crypto: Introduce x509 utils

An utility function for getting fingerprint from X.509 certificate
has been introduced. Implementation only provided using gnutls.

Signed-off-by: Dorjoy Chowdhury <dorjoychy111@gmail.com>
[DB: fixed missing gnutls_x509_crt_deinit in success path]
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: Support SHA384 hash when using glib

QEMU requires minimum glib version 2.66.0 as per the root meson.build
file and per glib documentation[1] G_CHECKSUM_SHA384 is available since
2.51.

[1] https://docs.gtk.org/glib/enum.ChecksumType.html

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Dorjoy Chowdhury <dorjoychy111@gmail.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: Define macros for hash algorithm digest lengths

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Dorjoy Chowdhury <dorjoychy111@gmail.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: use consistent error reporting pattern for unsupported cipher modes

Not all paths in qcrypto_cipher_ctx_new() were correctly distinguishing
between valid user input for cipher mode (which should report a user
facing error), vs program logic errors (which should assert).

Reported-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: avoid leak of ctx when bad cipher mode is given

Fixes: Coverity CID 1546884
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

tests/unit: build pbkdf test on macOS

Add CONFIG_DARWIN to the pbkdf test build condition, since we have a way
to measure CPU time on this platform since commit bf98afc75efedf1.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

tests/unit: always build the pbkdf crypto unit test

The meson rules were excluding the pbkdf crypto test when gnutls was the
crypto backend. It was then excluded again in #if statements in the test
file.

Rather than update these conditions, remove them all, and use the result
of the qcrypto_pbkdf_supports() function to determine whether to skip
test registration.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: check gnutls & gcrypt support the requested pbkdf hash

Both gnutls and gcrypt can be configured to exclude support for certain
algorithms via a runtime check against system crypto policies. Thus it
is not sufficient to have a compile time test for hash support in their
pbkdf implementations.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

crypto: run qcrypto_pbkdf2_count_iters in a new thread

CPU time accounting in the kernel has been demonstrated to have a
sawtooth pattern[1][2]. This can cause the getrusage system call to
not be as accurate as we are expecting, which can cause this calculation
to stall.

The kernel discussions shows that this inaccuracy happens when CPU time
gets big enough, so this patch changes qcrypto_pbkdf2_count_iters to run
in a fresh thread to avoid this inaccuracy. It also adds a sanity check
to fail the process if CPU time is not accounted.

[1] https://lore.kernel.org/lkml/159231011694.16989.16351419333851309713.tip-bot2@tip-bot2/
[2] https://lore.kernel.org/lkml/20221226031010.4079885-1-maxing.lan@bytedance.com/t/#m1c7f2fdc0ea742776a70fd1aa2a2e414c437f534

Resolves: #2398
Signed-off-by: Tiago Pasqualini <tiago.pasqualini@canonical.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

iotests: fix expected output from gnutls

Error reporting from gnutls was improved by:

  commit 57941c9c86357a6a642f9ee3279d881df4043b6d
  Author: Daniel P. Berrangé <berrange@redhat.com>
  Date:   Fri Mar 15 14:07:58 2024 +0000

    crypto: push error reporting into TLS session I/O APIs

This has the effect of changing the output from one of the NBD
tests.

Reported-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>

Merge tag 'pull-request-2024-09-08' of https://gitlab.com/huth/qemu into staging

* Fix Coverity issues in mcf5208evb and nubus machines
* Add URLs for mcf5208evb datasheets

# -----BEGIN PGP SIGNATURE-----
#
# iQJHBAABCAAxFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmbdl4gTHGh1dGhAdHV4
# ZmFtaWx5Lm9yZwAKCRAu2dd0/nAttaRsD/9qqE75Glk1/BYtbI5N4E1Q30vC9lTl
# /R+gOJod44E6/xeOgvvpliCTzsnrHM3sfk2n/EBoEUfQ3Ci3UqZe+wmBgsj+Wv4H
# IIM5Aviq05dwj0B2mgqFFpXLV/ilVTpl3vpkBL82hLmZH+5xre4P/qgn4JwwAAAi
# 7quHBcYCwTALB0bizaGj7djwABF7cThkI5z0qE46spju4N6YflfYOE3oRmcqW9Di
# gM5G5CcbJkUmAxeRpBaS+fD/pXjFgYSqZsfjHwaRBQLG+LLdy+EmNiS1UUBvZeE2
# J5w82qaFly+UMXcYEuOxApVP3Mf1M2sG4o2qoQaqGtvqjBvaXVwbUFcVEgYePZEf
# oRweSPjZTarsSCPx/11UOKPz9j3wh5wFME+j8I2TdOl5IZkYp1pL9hpls/Jyz+Ii
# /0JfWna8MlDEpoo/e0sNA+IEcg340BiNqBFMgWkl1TvzKyOJcSorlbbpq1rE0kr0
# Y2+g/rr9hEVh1E+Yla6Me7GOcNUvT77yhrpTdagX6NG7qZfJ4tB0/hADeZ6ic+sC
# Xf48RtD3uzAknBy9awL/hIJWx2sKrS36E01EZj8KvokQzhKawfU16P0++l1DMq9J
# 9npALGPIhpYm6JNTr4FZYd+88+ZEQEKkAjJ+a6t7/+gY5PINfRICpGMZtGB6LRvu
# NjVGIGuKHBYTPw==
# =3Pn7
# -----END PGP SIGNATURE-----
# gpg: Signature made Sun 08 Sep 2024 13:24:40 BST
# gpg:                using RSA key 27B88847EEE0250118F3EAB92ED9D774FE702DB5
# gpg:                issuer "huth@tuxfamily.org"
# 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-2024-09-08' of https://gitlab.com/huth/qemu:
  hw/nubus/nubus-device: Range check 'slot' property
  hw/m68k/mcf5208: Add URLs for datasheets
  hw/m68k/mcf5208: Avoid shifting off end of integer

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

hw/nubus/nubus-device: Range check 'slot' property

The TYPE_NUBUS_DEVICE class lets the user specify the nubus slot
using an int32 "slot" QOM property.  Its realize method doesn't do
any range checking on this value, which Coverity notices by way of
the possibility that 'nd->slot * NUBUS_SUPER_SLOT_SIZE' might
overflow the 32-bit arithmetic it is using.

Constrain the slot value to be less than NUBUS_SLOT_NB (16).

Resolves: Coverity CID 1464070
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-ID: <20240830173452.2086140-4-peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <huth@tuxfamily.org>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Thomas Huth <huth@tuxfamily.org>

hw/m68k/mcf5208: Add URLs for datasheets

The datasheets for the SoC and board we model here are still
available from the NXP website; add their URLs and titles for
future reference.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <huth@tuxfamily.org>
Message-ID: <20240830173452.2086140-3-peter.maydell@linaro.org>
Signed-off-by: Thomas Huth <huth@tuxfamily.org>

hw/m68k/mcf5208: Avoid shifting off end of integer

In m5208_sys_read(), we have a loop of n from 0 to 31, and we
calculate (2u << n).  For the n == 31 iteration this will shift off
the top of the unsigned 32 bit integer.

This is harmless, because we're going to stop the loop with n == 31
anyway, but we can avoid the error by using 64-bit arithmetic here.

(The SDCS0 register is documented at
https://www.nxp.com/docs/en/reference-manual/MCF5208RM.pdf
section 18.4.5; we want the lower 5 bits to indicate the
RAM size, where 31 == 4GB, 30 == 2GB, and so on down.
As it happens, the layout of the mcf5208evb board memory map
means it doesn't make sense to have more than 1GB of RAM
in any case.)

