Merge tag 'pull-tricore-20230929' of https://github.com/bkoppelmann/qemu into staging

- Add FTOU, CRCN, FTOHP, and HPTOF insns

# -----BEGIN PGP SIGNATURE-----
#
# iQJTBAABCgA9FiEEbmNqfoPy3Qz6bm43CtLGOWtpyhQFAmUWb2sfHGtiYXN0aWFu
# QG1haWwudW5pLXBhZGVyYm9ybi5kZQAKCRAK0sY5a2nKFPn0D/0S+Zth2okyfe6H
# YdoFB49PWlcafIvZHr1TDswp3LvSDnrjHLJfEW1Gx3mtDkw+/7uid0eMTQ8sDlxJ
# t7spJdZDZ5dkm+9K5MzGkW0zo0jDY6kbS1A3HJRPcpJJJk4zBBL1K4KC1FBUD6IK
# 7n41f5vExgWhIhOgZmT9WTMbBfh73/+Cu8h6M9RAI1VI0O6N5jOETpKTBFsPOx+A
# Kd429cB1c9QeAj0iEXdMn2/Xg2cAII86jrOcYkLYltxir/r6Cia9hfp/F6OXpcZI
# QqKzn11djvbCCL7m9OXhuI3ZP+TIcX7QOabSstfghHlNG1qs/RkXwIRqKHsfRXNG
# nywBTjwIDSiZ4cbZVJ6OjXxbU9OBRkmDgh+SYEVMlFi4E+t3WeTMC8gxUsjfITpK
# JXFoduN2P0yKRjkWQ2OSQ7xX4StFPikXBH1eC8RNnW4IY00wMiJ0tM/0+j+qJLLM
# Ft/bceIZhnGs+axN0jF1EtR03uLZ0kmy3YqsH/KnBnufrag3ytpC/kAtl9Scd6m+
# N4pAT9cfgxqXv/yXAKGupoNPwPGvvSKV6XQTJt2Hn7PBadHWlvlBkgYqGIejpHDM
# x9EghA8o4q5rTu9zTqBv36bOHJEDbJhmq5dYqJTS/q1ORjnWQQsLxv+6XGN3wrbb
# OuexPdD8fH3mWrjeJJ3KDKojOYyGyg==
# =gUyL
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 29 Sep 2023 02:32:11 EDT
# gpg:                using RSA key 6E636A7E83F2DD0CFA6E6E370AD2C6396B69CA14
# gpg:                issuer "kbastian@mail.uni-paderborn.de"
# gpg: Good signature from "Bastian Koppelmann <kbastian@mail.uni-paderborn.de>" [unknown]
# gpg: WARNING: The key's User ID is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: 6E63 6A7E 83F2 DD0C FA6E  6E37 0AD2 C639 6B69 CA14

* tag 'pull-tricore-20230929' of https://github.com/bkoppelmann/qemu:
  target/tricore: Change effective address (ea) to target_ulong
  target/tricore: Remove CSFRs from cpu.h
  tests/tcg: Reset result register after each test
  hw/tricore: Log failing test in testdevice
  tests/tcg/tricore: Extended and non-extened regs now match
  target/tricore: Fix FTOUZ being ISA v1.3.1 up
  target/tricore: Replace cpu_*_code with translator_*
  target/tricore: Swap src and dst reg for RCRR_INSERT
  target/tricore: Fix RCPW/RRPW_INSERT insns for width = 0
  target/tricore: Implement hptof insn
  target/tricore: Implement ftohp insn
  target/tricore: Clarify special case for FTOUZ insn
  target/tricore: Implement FTOU insn
  target/tricore: Correctly handle FPU RM from PSW
  target/tricore: Implement CRCN insn
  tests/tcg/tricore: Bump cpu to tc37x

