hw/pci/aer: Add missing routing for AER errors

PCIe r6.0 Figure 6-3 "Pseudo Logic Diagram for Selected Error Message Control
and Status Bits" includes a right hand branch under "All PCI Express devices"
that allows for messages to be generated or sent onwards without SERR#
being set as long as the appropriate per error class bit in the PCIe
Device Control Register is set.

Implement that branch thus enabling routing of ERR_COR, ERR_NONFATAL
and ERR_FATAL under OSes that set these bits appropriately (e.g. Linux)

Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Message-Id: <20230302133709.30373-3-Jonathan.Cameron@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>

hw/pci/aer: Implement PCI_ERR_UNCOR_MASK register

This register in AER should be both writeable and should
have a default value with a couple of the errors masked
including the Uncorrectable Internal Error used by CXL for
it's error reporting.

Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Message-Id: <20230302133709.30373-2-Jonathan.Cameron@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Fan Ni <fan.ni@samsung.com>

pcihp: add ACPI PCI hotplug specific is_hotpluggable_bus() callback

Provide pcihp specific callback to check if bus is hotpluggable
and consolidate its scattered hotplug criteria there.
While at it clean up no longer needed
   qbus_set_hotplug_handler(BUS(bus), NULL)
workarounds since callback makes qbus_is_hotpluggable() return
correct answer even if hotplug_handler is set on bus.

PS:
see ("pci: fix 'hotplugglable' property behavior") for details
why callback was introduced.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-35-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

pcihp: move fields enabling hotplug into AcpiPciHpState

... instead of duplicating them in piix4 and lpc and then
trying to pass them to pcihp routines as arguments.
it simplifies call sites and places pcihp specific in
its own structure.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-34-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi: pci: move out ACPI PCI hotplug generator from generic slot generator build_append_pci_bus_devices()

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-33-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi: pci: move BSEL into build_append_pcihp_slots()

Generic PCI enumeration code doesn't really need access to
BSEL value, it is only used as means to decide if hotplug
enumerator should be called.

Use stateless object_property_find() to do that, and move
the rest of BSEL handling into build_append_pcihp_slots()
where it belongs.

This cleans up generic code a bit from hotplug stuff
and follow up patch will remove remaining call to
build_append_pcihp_slots() from generic code, making
it possible to use without ACPI PCI hotplug dependencies.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-32-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi: pci: drop BSEL usage when deciding that device isn't hotpluggable

previous commit ("pci: fix 'hotplugglable' property behavior") fixed
pcie root port's 'hotpluggable' property to behave consistently.

So we don't need a BSEL crutch anymore to see of device is not
hotpluggable, drop it from 'generic' PCI slots description handling.

BSEL is still used to decide if hotplug part should be called
but that will be moved out of generic code to hotplug one by
followup patches.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-31-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

pci: move acpi-index uniqueness check to generic PCI device code

acpi-index is now working with non-hotpluggable buses
(pci/q35 machine hostbridge), it can be used even if
ACPI PCI hotplug is disabled and as result acpi-index
uniqueness check will be omitted (since the check is
done by ACPI PCI hotplug handler, which isn't wired
when ACPI PCI hotplug is disabled).
Move check and related code to generic PCIDevice so it
would be independent of ACPI PCI hotplug.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-30-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

an extra devices at non-zero function address with static
_DSM method get exposed, ex:

  +            Device (S15)
  +            {
  +                Name (_ADR, 0x00020005)  // _ADR: Address
  +                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  +                {
  +                    Local0 = Package (0x01)
  +                        {
  +                            0x66
  +                        }
  +                    Return (EDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  +                }
  +            }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-29-imammedo@redhat.com>

tests: acpi: add non zero function device with acpi-index on non-hotpluggble bus

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-28-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist DSDT before adding non-0 function device with acpi-index to testcases

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-27-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

in PC machine case piix3-ide and PIIX4_PM get exposed

  +            Device (S09)
  +            {
  +                Name (_ADR, 0x00010001)  // _ADR: Address
  +            }
  +
  +            Device (S0B)
  +            {
  +                Name (_ADR, 0x00010003)  // _ADR: Address
  +            }

in q35 machine case ich9-ahci gets exposed
  +            Device (SFA)
  +            {
  +                Name (_ADR, 0x001F0002)  // _ADR: Address
  +            }

and addtional pci-testdev, virtio-balloon exposed in q35 multi-bridge test case
  +            Device (S14)
  +            {
  +                Name (_ADR, 0x00020004)  // _ADR: Address
  +            }
  +
  ...
  +            Device (S22)
  +            {
  +                Name (_ADR, 0x00040002)  // _ADR: Address
  +            }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-26-imammedo@redhat.com>

acpi: pci: describe all functions on populated slots

describing all present devices on functions other than
0 was complicated when non hotplug and hotplug code
was intermixed. So QEMU has been excluding non zero
functions since they are not supported by hotplug code,
then a condition to whitelist coldplugged bridges was
added and later whitelisting of devices that advertise
presence of their own AML description.

With non hotplug and hotplug code separated, it is
possible to relax rules and allow describing all
non-hotpluggble functions and hence simplify
conditions whether PCI device should be enumerated by
generic (non-hotplug) code.

Price of that simplification is an extra few Device()
descriptors in DSDT exposing built-in chipset functions,
which has no functional effect on guest side.

Apart from that, the enumeration of non zero functions,
allows to attach more NICs with acpi-index enabled
directly on hostbridge (if hotplug is not required).

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-25-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist DSDT before exposing non zero functions

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-24-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

the only chenge is addition of _DSM- > EDSM method
on non-hotpluggable devices with configured acpi-index.
Something like:

  +                Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  +                {
  +                    Local0 = Package (0x01)
  +                        {
  +                            0x65
  +                        }
  +                    Return (EDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  +                }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-23-imammedo@redhat.com>

acpi: pci: support acpi-index for non-hotpluggable devices

Inject static _DSM (EDSM) if non-hotpluggable device has
acpi-index configured on it.
It lets use acpi-index non-hotpluggable devices / devices
attached to non-hotpluggable bus.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-22-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: add device with acpi-index on non-hotpluggble bus

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-21-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist DSDT before adding device with acpi-index to testcases

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-20-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-19-imammedo@redhat.com>

acpi: pci: add EDSM method to DSDT

it's a helper method for acpi-index support on PCI buses
that do no support or have disabled ACPI PCI hotplug
or for non-hotpluggble endpoint devices.
(like non-hotpluggble NICs, integrated endpoints and
later for machines that do not support ACPI PCI hotplug)

no functional change, commit adds only EDSM method in DSDT
without any users. (the follow up patches will use it)

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-18-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist DSDT before adding EDSM method

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-17-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

only following context change:
 -  Local1 = Zero
    If ((Arg0 != ToUUID ("e5c937d0-3553-4d7a-9117-ea4d19c3434d") /* Device Labeling Interface */))
    {
        Return (Local0)
 ...
        Return (Local0)
    }

 +  Local1 = Zero
    Local2 = AIDX (DerefOf (Arg4 [Zero]), DerefOf (Arg4 [One]

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-16-imammedo@redhat.com>

pcihp: move PCI _DSM function 0 prolog into separate function

it will be reused by follow up patches that will implement
static _DSM for non-hotpluggable devices.

no functional AML change, only context one, where 'cap' (Local1)
initialization is moved after UUID/revision checks.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-15-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist DSDT blobs before isolating PCI _DSM func 0 prolog

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-14-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

pci: fix 'hotplugglable' property behavior

Currently the property may flip its state
during VM bring up or just doesn't work as
the name implies.