Resolves: Coverity CID 1547727
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Thomas Huth <huth@tuxfamily.org>
Message-ID: <20240830173452.2086140-2-peter.maydell@linaro.org>
Signed-off-by: Thomas Huth <huth@tuxfamily.org>

Merge tag 'pull-ufs-20240906' of https://gitlab.com/jeuk20.kim/qemu into staging

ufs queue

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCgAdFiEEUBfYMVl8eKPZB+73EuIgTA5dtgIFAmba268ACgkQEuIgTA5d
# tgLndA//T7zvQboCWE2Aw+al4/cJmpfc/BP7pKrCvrwskhAo2H5JvbJ20WZ+/I6E
# sLqXjsAQ4qPWUNi46aty/tCCmFWatBRKIyWOg1E8w1N1PCqM/aKFElENgi28iclj
# 3TlIU+++a6VJXMtKKqGb/d6cxXM9QtRgkfpGEnVTCD4sRX25WuWcWu+hwCipgzsr
# dul1Ez+mp62SfHN2QLPUd+Ft0SvyxybDA65JP9fCEJPJ+2dtLWPN9XGY+6PzW4dT
# UEfUEV2V5k3w/QHTR8yG2i5s56wWVUhtQEhazbkj1VqgUSJ8PvIvBLhQpi4Gd51G
# 62/xHJaHXPxgVrVE6Or/5QF9npo1moG7UrLgP+FYX5kto280wEyh3KxNhlan+lmI
# IGo7V3Xv6UgGudJ/ZjR4dw24atFDcaqmdnWAOOp7mwxUIAq/5xLeDw2fzvuUw78a
# cc732SF4XwTJfXwgiXkJXa/Si248fDtecvlD+lQ9wezSIJZq8Ojpe9uFREA4jPVY
# jfgXEoopvam4w4ZKFRg93/0QErgwsYaJKKIKD8wZ99pi8/zrWlq5W1ujefQuvujt
# FL4IbF/8g6i22fZgBr8AlkRn3epxx4oRGV+Rr8OXoFYjNR/E7rjKZYfbsJx/gDbO
# zEMzSImQ48Tlxl9vkIx5kkUDxw3d7MDvrXEORcTMSW53CUDXJyA=
# =U151
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 06 Sep 2024 11:38:39 BST
# gpg:                using RSA key 5017D831597C78A3D907EEF712E2204C0E5DB602
# gpg: Good signature from "Jeuk Kim <jeuk20.kim@samsung.com>" [unknown]
# gpg:                 aka "Jeuk Kim <jeuk20.kim@gmail.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: 5017 D831 597C 78A3 D907  EEF7 12E2 204C 0E5D B602

* tag 'pull-ufs-20240906' of https://gitlab.com/jeuk20.kim/qemu:
  hw/ufs: ufs descriptor read test implemented
  hw/ufs: ufs attribute read/write test implemented
  hw/ufs: ufs flag read/write test implemented
  hw/ufs: minor bug fixes related to ufs-test
  hw/ufs: add basic info of query response upiu

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

Merge tag 'pull-target-arm-20240905' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * Implement FEAT_EBF16 emulation
 * accel/tcg: Remove dead code from rr_cpu_thread_fn()
 * hw: add compat machines for 9.2
 * virt: default to two-stage SMMU from virt-9.2
 * sbsa-ref: use two-stage SMMU
 * hw: Various minor memory leak fixes
 * target/arm: Correct names of VFP VFNMA and VFNMS insns
 * hw/arm/xilinx_zynq: Enable Security Extensions
 * hw/arm/boot: Report error msg if loading elf/dtb failed

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmbZqzEZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3lJ7D/9s/ZTkiCj/z+caHotwNJVt
# ECgEEVinitwZxSMINZd1f6bxTY8hYVjMewj6A6RvHtMJMr7SUOmL8wi0YlbhTm44
# jb8dZVf3pzPaZ399jxOeGnFipGyKmK0XM5rKc7CP6yJUS3B9RkUbLEHng8Q0ZBtl
# cnZqI12jJBdtHU8D4JIvBgM2N2ay4bKY8EQEPCv4S7ZTKawWcKgSR5pMd2TBIqIT
# 0gaDL3eOgCt2XWIrMzRjvaJK70obN/+n+vZQskJ/sIDsw+Kz8sZGlivdBXLRmQ+A
# OUgtdyZoD42Q8KtwM0bjoaoxz6VMNPJp5khB45EPjVgWyeyJ0L6ZcWCX7nT4hZsi
# 1C0NJaJU6HQbfsPiMIGxgHYJCbQue/mVBE02MPhmN8fZlsTRKWT9Miu67S0PI5Ib
# ZWo88Ew1coucBm25K2NWdoR3dCP8EFnxqL556L8M4iDWYQ/djf8cpFAN9QJBFrNw
# CaXS+vxIFUjZ6TSjf8gOYPAONmAg5DsCucgyO4MBKnvlY5h2J+GTq/FC+kWzL9jE
# UfhqOWSP34ol2lg319zOtKg4Ga+GOivo2DmgWQhDwZ2rmRR+xgN8rkQjpJKIT5Zj
# Ji+ucJrghBZ0sN622QYG0u0Ap9Jy4KCOxcFfS1b4gNhmMDWg27Tx9tIguXmjOE3M
# aAs4wmm4Nz4kpsf1KkB11Q==
# =gZuf
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 05 Sep 2024 13:59:29 BST
# gpg:                using RSA key E1A5C593CD419DE28E8315CF3C2525ED14360CDE
# gpg:                issuer "peter.maydell@linaro.org"
# gpg: Good signature from "Peter Maydell <peter.maydell@linaro.org>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@gmail.com>" [ultimate]
# gpg:                 aka "Peter Maydell <pmaydell@chiark.greenend.org.uk>" [ultimate]
# gpg:                 aka "Peter Maydell <peter@archaic.org.uk>" [ultimate]
# Primary key fingerprint: E1A5 C593 CD41 9DE2 8E83  15CF 3C25 25ED 1436 0CDE

* tag 'pull-target-arm-20240905' of https://git.linaro.org/people/pmaydell/qemu-arm: (25 commits)
  platform-bus: fix refcount leak
  hw/arm/boot: Explain why load_elf_hdr() error is ignored
  hw/arm/boot: Report error msg if loading elf/dtb failed
  hw/arm/xilinx_zynq: Enable Security Extensions
  target/arm: Correct names of VFP VFNMA and VFNMS insns
  hw/arm/sbsa-ref: Don't leak string in sbsa_fdt_add_gic_node()
  hm/nvram/xlnx-versal-efuse-ctrl: Call register_finalize_block
  hw/misc/xlnx-versal-trng: Call register_finalize_block
  hw/nvram/xlnx-zynqmp-efuse: Call register_finalize_block
  hw/nvram/xlnx-bbram: Call register_finalize_block
  hw/misc/xlnx-versal-trng: Free s->prng in finalize, not unrealize
  hw/misc/xlnx-versal-cfu: destroy fifo in finalize
  hw/arm/sbsa-ref: Use two-stage SMMU
  hw/arm/virt: Default to two-stage SMMU from virt-9.2
  hw/arm/smmuv3: Update comment documenting "stage" property
  hw: add compat machines for 9.2
  accel/tcg: Remove dead code from rr_cpu_thread_fn()
  target/arm: Enable FEAT_EBF16 in the "max" CPU
  target/arm: Implement FPCR.EBF=1 semantics for bfdotadd()
  target/arm: Prepare bfdotadd() callers for FEAT_EBF support
  ...