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-tcg-20230928' of https://gitlab.com/rth7680/qemu into staging

accel/tcg: Always require can_do_io, for #1866

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmUV1sgdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV/8UAgAjFSrUvW5cGTiaow7
# cTvvoFnQzVCM7gFbC2T9W29Hrv0s3tqJHSdUFftzzMXB1atylwb85XN5wegC98zk
# 7+OzrUvTN2fRHVTnWdUg2Tgj+pR/Sw+9HVy851c3B/e3oFegqAIQaNK5w9N3ZpWd
# gpGN8Lau6dEdte8ckMTzG0Sw4LryZyvcgz9+vDv+YX03JaiSMqziHoJGwHfjqr9Q
# FOI1f4gQw+Y2ct6Xop1vVILH2I26US40Y2DvEMEDuZpTkTN0L2V8ipz+tWgGuCMa
# 7HRySJAw5HiDdtAV9/H5dfGJVgjsgwuWoYm+viJtJKJ+X350a5YI77hkbglPxQu+
# Q+dCVA==
# =lQYd
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 28 Sep 2023 15:40:56 EDT
# gpg:                using RSA key 7A481E78868B4DB6A85A05C064DF38E8AF7E215F
# gpg:                issuer "richard.henderson@linaro.org"
# gpg: Good signature from "Richard Henderson <richard.henderson@linaro.org>" [full]
# Primary key fingerprint: 7A48 1E78 868B 4DB6 A85A  05C0 64DF 38E8 AF7E 215F

* tag 'pull-tcg-20230928' of https://gitlab.com/rth7680/qemu:
  accel/tcg: Always require can_do_io
  accel/tcg: Always set CF_LAST_IO with CF_NOIRQ
  accel/tcg: Improve setting of can_do_io at start of TB
  accel/tcg: Track current value of can_do_io in the TB
  accel/tcg: Hoist CF_MEMI_ONLY check outside translation loop
  accel/tcg: Avoid load of icount_decr if unused

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'tracing-pull-request' of https://gitlab.com/stefanha/qemu into staging

Pull request

- Mads Ynddal's improvements to simpletrace.py

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEhpWov9P5fNqsNXdanKSrs4Grc8gFAmUVgyAACgkQnKSrs4Gr
# c8jxkAf/TVtEZTPDvQMz7PNEsAdGLREwYKrsUXHtqOrINbVL1slrJ1qYmygeJltq
# 37CAcxBzvnaH7y2jF8JPKWK3sO3x1fc2ocA72BZt3kuim5T7MojJgFfAN/PdwRHf
# q1a+lYJHKOi0BV4yWPmPHnCSXOohHbJfRBxwZgwFEZMbWdlkAm6m+JFUqn+Vd5ee
# 4vEBxrNMY28H02OeoFNBQPCLRHrkeH4EVj+DP5TAk1H1ehoZeH45UoUsopEfD51z
# IyiXxzumtgpCMyC/jXS5BY6VStW7lDLoDscflMQ/l7ITEtm7Hj2fnPcp/aRKoEUk
# p1JuoSm8YrwnwMokFsQCbkYWkBWieQ==
# =rrnI
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 28 Sep 2023 09:44:00 EDT
# gpg:                using RSA key 8695A8BFD3F97CDAAC35775A9CA4ABB381AB73C8
# gpg: Good signature from "Stefan Hajnoczi <stefanha@redhat.com>" [ultimate]
# gpg:                 aka "Stefan Hajnoczi <stefanha@gmail.com>" [ultimate]
# Primary key fingerprint: 8695 A8BF D3F9 7CDA AC35  775A 9CA4 ABB3 81AB 73C8

* tag 'tracing-pull-request' of https://gitlab.com/stefanha/qemu:
  scripts/analyse-locks-simpletrace.py: changed iteritems() to items()
  MAINTAINERS: add maintainer of simpletrace.py
  simpletrace: added simplified Analyzer2 class
  simpletrace: move event processing to Analyzer class
  simpletrace: move logic of process into internal function
  simpletrace: refactor to separate responsibilities
  simpletrace: made Analyzer into context-manager
  simpletrace: define exception and add handling
  simpletrace: improved error handling on struct unpack
  simpletrace: update code for Python 3.11
  simpletrace: changed naming of edict and idtoname to improve readability
  simpletrace: improve parsing of sys.argv; fix files never closed.
  simpletrace: annotate magic constants from QEMU code
  simpletrace: add __all__ to define public interface

