more memslots support in libvhost-user
support PCIe Gen5/Gen6 link speeds in pcie
more traces in vdpa
network simulation devices support in vdpa
SMBIOS type 9 descriptor implementation
Bump max_cpus to 4096 vcpus in q35
aw-bits and granule options in VIRTIO-IOMMU
Support report NUMA nodes for device memory using GI in acpi
Beginning of shutdown event support in pvpanic
fixes, cleanups all over the place.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
-----BEGIN PGP SIGNATURE-----
iQFDBAABCAAtFiEEXQn9CHHI+FuUyooNKB8NuNKNVGkFAmXw0TMPHG1zdEByZWRo
YXQuY29tAAoJECgfDbjSjVRp8x4H+gLMoGwaGAX7gDGPgn2Ix4j/3kO77ZJ9X9k/
1KqZu/9eMS1j2Ei+vZqf05w7qRjxxhwDq3ilEXF/+UFqgAehLqpRRB8j5inqvzYt
+jv0DbL11PBp/oFjWcytm5CbiVsvq8KlqCF29VNzc162XdtcduUOWagL96y8lJfZ
uPrOoyeR7SMH9lp3LLLHWgu+9W4nOS03RroZ6Umj40y5B7yR0Rrppz8lMw5AoQtr
0gMRnFhYXeiW6CXdz+Tzcr7XfvkkYDi/j7ibiNSURLBfOpZa6Y8+kJGKxz5H1K1G
6ZY4PBcOpQzl+NMrktPHogczgJgOK10t+1i/R3bGZYw2Qn/93Eg=
=C0UU
-----END PGP SIGNATURE-----
Merge tag 'for_upstream' of https://git.kernel.org/pub/scm/virt/kvm/mst/qemu into staging
virtio,pc,pci: features, cleanups, fixes
more memslots support in libvhost-user
support PCIe Gen5/Gen6 link speeds in pcie
more traces in vdpa
network simulation devices support in vdpa
SMBIOS type 9 descriptor implementation
Bump max_cpus to 4096 vcpus in q35
aw-bits and granule options in VIRTIO-IOMMU
Support report NUMA nodes for device memory using GI in acpi
Beginning of shutdown event support in pvpanic
fixes, cleanups all over the place.
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iQFDBAABCAAtFiEEXQn9CHHI+FuUyooNKB8NuNKNVGkFAmXw0TMPHG1zdEByZWRo
# YXQuY29tAAoJECgfDbjSjVRp8x4H+gLMoGwaGAX7gDGPgn2Ix4j/3kO77ZJ9X9k/
# 1KqZu/9eMS1j2Ei+vZqf05w7qRjxxhwDq3ilEXF/+UFqgAehLqpRRB8j5inqvzYt
# +jv0DbL11PBp/oFjWcytm5CbiVsvq8KlqCF29VNzc162XdtcduUOWagL96y8lJfZ
# uPrOoyeR7SMH9lp3LLLHWgu+9W4nOS03RroZ6Umj40y5B7yR0Rrppz8lMw5AoQtr
# 0gMRnFhYXeiW6CXdz+Tzcr7XfvkkYDi/j7ibiNSURLBfOpZa6Y8+kJGKxz5H1K1G
# 6ZY4PBcOpQzl+NMrktPHogczgJgOK10t+1i/R3bGZYw2Qn/93Eg=
# =C0UU
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 12 Mar 2024 22:03:31 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
* tag 'for_upstream' of https://git.kernel.org/pub/scm/virt/kvm/mst/qemu: (68 commits)
docs/specs/pvpanic: document shutdown event
hw/cxl: Fix missing reserved data in CXL Device DVSEC
hmat acpi: Fix out of bounds access due to missing use of indirection
hmat acpi: Do not add Memory Proximity Domain Attributes Structure targetting non existent memory.
qemu-options.hx: Document the virtio-iommu-pci aw-bits option
hw/arm/virt: Set virtio-iommu aw-bits default value to 48
hw/i386/q35: Set virtio-iommu aw-bits default value to 39
virtio-iommu: Add an option to define the input range width
virtio-iommu: Trace domain range limits as unsigned int
qemu-options.hx: Document the virtio-iommu-pci granule option
virtio-iommu: Change the default granule to the host page size
virtio-iommu: Add a granule property
hw/i386/acpi-build: Add support for SRAT Generic Initiator structures
hw/acpi: Implement the SRAT GI affinity structure
qom: new object to associate device to NUMA node
hw/i386/pc: Inline pc_cmos_init() into pc_cmos_init_late() and remove it
hw/i386/pc: Set "normal" boot device order in pc_basic_device_init()
hw/i386/pc: Avoid one use of the current_machine global
hw/i386/pc: Remove "rtc_state" link again
Revert "hw/i386/pc: Confine system flash handling to pc_sysfw"
...