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

tests/qtest: Bump timeout on ahci-test

On my OpenBSD VM test system, the ahci-test sometimes hits its 60 second
timeout. It has 75 subtests and allowing at least two seconds per
subtest seems reasonable. Bump it to 150s.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 20240905165554.320577-3-peter.maydell@linaro.org

tests/qtest: Add missing qtest_quit() to stm32 tests

In the dm163-test and stm32l4x5_usart-test, a couple of subtests are
missing the qtest_quit() call.  The effect of this is that on hosts
other than Linux and FreeBSD the test will timeout after executing
all the tests:

242/845 qemu:qtest+qtest-arm / qtest-arm/dm163-test                               TIMEOUT         60.04s   3 subtests passed
100/845 qemu:qtest+qtest-arm / qtest-arm/stm32l4x5_usart-test                     TIMEOUT        600.02s   5 subtests passed

This happens because the qemu-system-arm binary which the test
starts does not exit, and because it shares the stdout with the
test binary, the overall meson test harness thinks the test is
still running. On Linux and FreeBSD we have an extra safety net
set up in qtest_spawn_qemu() which kills off any QEMU binary that
ends up without a parent. This is intended for the case where
QEMU crashed and didn't respond to a SIGTERM or polite request
to quit, but it also sidestepped the problem in this case.
However, OpenBSD doesn't have a PDEATHSIG equivalent, so we
see the timeouts when running a 'make vm-build-openbsd' run.

Add the missing qtest_quit() calls.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 20240905165554.320577-2-peter.maydell@linaro.org

Merge tag 'migration-20240904-pull-request' of https://gitlab.com/farosas/qemu into staging

Migration pull request

- Steve's cleanup of unused variable
- Peter Maydell's fixes for several leaks in migration-test
- Fabiano's flexibilization of multifd data structures for device
  state migration
- Arman Nabiev's fix for ppc e500 migration
- Thomas' fix for migration-test vs. --without-default-devices

# -----BEGIN PGP SIGNATURE-----
#
# iQJEBAABCAAuFiEEqhtIsKIjJqWkw2TPx5jcdBvsMZ0FAmbYVXwQHGZhcm9zYXNA
# c3VzZS5kZQAKCRDHmNx0G+wxnRucEAC1vo046UGdUmbb4PaF5vKAg97io6RB2nrH
# HMz56Yc0AcAKRUGwe2Z80e2jY8B6zi8Ha8b9l7cVsej095eGCF+tINIL4wRX4lHm
# alDY/LkhuqjE5g5c/DaeTztyBOFLvdWHPU5eJyDOC9r7kSlnUcL1gAslH23b8uL0
# xvhPVKaTWjGIzNL1q/XfBr1WgRGqfD6dYb32HJDTq85yOnUT5sEr55aoEEu0euKh
# MYbXPmi5AMbrp8nP21kzUopX8iYERRdoKwhF0ZssciGi/qJVevH70tNdbDEQSxyp
# +vtP54TnL3LrzD4uY5Snng9zT9h0QrZujY79OEcxu20U0s29OQaudWkIjp7yLLUv
# UnPZHS+bIyaS53DdpV94GKGGBX1wrjGC/sn8eGYzmb2yMlMjLTBoE8L5r9cadshX
# XTeF4MtKGqaS3xDM2fIgACHHFl6qr/l0nENspv0raFzpf9Jx/WbpekghvTuWN6/B
# pZHnoOTNiAqXS/Rnyy829vsQ0Pw4hi6wx79Z73RP+35ubZTgTmOsQx9f2FjuEh6k
# JS+q9k4VJ+nntUWsYn4GS1Jlt+FXJ2hfzNj1NNFN4xLT1oioc6pCHsQyV7SBArB1
# ml2zYyfKCTC3riIRhcv/ew6OcKbhHcPFOpd/v0y40LO3mx8S0LZnUWXkcrl3XIZS
# Mj5CBdlFgA==
# =SRN4
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 04 Sep 2024 13:41:32 BST
# gpg:                using RSA key AA1B48B0A22326A5A4C364CFC798DC741BEC319D
# gpg:                issuer "farosas@suse.de"
# gpg: Good signature from "Fabiano Rosas <farosas@suse.de>" [unknown]
# gpg:                 aka "Fabiano Almeida Rosas <fabiano.rosas@suse.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: AA1B 48B0 A223 26A5 A4C3  64CF C798 DC74 1BEC 319D

* tag 'migration-20240904-pull-request' of https://gitlab.com/farosas/qemu: (34 commits)
  tests/qtest/migration: Add a check for the availability of the "pc" machine
  target/ppc: Fix migration of CPUs with TLB_EMB TLB type
  migration/multifd: Add documentation for multifd methods
  migration/multifd: Add a couple of asserts for p->iov
  migration/multifd: Fix p->iov leak in multifd-uadk.c
  migration/multifd: Stop changing the packet on recv side
  migration/multifd: Make MultiFDMethods const
  migration/multifd: Move nocomp code into multifd-nocomp.c
  migration/multifd: Register nocomp ops dynamically
  migration/multifd: Standardize on multifd ops names
  migration/multifd: Allow multifd sync without flush
  migration/multifd: Replace multifd_send_state->pages with client data
  migration/multifd: Don't send ram data during SYNC
  migration/multifd: Isolate ram pages packet data
  migration/multifd: Remove total pages tracing
  migration/multifd: Move pages accounting into multifd_send_zero_page_detect()
  migration/multifd: Replace p->pages with an union pointer
  migration/multifd: Make MultiFDPages_t:offset a flexible array member
  migration/multifd: Introduce MultiFDSendData
  migration/multifd: Pass in MultiFDPages_t to file_write_ramblock_iov
  ...

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

hw/ufs: ufs descriptor read test implemented

New test function "ufstest_query_desc_request" added, which can check one's
virtual UFS device can properly read and its descriptor data.
(Writing descriptors are not implemented yet.)
The testcases attempt to read all kinds of descriptors at least once,
except for configuration descriptors (which are not implemented yet.)
There are some testcases that are intended to make an error caused by
an invalid index value or an invalid selector value.

Signed-off-by: Yoochan Jeong <yc01.jeong@samsung.com>
Reviewed-by: Jeuk Kim <jeuk20.kim@samsung.com>
Signed-off-by: Jeuk Kim <jeuk20.kim@samsung.com>

hw/ufs: ufs attribute read/write test implemented

New test function "ufstest_query_attr_request" added, which can check one's
virtual UFS device can properly read and write its attribute data.
It tests if reading and writing attributes work properly. There are
some testcases that are intended to make an error caused by writing an
invalid value, allocating an invalid selector and permission issues.

Signed-off-by: Yoochan Jeong <yc01.jeong@samsung.com>
Reviewed-by: Jeuk Kim <jeuk20.kim@samsung.com>
Signed-off-by: Jeuk Kim <jeuk20.kim@samsung.com>

hw/ufs: ufs flag read/write test implemented

New test function "ufstest_flag_request" added, which can check one's
virtual UFS device can properly read and write its flag data. It tests
if reading, setting, clearing and toggling flags work properly. There
are some testcases that are intended to make an error caused by
permission issues.

Signed-off-by: Yoochan Jeong <yc01.jeong@samsung.com>
Reviewed-by: Jeuk Kim <jeuk20.kim@samsung.com>
Signed-off-by: Jeuk Kim <jeuk20.kim@samsung.com>

hw/ufs: minor bug fixes related to ufs-test

Minor bugs and errors related to ufs-test are resolved. Some
permissions and code implementations that are not synchronized
with the ufs spec are edited.

Signed-off-by: Yoochan Jeong <yc01.jeong@samsung.com>
Reviewed-by: Jeuk Kim <jeuk20.kim@samsung.com>
Signed-off-by: Jeuk Kim <jeuk20.kim@samsung.com>

hw/ufs: add basic info of query response upiu

Modify to fill the opcode, idn, index, selector information of
all Query Response UPIU. because attr and flag operation of query
response upiu need these information too.