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

target/tricore: Change effective address (ea) to target_ulong

as this is an effective address and those cannot be signed,
it should not be a signed integer.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230913105326.40832-11-kbastian@mail.uni-paderborn.de>

target/tricore: Remove CSFRs from cpu.h

these are already defined in 'csfr.h.inc'. We don't need to duplicate
these registers.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230913105326.40832-10-kbastian@mail.uni-paderborn.de>

tests/tcg: Reset result register after each test

some insns use the result register implicitly as an input. Thus, we
could end up with data from the previous insn spilling over.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230913105326.40832-4-kbastian@mail.uni-paderborn.de>

hw/tricore: Log failing test in testdevice

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230913105326.40832-3-kbastian@mail.uni-paderborn.de>

accel/tcg: Always require can_do_io

Require i/o as the last insn of a TranslationBlock always,
not only with icount.  This is required for i/o that alters
the address space, such as a pci config space write.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1866
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Always set CF_LAST_IO with CF_NOIRQ

Without this we can get see loops through cpu_io_recompile,
in which the cpu makes no progress.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Improve setting of can_do_io at start of TB

Initialize can_do_io to true if this the TB has CF_LAST_IO
and will consist of a single instruction.  This avoids a
set to 0 followed immediately by a set to 1.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Track current value of can_do_io in the TB

Simplify translator_io_start by recording the current
known value of can_do_io within DisasContextBase.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Hoist CF_MEMI_ONLY check outside translation loop

The condition checked is loop invariant; check it only once.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

accel/tcg: Avoid load of icount_decr if unused

With CF_NOIRQ and without !CF_USE_ICOUNT, the load isn't used.
Avoid emitting it.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

tests/tcg/tricore: Extended and non-extened regs now match

RSx for d regs and e regs now use the same numbering. This makes sure
that mixing d and e registers in an insn test will not overwrite data
between registers.

Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230913105326.40832-2-kbastian@mail.uni-paderborn.de>

target/tricore: Fix FTOUZ being ISA v1.3.1 up

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-12-kbastian@mail.uni-paderborn.de>

target/tricore: Replace cpu_*_code with translator_*

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-11-kbastian@mail.uni-paderborn.de>

target/tricore: Swap src and dst reg for RCRR_INSERT

Acked-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-10-kbastian@mail.uni-paderborn.de>

target/tricore: Fix RCPW/RRPW_INSERT insns for width = 0

we would crash if width was 0 for these insns, as tcg_gen_deposit() is
undefined for that case. For TriCore, width = 0 is a mov from the src reg
to the dst reg, so we special case this here.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-9-kbastian@mail.uni-paderborn.de>

target/tricore: Implement hptof insn

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1667
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-8-kbastian@mail.uni-paderborn.de>

target/tricore: Implement ftohp insn

reported in https://gitlab.com/qemu-project/qemu/-/issues/1667

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-7-kbastian@mail.uni-paderborn.de>

target/tricore: Clarify special case for FTOUZ insn

this is not something other ISAs do, so clarify it with a comment.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-6-kbastian@mail.uni-paderborn.de>

target/tricore: Implement FTOU insn

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-5-kbastian@mail.uni-paderborn.de>

target/tricore: Correctly handle FPU RM from PSW

when we reconstructed PSW using psw_read(), we were trying to clear the
cached USB bits out of env->PSW. The mask was wrong and we would clear
PSW.RM as well.

when we write the PSW using psw_write() we update the rounding modes in
env->fp_status for softfloat. The order of bits used by TriCore is not
the one used by softfloat.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-4-kbastian@mail.uni-paderborn.de>

target/tricore: Implement CRCN insn

reported in https://gitlab.com/qemu-project/qemu/-/issues/1667

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-ID: <20230828112651.522058-3-kbastian@mail.uni-paderborn.de>

tests/tcg/tricore: Bump cpu to tc37x

we don't want to exclude ISA v1.6.2 insns from our tests.