In particular with PCIE root port that has
'hotplug={on|off}' property, and when it's
turned off, one would expect
  'hotpluggable' == false
for any devices attached to it.
Which is not the case since qbus_is_hotpluggable()
used by the property just checks for presence
of any hotplug_handler set on bus.

The problem is that name BusState::hotplug_handler
from its inception is misnomer, as it handles
not only hotplug but also in many cases coldplug
as well (i.e. generic wiring interface), and
it's fine to have hotplug_handler set on bus
while it doesn't support hotplug (ex. pcie-slot
with hotplug=off).

Another case of root port flipping 'hotpluggable'
state when ACPI PCI hotplug is enabled in this
case root port with 'hotplug=off' starts as
hotpluggable and then later on, pcihp
hotplug_handler clears hotplug_handler
explicitly after checking root port's 'hotplug'
property.

So root-port hotpluggablity check sort of works
if pcihp is enabled but is broken if pcihp is
disabled.

One way to deal with the issue is to ask
hotplug_handler if bus it controls is hotpluggable
or not. To do that add is_hotpluggable_bus()
hook to HotplugHandler interface and use it in
'hotpluggable' property + teach pcie-slot to
actually look into 'hotplug' property state
before deciding if bus is hotpluggable.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-13-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

pcihp: piix4: do not redirect hotplug controller to piix4 when ACPI hotplug is disabled

commit [1] added ability to disable ACPI PCI hotplug
on hostbridge but forgot to take into account that it
should disable all ACPI hotplug machinery in case both
hostbridge and bridge hotplug are disabled.

Commit [2] tried to fix that, however it forgot to
remove hotplug_handler override which hands hotplug
control over to piix4 hotplug controller
(uninitialized after [2]).

As result at the time bridge is plugged in, its default
(SHPC) hotplug handler is replaced by piix4 one in
  acpi_pcihp_device_plug_cb()
    ...
    if (!s->legacy_piix &&
       ...
       qbus_set_hotplug_handler(BUS(sec), OBJECT(hotplug_dev));

which is acting on uninitialized s->legacy_piix value
(0 by default) that was supposed to be initialized by
acpi_pcihp_init(), that is no longer called due to
following condition being false:

  piix4_acpi_system_hot_add_init()
    if (s->use_acpi_hotplug_bridge || s->use_acpi_root_pci_hotplug) {

and the bridge ends up with piix4 as hotplug handler
instead of shpc one.

Followup hotplug on that bridge as result yields
piix4 specific error:

  Error: Unsupported bus. Bus doesn't have property 'acpi-pcihp-bsel' set

1) 3d7e78aa777 (Introduce a new flag for i440fx to disable PCI hotplug on the root bus)
2) df4008c9c59 (piix4: don't reserve hw resources when hotplug is off globally)

Fixes: df4008c9c59 (piix4: don't reserve hw resources when hotplug is off globally)
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-12-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

BNUM numbering changes across DSDT due to addition of new bridges.

Fixed missing PCI tree brunch (q35/DSDT.multi-bridge case):

  //  -device pcie-root-port,id=rpnohp,chassis=8,addr=0xA.0,hotplug=off
  +            Device (S50)
  +            {
  +                Name (_ADR, 0x000A0000)  // _ADR: Address
  //  -device pcie-root-port,id=rp3,chassis=9,bus=rpnohp
  +                Device (S00)
  +                {
  +                    Name (_ADR, Zero)  // _ADR: Address
  +                    Name (BSEL, Zero)
  +                    Device (S00)
  +                    {
  +                        Name (_ADR, Zero)  // _ADR: Address
  +                        Name (ASUN, Zero)
  +                        Method (_DSM, 4, Serialized)  // _DSM: Device-Specific Method
  +                        {
  +                            Local0 = Package (0x02)
  +                                {
  +                                    BSEL,
  +                                    ASUN
  +                                }
  +                            Return (PDSM (Arg0, Arg1, Arg2, Arg3, Local0))
  +                        }
  +
  +                        Name (_SUN, Zero)  // _SUN: Slot User Number
  +                        Method (_EJ0, 1, NotSerialized)  // _EJx: Eject Device
  +                        {
  +                            PCEJ (BSEL, _SUN)
  +                        }
  +                    }
  +
  +                    Method (DVNT, 2, NotSerialized)
  +                    {
  +                        If ((Arg0 & One))
  +                        {
  +                            Notify (S00, Arg1)
  +                        }
  +                    }
  +                }
  +            }

Fixed hotplug notification for leaf root port (hotplug=on) attached to
intermediate root port (hotplug=off) (q35/DSDT.multi-bridge case)

  //  -device pcie-root-port,id=rpnohp,chassis=8,addr=0xA.0,hotplug=off
  +        Scope (S50)
  +        {
  //  -device pcie-root-port,id=rp3,chassis=9,bus=rpnohp
  +            Scope (S00)
  +            {
  +                Method (PCNT, 0, NotSerialized)
  +                {
  +                    BNUM = Zero
  +                    DVNT (PCIU, One)
  +                    DVNT (PCID, 0x03)
  +                }
  +            }
  +
  +            Method (PCNT, 0, NotSerialized)
  +            {
  +                ^S00.PCNT ()
  +            }
  +        }
  ...
           Method (PCNT, 0, NotSerialized)
           {
  +            ^S50.PCNT ()
               ^S13.PCNT ()

Populated slots being described on coldplugged bridges even if
ACPI bridge hotplug is disabled.
(pc/DSDT.hpbridge and pc/DSDT.hpbrroot)
  ...
               Device (S18)
               {
                   Name (_ADR, 0x00030000)  // _ADR: Address
  +                Device (S08)
  +                {
  +                    Name (_ADR, 0x00010000)  // _ADR: Address
  +                }
  +
  +                Device (S10)
  +                {
  +                    Name (_ADR, 0x00020000)  // _ADR: Address
  +                }
               }
  ...
               Device (S18)
               {
                   Name (_ADR, 0x00030000)  // _ADR: Address
  +                Device (S00)
  +                {
  +                    Name (_ADR, Zero)  // _ADR: Address
  +                }
               }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-11-imammedo@redhat.com>

x86: pcihp: fix missing bridge AML when intermediate root-port has 'hotplug=off' set

(I practice [1] hasn't broke anything since on hardware side we unset
hotplug_handler on such intermediate port => hotplug behind it has
never worked)

When deciding if bridge should be described, the original
condition was

  cold_plugged_bridge && pcihp_bridge_en

which was replaced [1] by

  bridge has ACPI_PCIHP_PROP_BSEL

the later however is not the same thing as the original
and flips to false if intermediate bridge has hotplug
turned off (root-port with 'hotplug=off' option).

