OSDN Git Service

qmiga/qemu.git
10 months agoMerge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging
Stefan Hajnoczi [Mon, 28 Aug 2023 19:53:30 +0000 (15:53 -0400)]
Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging

* separate accepted and auto-installed versions of Python dependencies
* bump tricore container to Debian 11
* small configure cleanups

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmTsZ14UHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroMlVQf+Juomqo/luBwWwwguEZp32s+c+CYI
# HZJJJSIycq/VY2OsT9e+H1eMJYsCsdzJxn1NcnmEIUSMRkIuCxV5F62gaMl6BjgF
# tH8v4y1ZBDc0i0zw6qkuZM4sydNkK1XohGeOp8NkTE7F2fX0DT2AO17rSKIHh77R
# enNE5yq+s0YGHfYz7PbNvT1G+YXqt9SEEfCqIHkCQccjgFx9PEJu7PPuWdIYLG5s
# VVIyrbZzcX7OmQCCWdEZCe5t8swbOHtzE5D3JUVvfnUDj3BONXQybp/14rEikrjU
# fuy9sf3qW4XlwzPOUWFlPfxJIg8KWB1fL2wIppDn2gKrBB7fekwz5hlJRA==
# =lZmw
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 28 Aug 2023 05:22:38 EDT
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* tag 'for-upstream' of https://gitlab.com/bonzini/qemu:
  configure: remove unnecessary mkdir -p
  configure: fix container_hosts misspellings and duplications
  target/i386: add support for VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE
  tests/docker: add python3-tomli dependency to containers
  Revert "tests: Use separate virtual environment for avocado"
  configure: switch to ensuregroup
  python: use vendored tomli
  configure: never use PyPI for Meson
  lcitool: bump libvirt-ci submodule and regenerate
  python: mkvenv: add ensuregroup command
  python: mkvenv: introduce TOML-like representation of dependencies
  python: mkvenv: tweak the matching of --diagnose to depspecs
  dockerfiles: bump tricore cross compiler container to Debian 11
  configure: fix and complete detection of tricore tools

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 months agoMerge tag 'devel-hppa-priv-cleanup2-pull-request' of https://github.com/hdeller/qemu...
Stefan Hajnoczi [Mon, 28 Aug 2023 19:12:01 +0000 (15:12 -0400)]
Merge tag 'devel-hppa-priv-cleanup2-pull-request' of https://github.com/hdeller/qemu-hppa into staging

target/hppa: Clean up conversion from/to MMU index and privilege level

Make the conversion between privilege level and QEMU MMU index
consistent, and afterwards switch to MMU indices 11-15.

Signed-off-by: Helge Deller <deller@gmx.de>
# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYKAB0WIQS86RI+GtKfB8BJu973ErUQojoPXwUCZOtpFAAKCRD3ErUQojoP
# X0lxAPwKfsMZOO/e81XXLgxeEZ5R4yjtIelErvOWmMvBfxEDUwEA6HgJt4gOe1uR
# Dw7d+wTqr+CSOj5I87+sJYl1FmihzQU=
# =01eA
# -----END PGP SIGNATURE-----
# gpg: Signature made Sun 27 Aug 2023 11:17:40 EDT
# gpg:                using EDDSA key BCE9123E1AD29F07C049BBDEF712B510A23A0F5F
# gpg: Good signature from "Helge Deller <deller@gmx.de>" [unknown]
# gpg:                 aka "Helge Deller <deller@kernel.org>" [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: 4544 8228 2CD9 10DB EF3D  25F8 3E5F 3D04 A7A2 4603
#      Subkey fingerprint: BCE9 123E 1AD2 9F07 C049  BBDE F712 B510 A23A 0F5F

* tag 'devel-hppa-priv-cleanup2-pull-request' of https://github.com/hdeller/qemu-hppa:
  target/hppa: Switch to use MMU indices 11-15
  target/hppa: Use privilege helper in hppa_get_physical_address()
  target/hppa: Do not use hardcoded value for tlb_flush_*()
  target/hppa: Add privilege to MMU index conversion helpers
  target/hppa: Add missing PL1 and PL2 privilege levels

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 months agoconfigure: remove unnecessary mkdir -p
Paolo Bonzini [Mon, 7 Aug 2023 12:17:43 +0000 (14:17 +0200)]
configure: remove unnecessary mkdir -p

It is already included in the symlink shell function.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agoconfigure: fix container_hosts misspellings and duplications
Paolo Bonzini [Mon, 7 Aug 2023 09:22:08 +0000 (11:22 +0200)]
configure: fix container_hosts misspellings and duplications

container_hosts is matched against $cpu, so it must contain QEMU
canonical architecture names, not Debian architecture names.
Also do not set $container_hosts inside the loop, since it is
already set before.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agotarget/i386: add support for VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE
Ake Koomsin [Mon, 7 Aug 2023 09:33:40 +0000 (18:33 +0900)]
target/i386: add support for VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE

Current QEMU can expose waitpkg to guests when it is available. However,
VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE is still not recognized and
masked by QEMU. This can lead to an unexpected situation when a L1
hypervisor wants to expose waitpkg to a L2 guest. The L1 hypervisor can
assume that VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE exists as waitpkg is
available. The L1 hypervisor then can accidentally expose waitpkg to the
L2 guest. This will cause invalid opcode exception in the L2 guest when
it executes waitpkg related instructions.

This patch adds VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE support, and
sets up dependency between the bit and CPUID_7_0_ECX_WAITPKG. QEMU should
not expose waitpkg feature if VMX_SECONDARY_EXEC_ENABLE_USER_WAIT_PAUSE is
not available to avoid unexpected invalid opcode exception in L2 guests.

Signed-off-by: Ake Koomsin <ake@igel.co.jp>
Message-ID: <20230807093339.32091-2-ake@igel.co.jp>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agotests/docker: add python3-tomli dependency to containers
Paolo Bonzini [Tue, 8 Aug 2023 21:35:47 +0000 (23:35 +0200)]
tests/docker: add python3-tomli dependency to containers

Instead of having CI pick tomli from the vendored wheel at configure
time, place it in the containers.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agoRevert "tests: Use separate virtual environment for avocado"
Paolo Bonzini [Tue, 8 Aug 2023 09:28:08 +0000 (11:28 +0200)]
Revert "tests: Use separate virtual environment for avocado"

This reverts commit e8e4298feadae7924cf7600bb3bcc5b0a8d7cbe9.

ensuregroup allows to specify both the acceptable versions of avocado,
and a locked version to be used when avocado is not installed as a system
pacakge.  This lets us install avocado in pyvenv/ using "mkvenv.py" and
reuse the distro package on Fedora and CentOS Stream (the only distros
where it's available).

ensuregroup's usage of "(>=..., <=...)" constraints when evaluating
the distro package, and "==" constraints when installing it from PyPI,
makes it possible to avoid conflicts between the known-good version and
a package plugins included in the distro.

This is because package plugins have "==" constraints on the version
that is included in the distro, and, using "pip install avocado==88.1"
on a venv that includes system packages will result in an error:

   avocado-framework-plugin-varianter-yaml-to-mux 98.0 requires avocado-framework==98.0, but you have avocado-framework 88.1 which is incompatible.
   avocado-framework-plugin-result-html 98.0 requires avocado-framework==98.0, but you have avocado-framework 88.1 which is incompatible.

But at the same time, if the venv does not include a system distribution
of avocado then we can install a known-good version and stick to LTS
releases.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1663
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agoconfigure: switch to ensuregroup
Paolo Bonzini [Tue, 8 Aug 2023 09:23:48 +0000 (11:23 +0200)]
configure: switch to ensuregroup

Using the new ensuregroup command, the desired versions of meson and
sphinx can be placed in pythondeps.toml rather than configure.

The meson.install entry in pythondeps.toml matches the version that is
found in python/wheels.  This ensures that mkvenv.py uses the bundled
wheel even if PyPI is enabled; thus not introducing warnings or errors
from versions that are more recent than the one used in CI.

The sphinx entries match what is shipped in Fedora 38.  It's the
last release that has support for older versions of Python (sphinx 6.0
requires Python 3.8) and especially docutils (of which sphinx 6.0 requires
version 0.18).  This is important because Ubuntu 20.04 has docutils 0.14
and Debian 11 has docutils 0.16.

"mkvenv.py ensure" is only used to bootstrap tomli.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agopython: use vendored tomli
Paolo Bonzini [Tue, 8 Aug 2023 18:19:43 +0000 (20:19 +0200)]
python: use vendored tomli

Debian only introduced tomli in the bookworm release.  Use a
vendored wheel to avoid requiring a package that is only in
bullseye-backports and is also absent in Ubuntu 20.04.