Acked-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Bastian Koppelmann <kbastian@mail.uni-paderborn.de>
Message-Id: <20230828112651.522058-2-kbastian@mail.uni-paderborn.de>

migration: Move return path cleanup to main migration thread

Now that the return path thread is allowed to finish during a paused
migration, we can move the cleanup of the QEMUFiles to the main
migration thread.

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-9-farosas@suse.de>

migration: Replace the return path retry logic

Replace the return path retry logic with finishing and restarting the
thread. This fixes a race when resuming the migration that leads to a
segfault.

Currently when doing postcopy we consider that an IO error on the
return path file could be due to a network intermittency. We then keep
the thread alive but have it do cleanup of the 'from_dst_file' and
wait on the 'postcopy_pause_rp' semaphore. When the user issues a
migrate resume, a new return path is opened and the thread is allowed
to continue.

There's a race condition in the above mechanism. It is possible for
the new return path file to be setup *before* the cleanup code in the
return path thread has had a chance to run, leading to the *new* file
being closed and the pointer set to NULL. When the thread is released
after the resume, it tries to dereference 'from_dst_file' and crashes:

Thread 7 "return path" received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7fffd1dbf700 (LWP 9611)]
0x00005555560e4893 in qemu_file_get_error_obj (f=0x0, errp=0x0) at ../migration/qemu-file.c:154
154         return f->last_error;

(gdb) bt
 #0  0x00005555560e4893 in qemu_file_get_error_obj (f=0x0, errp=0x0) at ../migration/qemu-file.c:154
 #1  0x00005555560e4983 in qemu_file_get_error (f=0x0) at ../migration/qemu-file.c:206
 #2  0x0000555555b9a1df in source_return_path_thread (opaque=0x555556e06000) at ../migration/migration.c:1876
 #3  0x000055555602e14f in qemu_thread_start (args=0x55555782e780) at ../util/qemu-thread-posix.c:541
 #4  0x00007ffff38d76ea in start_thread (arg=0x7fffd1dbf700) at pthread_create.c:477
 #5  0x00007ffff35efa6f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95

Here's the race (important bit is open_return_path happening before
migration_release_dst_files):

migration                 | qmp                         | return path
--------------------------+-----------------------------+---------------------------------
    qmp_migrate_pause()
     shutdown(ms->to_dst_file)
      f->last_error = -EIO
migrate_detect_error()
 postcopy_pause()
  set_state(PAUSED)
  wait(postcopy_pause_sem)
    qmp_migrate(resume)
    migrate_fd_connect()
     resume = state == PAUSED
     open_return_path <-- TOO SOON!
     set_state(RECOVER)
     post(postcopy_pause_sem)
(incoming closes to_src_file)
res = qemu_file_get_error(rp)
migration_release_dst_files()
ms->rp_state.from_dst_file = NULL
  post(postcopy_pause_rp_sem)
postcopy_pause_return_path_thread()
  wait(postcopy_pause_rp_sem)
rp = ms->rp_state.from_dst_file
goto retry
qemu_file_get_error(rp)
SIGSEGV
-------------------------------------------------------------------------------------------

We can keep the retry logic without having the thread alive and
waiting. The only piece of data used by it is the 'from_dst_file' and
it is only allowed to proceed after a migrate resume is issued and the
semaphore released at migrate_fd_connect().

Move the retry logic to outside the thread by waiting for the thread
to finish before pausing the migration.

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-8-farosas@suse.de>

migration: Consolidate return path closing code

We'll start calling the await_return_path_close_on_source() function
from other parts of the code, so move all of the related checks and
tracepoints into it.