Since we already in build_pci_bridge_aml(), the question
if it's bridge is answered. Use DeviceState::hotplugged
to make decision if bridge should describe its slots.

What's left out is pcihp_bridge_en, which tells us if
ACPI bridge hotplug is enabled.

With hotplug and non hotplug part now being mostly
separated, omitting this check will only lead to
colplugged bridges describe occupied slots in case
when ACPI bridge hotplug is disabled.
Which makes behavior consistent with occupied slots
on hostbridge.

Ex (pc/DSDT.hpbrroot diff):
  ...
               Device (S20)
               {
                   Name (_ADR, 0x00040000)  // _ADR: Address
  +                Device (S08)
  +                {
  +                    Name (_ADR, 0x00010000)  // _ADR: Address
  +                }
  +
  +                Device (S10)
  +                {
  +                    Name (_ADR, 0x00020000)  // _ADR: Address
  +                }
               }
  ...

PS:
testing shows that above doesn't affect adversely guest OS
behavior: i.e. if ACPI bridge hotplug is enabled it's
expected behaviour, and with ACPI bridge hotplug is disabled
(a.k. native hotplug), it doesn't break slot enumeration
nor native hotplug. (tested with RHEL9.0 and WS2022).

1)
Fixes: 6c36ec46b0d ("pcihp: make bridge describe itself using AcpiDevAmlIfClass:build_dev_aml")
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-10-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist pc/DSDT.hpbrroot and pc/DSDT.hpbridge tests

follow up fix for missing root-port AML will affect these tests
by adding non-hotpluggable Device descriptors of colplugged
bridges when bridge hotplug is disabled.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-9-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

x86: pcihp: fix missing PCNT callchain when intermediate root-port has 'hotplug=off' set

Beside BSEL numbers change (due to 2 extra root-ports in q35/miltibridge test),
following change is expected:

       Scope (\_SB.PCI0)
       {
  ...
  +        Scope (S50)
  +        {
  +            Scope (S00)
  +            {
  +                Method (PCNT, 0, NotSerialized)
  +                {
  +                    BNUM = Zero
  +                    DVNT (PCIU, One)
  +                    DVNT (PCID, 0x03)
  +                }
  +            }
  +
  +            Method (PCNT, 0, NotSerialized)
  +            {
  +                ^S00.PCNT
  +            }
  +        }
  ...
           Method (PCNT, 0, NotSerialized)
           {
  +            ^S50.PCNT ()
               ^S13.PCNT ()
               ^S12.PCNT ()
               ^S11.PCNT ()

I practice [1] hasn't broke anything since on hardware side we unset
hotplug_handler on such intermediate port => hotplug behind it has
not been properly wired and as result not worked.

1)
Fixes: ddab4d3fae4e8 ("pcihp: compose PCNT callchain right before its user _GPE._E01")
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-8-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: extend multi-bridge case with case 'root-port,id=HOHP,hotplug=off root-port,bus=NOHP'

Following corner case wasn't covered:

  -device pcie-root-port,id=NO_HOTPLUG,hotplug=off
  -device pcie-root-port,bus=NO_HOTPLUG

when intermediate root-port has explicitly disabled hotplug,
all hierarchy below it is not described anymore (used to be
described in 7.2)

So as result we see only NO_HOTPLUG root-port described

  +            Device (S50)
  +            {
  +                Name (_ADR, 0x000A0000)  // _ADR: Address
  +            }

and no children nor notification chain for them are being composed.
Follow up patches will fix missing leaf root-port descriptor
and notification chain that should accompany it.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-7-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist q35/DSDT.multi-bridge before extending testcase

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-6-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: update expected blobs

expected changes:
Basically adds devices present on root bus in form:
  Device (SXX)
  {
     Name (_ADR, 0xYYYYYYYY)  // _ADR: Address
  }

On top of that For q35.noacpihp, all ACPI PCI hotplug
AML is removed and _OSC get native hotplug enabled:

                       CreateDWordField (Arg3, 0x04, CDW2)
                       CreateDWordField (Arg3, 0x08, CDW3)
                       Local0 = CDW3 /* \_SB_.PCI0._OSC.CDW3 */
  -                    Local0 &= 0x1E
  +                    Local0 &= 0x1F
                       If ((Arg1 != One))
                       {
                           CDW1 |= 0x08
Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-5-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: add test_acpi_q35_tcg_no_acpi_hotplug test and extend test_acpi_piix4_no_acpi_pci_hotplug

test bridge AML generator with ACPI PCI hotplug disabled
(i.e. with native hotplug enabled/disabled per bridge/root port)

PS:
while at make sure that devices on pci-bridge are starting
from addr=1.0 as slot 0 is not available there and test
passes only because of a bug in ACPI hotplug that will be
fixed by follow up patch

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-4-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests: acpi: whitelist new q35.noacpihp test and pc.hpbrroot

for q35.noacpihp use plain default Q35 DSDT table as a starting point.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-3-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Revert "tests/qtest: Check for devices in bios-tables-test"

This reverts commit c471eb4f40445908c1be7bb11a37ac676a0edae7.

which broke acpi tables test and rebuild due to skipping some tests
even thought none of devices tests depend on weren't disabled.

As result it leads to some expected tables not being updated,
merge conflicts and tests failure.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20230302161543.286002-2-imammedo@redhat.com>
Acked-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: return VHOST_F_LOG_ALL in vhost-vdpa devices

vhost-vdpa devices can return this feature now that blockers have been
set in case some features are not met.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20230303172445.1089785-15-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa net: allow VHOST_F_LOG_ALL

Since some actions move to the start function instead of init, the
device features may not be the parent vdpa device's, but the one
returned by vhost backend.  If transition to SVQ is supported, the vhost
backend will return _F_LOG_ALL to signal the device is migratable.

Add VHOST_F_LOG_ALL.  HW dirty page tracking can be added on top of this
change if the device supports it in the future.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20230303172445.1089785-14-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: block migration if SVQ does not admit a feature

Next patches enable devices to be migrated even if vdpa netdev has not
been started with x-svq. However, not all devices are migratable, so we
need to block migration if we detect that.

Block migration if we detect the device expose a feature SVQ does not
know how to work with.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-13-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: block migration if device has unsupported features

A vdpa net device must initialize with SVQ in order to be migratable at
this moment, and initialization code verifies some conditions.  If the
device is not initialized with the x-svq parameter, it will not expose
_F_LOG so the vhost subsystem will block VM migration from its
initialization.

Next patches change this, so we need to verify migration conditions
differently.

QEMU only supports a subset of net features in SVQ, and it cannot
migrate state that cannot track or restore in the destination.  Add a
migration blocker if the device offers an unsupported feature.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-12-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa net: block migration if the device has CVQ

Devices with CVQ need to migrate state beyond vq state.  Leaving this to
future series.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-11-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: disable RAM block discard only for the first device

Although it does not make a big difference, its more correct and
simplifies the cleanup path in subsequent patches.

