target/arm: Implement VLDR/VSTR system register

Implement the new-in-v8.1M VLDR/VSTR variants which directly
read or write FP system registers to memory.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-10-peter.maydell@linaro.org

target/arm: Move general-use constant expanders up in translate.c

The constant-expander functions like negate, plus_2, etc, are
generally useful; move them up in translate.c so we can use them in
the VFP/Neon decoders as well as in the A32/T32/T16 decoders.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-9-peter.maydell@linaro.org

target/arm: Refactor M-profile VMSR/VMRS handling

Currently M-profile borrows the A-profile code for VMSR and VMRS
(access to the FP system registers), because all it needs to support
is the FPSCR.  In v8.1M things become significantly more complicated
in two ways:

 * there are several new FP system registers; some have side effects
   on read, and one (FPCXT_NS) needs to avoid the usual
   vfp_access_check() and the "only if FPU implemented" check

 * all sysregs are now accessible both by VMRS/VMSR (which
   reads/writes a general purpose register) and also by VLDR/VSTR
   (which reads/writes them directly to memory)

Refactor the structure of how we handle VMSR/VMRS to cope with this:

 * keep the M-profile code entirely separate from the A-profile code

 * abstract out the "read or write the general purpose register" part
   of the code into a loadfn or storefn function pointer, so we can
   reuse it for VLDR/VSTR.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-8-peter.maydell@linaro.org

target/arm: Enforce M-profile VMRS/VMSR register restrictions

For M-profile before v8.1M, the only valid register for VMSR/VMRS is
the FPSCR.  We have a comment that states this, but the actual logic
to forbid accesses for any other register value is missing, so we
would end up with A-profile style behaviour.  Add the missing check.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-7-peter.maydell@linaro.org

target/arm: Implement CLRM instruction

In v8.1M the new CLRM instruction allows zeroing an arbitrary set of
the general-purpose registers and APSR.  Implement this.

The encoding is a subset of the LDMIA T2 encoding, using what would
be Rn=0b1111 (which UNDEFs for LDMIA).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-6-peter.maydell@linaro.org

target/arm: Implement VSCCLRM insn

Implement the v8.1M VSCCLRM insn, which zeros floating point
registers if there is an active floating point context.
This requires support in write_neon_element32() for the MO_32
element size, so add it.

Because we want to use arm_gen_condlabel(), we need to move
the definition of that function up in translate.c so it is
before the #include of translate-vfp.c.inc.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-5-peter.maydell@linaro.org

target/arm: Don't clobber ID_PFR1.Security on M-profile cores

In arm_cpu_realizefn() we check whether the board code disabled EL3
via the has_el3 CPU object property, which we create if the CPU
starts with the ARM_FEATURE_EL3 feature bit.  If it is disabled, then
we turn off ARM_FEATURE_EL3 and also zero out the relevant fields in
the ID_PFR1 and ID_AA64PFR0 registers.

This codepath was incorrectly being taken for M-profile CPUs, which
do not have an EL3 and don't set ARM_FEATURE_EL3, but which may have
the M-profile Security extension and so should have non-zero values
in the ID_PFR1.Security field.

Restrict the handling of the feature flag to A/R-profile cores.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-4-peter.maydell@linaro.org

target/arm: Implement v8.1M PXN extension

In v8.1M the PXN architecture extension adds a new PXN bit to the
MPU_RLAR registers, which forbids execution of code in the region
from a privileged mode.

This is another feature which is just in the generic "in v8.1M" set
and has no ID register field indicating its presence.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-3-peter.maydell@linaro.org

hw/intc/armv7m_nvic: Make all of system PPB range be RAZWI/BusFault

For M-profile CPUs, the range from 0xe0000000 to 0xe00fffff is the
Private Peripheral Bus range, which includes all of the memory mapped
devices and registers that are part of the CPU itself, including the
NVIC, systick timer, and debug and trace components like the Data
Watchpoint and Trace unit (DWT).  Within this large region, the range
0xe000e000 to 0xe000efff is the System Control Space (NVIC, system
registers, systick) and 0xe002e000 to 0exe002efff is its Non-secure
alias.

The architecture is clear that within the SCS unimplemented registers
should be RES0 for privileged accesses and generate BusFault for
unprivileged accesses, and we currently implement this.

It is less clear about how to handle accesses to unimplemented
regions of the wider PPB.  Unprivileged accesses should definitely
cause BusFaults (R_DQQS), but the behaviour of privileged accesses is
not given as a general rule.  However, the register definitions of
individual registers for components like the DWT all state that they
are RES0 if the relevant component is not implemented, so the
simplest way to provide that is to provide RAZ/WI for the whole range
for privileged accesses.  (The v7M Arm ARM does say that reserved
registers should be UNK/SBZP.)