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-7-farosas@suse.de>

migration: Remove redundant cleanup of postcopy_qemufile_src

This file is owned by the return path thread which is already doing
cleanup.

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-6-farosas@suse.de>

migration: Fix possible race when shutting down to_dst_file

It's not safe to call qemu_file_shutdown() on the to_dst_file without
first checking for the file's presence under the lock. The cleanup of
this file happens at postcopy_pause() and migrate_fd_cleanup() which
are not necessarily running in the same thread as migrate_fd_cancel().

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-5-farosas@suse.de>

migration: Fix possible races when shutting down the return path

We cannot call qemu_file_shutdown() on the return path file without
taking the file lock. The return path thread could be running it's
cleanup code and have just cleared the from_dst_file pointer.

Checking ms->to_dst_file for errors could also race with
migrate_fd_cleanup() which clears the to_dst_file pointer.

Protect both accesses by taking the file lock.

This was caught by inspection, it should be rare, but the next patches
will start calling this code from other places, so let's do the
correct thing.

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-4-farosas@suse.de>

migration: Fix possible race when setting rp_state.error

We don't need to set the rp_state.error right after a shutdown because
qemu_file_shutdown() always sets the QEMUFile error, so the return
path thread would have seen it and set the rp error itself.

Setting the error outside of the thread is also racy because the
thread could clear it after we set it.

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-3-farosas@suse.de>

migration: Fix race that dest preempt thread close too early

We hit intermit CI issue on failing at migration-test over the unit test
preempt/plain:

qemu-system-x86_64: Unable to read from socket: Connection reset by peer
Memory content inconsistency at 5b43000 first_byte = bd last_byte = bc current = 4f hit_edge = 1
**
ERROR:../tests/qtest/migration-test.c:300:check_guests_ram: assertion failed: (bad == 0)
(test program exited with status code -6)

Fabiano debugged into it and found that the preempt thread can quit even
without receiving all the pages, which can cause guest not receiving all
the pages and corrupt the guest memory.

To make sure preempt thread finished receiving all the pages, we can rely
on the page_requested_count being zero because preempt channel will only
receive requested page faults. Note, not all the faulted pages are required
to be sent via the preempt channel/thread; imagine the case when a
requested page is just queued into the background main channel for
migration, the src qemu will just still send it via the background channel.

Here instead of spinning over reading the count, we add a condvar so the
main thread can wait on it if that unusual case happened, without burning
the cpu for no good reason, even if the duration is short; so even if we
spin in this rare case is probably fine.  It's just better to not do so.

The condvar is only used when that special case is triggered.  Some memory
ordering trick is needed to guarantee it from happening (against the
preempt thread status field), so the main thread will always get a kick
when that triggers correctly.

Closes: https://gitlab.com/qemu-project/qemu/-/issues/1886
Debugged-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-ID: <20230918172822.19052-2-farosas@suse.de>

Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging

* new round of audio cleanups
* various shadowed local variable fixes in vl, mptsas, pm_smbus, target/i386
* remove deprecated pc-i440fx-1.4 up to pc-i440fx-1.7
* remove PCI drivers from 128K bios.bin
* remove unused variable in user-exec-stub.c
* small fixes for ui/vnc
* scsi-disk: Disallow block sizes smaller than 512 [CVE-2023-42467]