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
# Conflicts:
# hw/core/machine.c
Shutdown requests are normally hardware dependent.
By extending pvpanic to also handle shutdown requests, guests can
submit such requests with an easily implementable and cross-platform
mechanism.
Signed-off-by: Thomas Weißschuh <thomas@t-8ch.de>
Message-Id: <20240310-pvpanic-shutdown-spec-v1-1-b258e182ce55@t-8ch.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The r3.1 specification introduced a new 2 byte field, but
to maintain DWORD alignment, a additional 2 reserved bytes
were added. Forgot those in updating the structure definition
but did include them in the size define leading to a buffer
overrun.
Also use the define so that we don't duplicate the value.
Fixes: Coverity ID 1534095 buffer overrun
Fixes: 8700ee15de ("hw/cxl: Standardize all references on CXL r3.1 and minor updates")
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-Id: <20240308143831.6256-1-Jonathan.Cameron@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
With a numa set up such as
-numa nodeid=0,cpus=0 \
-numa nodeid=1,memdev=mem \
-numa nodeid=2,cpus=1
and appropriate hmat_lb entries the initiator list is correctly
computed and writen to HMAT as 0,2 but then the LB data is accessed
using the node id (here 2), landing outside the entry_list array.
Stash the reverse lookup when writing the initiator list and use
it to get the correct array index index.
Fixes: 4586a2cb83 ("hmat acpi: Build System Locality Latency and Bandwidth Information Structure(s)")
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-Id: <20240307160326.31570-3-Jonathan.Cameron@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
If qemu is started with a proximity node containing CPUs alone,
it will provide one of these structures to say memory in this
node is directly connected to itself.
This description is arguably pointless even if there is memory
in the node. If there is no memory present, and hence no SRAT
entry it breaks Linux HMAT passing and the table is rejected.
https://elixir.bootlin.com/linux/v6.7/source/drivers/acpi/numa/hmat.c#L444
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-Id: <20240307160326.31570-2-Jonathan.Cameron@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Document the new aw-bits option.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20240307134445.92296-10-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
On ARM we set 48b as a default (matching SMMUv3 SMMU_IDR5.VAX == 0).
hw_compat_8_2 is used to handle the compatibility for machine types
before 9.0 (default was 64 bits).
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Zhenzhong Duan <Zhenzhong.duan@intel.com>
Message-Id: <20240307134445.92296-9-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently the default input range can extend to 64 bits. On x86,
when the virtio-iommu protects vfio devices, the physical iommu
may support only 39 bits. Let's set the default to 39, as done
for the intel-iommu.
We use hw_compat_8_2 to handle the compatibility for machines
before 9.0 which used to have a virtio-iommu default input range
of 64 bits.
Of course if aw-bits is set from the command line, the default
is overriden.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20240307134445.92296-8-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
aw-bits is a new option that allows to set the bit width of
the input address range. This value will be used as a default for
the device config input_range.end. By default it is set to 64 bits
which is the current value.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Message-Id: <20240307134445.92296-7-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Use %u format to trace domain_range limits.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Message-Id: <20240307134445.92296-6-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
We are missing an entry for the virtio-iommu-pci device. Add the
information on which machine it is currently supported and document
the new granule option.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Message-Id: <20240307134445.92296-5-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
We used to set the default granule to 4KB but with VFIO assignment
it makes more sense to use the actual host page size.
Indeed when hotplugging a VFIO device protected by a virtio-iommu
on a 64kB/64kB host/guest config, we current get a qemu crash:
"vfio: DMA mapping failed, unable to continue"
This is due to the hot-attached VFIO device calling
memory_region_iommu_set_page_size_mask() with 64kB granule
whereas the virtio-iommu granule was already frozen to 4KB on
machine init done.
Set the granule property to "host" and introduce a new compat.
The page size mask used before 9.0 was qemu_target_page_mask().
Since the virtio-iommu currently only supports x86_64 and aarch64,
this matched a 4KB granule.
Note that the new default will prevent 4kB guest on 64kB host
because the granule will be set to 64kB which would be larger
than the guest page size. In that situation, the virtio-iommu
driver fails on viommu_domain_finalise() with
"granule 0x10000 larger than system page size 0x1000".
In that case the workaround is to request 4K granule.
The current limitation of global granule in the virtio-iommu
should be removed and turned into per domain granule. But
until we get this upgraded, this new default is probably
better because I don't think anyone is currently interested in
running a 4KB page size guest with virtio-iommu on a 64KB host.
However supporting 64kB guest on 64kB host with virtio-iommu and
VFIO looks a more important feature.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Message-Id: <20240307134445.92296-4-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This allows to choose which granule will be used by
default by the virtio-iommu. Current page size mask
default is qemu_target_page_mask so this translates
into a 4k granule on ARM and x86_64 where virtio-iommu
is supported.
