Merge tag 'pull-loongarch-20240912' of https://gitlab.com/gaosong/qemu into staging

pull-loongarch-20240912

# -----BEGIN PGP SIGNATURE-----
#
# iLMEAAEKAB0WIQS4/x2g0v3LLaCcbCxAov/yOSY+3wUCZuLmLgAKCRBAov/yOSY+
# 38JNA/9UdorT4a7H+H5PhNeEu2EHDgMPb7+gxyYKw03mOG2MB3KFzkK0LRQShaPt
# ADJmIqAFlc9SJLkbo6ELMDl+ZnUU9OdC/P6YU5iBG71zx1PonMwuyJTWhlBwxWcG
# +OB8aDBUALoe/Gb4za152I84cR08g58TgLnXNfEkCM8lnPfAug==
# =Plwu
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 12 Sep 2024 14:01:34 BST
# gpg:                using RSA key B8FF1DA0D2FDCB2DA09C6C2C40A2FFF239263EDF
# gpg: Good signature from "Song Gao <m17746591750@163.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B8FF 1DA0 D2FD CB2D A09C  6C2C 40A2 FFF2 3926 3EDF

* tag 'pull-loongarch-20240912' of https://gitlab.com/gaosong/qemu:
  hw/loongarch: Add acpi SPCR table support
  hw/loongarch: virt: pass random seed to fdt
  hw/loongarch: virt: support up to 4 serial ports
  target/loongarch: Support QMP dump-guest-memory
  target/loongarch/kvm: Add vCPU reset function
  hw/loongarch: Remove default enable with VIRTIO_VGA device
  target/loongarch: Add compatible support about VM reboot

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

Merge tag 'pull-sparc-20240911' of https://gitlab.com/rth7680/qemu into staging

target/sparc: Implement single entry FP Queue

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmbifAAdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV+XAwgAlj//8JuNoRB/2hi0
# gU3Ifjrs+r+AZrcsG7pTOmYTZa6cYqJX4XsYoNq1S4FHky239vNKPQOQEadkmLGv
# wKH0fBjzvydOKRfrhEK2VLlhMyhGyuv59psfCCUB5HZEiueSHFFAvfjUtKNpjzRT
# KE2fwL6iKK3IXeKC6ynq0bkC/OymnLUYSgSslA6C1x1sReNz5Y6ZsGUEZRwODY4f
# q6s6JS2aBn1L9nJTzwXH/J5Ue8iix53d6EZ42QHqqwzRvAWHtfFqoMLc9P6Dg8P7
# FmiwHAErwr7Pj5cqcnl2C0zTp3LXg5xXpTJysi8CFJvCsObNRh9gL15W3xy9qBFX
# 2WfqWQ==
# =kxM7
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 12 Sep 2024 06:28:32 BST
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* tag 'pull-sparc-20240911' of https://gitlab.com/rth7680/qemu:
  target/sparc: Add gen_trap_if_nofpu_fpexception
  target/sparc: Implement STDFQ
  target/sparc: Add FSR_QNE to tb_flags
  target/sparc: Populate sparc32 FQ when raising fp exception
  target/sparc: Add FQ and FSR.QNE

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

.gitlab-ci.d/crossbuilds.yml: Force 'make check' single-threaded for cross-i686-tci

The cross-i686-tci CI job is persistently flaky with various tests
hitting timeouts.  One theory for why this is happening is that we're
running too many tests in parallel and so sometimes a test gets
starved of CPU and isn't able to complete within the timeout.

(The environment this CI job runs in seems to cause us to default
to a parallelism of 9 in the main CI.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-id: 20240912151003.2045031-1-peter.maydell@linaro.org

Merge tag 'for_upstream' of https://git./virt/kvm/mst/qemu into staging

virtio,pc,pci: features, fixes, cleanups

i286 acpi speedup by precomputing _PRT by Ricardo Ribalda
vhost_net speedup by using MR transactions by Zuo Boqun
ich9 gained support for periodic and swsmi timer by Dominic Prinz

Fixes, cleanups all over the place.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
# -----BEGIN PGP SIGNATURE-----
#
# iQFDBAABCAAtFiEEXQn9CHHI+FuUyooNKB8NuNKNVGkFAmbhoCUPHG1zdEByZWRo
# YXQuY29tAAoJECgfDbjSjVRptpUH/iR5AmJFpvAItqlPOvJiYDEch46C73tyrSws
# Kk/1EbGSL7mFFD5sfdSSV4Rw8CQBsmM/Dt5VDkJKsWnOLjkBQ2CYH0MYHktnrKcJ
# LlSk32HnY5p1DsXnJhgm5M7St8T3mV/oFdJCJAFgCmpx5uT8IRLrKETN8+30OaiY
# xo35xAKOAS296+xsWeVubKkMq7H4y2tdZLE/22gb8rlA8d96BJIeVLQ3y3IjeUPR
# 24q6c7zpObzGhYNZ/PzAKOn+YcVsV/lLAzKRZJTzTUPyG24BcjJTyyr/zNSYAgfk
# lLXzIZID3GThBmrCAiDZ1z6sfo3MRg2wNS/FBXtK6fPIuFxed+8=
# =ySRy
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 11 Sep 2024 14:50:29 BST
# 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:
  hw/acpi/ich9: Add periodic and swsmi timer
  virtio-mem: don't warn about THP sizes on a kernel without THP support
  hw/audio/virtio-sound: fix heap buffer overflow
  hw/cxl: fix physical address field in get scan media results output
  virtio-pci: Add lookup subregion of VirtIOPCIRegion MR
  vhost_net: configure all host notifiers in a single MR transaction
  tests/acpi: pc: update golden masters for DSDT
  hw/i386/acpi-build: Return a pre-computed _PRT table
  tests/acpi: pc: allow DSDT acpi table changes
  intel_iommu: Make PASID-cache and PIOTLB type invalid in legacy mode
  intel_iommu: Fix invalidation descriptor type field
  virtio: rename virtio_split_packed_update_used_idx
  hw/pci/pci-hmp-cmds: Avoid displaying bogus size in 'info pci'
  pci: don't skip function 0 occupancy verification for devfn auto assign
  hw/isa/vt82c686.c: Embed i8259 irq in device state instead of allocating
  hw: Move declaration of IRQState to header and add init function
  virtio: Always reset vhost devices
  virtio: Allow .get_vhost() without vhost_started

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

hw/loongarch: Add acpi SPCR table support

Serial port console redirection table can be used for default serial
port selection, like chosen stdout-path selection with FDT method.

With acpi SPCR table added, early debug console can be parsed from
SPCR table with simple kernel parameter earlycon rather than
earlycon=uart,mmio,0x1fe001e0

Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240907073037.243353-1-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>

hw/loongarch: virt: pass random seed to fdt

If the FDT contains /chosen/rng-seed, then the Linux RNG will use it to
initialize early. Set this using the usual guest random number
generation function.

This is the same procedure that's done in b91b6b5a2c ("hw/microblaze:
pass random seed to fdt"), e4b4f0b71c ("hw/riscv: virt: pass random seed
to fdt"), c6fe3e6b4c ("hw/openrisc: virt: pass random seed to fdt"),
67f7e426e5 ("hw/i386: pass RNG seed via setup_data entry"), c287941a4d
("hw/rx: pass random seed to fdt"), 5e19cc68fb ("hw/mips: boston: pass
random seed to fdt"), 6b23a67916 ("hw/nios2: virt: pass random seed to fdt")
c4b075318e ("hw/ppc: pass random seed to fdt"), and 5242876f37
("hw/arm/virt: dt: add rng-seed property").

These earlier commits later were amended to rerandomize the RNG seed on
snapshot load, but the LoongArch code somehow already does that, despite
not having this patch here, presumably due to some lucky copy and
pasting.

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240905153316.2038769-1-Jason@zx2c4.com>
Signed-off-by: Song Gao <gaosong@loongson.cn>

hw/loongarch: virt: support up to 4 serial ports

In order to support additional channels of communication using
`-serial`, add several serial ports, up to the standard 4 generally
supported by the 8250 driver.

Fixed: https://lore.kernel.org/all/20240907143439.2792924-1-Jason@zx2c4.com/

Signed-off-by: Jason A. Donenfeld <Jason@zx2c4.com>
Tested-by: Bibo Mao <maobibo@loongson.cn>
[gaosong: ACPI uart need't reverse order]
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240907143439.2792924-1-Jason@zx2c4.com>

target/loongarch: Support QMP dump-guest-memory

Add the support needed for creating prstatus elf notes. This allows
us to use QMP dump-guest-memory.

Now ELF notes of LoongArch only supports general elf notes, LSX and
LASX is not supported, since it is mainly used to dump guest memory.

Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Tested-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240822065245.2286214-1-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>

target/loongarch/kvm: Add vCPU reset function

KVM provides interface KVM_REG_LOONGARCH_VCPU_RESET to reset vCPU,
it can be used to clear internal state about kvm kernel. vCPU reset
function is added here for kvm mode.

Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240822022827.2273534-1-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>

hw/loongarch: Remove default enable with VIRTIO_VGA device

For virtio VGA deivce libvirt will select VIRTIO_VGA firstly rather than
VIRTIO_GPU, VIRTIO_VGA device supports frame buffer however it requires
legacy VGA compatible support. Frame buffer area 0xa0000 -- 0xc0000
conflicts with low memory area 0 -- 0x10000000.

Here remove default support for VIRTIO_VGA device, VIRTIO_GPU is prefered
on LoongArch system. For frame buffer video card support, standard VGA can
be used.

Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240823073050.2619484-1-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>

target/loongarch: Add compatible support about VM reboot

With edk2-stable202408 LoongArch UEFI bios, CSR PGD register is set only
if its value is equal to zero for boot cpu, it causes reboot issue. Since
CSR PGD register is changed with linux kernel, UEFI BIOS cannot use it.

Add workaround to clear CSR registers relative with TLB in function
loongarch_cpu_reset_hold(), so that VM can reboot with edk2-stable202408
UEFI bios.

Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20240827035807.3326293-1-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>

target/sparc: Add gen_trap_if_nofpu_fpexception

Model fp_exception state, in which only fp stores are allowed
until such time as the FQ has been flushed.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Tested-by: Carl Hauser <chauser@pullman.com>

target/sparc: Implement STDFQ

Invalid encoding of addr should raise TT_ILL_INSN, so
check before supervisor, which might raise TT_PRIV_INSN.
Clear QNE after execution.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2340
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Tested-by: Carl Hauser <chauser@pullman.com>

target/sparc: Add FSR_QNE to tb_flags

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Tested-by: Carl Hauser <chauser@pullman.com>

target/sparc: Populate sparc32 FQ when raising fp exception

Implement a single instruction floating point queue,
populated while delivering an fp exception.

Signed-off-by: Carl Hauser <chauser@pullman.com>
[rth: Split from a larger patch]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Tested-by: Carl Hauser <chauser@pullman.com>

target/sparc: Add FQ and FSR.QNE

Add support for, and migrate, a single-entry fp
instruction queue for sparc32.

Signed-off-by: Carl Hauser <chauser@pullman.com>
[rth: Split from a larger patch;
      adjust representation with union;
      add migration state]
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Tested-by: Carl Hauser <chauser@pullman.com>

Merge tag 'pull-request-2024-09-11' of https://gitlab.com/thuth/qemu into staging

* Split --enable-sanitizers to --enable-{asan, ubsan}
* Build MSYS2 job using multiple CPUs
* Fix "make distclean" wrt contrib/plugins/
* Convert more Avocado tests to plain standalone functional tests
* Fix bug that breaks "make check-functional" when tesseract is missing
* Use builtin hashlib of Python in the functional tests
* Update the FreeBSD CI jobs to 14.1

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmbhY4YRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbU/aw/9HXl9H8BUDn8lnoEmxuuQSk8F19n/l5pt
# en3L8pMBt4dGFe/9KaGes2GFfid+cp2zlx+qQhA4HW35ntMJorF/qinOH/JGDtoM
# 3O6RGZrQPn60zD9P2EbFVCrVYysVYCEu0U3Uglj6tf33bE0L7SJsQxqcbIciyIj5
# aq3Te0yMM2lqzCdMqNpWHGn3VMZRvbRaGBPDU4RLP8V2Bpz1iiRE+6HCH9Kg7HzS
# OmleeXtvcyInG+54onjfTcn4/XA27pl1UU04KFv5PrRPB3M2FspHn7oOT2yyQ+ls
# 79mqIcd8PvycCT+3ch9p8KhVtbVBgZGmeemALLvk5FxysaWnl4KtSqmQNdqSvvpV
# waDDKlLaSnjEHDUse3bCJX0m4d7/vTBY5fOYxqZ4z5dl63csDtgPY4/VF4XR08sP
# tR1mW+2qEH9eygsxuKcBjx/j7Etpy+jL9pX2ii1V3ElhjjYuEnpEiURa+TaqPjpZ
# jmPtBEszzUdPbrD707tDkW3/ezT7VAnASQeYneJXB/JQG6K6Z//05iX6oCzCbRm3
# ceW/fem3UaeGYpzbMdoZToTuNlXEyS7NDcr39xJjH4LyRTPJAX4zeqUEdzces9g/
# u4Dw6rJ0Yhj4rscKxRvGl3/BH6CTI+8IAsbju2B/CnVLTqaABB0q/MDB90aB44xX
# bAVsl4P03Uk=
# =5TR0
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 11 Sep 2024 10:31:50 BST
# 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

* tag 'pull-request-2024-09-11' of https://gitlab.com/thuth/qemu: (24 commits)
  Update FreeBSD CI jobs FreeBSD 14.1
  tests/functional/qemu_test: Use Python hashlib instead of external programs
  tests/functional: Fix bad usage of has_cmd
  tests/functional: Convert the multiprocess avocado test into a standalone test
  tests/functional: Convert the or1k-sim Avocado test
  tests/functional: Convert the m68k MCF5208EVB Avocado test
  tests/functional: Convert the Alpha Clipper Avocado test
  tests/functional: Convert Aarch64 Raspi4 avocado tests
  tests/functional: Convert Aarch64 Raspi3 avocado tests
  tests/functional: Convert ARM Raspi2 avocado tests
  tests/functional: Convert mips32eb 4Kc Malta avocado tests
  tests/functional: Convert nanomips Malta avocado tests
  tests/functional: Convert mips32el Malta YAMON avocado test
  tests/functional: Convert mips64el 5KEc Malta avocado tests
  tests/functional: Convert mips64el I6400 Malta avocado tests
  tests/functional: Convert mips64el Fuloong2e avocado test (2/2)
  tests/functional: Convert the m68k Q800 Avocado test into a functional test
  tests/functional: Add the LinuxKernelTest for testing the Linux boot process
  MAINTAINERS: Remove myself from the Meson section
  MAINTAINERS: Remove myself as reviewer
  ...

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

hw/acpi/ich9: Add periodic and swsmi timer

This patch implements the periodic and the swsmi ICH9 chipset timers. They are
especially useful when prototyping UEFI firmware (e.g. with EDK2's OVMF)
using QEMU.

For backwards compatibility, the compat properties "x-smi-swsmi-timer",
and "x-smi-periodic-timer" are introduced.

Additionally, writes to the SMI_STS register are enabled for the
corresponding two bits using a write mask to make future work easier.

Signed-off-by: Dominic Prinz <git@dprinz.de>
Message-Id: <1d90ea69e01ab71a0f2ced116801dc78e04f4448.1725991505.git.git@dprinz.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

virtio-mem: don't warn about THP sizes on a kernel without THP support

If the config directory in sysfs does not exist at all, we are dealing
with a system that does not support THPs. Simply use 1 MiB block size
then, instead of warning "Could not detect THP size, falling back to
..." and falling back to the default THP size.

Cc: "Michael S. Tsirkin" <mst@redhat.com>
Cc: Gavin Shan <gshan@redhat.com>
Cc: Juraj Marcin <jmarcin@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Message-Id: <20240910163433.2100295-1-david@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/audio/virtio-sound: fix heap buffer overflow

Currently, the guest may write to the device configuration space,
whereas the virtio sound device specification in chapter 5.14.4
clearly states that the fields in the device configuration space
are driver-read-only.

Remove the set_config function from the virtio_snd class.

This also prevents a heap buffer overflow. See QEMU issue #2296.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2296
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20240901130112.8242-1-vr_qemu@t-online.de>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/cxl: fix physical address field in get scan media results output

When using the mailbox command get scan media results, the scan media
restart physical address field in the ouput palyload is not 64-byte
aligned.

This patch removed the error source of the restart physical address.

The Scan Media Restart Physical Address is the location from which the
host should restart the Scan Media operation. [5:0] bits are reserved.
Refer to CXL spec r3.1 Table 8-146

Fixes: 89b5cfcc31e6 ("hw/cxl: Add get scan media results cmd support")
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Link: https://lore.kernel.org/linux-cxl/20240819154206.16456-1-engguopeng@buaa.edu.cn/
Signed-off-by: peng guo <engguopeng@buaa.edu.cn>
Message-Id: <20240825102212.3871-1-engguopeng@buaa.edu.cn>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

virtio-pci: Add lookup subregion of VirtIOPCIRegion MR

Now virtio_address_space_lookup only lookup common/isr/device/notify
MR and exclude their subregions.

When VHOST_USER_PROTOCOL_F_HOST_NOTIFIER enable, the notify MR has
host-notifier subregions and we need use host-notifier MR to
notify the hardware accelerator directly instead of eventfd notify.