While at it, fix an issue in the vendor.py scripts which does
not add a newline after each package and hash.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agoconfigure: never use PyPI for Meson
Paolo Bonzini [Tue, 8 Aug 2023 18:28:25 +0000 (20:28 +0200)]
configure: never use PyPI for Meson

Since there is a vendored copy, there is no point in choosing online
operation.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agolcitool: bump libvirt-ci submodule and regenerate
Paolo Bonzini [Tue, 8 Aug 2023 13:31:22 +0000 (15:31 +0200)]
lcitool: bump libvirt-ci submodule and regenerate

This brings in a newer version of the pipewire mapping, so rename it.

Python 3.9 and 3.10 do not seem to work in OpenSUSE LEAP 15.5 (weird,
because 3.9 persisted from 15.3 to 15.4) so bump the Python runtime
version to 3.11.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agopython: mkvenv: add ensuregroup command
Paolo Bonzini [Tue, 8 Aug 2023 08:03:42 +0000 (10:03 +0200)]
python: mkvenv: add ensuregroup command

Introduce a new subcommand that retrieves the packages to be installed
from a TOML file. This allows being more flexible in using the system
version of a package, while at the same time using a known-good version
when installing the package.  This is important for packages that
sometimes have backwards-incompatible changes or that depend on
specific versions of their dependencies.

Compared to JSON, TOML is more human readable and easier to edit.  A
parser is available in 3.11 but also available as a small (12k) package
for older versions, tomli.  While tomli is bundled with pip, this is only
true of recent versions of pip.  Of all the supported OSes pretty much
only FreeBSD has a recent enough version of pip while staying on Python
<3.11.  So we cannot use the same trick that is in place for distlib.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agopython: mkvenv: introduce TOML-like representation of dependencies
Paolo Bonzini [Tue, 8 Aug 2023 07:47:25 +0000 (09:47 +0200)]
python: mkvenv: introduce TOML-like representation of dependencies

We would like to place all Python dependencies in the same file, so that
we can add more information without having long and complex command lines.
The plan is to have a TOML file with one entry per package, for example

  [avocado]
  avocado-framework = {
    accepted = "(>=88.1, <93.0)",
    installed = "88.1",
    canary = "avocado"
  }

Each TOML section will thus be a dictionary of dictionaries.  Modify
mkvenv.py's workhorse function, _do_ensure, to already operate on such
a data structure.  The "ensure" subcommand is modified to separate the
depspec into a name and a version part, and use the result (plus the
--diagnose argument) to build a dictionary for each command line argument.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agopython: mkvenv: tweak the matching of --diagnose to depspecs
Paolo Bonzini [Tue, 8 Aug 2023 11:25:09 +0000 (13:25 +0200)]
python: mkvenv: tweak the matching of --diagnose to depspecs

Move the matching between the "absent" array and dep_specs[0] inside
the loop, preparing for the possibility of having multiple canaries
among the installed packages.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agodockerfiles: bump tricore cross compiler container to Debian 11
Paolo Bonzini [Tue, 8 Aug 2023 14:02:57 +0000 (16:02 +0200)]
dockerfiles: bump tricore cross compiler container to Debian 11

With the release of version 12 on June 10, 2023, Debian 10 is
not supported anymore.  Modify the cross compiler container to
build on a newer version.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agoconfigure: fix and complete detection of tricore tools
Paolo Bonzini [Wed, 9 Aug 2023 08:13:43 +0000 (10:13 +0200)]
configure: fix and complete detection of tricore tools

The tricore tools are not detected when they are installed in
the host system, only if they are taken from an external
container.  For this reason the build-tricore-softmmu job
was not running the TCG tests.

In addition the container provides all tools, not just as/ld/gcc,
so there is no need to special case tricore.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
10 months agotarget/hppa: Switch to use MMU indices 11-15
Helge Deller [Mon, 7 Aug 2023 10:14:36 +0000 (12:14 +0200)]
target/hppa: Switch to use MMU indices 11-15

The MMU indices 9-15 will use shorter assembler instructions
when run on a x86-64 host. So, switch over to those to get
smaller code and maybe minimally faster emulation.

Signed-off-by: Helge Deller <deller@gmx.de>
10 months agotarget/hppa: Use privilege helper in hppa_get_physical_address()
Helge Deller [Mon, 7 Aug 2023 09:52:39 +0000 (11:52 +0200)]
target/hppa: Use privilege helper in hppa_get_physical_address()

Convert hppa_get_physical_address() to use the privilege helper macro.

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agotarget/hppa: Do not use hardcoded value for tlb_flush_*()
Helge Deller [Mon, 7 Aug 2023 09:42:11 +0000 (11:42 +0200)]
target/hppa: Do not use hardcoded value for tlb_flush_*()

Avoid using hardcoded values when calling the tlb_flush*() functions.
Instead, define and use HPPA_MMU_FLUSH_MASK (keeping the current
behavior, which doesn't flush the physical address MMU).

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agotarget/hppa: Add privilege to MMU index conversion helpers
Helge Deller [Mon, 7 Aug 2023 09:32:09 +0000 (11:32 +0200)]
target/hppa: Add privilege to MMU index conversion helpers

Add two macros which convert privilege level to/from MMU index:

- PRIV_TO_MMU_IDX(priv)
    returns the MMU index for the given privilege level

- MMU_IDX_TO_PRIV(mmu_idx)
    returns the corresponding privilege level for this MMU index

The introduction of those macros make the code easier to read and
will help to improve performance in follow-up patch.

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agotarget/hppa: Add missing PL1 and PL2 privilege levels
Helge Deller [Mon, 7 Aug 2023 09:17:59 +0000 (11:17 +0200)]
target/hppa: Add missing PL1 and PL2 privilege levels

The hppa CPU has 4 privilege levels (0-3).
Mention the missing PL1 and PL2 levels, although the Linux kernel
uses only 0 (KERNEL) and 3 (USER). Not sure about HP-UX.

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agoMerge tag 'pull-target-arm-20230824' of https://git.linaro.org/people/pmaydell/qemu...
Stefan Hajnoczi [Thu, 24 Aug 2023 14:08:33 +0000 (10:08 -0400)]
Merge tag 'pull-target-arm-20230824' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * hw/gpio/nrf51: implement DETECT signal
 * accel/kvm: Specify default IPA size for arm64
 * ptw: refactor, fix some FEAT_RME bugs
 * target/arm: Adjust PAR_EL1.SH for Device and Normal-NC memory types
 * target/arm/helper: Implement CNTHCTL_EL2.CNT[VP]MASK
 * Fix SME ST1Q
 * Fix 64-bit SSRA

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmTnIoUZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3vufEACPJcwyFvSBHDv4VQ6tbgOU
# zwjpUMv4RMKhCOjuxBlJ2DICwOcGNuKer0tc6wkH2T5Ebhoego1osYbRZZoawAJf
# ntg+Ndrx1QH9ORuGqYccLXtHnP741KiKggDHM05BJqB7rqtuH+N4fEn7Cdsw/DNg
# XuCYD5QrxMYvkSOD1l8W0aqp81ucYPgkFqLufypgxrXUiRZ1RBAmPF47BFFdnM8f
# NmrmT1LTF5jr70ySRB+ukK6BAGDc0CUfs6R6nYRwUjRPmSG2rrtUDGo+nOQGDqJo
# PHWmt7rdZQG2w7HVyE/yc3h/CQ3NciwWKbCkRlaoujxHx/B6DRynSeO3NXsP8ELu
# Gizoi3ltwHDQVIGQA19P5phZKHZf7x3MXmK4fDBGB9znvoSFTcjJqkdaN/ARXXO3
# e1vnK1MqnPI8Z1nGdeVIAUIrqhtLHnrrM7jf1tI/e4sjpl3prHq2PvQkakXu8clr
# H8bPZ9zZzyrrSbl4NhpaFTsUiYVxeLoJsNKAmG8dHb+9YsFGXTvEBhtR9eUxnbaV
# XyZ3jEdeW7/ngQ4C6XMD2ZDiKVdx2xJ2Pp5npvljldjmtGUvwQabKo+fPDt2fKjM
# BwjhHA50I633k4fYIwm8YOb70I4oxoL9Lr6PkKriWPMTI5r7+dtwgigREVwnCn+Y
# RsiByKMkDO2TcoQjvBZlCA==
# =3MJ8
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 24 Aug 2023 05:27:33 EDT
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [full]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [full]
# gpg:                 aka "Peter Maydell <peter@archaic.org.uk>" [unknown]
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* tag 'pull-target-arm-20230824' of https://git.linaro.org/people/pmaydell/qemu-arm: (35 commits)
  target/arm: Fix 64-bit SSRA
  target/arm: Fix SME ST1Q
  target/arm/helper: Implement CNTHCTL_EL2.CNT[VP]MASK
  target/arm/helper: Check SCR_EL3.{NSE, NS} encoding for AT instructions
  target/arm: Pass security space rather than flag for AT instructions
  target/arm: Skip granule protection checks for AT instructions
  target/arm/helper: Fix tlbmask and tlbbits for TLBI VAE2*
  target/arm/ptw: Load stage-2 tables from realm physical space
  target/arm: Adjust PAR_EL1.SH for Device and Normal-NC memory types
  target/arm/ptw: Report stage 2 fault level for stage 2 faults on stage 1 ptw
  target/arm/ptw: Check for block descriptors at invalid levels
  target/arm/ptw: Set attributes correctly for MMU disabled data accesses
  target/arm/ptw: Drop S1Translate::out_secure
  target/arm/ptw: Remove S1Translate::in_secure
  target/arm/ptw: Remove last uses of ptw->in_secure
  target/arm/ptw: Only fold in NSTable bit effects in Secure state
  target/arm: Pass an ARMSecuritySpace to arm_is_el2_enabled_secstate()
  target/arm/ptw: Pass an ARMSecuritySpace to arm_hcr_el2_eff_secstate()
  target/arm/ptw: Pass ARMSecurityState to regime_translation_disabled()
  target/arm/ptw: Pass ptw into get_phys_addr_pmsa*() and get_phys_addr_disabled()
  ...