Signed-off-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Message-Id: <20240307134445.92296-3-eric.auger@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The acpi-generic-initiator object is added to allow a host device
to be linked with a NUMA node. Qemu use it to build the SRAT
Generic Initiator Affinity structure [1]. Add support for i386.
[1] ACPI Spec 6.3, Section 5.2.16.6
Suggested-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ankit Agrawal <ankita@nvidia.com>
Message-Id: <20240308145525.10886-4-ankita@nvidia.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
ACPI spec provides a scheme to associate "Generic Initiators" [1]
(e.g. heterogeneous processors and accelerators, GPUs, and I/O devices with
integrated compute or DMA engines GPUs) with Proximity Domains. This is
achieved using Generic Initiator Affinity Structure in SRAT. During bootup,
Linux kernel parse the ACPI SRAT to determine the PXM ids and create a NUMA
node for each unique PXM ID encountered. Qemu currently do not implement
these structures while building SRAT.
Add GI structures while building VM ACPI SRAT. The association between
device and node are stored using acpi-generic-initiator object. Lookup
presence of all such objects and use them to build these structures.
The structure needs a PCI device handle [2] that consists of the device BDF.
The vfio-pci device corresponding to the acpi-generic-initiator object is
located to determine the BDF.
[1] ACPI Spec 6.3, Section 5.2.16.6
[2] ACPI Spec 6.3, Table 5.80
Cc: Jonathan Cameron <qemu-devel@nongnu.org>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Cedric Le Goater <clg@redhat.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Signed-off-by: Ankit Agrawal <ankita@nvidia.com>
Message-Id: <20240308145525.10886-3-ankita@nvidia.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
NVIDIA GPU's support MIG (Mult-Instance GPUs) feature [1], which allows
partitioning of the GPU device resources (including device memory) into
several (upto 8) isolated instances. Each of the partitioned memory needs
a dedicated NUMA node to operate. The partitions are not fixed and they
can be created/deleted at runtime.
Unfortunately Linux OS does not provide a means to dynamically create/destroy
NUMA nodes and such feature implementation is not expected to be trivial. The
nodes that OS discovers at the boot time while parsing SRAT remains fixed. So
we utilize the Generic Initiator (GI) Affinity structures that allows
association between nodes and devices. Multiple GI structures per BDF is
possible, allowing creation of multiple nodes by exposing unique PXM in each
of these structures.
Implement the mechanism to build the GI affinity structures as Qemu currently
does not. Introduce a new acpi-generic-initiator object to allow host admin
link a device with an associated NUMA node. Qemu maintains this association
and use this object to build the requisite GI Affinity Structure.
When multiple NUMA nodes are associated with a device, it is required to
create those many number of acpi-generic-initiator objects, each representing
a unique device:node association.
Following is one of a decoded GI affinity structure in VM ACPI SRAT.
[0C8h 0200 1] Subtable Type : 05 [Generic Initiator Affinity]
[0C9h 0201 1] Length : 20
[0CAh 0202 1] Reserved1 : 00
[0CBh 0203 1] Device Handle Type : 01
[0CCh 0204 4] Proximity Domain : 00000007
[0D0h 0208 16] Device Handle : 00 00 20 00 00 00 00 00 00 00 00
00 00 00 00 00
[0E0h 0224 4] Flags (decoded below) : 00000001
Enabled : 1
[0E4h 0228 4] Reserved2 : 00000000
[0E8h 0232 1] Subtable Type : 05 [Generic Initiator Affinity]
[0E9h 0233 1] Length : 20
An admin can provide a range of acpi-generic-initiator objects, each
associating a device (by providing the id through pci-dev argument)
to the desired NUMA node (using the node argument). Currently, only PCI
device is supported.