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmUTDaoUHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroMvEgf+NrSaP4pmHrYcVtm43fnKXoLHFrCx
# KYfoK9Lke/DDkTff6rrcfW/Wyqid6Pp9Ch4Rrpr/X71X5gi+c6xb5klC8cpSfLg4
# gtuGctj7WL7KR/067EsLqHvzBob/iebFhZwhtsBrI+z65X+J9pOK78efBTdhezq4
# EEHTWohMAg1I/MWBK5VnOk2fI4+9z9K9zP5AtWmJzwwJkQUoEyl+YDkVmIhMYoGn
# CapRO7i2wIvtoF4wuQUCGsOLmrcWTvRIOcV13k3b6PYCPC40/N9AOpiiyg3XqNah
# UKKM9CcgVnCzCc4Jar2QD+MzkTDxhmQSyLFJgtzrW7CQSE5YB3sUHj3CXg==
# =8nvs
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 26 Sep 2023 12:58:18 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:
  audio: remove shadowed locals
  compiler: introduce QEMU_ANNOTATE
  block: mark mixed functions that can suspend
  target/i386/svm_helper: eliminate duplicate local variable
  target/i386/seg_helper: remove shadowed variable
  target/i386/seg_helper: introduce tss_set_busy
  target/i386/translate: avoid shadowed local variables
  target/i386/cpu: avoid shadowed local variables
  target/i386/kvm: eliminate shadowed local variables
  m48t59-test: avoid possible overflow on ABS
  pm_smbus: rename variable to avoid shadowing
  mptsas: avoid shadowed local variables
  ui/vnc: fix handling of VNC_FEATURE_XVP
  ui/vnc: fix debug output for invalid audio message
  vl: remove shadowed local variables
  hw/scsi/scsi-disk: Disallow block sizes smaller than 512 [CVE-2023-42467]
  user-exec-stub: remove unused variable
  seabios: remove PCI drivers from bios.bin
  pc_piix: remove pc-i440fx-1.4 up to pc-i440fx-1.7

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

scripts/analyse-locks-simpletrace.py: changed iteritems() to items()

Python 3 removed `dict.iteritems()` in favor of `dict.items()`. This
means the script currently doesn't work on Python 3.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-15-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

MAINTAINERS: add maintainer of simpletrace.py

In my work to refactor simpletrace.py, I noticed that there's no
maintainer of it, and has the status of "odd fixes". I'm using it from
time to time, so I'd like to maintain the script.

I've added myself as reviewer under "Tracing" to be informed of changes
that might affect simpletrace.py.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-14-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: added simplified Analyzer2 class

By moving the dynamic argument construction to keyword-arguments,
we can remove all of the specialized handling, and streamline it.
If a tracing method wants to access these, they can define the
kwargs, or ignore it be placing `**kwargs` at the end of the
function's arguments list.

Added deprecation warning to Analyzer class to make users aware
of the Analyzer2 class. No removal date is planned.

Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-13-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: move event processing to Analyzer class

Moved event processing to the Analyzer class to separate specific analyzer
logic (like caching and function signatures) from the _process function.
This allows for new types of Analyzer-based subclasses without changing
the core code.

Note, that the fn_cache is important for performance in cases where the
analyzer is branching away from the catch-all a lot. The cache has no
measurable performance penalty.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-12-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: move logic of process into internal function

To avoid duplicate code depending on input types and to better handle
open/close of log with a context-manager, we move the logic of process into
_process.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-11-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: refactor to separate responsibilities

Moved event_mapping and event_id_to_name down one level in the function
call-stack to keep variable instantiation and usage closer (`process`
and `run` has no use of the variables; `read_trace_records` does).

Instead of passing event_mapping and event_id_to_name to the bottom of
the call-stack, we move their use to `read_trace_records`. This
separates responsibility and ownership of the information.

`read_record` now just reads the arguments from the file-object by
knowning the total number of bytes. Parsing it to specific arguments is
moved up to `read_trace_records`.

Special handling of dropped events removed, as they can be handled
by the general code.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-10-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: made Analyzer into context-manager

Instead of explicitly calling `begin` and `end`, we can change the class
to use the context-manager paradigm. This is mostly a styling choice,
used in modern Python code. But it also allows for more advanced analyzers
to handle exceptions gracefully in the `__exit__` method (not
demonstrated here).