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 months agoMerge tag 'pull-loongarch-20230824' of https://gitlab.com/gaosong/qemu into staging
Stefan Hajnoczi [Thu, 24 Aug 2023 13:17:05 +0000 (09:17 -0400)]
Merge tag 'pull-loongarch-20230824' of https://gitlab.com/gaosong/qemu into staging

pull-loongarch-20230824

# -----BEGIN PGP SIGNATURE-----
#
# iLMEAAEIAB0WIQS4/x2g0v3LLaCcbCxAov/yOSY+3wUCZOcdAwAKCRBAov/yOSY+
# 3w3CA/sH8+Ay+Qnaqa2vEyuhOlFQuxHKeR7mYfsitAdzh8yMK2K8C2iBUzDzL1H3
# kZmZbCcYX7ko9RLhsuXmvfBJ7iwzY55ozSHLIjJ/VS4JVE5B0cUSZ5jjIPDqpzDs
# 7TUt9qpTkwg0e+klzVREWLSWP5xopvkRvFHZM3KZZhGMphOTUQ==
# =/HHZ
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 24 Aug 2023 05:04:03 EDT
# 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-20230824' of https://gitlab.com/gaosong/qemu: (31 commits)
  hw/loongarch: Fix ACPI processor id off-by-one error
  target/loongarch: Split fcc register to fcc0-7 in gdbstub
  hw/intc/loongarch_pch: fix edge triggered irq handling
  target/loongarch: cpu: Implement get_arch_id callback
  target/loongarch: Add avail_IOCSR to check iocsr instructions
  target/loongarch: Add avail_LSX to check LSX instructions
  target/loongarch: Add avail_LAM to check atomic instructions
  target/loongarch: Add avail_LSPW to check LSPW instructions
  target/loongarch: Add avail_FP/FP_SP/FP_DP to check fpu instructions
  hw/loongarch: Remove restriction of la464 cores in the virt machine
  target/loongarch: Add LoongArch32 cpu la132
  target/loongarch: Add avail_64 to check la64-only instructions
  target/loongarch: Add a check parameter to the TRANS macro
  target/loongarch: Sign extend results in VA32 mode
  target/loongarch: Truncate high 32 bits of address in VA32 mode
  target/loongarch: Extract set_pc() helper
  target/loongarch: Extract make_address_pc() helper
  target/loongarch: Extract make_address_i() helper
  target/loongarch: Extract make_address_x() helper
  target/loongarch: Add LA64 & VA32 to DisasContext
  ...

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 months agohw/loongarch: Fix ACPI processor id off-by-one error
Jiajie Chen [Sun, 20 Aug 2023 10:56:59 +0000 (18:56 +0800)]
hw/loongarch: Fix ACPI processor id off-by-one error

In hw/acpi/aml-build.c:build_pptt() function, the code assumes that the
ACPI processor id equals to the cpu index, for example if we have 8
cpus, then the ACPI processor id should be in range 0-7.

However, in hw/loongarch/acpi-build.c:build_madt() function we broke the
assumption. If we have 8 cpus again, the ACPI processor id in MADT table
would be in range 1-8. It violates the following description taken from
ACPI spec 6.4 table 5.138:

If the processor structure represents an actual processor, this field
must match the value of ACPI processor ID field in the processor’s entry
in the MADT.

It will break the latest Linux 6.5-rc6 with the
following error message:

ACPI PPTT: PPTT table found, but unable to locate core 7 (8)
Invalid BIOS PPTT

Here 7 is the last cpu index, 8 is the ACPI processor id learned from
MADT.

With this patch, Linux can properly detect SMT threads when "-smp
8,sockets=1,cores=4,threads=2" is passed:

Thread(s) per core:  2
Core(s) per socket:  2
Socket(s):           2

The detection of number of sockets is still wrong, but that is out of
scope of the commit.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Bibo Mao <maobibo@loongson.cn>
Message-Id: <20230820105658.99123-2-c@jia.je>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agotarget/loongarch: Split fcc register to fcc0-7 in gdbstub
Jiajie Chen [Tue, 8 Aug 2023 05:42:47 +0000 (13:42 +0800)]
target/loongarch: Split fcc register to fcc0-7 in gdbstub

Since GDB 13.1(GDB commit ea3352172), GDB LoongArch changed to use
fcc0-7 instead of fcc register. This commit partially reverts commit
2f149c759 (`target/loongarch: Update gdb_set_fpu() and gdb_get_fpu()`)
to match the behavior of GDB.

Note that it is a breaking change for GDB 13.0 or earlier, but it is
also required for GDB 13.1 or later to work.

Signed-off-by: Jiajie Chen <c@jia.je>
Acked-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230808054315.3391465-1-c@jia.je>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agohw/intc/loongarch_pch: fix edge triggered irq handling
Bibo Mao [Fri, 7 Jul 2023 09:15:57 +0000 (17:15 +0800)]
hw/intc/loongarch_pch: fix edge triggered irq handling

For edge triggered irq, qemu_irq_pulse is used to inject irq. It will
set irq with high level and low level soon to simluate pulse irq.

For edge triggered irq, irq is injected and set as pending at rising
level, do not clear irq at lowering level. LoongArch pch interrupt will
clear irq for lowering level irq, there will be problem. ACPI ged deivce
is edge-triggered irq, it is used for cpu/memory hotplug.

This patch fixes memory hotplug issue on LoongArch virt machine.

Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230707091557.1474790-1-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agotarget/loongarch: cpu: Implement get_arch_id callback
Bibo Mao [Thu, 24 Aug 2023 00:50:07 +0000 (08:50 +0800)]
target/loongarch: cpu: Implement get_arch_id callback

Implement the callback for getting the architecture-dependent CPU
ID, the cpu ID is physical id described in ACPI MADT table, this
will be used for cpu hotplug.

Signed-off-by: Bibo Mao <maobibo@loongson.cn>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230824005007.2000525-1-maobibo@loongson.cn>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agotarget/loongarch: Add avail_IOCSR to check iocsr instructions
Song Gao [Tue, 22 Aug 2023 07:22:19 +0000 (09:22 +0200)]
target/loongarch: Add avail_IOCSR to check iocsr instructions

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20230822032724.1353391-16-gaosong@loongson.cn>
Message-Id: <20230822072219.35719-1-philmd@linaro.org>

10 months agotarget/loongarch: Add avail_LSX to check LSX instructions
Song Gao [Tue, 22 Aug 2023 07:30:26 +0000 (09:30 +0200)]
target/loongarch: Add avail_LSX to check LSX instructions

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20230822032724.1353391-15-gaosong@loongson.cn>
Message-Id: <20230822073026.35776-1-philmd@linaro.org>

10 months agotarget/loongarch: Add avail_LAM to check atomic instructions
Song Gao [Tue, 22 Aug 2023 07:19:57 +0000 (09:19 +0200)]
target/loongarch: Add avail_LAM to check atomic instructions

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20230822032724.1353391-14-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-8-philmd@linaro.org>

10 months agotarget/loongarch: Add avail_LSPW to check LSPW instructions
Song Gao [Tue, 22 Aug 2023 07:19:56 +0000 (09:19 +0200)]
target/loongarch: Add avail_LSPW to check LSPW instructions

Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20230822032724.1353391-13-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-7-philmd@linaro.org>

10 months agotarget/loongarch: Add avail_FP/FP_SP/FP_DP to check fpu instructions
Song Gao [Tue, 22 Aug 2023 07:19:55 +0000 (09:19 +0200)]
target/loongarch: Add avail_FP/FP_SP/FP_DP to check fpu instructions

Signed-off-by: Song Gao <gaosong@loongson.cn>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20230822032724.1353391-12-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-6-philmd@linaro.org>

10 months agohw/loongarch: Remove restriction of la464 cores in the virt machine
Song Gao [Tue, 22 Aug 2023 07:19:54 +0000 (09:19 +0200)]
hw/loongarch: Remove restriction of la464 cores in the virt machine

Allow virt machine to be used with la132 instead of la464.