Further more, maybe common/isr/device MR also has subregions in
the future, so need memory_region_find for each MR incluing
their subregions.

Add lookup subregion of VirtIOPCIRegion MR instead of only lookup container MR.

Fixes: a93c8d8 ("virtio-pci: Replace modern_as with direct access to modern_bar")
Co-developed-by: Zuo Boqun <zuoboqun@baidu.com>
Signed-off-by: Gao Shiyuan <gaoshiyuan@baidu.com>
Signed-off-by: Zuo Boqun <zuoboqun@baidu.com>
Message-Id: <20240903120304.97833-1-gaoshiyuan@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

vhost_net: configure all host notifiers in a single MR transaction

This allows the vhost_net device which has multiple virtqueues to batch
the setup of all its host notifiers. This significantly reduces the
vhost_net device starting and stoping time, e.g. the time spend
on enabling notifiers reduce from 630ms to 75ms and the time spend on
disabling notifiers reduce from 441ms to 45ms for a VM with 192 vCPUs
and 15 vhost-user-net devices (64vq per device) in our case.

Signed-off-by: zuoboqun <zuoboqun@baidu.com>
Message-Id: <20240816070835.8309-1-zuoboqun@baidu.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests/acpi: pc: update golden masters for DSDT

Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
Message-Id: <20240814115736.1580337-4-ribalda@chromium.org>
Acked-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/i386/acpi-build: Return a pre-computed _PRT table

When qemu runs without kvm acceleration the ACPI executions take a great
amount of time. If they take more than the default time (30sec), the
ACPI calls fail and the system might not behave correctly.

Now the _PRT table is computed on the fly. We can drastically reduce the
execution of the _PRT method if we return a pre-computed table.

Without this patch:
[   51.343484] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
[   51.527032] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
[   51.530049] virtio-pci 0000:00:02.0: can't derive routing for PCI INT A
[   51.530797] virtio-pci 0000:00:02.0: PCI INT A: no GSI
[   81.922901] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
[   82.103534] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
[   82.106088] virtio-pci 0000:00:04.0: can't derive routing for PCI INT A
[   82.106761] virtio-pci 0000:00:04.0: PCI INT A: no GSI
[  112.192568] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
[  112.486687] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
[  112.489554] virtio-pci 0000:00:05.0: can't derive routing for PCI INT A
[  112.490027] virtio-pci 0000:00:05.0: PCI INT A: no GSI
[  142.559448] ACPI Error: Aborting method \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/psparse-529)
[  142.718596] ACPI Error: Method execution failed \_SB.PCI0._PRT due to previous error (AE_AML_LOOP_TIMEOUT) (20230628/uteval-68)
[  142.722889] virtio-pci 0000:00:06.0: can't derive routing for PCI INT A
[  142.724578] virtio-pci 0000:00:06.0: PCI INT A: no GSI

With this patch:
[   22.938076] ACPI: \_SB_.LNKB: Enabled at IRQ 10
[   24.214002] ACPI: \_SB_.LNKD: Enabled at IRQ 11
[   25.465170] ACPI: \_SB_.LNKA: Enabled at IRQ 10
[   27.944920] ACPI: \_SB_.LNKC: Enabled at IRQ 11

ACPI disassembly:
        Scope (PCI0)
        {
            Method (_PRT, 0, NotSerialized)  // _PRT: PCI Routing Table
            {
                Return (Package (0x80)
                {
                    Package (0x04)
                    {
                        0xFFFF,
                        Zero,
                        LNKD,
                        Zero
                    },

                    Package (0x04)
                    {
                        0xFFFF,
                        One,
                        LNKA,
                        Zero
                    },

                    Package (0x04)
                    {
                        0xFFFF,
                        0x02,
                        LNKB,
                        Zero
                    },

                    Package (0x04)
                    {
                        0xFFFF,
                        0x03,
                        LNKC,
                        Zero
                    },

                    Package (0x04)
                    {
                        0x0001FFFF,
                        Zero,
                        LNKS,
                        Zero
                    },
Context: https://lore.kernel.org/virtualization/20240417145544.38d7b482@imammedo.users.ipa.redhat.com/T/#t

Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240814115736.1580337-3-ribalda@chromium.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

tests/acpi: pc: allow DSDT acpi table changes

Signed-off-by: Ricardo Ribalda <ribalda@chromium.org>
Message-Id: <20240814115736.1580337-2-ribalda@chromium.org>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

intel_iommu: Make PASID-cache and PIOTLB type invalid in legacy mode

In vtd_process_inv_desc(), VTD_INV_DESC_PC and VTD_INV_DESC_PIOTLB are
bypassed without scalable mode check. These two types are not valid
in legacy mode and we should report error.

Fixes: 4a4f219e8a10 ("intel_iommu: add scalable-mode option to make scalable mode work")
Suggested-by: Yi Liu <yi.l.liu@intel.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Clément Mathieu--Drif<clement.mathieu--drif@eviden.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Message-Id: <20240814071321.2621384-3-zhenzhong.duan@intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

intel_iommu: Fix invalidation descriptor type field

According to spec, invalidation descriptor type is 7bits which is
concatenation of bits[11:9] and bits[3:0] of invalidation descriptor.

Currently we only pick bits[3:0] as the invalidation type and treat
bits[11:9] as reserved zero. This is not a problem for now as bits[11:9]
is zero for all current invalidation types. But it will break if newer
type occupies bits[11:9].

Fix it by taking bits[11:9] into type and make reserved bits check accurate.

Suggested-by: Clément Mathieu--Drif<clement.mathieu--drif@eviden.com>
Signed-off-by: Zhenzhong Duan <zhenzhong.duan@intel.com>
Reviewed-by: Yi Liu <yi.l.liu@intel.com>
Reviewed-by: Clément Mathieu--Drif<clement.mathieu--drif@eviden.com>
Message-Id: <20240814071321.2621384-2-zhenzhong.duan@intel.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

virtio: rename virtio_split_packed_update_used_idx

virtio_split_packed_update_used_idx should be
virtio_queue_split_update_used_idx like
virtio_split_packed_update_used_idx.

Signed-off-by: Wenyu Huang <huangwenyuu@outlook.com>
Message-Id: <TYBP286MB036536B9015994AA5F3E4495ACB22@TYBP286MB0365.JPNP286.PROD.OUTLOOK.COM>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/pci/pci-hmp-cmds: Avoid displaying bogus size in 'info pci'

When BAR aren't mapped, we get:

  (qemu) info pci
    Bus  0, device   0, function 0:
      Host bridge: PCI device dead:beef
        ...
        BAR4: 32 bit memory at 0xffffffffffffffff [0x00000ffe].
        BAR5: I/O at 0xffffffffffffffff [0x0ffe].

Check the BAR is mapped comparing its address to PCI_BAR_UNMAPPED
which is what the PCI layer uses for unmapped BARs.
See pci_bar_address and pci_update_mappings implementations and
in "hw/pci/pci.h":

  typedef struct PCIIORegion {
      pcibus_t addr; /* current PCI mapping address. -1 means not mapped */
  #define PCI_BAR_UNMAPPED (~(pcibus_t)0)
      ...

This improves the logging, not displaying bogus sizes:

  (qemu) info pci
    Bus  0, device   0, function 0:
      Host bridge: PCI device dead:beef
        ...
      BAR4: 32 bit memory (not mapped)
      BAR5: I/O (not mapped)

Remove trailing dot which is not used in other commands format.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20240801131449.51328-1-philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

Merge tag 'pull-testing-gdbstub-oct-100924-1' of https://gitlab.com/stsquad/qemu into staging

testing and gdbstub updates:

  - remove docker-armel-cross
  - update i686 and mipsel images to bookworm
  - use docker-all-test-cross for mips64le tests
  - fix duplicated line in docs
  - update gitlab-runner ansible script
  - support MTE in gdbstub for system mode

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEZoWumedRZ7yvyN81+9DbCVqeKkQFAmbgye8ACgkQ+9DbCVqe
# KkTesQf/WSTYAelzJWlEo0EPg5agokephfza4vdmweDujOT8MYPF9qxfsxoiTVA8
# GTtTOod9iqmY/4/EPKIqUtZH38oaX5h9on2FhSssOMy+N4lUADJ+CcHHMSj4BuUt
# jTXDSa9e5aj0m/yqg2PjF8U12Sygf7dKJturGLOWoWR5qa3xpQ2a6c3CkfxO3RQK
# yTBfIZk47iOrVvEX8chsRzpkpiXY6/S5hkZZwcqbXcUMKH2s0po9Yg031vE3yN+g
# kxJ7/mFNl49E/fqYdRahhyBDORlltCglCHsacxxa/4a216wOsNKyV3QLCJMjq8yO
# 3/SPu0p+UouSFcASwTUt5XIo0G0TcA==
# =7W1s
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 10 Sep 2024 23:36:31 BST
# gpg:                using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>" [full]
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8  DF35 FBD0 DB09 5A9E 2A44

* tag 'pull-testing-gdbstub-oct-100924-1' of https://gitlab.com/stsquad/qemu:
  tests/tcg/aarch64: Extend MTE gdbstub tests to system mode
  tests/tcg/aarch64: Improve linker script organization
  tests/guest-debug: Support passing arguments to the GDB test script
  gdbstub: Add support for MTE in system mode
  gdbstub: Use specific MMU index when probing MTE addresses
  scripts/ci: update the gitlab-runner playbook
  docs/devel: fix duplicate line
  tests/docker: use debian-all-test-cross for mips64el tests
  tests/docker: update debian i686 and mipsel images to bookworm
  tests/docker: remove debian-armel-cross

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

pci: don't skip function 0 occupancy verification for devfn auto assign

When the devfn is already assigned in the command line, the
do_pci_register_device() may verify if the function 0 is already occupied.