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-9-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: define exception and add handling

Define `SimpleException` to differentiate our exceptions from generic
exceptions (IOError, etc.). Adapted simpletrace to support this and
output to stderr.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-8-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: improved error handling on struct unpack

A failed call to `read_header` wouldn't be handled the same for the two
different code paths (one path would try to use `None` as a list).
Changed to raise exception to be handled centrally. This also allows for
easier unpacking, as errors has been filtered out.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-7-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: update code for Python 3.11

The call to `getargspec` was deprecated and in Python 3.11 it has been
removed in favor of `getfullargspec`. `getfullargspec` is compatible
with QEMU's requirement of at least Python version 3.6.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-6-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: changed naming of edict and idtoname to improve readability

Readability is subjective, but I've expanded the naming of the variables
and arguments, to help with understanding for new eyes on the code.

Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Message-id: 20230926103436.25700-5-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: improve parsing of sys.argv; fix files never closed.

The arguments extracted from `sys.argv` named and unpacked to make it
clear what the arguments are and what they're used for.

The two input files were opened, but never explicitly closed. File usage
changed to use `with` statement to take care of this. At the same time,
ownership of the file-object is moved up to `run` function. Added option
to process to support file-like objects.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-4-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: annotate magic constants from QEMU code

It wasn't clear where the constants and structs came from, so I added
comments to help.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-3-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

simpletrace: add __all__ to define public interface

It was unclear what was the supported public interface. I.e. when
refactoring the code, what functions/classes are important to retain.

Reviewed-by: Stefan Hajnoczi <stefanha@redhat.com>
Signed-off-by: Mads Ynddal <m.ynddal@samsung.com>
Message-id: 20230926103436.25700-2-mads@ynddal.dk
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

audio: remove shadowed locals

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

compiler: introduce QEMU_ANNOTATE

Allow a more shorter syntax when defining wrapper macros for
__attribute__((annotate(...))).

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

block: mark mixed functions that can suspend

The marking should be extended transitively to all functions that call
these ones, so that static analysis can be done much more efficiently.
However, this is a start and makes it possible to use vrc's path-based
searches to find potential bugs where coroutine_fns call blocking functions.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386/svm_helper: eliminate duplicate local variable

This shadows an outer "cs" variable that is initialized to the
same expression.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386/seg_helper: remove shadowed variable

Return the width of the new task directly from switch_tss_ra.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386/seg_helper: introduce tss_set_busy

Eliminate a shadowed local variable in the process.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386/translate: avoid shadowed local variables

Just remove the declaration.  There is nothing in the function after the
switch statement, so it is safe to do.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386/cpu: avoid shadowed local variables

Reuse the pointer variable used for the unversioned model.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

target/i386/kvm: eliminate shadowed local variables

These are harmless are they die immediately after their use.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

m48t59-test: avoid possible overflow on ABS

Originally meant to avoid a shadowed variable "s", which was fixed by
renaming the outer declaration to "qts".  Avoid the chance of an overflow
in the computation of ABS(t - s).

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

pm_smbus: rename variable to avoid shadowing

Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge tag 'pull-nbd-2023-09-25' of https://repo.or.cz/qemu/ericb into staging

NBD patches through 2023-09-25

- Denis V. Lunev: iotest improvements
- Eric Blake: further work towards 64-bit NBD extensions

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCAAdFiEEccLMIrHEYCkn0vOqp6FrSiUnQ2oFAmUR2MUACgkQp6FrSiUn
# Q2q6jAf+PT65XzMAhgKvu1vIeMSQqyCocNB2MCOzNp+46uB9bNbPPLQSH2EX+t6p
# kQfHyHUl4YMi0EqgCfodiewlaUKeMxP3cPWMGYaYZ16uNMOIYL1boreDAcM25rb5
# P3TV3DAWTWSclUxrkTC2DxAIBPgsPsGG/2daqOMDEdinxlIywCMJDEIHc9gwwd/t
# 7laz9V1cOW9NbQXrM7eTofJKPKIeqZ+w0kvqrf9HBvZl9CqwHADi7xoz9xP+fN+f
# 713ED/hwt0FIlixtIm2/8vu7nn09cu6m9NaKsMOomsYg9Z6wU3ctivViG5NLq3MD
# OOUu51dV8gRRAXAFU5vKb0d93D27zQ==
# =Ik02
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 25 Sep 2023 15:00:21 EDT
# gpg:                using RSA key 71C2CC22B1C4602927D2F3AAA7A16B4A2527436A
# gpg: Good signature from "Eric Blake <eblake@redhat.com>" [full]
# gpg:                 aka "Eric Blake (Free Software Programmer) <ebb9@byu.net>" [full]
# gpg:                 aka "[jpeg image of size 6874]" [full]
# Primary key fingerprint: 71C2 CC22 B1C4 6029 27D2  F3AA A7A1 6B4A 2527 436A