Expand the container MemoryRegion that the NVIC exposes so that
it covers the whole PPB space. This means:
 * moving the address that the ARMV7M device maps it to down by
   0xe000 bytes
 * moving the off and the offsets within the container of all the
   subregions forward by 0xe000 bytes
 * adding a new default MemoryRegion that covers the whole container
   at a lower priority than anything else and which provides the
   RAZWI/BusFault behaviour

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201119215617.29887-2-peter.maydell@linaro.org

i.MX6ul: Fix bad printf format specifiers

We should use printf format specifier "%u" instead of "%d" for
argument of type "unsigned int".

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Alex Chen <alex.chen@huawei.com>
Message-id: 20201126111109.112238-5-alex.chen@huawei.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

i.MX6: Fix bad printf format specifiers

We should use printf format specifier "%u" instead of "%d" for
argument of type "unsigned int".

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Alex Chen <alex.chen@huawei.com>
Message-id: 20201126111109.112238-4-alex.chen@huawei.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

i.MX31: Fix bad printf format specifiers

We should use printf format specifier "%u" instead of "%d" for
argument of type "unsigned int".

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Alex Chen <alex.chen@huawei.com>
Message-id: 20201126111109.112238-3-alex.chen@huawei.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

i.MX25: Fix bad printf format specifiers

We should use printf format specifier "%u" instead of "%d" for
argument of type "unsigned int".

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Alex Chen <alex.chen@huawei.com>
Message-id: 20201126111109.112238-2-alex.chen@huawei.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/qtest/npcm7xx_rng-test: dump random data on failure

Dump the collected random data after a randomness test failure.

Note that this relies on the test having called
g_test_set_nonfatal_assertions() so we don't abort immediately on the
assertion failure.

Signed-off-by: Havard Skinnemoen <hskinnemoen@google.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: minor commit message tweak]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sbsa-ref: allow to use Cortex-A53/57/72 cpus

Trusted Firmware now supports A72 on sbsa-ref by default [1] so enable
it for QEMU as well. A53 was already enabled there.

1. https://review.trustedfirmware.org/c/TF-A/trusted-firmware-a/+/7117

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20201120141705.246690-1-marcin.juszkiewicz@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

MAINTAINERS: Add maintainer entry for Xilinx ZynqMP CAN controller

Reviewed-by: Francisco Iglesias <francisco.iglesias@xilinx.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Signed-off-by: Vikram Garhwal <fnu.vikram@xilinx.com>
Message-id: 1605728926-352690-5-git-send-email-fnu.vikram@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/qtest: Introduce tests for Xilinx ZynqMP CAN controller

The QTests perform five tests on the Xilinx ZynqMP CAN controller:
    Tests the CAN controller in loopback, sleep and snoop mode.
    Tests filtering of incoming CAN messages.

Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Reviewed-by: Francisco Iglesias <francisco.iglesias@xilinx.com>
Signed-off-by: Vikram Garhwal <fnu.vikram@xilinx.com>
Message-id: 1605728926-352690-4-git-send-email-fnu.vikram@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

xlnx-zynqmp: Connect Xilinx ZynqMP CAN controllers

Connect CAN0 and CAN1 on the ZynqMP.

Reviewed-by: Francisco Iglesias <francisco.iglesias@xilinx.com>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@xilinx.com>
Signed-off-by: Vikram Garhwal <fnu.vikram@xilinx.com>
Message-id: 1605728926-352690-3-git-send-email-fnu.vikram@xilinx.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/net/can: Introduce Xilinx ZynqMP CAN controller

The Xilinx ZynqMP CAN controller is developed based on SocketCAN, QEMU CAN bus
implementation. Bus connection and socketCAN connection for each CAN module
can be set through command lines.

Example for using single CAN:
    -object can-bus,id=canbus0 \
    -machine xlnx-zcu102.canbus0=canbus0 \
    -object can-host-socketcan,id=socketcan0,if=vcan0,canbus=canbus0

Example for connecting both CAN to same virtual CAN on host machine:
    -object can-bus,id=canbus0 -object can-bus,id=canbus1 \
    -machine xlnx-zcu102.canbus0=canbus0 \
    -machine xlnx-zcu102.canbus1=canbus1 \
    -object can-host-socketcan,id=socketcan0,if=vcan0,canbus=canbus0 \
    -object can-host-socketcan,id=socketcan1,if=vcan0,canbus=canbus1

To create virtual CAN on the host machine, please check the QEMU CAN docs:
https://github.com/qemu/qemu/blob/master/docs/can.txt

Signed-off-by: Vikram Garhwal <fnu.vikram@xilinx.com>
Message-id: 1605728926-352690-2-git-send-email-fnu.vikram@xilinx.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/smmuv3: Fix up L1STD_SPAN decoding

Accroding to the SMMUv3 spec, the SPAN field of Level1 Stream Table
Descriptor is 5 bits([4:0]).