Co-authored-by: Jiajie Chen <c@jia.je>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-11-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-5-philmd@linaro.org>

10 months agotarget/loongarch: Add LoongArch32 cpu la132
Jiajie Chen [Tue, 22 Aug 2023 07:19:53 +0000 (09:19 +0200)]
target/loongarch: Add LoongArch32 cpu la132

Add LoongArch32 cpu la132.

Due to lack of public documentation of la132, it is currently a
synthetic LoongArch32 cpu model. Details need to be added in the future.

Signed-off-by: Jiajie Chen <c@jia.je>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Acked-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-10-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-4-philmd@linaro.org>

10 months agotarget/loongarch: Add avail_64 to check la64-only instructions
Song Gao [Tue, 22 Aug 2023 07:19:52 +0000 (09:19 +0200)]
target/loongarch: Add avail_64 to check la64-only instructions

The la32 instructions listed in Table 2 at
https://loongson.github.io/LoongArch-Documentation/LoongArch-Vol1-EN.html#overview-of-basic-integer-instructions

Co-authored-by: Jiajie Chen <c@jia.je>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20230822032724.1353391-9-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-3-philmd@linaro.org>

10 months agotarget/loongarch: Add a check parameter to the TRANS macro
Song Gao [Tue, 22 Aug 2023 07:19:51 +0000 (09:19 +0200)]
target/loongarch: Add a check parameter to the TRANS macro

The default check parmeter is ALL.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20230822032724.1353391-8-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-2-philmd@linaro.org>

10 months agotarget/loongarch: Sign extend results in VA32 mode
Jiajie Chen [Tue, 22 Aug 2023 07:19:50 +0000 (09:19 +0200)]
target/loongarch: Sign extend results in VA32 mode

In VA32 mode, BL, JIRL and PC* instructions should sign-extend the low
32 bit result to 64 bits.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-7-gaosong@loongson.cn>
Message-Id: <20230822071959.35620-1-philmd@linaro.org>

10 months agotarget/loongarch: Truncate high 32 bits of address in VA32 mode
Jiajie Chen [Tue, 22 Aug 2023 07:13:55 +0000 (09:13 +0200)]
target/loongarch: Truncate high 32 bits of address in VA32 mode

When running in VA32 mode(!LA64 or VA32L[1-3] matching PLV), virtual
address is truncated to 32 bits before address mapping.

Signed-off-by: Jiajie Chen <c@jia.je>
Co-authored-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-6-gaosong@loongson.cn>
Message-Id: <20230822071405.35386-10-philmd@linaro.org>

10 months agotarget/loongarch: Extract set_pc() helper
Jiajie Chen [Tue, 22 Aug 2023 07:13:54 +0000 (09:13 +0200)]
target/loongarch: Extract set_pc() helper

Signed-off-by: Jiajie Chen <c@jia.je>
Co-authored-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-6-gaosong@loongson.cn>
[PMD: Extract helper from bigger patch]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230822071405.35386-9-philmd@linaro.org>

10 months agotarget/loongarch: Extract make_address_pc() helper
Jiajie Chen [Tue, 22 Aug 2023 07:13:53 +0000 (09:13 +0200)]
target/loongarch: Extract make_address_pc() helper

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-7-gaosong@loongson.cn>
[PMD: Extract helper from bigger patch]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230822071405.35386-8-philmd@linaro.org>

10 months agotarget/loongarch: Extract make_address_i() helper
Jiajie Chen [Tue, 22 Aug 2023 07:13:52 +0000 (09:13 +0200)]
target/loongarch: Extract make_address_i() helper

Signed-off-by: Jiajie Chen <c@jia.je>
Co-authored-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-6-gaosong@loongson.cn>
[PMD: Extract helper from bigger patch]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230822071405.35386-7-philmd@linaro.org>

10 months agotarget/loongarch: Extract make_address_x() helper
Jiajie Chen [Tue, 22 Aug 2023 07:13:51 +0000 (09:13 +0200)]
target/loongarch: Extract make_address_x() helper

Signed-off-by: Jiajie Chen <c@jia.je>
Co-authored-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230822032724.1353391-6-gaosong@loongson.cn>
[PMD: Extract helper from bigger patch]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230822071405.35386-6-philmd@linaro.org>

10 months agotarget/loongarch: Add LA64 & VA32 to DisasContext
Jiajie Chen [Tue, 22 Aug 2023 07:13:50 +0000 (09:13 +0200)]
target/loongarch: Add LA64 & VA32 to DisasContext

Add LA64 and VA32(32-bit Virtual Address) to DisasContext to allow the
translator to reject doubleword instructions in LA32 mode for example.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20230822032724.1353391-5-gaosong@loongson.cn>
Message-Id: <20230822071405.35386-5-philmd@linaro.org>

10 months agotarget/loongarch: Support LoongArch32 VPPN
Jiajie Chen [Tue, 22 Aug 2023 07:13:49 +0000 (09:13 +0200)]
target/loongarch: Support LoongArch32 VPPN

VPPN of TLBEHI/TLBREHI is limited to 19 bits in LA32.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20230822032724.1353391-4-gaosong@loongson.cn>
Message-Id: <20230822071405.35386-4-philmd@linaro.org>

10 months agotarget/loongarch: Support LoongArch32 DMW
Jiajie Chen [Tue, 22 Aug 2023 07:13:48 +0000 (09:13 +0200)]
target/loongarch: Support LoongArch32 DMW

LA32 uses a different encoding for CSR.DMW and a new direct mapping
mechanism.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20230822032724.1353391-3-gaosong@loongson.cn>
Message-Id: <20230822071405.35386-3-philmd@linaro.org>

10 months agotarget/loongarch: Support LoongArch32 TLB entry
Jiajie Chen [Tue, 22 Aug 2023 07:13:47 +0000 (09:13 +0200)]
target/loongarch: Support LoongArch32 TLB entry

The TLB entry of LA32 lacks NR, NX and RPLV and they are hardwired to
zero in LoongArch32.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20230822032724.1353391-2-gaosong@loongson.cn>
Message-Id: <20230822071405.35386-2-philmd@linaro.org>

10 months agotarget/loongarch: Add GDB support for loongarch32 mode
Jiajie Chen [Mon, 21 Aug 2023 12:59:59 +0000 (14:59 +0200)]
target/loongarch: Add GDB support for loongarch32 mode

GPRs and PC are 32-bit wide in loongarch32 mode.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20230817093121.1053890-4-gaosong@loongson.cn>
[PMD: Rebased, set gdb_num_core_regs]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230821125959.28666-9-philmd@linaro.org>

10 months agotarget/loongarch: Add new object class for loongarch32 cpus
Jiajie Chen [Mon, 21 Aug 2023 12:59:58 +0000 (14:59 +0200)]
target/loongarch: Add new object class for loongarch32 cpus

Add object class stub for future loongarch32 cpus.

Signed-off-by: Jiajie Chen <c@jia.je>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20230817093121.1053890-3-gaosong@loongson.cn>
[Rebased on TYPE_LOONGARCH64_CPU introduction]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230821125959.28666-8-philmd@linaro.org>

10 months agotarget/loongarch: Add function to check current arch
Jiajie Chen [Mon, 21 Aug 2023 12:59:57 +0000 (14:59 +0200)]
target/loongarch: Add function to check current arch

Add is_la64 function to check if the current cpucfg[1].arch equals to
2(LA64).