Move ram_block_discard_disable(false) call to the top of
vhost_vdpa_cleanup because:
* We cannot use vhost_vdpa_first_dev after dev->opaque = NULL
  assignment.
* Improve the stack order in cleanup: since it is the last action taken
  in init, it should be the first at cleanup.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-10-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: add vdpa net migration state notifier

This allows net to restart the device backend to configure SVQ on it.

Ideally, these changes should not be net specific and they could be done
in:
* vhost_vdpa_set_features (with VHOST_F_LOG_ALL)
* vhost_vdpa_set_vring_addr (with .enable_log)
* vhost_vdpa_set_log_base.

However, the vdpa net backend is the one with enough knowledge to
configure everything because of some reasons:
* Queues might need to be shadowed or not depending on its kind (control
  vs data).
* Queues need to share the same map translations (iova tree).

Also, there are other problems that may have solutions but complicates
the implementation at this stage:
* We're basically duplicating vhost_dev_start and vhost_dev_stop, and
  they could go out of sync.  If we want to reuse them, we need a way to
  skip some function calls to avoid recursiveness (either vhost_ops ->
  vhost_set_features, vhost_set_vring_addr, ...).
* We need to traverse all vhost_dev of a given net device twice: one to
  stop and get the vq state and another one after the reset to
  configure properties like address, fd, etc.

Because of that it is cleaner to restart the whole net backend and
configure again as expected, similar to how vhost-kernel moves between
userspace and passthrough.

If more kinds of devices need dynamic switching to SVQ we can:
* Create a callback struct like VhostOps and move most of the code
  there.  VhostOps cannot be reused since all vdpa backend share them,
  and to personalize just for networking would be too heavy.
* Add a parent struct or link all the vhost_vdpa or vhost_dev structs so
  we can traverse them.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-9-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: move vhost reset after get vring base

The function vhost.c:vhost_dev_stop calls vhost operation
vhost_dev_start(false). In the case of vdpa it totally reset and wipes
the device, making the fetching of the vring base (virtqueue state) totally
useless.

The kernel backend does not use vhost_dev_start vhost op callback, but
vhost-user do. A patch to make vhost_user_dev_start more similar to vdpa
is desirable, but it can be added on top.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-8-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: add vhost_vdpa_suspend

The function vhost.c:vhost_dev_stop fetches the vring base so the vq
state can be migrated to other devices.  However, this is unreliable in
vdpa, since we didn't signal the device to suspend the queues, making
the value fetched useless.

Suspend the device if possible before fetching first and subsequent
vring bases.

Moreover, vdpa totally reset and wipes the device at the last device
before fetch its vrings base, making that operation useless in the last
device. This will be fixed in later patches of this series.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-7-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: add vhost_vdpa->suspended parameter

This allows vhost_vdpa to track if it is safe to get the vring base from
the device or not.  If it is not, vhost can fall back to fetch idx from
the guest buffer again.

No functional change intended in this patch, later patches will use this
field.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-6-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: rewind at get_base, not set_base

At this moment it is only possible to migrate to a vdpa device running
with x-svq=on. As a protective measure, the rewind of the inflight
descriptors was done at the destination. That way if the source sent a
virtqueue with inuse descriptors they are always discarded.

Since this series allows to migrate also to passthrough devices with no
SVQ, the right thing to do is to rewind at the source so the base of
vrings are correct.

Support for inflight descriptors may be added in the future.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20230303172445.1089785-5-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: Negotiate _F_SUSPEND feature

This is needed for qemu to know it can suspend the device to retrieve
its status and enable SVQ with it, so all the process is transparent to
the guest.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20230303172445.1089785-4-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa: Remember last call fd set

As SVQ can be enabled dynamically at any time, it needs to store call fd
always.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20230303172445.1089785-3-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vdpa net: move iova tree creation from init to start

Only create iova_tree if and when it is needed.

The cleanup keeps being responsible for the last VQ but this change
allows it to merge both cleanup functions.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20230303172445.1089785-2-eperezma@redhat.com>
Tested-by: Lei Yang <leiyang@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

MAINTAINERS: add myself as the maintainer for cryptodev

I developed the akcipher service, QoS setting, QMP/HMP commands and
statistics accounting for crypto device. Making myself as the
maintainer for QEMU's cryptodev.

Cc: Gonglei <arei.gonglei@huawei.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-13-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: Support query-stats QMP command

Now we can use "query-stats" QMP command to query statistics of
crypto devices. (Originally this was designed to show statistics
by '{"execute": "query-cryptodev"}'. Daniel Berrangé suggested that
querying configuration info by "query-cryptodev", and querying
runtime performance info by "query-stats". This makes sense!)

Example:
~# virsh qemu-monitor-command vm '{"execute": "query-stats", \
   "arguments": {"target": "cryptodev"} }' | jq
{
  "return": [
    {
      "provider": "cryptodev",
      "stats": [
        {
          "name": "asym-verify-bytes",
          "value": 7680
        },
        ...
        {
          "name": "asym-decrypt-ops",
          "value": 32
        },
        {
          "name": "asym-encrypt-ops",
          "value": 48
        }
      ],
      "qom-path": "/objects/cryptodev0" # support asym only
    },
    {
      "provider": "cryptodev",
      "stats": [
        {
          "name": "asym-verify-bytes",
          "value": 0
        },
        ...
        {
          "name": "sym-decrypt-bytes",
          "value": 5376
        },
        ...
      ],
      "qom-path": "/objects/cryptodev1" # support asym/sym
    }
  ],
  "id": "libvirt-422"
}

Suggested-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-12-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: support QoS

Add 'throttle-bps' and 'throttle-ops' limitation to set QoS. The
two arguments work with both QEMU command line and QMP command.

Example of QEMU command line:
-object cryptodev-backend-builtin,id=cryptodev1,throttle-bps=1600,\
throttle-ops=100

Example of QMP command:
virsh qemu-monitor-command buster --hmp qom-set /objects/cryptodev1 \
throttle-ops 100

or cancel limitation:
virsh qemu-monitor-command buster --hmp qom-set /objects/cryptodev1 \
throttle-ops 0

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-11-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: Account statistics

Account OPS/BPS for crypto device, this will be used for 'query-stats'
QEMU monitor command and QoS in the next step.

Note that a crypto device may support symmetric mode, asymmetric mode,
both symmetric and asymmetric mode. So we use two structure to
describe the statistics of a crypto device.

Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-10-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>

cryptodev: Use CryptoDevBackendOpInfo for operation

Move queue_index, CryptoDevCompletionFunc and opaque into struct
CryptoDevBackendOpInfo, then cryptodev_backend_crypto_operation()
needs an argument CryptoDevBackendOpInfo *op_info only. And remove
VirtIOCryptoReq from cryptodev. It's also possible to hide
VirtIOCryptoReq into virtio-crypto.c in the next step. (In theory,
VirtIOCryptoReq is a private structure used by virtio-crypto only)

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-9-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hmp: add cryptodev info command

Example of this command:
 # virsh qemu-monitor-command vm --hmp info cryptodev
cryptodev1: service=[akcipher|mac|hash|cipher]
    queue 0: type=builtin
cryptodev0: service=[akcipher]
    queue 0: type=lkcf

Acked-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-8-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev-builtin: Detect akcipher capability

Rather than exposing akcipher service/RSA algorithm to virtio crypto
device unconditionally, detect akcipher capability from akcipher
crypto framework. This avoids unsuccessful requests.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-7-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: Introduce 'query-cryptodev' QMP command

Now we have a QMP command to query crypto devices:
virsh qemu-monitor-command vm '{"execute": "query-cryptodev"}' | jq
{
  "return": [
    {
      "service": [
        "akcipher",
        "mac",
        "hash",
        "cipher"
      ],
      "id": "cryptodev1",
      "client": [
        {
          "queue": 0,
          "type": "builtin"
        }
      ]
    },
    {
      "service": [
        "akcipher"
      ],
      "id": "cryptodev0",
      "client": [
        {
          "queue": 0,
          "type": "lkcf"
        }
      ]
    }
  ],
  "id": "libvirt-417"
}

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-6-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: Introduce server type in QAPI

Introduce cryptodev service type in cryptodev.json, then apply this
to related codes. Now we can remove VIRTIO_CRYPTO_SERVICE_xxx
dependence from QEMU cryptodev.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-5-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: Introduce cryptodev alg type in QAPI

Introduce cryptodev alg type in cryptodev.json, then apply this to
related codes, and drop 'enum CryptoDevBackendAlgType'.

There are two options:
1, { 'enum': 'QCryptodevBackendAlgType',
  'prefix': 'CRYPTODEV_BACKEND_ALG',
  'data': ['sym', 'asym']}
Then we can keep 'CRYPTODEV_BACKEND_ALG_SYM' and avoid lots of
changes.
2, changes in this patch(with prefix 'QCRYPTODEV_BACKEND_ALG').

To avoid breaking the rule of QAPI, use 2 here.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-4-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: Remove 'name' & 'model' fields

We have already used qapi to generate crypto device types, this allows
to convert type to a string 'model', so the 'model' field is not
needed.

And the 'name' field is not used by any backend driver, drop it.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-3-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

cryptodev: Introduce cryptodev.json

Introduce QCryptodevBackendType in cryptodev.json, also apply this to
related codes. Then we can drop 'enum CryptoDevBackendOptionsType'.

Note that `CRYPTODEV_BACKEND_TYPE_NONE` is *NOT* used by anywhere, so
drop it(no 'none' enum in QCryptodevBackendType).

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: zhenwei pi <pizhenwei@bytedance.com>
Message-Id: <20230301105847.253084-2-pizhenwei@bytedance.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Merge tag 'pull-riscv-to-apply-20230306' of https://gitlab.com/palmer-dabbelt/qemu into staging

Sixth RISC-V PR for 8.0

* Support for the Zicbiom, ZCicboz, and Zicbop extensions.
* OpenSBI has been updated to version 1.2, see
  <https://github.com/riscv-software-src/opensbi/releases/tag/v1.2> for
  the release notes.
* Support for setting the virtual address width (ie, sv39/sv48/sv57) on
  the command line.
* Support for ACPI on RISC-V.

# -----BEGIN PGP SIGNATURE-----
#
# iQJHBAABCAAxFiEEKzw3R0RoQ7JKlDp6LhMZ81+7GIkFAmQGYGgTHHBhbG1lckBk
# YWJiZWx0LmNvbQAKCRAuExnzX7sYidmyEAC6FEMbbFM5D++qR6w6xM6hXgzcrev6
# s1kyRRNVa45uSA78ti/Zi0hsDLNf7ZsNPndF0OIkkO5iAE0OVm3LU7tV1TqKcT82
# Dd9VXxe93zEmfnuJazHrMa54SXPhhnNdWHtKlZ6vBfZpbxgx0FFs50xkCsrM5LQZ
# hYHxQUqPWQTvF2MdDHrxCuLcdKl+Wg3ysCcgRh2d049KUBrIu6vNaHC2+AGRjCbj
# BkrGCkB82fTmVJjzAcVWQxLoAV12pCbJS4og1GtP8hA7WevtB39tbPin9siBKRZp
# QBeiIsg0nebkpmZGrb+xWVwlIBNe9yYwJa0KmveQk8v7L5RIzjM1mtDL91VrVljC
# KC2tfT570m0Iq2NoFMb3wd/kESHFzVDM/g+XYqRd4KSoiCNP/RbqYNQBwbMc31Tr
# E27xfA1D8w2vem0Rk20x3KgPf1Z5OmGXjq6YObTpnAzG8cZlA37qKBP+ortt5aHX
# GZSg3CAwknHHVajd4aaegkPsHxm1tRvoTfh38MwkPSNxaA9GD0nz0k9xaYDmeZ2L
# olfanNsaQEwcVUId31+7sAENg1TZU0fnj879/nxkMUCazVTdL8/mz+IoTTx0QCST
# 3+9ATWcyJUlmjbDKIs7kr1L+wJdvvHEJggPAbbPI8ekpXaLZvUYOT6ObzYKNAmwY
# wELQBn8QKXcLVA==
# =5gAt
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 06 Mar 2023 21:51:36 GMT
# gpg:                using RSA key 2B3C3747446843B24A943A7A2E1319F35FBB1889
# gpg:                issuer "palmer@dabbelt.com"
# gpg: Good signature from "Palmer Dabbelt <palmer@dabbelt.com>" [unknown]
# gpg:                 aka "Palmer Dabbelt <palmerdabbelt@google.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 00CE 76D1 8349 60DF CE88  6DF8 EF4C A150 2CCB AB41
#      Subkey fingerprint: 2B3C 3747 4468 43B2 4A94  3A7A 2E13 19F3 5FBB 1889

* tag 'pull-riscv-to-apply-20230306' of https://gitlab.com/palmer-dabbelt/qemu: (22 commits)
  MAINTAINERS: Add entry for RISC-V ACPI
  hw/riscv/virt.c: Initialize the ACPI tables
  hw/riscv/virt: virt-acpi-build.c: Add RHCT Table
  hw/riscv/virt: virt-acpi-build.c: Add RINTC in MADT
  hw/riscv/virt: Enable basic ACPI infrastructure
  hw/riscv/virt: Add memmap pointer to RiscVVirtState
  hw/riscv/virt: Add a switch to disable ACPI
  hw/riscv/virt: Add OEM_ID and OEM_TABLE_ID fields
  riscv: Correctly set the device-tree entry 'mmu-type'
  riscv: Introduce satp mode hw capabilities
  riscv: Allow user to set the satp mode
  riscv: Change type of valid_vm_1_10_[32|64] to bool
  riscv: Pass Object to register_cpu_props instead of DeviceState
  roms/opensbi: Upgrade from v1.1 to v1.2
  gitlab/opensbi: Move to docker:stable
  hw: intc: Use cpu_by_arch_id to fetch CPU state
  target/riscv: cpu: Implement get_arch_id callback
  disas/riscv Fix ctzw disassemble
  hw/riscv/virt.c: add cbo[mz]-block-size fdt properties
  target/riscv: add Zicbop cbo.prefetch{i, r, m} placeholder
  ...