Signed-off-by: KyoungrulKim <k831.kim@samsung.com>
Reviewed-by: Minwoo Im <minwoo.im@samsung.com>
Reviewed-by: Jeuk Kim <jeuk20.kim@samsung.com>
Signed-off-by: Jeuk Kim <jeuk20.kim@samsung.com>

Merge tag 'pull-request-2024-09-04' of https://gitlab.com/thuth/qemu into staging

* Bump Avocado to version 103
* Introduce new functional test framework for Python-based tests
* Convert many Avocado tests to the new functional test framework

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmbYOEsRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbUDAA/+Kdlak/nCrK5gXDyDasfy86IxgMD0QlDR
# U0MOpQyfXbM2EJjwCUhmgo8pui8qV23dKzfCwbDmkjB7mJ+yKi2ZdiFEp6onq/ke
# aAdaaZwENtWcFglRD80TOSQX6oyeNmE/PuvJGG0BfwWXyyhaEa6kCdytEPORipQs
# lZ+ZndHgXtcM3roXtgI3kp2V1nY5LLCJ044UrasKRq2xWfD/Ken90uWP5/nMLV7f
# 7YLRUIb0sgV7IdjZiT1UkXJZRB7MatV7+OsojYbG8BPbQEvXqpryXMIeygHVR9a0
# yxNDUpTZR6JoS1IaLKkHh1mTM+L1JpFltKadKkXa0zqJHHSur7Tp0xVO/GeqCek4
# 9N8K4zw2CoO/AKmN8JjW5i4GnMrFMdcvxxNwLdRoVgYt4YA731wnHrbosXZOXcuv
# H0z8Tm6ueKvfBtrQErdvqsGrP/8FUYRqZP4H6XaaC+wEis++7OmVR2nlQ/gAyr6/
# mMJtmxqVHCIcEVjDu1jYltrW3BN2CcxN2M9gxyOScq2/Xmzqtaeb4iyjxeCUjIBW
# Pc4LXlSafIg3hPrdH3EKN275ev8cx/5jp8oEgXD5We25Mj3W930zde6/STXoX318
# NVNlbrIQjGjQN7rN5oxTFxTlIN8ax2tuuzpQDFvS/4bLyMYXcZ4I5gUrM5tvWTGv
# +0UN45pJ7Nk=
# =l6Ki
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 04 Sep 2024 11:36:59 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-2024-09-04' of https://gitlab.com/thuth/qemu: (42 commits)
  docs/devel/testing: Add documentation for functional tests
  docs/devel/testing: Rename avocado_qemu.Test class
  docs/devel/testing: Split the Avocado documentation into a separate file
  docs/devel: Split testing docs from the build docs and move to separate folder
  gitlab-ci: Add "check-functional" to the build tests
  tests/avocado: Remove unused QemuUserTest class
  tests/functional: Convert ARM bFLT linux-user avocado test
  tests/functional: Add QemuUserTest class
  tests/functional: Convert mips64el Fuloong2e avocado test (1/2)
  tests/functional: Convert Aarch64 Virt machine avocado tests
  tests/functional: Convert Aarch64 SBSA-Ref avocado tests
  tests/functional: Convert ARM Integrator/CP avocado tests
  tests/functional: Convert the linux_initrd avocado test into a standalone test
  tests/functional: Convert the rx_gdbsim avocado test into a standalone test
  tests/functional: Convert the acpi-bits test into a standalone test
  tests/functional: Convert the m68k nextcube test with tesseract
  tests/functional: Convert the ppc_hv avocado test into a standalone test
  tests/functional: Convert the ppc_amiga avocado test into a standalone test
  tests/functional: Convert most ppc avocado tests into standalone tests
  tests/functional: Convert the virtio_gpu avocado test into a standalone test
  ...

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

platform-bus: fix refcount leak

memory_region_find() returns an MR which it is the caller's
responsibility to unref, but platform_bus_map_mmio() was
forgetting to do so, thus leaking the MR.

Signed-off-by: Gao Shiyuan <gaoshiyuan@baidu.com>
Message-id: 20240829131005.9196-1-gaoshiyuan@baidu.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: tweaked commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/boot: Explain why load_elf_hdr() error is ignored

If the file is not an ELF file, arm_setup_direct_kernel_boot()
falls back to try it as a uimage or an AArch64 Image file or as
last resort a bare raw binary. We can discard load_elf_hdr()
error and silently return.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240903144154.17135-1-philmd@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/boot: Report error msg if loading elf/dtb failed

Print errors before exit. Do not exit silently.

Cc: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Changbin Du <changbin.du@huawei.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240903133940.3447430-1-changbin.du@huawei.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/xilinx_zynq: Enable Security Extensions

The system supports the Security Extensions (core and GIC).  This change is
necessary to run tests which pass on the real hardware.

Signed-off-by: Sebastian Huber <sebastian.huber@embedded-brains.de>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Tested-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Message-id: 20240828005019.57705-1-sebastian.huber@embedded-brains.de
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Correct names of VFP VFNMA and VFNMS insns

In vfp.decode we have the names of the VFNMA and VFNMS instructions
the wrong way around.  The architecture says that bit 6 is the 'op'
bit, which is 1 for VFNMA and 0 for VFNMS, but we label these two
lines of decode the other way around.  This doesn't cause any
user-visible problem because in the handling of these functions in
translate-vfp.c we give VFNMA the behaviour specified for VFNMS and
vice-versa, but it's confusing when reading the code.

Switch the names of the VFP VFNMA and VFNMS instructions in
the decode file and flip the behaviour also.

NB: the instructions VFMA and VFMS *are* decoded with op=0 for
VFMA and op=1 for VFMS; the confusion probably arose because
we assumed VFNMA and VFNMS to be the same way around.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2536
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240830152156.2046590-1-peter.maydell@linaro.org
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

hw/arm/sbsa-ref: Don't leak string in sbsa_fdt_add_gic_node()

In sbsa_fdt_add_gic_node() we g_strdup_printf() two nodename
strings, but only free one.

Since the string is actually entirely constant and we don't
make any use of printf's format-string operations, we can
drop the g_strdup_printf() use entirely.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-id: 20240822162323.706382-1-peter.maydell@linaro.org

hm/nvram/xlnx-versal-efuse-ctrl: Call register_finalize_block

The TYPE_XLNX_VERSAL_EFUSE_CTRL device creates a register block with
register_init_block32() in its instance_init method; we must
therefore destroy it in our instance_finalize method to avoid a leak
in the QOM introspection "init-inspect-finalize" lifecycle:

Direct leak of 304 byte(s) in 1 object(s) allocated from:
    #0 0x55f222b5b9d8 in __interceptor_calloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-aarch64+0x294e9d8) (BuildId: 420
43d49e1139e3f3071b1f22fac1e3e7249c9a6)
    #1 0x7fbb10669c50 in g_malloc0 debian/build/deb/../../../glib/gmem.c:161:13
    #2 0x55f222f90c5d in register_init_block hw/core/register.c:248:34
    #3 0x55f222f916be in register_init_block32 hw/core/register.c:299:12
    #4 0x55f223bbdd15 in efuse_ctrl_init hw/nvram/xlnx-versal-efuse-ctrl.c:718:9
    #5 0x55f225b23391 in object_init_with_type qom/object.c:420:9
    #6 0x55f225b0a66b in object_initialize_with_type qom/object.c:562:5
    #7 0x55f225b0bf0d in object_new_with_type qom/object.c:782:5
    #8 0x55f225b0bfe1 in object_new qom/object.c:797:12
    #9 0x55f226309e0d in qmp_device_list_properties qom/qom-qmp-cmds.c:144:11

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20240822162127.705879-7-peter.maydell@linaro.org

hw/misc/xlnx-versal-trng: Call register_finalize_block

The TYPE_XLNX_VERSAL_TRNG device creates a register block with
register_init_block32() in its instance_init method; we must
therefore destroy it in our instance_finalize method to avoid a leak
in the QOM introspection "init-inspect-finalize" lifecycle:

Direct leak of 304 byte(s) in 1 object(s) allocated from:
    #0 0x55842ec799d8 in __interceptor_calloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-aarch64+0x294e9d8) (BuildId: 47496e53f3e779f1c7e9b82cbea07407152b498b)
    #1 0x7fe793c75c50 in g_malloc0 debian/build/deb/../../../glib/gmem.c:161:13
    #2 0x55842f0aec5d in register_init_block hw/core/register.c:248:34
    #3 0x55842f0af6be in register_init_block32 hw/core/register.c:299:12
    #4 0x55842f801588 in trng_init hw/misc/xlnx-versal-trng.c:614:9
    #5 0x558431c411a1 in object_init_with_type qom/object.c:420:9
    #6 0x558431c2847b in object_initialize_with_type qom/object.c:562:5
    #7 0x558431c29d1d in object_new_with_type qom/object.c:782:5
    #8 0x558431c29df1 in object_new qom/object.c:797:12
    #9 0x558432427c1d in qmp_device_list_properties qom/qom-qmp-cmds.c:144:11

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20240822162127.705879-6-peter.maydell@linaro.org

hw/nvram/xlnx-zynqmp-efuse: Call register_finalize_block

The TYPE_XLNX_ZYNQMP_EFUSE device creates a register block with
register_init_block32() in its instance_init method; we must
therefore destroy it in our instance_finalize method to avoid a leak
in the QOM introspection "init-inspect-finalize" lifecycle:

Direct leak of 304 byte(s) in 1 object(s) allocated from:
    #0 0x55f3ff5839d8 in __interceptor_calloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-aarch64+0x294d9d8) (BuildId: 23cf931c66865a71b6cc4da95156d03bc106fa72)
    #1 0x7f3f31c6bc50 in g_malloc0 debian/build/deb/../../../glib/gmem.c:161:13
    #2 0x55f3ff9b8c5d in register_init_block hw/core/register.c:248:34
    #3 0x55f3ff9b96be in register_init_block32 hw/core/register.c:299:12
    #4 0x55f4005e5b25 in efuse_ctrl_init hw/nvram/xlnx-versal-efuse-ctrl.c:718:9
    #5 0x55f40254afb1 in object_init_with_type qom/object.c:420:9
    #6 0x55f40253228b in object_initialize_with_type qom/object.c:562:5
    #7 0x55f402533b2d in object_new_with_type qom/object.c:782:5
    #8 0x55f402533c01 in object_new qom/object.c:797:12
    #9 0x55f402d31a2d in qmp_device_list_properties qom/qom-qmp-cmds.c:144:11

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20240822162127.705879-5-peter.maydell@linaro.org

hw/nvram/xlnx-bbram: Call register_finalize_block

The TYPE_XLNX_BBRAM device creates a register block with
register_init_block32() in its instance_init method; we must
therefore destroy it in our instance_finalize method to avoid a leak
in the QOM introspection "init-inspect-finalize" lifecycle:

Direct leak of 304 byte(s) in 1 object(s) allocated from:
    #0 0x5641518ca9d8 in __interceptor_calloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-aarch64+0x294d9d8) (BuildId: 4a6
18cb63d57d5a19ed45cfc262b08da47eaafe5)
    #1 0x7ff1aab31c50 in g_malloc0 debian/build/deb/../../../glib/gmem.c:161:13
    #2 0x564151cffc5d in register_init_block hw/core/register.c:248:34
    #3 0x564151d006be in register_init_block32 hw/core/register.c:299:12
    #4 0x56415293df75 in bbram_ctrl_init hw/nvram/xlnx-bbram.c:462:9
    #5 0x564154891dc1 in object_init_with_type qom/object.c:420:9
    #6 0x56415487909b in object_initialize_with_type qom/object.c:562:5
    #7 0x56415487a93d in object_new_with_type qom/object.c:782:5
    #8 0x56415487aa11 in object_new qom/object.c:797:12
    #9 0x56415507883d in qmp_device_list_properties qom/qom-qmp-cmds.c:144:11

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20240822162127.705879-4-peter.maydell@linaro.org

hw/misc/xlnx-versal-trng: Free s->prng in finalize, not unrealize

The TYPE_XLNX_VERSAL_TRNG device creates s->prng with g_rand_new()
in its init method, but it frees it in its unrealize method. This
results in a leak in the QOM introspection "initialize-inspect-finalize"
lifecycle:

Direct leak of 2500 byte(s) in 1 object(s) allocated from:
    #0 0x55ec89eae9d8 in __interceptor_calloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-aarch64+0x294d9d8) (BuildId: 6d5
08874816cc47d17c8dd775e8f809ae520e8cb)
    #1 0x7f697018fc50 in g_malloc0 debian/build/deb/../../../glib/gmem.c:161:13
    #2 0x7f6970197738 in g_rand_new_with_seed_array debian/build/deb/../../../glib/grand.c:202:17
    #3 0x7f6970197816 in g_rand_new debian/build/deb/../../../glib/grand.c:286:10
    #4 0x55ec8aa3656a in trng_init hw/misc/xlnx-versal-trng.c:624:15
    #5 0x55ec8ce75da1 in object_init_with_type qom/object.c:420:9
    #6 0x55ec8ce5d07b in object_initialize_with_type qom/object.c:562:5
    #7 0x55ec8ce5e91d in object_new_with_type qom/object.c:782:5
    #8 0x55ec8ce5e9f1 in object_new qom/object.c:797:12
    #9 0x55ec8d65c81d in qmp_device_list_properties qom/qom-qmp-cmds.c:144:11

Move the free to finalize so it matches where we are initing
s->prng. Since that's the only thing our unrealize method was
doing, this essentially switches the whole function to be
a finalize implementation.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20240822162127.705879-3-peter.maydell@linaro.org

hw/misc/xlnx-versal-cfu: destroy fifo in finalize

Since the TYPE_XNLX_VERSAL_CFU_FDRO device creates a FIFO in its
instance_init method, we must destroy the FIFO in instance_finalize
to avoid a memory leak for the QOM introspection
"instantiate-examine-finalize" cycle:

Direct leak of 8192 byte(s) in 1 object(s) allocated from:
    #0 0x55ec89eae7ee in malloc (/mnt/nvmedisk/linaro/qemu-from-laptop/qemu/build/asan/qemu-system-aarch64+0x294d7ee) (BuildId: 6d508874816cc47d17c8dd775e8f809ae520e8cb)
    #1 0x7f697018f738 in g_malloc debian/build/deb/../../../glib/gmem.c:128:13
    #2 0x55ec8d98d98d in fifo8_create util/fifo8.c:27:18
    #3 0x55ec8aa2a624 in fifo32_create /mnt/nvmedisk/linaro/qemu-from-laptop/qemu/include/qemu/fifo32.h:35:5
    #4 0x55ec8aa2a33c in cfu_fdro_init hw/misc/xlnx-versal-cfu.c:397:5
    #5 0x55ec8ce75da1 in object_init_with_type qom/object.c:420:9
    #6 0x55ec8ce5d07b in object_initialize_with_type qom/object.c:562:5
    #7 0x55ec8ce5e91d in object_new_with_type qom/object.c:782:5
    #8 0x55ec8ce5e9f1 in object_new qom/object.c:797:12
    #9 0x55ec8d65c81d in qmp_device_list_properties qom/qom-qmp-cmds.c:144:11

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Francisco Iglesias <francisco.iglesias@amd.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20240822162127.705879-2-peter.maydell@linaro.org

hw/arm/sbsa-ref: Use two-stage SMMU

Now that our SMMU model supports enabling both stages of translation
at once, we can enable this in the sbsa-ref board.  Existing guest
code that only programs stage 1 and doesn't care about stage 2 should
continue to run with the same behaviour, but guests that do want to
do nested SMMU configurations can now do so.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Message-id: 20240816161350.3706332-5-peter.maydell@linaro.org

hw/arm/virt: Default to two-stage SMMU from virt-9.2

Now that our SMMU model supports enabling both stages of translation
at once, we can enable this in the virt board.  This is no change in
behaviour for guests, because if they simply ignore stage 2 and never
configure it then it has no effect.  For the usual backwards
compatibility reasons we enable this only for machine types starting
with 9.2.