However, when devfn < 0, the verification is skipped because it is part of
the last "else if".

For instance, suppose there is already a device at addr=00.00 of a port.

-device pcie-root-port,bus=pcie.0,chassis=115,id=port01,addr=0e.00 \
-device virtio-net-pci,bus=port01,id=vnet01,addr=00.00 \

When 'addr' is specified for the 2nd device, the hotplug is denied.

(qemu) device_add virtio-net-pci,bus=port01,id=vnet02,addr=01.00
Error: PCI: slot 0 function 0 already occupied by virtio-net-pci, new func virtio-net-pci cannot be exposed to guest.

When 'addr' is automatically assigned, the hotplug is not denied. This is
because the verification is skipped.

(qemu) device_add virtio-net-pci,bus=port01,id=vnet02
warning: PCI: slot 1 is not valid for virtio-net-pci, parent device only allows plugging into slot 0.

Fix the issue by moving the verification into an independent 'if'
statement.

Fixes: 3f1e1478db2d ("enable multi-function hot-add")
Reported-by: Aswin Unnikrishnan <aswin.u.unnikrishnan@oracle.com>
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Message-Id: <20240708041056.54504-1-dongli.zhang@oracle.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

hw/isa/vt82c686.c: Embed i8259 irq in device state instead of allocating

To avoid a warning about unfreed qemu_irq embed the i8259 irq in the
device state instead of allocating it.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <b70b9e72063b4dd4005bf4bc040b84f2bb617bf4.1719690591.git.balaton@eik.bme.hu>
Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>

hw: Move declaration of IRQState to header and add init function

To allow embedding a qemu_irq in a struct move its definition to the
header and add a function to init it in place without allocating it.

Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>
Message-Id: <e3ffd0f6ef8845d0f7247c9b6ff33f7ee8b432cf.1719690591.git.balaton@eik.bme.hu>
Signed-off-by: BALATON Zoltan <balaton@eik.bme.hu>

Update FreeBSD CI jobs FreeBSD 14.1

The current FreeBSD CI jobs are failing installation since the
"opencv" package is now missing there. Updating to 14.1 fixes
the issue.

Message-Id: <20240911090149.286257-1-thuth@redhat.com>
Reviewed-by: Li-Wen Hsu <lwhsu@FreeBSD.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional/qemu_test: Use Python hashlib instead of external programs

Some systems (like OpenBSD) do not have the sha256sum or sha512sum programs
installed by default, or use different names for those. Use the Python
hashlib instead so we don't have to rely on the external programs.

Reported-by: Peter Maydell <peter.maydell@linaro.org>
Message-ID: <20240910201742.239559-1-thuth@redhat.com>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Fix bad usage of has_cmd

has_cmd returns a tuple, not a boolean value. This fixes a crash when
e.g. "tesseract" is not available in the test_m68k_nextcube test.

Reported-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20240910075820.51346-1-thuth@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the multiprocess avocado test into a standalone test

This test handles both, aarch64 and x86_64, with the same test code
(apart from some initial setup), so don't split this file by target
but add a check for self.arch in the main test function.

Message-ID: <20240903051333.102494-1-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the or1k-sim Avocado test

We've got to do_test_advcal_2018() here now that the test resides
in a separate file. Also switch back to the original URL (since
the site did not vanish as originally expected) and update the
hashsum to use SHA256.

Message-ID: <20240906180549.792832-15-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the m68k MCF5208EVB Avocado test

We've got to do_test_advcal_2018() here now that the test resides
in a separate file. Also switch back to the original URL (since
the site did not vanish as originally expected) and update the
hashsum to use SHA256.

Message-ID: <20240906180549.792832-14-thuth@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the Alpha Clipper Avocado test

Straight forward conversion, just switch to SHA256 hashsum now.

Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240906180549.792832-13-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert Aarch64 Raspi4 avocado tests

Straight forward conversion. Update the SHA1 hashes to SHA256
hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240826221058.75126-5-philmd@linaro.org>
[thuth: Use LinuxKernelTest class]
Message-ID: <20240906180549.792832-12-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert Aarch64 Raspi3 avocado tests

Straight forward conversion. Update the SHA1 hashes to SHA256
hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240826221058.75126-4-philmd@linaro.org>
[thuth: Use the LinuxKernelTest class]
Message-ID: <20240906180549.792832-11-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert ARM Raspi2 avocado tests

Straight forward conversion. Update the SHA1 hashes to SHA256
hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240826221058.75126-3-philmd@linaro.org>
[thuth: Use LinuxKernelTest class]
Message-ID: <20240906180549.792832-10-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert mips32eb 4Kc Malta avocado tests

Straight forward conversion. Update the SHA1 hashes to
SHA256 hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240827094905.80648-8-philmd@linaro.org>
[thuth: Use LinuxKernelTest class]
Message-ID: <20240906180549.792832-9-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert nanomips Malta avocado tests

Straight forward conversion. Update the SHA1 hashes to
SHA256 hashes since SHA1 should not be used anymore nowadays.

  $ QEMU_TEST_ALLOW_UNTRUSTED_CODE=1 \
        make check-functional-mipsel
  ...
  ▶ 4/4 test_mipsel_malta.MaltaMachineConsole.test_mips_malta32el_nanomips_16k_up   OK
  ▶ 4/4 test_mipsel_malta.MaltaMachineConsole.test_mips_malta32el_nanomips_4k       OK
  ▶ 4/4 test_mipsel_malta.MaltaMachineConsole.test_mips_malta32el_nanomips_64k_dbg  OK
  ▶ 4/4 test_mipsel_malta.MaltaMachineYAMON.test_mipsel_malta_yamon                 OK
  4/4 qemu:func-thorough+func-mipsel-thorough+thorough / func-mipsel-mipsel_malta   OK   9.95s   4 subtests passed

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240827094905.80648-7-philmd@linaro.org>
[thuth: Use LinuxKernelTest]
Message-ID: <20240906180549.792832-8-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert mips32el Malta YAMON avocado test

Straight forward conversion using the Python standard zipfile
module instead of avocado.utils package. Update the SHA1 hashes
to SHA256 hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240827094905.80648-6-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240906180549.792832-7-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert mips64el 5KEc Malta avocado tests

Straight forward conversion. Update the SHA1 hashes to
SHA256 hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240827094905.80648-5-philmd@linaro.org>
Message-ID: <20240906180549.792832-6-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert mips64el I6400 Malta avocado tests

Straight forward conversion. Update the SHA1 hashes to SHA256
hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240827094905.80648-4-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240906180549.792832-5-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert mips64el Fuloong2e avocado test (2/2)

Straight forward conversion. Update the SHA1 hashes to
SHA256 hashes since SHA1 should not be used anymore nowadays.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240827094905.80648-3-philmd@linaro.org>
[thuth: Use LinuxKernelTest class for this test]
Message-ID: <20240906180549.792832-4-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the m68k Q800 Avocado test into a functional test

Just had to update the asset checksum to use SHA256 instead of SHA1,
but apart from that it is a pretty much straightforward conversion.

Message-ID: <20240906180549.792832-3-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Add the LinuxKernelTest for testing the Linux boot process

Copy the LinuxKernelTest from tests/acceptance/boot_linux_console.py
to be able to convert the related tests to the functional test framework
in the following patches.

Message-ID: <20240906180549.792832-2-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

MAINTAINERS: Remove myself from the Meson section

The Meson section has plenty of reviewers, and in order to reduce the
load on my Inbox a little bit, I'm stepping back as a reviewer here.

Message-ID: <20240906155900.742157-1-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

MAINTAINERS: Remove myself as reviewer

Finally taking this off my to-do list. It’s been a privilege to be part
of this project, but I am no longer actively involved in reviewing
Python code here, so I believe it's best to update the list to reflect
the current maintainers.

Please, feel free to reach out if any questions arise.