For the grace hopper system, create a range of 8 nodes and associate that
with the device using the acpi-generic-initiator object. While a configuration
of less than 8 nodes per device is allowed, such configuration will prevent
utilization of the feature to the fullest. The following sample creates 8
nodes per PCI device for a VM with 2 PCI devices and link them to the
respecitve PCI device using acpi-generic-initiator objects:
-numa node,nodeid=2 -numa node,nodeid=3 -numa node,nodeid=4 \
-numa node,nodeid=5 -numa node,nodeid=6 -numa node,nodeid=7 \
-numa node,nodeid=8 -numa node,nodeid=9 \
-device vfio-pci-nohotplug,host=0009:01:00.0,bus=pcie.0,addr=04.0,rombar=0,id=dev0 \
-object acpi-generic-initiator,id=gi0,pci-dev=dev0,node=2 \
-object acpi-generic-initiator,id=gi1,pci-dev=dev0,node=3 \
-object acpi-generic-initiator,id=gi2,pci-dev=dev0,node=4 \
-object acpi-generic-initiator,id=gi3,pci-dev=dev0,node=5 \
-object acpi-generic-initiator,id=gi4,pci-dev=dev0,node=6 \
-object acpi-generic-initiator,id=gi5,pci-dev=dev0,node=7 \
-object acpi-generic-initiator,id=gi6,pci-dev=dev0,node=8 \
-object acpi-generic-initiator,id=gi7,pci-dev=dev0,node=9 \
-numa node,nodeid=10 -numa node,nodeid=11 -numa node,nodeid=12 \
-numa node,nodeid=13 -numa node,nodeid=14 -numa node,nodeid=15 \
-numa node,nodeid=16 -numa node,nodeid=17 \
-device vfio-pci-nohotplug,host=0009:01:01.0,bus=pcie.0,addr=05.0,rombar=0,id=dev1 \
-object acpi-generic-initiator,id=gi8,pci-dev=dev1,node=10 \
-object acpi-generic-initiator,id=gi9,pci-dev=dev1,node=11 \
-object acpi-generic-initiator,id=gi10,pci-dev=dev1,node=12 \
-object acpi-generic-initiator,id=gi11,pci-dev=dev1,node=13 \
-object acpi-generic-initiator,id=gi12,pci-dev=dev1,node=14 \
-object acpi-generic-initiator,id=gi13,pci-dev=dev1,node=15 \
-object acpi-generic-initiator,id=gi14,pci-dev=dev1,node=16 \
-object acpi-generic-initiator,id=gi15,pci-dev=dev1,node=17 \
Link: https://www.nvidia.com/en-in/technologies/multi-instance-gpu [1]
Cc: Jonathan Cameron <qemu-devel@nongnu.org>
Cc: Alex Williamson <alex.williamson@redhat.com>
Cc: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Acked-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Ankit Agrawal <ankita@nvidia.com>
Message-Id: <20240308145525.10886-2-ankita@nvidia.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Now that pc_cmos_init() doesn't populate the X86MachineState::rtc attribute any
longer, its duties can be merged into pc_cmos_init_late() which is called within
machine_done notifier. This frees pc_piix and pc_q35 from explicit CMOS
initialization.
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-Id: <20240303185332.1408-5-shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The boot device order may change during the lifetime of a VM. Usually, the
"normal" order is set once during machine init(). However, if a user specifies
`-boot once=...`, the "normal" order is overwritten by the "once" order just
before machine_done, and a reset handler is registered which restores the
"normal" order during the next reset.
In the next patch, pc_cmos_init() will be inlined into pc_cmos_init_late() which
runs during machine_done. This means that the "once" boot order would be
overwritten again with the "normal" boot order -- which renders the user's
choice ineffective. Fix this by setting the "normal" boot order in
pc_basic_device_init() which already registers the boot_set() handler.
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-Id: <20240303185332.1408-4-shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
The RTC can be accessed through the X86 machine instance, so rather than passing
the RTC it's possible to pass the machine state instead. This avoids
pc_boot_set() from having to access the current_machine global.
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-Id: <20240303185332.1408-3-shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Commit 99e1c1137b "hw/i386/pc: Populate RTC attribute directly" made linking
the "rtc_state" property unnecessary and removed it. Commit 84e945aad2 "vl,
pc: turn -no-fd-bootchk into a machine property" accidently reintroduced the
link. Remove it again since it is not needed.
Fixes: 84e945aad2 "vl, pc: turn -no-fd-bootchk into a machine property"
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-Id: <20240303185332.1408-2-shentey@gmail.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Specifying the property `-M pflash0` results in a regression:
qemu-system-x86_64: Property 'pc-q35-9.0-machine.pflash0' not found
Revert the change for now until a solution is found.
This reverts commit 6f6ad2b245.
Reported-by: Volker Rümelin <vr_qemu@t-online.de>
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-Id: <20240226215909.30884-3-shentey@gmail.com>
Tested-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Commit 6f6ad2b245 "hw/i386/pc: Confine system flash handling to pc_sysfw"
causes a regression when specifying the property `-M pflash0` in the PCI PC
machines:
qemu-system-x86_64: Property 'pc-q35-9.0-machine.pflash0' not found
In order to revert the commit, the commit below must be reverted first.
This reverts commit cb05cc1602.