(Note that the SMMU is disabled by default on the virt board and is
only created if the user passes the 'iommu=smmuv3' machine option.)

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 20240816161350.3706332-4-peter.maydell@linaro.org

hw/arm/smmuv3: Update comment documenting "stage" property

When we added support for nested (stage 1 + stage 2) translation
to the SMMU in commit 58377c363291d we forgot to update the
comment that documents the valid values of the "stage" property.
Add the new "nested" value to it.

Fixes: 58377c363291d ("hw/arm/smmuv3: Support and advertise nesting")
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-id: 20240816161350.3706332-3-peter.maydell@linaro.org

hw: add compat machines for 9.2

Add 9.2 machine types for arm/i440fx/m68k/q35/s390x/spapr.

Signed-off-by: Cornelia Huck <cohuck@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Eric Auger <eric.auger@redhat.com>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240816161350.3706332-2-peter.maydell@linaro.org
Message-id: 20240816103723.2325982-1-cohuck@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

accel/tcg: Remove dead code from rr_cpu_thread_fn()

The main loop in rr_cpu_thread_fn() can never terminate, so the
code at the end of the function to clean up the RCU subsystem is
dead code. Replace it with g_assert_not_reached().

(This is different from the other cpu_thread_fn for e.g. MTTCG or
for the KVM accelerator -- those can exit, if the vCPU they
are responsible for is unplugged. But the RR cpu thread fn
handles all CPUs in the system in a round-robin way, so even
if one is unplugged it keeps looping.)

Resolves: Coverity CID 1547782
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240815143634.3413679-1-peter.maydell@linaro.org

target/arm: Enable FEAT_EBF16 in the "max" CPU

Now that we've implemented the required behaviour for FEAT_EBF16, we
can enable it for the "max" CPU type, list it in our documentation,
and delete a TODO comment about it being missing.

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

target/arm: Implement FPCR.EBF=1 semantics for bfdotadd()

Implement the FPCR.EBF=1 semantics for bfdotadd() operations:
 * is_ebf() sets up fpst and fpst_odd
 * bfdotadd_ebf() implements the fused paired-multiply-and-add
   operation that we need

The paired-multiply-and-add is similar to f16_dotadd() and
we use the same trick here as in that function, but the inputs
here are bfloat16 rather than float16.

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

target/arm: Prepare bfdotadd() callers for FEAT_EBF support

We use bfdotadd() in four callsites for various helper functions. Currently
this all assumes that we have the FPCR.EBF=0 semantics. For FPCR.EBF=1
we will need to:
 * call a different routine to bfdotadd() because we need to do a
   fused multiply-add rather than separate multiply and add steps
 * use a different float_status that honours the FPCR rounding mode
   and denormal-flushing fields
 * pass in an extra float_status that has been set up to perform
   round-to-odd rounding

To prepare for this, refactor all the callsites so that instead of
   for (...) {
       x = bfdotadd(...);
   }

they are:
   float_status fpst, fpst_odd;
   if (is_ebf(env, &fpst, &fpst_odd)) {
       for (...) {
           x = bfdotadd_ebf(..., &fpst, &fpst_odd);
       }
   } else {
       for (...) {
           x = bfdotadd(..., &fpst);
       }
   }

For the moment the is_ebf() function always returns false, sets up
fpst for EBF=0 semantics and never sets up fpst_odd; bfdotadd_ebf()
will assert if called. We'll fill in the handling for EBF=1 in the
next commit.

This change should be a zero-behaviour-change refactor.

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

target/arm: Pass env pointer through to gvec_bfmmla helper

Pass the env pointer through to the gvec_bfmmla helper,
so we can use it to add support for FEAT_EBF16.

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

target/arm: Pass env pointer through to gvec_bfdot_idx helper

Pass the env pointer through to the gvec_bfdot_idx helper,
so we can use it to add support for FEAT_EBF16.

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

target/arm: Pass env pointer through to gvec_bfdot helper

Pass the env pointer through to the gvec_bfdot helper,
so we can use it to add support for FEAT_EBF16.

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

target/arm: Pass env pointer through to sme_bfmopa helper

To implement the FEAT_EBF16 semantics, we are going to need
the CPUARMState env pointer in every helper function which calls
bfdotadd().

Pass the env pointer through from generated code to the sme_bfmopa
helper. (We'll add the code that uses it when we've adjusted
all the helpers to have access to the env pointer.)

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

target/arm: Allow setting the FPCR.EBF bit for FEAT_EBF16

FEAT_EBF16 adds one new bit to the FPCR floating point control
register.  Allow this bit to be read and written when the ID
registers indicate the presence of the feature.

Note that because this new bit is not in FPSCR_FPCR_MASK the bit is
not visible in the AArch32 FPSCR, and FPSCR writes do not affect it.

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

Merge tag 'edgar/xen-queue-2024-09-04.for-upstream' of https://gitlab.com/edgar.iglesias/qemu into staging

Edgars Xen queue.

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEErET+3BT38evtv0FRKcWWeA9ryoMFAmbYfMIACgkQKcWWeA9r
# yoNPfwgAuK6MyPGEJh75Pe7yEmkVeuL0RPTmet5Ie1WrywNsn91IybQGLctpxr1s
# 7m2Zhl3IWXV5Jezfr9gEnLF+LYBQgK0ENJWDTPuIqf7D5ZRtOVeaNqrQVJ10Jomn
# s1pn4kWtHxKsgWTzcdi3qP7vhxn1PllxE+yuOcoYQIHcUp1oQAHr6ApbAcxseWBE
# qPdgxT7nlvFdcKqkzxSgKy5MWjs3xcBg6R5Ywoy+t7lb003swivPnkK6MSt1P03h
# EkQsTWr0Ox4nACOWt15U0MoK5rtBEN7Gsox0FUEPF3QhmAJt75FTFLs8+JhqcnKH
# LMxud5C6t6FcI+kxPqPoIdEWy5uM6g==
# =hW3R
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 04 Sep 2024 16:29:06 BST
# gpg:                using RSA key AC44FEDC14F7F1EBEDBF415129C596780F6BCA83
# gpg: Good signature from "Edgar E. Iglesias (Xilinx key) <edgar.iglesias@xilinx.com>" [unknown]
# gpg:                 aka "Edgar E. Iglesias <edgar.iglesias@gmail.com>" [full]
# Primary key fingerprint: AC44 FEDC 14F7 F1EB EDBF  4151 29C5 9678 0F6B CA83

* tag 'edgar/xen-queue-2024-09-04.for-upstream' of https://gitlab.com/edgar.iglesias/qemu:
  docs/system/i386: xenpvh: Add a basic description
  hw/i386/xen: Add a Xen PVH x86 machine
  hw/xen: pvh-common: Add support for creating PCIe/GPEX
  hw/arm: xenpvh: Reverse virtio-mmio creation order
  hw/arm: xenpvh: Rename xen_arm.c -> xen-pvh.c
  hw/arm: xenpvh: Break out a common PVH machine
  hw/arm: xenpvh: Move stubbed functions to xen-stubs.c
  hw/arm: xenpvh: Remove double-negation in warning
  hw/arm: xenpvh: Add support for SMP guests
  hw/arm: xenpvh: Tweak machine description
  hw/arm: xenpvh: Update file header to use SPDX
  MAINTAINERS: Add docs/system/arm/xenpvh.rst

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

docs/system/i386: xenpvh: Add a basic description

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/i386/xen: Add a Xen PVH x86 machine

Add a Xen PVH x86 machine based on the abstract PVH Machine.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/xen: pvh-common: Add support for creating PCIe/GPEX

Add support for optionally creating a PCIe/GPEX controller.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Reverse virtio-mmio creation order

We've been creating the virtio-mmio devices in forwards order
but since the qbus lists prepend (rather than append) entries,
the virtio busses end up with decreasing base address order.