Fixes: 9bde7f0674f(hw/arm/smmuv3: Implement translate callback)
Signed-off-by: Kunkun Jiang <jiangkunkun@huawei.com>
Message-id: 20201124023711.1184-1-jiangkunkun@huawei.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Acked-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge remote-tracking branch 'remotes/mst/tags/for_upstream' into staging

pc,pci,virtio: fixes, cleanups

Lots of fixes, cleanups.
CPU hot-unplug improvements.
A new AER property for virtio devices, adding a dummy AER capability.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# gpg: Signature made Wed 09 Dec 2020 18:04:28 GMT
# gpg:                using RSA key 5D09FD0871C8F85B94CA8A0D281F0DB8D28D5469
# gpg:                issuer "mst@redhat.com"
# gpg: Good signature from "Michael S. Tsirkin <mst@kernel.org>" [full]
# gpg:                 aka "Michael S. Tsirkin <mst@redhat.com>" [full]
# Primary key fingerprint: 0270 606B 6F3C DF3D 0B17  0970 C350 3912 AFBE 8E67
#      Subkey fingerprint: 5D09 FD08 71C8 F85B 94CA  8A0D 281F 0DB8 D28D 5469

* remotes/mst/tags/for_upstream: (65 commits)
  hw/virtio-pci Added AER capability.
  hw/virtio-pci Added counter for pcie capabilities offsets.
  pcie_aer: Fix help message of pcie_aer_inject_error command
  x86: ich9: let firmware negotiate 'CPU hot-unplug with SMI' feature
  x86: ich9: factor out "guest_cpu_hotplug_features"
  tests/acpi: update expected files
  x86: acpi: let the firmware handle pending "CPU remove" events in SMM
  tests/acpi: allow expected files change
  x86: acpi: introduce AcpiPmInfo::smi_on_cpu_unplug
  acpi: cpuhp: introduce 'firmware performs eject' status/control bits
  hw/i386/pc: add max combined fw size as machine configuration option
  block/export: avoid g_return_val_if() input validation
  contrib/vhost-user-input: avoid g_return_val_if() input validation
  contrib/vhost-user-gpu: avoid g_return_val_if() input validation
  contrib/vhost-user-blk: avoid g_return_val_if() input validation
  .gitlab-ci: add build-libvhost-user
  libvhost-user: add a simple link test without glib
  libvhost-user: make it a meson subproject
  libvhost-user: drop qemu/osdep.h dependency
  libvhost-user: remove qemu/compiler.h usage
  ...

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

hw/virtio-pci Added AER capability.

Added AER capability for virtio-pci devices.
Also added property for devices, by default AER is disabled.

Signed-off-by: Andrew Melnychenko <andrew@daynix.com>
Message-Id: <20201203110713.204938-3-andrew@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/virtio-pci Added counter for pcie capabilities offsets.

Removed hardcoded offset for ats. Added cap offset counter
for future capabilities like AER.

Signed-off-by: Andrew Melnychenko <andrew@daynix.com>
Message-Id: <20201203110713.204938-2-andrew@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

pcie_aer: Fix help message of pcie_aer_inject_error command

There is an interesting typo in the help message of pcie_aer_inject_error
command. Use 'tlp' instead of 'tlb' to match the PCIe AER term.

Signed-off-by: Zenghui Yu <yuzenghui@huawei.com>
Message-Id: <20201204030953.837-1-yuzenghui@huawei.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

x86: ich9: let firmware negotiate 'CPU hot-unplug with SMI' feature

Keep CPU hotunplug with SMI disabled on 5.2 and older and enable
it by default on newer machine types.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20201207140739.3829993-9-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

x86: ich9: factor out "guest_cpu_hotplug_features"

it will be reused by next patch to check validity of unplug
feature.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20201207140739.3829993-8-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests/acpi: update expected files

update expected files with following change:

@@ -557,6 +557,7 @@ DefinitionBlock ("", "DSDT", 1, "BOCHS ", "BXPCDSDT", 0x00000001)
                 CINS,   1,
                 CRMV,   1,
                 CEJ0,   1,
+                CEJF,   1,
                 Offset (0x05),
                 CCMD,   8
             }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20201207140739.3829993-7-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

x86: acpi: let the firmware handle pending "CPU remove" events in SMM

if firmware and QEMU negotiated CPU hotunplug support, generate
_EJ0 method so that it will mark CPU for removal by firmware and
pass control to it by triggering SMI.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20201207140739.3829993-6-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests/acpi: allow expected files change