Signed-off-by: Bernhard Beschow <shentey@gmail.com>
Message-Id: <20240226215909.30884-2-shentey@gmail.com>
Tested-by: Alex Williamson <alex.williamson@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Since commit f10a570b093e6 ("KVM: x86: Add CONFIG_KVM_MAX_NR_VCPUS to allow up to 4096 vCPUs")
Linux kernel can support upto a maximum number of 4096 vcpus when MAXSMP is
enabled in the kernel. At present, QEMU has been tested to correctly boot a
linux guest with 4096 vcpus using the current edk2 upstream master branch that
has the fixes corresponding to the following two PRs:
https://github.com/tianocore/edk2/pull/5410https://github.com/tianocore/edk2/pull/5418
The changes merged into edk2 with the above PRs will be in the upcoming 2024-05
release. With current seabios firmware, it boots fine with 4096 vcpus already.
So bump up the value max_cpus to 4096 for q35 machines versions 9 and newer.
Q35 machines versions 8.2 and older continue to support 1024 maximum vcpus
as before for compatibility reasons.
If KVM is not able to support the specified number of vcpus, QEMU would
return the following error messages:
$ ./qemu-system-x86_64 -cpu host -accel kvm -machine q35 -smp 1728
qemu-system-x86_64: -accel kvm: warning: Number of SMP cpus requested (1728) exceeds the recommended cpus supported by KVM (12)
qemu-system-x86_64: -accel kvm: warning: Number of hotpluggable cpus requested (1728) exceeds the recommended cpus supported by KVM (12)
Number of SMP cpus requested (1728) exceeds the maximum cpus supported by KVM (1024)
Cc: Daniel P. Berrangé <berrange@redhat.com>
Cc: Igor Mammedov <imammedo@redhat.com>
Cc: Michael S. Tsirkin <mst@redhat.com>
Cc: Julia Suvorova <jusual@redhat.com>
Cc: kraxel@redhat.com
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Gerd Hoffmann <kraxel@redhat.com>
Signed-off-by: Ani Sinha <anisinha@redhat.com>
Message-Id: <20240228143351.3967-1-anisinha@redhat.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Call pcie_sriov_pf_reset() from pci_do_device_reset() just as we do
for msi_reset() and msix_reset() to prevent duplicating code for each
SR-IOV PF.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20240228-reuse-v8-5-282660281e60@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Sriram Yagnaraman <sriram.yagnaraman@ericsson.com>
The spec does not NumVFs is reset after disabling VFs except when
resetting the PF. Clearing it is guest visible and out of spec, even
though Linux doesn't rely on this value being preserved, so we never
noticed.
Fixes: 7c0fa8dff8 ("pcie: Add support for Single Root I/O Virtualization (SR/IOV)")
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20240228-reuse-v8-4-282660281e60@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
pcie_sriov_pf_disable_vfs() is called when resetting the PF, but it only
disables VFs and does not reset SR-IOV extended capability, leaking the
state and making the VF Enable register inconsistent with the actual
state.
Replace pcie_sriov_pf_disable_vfs() with pcie_sriov_pf_reset(), which
does not only disable VFs but also resets the capability.
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20240228-reuse-v8-3-282660281e60@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Sriram Yagnaraman <sriram.yagnaraman@ericsson.com>
The guest may write NumVFs greater than TotalVFs and that can lead
to buffer overflow in VF implementations.
Cc: qemu-stable@nongnu.org
Fixes: CVE-2024-26327
Fixes: 7c0fa8dff8 ("pcie: Add support for Single Root I/O Virtualization (SR/IOV)")
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20240228-reuse-v8-2-282660281e60@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Sriram Yagnaraman <sriram.yagnaraman@ericsson.com>
nvme_sriov_pre_write_ctrl() used to directly inspect SR-IOV
configurations to know the number of VFs being disabled due to SR-IOV
configuration writes, but the logic was flawed and resulted in
out-of-bound memory access.
It assumed PCI_SRIOV_NUM_VF always has the number of currently enabled
VFs, but it actually doesn't in the following cases:
- PCI_SRIOV_NUM_VF has been set but PCI_SRIOV_CTRL_VFE has never been.
- PCI_SRIOV_NUM_VF was written after PCI_SRIOV_CTRL_VFE was set.
- VFs were only partially enabled because of realization failure.
It is a responsibility of pcie_sriov to interpret SR-IOV configurations
and pcie_sriov does it correctly, so use pcie_sriov_num_vfs(), which it
provides, to get the number of enabled VFs before and after SR-IOV
configuration writes.
Cc: qemu-stable@nongnu.org
Fixes: CVE-2024-26328
Fixes: 11871f53ef ("hw/nvme: Add support for the Virtualization Management command")
Suggested-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-Id: <20240228-reuse-v8-1-282660281e60@daynix.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Felix Wu <flwu@google.com>
Signed-off-by: Nabih Estefan <nabihestefan@google.com>
Message-Id: <20240221170027.1027325-3-nabihestefan@google.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Version 2.1+.