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

Merge tag 'pull-target-arm-20230306' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * allwinner-h3: Fix I2C controller model for Sun6i SoCs
 * allwinner-h3: Add missing i2c controllers
 * Expose M-profile system registers to gdbstub
 * Expose pauth information to gdbstub
 * Support direct boot for Linux/arm64 EFI zboot images
 * Fix incorrect stage 2 MMU setup validation

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmQGB+wZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3gdQEACVfgbs77mxbOb6u8yWHKGZ
# tVnQr9KZMv2lmwt5H3ROJPXznchrIIAwdMeRgKnbI+lC5jTq9L+Q8RJch3t/EbAd
# f0VMyiPe3DzCbCrAR9cW6EWzbYnEVo3Ioj4k7qjxK6u1BIKhXz99DLYd1KRdTxnx
# BAYmcl857Uir1q2FrBVMZ/ItCLbk4ejn+YaDIawNue2/s1oGa+we473x9rosCFvp
# L9bzT3R46e0o+Mfkn1OYRmgCmURTalWPpWAxyOUFR9YbrzXleLgAKEB3o3PPcvls
# u26uxztyRMqje1q06VjUzwaLw7zN9XPhmir+NXX7KXp2/x9PZjApOpPtt0kl+6qe
# FbByKfl24O9w/OKewsJw+udCBYdYrRPm6tWv2D71iAwjBUzBJgNGe5VPRdPFtPDn
# uSRO65o34w1nPzRpAheUciZueiabYrVmIgVltFxj0JlrKGfgiYHPLVyU0Uu0K/A7
# F2kUEQIzIcWdo+c8SlvlWOEA2ojVd/KoLVLgndqr40Tk5pbc65TRS08kkVVl4cMT
# jUGscl7Dyxe+yo8+nHdycAJpnKYDllJOh2JbGv3r2FqCy5FMuIqW4hHeuUxwpE+O
# nxm7lzjnaVHSAFHdzhk9x4E4uH/GTcdWzX1EsmpgGqe5oejLJOrCINb+Dj44+Y8h
# 8aGRvE7kxMs11upxc7BcAw==
# =KIMt
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 06 Mar 2023 15:34:04 GMT
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [ultimate]
# gpg:                 aka "Peter Maydell <peter@archaic.org.uk>" [ultimate]
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* tag 'pull-target-arm-20230306' of https://git.linaro.org/people/pmaydell/qemu-arm: (21 commits)
  hw: arm: allwinner-h3: Fix and complete H3 i2c devices
  hw: allwinner-i2c: Fix TWI_CNTR_INT_FLAG on SUN6i SoCs
  hw: arm: Support direct boot for Linux/arm64 EFI zboot images
  target/arm: Rewrite check_s2_mmu_setup
  target/arm: Diagnose incorrect usage of arm_is_secure subroutines
  target/arm: Stub arm_hcr_el2_eff for m-profile
  target/arm: Handle m-profile in arm_is_secure
  target/arm: Implement gdbstub m-profile systemreg and secext
  target/arm: Export arm_v7m_get_sp_ptr
  target/arm: Export arm_v7m_mrs_control
  target/arm: Implement gdbstub pauth extension
  target/arm: Create pauth_ptr_mask
  target/arm: Simplify iteration over bit widths
  target/arm: Add name argument to output_vector_union_type
  target/arm: Fix svep width in arm_gen_dynamic_svereg_xml
  target/arm: Hoist pred_width in arm_gen_dynamic_svereg_xml
  target/arm: Simplify register counting in arm_gen_dynamic_svereg_xml
  target/arm: Split out output_vector_union_type
  target/arm: Move arm_gen_dynamic_svereg_xml to gdbstub64.c
  target/arm: Unexport arm_gen_dynamic_sysreg_xml
  ...

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

Merge tag 'nvme-next-pull-request' of https://gitlab.com/birkelund/qemu into staging

hw/nvme updates

* basic support for directives
* simple support for endurance groups
* emulation of flexible data placement (tp4146)

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEUigzqnXi3OaiR2bATeGvMW1PDekFAmQF+doACgkQTeGvMW1P
# DenIEgf+MLsRQ3kKUmsgVNnPuR69M0COfyaz0AnfX6YEIL9ukFJQPsmASfPmHof5
# tCYIFyKEpZt/givmzSI1jdpm0uX2MRwLGLYRdNhEPVjo+TfGda15x7DgpBEduqjq
# mChUS2wrmgP9TZne+kTAU28pUpU7hcfrt1RkDOO86W8oJmpBeIyGe6vikVhQppKW
# fAIKvhNfN3p5Kxq1fhE6I5YzKd2vvKtBvPpZp2uFe6LHXEcVV/FPcTx3Ph+um/o6
# ScmmxowT4Wqk4EgXh1ohephlxB89aWgwLNHLHcfte6UCU9x4eSmTC2T3pf7piBaE
# pGLpzPoYk6BAurwrMuxCxYgStl6SzQ==
# =CNSk
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 06 Mar 2023 14:34:02 GMT
# gpg:                using RSA key 522833AA75E2DCE6A24766C04DE1AF316D4F0DE9
# gpg: Good signature from "Klaus Jensen <its@irrelevant.dk>" [full]
# gpg:                 aka "Klaus Jensen <k.jensen@samsung.com>" [full]
# Primary key fingerprint: DDCA 4D9C 9EF9 31CC 3468  4272 63D5 6FC5 E55D A838
#      Subkey fingerprint: 5228 33AA 75E2 DCE6 A247  66C0 4DE1 AF31 6D4F 0DE9

* tag 'nvme-next-pull-request' of https://gitlab.com/birkelund/qemu:
  hw/nvme: flexible data placement emulation
  hw/nvme: basic directives support
  hw/nvme: add basic endurance group support
  hw/nvme: store a pointer to the NvmeSubsystem in the NvmeNamespace
  hw/nvme: move adjustment of data_units{read,written}

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

Merge tag 'pull-xen-20230306' of https://xenbits.xen.org/git-http/people/aperard/qemu-dm into staging

Xen queue:

- fix for graphic passthrough with 'xenfv' machine
- fix uninitialized variable

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEE+AwAYwjiLP2KkueYDPVXL9f7Va8FAmQF8fgACgkQDPVXL9f7
# Va/nSQf/XVfmhe2W1ailKJxuvGeMLRW/tmY/dsNAZNXXBMjRYEaF4Eps51pjYdb7
# 6UUY/atT1fm9v/AYhxc+k8weIE/mxCDbaRStQUzHlrWPof1NsmEeYZ3NVdVq5w7s
# FmDCR+yiP2tcrBPhPD0aFBB7Lsayfy0P5qLFMMeeerlkZmk1O3fB04EKtus3YD1r
# hVSH+H8i5b8vg0d/5fGGrRzKalh5E2xGGUfz4ukp3+AYWNCl2m65K0JsX42+G79b
# Cg+OpeNp9CEXZSUvkfVoRxH9OJp6GpGZIHA9U3nvH31KR4OnDeCSZuCiPvoUuvZT
# Q0fd8eA4DRTEtt9gJ+ecQEpON5dcSA==
# =kvNV
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 06 Mar 2023 14:00:24 GMT
# gpg:                using RSA key F80C006308E22CFD8A92E7980CF5572FD7FB55AF
# gpg: Good signature from "Anthony PERARD <anthony.perard@gmail.com>" [marginal]
# gpg:                 aka "Anthony PERARD <anthony.perard@citrix.com>" [marginal]
# 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: 5379 2F71 024C 600F 778A  7161 D8D5 7199 DF83 42C8
#      Subkey fingerprint: F80C 0063 08E2 2CFD 8A92  E798 0CF5 572F D7FB 55AF