Signed-off-by: Jiajie Chen <c@jia.je>
Co-authored-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Message-ID: <20230817093121.1053890-2-gaosong@loongson.cn>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230821125959.28666-7-philmd@linaro.org>

10 months agotarget/loongarch: Extract 64-bit specifics to loongarch64_cpu_class_init
Philippe Mathieu-Daudé [Mon, 21 Aug 2023 12:59:56 +0000 (14:59 +0200)]
target/loongarch: Extract 64-bit specifics to loongarch64_cpu_class_init

Extract loongarch64 specific code from loongarch_cpu_class_init()
to a new loongarch64_cpu_class_init().

In preparation of supporting loongarch32 cores, rename these
functions using the '64' suffix.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230821125959.28666-6-philmd@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agotarget/loongarch: Introduce abstract TYPE_LOONGARCH64_CPU
Philippe Mathieu-Daudé [Mon, 21 Aug 2023 12:59:55 +0000 (14:59 +0200)]
target/loongarch: Introduce abstract TYPE_LOONGARCH64_CPU

In preparation of introducing TYPE_LOONGARCH32_CPU, introduce
an abstract TYPE_LOONGARCH64_CPU.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230821125959.28666-5-philmd@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agotarget/loongarch: Fix loongarch_la464_initfn() misses setting LSPW
Song Gao [Mon, 21 Aug 2023 12:59:54 +0000 (14:59 +0200)]
target/loongarch: Fix loongarch_la464_initfn() misses setting LSPW

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230817093121.1053890-11-gaosong@loongson.cn>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230821125959.28666-4-philmd@linaro.org>

10 months agotarget/loongarch: Remove duplicated disas_set_info assignment
Philippe Mathieu-Daudé [Mon, 21 Aug 2023 12:59:53 +0000 (14:59 +0200)]
target/loongarch: Remove duplicated disas_set_info assignment

Commit 228021f05e ("target/loongarch: Add core definition") sets
disas_set_info to loongarch_cpu_disas_set_info. Probably due to
a failed git-rebase, commit ca61e75071 ("target/loongarch: Add gdb
support") also sets it to the same value. Remove the duplication.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230821125959.28666-3-philmd@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agotarget/loongarch: Log I/O write accesses to CSR registers
Philippe Mathieu-Daudé [Mon, 21 Aug 2023 12:59:52 +0000 (14:59 +0200)]
target/loongarch: Log I/O write accesses to CSR registers

Various CSR registers have Read/Write fields. We might
want to see guest trying to change such registers.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Song Gao <gaosong@loongson.cn>
Message-Id: <20230821125959.28666-2-philmd@linaro.org>
Signed-off-by: Song Gao <gaosong@loongson.cn>
10 months agoMerge tag 'pull-request-2023-08-23' of https://gitlab.com/thuth/qemu into staging
Stefan Hajnoczi [Wed, 23 Aug 2023 13:17:41 +0000 (09:17 -0400)]
Merge tag 'pull-request-2023-08-23' of https://gitlab.com/thuth/qemu into staging

* Add compat machines for QEMU 8.2
* Convert some DPRINTFs in s390x code into trace events
* Fix VFMIN/VFMAX, VSTL, VREP and VSTRS s390x instructions
* Fix virtio-gpu on big endian (i.e. s390x) hosts

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmTl8McRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbVXTQ/+OLnw+5+rCA/WbVWfWwEragtmSZuo5302
# ByARdgv8BdD7AobUdSLkWbz9fgU7TAGPnv4aRXBs1K1HE77e63sg6ZfmGUJQllx8
# T/86LPB0dPHZHPt39t3zv/ZTfj+yoXF+7+MIzhSbgV9sumKRqIW/w/BsTI3Rkkwd
# yp1mpkNvYzCkO66nZWYDgKYLTvSmSJx+GUC6dgWswwXSmGP9UU+jutX62lDQS4k4
# l3VjHHhl5V9LENQAvHQ1x48tMIjR4vra8T4fhYLAr1nVsHhsBONRX9qxUHhpy0c3
# zrWA95kO0CPoJLqMNdY9CSyBRgrH/BCmM6Z5+GvBI0RWo+bdgYzF5QpNW6Sbfb/L
# NfE7PK0EYRFk8Q1LK+pYQ0wCjw/a5tOj3NtwZQUUMqqiNL6zmaQpOXujw3PTtIDN
# 6qS9aiAVlL+taIsk9av1So5Mgrr97BsptPKhe22BCYxv832Vj8mPOhjkbiTs8OYD
# PCr+sfJvpwcDBdDhQ1xi4M5tkxg26CPtntVDJdl/pXM3dmpxQ/D8ciok+f2/EeWU
# VeiJ4/tbelm3u0zfGqYfRGSxvPYZM9aJJCLloXuffeT+UEJXUVze6MgMbSuf4vji
# f+hWxA38WUAaoZjoBguMgwLOp/hvBtbHF+Hdk+iT0yE5uc3Ajo619YRVAspJbNi4
# qSYIMAJoGqM=
# =cIfC
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 23 Aug 2023 07:43:03 EDT
# 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-2023-08-23' of https://gitlab.com/thuth/qemu:
  tests/tcg/s390x: Test VSTRS
  target/s390x: Fix the "ignored match" case in VSTRS
  linux-user/elfload: Enable vxe2 on s390x
  include/hw/virtio/virtio-gpu: Fix virtio-gpu with blob on big endian hosts
  hw/s390x/s390-virtio-ccw: Remove superfluous code to set the NIC model
  tests/tcg/s390x: Test VREP
  target/s390x: Use a 16-bit immediate in VREP
  tests/tcg/s390x: Test VSTL
  target/s390x: Fix VSTL with a large length
  target/s390x: Check reserved bits of VFMIN/VFMAX's M5
  s390x: Convert DPRINTF to trace events
  hw: Add compat machines for 8.2

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
10 months agodocs/about/license: Update LICENSE URL
Philippe Mathieu-Daudé [Tue, 22 Aug 2023 12:57:16 +0000 (14:57 +0200)]
docs/about/license: Update LICENSE URL

In early 2021 (see commit 2ad784339e "docs: update README to use
GitLab repo URLs") almost all of the code base was converted to
point to GitLab instead of git.qemu.org. During 2023, git.qemu.org
switched from a git mirror to a http redirect to GitLab (see [1]).

Update the LICENSE URL to match its previous content, displaying
the file raw content similarly to gitweb 'blob_plain' format ([2]).

[1] https://lore.kernel.org/qemu-devel/CABgObfZu3mFc8tM20K-yXdt7F-7eV-uKZN4sKDarSeu7DYoRbA@mail.gmail.com/
[2] https://git-scm.com/docs/gitweb#Documentation/gitweb.txt-blobplain

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230822125716.55295-1-philmd@linaro.org>

10 months agotests/tcg/s390x: Test VSTRS
Ilya Leoshkevich [Fri, 4 Aug 2023 23:03:19 +0000 (01:03 +0200)]
tests/tcg/s390x: Test VSTRS

Add a small test to prevent regressions.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230804233748.218935-4-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agotarget/s390x: Fix the "ignored match" case in VSTRS
Ilya Leoshkevich [Fri, 4 Aug 2023 23:03:18 +0000 (01:03 +0200)]
target/s390x: Fix the "ignored match" case in VSTRS

Currently the emulation of VSTRS recognizes partial matches in presence
of \0 in the haystack, which, according to PoP, is not correct:

    If the ZS flag is one and a zero byte was detected
    in the second operand, then there can not be a
    partial match ...

Add a check for this. While at it, fold a number of explicitly handled
special cases into the generic logic.

Cc: qemu-stable@nongnu.org
Reported-by: Claudio Fontana <cfontana@suse.de>
Closes: https://lists.gnu.org/archive/html/qemu-devel/2023-08/msg00633.html
Fixes: 1d706f314191 ("target/s390x: vxeh2: vector string search")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230804233748.218935-3-iii@linux.ibm.com>
Tested-by: Claudio Fontana <cfontana@suse.de>
Acked-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agolinux-user/elfload: Enable vxe2 on s390x
Ilya Leoshkevich [Fri, 4 Aug 2023 23:03:17 +0000 (01:03 +0200)]
linux-user/elfload: Enable vxe2 on s390x

The vxe2 hwcap is not set for programs running in linux-user, but is
set by a Linux kernel running in softmmu. Add it to the former.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230804233748.218935-2-iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Reviewed-by: Claudio Fontana <cfontana@suse.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agoinclude/hw/virtio/virtio-gpu: Fix virtio-gpu with blob on big endian hosts
Thomas Huth [Tue, 15 Aug 2023 12:20:07 +0000 (14:20 +0200)]
include/hw/virtio/virtio-gpu: Fix virtio-gpu with blob on big endian hosts

Using "-device virtio-gpu,blob=true" currently does not work on big
endian hosts (like s390x). The guest kernel prints an error message
like:

 [drm:virtio_gpu_dequeue_ctrl_func [virtio_gpu]] *ERROR* response 0x1200 (command 0x10c)

and the display stays black. When running QEMU with "-d guest_errors",
it shows an error message like this:

 virtio_gpu_create_mapping_iov: nr_entries is too big (83886080 > 16384)

which indicates that this value has not been properly byte-swapped.
And indeed, the virtio_gpu_create_blob_bswap() function (that should
swap the fields in the related structure) fails to swap some of the
entries. After correctly swapping all missing values here, too, the
virtio-gpu device is now also working with blob=true on s390x hosts.