Signed-off-by: Felix Wu <flwu@google.com>
Signed-off-by: Nabih Estefan <nabihestefan@google.com>
Message-Id: <20240221170027.1027325-2-nabihestefan@google.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
IOAPICCommonClass implements its own private realize(), and this private
realize() allows error.
Since IOAPICCommonClass.realize() returns void, to check the error,
dereference @errp with ERRP_GUARD().
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20240223085653.1255438-8-zhao1.liu@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
As the comment in qapi/error, dereferencing @errp requires
ERRP_GUARD():
* = Why, when and how to use ERRP_GUARD() =
*
* Without ERRP_GUARD(), use of the @errp parameter is restricted:
* - It must not be dereferenced, because it may be null.
...
* ERRP_GUARD() lifts these restrictions.
*
* To use ERRP_GUARD(), add it right at the beginning of the function.
* @errp can then be used without worrying about the argument being
* NULL or &error_fatal.
*
* Using it when it's not needed is safe, but please avoid cluttering
* the source with useless code.
But in iommufd_cdev_getfd(), @errp is dereferenced without ERRP_GUARD():
if (*errp) {
error_prepend(errp, VFIO_MSG_PREFIX, path);
}
Currently, since vfio_attach_device() - the caller of
iommufd_cdev_getfd() - is always called in DeviceClass.realize() context
and doesn't get the NULL @errp parameter, iommufd_cdev_getfd()
hasn't triggered the bug that dereferencing the NULL @errp.
To follow the requirement of @errp, add missing ERRP_GUARD() in
iommufd_cdev_getfd().
Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20240223085653.1255438-7-zhao1.liu@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
As the comment in qapi/error, dereferencing @errp requires
ERRP_GUARD():
* = Why, when and how to use ERRP_GUARD() =
*
* Without ERRP_GUARD(), use of the @errp parameter is restricted:
* - It must not be dereferenced, because it may be null.
...
* ERRP_GUARD() lifts these restrictions.
*
* To use ERRP_GUARD(), add it right at the beginning of the function.
* @errp can then be used without worrying about the argument being
* NULL or &error_fatal.
*
* Using it when it's not needed is safe, but please avoid cluttering
* the source with useless code.
But in cxl_usp_realize(), @errp is dereferenced without ERRP_GUARD():
cxl_doe_cdat_init(cxl_cstate, errp);
if (*errp) {
goto err_cap;
}
Here we check *errp, because cxl_doe_cdat_init() returns void. And since
cxl_usp_realize() - as a PCIDeviceClass.realize() method - doesn't get
the NULL @errp parameter, it hasn't triggered the bug that dereferencing
the NULL @errp.
To follow the requirement of @errp, add missing ERRP_GUARD() in
cxl_usp_realize().
Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20240223085653.1255438-6-zhao1.liu@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
As the comment in qapi/error, dereferencing @errp requires
ERRP_GUARD():
* = Why, when and how to use ERRP_GUARD() =
*
* Without ERRP_GUARD(), use of the @errp parameter is restricted:
* - It must not be dereferenced, because it may be null.
...
* ERRP_GUARD() lifts these restrictions.
*
* To use ERRP_GUARD(), add it right at the beginning of the function.
* @errp can then be used without worrying about the argument being
* NULL or &error_fatal.
*
* Using it when it's not needed is safe, but please avoid cluttering
* the source with useless code.
But in trng_prop_fault_event_set, @errp is dereferenced without
ERRP_GUARD():
visit_type_uint32(v, name, events, errp);
if (*errp) {
return;
}
Currently, since trng_prop_fault_event_set() doesn't get the NULL @errp
parameter as a "set" method of object property, it hasn't triggered the
bug that dereferencing the NULL @errp.
And since visit_type_uint32() returns bool, check the returned bool
directly instead of dereferencing @errp, then we needn't the add missing
ERRP_GUARD().
Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20240223085653.1255438-5-zhao1.liu@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
As the comment in qapi/error, dereferencing @errp requires
ERRP_GUARD():
* = Why, when and how to use ERRP_GUARD() =
*
* Without ERRP_GUARD(), use of the @errp parameter is restricted:
* - It must not be dereferenced, because it may be null.
...
* ERRP_GUARD() lifts these restrictions.
*
* To use ERRP_GUARD(), add it right at the beginning of the function.
* @errp can then be used without worrying about the argument being
* NULL or &error_fatal.
*
* Using it when it's not needed is safe, but please avoid cluttering
* the source with useless code.
But in ct3_realize(), @errp is dereferenced without ERRP_GUARD():
cxl_doe_cdat_init(cxl_cstate, errp);
if (*errp) {
goto err_free_special_ops;
}
Here we check *errp, because cxl_doe_cdat_init() returns void. And
ct3_realize() - as a PCIDeviceClass.realize() method - doesn't get the
NULL @errp parameter, it hasn't triggered the bug that dereferencing
the NULL @errp.