Xen enables virtio-mmio nodes in forwards order so there's been
a missmatch. So far, we've been working around this with an
out-of-tree patch to Xen.

This reverses the order making sure the virtio busses end up
ordered with increasing base addresses avoiding the need to
patch Xen.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Rename xen_arm.c -> xen-pvh.c

Rename xen_arm.c -> xen-pvh.c to better express that this
is a PVH machine and to align with x86 HVM and future PVH
machine filenames:
hw/i386/xen/xen-hvm.c
hw/i386/xen/xen-pvh.c (in preparation)

No functional changes.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Break out a common PVH machine

Break out a common Xen PVH machine in preparation for
adding a x86 Xen PVH machine.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Move stubbed functions to xen-stubs.c

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Remove double-negation in warning

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Add support for SMP guests

Add SMP support for Xen PVH ARM guests.
Create ms->smp.max_cpus ioreq servers to handle hotplug.

Note that ms->smp.max_cpus will be passed to us by the
user (Xen tools) set to the guests maxvcpus.

The value in mc->max_cpus is an absolute maximum for the
-smp option and won't be used to setup ioreq servers unless
the user explicitly specifies it with -smp.

If the user doesn't pass -smp on the command-line, smp.cpus
and smp.max_cpus will default to 1.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Tweak machine description

Tweak machine description to better express that this is
a Xen PVH machine for ARM.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Reviewed-by: Stefano Stabellini <sstabellini@kernel.org>

hw/arm: xenpvh: Update file header to use SPDX

Update file header to use SPDX and remove stray empty
comment line.

No functional changes.

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>

MAINTAINERS: Add docs/system/arm/xenpvh.rst

Signed-off-by: Edgar E. Iglesias <edgar.iglesias@amd.com>
Acked-by: Stefano Stabellini <sstabellini@kernel.org>

Merge tag 'hppa-v9.1-fixes-pull-request' of https://github.com/hdeller/qemu-hppa into staging

hppa target fixes

Two important patches for the hppa target which missed qemu-v9.1:
- One fix for random linux-user crashes
- One fix for random issues due to loosing the division V-bit
  during delivery of hardware interrupts. This triggers all sorts
  of random faults when running in system mode.

Helge

# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYKAB0WIQS86RI+GtKfB8BJu973ErUQojoPXwUCZtgy9AAKCRD3ErUQojoP
# X068AQCDCwbzsoQpFX/6Kx2lm3UsW4wrESh/CYcP+epd+X59dAD+LxNwN0ol5Gvl
# kPzM+7QdKC7geIXvBxIG6UuggTscJgM=
# =9roy
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 04 Sep 2024 11:14:12 BST
# 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 'hppa-v9.1-fixes-pull-request' of https://github.com/hdeller/qemu-hppa:
  target/hppa: Fix random 32-bit linux-user crashes
  target/hppa: Fix PSW V-bit packaging in cpu_hppa_get for hppa64

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

docs/devel/testing: Add documentation for functional tests

Document the new functional testing framework. The text is originally
based on the Avocado documentation, but heavily modified to match the
new framework.

Message-ID: <20240830133841.142644-45-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

docs/devel/testing: Rename avocado_qemu.Test class

The avocado_qemu.Test class has been renamed a while back in commit
2283b627bc ("tests/avocado: Rename avocado_qemu.Test -> QemuSystemTest"),
so we should reflect this now in the documentation, too.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240830133841.142644-44-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

docs/devel/testing: Split the Avocado documentation into a separate file

The main testing documentation file got very overloaded already.
Thus let's split the Avocado information into a separate file.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240830133841.142644-43-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

docs/devel: Split testing docs from the build docs and move to separate folder

Building and testing are two separate topics, so let's split the testing
into a separate category and move the related files into a separate folder.

Message-ID: <20240830133841.142644-42-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

gitlab-ci: Add "check-functional" to the build tests

Now that we converted many tests from the "check-avocado" test suite
to the "check-functional" test suite, we should make sure that these
also get tested in the CI.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-41-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/avocado: Remove unused QemuUserTest class

The single test that was using the QemuUserTest class
has been converted to the functional test framework.
This class is now unused, remove it.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240822104238.75045-4-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240830133841.142644-40-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert ARM bFLT linux-user avocado test

Straight forward conversion. Update the SHA1 hashes to
SHA256 hashes since SHA1 should not be used anymore nowadays.
Expose cpio_extract() in qemu_test.utils for possible reuse.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240822104238.75045-3-philmd@linaro.org>
[thuth: Add test to meson.build]
Message-ID: <20240830133841.142644-39-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Add QemuUserTest class

Per commit 5334df4822 ("tests/avocado: Introduce
QemuUserTest base class"):

  Similarly to the 'System' Test base class with methods
  for testing system emulation, the QemuUserTest class
  contains methods useful to test user-mode emulation.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240822104238.75045-2-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240830133841.142644-38-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert mips64el Fuloong2e avocado test (1/2)

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

Since the asset is expected locally and the test is guarded
with RESCUE_YL_PATH, keep it under the 'quick' category.

  $ RESCUE_YL_PATH=/path/to/rescue-yl QEMU_TEST_ALLOW_UNTRUSTED_CODE=1 \
    make check-functional-mips64el
  1/4 qemu:func-quick+func-mips64el / func-mips64el-empty_cpu_model      OK   0.12s   1 subtests passed
  2/4 qemu:func-quick+func-mips64el / func-mips64el-version              OK   0.13s   1 subtests passed
  3/4 qemu:func-quick+func-mips64el / func-mips64el-info_usernet         OK   0.15s   1 subtests passed
  4/4 qemu:func-quick+func-mips64el / func-mips64el-mips64el_fuloong2e   OK   0.19s   1 subtests passed

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240827094905.80648-2-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240830133841.142644-37-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert Aarch64 Virt machine avocado tests

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

Provide get_qemu_img() helper in qemu_test.

  $ make check-functional-aarch64 V=1
  ▶ 1/4 test_aarch64_virt.Aarch64VirtMachine.test_aarch64_virt_gicv2                 OK
  ▶ 1/4 test_aarch64_virt.Aarch64VirtMachine.test_aarch64_virt_gicv3                 OK
  ▶ 1/4 test_aarch64_virt.Aarch64VirtMachine.test_alpine_virt_tcg_gic_max            OK
  1/4 qemu:func-thorough+func-aarch64-thorough+thorough / func-aarch64-aarch64_virt  OK   30.44s  3 subtests passed
  ...

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240826103634.52384-1-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240830133841.142644-36-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert Aarch64 SBSA-Ref avocado tests

Straight forward conversion. Since SBSA_FLASH files are not
tarballs, use lzma_uncompress() method.

Avocado used to set a timeout of 11 tests * 180s = 1980s.
Hopefully 600s should be sufficient.