Fixes: e0933d91b1 ("virtio-gpu: Add virtio_gpu_resource_create_blob")
Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2230469
Message-Id: <20230815122007.928049-1-thuth@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agohw/s390x/s390-virtio-ccw: Remove superfluous code to set the NIC model
Thomas Huth [Fri, 4 Aug 2023 07:35:25 +0000 (09:35 +0200)]
hw/s390x/s390-virtio-ccw: Remove superfluous code to set the NIC model

The check for nd->model being NULL was originally required, but in
commit e11f463295d95aba ("s390x/virtio: use qemu_check_nic_model()")
the corresponding code had been replaced by a call to the function
qemu_check_nic_model() - and this in turn calls qemu_find_nic_model()
which contains the same check for nd->model being NULL again. So we
can remove this from the calling site now.

Message-Id: <20230804073525.11857-1-thuth@redhat.com>
Reviewed-by: Halil Pasic <pasic@linux.ibm.com>
Reviewed-by: Cornelia Huck <cohuck@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agotests/tcg/s390x: Test VREP
Ilya Leoshkevich [Mon, 7 Aug 2023 16:34:32 +0000 (18:34 +0200)]
tests/tcg/s390x: Test VREP

Add a small test to prevent regressions.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230807163459.849766-2-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agotarget/s390x: Use a 16-bit immediate in VREP
Ilya Leoshkevich [Mon, 7 Aug 2023 16:34:31 +0000 (18:34 +0200)]
target/s390x: Use a 16-bit immediate in VREP

Unlike most other instructions that contain an immediate element index,
VREP's one is 16-bit, and not 4-bit. The code uses only 8 bits, so
using, e.g., 0x101 does not lead to a specification exception.

Fix by checking all 16 bits.

Cc: qemu-stable@nongnu.org
Fixes: 28d08731b1d8 ("s390x/tcg: Implement VECTOR REPLICATE")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230807163459.849766-1-iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agotests/tcg/s390x: Test VSTL
Ilya Leoshkevich [Fri, 4 Aug 2023 23:55:34 +0000 (01:55 +0200)]
tests/tcg/s390x: Test VSTL

Add a small test to prevent regressions.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230804235624.263260-2-iii@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agotarget/s390x: Fix VSTL with a large length
Ilya Leoshkevich [Fri, 4 Aug 2023 23:55:33 +0000 (01:55 +0200)]
target/s390x: Fix VSTL with a large length

The length is always truncated to 16 bytes. Do not probe more than
that.

Cc: qemu-stable@nongnu.org
Fixes: 0e0a5b49ad58 ("s390x/tcg: Implement VECTOR STORE WITH LENGTH")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230804235624.263260-1-iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agotarget/s390x: Check reserved bits of VFMIN/VFMAX's M5
Ilya Leoshkevich [Fri, 4 Aug 2023 23:46:10 +0000 (01:46 +0200)]
target/s390x: Check reserved bits of VFMIN/VFMAX's M5

VFMIN and VFMAX should raise a specification exceptions when bits 1-3
of M5 are set.

Cc: qemu-stable@nongnu.org
Fixes: da4807527f3b ("s390x/tcg: Implement VECTOR FP (MAXIMUM|MINIMUM)")
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20230804234621.252522-1-iii@linux.ibm.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agos390x: Convert DPRINTF to trace events
Cédric Le Goater [Fri, 4 Aug 2023 08:04:15 +0000 (10:04 +0200)]
s390x: Convert DPRINTF to trace events

Output message are slightly modified to ease selection with wildcards
and to report extra parameters.

Signed-off-by: Cédric Le Goater <clg@redhat.com>
Message-Id: <20230804080415.56852-1-clg@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Matthew Rosato <mjrosato@linux.ibm.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agohw: Add compat machines for 8.2
Cornelia Huck [Tue, 18 Jul 2023 14:22:35 +0000 (16:22 +0200)]
hw: Add compat machines for 8.2

Add 8.2 machine types for arm/i440fx/m68k/q35/s390x/spapr.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Message-Id: <20230718142235.135319-1-cohuck@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Acked-by: Laurent Vivier <laurent@vivier.eu>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
10 months agotarget/arm: Fix 64-bit SSRA
Richard Henderson [Tue, 22 Aug 2023 16:31:14 +0000 (17:31 +0100)]
target/arm: Fix 64-bit SSRA

Typo applied byte-wise shift instead of double-word shift.

Cc: qemu-stable@nongnu.org
Fixes: 631e565450c ("target/arm: Create gen_gvec_[us]sra")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1737
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230821022025.397682-1-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm: Fix SME ST1Q
Richard Henderson [Tue, 22 Aug 2023 16:31:13 +0000 (17:31 +0100)]
target/arm: Fix SME ST1Q

A typo, noted in the bug report, resulting in an
incorrect write offset.

Cc: qemu-stable@nongnu.org
Fixes: 7390e0e9ab8 ("target/arm: Implement SME LD1, ST1")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1833
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20230818214255.146905-1-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm/helper: Implement CNTHCTL_EL2.CNT[VP]MASK
Jean-Philippe Brucker [Tue, 22 Aug 2023 16:31:13 +0000 (17:31 +0100)]
target/arm/helper: Implement CNTHCTL_EL2.CNT[VP]MASK

When FEAT_RME is implemented, these bits override the value of
CNT[VP]_CTL_EL0.IMASK in Realm and Root state. Move the IRQ state update
into a new gt_update_irq() function and test those bits every time we
recompute the IRQ state.

Since we're removing the IRQ state from some trace events, add a new
trace event for gt_update_irq().

Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Message-id: 20230809123706.1842548-7-jean-philippe@linaro.org
[PMM: only register change hook if not USER_ONLY and if TCG]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm/helper: Check SCR_EL3.{NSE, NS} encoding for AT instructions
Jean-Philippe Brucker [Tue, 22 Aug 2023 16:31:13 +0000 (17:31 +0100)]
target/arm/helper: Check SCR_EL3.{NSE, NS} encoding for AT instructions

The AT instruction is UNDEFINED if the {NSE,NS} configuration is
invalid. Add a function to check this on all AT instructions that apply
to an EL lower than 3.

Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Message-id: 20230809123706.1842548-6-jean-philippe@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm: Pass security space rather than flag for AT instructions
Jean-Philippe Brucker [Tue, 22 Aug 2023 16:31:12 +0000 (17:31 +0100)]
target/arm: Pass security space rather than flag for AT instructions

At the moment we only handle Secure and Nonsecure security spaces for
the AT instructions. Add support for Realm and Root.

For AArch64, arm_security_space() gives the desired space. ARM DDI0487J
says (R_NYXTL):

  If EL3 is implemented, then when an address translation instruction
  that applies to an Exception level lower than EL3 is executed, the
  Effective value of SCR_EL3.{NSE, NS} determines the target Security
  state that the instruction applies to.

For AArch32, some instructions can access NonSecure space from Secure,
so we still need to pass the state explicitly to do_ats_write().

Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230809123706.1842548-5-jean-philippe@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm: Skip granule protection checks for AT instructions
Jean-Philippe Brucker [Tue, 22 Aug 2023 16:31:12 +0000 (17:31 +0100)]
target/arm: Skip granule protection checks for AT instructions

GPC checks are not performed on the output address for AT instructions,
as stated by ARM DDI 0487J in D8.12.2:

  When populating PAR_EL1 with the result of an address translation
  instruction, granule protection checks are not performed on the final
  output address of a successful translation.

Rename get_phys_addr_with_secure(), since it's only used to handle AT
instructions.

Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230809123706.1842548-4-jean-philippe@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm/helper: Fix tlbmask and tlbbits for TLBI VAE2*
Jean-Philippe Brucker [Tue, 22 Aug 2023 16:31:11 +0000 (17:31 +0100)]
target/arm/helper: Fix tlbmask and tlbbits for TLBI VAE2*

When HCR_EL2.E2H is enabled, TLB entries are formed using the EL2&0
translation regime, instead of the EL2 translation regime. The TLB VAE2*
instructions invalidate the regime that corresponds to the current value
of HCR_EL2.E2H.