Change that will be introduced by following patch:

@@ -557,6 +557,7 @@ DefinitionBlock ("", "DSDT", 1, "BOCHS ", "BXPCDSDT", 0x00000001)
                 CINS,   1,
                 CRMV,   1,
                 CEJ0,   1,
+                CEJF,   1,
                 Offset (0x05),
                 CCMD,   8
             }

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20201207140739.3829993-5-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

x86: acpi: introduce AcpiPmInfo::smi_on_cpu_unplug

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20201207140739.3829993-4-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi: cpuhp: introduce 'firmware performs eject' status/control bits

Adds bit #4 to status/control field of CPU hotplug MMIO interface.
New bit will be used OSPM to mark CPUs as pending for removal by firmware,
when it calls _EJ0 method on CPU device node. Later on, when firmware
sees this bit set, it will perform CPU eject which will clear bit #4
as well.

Signed-off-by: Igor Mammedov <imammedo@redhat.com>
Message-Id: <20201207140739.3829993-3-imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/i386/pc: add max combined fw size as machine configuration option

At Hewlett Packard Inc. we have a need for increased fw size to enable testing of our custom fw.

Rebase v6 patch to d73c46e4

Signed-off-by: Erich McMillan <erich.mcmillan@hp.com>
Message-Id: <20201208155338.14-1-erich.mcmillan@hp.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

block/export: avoid g_return_val_if() input validation

Do not validate input with g_return_val_if(). This API is intended for
checking programming errors and is compiled out with -DG_DISABLE_CHECKS.

Use an explicit if statement for input validation so it cannot
accidentally be compiled out.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20201118091644.199527-5-stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

contrib/vhost-user-input: avoid g_return_val_if() input validation

Do not validate input with g_return_val_if(). This API is intended for
checking programming errors and is compiled out with -DG_DISABLE_CHECKS.

Use an explicit if statement for input validation so it cannot
accidentally be compiled out.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20201118091644.199527-4-stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

contrib/vhost-user-gpu: avoid g_return_val_if() input validation

Do not validate input with g_return_val_if(). This API is intended for
checking programming errors and is compiled out with -DG_DISABLE_CHECKS.

Use an explicit if statement for input validation so it cannot
accidentally be compiled out.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20201118091644.199527-3-stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

contrib/vhost-user-blk: avoid g_return_val_if() input validation

Do not validate input with g_return_val_if(). This API is intended for
checking programming errors and is compiled out with -DG_DISABLE_CHECKS.

Use an explicit if statement for input validation so it cannot
accidentally be compiled out.

Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20201118091644.199527-2-stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

.gitlab-ci: add build-libvhost-user

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20201125100640.366523-9-marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: add a simple link test without glib

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20201125100640.366523-8-marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Merge remote-tracking branch 'remotes/huth-gitlab/tags/pull-request-2020-12-09' into staging

* Gitlab-CI improvement patches by Philippe
* Some small fixes for tests
* Fix coverity warning in the mcf5206 code

# gpg: Signature made Wed 09 Dec 2020 07:13:12 GMT
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "thuth@redhat.com"
# gpg: Good signature from "Thomas Huth <th.huth@gmx.de>" [full]
# gpg:                 aka "Thomas Huth <thuth@redhat.com>" [full]
# gpg:                 aka "Thomas Huth <huth@tuxfamily.org>" [full]
# gpg:                 aka "Thomas Huth <th.huth@posteo.de>" [unknown]
# Primary key fingerprint: 27B8 8847 EEE0 2501 18F3  EAB9 2ED9 D774 FE70 2DB5

* remotes/huth-gitlab/tags/pull-request-2020-12-09:
  hw/m68k/mcf5206: Don't leak IRQs in mcf5206_mbar_realize()
  gitlab-ci: Move coroutine tests across to gitlab
  gitlab-ci: Move user-static test across to gitlab
  gitlab-ci: Update 'build-disabled' to cover all configurable options
  gitlab-ci: Split CONFIGURE_ARGS one argument per line for build-disabled
  fuzz: avoid double-fetches by default
  tests/qtest/fuzz-test: Quit test_lp1878642 once done
  test-qga: fix a resource leak in test_qga_guest_get_osinfo()
  gitlab-ci: Add Xen cross-build jobs
  gitlab-ci: Add KVM s390x cross-build jobs
  gitlab-ci: Introduce 'cross_accel_build_job' template
  gitlab-ci: Replace YAML anchors by extends (cross_system_build_job)
  gitlab-ci: Document 'build-tcg-disabled' is a KVM X86 job

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

hw/m68k/mcf5206: Don't leak IRQs in mcf5206_mbar_realize()

Coverity points out that the realize function for the TYPE_MCF5206_MBAR
device leaks the IRQ array it allocates with qemu_allocate_irqs().
Keep a pointer to it in the device state struct to avoid the leak.
(Since it needs to stay around for the life of the simulation there
is no need to actually free it, and the leak was harmless.)