To follow the requirement of @errp, add missing ERRP_GUARD() in
ct3_realize().
Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20240223085653.1255438-4-zhao1.liu@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
As the comment in qapi/error, dereferencing @errp requires
ERRP_GUARD():
* = Why, when and how to use ERRP_GUARD() =
*
* Without ERRP_GUARD(), use of the @errp parameter is restricted:
* - It must not be dereferenced, because it may be null.
...
* ERRP_GUARD() lifts these restrictions.
*
* To use ERRP_GUARD(), add it right at the beginning of the function.
* @errp can then be used without worrying about the argument being
* NULL or &error_fatal.
*
* Using it when it's not needed is safe, but please avoid cluttering
* the source with useless code.
But in macfb_nubus_realize(), @errp is dereferenced without
ERRP_GUARD():
ndc->parent_realize(dev, errp);
if (*errp) {
return;
}
Here we check *errp, because the ndc->parent_realize(), as a
DeviceClass.realize() callback, returns void. And since
macfb_nubus_realize(), also as a DeviceClass.realize(), doesn't get the
NULL @errp parameter, it hasn't triggered the bug that dereferencing the
NULL @errp.
To follow the requirement of @errp, add missing ERRP_GUARD() in
macfb_nubus_realize().
Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20240223085653.1255438-3-zhao1.liu@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
As the comment in qapi/error, dereferencing @errp requires
ERRP_GUARD():
* = Why, when and how to use ERRP_GUARD() =
*
* Without ERRP_GUARD(), use of the @errp parameter is restricted:
* - It must not be dereferenced, because it may be null.
...
* ERRP_GUARD() lifts these restrictions.
*
* To use ERRP_GUARD(), add it right at the beginning of the function.
* @errp can then be used without worrying about the argument being
* NULL or &error_fatal.
*
* Using it when it's not needed is safe, but please avoid cluttering
* the source with useless code.
But in cxl_fixed_memory_window_config(), @errp is dereferenced in 2
places without ERRP_GUARD():
fw->enc_int_ways = cxl_interleave_ways_enc(fw->num_targets, errp);
if (*errp) {
return;
}
and
fw->enc_int_gran =
cxl_interleave_granularity_enc(object->interleave_granularity,
errp);
if (*errp) {
return;
}
For the above 2 places, we check "*errp", because neither function
returns a suitable error code. And since machine_set_cfmw() - the caller
of cxl_fixed_memory_window_config() - doesn't get the NULL @errp
parameter as the "set" method of object property,
cxl_fixed_memory_window_config() hasn't triggered the bug that
dereferencing the NULL @errp.
To follow the requirement of @errp, add missing ERRP_GUARD() in
cxl_fixed_memory_window_config().
Suggested-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20240223085653.1255438-2-zhao1.liu@linux.intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Acked-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
This patch adds support for VDPA network simulation devices.
The device is developed based on virtio-net and tap backend,
and supports hardware live migration function.
For more details, please refer to "docs/system/devices/vdpa-net.rst"
Signed-off-by: Hao Chen <chenh@yusur.tech>
Message-Id: <20240221073802.2888022-1-chenh@yusur.tech>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Shared objects lack spoofing protection.
For VHOST_USER_BACKEND_SHARED_OBJECT_REMOVE messages
received by the vhost-user interface, any backend was
allowed to remove entries from the shared table just
by knowing the UUID. Only the owner of the entry
shall be allowed to removed their resources
from the table.
To fix that, add a check for all
*SHARED_OBJECT_REMOVE messages received.
A vhost device can only remove TYPE_VHOST_DEV
entries that are owned by them, otherwise skip
the removal, and inform the device that the entry
has not been removed in the answer.
Signed-off-by: Albert Esteve <aesteve@redhat.com>
Acked-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-Id: <20240219143423.272012-2-aesteve@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>
The payload size returned by command VIRTIO_SND_R_PCM_INFO is
wrong. The code in process_cmd() assumes that all commands
return only a virtio_snd_hdr payload, but some commands like
VIRTIO_SND_R_PCM_INFO may return an additional payload.
Add a zero initialized payload_size variable to struct
virtio_snd_ctrl_command to allow for additional payloads.
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20240218083351.8524-1-vr_qemu@t-online.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This CXL component isn't allowed to have a RAS capability.
Whilst this should be harmless as software is not expected to look
here, good to clean it up.
Signed-off-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Message-Id: <20240215155206.2736-1-Jonathan.Cameron@huawei.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Sometimes, certain parts are not being skipped in
vhost_vdpa_listener_region_del, but they are skipped in
vhost_vdpa_listener_region_add, or vice versa. The vhost-vdpa code
expects all parts to maintain their properties, so we're adding a trace
to help with debugging when any part is skipped.