At the moment we only invalidate the EL2 translation regime. This causes
problems with RMM, which issues TLBI VAE2IS instructions with
HCR_EL2.E2H enabled. Update vae2_tlbmask() to take HCR_EL2.E2H into
account.

Add vae2_tlbbits() as well, since the top-byte-ignore configuration is
different between the EL2&0 and EL2 regime.

Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230809123706.1842548-3-jean-philippe@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm/ptw: Load stage-2 tables from realm physical space
Jean-Philippe Brucker [Tue, 22 Aug 2023 16:31:11 +0000 (17:31 +0100)]
target/arm/ptw: Load stage-2 tables from realm physical space

In realm state, stage-2 translation tables are fetched from the realm
physical address space (R_PGRQD).

Signed-off-by: Jean-Philippe Brucker <jean-philippe@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230809123706.1842548-2-jean-philippe@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agotarget/arm: Adjust PAR_EL1.SH for Device and Normal-NC memory types
Peter Maydell [Tue, 22 Aug 2023 16:31:10 +0000 (17:31 +0100)]
target/arm: Adjust PAR_EL1.SH for Device and Normal-NC memory types

The PAR_EL1.SH field documents that for the cases of:
 * Device memory
 * Normal memory with both Inner and Outer Non-Cacheable
the field should be 0b10 rather than whatever was in the
translation table descriptor field. (In the pseudocode this
is handled by PAREncodeShareability().) Perform this
adjustment when assembling a PAR value.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-16-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Report stage 2 fault level for stage 2 faults on stage 1 ptw
Peter Maydell [Tue, 22 Aug 2023 16:31:10 +0000 (17:31 +0100)]
target/arm/ptw: Report stage 2 fault level for stage 2 faults on stage 1 ptw

When we report faults due to stage 2 faults during a stage 1
page table walk, the 'level' parameter should be the level
of the walk in stage 2 that faulted, not the level of the
walk in stage 1. Correct the reporting of these faults.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-15-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Check for block descriptors at invalid levels
Peter Maydell [Tue, 22 Aug 2023 16:31:10 +0000 (17:31 +0100)]
target/arm/ptw: Check for block descriptors at invalid levels

The architecture doesn't permit block descriptors at any arbitrary
level of the page table walk; it depends on the granule size which
levels are permitted.  We implemented only a partial version of this
check which assumes that block descriptors are valid at all levels
except level 3, which meant that we wouldn't deliver the Translation
fault for all cases of this sort of guest page table error.

Implement the logic corresponding to the pseudocode
AArch64.DecodeDescriptorType() and AArch64.BlockDescSupported().

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-14-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Set attributes correctly for MMU disabled data accesses
Peter Maydell [Tue, 22 Aug 2023 16:31:09 +0000 (17:31 +0100)]
target/arm/ptw: Set attributes correctly for MMU disabled data accesses

When the MMU is disabled, data accesses should be Device nGnRnE,
Outer Shareable, Untagged.  We handle the other cases from
AArch64.S1DisabledOutput() correctly but missed this one.
Device nGnRnE is memattr == 0, so the only part we were missing
was that shareability should be set to 2 for both insn fetches
and data accesses.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-13-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Drop S1Translate::out_secure
Peter Maydell [Tue, 22 Aug 2023 16:31:09 +0000 (17:31 +0100)]
target/arm/ptw: Drop S1Translate::out_secure

We only use S1Translate::out_secure in two places, where we are
setting up MemTxAttrs for a page table load. We can use
arm_space_is_secure(ptw->out_space) instead, which guarantees
that we're setting the MemTxAttrs secure and space fields
consistently, and allows us to drop the out_secure field in
S1Translate entirely.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-12-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Remove S1Translate::in_secure
Peter Maydell [Tue, 22 Aug 2023 16:31:08 +0000 (17:31 +0100)]
target/arm/ptw: Remove S1Translate::in_secure

We no longer look at the in_secure field of the S1Translate struct
anyway, so we can remove it and all the code which sets it.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-11-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Remove last uses of ptw->in_secure
Peter Maydell [Tue, 22 Aug 2023 16:31:08 +0000 (17:31 +0100)]
target/arm/ptw: Remove last uses of ptw->in_secure

Replace the last uses of ptw->in_secure with appropriate
checks on ptw->in_space.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-10-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Only fold in NSTable bit effects in Secure state
Peter Maydell [Tue, 22 Aug 2023 16:31:08 +0000 (17:31 +0100)]
target/arm/ptw: Only fold in NSTable bit effects in Secure state

When we do a translation in Secure state, the NSTable bits in table
descriptors may downgrade us to NonSecure; we update ptw->in_secure
and ptw->in_space accordingly.  We guard that check correctly with a
conditional that means it's only applied for Secure stage 1
translations.  However, later on in get_phys_addr_lpae() we fold the
effects of the NSTable bits into the final descriptor attributes
bits, and there we do it unconditionally regardless of the CPU state.
That means that in Realm state (where in_secure is false) we will set
bit 5 in attrs, and later use it to decide to output to non-secure
space.

We don't in fact need to do this folding in at all any more (since
commit 2f1ff4e7b9f30c): if an NSTable bit was set then we have
already set ptw->in_space to ARMSS_NonSecure, and in that situation
we don't look at attrs bit 5.  The only thing we still need to deal
with is the real NS bit in the final descriptor word, so we can just
drop the code that ORed in the NSTable bit.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-9-peter.maydell@linaro.org

10 months agotarget/arm: Pass an ARMSecuritySpace to arm_is_el2_enabled_secstate()
Peter Maydell [Tue, 22 Aug 2023 16:31:07 +0000 (17:31 +0100)]
target/arm: Pass an ARMSecuritySpace to arm_is_el2_enabled_secstate()

Pass an ARMSecuritySpace instead of a bool secure to
arm_is_el2_enabled_secstate(). This doesn't change behaviour.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-8-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Pass an ARMSecuritySpace to arm_hcr_el2_eff_secstate()
Peter Maydell [Tue, 22 Aug 2023 16:31:07 +0000 (17:31 +0100)]
target/arm/ptw: Pass an ARMSecuritySpace to arm_hcr_el2_eff_secstate()

arm_hcr_el2_eff_secstate() takes a bool secure, which it uses to
determine whether EL2 is enabled in the current security state.
With the advent of FEAT_RME this is no longer sufficient, because
EL2 can be enabled for Secure state but not for Root, and both
of those will pass 'secure == true' in the callsites in ptw.c.

As it happens in all of our callsites in ptw.c we either avoid making
the call or else avoid using the returned value if we're doing a
translation for Root, so this is not a behaviour change even if the
experimental FEAT_RME is enabled.  But it is less confusing in the
ptw.c code if we avoid the use of a bool secure that duplicates some
of the information in the ArmSecuritySpace argument.

Make arm_hcr_el2_eff_secstate() take an ARMSecuritySpace argument
instead. Because we always want to know the HCR_EL2 for the
security state defined by the current effective value of
SCR_EL3.{NSE,NS}, it makes no sense to pass ARMSS_Root here,
and we assert that callers don't do that.

To avoid the assert(), we thus push the call to
arm_hcr_el2_eff_secstate() down into the cases in
regime_translation_disabled() that need it, rather than calling the
function and ignoring the result for the Root space translations.
All other calls to this function in ptw.c are already in places
where we have confirmed that the mmu_idx is a stage 2 translation
or that the regime EL is not 3.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-7-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Pass ARMSecurityState to regime_translation_disabled()
Peter Maydell [Tue, 22 Aug 2023 16:31:06 +0000 (17:31 +0100)]
target/arm/ptw: Pass ARMSecurityState to regime_translation_disabled()

Plumb the ARMSecurityState through to regime_translation_disabled()
rather than just a bool is_secure.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-6-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Pass ptw into get_phys_addr_pmsa*() and get_phys_addr_disabled()
Peter Maydell [Tue, 22 Aug 2023 16:31:06 +0000 (17:31 +0100)]
target/arm/ptw: Pass ptw into get_phys_addr_pmsa*() and get_phys_addr_disabled()

In commit 6d2654ffacea813916176 we created the S1Translate struct and
used it to plumb through various arguments that we were previously
passing one-at-a-time to get_phys_addr_v5(), get_phys_addr_v6(), and
get_phys_addr_lpae().  Extend that pattern to get_phys_addr_pmsav5(),
get_phys_addr_pmsav7(), get_phys_addr_pmsav8() and
get_phys_addr_disabled(), so that all the get_phys_addr_* functions
we call from get_phys_addr_nogpc() take the S1Translate struct rather
than the mmu_idx and is_secure bool.