Signed-off-by: Beraldo Leal <bleal@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240819150035.2180786-1-bleal@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

contrib/plugins/Makefile: Add a 'distclean' target

Running "make distclean" in the build tree currently fails since this
tries to run the "distclean" target in the contrib/plugins/ folder, too,
but the Makefile there is missing this target. Thus add 'distclean' there
to fix this issue.

And to avoid regressions with "make distclean", add this command to one
of the build jobs, too.

Message-ID: <20240902154749.73876-1-thuth@redhat.com>
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Build MSYS2 job using multiple CPUs

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240819112141.13786-1-philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

meson: Move -fsanitize=undefined into normal configuraton

With 8e466dd09246 and 23ef50ae2d0c, we disable function pointer
sanitization in CI because the qemu code base does not support it.
We must disable this for normal usage of --enable-ubsan as well,
so move it there.

Append options rather than prepend, since all of this requires
proper ordering of options.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240813095216.306555-3-richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

meson: Split --enable-sanitizers to --enable-{asan, ubsan}

We do not always want both address and undefined behavior
sanitizers running at the same time.

For the gitlab custom-runners, drop to only --enable-ubsan.
These jobs are not run by default, but as will be obvious in the
next patch, we don't run ASan on x86 either, and it seems wrong
to hold aarch64 and s390x to a different standard.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240813095216.306555-2-richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/tcg/aarch64: Extend MTE gdbstub tests to system mode

Extend MTE gdbstub tests to also run in system mode (share tests between
user mode and system mode). The tests will only run if a version of GDB
that supports MTE on baremetal is available in the test environment and
if available compiler supports the 'memtag' flag
(-march=armv8.5-a+memtag).

For the tests running in system mode, a page that supports MTE ops. is
necessary. Therefore, an MTE-enabled page is made available (mapped) in
the third 2 MB chunk of the second 1 GB space in the flat mapping set in
boot.S. A new binary, mte.S, is also introduced for the tests. It links
against boot.S and is executed by QEMU in system mode.

Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Message-Id: <20240906143316.657436-6-gustavo.romero@linaro.org>
[AJB: fix stray _]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-11-alex.bennee@linaro.org>

tests/tcg/aarch64: Improve linker script organization

Improve kernel.ld linker script organization by using MEMORY command.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Message-Id: <20240906143316.657436-5-gustavo.romero@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-10-alex.bennee@linaro.org>

tests/guest-debug: Support passing arguments to the GDB test script

This commit adds support for passing arguments to the GDB test scripts
so it's possible to parse the args in an "argparse way" in the test
scripts launched by the runner. The arguments should be preceded by --
when passed to the runner. For example, passing "--help" arg to the
GDB_TEST_SCRIPT:

run-test.py [...] --test <GDB_TEST_SCRIPT> -- --help

The test script should not use the argparse module directly but import
arg_parser from test_gdbstub module. arg_parser then can be used just
like the argparse.ArgumentParser class:

from test_gdbstub import arg_parser

p = arg_parser(prog="test-mytest.py", description="My test.")
p.add_argument("--vowel", help="Select vowel",
               required=True, choices=['a','e','i','o','u'])
[...]

The arg_parser allows a smooth and informative exit if, for instance,
the caller of the runner script passes an invalid argument or misses a
required argument by the test script.

Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240906143316.657436-4-gustavo.romero@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-9-alex.bennee@linaro.org>

gdbstub: Add support for MTE in system mode

This commit makes handle_q_memtag, handle_q_isaddresstagged, and
handle_Q_memtag stubs build for system mode, allowing all GDB
'memory-tag' subcommands to work with QEMU gdbstub on aarch64 system
mode.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/620
Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240906143316.657436-3-gustavo.romero@linaro.org>
[AJB: add #ifdef CONFIG_TCG guards]
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-8-alex.bennee@linaro.org>

gdbstub: Use specific MMU index when probing MTE addresses

Use cpu_mmu_index() to determine the specific translation regime (MMU
index) before probing addresses using allocation_tag_mem_probe().

Currently, the MMU index is hardcoded to 0 and only works for user mode.
By obtaining the specific MMU index according to the translation regime,
future use of the stubs relying on allocation_tag_mem_probe in other
regimes will be possible, like in EL1.

This commit also changes the ptr_size value passed to
allocation_tag_mem_probe() from 8 to 1. The ptr_size parameter actually
represents the number of bytes in the memory access (which can be as
small as 1 byte), rather than the number of bits used in the address
space pointed to by ptr.

Signed-off-by: Gustavo Romero <gustavo.romero@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240906143316.657436-2-gustavo.romero@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-7-alex.bennee@linaro.org>

scripts/ci: update the gitlab-runner playbook

The upstream install instructions:

  https://docs.gitlab.com/runner/install/linux-repository.html

Now refer to repositories and a setup script. Modernise the playbook
to use the preferred delivery method.

Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-6-alex.bennee@linaro.org>

docs/devel: fix duplicate line

I guess the same change came in via two patch series. Remove the
repetition.

Fixes: 2a851fca9f (docs/devel: remind developers to run CI container pipeline when updating images)
Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-5-alex.bennee@linaro.org>

tests/docker: use debian-all-test-cross for mips64el tests

While the mips64el cross compilation environment is busted in Debian
we can use the debian-all-test-cross image for building TCG tests like
we do in the CI.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-4-alex.bennee@linaro.org>

tests/docker: update debian i686 and mipsel images to bookworm

Whatever issues there were which stopped these being updates when the
rest were have now been resolved. However mips64el continues to be
broken so don't update it here.

Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-3-alex.bennee@linaro.org>

tests/docker: remove debian-armel-cross

As debian-11 transitions to LTS we are starting to have problems
building the image. While we could update to a later Debian building a
32 bit QEMU without modern floating point is niche host amongst the
few remaining 32 bit hosts we regularly build for. For now we still
have armhf-debian-cross-container which is currently built from the
more recent debian-12.

Reviewed-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240910173900.4154726-2-alex.bennee@linaro.org>

virtio: Always reset vhost devices

Requiring `vhost_started` to be true for resetting vhost devices in
`virtio_reset()` seems like the wrong condition: Most importantly, the
preceding `virtio_set_status(vdev, 0)` call will (for vhost devices) end
up in `vhost_dev_stop()` (through vhost devices' `.set_status`
implementations), setting `vdev->vhost_started = false`.  Therefore, the
gated `vhost_reset_device()` call is unreachable.

`vhost_started` is not documented, so it is hard to say what exactly it
is supposed to mean, but judging from the fact that `vhost_dev_start()`
sets it and `vhost_dev_stop()` clears it, it seems like it indicates
whether there is a vhost back-end, and whether that back-end is
currently running and processing virtio requests.

Making a reset conditional on whether the vhost back-end is processing
virtio requests seems wrong; in fact, it is probably better to reset it
only when it is not currently processing requests, which is exactly the
current order of operations in `virtio_reset()`: First, the back-end is
stopped through `virtio_set_status(vdev, 0)`, then we want to send a
reset.

Therefore, we should drop the `vhost_started` condition, but in its
stead we then have to verify that we can indeed send a reset to this
vhost device, by not just checking `k->get_vhost != NULL` (introduced by
commit 95e1019a4a9), but also that the vhost back-end is connected
(`hdev = k->get_vhost(); hdev != NULL && hdev->vhost_ops != NULL`).

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
Message-Id: <20240723163941.48775-3-hreitz@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>

virtio: Allow .get_vhost() without vhost_started

Historically, .get_vhost() was probably only called when
vdev->vhost_started is true.  However, we now decidedly want to call it
also when vhost_started is false, specifically so we can issue a reset
to the vhost back-end while device operation is stopped.

Some .get_vhost() implementations dereference some pointers (or return
offsets from them) that are probably guaranteed to be non-NULL when
vhost_started is true, but not necessarily otherwise.  This patch makes
all such implementations check all such pointers, returning NULL if any
is NULL.