Fixes: Coverity CID 1432412
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <huth@tuxfamily.org>
Reviewed-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
Message-Id: <20201120172314.14725-1-peter.maydell@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Move coroutine tests across to gitlab

Similarly to commit 8cdb2cef3f1, move the coroutine tests to GitLab.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201108204535.2319870-13-philmd@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
[thuth: Replaced Travis by Gitlab-CI in comment]
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Move user-static test across to gitlab

Similarly to commit 8cdb2cef3f1, move the user-static test to GitLab.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201108204535.2319870-11-philmd@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Update 'build-disabled' to cover all configurable options

Add these missing options to the 'build-disabled' job:
  --disable-auth-pam             (commit 8953caf3cd38)
  --disable-gcrypt               (commit 91bfcdb01d48)
  --disable-keyring              (commit 54e7aac05624)
  --disable-libudev              (commit 5c53015a480b)
  --disable-opengl               (commit da076ffed6b9)
  --disable-sparse               (commit 03b4fe7de226)

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
[thuth: Dropped the unnecessary hunk with --disable-vhost-user-blk-server]
Message-Id: <20201108204535.2319870-7-philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Split CONFIGURE_ARGS one argument per line for build-disabled

We will keep adding/removing options to our 'configure' script,
so for easier maintainability it makes sense to have CONFIGURE_ARGS
declared as one option per line. This way we can review diff easily
(or rebase/cherry-pick).

No logical change.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201108204535.2319870-6-philmd@redhat.com>
Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

fuzz: avoid double-fetches by default

The generic fuzzer can find double-fetch bugs. However:
* We currently have no good way of producing qemu-system reproducers for
  double-fetch bugs. Even if we can get developers to run the binary-blob
  reproducers with the qemu-fuzz builds, we currently don't have a minimizer for
  these reproducers, so they are usually not easy to follow.
* Often times the fuzzer will provide a reproducer containing a
  double-fetch for a bug that can be reproduced without double-fetching.

Until we find a way to build nice double-fetch reproducers that
developers are willing to look at, lets tell OSS-Fuzz to avoid
double-fetches.

Signed-off-by: Alexander Bulekov <alxndr@bu.edu>
Reviewed-by: Darren Kenny <darren.kenny@oracle.com>
Message-Id: <20201202164214.93867-1-alxndr@bu.edu>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest/fuzz-test: Quit test_lp1878642 once done

Missed in fd250172842 ("qtest: add a reproducer for LP#1878642").

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201201191026.4149955-2-philmd@redhat.com>
Reviewed-by: Li Qiang <liq3ea@gmail.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

test-qga: fix a resource leak in test_qga_guest_get_osinfo()

The fixture->fd is created in fixture_setup() and, likewise, needs to be closed
in fixture_tear_down().

Reported-by: Euler Robot <euler.robot@huawei.com>
Signed-off-by: Alex Chen <alex.chen@huawei.com>
Reviewed-by: Li Qiang <liq3ea@gmail.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20201125102403.57709-1-alex.chen@huawei.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Add Xen cross-build jobs

Cross-build ARM and X86 targets with only Xen accelerator enabled.

Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201207131503.3858889-6-philmd@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Add KVM s390x cross-build jobs

Cross-build s390x target with only KVM accelerator enabled.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201207131503.3858889-5-philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Introduce 'cross_accel_build_job' template

Introduce a job template to cross-build accelerator specific
jobs (enable a specific accelerator, disabling the others).

The specific accelerator is selected by the $ACCEL environment
variable (default to KVM).

Extra options such disabling other accelerators are passed
via the $ACCEL_CONFIGURE_OPTS environment variable.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201207131503.3858889-4-philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Replace YAML anchors by extends (cross_system_build_job)

'extends' is an alternative to using YAML anchors
and is a little more flexible and readable. See:
https://docs.gitlab.com/ee/ci/yaml/#extends

More importantly it allows exploding YAML jobs.

Reviewed-by: Wainer dos Santos Moschetta <wainersm@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201207131503.3858889-3-philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Document 'build-tcg-disabled' is a KVM X86 job

Document what this job cover (build X86 targets with
KVM being the single accelerator available).

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@redhat.com>
Message-Id: <20201207131503.3858889-2-philmd@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

Open 6.0 development tree

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

libvhost-user: make it a meson subproject

By making libvhost-user a subproject, check it builds
standalone (without the global QEMU cflags etc).