* tag 'pull-xen-20230306' of https://xenbits.xen.org/git-http/people/aperard/qemu-dm:
  hw/xen/xen_pt: fix uninitialized variable
  xen/pt: reserve PCI slot 2 for Intel igd-passthru

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

MAINTAINERS: Add entry for RISC-V ACPI

RISC-V ACPI related functionality for virt machine is added in
virt-acpi-build.c. Add the maintainer entry after moving the
ARM ACPI entry under the main ACPI entry.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-9-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw/riscv/virt.c: Initialize the ACPI tables

Initialize the ACPI tables if the acpi option is not
disabled.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-8-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw/riscv/virt: virt-acpi-build.c: Add RHCT Table

RISC-V ACPI platforms need to provide RISC-V Hart Capabilities
Table (RHCT). Add this to the ACPI tables.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-7-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw/riscv/virt: virt-acpi-build.c: Add RINTC in MADT

Add Multiple APIC Description Table (MADT) with the
RINTC structure for each cpu.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-6-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw/riscv/virt: Enable basic ACPI infrastructure

Add basic ACPI infrastructure for RISC-V with below tables.
        1) DSDT with below basic objects
                - CPUs
                - fw_cfg
        2) FADT revision 6 with HW_REDUCED flag
        3) XSDT
        4) RSDP

Add this functionality in a new file virt-acpi-build.c and enable
building this infrastructure.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-5-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw/riscv/virt: Add memmap pointer to RiscVVirtState

memmap needs to be exported outside of virt.c so that
modules like acpi can use it. Hence, add a pointer field
in RiscVVirtState structure and initialize it with the
memorymap.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-4-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw/riscv/virt: Add a switch to disable ACPI

ACPI will be enabled by default. Add a switch to turn off
for testing and debug purposes.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-3-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw/riscv/virt: Add OEM_ID and OEM_TABLE_ID fields

ACPI needs OEM_ID and OEM_TABLE_ID for the machine. Add these fields
in the RISCVVirtState structure and initialize with default values.

Signed-off-by: Sunil V L <sunilvl@ventanamicro.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Message-ID: <20230302091212.999767-2-sunilvl@ventanamicro.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

riscv: Correctly set the device-tree entry 'mmu-type'

The 'mmu-type' should reflect what the hardware is capable of so use the
new satp_mode field in RISCVCPUConfig to do that.

Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Message-ID: <20230303131252.892893-6-alexghiti@rivosinc.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

riscv: Introduce satp mode hw capabilities

Currently, the max satp mode is set with the only constraint that it must be
implemented in QEMU, i.e. set in valid_vm_1_10_[32|64].

But we actually need to add another level of constraint: what the hw is
actually capable of, because currently, a linux booting on a sifive-u54
boots in sv57 mode which is incompatible with the cpu's sv39 max
capability.

So add a new bitmap to RISCVSATPMap which contains this capability and
initialize it in every XXX_cpu_init.

Finally:
- valid_vm_1_10_[32|64] constrains which satp mode the CPU can use
- the CPU hw capabilities constrains what the user may select
- the user's selection then constrains what's available to the guest
  OS.

Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20230303131252.892893-5-alexghiti@rivosinc.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

riscv: Allow user to set the satp mode

RISC-V specifies multiple sizes for addressable memory and Linux probes for
the machine's support at startup via the satp CSR register (done in
csr.c:validate_vm).

As per the specification, sv64 must support sv57, which in turn must
support sv48...etc. So we can restrict machine support by simply setting the
"highest" supported mode and the bare mode is always supported.

You can set the satp mode using the new properties "sv32", "sv39", "sv48",
"sv57" and "sv64" as follows:
-cpu rv64,sv57=on  # Linux will boot using sv57 scheme
-cpu rv64,sv39=on  # Linux will boot using sv39 scheme
-cpu rv64,sv57=off # Linux will boot using sv48 scheme
-cpu rv64          # Linux will boot using sv57 scheme by default

We take the highest level set by the user:
-cpu rv64,sv48=on,sv57=on # Linux will boot using sv57 scheme

We make sure that invalid configurations are rejected:
-cpu rv64,sv39=off,sv48=on # sv39 must be supported if higher modes are
                           # enabled

We accept "redundant" configurations:
-cpu rv64,sv48=on,sv57=off # Linux will boot using sv48 scheme

And contradictory configurations:
-cpu rv64,sv48=on,sv48=off # Linux will boot using sv39 scheme

Co-Developed-by: Ludovic Henry <ludovic@rivosinc.com>
Signed-off-by: Ludovic Henry <ludovic@rivosinc.com>
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Message-ID: <20230303131252.892893-4-alexghiti@rivosinc.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

riscv: Change type of valid_vm_1_10_[32|64] to bool

This array is actually used as a boolean so swap its current char type
to a boolean and at the same time, change the type of validate_vm to
bool since it returns valid_vm_1_10_[32|64].

Suggested-by: Andrew Jones <ajones@ventanamicro.com>
Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Message-ID: <20230303131252.892893-3-alexghiti@rivosinc.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

riscv: Pass Object to register_cpu_props instead of DeviceState

One can extract the DeviceState pointer from the Object pointer, so pass
the Object for future commits to access other fields of Object.

No functional changes intended.

Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Bin Meng <bmeng@tinylab.org>
Message-ID: <20230303131252.892893-2-alexghiti@rivosinc.com>
Signed-off-by: Palmer Dabbelt <palmer@rivosinc.com>

hw: arm: allwinner-h3: Fix and complete H3 i2c devices

Allwinner h3 has 4 twi(i2c) devices named twi0, twi1, twi2 and r_twi.
The registers are compatible with TYPE_AW_I2C_SUN6I, write 1 to clear
control register's INT_FLAG bit.

Signed-off-by: qianfan Zhao <qianfanguijin@163.com>
Reviewed-by: Strahinja Jankovic <strahinja.p.jankovic@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/nvme: flexible data placement emulation

Add emulation of TP4146 ("Flexible Data Placement").

Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Jesper Devantier <j.devantier@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>

hw/nvme: basic directives support

Add support for the Directive Send and Recv commands and the Identify
directive.

Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Gollu Appalanaidu <anaidu.gollu@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>

hw/nvme: add basic endurance group support

Add the mandatory Endurance Group identify data structures and log
pages.

For now, all namespaces in a subsystem belongs to a single Endurance
Group.

Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>

hw/nvme: store a pointer to the NvmeSubsystem in the NvmeNamespace

Each NvmeNamespace can be used by serveral controllers,
but a NvmeNamespace can at most belong to a single NvmeSubsystem.
Store a pointer to the NvmeSubsystem, if the namespace was realized
with a NvmeSubsystem.

