/pc.h.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

x86: move more x86-generic functions out of PC files

These are needed by microvm too, so move them outside of PC-specific files.
With this patch, microvm.c need not include pc.h anymore.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/i386/pc: Convert DPRINTF() to trace events

Convert the deprecated DPRINTF() macro to trace events.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

x86: move SMM property to X86MachineState

Add it to microvm as well, it is a generic property of the x86
architecture.

Suggested-by: Sergio Lopez <slp@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

pci-stub: add more MSI functions

On x86, KVM needs some function from the PCI subsystem in order to set
up interrupt routes.  Provide some stubs to support x86 machines that
lack PCI.

Reviewed-by: Sergio Lopez <slp@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw: replace hw/i386/pc.h with a header just for the i8259

Remove the need to include i386/pc.h to get to the i8259 functions.
This is enough to remove the inclusion of hw/i386/pc.h from all non-x86
files.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

fw_cfg: allow building without other devices

The microvm machine type uses fw_cfg but lacks SMBIOS and ACPI.  Do not
include the files if the symbol is not present in QEMU and remove
dependencies on machine-specific files.

Reviewed-by: Sergio Lopez <slp@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

i386: conditionally compile more files

Reviewed-by: Sergio Lopez <slp@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: simplify vhost condition with Kconfig

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

configure: set $PYTHON to a full path

This will make it possible to replace it in a shebang line.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

stubs: replace stubs with lnot if applicable

The stubs mechanism relies on static libraries and compilation order,
which is a bit brittle and should be avoided unless necessary.
Replace it with Boolean operations on CONFIG_* symbols.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

crypto: move common bits for all emulators to libqemuutil

qcrypto_random_*, AES and qcrypto_init do not need to be linked as a whole
and are the only parts that are used by user-mode emulation.  Place them
in libqemuutil, so that whatever needs them will pick them up automatically.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

libvixl: remove per-target compiler flags

We are already including -D__STDC_LIMIT_MACROS in the global CXXFLAGS,
so it makes sense to do the same for -D__STDC_CONSTANT_MACROS and
-D__STDC_FORMAT_MACROS instead of limiting that to libvixl.

The -Wno-sign-compare option can also be removed since GCC 4.6 is not
supported anymore.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

tests: skip block layer tests if !CONFIG_TOOLS

The block tests, as well as ahci-test needs qemu-img.  Do not run
them if it wasn't built.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

os-posix: simplify os_find_datadir

Use g_build_filename instead of sprintf, and g_autofree instead of
manual freeing.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

vhost-user-scsi: fix printf format warning

Fixes:
../contrib/vhost-user-scsi/vhost-user-scsi.c:118:57: error: format specifies
      type 'unsigned char' but the argument has type 'int' [-Werror,-Wformat]
    g_warning("Unable to determine cdb len (0x%02hhX)", cdb[0] >> 5);

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

monitor: fix maybe-uninitialized

../monitor/misc.c: In function ‘mon_get_cpu_sync’:
/home/elmarco/src/qq/include/sysemu/hw_accel.h:22:9: error: ‘cpu’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
   22 |         kvm_cpu_synchronize_state(cpu);
      |         ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
../monitor/misc.c:397:15: note: ‘cpu’ was declared here
  397 |     CPUState *cpu;
      |               ^~~

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

migration: fix maybe-uninitialized warning

../migration/ram.c: In function ‘multifd_recv_thread’:
/home/elmarco/src/qq/include/qapi/error.h:165:5: error: ‘block’ may be used uninitialized in this function [-Werror=maybe-uninitialized]
  165 |     error_setg_internal((errp), __FILE__, __LINE__, __func__,   \
      |     ^~~~~~~~~~~~~~~~~~~
../migration/ram.c:818:15: note: ‘block’ was declared here
  818 |     RAMBlock *block;
      |               ^~~~~

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

build-sys: do not include Windows SLIRP dependencies in $LIBS

When including the internal SLIRP library, we should add all the libraries that
it needs for the build.  Right now they are all included by QEMU, but -liphlpapi
is not needed without slirp.  Move it from LIBS to slirp_libs.

Based on a patch by Marc-André Lureau.

Acked-by: Samuel Thibault <samuel.thibault@ens-lyon.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

build-sys: build vhost-user-gpu only if CONFIG_TOOLS

vhost-user-gpu is always built and installed, but it is not part of the emulator
proper.  Cut it if --disable-tools is specified.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

object: Improve documentation of interfaces

QOM interfaces allow a limited form of multiple inheritance, at the
condition of being stateless. That is, they cannot be instantiated
and a pointer to an interface shouldn't be dereferenceable in any way.
This is achieved by making the QOM instance type an incomplete type,
which is, as mentioned by Markus Armbruster, the closest you can get
to abstract class in C.

Incomplete types are widely used to hide implementation details, but
people usually expect to find at least one place where the type is
fully defined. The fact that it doesn't happen with QOM interfaces is
quite disturbing, especially since it isn't documented anywhere as
recently discussed in this thread:

https://lists.gnu.org/archive/html/qemu-devel/2019-12/msg01579.html

Amend the documentation in the object.h header file to provide more
details about why and how to implement QOM interfaces using incomplete
types.

Signed-off-by: Greg Kurz <groug@kaod.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Makefile: remove unused variables

Reviewed-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

kvm: convert "-machine kernel_irqchip" to an accelerator property

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

kvm: introduce kvm_kernel_irqchip_* functions

The KVMState struct is opaque, so provide accessors for the fields
that will be moved from current_machine to the accelerator.  For now
they just forward to the machine object, but this will change.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

kvm: convert "-machine kvm_shadow_mem" to an accelerator property

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

xen: convert "-machine igd-passthru" to an accelerator property

The first machine property to fall is Xen's Intel integrated graphics
passthrough.  The "-machine igd-passthru" option does not set anymore
a property on the machine object, but desugars to a GlobalProperty on
accelerator objects.

The setter is very simple, since the value ends up in a
global variable, so this patch also provides an example before the more
complicated cases that follow it.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>