Note that the library still relies on QEMU include/qemu/atomic.h and
linux_headers/.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20201125100640.366523-6-marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: drop qemu/osdep.h dependency

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20201125100640.366523-5-marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: remove qemu/compiler.h usage

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20201125100640.366523-4-marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: replace qemu/memfd.h usage

Undo the damage from commit 5f9ff1eff3 ("libvhost-user: Support tracking
inflight I/O in shared memory") which introduced glib dependency through
osdep.h inclusion.

libvhost-user.c tries to stay free from glib usage.

Use glibc memfd_create directly when available (assumed so when
MFD_ALLOW_SEALING is defined). A following commit will make the project
standalone and check for memfd API at configure time, instead of a
panic at runtime.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20201125100640.366523-3-marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

libvhost-user: replace qemu/bswap.h with glibc endian.h

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Dr. David Alan Gilbert <dgilbert@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20201125100640.366523-2-marcandre.lureau@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw: add compat machines for 6.0

Add 6.0 machine types for arm/i440fx/q35/s390x/spapr.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Message-Id: <20201109173928.1001764-1-cohuck@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>

failover: Remove primary_dev member

Only three uses remained, and we can remove them on that case.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-28-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: simplify failover_unplug_primary

We can calculate device just once.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-27-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Caller of this two functions already have primary_dev

Pass it as an argument.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-26-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: We don't need to cache primary_device_id anymore

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-25-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: split failover_find_primary_device_id()

So we can calculate the device id when we need it.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-24-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: remove failover_find_primary_device() error parameter

It can never give one error.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-23-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: make sure that id always exist

We check that it exist at device creation time, so we don't have to
check anywhere else.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-22-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: simplify qdev_device_add()

We don't need to walk the opts by hand.  qmp_opt_get() already does
that.  And then we can remove the functions that did that walk.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-21-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: simplify qdev_device_add() failover case

Just put allthe logic inside the same if.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-20-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Rename to failover_find_primary_device()

This commit:
* Rename them to failover_find_primary_devices() so
  - it starts with failover_
  - it don't connect anything, just find the primary device
* Create documentation for the function

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-19-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: virtio_net_connect_failover_devices() does nothing

It just calls virtio_net_find_primary(), so just update the callers.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-18-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Rename function to hide_device()

You should not use pasive.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-17-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: should_be_hidden() should take a bool

We didn't use at all the -1 value, and we don't really care.  It was
only used for the cases when this is not the device that we are
searching for.  And in that case we should not hide the device.

Once there, simplify virtio-Snet_primary_should_be_hidden.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-16-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: simplify virtio_net_find_primary()

a - is_my_primary() never sets one error
b - If we return 1, primary_device_id is always set

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-15-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Remove memory leak

Two things, at this point:

* n->primary_device_id has to be set, otherwise
  virtio_net_find_primary don't work.  So we have a leak here.

* it has to be exactly the same that prim_dev->id because what
  qdev_find_recursive() does is just compare this two values.

So remove the unneeded assignment and leaky bits.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-14-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Remove primary_device_dict

It was only used once.  And we have there opts->id, so no need for it.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-13-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: remove standby_id variable

We can calculate it, and we only use it once anyways.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-12-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Remove primary_device_opts

It was really only used once, in failover_add_primary().  Just search
for it on global opts when it is needed.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-11-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: g_strcmp0() knows how to handle NULL

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-10-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Rename bool to failover_primary_hidden

You should not use passive naming variables.
And once there, be able to search for them.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-9-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: qdev_device_add() returns err or dev set

Never both.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-8-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Remove external partially_hotplugged property

It was only set "once", and with the wrong value. As far as I can see,
libvirt still don't use it.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-7-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Remove unused parameter

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-6-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: primary bus is only used once, and where it is set

Just remove the struct member.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-5-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: Use always atomics for primary_should_be_hidden

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-4-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

failover: fix indentantion

Once there, remove not needed cast.

Signed-off-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20201118083748.1328-3-quintela@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

unit-test: Add the binary file and clear diff.h

Add the binary file DSDT.pxb and clear bios-tables-test-allowed-diff.h

Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Message-Id: <20201119014841.7298-10-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

unit-test: Add testcase for pxb

Add testcase for pxb to make sure the ACPI table is correct for guest.

Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Message-Id: <20201119014841.7298-9-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

unit-test: The files changed.

The unit-test is seperated into three patches:
1. The files changed and list in bios-tables-test-allowed-diff.h
2. The unit-test
3. The binary file and clear bios-tables-test-allowed-diff.h

The ASL diff would also be listed.
Sice there are 1000+lines diff, some changes would be omitted.