Signed-off-by: Hanna Czenczek <hreitz@redhat.com>
Message-Id: <20240723163941.48775-2-hreitz@redhat.com>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-qapi-2024-09-10' of https://repo.or.cz/qemu/armbru into staging

QAPI patches patches for 2024-09-10

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEENUvIs9frKmtoZ05fOHC0AOuRhlMFAmbgS0ASHGFybWJydUBy
# ZWRoYXQuY29tAAoJEDhwtADrkYZTm+sP/1kEuGdbjjVLQC7U+gCClhALpYUtIMHQ
# qNZWmnxFyW9Jx6Lk8+FZBRJegNlmA9AJ2ZzD/2I/WoNtcrNiE53rIRG71Lcy8I6U
# CRjekLPn0jhFuYhmRfwz6ilcLkyMJ15Yi7iG6ssFsO1joBtU5Q1qpzQmFUl9UbV9
# 9iBe7mE99VzNwi3hMELEM9YuSgAs8st4Itnn8SiHkJ3s6hlmF8N4HwNKfQrseGfb
# FuJNN4p8Gcu0aF/dkGE8ADJBvtgzkFnavXt0GQPC7SjGOF7rUXxnUQlszNZM0OcK
# UAfWU06sIyiOholQQt8H8jawT+rGCLomfgPUBVq5K8WByd21IeQeS0upSPeTjxS+
# d/C8wzpcbhdNOECwC/wDtgZHvCC95cHNrxuC4+4/Q1KIVVr+1cWqe8hQzFvm3TIg
# dduEFTQDIxhpE0GO1ZWNf90upzBYBWwIsh4bqsMZS7SpaYIZ6QV45yViZ1w2WfbH
# m3/F34Z3yqgFuZQrZv4OPo7xHK0+y7uZ60RrhSJtE0X080syRJKBbGmNBRidoVyu
# JOONWH44I/XN5enZV8StJnqJr9MCV0DBZUCi9ZhP/kAhBjLc5cQ6NByEa9/rebYX
# 1bLTiA1JDLzDVIan+A8dz0riWmHBvTyBwhLnEXvXb9lcB3ozgHTb7axE5RnZSYLI
# YQgBOBMFuQKM
# =YO/Q
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 10 Sep 2024 14:36:00 BST
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [full]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [full]
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* tag 'pull-qapi-2024-09-10' of https://repo.or.cz/qemu/armbru:
  qapi/vfio: Rename VfioMigrationState to Qapi*, and drop prefix
  qapi/cryptodev: Rename QCryptodevBackendAlgType to *Algo, and drop prefix
  qapi/cryptodev: Drop unwanted 'prefix'
  qapi/crypto: Rename QCryptoAFAlg to QCryptoAFAlgo
  qapi/crypto: Rename QCryptoRSAPaddingAlgorithm to *Algo, and drop prefix
  qapi/crypto: Rename QCryptoAkCipherAlgorithm to *Algo, and drop prefix
  qapi/crypto: Rename QCryptoIVGenAlgorithm to *Algo, and drop prefix
  qapi/crypto: Rename QCryptoCipherAlgorithm to *Algo, and drop prefix
  qapi/crypto: Rename QCryptoHashAlgorithm to *Algo, and drop prefix
  qapi/crypto: Drop unwanted 'prefix'
  qapi/machine: Rename CpuS390* to S390Cpu*, and drop 'prefix'
  qapi/ui: Drop temporary 'prefix'
  qapi/machine: Drop temporary 'prefix'
  qapi/ebpf: Drop temporary 'prefix'
  qapi/crypto: Drop temporary 'prefix'
  qapi/common: Drop temporary 'prefix'
  qapi/block-core: Drop temporary 'prefix'
  tests/qapi-schema: Drop temporary 'prefix'
  qapi: Smarter camel_to_upper() to reduce need for 'prefix'

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

qapi/vfio: Rename VfioMigrationState to Qapi*, and drop prefix

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

VfioMigrationState has a 'prefix' that overrides the generated
enumeration constants' prefix to QAPI_VFIO_MIGRATION_STATE.

We could simply drop 'prefix', but then the enumeration constants
would look as if they came from kernel header linux/vfio.h.

Rename the type to QapiVfioMigrationState instead, so that 'prefix' is
not needed.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240904111836.3273842-20-armbru@redhat.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>

qapi/cryptodev: Rename QCryptodevBackendAlgType to *Algo, and drop prefix

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptodevBackendAlgType has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTODEV_BACKEND_ALG.

We could simply drop 'prefix', but I think the abbreviation "alg" is
less than clear.

Additionally rename the type to QCryptodevBackendAlgoType.  The prefix
becomes QCRYPTODEV_BACKEND_ALGO_TYPE.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-19-armbru@redhat.com>

qapi/cryptodev: Drop unwanted 'prefix'

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptodevBackendServiceType has a 'prefix' that overrides the
generated enumeration constants' prefix to QCRYPTODEV_BACKEND_SERVICE.

Drop it.  The prefix becomes QCRYPTODEV_BACKEND_SERVICE_TYPE.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-18-armbru@redhat.com>

qapi/crypto: Rename QCryptoAFAlg to QCryptoAFAlgo

For consistency with other types names *Algo.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-17-armbru@redhat.com>

qapi/crypto: Rename QCryptoRSAPaddingAlgorithm to *Algo, and drop prefix

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptoRSAPaddingAlgorithm has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTO_RSA_PADDING_ALG.

We could simply drop 'prefix', but then the prefix becomes
QCRYPTO_RSA_PADDING_ALGORITHM, which is rather long.

We could additionally rename the type to QCryptoRSAPaddingAlg, but I
think the abbreviation "alg" is less than clear.

Rename the type to QCryptoRSAPaddingAlgo instead.  The prefix becomes
QCRYPTO_RSA_PADDING_ALGO.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-16-armbru@redhat.com>

qapi/crypto: Rename QCryptoAkCipherAlgorithm to *Algo, and drop prefix

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptoAkCipherAlgorithm has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTO_AKCIPHER_ALG.

We could simply drop 'prefix', but then the prefix becomes
QCRYPTO_AK_CIPHER_ALGORITHM, which is rather long.

We could additionally rename the type to QCryptoAkCipherAlg, but I
think the abbreviation "alg" is less than clear.

Rename the type to QCryptoAkCipherAlgo instead.  The prefix becomes
QCRYPTO_AK_CIPHER_ALGO.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-15-armbru@redhat.com>

qapi/crypto: Rename QCryptoIVGenAlgorithm to *Algo, and drop prefix

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptoIVGenAlgorithm has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTO_IVGEN_ALG.

We could simply drop 'prefix', but then the prefix becomes
QCRYPTO_IV_GEN_ALGORITHM, which is rather long.

We could additionally rename the type to QCryptoIVGenAlg, but I think
the abbreviation "alg" is less than clear.

Rename the type to QCryptoIVGenAlgo instead.  The prefix becomes
QCRYPTO_IV_GEN_ALGO.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-14-armbru@redhat.com>

qapi/crypto: Rename QCryptoCipherAlgorithm to *Algo, and drop prefix

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptoCipherAlgorithm has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTO_CIPHER_ALG.

We could simply drop 'prefix', but then the prefix becomes
QCRYPTO_CIPHER_ALGORITHM, which is rather long.

We could additionally rename the type to QCryptoCipherAlg, but I think
the abbreviation "alg" is less than clear.

Rename the type to QCryptoCipherAlgo instead.  The prefix becomes
QCRYPTO_CIPHER_ALGO.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-13-armbru@redhat.com>

qapi/crypto: Rename QCryptoHashAlgorithm to *Algo, and drop prefix

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptoHashAlgorithm has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTO_HASH_ALG.

We could simply drop 'prefix', but then the prefix becomes
QCRYPTO_HASH_ALGORITHM, which is rather long.

We could additionally rename the type to QCryptoHashAlg, but I think
the abbreviation "alg" is less than clear.

Rename the type to QCryptoHashAlgo instead.  The prefix becomes to
QCRYPTO_HASH_ALGO.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-12-armbru@redhat.com>
[Conflicts with merge commit 7bbadc60b58b resolved]

qapi/crypto: Drop unwanted 'prefix'

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

QCryptoAkCipherKeyType has a 'prefix' that overrides the generated
enumeration constants' prefix to QCRYPTO_AKCIPHER_KEY_TYPE.

Drop it.  The prefix becomes QCRYPTO_AK_CIPHER_KEY_TYPE.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-11-armbru@redhat.com>

qapi/machine: Rename CpuS390* to S390Cpu*, and drop 'prefix'

QAPI's 'prefix' feature can make the connection between enumeration
type and its constants less than obvious.  It's best used with
restraint.

CpuS390Entitlement has a 'prefix' to change the generated enumeration
constants' prefix from CPU_S390_ENTITLEMENT to S390_CPU_ENTITLEMENT.
Rename the type to S390CpuEntitlement, so that 'prefix' is not needed.

Likewise change CpuS390Polarization to S390CpuPolarization, and
CpuS390State to S390CpuState.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240904111836.3273842-10-armbru@redhat.com>

qapi/ui: Drop temporary 'prefix'

Recent commit "qapi: Smarter camel_to_upper() to reduce need for
'prefix'" added a temporary 'prefix' to delay changing the generated
code.

Revert it.  This improves DisplayGLMode's generated enumeration
constant prefix from DISPLAYGL_MODE to DISPLAY_GL_MODE.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240904111836.3273842-9-armbru@redhat.com>

qapi/machine: Drop temporary 'prefix'

Recent commit "qapi: Smarter camel_to_upper() to reduce need for
'prefix'" added a temporary 'prefix' to delay changing the generated
code.