Running on macOS Sonoma / Apple silicon M1:

  $ QEMU_TEST_TIMEOUT_EXPECTED=1 make check-functional-aarch64 V=1
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_alpine_linux_max              OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_alpine_linux_max_pauth_impdef OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_alpine_linux_max_pauth_off    OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_alpine_linux_neoverse_n1      OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_edk2_firmware                 OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_openbsd73_cortex_a57          OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_openbsd73_max                 OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_openbsd73_max_pauth_impdef    OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_openbsd73_max_pauth_off       OK
  ▶ 1/5 test_aarch64_sbsaref.Aarch64SbsarefMachine.test_sbsaref_openbsd73_neoverse_n1         OK
  1/5 qemu:func-thorough+func-aarch64-thorough+thorough / func-aarch64-aarch64_sbsaref        OK   241.79s  11 subtests passed
  ...

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240823131614.10269-1-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Message-ID: <20240830133841.142644-35-thuth@redhat.com>
[thuth: remove remaining Avocado tags]
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert ARM Integrator/CP 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-arm
  ...
  6/6 qemu:func-thorough+func-arm-thorough+thorough / func-arm-arm_integratorcp   OK   3.90s   2 subtests passed

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240822110238.82312-1-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20240830133841.142644-34-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

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

While we're at it, update the SHA1 hashes to SHA256 hashes since
SHA1 should not be used anymore nowadays.

Message-ID: <20240830133841.142644-32-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

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

Provide a "gzip_uncompress" function based on the standard "gzip" module
to avoid the usage of avocado.utils here.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-31-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the acpi-bits test into a standalone test

Mostly a straight-forward conversion. Looks like we can simply drop
the avocado datadrainer stuff when not using the avocado framework
anymore.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-30-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the m68k nextcube test with tesseract

The code that handles running of tesseract needs to be tweaked a little
bit to be able to run without the functions from avocado.utils, and
while we're at it, drop some legacy stuff that was still there due to
Tesseract 3 support that we already dropped a while ago.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-29-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

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

A straight forward conversion, we just also have to remove the decorator
@skipUnless(os.getenv('SPEED')) since all non-trivial functional tests
are running in SPEED=thorough mode now. Also make sure that the extracted
assets are writable, so that the test does not fail if it gets re-run
and there are stale read-only files already around.

Message-ID: <20240830133841.142644-28-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

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

Use the Python standard zipfile module instead of avocado.utils for
extracting the ZIP file that we download here, and use the standard
subprocess module for running the "tail" command.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-27-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert most ppc avocado tests into standalone tests

Nothing thrilling in here, just straight forward conversions.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-26-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

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

Nothing thrilling in here, it's just a straight forward conversion.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-25-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the microblaze avocado tests into standalone tests

The machine_microblaze.py file contained two tests, one for each
endianness. Since we only support one QEMU target binary per file
in the new functional test environment, we have to split this file
up into two files now.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-23-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the x86_cpu_model_versions test

Nothing thrilling in here, it's just a straight forward conversion.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-22-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert the s390x avocado tests into standalone tests

These tests use archive.lzma_uncompress() from the Avocado utils,
so provide a small helper function for this, based on the
standard lzma module from Python instead.

And while we're at it, replace the MD5 hashes in the topology test
with proper SHA256 hashes, since MD5 should not be used anymore
nowadays.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-21-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert some avocado tests that needed avocado.utils.archive

Instead of using the "archive" module from avocado.utils, switch
these tests to use the new wrapper function that is based on the
"tarfile" module instead.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-20-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Add a function for extracting files from an archive

Some Avocado-based tests use the "archive" module from avocado.utils
to extract files from an archive. To be able to use these tests
without Avocado, we have to provide our own function for extracting
files. Fortunately, there is already the tarfile module that will
provide us with this functionality, so let's just add a nice wrapper
function around that.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240830133841.142644-19-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert some tests that download files via fetch_asset()

Now that we've got the Asset class with pre-caching, we can convert
some Avocado tests that use fetch_asset() for downloading their
required files.

Message-ID: <20240830133841.142644-18-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Allow asset downloading with concurrent threads

When running "make -j$(nproc) check-functional", tests that use the
same asset might be running in parallel. Improve the downloading to
detect this situation and wait for the other thread to finish the
download.

Message-ID: <20240830133841.142644-17-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: enable pre-emptive caching of assets

Many tests need to access assets stored on remote sites. We don't want
to download these during test execution when run by meson, since this
risks hitting test timeouts when data transfers are slow.

Add support for pre-emptive caching of assets by setting the env var
QEMU_TEST_PRECACHE to point to a timestamp file. When this is set,
instead of running the test, the assets will be downloaded and saved
to the cache, then the timestamp file created.

A meson custom target is created as a dependency of each test suite
to trigger the pre-emptive caching logic before the test runs.

When run in caching mode, it will locate assets by looking for class
level variables with a name prefix "ASSET_", and type "Asset".

At the ninja level

   ninja test --suite functional

will speculatively download any assets that are not already cached,
so it is advisable to set a timeout multiplier.

   QEMU_TEST_NO_DOWNLOAD=1 ninja test --suite functional

will fail the test if a required asset is not already cached

   ninja precache-functional

will download and cache all assets required by the functional
tests

At the make level, precaching is always done by

   make check-functional

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Tested-by: Richard Henderson <richard.henderson@linaro.org>
[thuth: Remove the duplicated "path = os.path.basename(...)" line]
Message-ID: <20240830133841.142644-16-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: add a module for handling asset download & caching

The 'Asset' class is a simple module that declares a downloadable
asset that can be cached locally. Downloads are stored in the user's
home dir at ~/.cache/qemu/download, using a sha256 sum of the URL.

[thuth: Drop sha1 support, use hash on file content for naming instead of URL,
        add the possibility to specify the cache dir via environment variable]

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-15-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert avocado tests that just need a small adjustment

These simple tests can be converted to stand-alone tests quite easily,
e.g. by just setting the machine to 'none' now manually or by adding
"-cpu" command line parameters, since we don't support the corresponding
avocado tags in the new python test framework.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-14-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Convert simple avocado tests into standalone python tests

These test are rather simple and don't need any modifications apart
from adjusting the "from avocado_qemu" line. To ease debugging, make
the files executable and add a shebang line and Python '__main__'
handling, too, so that these tests can now be run by executing them
directly.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-13-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Prepare the meson build system for the functional tests

Provide a meson.build file for the upcoming python-based functional
tests, and add some wrapper glue targets to the tests/Makefile.include
file. We are going to use two "speed" modes for the functional tests:
The "quick" tests can be run at any time (i.e. also during "make check"),
while the "thorough" tests should only be run when running a
"make check-functional" test run (since these tests might download
additional assets from the internet).

The changes to the meson.build files are partly based on an earlier
patch by Ani Sinha.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240830133841.142644-12-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/Makefile.include: Increase the level of indentation in the help text

The next patch is going to add some entries that need more space between
the command and the help text, so let's increase the indentation here
first.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Message-ID: <20240830133841.142644-11-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Set up logging

Create log files for each test separately, one file that contains
the basic logging and one that contains the console output.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240830133841.142644-10-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/functional: Add base classes for the upcoming pytest-based tests

The files are mostly a copy of the tests/avocado/avocado_qemu/__init__.py
file with some adjustments to get rid of the Avocado dependencies (i.e.
we also have to drop the LinuxSSHMixIn and LinuxTest for now).

The emulator binary and build directory are now passed via
environment variables that will be set via meson.build later.

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240830133841.142644-9-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

python: Install pycotap in our venv if necessary

The upcoming functional tests will require pycotap for providing
TAP output from the python-based tests. Since we want to be able
to run some of the tests offline by default, too, let's install
it along with meson in our venv if necessary (it's size is only
5 kB, so adding the wheel here should not really be a problem).

The wheel file has been obtained with:

 pip download --only-binary :all: --dest . --no-cache pycotap

Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20240830133841.142644-8-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>