  * Original Table Header:
  *     Signature        "DSDT"
- *     Length           0x000014BB (5307)
+ *     Length           0x00001E7A (7802)
  *     Revision         0x02
- *     Checksum         0xD1
+ *     Checksum         0x57
  *     OEM ID           "BOCHS "
  *     OEM Table ID     "BXPCDSDT"
  *     OEM Revision     0x00000001 (1)

+        Device (PC80)
+        {
+            Name (_HID, "PNP0A08" /* PCI Express Bus */)  // _HID: Hardware ID
+            Name (_CID, "PNP0A03" /* PCI Bus */)  // _CID: Compatible ID
+            Name (_ADR, Zero)  // _ADR: Address
+            Name (_CCA, One)  // _CCA: Cache Coherency Attribute
+            Name (_SEG, Zero)  // _SEG: PCI Segment
+            Name (_BBN, 0x80)  // _BBN: BIOS Bus Number
+            Name (_UID, 0x80)  // _UID: Unique ID
+            Name (_STR, Unicode ("pxb Device"))  // _STR: Description String
+            Name (_PRT, Package (0x80)  // _PRT: PCI Routing Table
+            {
+                Package (0x04)
+                {
+                    0xFFFF,
+                    Zero,
+                    GSI0,
+                    Zero
+                },
+

Packages are omitted.

+                Package (0x04)
+                {
+                    0x001FFFFF,
+                    0x03,
+                    GSI2,
+                    Zero
+                }
+            })
+            Device (GSI0)
+            {
+                Name (_HID, "PNP0C0F" /* PCI Interrupt Link Device */)  // _HID: Hardware ID
+                Name (_UID, Zero)  // _UID: Unique ID
+                Name (_PRS, ResourceTemplate ()  // _PRS: Possible Resource Settings
+                {
+                    Interrupt (ResourceConsumer, Level, ActiveHigh, Exclusive, ,, )
+                    {
+                        0x00000023,
+                    }
+                })
+                Name (_CRS, ResourceTemplate ()  // _CRS: Current Resource Settings
+                {
+                    Interrupt (ResourceConsumer, Level, ActiveHigh, Exclusive, ,, )
+                    {
+                        0x00000023,
+                    }
+                })
+                Method (_SRS, 1, NotSerialized)  // _SRS: Set Resource Settings
+                {
+                }
+            }

GSI1,2,3 are omitted.

+            Name (_CRS, ResourceTemplate ()  // _CRS: Current Resource Settings
+            {
+                WordBusNumber (ResourceProducer, MinFixed, MaxFixed, PosDecode,
+                    0x0000,             // Granularity
+                    0x0080,             // Range Minimum
+                    0x0080,             // Range Maximum
+                    0x0000,             // Translation Offset
+                    0x0001,             // Length
+                    ,, )
+            })
+            Name (SUPP, Zero)
+            Name (CTRL, Zero)
+            Method (_OSC, 4, NotSerialized)  // _OSC: Operating System Capabilities
+            {
+                CreateDWordField (Arg3, Zero, CDW1)
+                If ((Arg0 == ToUUID ("33db4d5b-1ff7-401c-9657-7441c03dd766") /* PCI Host Bridge Device */))
+                {
+                    CreateDWordField (Arg3, 0x04, CDW2)
+                    CreateDWordField (Arg3, 0x08, CDW3)
+                    SUPP = CDW2 /* \_SB_.PC80._OSC.CDW2 */
+                    CTRL = CDW3 /* \_SB_.PC80._OSC.CDW3 */
+                    CTRL &= 0x1F
+                    If ((Arg1 != One))
+                    {
+                        CDW1 |= 0x08
+                    }
+
+                    If ((CDW3 != CTRL))
+                    {
+                        CDW1 |= 0x10
+                    }
+
+                    CDW3 = CTRL /* \_SB_.PC80.CTRL */
+                    Return (Arg3)
+                }
+                Else
+                {
+                    CDW1 |= 0x04
+                    Return (Arg3)
+                }
+            }

DSM is are omitted

         Device (PCI0)
         {
             Name (_HID, "PNP0A08" /* PCI Express Bus */)  // _HID: Hardware ID
                     WordBusNumber (ResourceProducer, MinFixed, MaxFixed, PosDecode,
                         0x0000,             // Granularity
                         0x0000,             // Range Minimum
-                        0x00FF,             // Range Maximum
+                        0x007F,             // Range Maximum
                         0x0000,             // Translation Offset
-                        0x0100,             // Length
+                        0x0080,             // Length

Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Message-Id: <20201119014841.7298-8-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi: Align the size to 128k

If table size is changed between virt_acpi_build and
virt_acpi_build_update, the table size would not be updated to
UEFI, therefore, just align the size to 128kb, which is enough
and same with x86. It would warn if 64k is not enough and the
align size should be updated.

Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Message-Id: <20201119014841.7298-7-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi/gpex: Build tables for pxb

The resources of pxbs are obtained by crs_build and the resources
used by pxbs would be moved from the resources defined for host-bridge.

The resources for pxb are composed of following two parts:
1. The bar space of the pci-bridge/pcie-root-port behined it
2. The config space of devices behind it.

Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Message-Id: <20201119014841.7298-6-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi: Extract crs build form acpi_build.c

Extract crs build form acpi_build.c, the function could also be used
to build the crs for pxbs for arm. The resources are composed by two parts:
1. The bar space of pci-bridge/pcie-root-ports
2. The resources needed by devices behind PXBs.
The base and limit of memory/io are obtained from the config via two APIs:
pci_bridge_get_base and pci_bridge_get_limit

Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Message-Id: <20201119014841.7298-5-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/arm/virt: Write extra pci roots into fw_cfg

Add bus property to virt machine for primary PCI root bus and use it to add
extra pci roots behind it.

Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Message-Id: <20201119014841.7298-4-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

fw_cfg: Refactor extra pci roots addition

Extract extra pci roots addition from pc machine, which could be used by
other machines.

In order to make uefi get the extra roots, it is necessary to write extra
roots into fw_cfg. And only if the uefi knows there are extra roots,
the config spaces of devices behind the root could be obtained.

Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Message-Id: <20201119014841.7298-3-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

acpi/gpex: Extract two APIs from acpi_dsdt_add_pci

Extract two APIs acpi_dsdt_add_pci_route_table and
acpi_dsdt_add_pci_osc from acpi_dsdt_add_pci. The first
API is used to specify the pci route table and the second
API is used to declare the operation system capabilities.
These two APIs would be used to specify the pxb-pcie in DSDT.

Signed-off-by: Yubo Miao <miaoyubo@huawei.com>
Signed-off-by: Jiahui Cen <cenjiahui@huawei.com>
Message-Id: <20201119014841.7298-2-cenjiahui@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

virtio: reset device on bad guest index in virtio_load()

If we find a queue with an inconsistent guest index value, explicitly mark the
device as needing a reset - and broken - via virtio_error().

There's at least one driver implementation - the virtio-win NetKVM driver - that
is able to handle a VIRTIO_CONFIG_S_NEEDS_RESET notification and successfully
restore the device to a working state. Other implementations do not correctly
handle this, but as the VQ is not in a functional state anyway, this is still
worth doing.

Signed-off-by: John Levon <john.levon@nutanix.com>
Message-Id: <20201120185103.GA442386@sent>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

memory: Skip bad range assertion if notifier is DEVIOTLB_UNMAP type

Device IOTLB invalidations can unmap arbitrary ranges, eiter outside of
the memory region or even [0, ~0ULL] for all the space. The assertion
could be hit by a guest, and rhel7 guest effectively hit it.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20201116165506.31315-6-eperezma@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

intel_iommu: Skip page walking on device iotlb invalidations

Although they didn't reach the notifier because of the filtering in
memory_region_notify_iommu_one, the vt-d was still splitting huge
memory invalidations in chunks. Skipping it.

This improves performance in case of netperf with vhost-net:
* TCP_STREAM: From 1923.6Mbit/s to 2175.13Mbit/s (13%)
* TCP_RR: From 8464.73 trans/s to 8932.703333 trans/s (5.5%)
* UDP_RR: From 8562.08 trans/s to 9005.62/s (5.1%)
* UDP_STREAM: No change observed (insignificant 0.1% improvement)

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20201116165506.31315-5-eperezma@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

memory: Add IOMMU_NOTIFIER_DEVIOTLB_UNMAP IOMMUTLBNotificationType

This allows us to differentiate between regular IOMMU map/unmap events
and DEVIOTLB unmap. Doing so, notifiers that only need device IOTLB
invalidations will not receive regular IOMMU unmappings.

Adapt intel and vhost to use it.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20201116165506.31315-4-eperezma@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

memory: Add IOMMUTLBEvent

This way we can tell between regular IOMMUTLBEntry (entry of IOMMU
hardware) and notifications.

In the notifications, we set explicitly if it is a MAPs or an UNMAP,
instead of trusting in entry permissions to differentiate them.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20201116165506.31315-3-eperezma@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Acked-by: David Gibson <david@gibson.dropbear.id.au>

memory: Rename memory_region_notify_one to memory_region_notify_iommu_one

Previous name didn't reflect the iommu operation.

Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: David Gibson <david@gibson.dropbear.id.au>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Acked-by: Jason Wang <jasowang@redhat.com>
Message-Id: <20201116165506.31315-2-eperezma@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>