Revert it.  This improves HmatLBDataType's generated enumeration
constant prefix from HMATLB_DATA_TYPE to HMAT_LB_DATA_TYPE, and
HmatLBMemoryHierarchy's from HMATLB_MEMORY_HIERARCHY to
HMAT_LB_MEMORY_HIERARCHY.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-ID: <20240904111836.3273842-8-armbru@redhat.com>

qapi/ebpf: Drop temporary 'prefix'

Recent commit "qapi: Smarter camel_to_upper() to reduce need for
'prefix'" added a temporary 'prefix' to delay changing the generated
code.

Revert it.  This improves EbpfProgramID's generated enumeration
constant prefix from EBPF_PROGRAMID to EBPF_PROGRAM_ID.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240904111836.3273842-7-armbru@redhat.com>

qapi/crypto: Drop temporary 'prefix'

Recent commit "qapi: Smarter camel_to_upper() to reduce need for
'prefix'" added two temporary 'prefix' to delay changing the generated
code.

Revert them.  This improves QCryptoBlockFormat's generated enumeration
constant prefix from Q_CRYPTO_BLOCK_FORMAT to QCRYPTO_BLOCK_FORMAT,
and QCryptoBlockLUKSKeyslotState's from
Q_CRYPTO_BLOCKLUKS_KEYSLOT_STATE to QCRYPTO_BLOCK_LUKS_KEYSLOT_STATE.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Acked-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-6-armbru@redhat.com>

qapi/common: Drop temporary 'prefix'

Recent commit "qapi: Smarter camel_to_upper() to reduce need for
'prefix'" added a temporary 'prefix' to delay changing the generated
code.

Revert it.  This improves OffAutoPCIBAR's generated enumeration
constant prefix from OFF_AUTOPCIBAR to OFF_AUTO_PCIBAR.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Message-ID: <20240904111836.3273842-5-armbru@redhat.com>

qapi/block-core: Drop temporary 'prefix'

Recent commit "qapi: Smarter camel_to_upper() to reduce need for
'prefix'" added a temporary 'prefix' to delay changing the generated
code.

Revert it.  This improves XDbgBlockGraphNodeType's generated
enumeration constant prefix from
X_DBG_BLOCK_GRAPH_NODE_TYPE_BLOCK_BACKEND to
XDBG_BLOCK_GRAPH_NODE_TYPE_BLOCK_BACKEND.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240904111836.3273842-4-armbru@redhat.com>

tests/qapi-schema: Drop temporary 'prefix'

Recent commit "qapi: Smarter camel_to_upper() to reduce need for
'prefix'" added a temporary 'prefix' to delay changing the generated
code.

Revert it.  This changes TestUnionEnumA's generated enumeration
constant prefix from TEST_UNION_ENUMA to TEST_UNION_ENUM_A.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240904111836.3273842-3-armbru@redhat.com>

qapi: Smarter camel_to_upper() to reduce need for 'prefix'

camel_to_upper() converts its argument from camel case to upper case
with '_' between words.  Used for generated enumeration constant
prefixes.

When some of the words are spelled all caps, where exactly to insert
'_' is guesswork.  camel_to_upper()'s guesses are bad enough in places
to make people override them with a 'prefix' in the schema.

Rewrite it to guess better:

1. Insert '_' after a non-upper case character followed by an upper
   case character:

       OneTwo -> ONE_TWO
       One2Three -> ONE2_THREE

2. Insert '_' before the last upper case character followed by a
   non-upper case character:

       ACRONYMWord -> ACRONYM_Word

   Except at the beginning (as in OneTwo above), or when there is
   already one:

       AbCd -> AB_CD

This changes the default enumeration constant prefix for a number of
enums.  Generated enumeration constants change only where the default
is not overridden with 'prefix'.

The following enumerations without a 'prefix' change:

    enum                            old camel_to_upper()
                                    new camel_to_upper()
    ------------------------------------------------------------------
    DisplayGLMode                   DISPLAYGL_MODE
                                    DISPLAY_GL_MODE
    EbpfProgramID                   EBPF_PROGRAMID
                                    EBPF_PROGRAM_ID
    HmatLBDataType                  HMATLB_DATA_TYPE
                                    HMAT_LB_DATA_TYPE
    HmatLBMemoryHierarchy           HMATLB_MEMORY_HIERARCHY
                                    HMAT_LB_MEMORY_HIERARCHY
    MultiFDCompression              MULTIFD_COMPRESSION
                                    MULTI_FD_COMPRESSION
    OffAutoPCIBAR                   OFF_AUTOPCIBAR
                                    OFF_AUTO_PCIBAR
    QCryptoBlockFormat              Q_CRYPTO_BLOCK_FORMAT
                                    QCRYPTO_BLOCK_FORMAT
    QCryptoBlockLUKSKeyslotState    Q_CRYPTO_BLOCKLUKS_KEYSLOT_STATE
                                    QCRYPTO_BLOCK_LUKS_KEYSLOT_STATE
    QKeyCode                        Q_KEY_CODE
                                    QKEY_CODE
    XDbgBlockGraphNodeType          X_DBG_BLOCK_GRAPH_NODE_TYPE
                                    XDBG_BLOCK_GRAPH_NODE_TYPE
    TestUnionEnumA                  TEST_UNION_ENUMA
                                    TEST_UNION_ENUM_A

Add a 'prefix' so generated code doesn't change now.  Subsequent
commits will remove most of them again.  Two will remain:
MULTIFD_COMPRESSION, because migration code generally spells "multifd"
that way, and Q_KEY_CODE, because that one is baked into
subprojects/keycodemapdb/tools/keymap-gen.

The following enumerations with a 'prefix' change so that the prefix
is now superfluous:

    enum                            old camel_to_upper()
                                    new camel_to_upper() [equal to prefix]
    ------------------------------------------------------------------
    BlkdebugIOType                  BLKDEBUGIO_TYPE
                                    BLKDEBUG_IO_TYPE
    QCryptoTLSCredsEndpoint         Q_CRYPTOTLS_CREDS_ENDPOINT
                                    QCRYPTO_TLS_CREDS_ENDPOINT
    QCryptoSecretFormat             Q_CRYPTO_SECRET_FORMAT
                                    QCRYPTO_SECRET_FORMAT
    QCryptoCipherMode               Q_CRYPTO_CIPHER_MODE
                                    QCRYPTO_CIPHER_MODE
    QCryptodevBackendType           Q_CRYPTODEV_BACKEND_TYPE
                                    QCRYPTODEV_BACKEND_TYPE
    QType [builtin]                 Q_TYPE
                                    QTYPE

Drop these prefixes.

The following enumerations with a 'prefix' change without making the
'prefix' superfluous:

    enum                            old camel_to_upper()
                                    new camel_to_upper() [equal to prefix]
                                    prefix
    ------------------------------------------------------------------
    CpuS390Entitlement              CPUS390_ENTITLEMENT
                                    CPU_S390_ENTITLEMENT
                                    S390_CPU_ENTITLEMENT
    CpuS390Polarization             CPUS390_POLARIZATION
                                    CPU_S390_POLARIZATION
                                    S390_CPU_POLARIZATION
    CpuS390State                    CPUS390_STATE
                                    CPU_S390_STATE
                                    S390_CPU_STATE
    QAuthZListFormat                Q_AUTHZ_LIST_FORMAT
                                    QAUTH_Z_LIST_FORMAT
                                    QAUTHZ_LIST_FORMAT
    QAuthZListPolicy                Q_AUTHZ_LIST_POLICY
                                    QAUTH_Z_LIST_POLICY
                                    QAUTHZ_LIST_POLICY
    QCryptoAkCipherAlgorithm        Q_CRYPTO_AK_CIPHER_ALGORITHM
                                    QCRYPTO_AK_CIPHER_ALGORITHM
                                    QCRYPTO_AKCIPHER_ALG
    QCryptoAkCipherKeyType          Q_CRYPTO_AK_CIPHER_KEY_TYPE
                                    QCRYPTO_AK_CIPHER_KEY_TYPE
                                    QCRYPTO_AKCIPHER_KEY_TYPE
    QCryptoCipherAlgorithm          Q_CRYPTO_CIPHER_ALGORITHM
                                    QCRYPTO_CIPHER_ALGORITHM
                                    QCRYPTO_CIPHER_ALG
    QCryptoHashAlgorithm            Q_CRYPTO_HASH_ALGORITHM
                                    QCRYPTO_HASH_ALGORITHM
                                    QCRYPTO_HASH_ALG
    QCryptoIVGenAlgorithm           Q_CRYPTOIV_GEN_ALGORITHM
                                    QCRYPTO_IV_GEN_ALGORITHM
                                    QCRYPTO_IVGEN_ALG
    QCryptoRSAPaddingAlgorithm      Q_CRYPTORSA_PADDING_ALGORITHM
                                    QCRYPTO_RSA_PADDING_ALGORITHM
                                    QCRYPTO_RSA_PADDING_ALG
    QCryptodevBackendAlgType        Q_CRYPTODEV_BACKEND_ALG_TYPE
                                    QCRYPTODEV_BACKEND_ALG_TYPE
                                    QCRYPTODEV_BACKEND_ALG
    QCryptodevBackendServiceType    Q_CRYPTODEV_BACKEND_SERVICE_TYPE
                                    QCRYPTODEV_BACKEND_SERVICE_TYPE
                                    QCRYPTODEV_BACKEND_SERVICE