(This refactoring is a prelude to having the called functions look
at ptw->is_space rather than using an is_secure boolean.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-5-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Set s1ns bit in fault info more consistently
Peter Maydell [Tue, 22 Aug 2023 16:31:05 +0000 (17:31 +0100)]
target/arm/ptw: Set s1ns bit in fault info more consistently

The s1ns bit in ARMMMUFaultInfo is documented as "true if
we faulted on a non-secure IPA while in secure state". Both the
places which look at this bit only do so after having confirmed
that this is a stage 2 fault and we're dealing with Secure EL2,
which leaves the ptw.c code free to set the bit to any random
value in the other cases.

Instead of taking advantage of that freedom, consistently
make the bit be set to false for the "not a stage 2 fault
for Secure EL2" cases. This removes some cases where we
were using an 'is_secure' boolean and leaving the reader
guessing about whether that was the right thing for Realm
and Root cases.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-4-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Don't report GPC faults on stage 1 ptw as stage2 faults
Peter Maydell [Tue, 22 Aug 2023 16:31:05 +0000 (17:31 +0100)]
target/arm/ptw: Don't report GPC faults on stage 1 ptw as stage2 faults

In S1_ptw_translate() we set up the ARMMMUFaultInfo if the attempt to
translate the page descriptor address into a physical address fails.
This used to only be possible if we are doing a stage 2 ptw for that
descriptor address, and so the code always sets fi->stage2 and
fi->s1ptw to true.  However, with FEAT_RME it is also possible for
the lookup of the page descriptor address to fail because of a
Granule Protection Check fault.  These should not be reported as
stage 2, otherwise arm_deliver_fault() will incorrectly set
HPFAR_EL2.  Similarly the s1ptw bit should only be set for stage 2
faults on stage 1 translation table walks, i.e.  not for GPC faults.

Add a comment to the the other place where we might detect a
stage2-fault-on-stage-1-ptw, in arm_casq_ptw(), noting why we know in
that case that it must really be a stage 2 fault and not a GPC fault.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-3-peter.maydell@linaro.org

10 months agotarget/arm/ptw: Don't set fi->s1ptw for UnsuppAtomicUpdate fault
Peter Maydell [Tue, 22 Aug 2023 16:31:05 +0000 (17:31 +0100)]
target/arm/ptw: Don't set fi->s1ptw for UnsuppAtomicUpdate fault

For an Unsupported Atomic Update fault where the stage 1 translation
table descriptor update can't be done because it's to an unsupported
memory type, this is a stage 1 abort (per the Arm ARM R_VSXXT).  This
means we should not set fi->s1ptw, because this will cause the code
in the get_phys_addr_lpae() error-exit path to mark it as stage 2.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230807141514.19075-2-peter.maydell@linaro.org

10 months agoaccel/kvm: Make kvm_dirty_ring_reaper_init() void
Akihiko Odaki [Tue, 22 Aug 2023 16:31:04 +0000 (17:31 +0100)]
accel/kvm: Make kvm_dirty_ring_reaper_init() void

The returned value was always zero and had no meaning.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230727073134.134102-7-akihiko.odaki@daynix.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agoaccel/kvm: Free as when an error occurred
Akihiko Odaki [Tue, 22 Aug 2023 16:31:04 +0000 (17:31 +0100)]
accel/kvm: Free as when an error occurred

An error may occur after s->as is allocated, for example if the
KVM_CREATE_VM ioctl call fails.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230727073134.134102-6-akihiko.odaki@daynix.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: tweaked commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agoaccel/kvm: Use negative KVM type for error propagation
Akihiko Odaki [Tue, 22 Aug 2023 16:31:03 +0000 (17:31 +0100)]
accel/kvm: Use negative KVM type for error propagation

On MIPS, kvm_arch_get_default_type() returns a negative value when an
error occurred so handle the case. Also, let other machines return
negative values when errors occur and declare returning a negative
value as the correct way to propagate an error that happened when
determining KVM type.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230727073134.134102-5-akihiko.odaki@daynix.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agomips: Report an error when KVM_VM_MIPS_VZ is unavailable
Akihiko Odaki [Tue, 22 Aug 2023 16:31:03 +0000 (17:31 +0100)]
mips: Report an error when KVM_VM_MIPS_VZ is unavailable

On MIPS, QEMU requires KVM_VM_MIPS_VZ type for KVM. Report an error in
such a case as other architectures do when an error occurred during KVM
type decision.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230727073134.134102-4-akihiko.odaki@daynix.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agoaccel/kvm: Specify default IPA size for arm64
Akihiko Odaki [Tue, 22 Aug 2023 16:31:02 +0000 (17:31 +0100)]
accel/kvm: Specify default IPA size for arm64

Before this change, the default KVM type, which is used for non-virt
machine models, was 0.

The kernel documentation says:
> On arm64, the physical address size for a VM (IPA Size limit) is
> limited to 40bits by default. The limit can be configured if the host
> supports the extension KVM_CAP_ARM_VM_IPA_SIZE. When supported, use
> KVM_VM_TYPE_ARM_IPA_SIZE(IPA_Bits) to set the size in the machine type
> identifier, where IPA_Bits is the maximum width of any physical
> address used by the VM. The IPA_Bits is encoded in bits[7-0] of the
> machine type identifier.
>
> e.g, to configure a guest to use 48bit physical address size::
>
>     vm_fd = ioctl(dev_fd, KVM_CREATE_VM, KVM_VM_TYPE_ARM_IPA_SIZE(48));
>
> The requested size (IPA_Bits) must be:
>
>  ==   =========================================================
>   0   Implies default size, 40bits (for backward compatibility)
>   N   Implies N bits, where N is a positive integer such that,
>       32 <= N <= Host_IPA_Limit
>  ==   =========================================================

> Host_IPA_Limit is the maximum possible value for IPA_Bits on the host
> and is dependent on the CPU capability and the kernel configuration.
> The limit can be retrieved using KVM_CAP_ARM_VM_IPA_SIZE of the
> KVM_CHECK_EXTENSION ioctl() at run-time.
>
> Creation of the VM will fail if the requested IPA size (whether it is
> implicit or explicit) is unsupported on the host.
https://docs.kernel.org/virt/kvm/api.html#kvm-create-vm

So if Host_IPA_Limit < 40, specifying 0 as the type will fail. This
actually confused libvirt, which uses "none" machine model to probe the
KVM availability, on M2 MacBook Air.

Fix this by using Host_IPA_Limit as the default type when
KVM_CAP_ARM_VM_IPA_SIZE is available.

Cc: qemu-stable@nongnu.org
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230727073134.134102-3-akihiko.odaki@daynix.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agokvm: Introduce kvm_arch_get_default_type hook
Akihiko Odaki [Tue, 22 Aug 2023 16:31:02 +0000 (17:31 +0100)]
kvm: Introduce kvm_arch_get_default_type hook

kvm_arch_get_default_type() returns the default KVM type. This hook is
particularly useful to derive a KVM type that is valid for "none"
machine model, which is used by libvirt to probe the availability of
KVM.

For MIPS, the existing mips_kvm_type() is reused. This function ensures
the availability of VZ which is mandatory to use KVM on the current
QEMU.

Cc: qemu-stable@nongnu.org
Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230727073134.134102-2-akihiko.odaki@daynix.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: added doc comment for new function]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
10 months agoqtest: microbit-test: add tests for nRF51 DETECT
Chris Laplante [Tue, 22 Aug 2023 16:31:02 +0000 (17:31 +0100)]
qtest: microbit-test: add tests for nRF51 DETECT

Exercise the DETECT mechanism of the GPIO peripheral.

Signed-off-by: Chris Laplante <chris@laplante.io>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230728160324.1159090-7-chris@laplante.io
[PMM: fixed coding style nits]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agoqtest: irq_intercept_[out/in]: return FAIL if no intercepts are installed
Chris Laplante [Tue, 22 Aug 2023 16:31:01 +0000 (17:31 +0100)]
qtest: irq_intercept_[out/in]: return FAIL if no intercepts are installed

This is much better than just silently failing with OK.

Signed-off-by: Chris Laplante <chris@laplante.io>
Message-id: 20230728160324.1159090-6-chris@laplante.io
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
10 months agoqtest: bail from irq_intercept_in if name is specified
Chris Laplante [Tue, 22 Aug 2023 16:31:01 +0000 (17:31 +0100)]
qtest: bail from irq_intercept_in if name is specified

Named interception of in-GPIOs is not supported yet.

Signed-off-by: Chris Laplante <chris@laplante.io>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20230728160324.1159090-5-chris@laplante.io
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>