Reviewed-by: Keith Busch <kbusch@kernel.org>
Signed-off-by: Niklas Cassel <niklas.cassel@wdc.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>

hw/nvme: move adjustment of data_units{read,written}

Move the rounding of bytes read/written into nvme_smart_log which
reports in units of 512 bytes, rounded up in thousands. This is in
preparation for adding the Endurance Group Information log page which
reports in units of billions, rounded up.

Reviewed-by: Keith Busch <kbusch@kernel.org>
Reviewed-by: Klaus Jensen <k.jensen@samsung.com>
Signed-off-by: Joel Granados <j.granados@samsung.com>
Signed-off-by: Klaus Jensen <k.jensen@samsung.com>

hw: allwinner-i2c: Fix TWI_CNTR_INT_FLAG on SUN6i SoCs

TWI_CNTR_INT_FLAG is W1C(write 1 to clear and write 0 has non-effect)
register on SUN6i based SoCs, we should lower interrupt when the guest
set this bit.

The linux kernel will hang in irq handler(mv64xxx_i2c_intr) if no
device connected on the i2c bus, next is the trace log:

allwinner_i2c_write write   CNTR(0x0c): 0xc4 A_ACK BUS_EN INT_EN
allwinner_i2c_write write   CNTR(0x0c): 0xcc A_ACK INT_FLAG BUS_EN INT_EN
allwinner_i2c_read  read    CNTR(0x0c): 0xcc A_ACK INT_FLAG BUS_EN INT_EN
allwinner_i2c_read  read    STAT(0x10): 0x20 STAT_M_ADDR_WR_NACK
allwinner_i2c_write write   CNTR(0x0c): 0x54 A_ACK M_STP BUS_EN
allwinner_i2c_write write   CNTR(0x0c): 0x4c A_ACK INT_FLAG BUS_EN
allwinner_i2c_read  read    CNTR(0x0c): 0x4c A_ACK INT_FLAG BUS_EN
allwinner_i2c_read  read    STAT(0x10): 0xf8 STAT_IDLE
allwinner_i2c_write write   CNTR(0x0c): 0x54 A_ACK M_STP BUS_EN
allwinner_i2c_write write   CNTR(0x0c): 0x4c A_ACK INT_FLAG BUS_EN
allwinner_i2c_read  read    CNTR(0x0c): 0x4c A_ACK INT_FLAG BUS_EN
allwinner_i2c_read  read    STAT(0x10): 0xf8 STAT_IDLE
...

Fix it.

Signed-off-by: qianfan Zhao <qianfanguijin@163.com>
Reviewed-by: Strahinja Jankovic <strahinja.p.jankovic@gmail.com>
Tested-by: Strahinja Jankovic <strahinja.p.jankovic@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw: arm: Support direct boot for Linux/arm64 EFI zboot images

Fedora 39 will ship its arm64 kernels in the new generic EFI zboot
format, using gzip compression for the payload.

For doing EFI boot in QEMU, this is completely transparent, as the
firmware or bootloader will take care of this. However, for direct
kernel boot without firmware, we will lose the ability to boot such
distro kernels unless we deal with the new format directly.

EFI zboot images contain metadata in the header regarding the placement
of the compressed payload inside the image, and the type of compression
used. This means we can wire up the existing gzip support without too
much hassle, by parsing the header and grabbing the payload from inside
the loaded zboot image.

Cc: Peter Maydell <peter.maydell@linaro.org>
Cc: Alex Bennée <alex.bennee@linaro.org>
Cc: Richard Henderson <richard.henderson@linaro.org>
Cc: Philippe Mathieu-Daudé <f4bug@amsat.org>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Message-id: 20230303160109.3626966-1-ardb@kernel.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: tweaked comment formatting, fixed checkpatch nits]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Rewrite check_s2_mmu_setup

Integrate neighboring code from get_phys_addr_lpae which computed
starting level, as it is easier to validate when doing both at the
same time.  Mirror the checks at the start of AArch{64,32}.S2Walk,
especially S2InvalidSL and S2InconsistentSL.

This reverts 49ba115bb74, which was incorrect -- there is nothing
in the ARM pseudocode that depends on TxSZ, i.e. outputsize; the
pseudocode is consistent in referencing PAMax.

Fixes: 49ba115bb74 ("target/arm: Pass outputsize down to check_s2_mmu_setup")
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227225832.816605-5-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Diagnose incorrect usage of arm_is_secure subroutines

In several places we use arm_is_secure_below_el3 and
arm_is_el3_or_mon separately from arm_is_secure.
These functions make no sense for m-profile, and
would indicate prior incorrect feature testing.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227225832.816605-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Stub arm_hcr_el2_eff for m-profile

M-profile doesn't have HCR_EL2.  While we could test features
before each call, zero is a generally safe return value to
disable the code in the caller.  This test is required to
avoid an assert in arm_is_secure_below_el3.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227225832.816605-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Handle m-profile in arm_is_secure

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1421
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227225832.816605-2-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement gdbstub m-profile systemreg and secext

The upstream gdb xml only implements {MSP,PSP}{,_NS,S}, but
go ahead and implement the other system registers as well.

Since there is significant overlap between the two, implement
them with common code.  The only exception is the systemreg
view of CONTROL, which merges the banked bits as per MRS.

Signed-off-by: David Reiss <dreiss@meta.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-15-richard.henderson@linaro.org
[rth: Substatial rewrite using enumerator and shared code.]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Export arm_v7m_get_sp_ptr

Allow the function to be used outside of m_helper.c.
Move to be outside of ifndef CONFIG_USER_ONLY block.
Rename from get_v7m_sp_ptr.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: David Reiss <dreiss@meta.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-14-richard.henderson@linaro.org
[rth: Split out of a larger patch]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Export arm_v7m_mrs_control

Allow the function to be used outside of m_helper.c.
Rename with an "arm_" prefix.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: David Reiss <dreiss@meta.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-13-richard.henderson@linaro.org
[rth: Split out of a larger patch]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Implement gdbstub pauth extension

The extension is primarily defined by the Linux kernel NT_ARM_PAC_MASK
ptrace register set.

The original gdb feature consists of two masks, data and code, which are
used to mask out the authentication code within a pointer.  Following
discussion with Luis Machado, add two more masks in order to support
pointers within the high half of the address space (i.e. TTBR1 vs TTBR0).

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1105
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-12-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Create pauth_ptr_mask

Keep the logic for pauth within pauth_helper.c, and expose
a helper function for use with the gdbstub pac extension.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-11-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Simplify iteration over bit widths

Order suf[] by the log8 of the width.
Use ARRAY_SIZE instead of hard-coding 128.

This changes the order of the union definitions,
but retains the order of the union-of-union members.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-10-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Add name argument to output_vector_union_type

This will make the function usable between SVE and SME.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-9-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Fix svep width in arm_gen_dynamic_svereg_xml

Define svep based on the size of the predicates,
not the primary vector registers.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-8-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Hoist pred_width in arm_gen_dynamic_svereg_xml

Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230227213329.793795-7-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>