Subsequent commits will tweak things to remove most of these prefixes.
Only QAUTHZ_LIST_FORMAT and QAUTHZ_LIST_POLICY will remain.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240904111836.3273842-2-armbru@redhat.com>

Merge tag 'migration-20240909-pull-request' of https://gitlab.com/peterx/qemu into staging

Migration pull request for 9.2

- Mattias's patch to support concurrent bounce buffers for PCI devices
- David's memory leak fix in dirty_memory_extend()
- Fabiano's CI fix to disable vmstate-static-checker test in compat tests
- Denis's patch that adds one more trace point for cpu throttle changes
- Yichen's multifd qatzip compressor support

# -----BEGIN PGP SIGNATURE-----
#
# iIgEABYKADAWIQS5GE3CDMRX2s990ak7X8zN86vXBgUCZt9VlhIccGV0ZXJ4QHJl
# ZGhhdC5jb20ACgkQO1/MzfOr1wZ+4QD+NPzprFD7RF2DPHT5bdo6NTWFSZxW4dyD
# oFp2vhYDEAYA/A5TTfOh3QpYBOaP2PxztZIZSLgs1bokhp+sLM3/PVsK
# =6JYP
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 09 Sep 2024 21:07:50 BST
# gpg:                using EDDSA key B9184DC20CC457DACF7DD1A93B5FCCCDF3ABD706
# gpg:                issuer "peterx@redhat.com"
# gpg: Good signature from "Peter Xu <xzpeter@gmail.com>" [marginal]
# gpg:                 aka "Peter Xu <peterx@redhat.com>" [marginal]
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: B918 4DC2 0CC4 57DA CF7D  D1A9 3B5F CCCD F3AB D706

* tag 'migration-20240909-pull-request' of https://gitlab.com/peterx/qemu:
  system: improve migration debug
  tests/migration: Add integration test for 'qatzip' compression method
  migration: Introduce 'qatzip' compression method
  migration: Add migration parameters for QATzip
  meson: Introduce 'qatzip' feature to the build system
  docs/migration: add qatzip compression feature
  ci: migration: Don't run python tests in the compat job
  softmmu/physmem: fix memory leak in dirty_memory_extend()
  softmmu: Support concurrent bounce buffers

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

system: improve migration debug

Right now migration_throttle() tracepoint lacks very important
important information, i.e. no one could easily say how much the guest
is throttled. This makes difficult to debug guest quality of service
during migration.

This patch adds one more tracepoint into cpu_throttle_set() which is
actually doing this job.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Peter Xu <peterx@redhat.com>
CC: Fabiano Rosas <farosas@suse.de>
CC: Paolo Bonzini <pbonzini@redhat.com>
Link: https://lore.kernel.org/r/20240905191941.310592-1-den@openvz.org
Signed-off-by: Peter Xu <peterx@redhat.com>

tests/migration: Add integration test for 'qatzip' compression method

Adds an integration test for 'qatzip'.

Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Bryan Zhang <bryan.zhang@bytedance.com>
Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
Signed-off-by: Yichen Wang <yichen.wang@bytedance.com>
Link: https://lore.kernel.org/r/20240830232722.58272-6-yichen.wang@bytedance.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Introduce 'qatzip' compression method

Adds support for 'qatzip' as an option for the multifd compression
method parameter, and implements using QAT for 'qatzip' compression and
decompression.

Acked-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Prasad Pandit <pjp@fedoraproject.org>
Signed-off-by: Bryan Zhang <bryan.zhang@bytedance.com>
Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
Signed-off-by: Yichen Wang <yichen.wang@bytedance.com>
Link: https://lore.kernel.org/r/20240830232722.58272-5-yichen.wang@bytedance.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Add migration parameters for QATzip

Adds support for migration parameters to control QATzip compression
level.

Acked-by: Markus Armbruster <armbru@redhat.com>
Signed-off-by: Bryan Zhang <bryan.zhang@bytedance.com>
Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
Signed-off-by: Yichen Wang <yichen.wang@bytedance.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Prasad Pandit <pjp@fedoraproject.org>
Link: https://lore.kernel.org/r/20240830232722.58272-4-yichen.wang@bytedance.com
Signed-off-by: Peter Xu <peterx@redhat.com>

meson: Introduce 'qatzip' feature to the build system

Add a 'qatzip' feature, which is automatically disabled, and which
depends on the QATzip library if enabled.

Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Bryan Zhang <bryan.zhang@bytedance.com>
Signed-off-by: Hao Xiang <hao.xiang@linux.dev>
Signed-off-by: Yichen Wang <yichen.wang@bytedance.com>
Link: https://lore.kernel.org/r/20240830232722.58272-3-yichen.wang@bytedance.com
Signed-off-by: Peter Xu <peterx@redhat.com>

docs/migration: add qatzip compression feature

add Intel QATzip compression method introduction

Reviewed-by: Nanhai Zou <nanhai.zou@intel.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Yuan Liu <yuan1.liu@intel.com>
Signed-off-by: Yichen Wang <yichen.wang@bytedance.com>
Link: https://lore.kernel.org/r/20240830232722.58272-2-yichen.wang@bytedance.com
Signed-off-by: Peter Xu <peterx@redhat.com>

ci: migration: Don't run python tests in the compat job

The vmstate-checker-script test has a bug that makes it flaky. It was
also committed by mistake and will be removed.

Since the migration-compat job takes the tests from the build-previous
job instead of the current HEAD, neither a fix or a removal of the
test will take effect for this release.

Disable the faulty/undesirable test by taking advantage that it only
runs if the PYTHON environment variable is set. This also disables the
analyze-migration-script test, but this is fine because that test
doesn't have migration compatibility implications.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Link: https://lore.kernel.org/r/20240905185445.8179-1-farosas@suse.de
[peterx: Added a TODO to remove the line after 9.2 release, per thuth]
Signed-off-by: Peter Xu <peterx@redhat.com>

softmmu/physmem: fix memory leak in dirty_memory_extend()

As reported by Peter, we might be leaking memory when removing the
highest RAMBlock (in the weird ram_addr_t space), and adding a new one.

We will fail to realize that we already allocated bitmaps for more
dirty memory blocks, and effectively discard the pointers to them.

Fix it by getting rid of last_ram_page() and by remembering the number
of dirty memory blocks that have been allocated already.

While at it, let's use "unsigned int" for the number of blocks, which
should be sufficient until we reach ~32 exabytes.

Looks like this leak was introduced as we switched from using a single
bitmap_zero_extend() to allocating multiple bitmaps:
bitmap_zero_extend() relies on g_renew() which should have taken care of
this.

Resolves: https://lkml.kernel.org/r/CAFEAcA-k7a+VObGAfCFNygQNfCKL=AfX6A4kScq=VSSK0peqPg@mail.gmail.com
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Fixes: 5b82b703b69a ("memory: RCU ram_list.dirty_memory[] for safe RAM hotplug")
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Tested-by: Peter Maydell <peter.maydell@linaro.org>
Cc: qemu-stable@nongnu.org
Cc: Stefan Hajnoczi <stefanha@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Peter Xu <peterx@redhat.com>
Cc: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: David Hildenbrand <david@redhat.com>
Link: https://lore.kernel.org/r/20240828090743.128647-1-david@redhat.com
Signed-off-by: Peter Xu <peterx@redhat.com>

softmmu: Support concurrent bounce buffers

When DMA memory can't be directly accessed, as is the case when
running the device model in a separate process without shareable DMA
file descriptors, bounce buffering is used.

It is not uncommon for device models to request mapping of several DMA
regions at the same time. Examples include:
 * net devices, e.g. when transmitting a packet that is split across
   several TX descriptors (observed with igb)
 * USB host controllers, when handling a packet with multiple data TRBs
   (observed with xhci)

Previously, qemu only provided a single bounce buffer per AddressSpace
and would fail DMA map requests while the buffer was already in use. In
turn, this would cause DMA failures that ultimately manifest as hardware
errors from the guest perspective.

This change allocates DMA bounce buffers dynamically instead of
supporting only a single buffer. Thus, multiple DMA mappings work
correctly also when RAM can't be mmap()-ed.

The total bounce buffer allocation size is limited individually for each
AddressSpace. The default limit is 4096 bytes, matching the previous
maximum buffer size. A new x-max-bounce-buffer-size parameter is
provided to configure the limit for PCI devices.

Signed-off-by: Mattias Nissler <mnissler@rivosinc.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/20240819135455.2957406-1-mnissler@rivosinc.com
Signed-off-by: Peter Xu <peterx@redhat.com>