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20240215103616.330518-3-eperezma@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Next changes uses this variables, so avoid call repeatedly to memory
region functions. No functional change intended.
Signed-off-by: Eugenio Pérez <eperezma@redhat.com>
Message-Id: <20240215103616.330518-2-eperezma@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
This patch extends the PCIe link speed option so that slots can be
configured as supporting 32GT/s (Gen5) or 64GT/s (Gen5) speeds.
This is as simple as setting the appropriate bit in LnkCap2 and
the appropriate value in LnkCap and LnkCtl2.
Signed-off-by: Lukas Stockner <lstockner@genesiscloud.com>
Message-Id: <20240215012326.3272366-1-lstockner@genesiscloud.com>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
We already use MADV_NORESERVE to deal with sparse memory regions. Let's
also set madvise(MADV_DONTDUMP), otherwise a crash of the process can
result in us allocating all memory in the mmap'ed region for dumping
purposes.
This change implies that the mmap'ed rings won't be included in a
coredump. If ever required for debugging purposes, we could mark only
the mapped rings MADV_DODUMP.
Ignore errors during madvise() for now.
Reviewed-by: Raphael Norwitz <raphael@enfabrica.net>
Acked-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20240214151701.29906-15-david@redhat.com>
Tested-by: Mario Casquero <mcasquer@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Currently, we try to remap all rings whenever we add a single new memory
region. That doesn't quite make sense, because we already map rings when
setting the ring address, and panic if that goes wrong. Likely, that
handling was simply copied from set_mem_table code, where we actually
have to remap all rings.
Remapping all rings might require us to walk quite a lot of memory
regions to perform the address translations. Ideally, we'd simply remove
that remapping.
However, let's be a bit careful. There might be some weird corner cases
where we might temporarily remove a single memory region (e.g., resize
it), that would have worked for now. Further, a ring might be located on
hotplugged memory, and as the VM reboots, we might unplug that memory, to
hotplug memory before resetting the ring addresses.
So let's unmap affected rings as we remove a memory region, and try
dynamically mapping the ring again when required.
Acked-by: Raphael Norwitz <raphael@enfabrica.net>
Acked-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20240214151701.29906-14-david@redhat.com>
Tested-by: Mario Casquero <mcasquer@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Let's factor it out to prepare for further changes.
Reviewed-by: Raphael Norwitz <raphael@enfabrica.net>
Acked-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20240214151701.29906-13-david@redhat.com>
Tested-by: Mario Casquero <mcasquer@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
In the past, QEMU would create memory regions that could partially cover
hugetlb pages, making mmap() fail if we would use the mmap_offset as an
fd_offset. For that reason, we never used the mmap_offset as an offset into
the fd and instead always mapped the fd from the very start.
However, that can easily result in us mmap'ing a lot of unnecessary
parts of an fd, possibly repeatedly.
QEMU nowadays does not create memory regions that partially cover huge
pages -- it never really worked with postcopy. QEMU handles merging of
regions that partially cover huge pages (due to holes in boot memory) since
2018 in c1ece84e7c ("vhost: Huge page align and merge").
Let's be a bit careful and not unconditionally convert the
mmap_offset into an fd_offset. Instead, let's simply detect the hugetlb
size and pass as much as we can as fd_offset, making sure that we call
mmap() with a properly aligned offset.
With QEMU and a virtio-mem device that is fully plugged (50GiB using 50
memslots) the qemu-storage daemon process consumes in the VA space
1281GiB before this change and 58GiB after this change.
================ Vhost user message ================
Request: VHOST_USER_ADD_MEM_REG (37)
Flags: 0x9
Size: 40
Fds: 59
Adding region 4
guest_phys_addr: 0x0000000200000000
memory_size: 0x0000000040000000
userspace_addr: 0x00007fb73bffe000
old mmap_offset: 0x0000000080000000
fd_offset: 0x0000000080000000
new mmap_offset: 0x0000000000000000
mmap_addr: 0x00007f02f1bdc000
Successfully added new region
================ Vhost user message ================
Request: VHOST_USER_ADD_MEM_REG (37)
Flags: 0x9
Size: 40
Fds: 59
Adding region 5
guest_phys_addr: 0x0000000240000000
memory_size: 0x0000000040000000
userspace_addr: 0x00007fb77bffe000
old mmap_offset: 0x00000000c0000000
fd_offset: 0x00000000c0000000
new mmap_offset: 0x0000000000000000
mmap_addr: 0x00007f0284000000
Successfully added new region
Reviewed-by: Raphael Norwitz <raphael@enfabrica.net>
Acked-by: Stefano Garzarella <sgarzare@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20240214151701.29906-12-david@redhat.com>
Tested-by: Mario Casquero <mcasquer@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