* tag 'pull-nbd-2023-09-25' of https://repo.or.cz/qemu/ericb:
  nbd/server: Refactor handling of command sanity checks
  nbd: Prepare for 64-bit request effect lengths
  nbd: Add types for extended headers
  nbd/client: Pass mode through to nbd_send_request
  nbd: Replace bool structured_reply with mode enum
  iotests: improve 'not run' message for nbd-multiconn test
  iotests: use TEST_IMG_FILE instead of TEST_IMG in _require_large_file

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

mptsas: avoid shadowed local variables

Rename the argument so that "addr" is only used inside the for loop.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

ui/vnc: fix handling of VNC_FEATURE_XVP

VNC_FEATURE_XVP was not shifted left before adding it to vs->features,
so it was never enabled; but it was also checked the wrong way with
a logical AND instead of vnc_has_feature.  Fix both places.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

ui/vnc: fix debug output for invalid audio message

The debug message was cut and pasted from the invalid audio format
case, but the audio message is at bytes 2-3.

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

vl: remove shadowed local variables

Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/scsi/scsi-disk: Disallow block sizes smaller than 512 [CVE-2023-42467]

We are doing things like

    nb_sectors /= (s->qdev.blocksize / BDRV_SECTOR_SIZE);

in the code here (e.g. in scsi_disk_emulate_mode_sense()), so if
the blocksize is smaller than BDRV_SECTOR_SIZE (=512), this crashes
with a division by 0 exception. Thus disallow block sizes of 256
bytes to avoid this situation.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1813
CVE: 2023-42467
Signed-off-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20230925091854.49198-1-thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

user-exec-stub: remove unused variable

enable_cpu_pm is only used by softmmu-specific code, namely target/i386/host-cpu.c
and target/i386/kvm/*.  It does not need a stub definition anymore.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

seabios: remove PCI drivers from bios.bin

bios.bin is now used only by ISA PC, so PCI drivers are not necessary.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

pc_piix: remove pc-i440fx-1.4 up to pc-i440fx-1.7

These are the last users of the 128K SeaBIOS blob in the i440FX family.
Removing them allows us to drop PCI support from the 128K blob,
thus making it easier to update SeaBIOS to newer versions.

Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Merge tag 'pull-request-2023-09-25' of https://gitlab.com/thuth/qemu into staging

* Make keyutils independent from keyring in meson.build
* Simplify the NIC init code of the jazz machine a little bit
* Minor qtest and avocado fixes

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmURS8gRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbVn4A/+NQKFZcN7gVn5JXkK7kf6i01LNmAoqjj9
# QeQL+WCoNC68OApw7DxIEnpBYT0G42NTHHx4SYeOvzJUzCpeWcxYzQUz58ObZML7
# +OKsiOsaHu3/qOuihBCn43et6moLdDCWbee5Zr6JQv/Fjn3q3nEQZnJDWdw8vm1v
# csYQJZOD6HelLVMmbLfl1szzrykDTT53NhPncH/SjPz6we17sKqHqmT6LBUIsXcV
# u2LaowppKmT7Ooexu6SmsCagLhtWuYo1iGGcRqoojtRWo7eZtWLrAy2DJpyFkPBW
# AIYBfntRISZv4eBGCxcVfvODD/Q4OXHuYTfGzD3m+ELJ6hUk/+d4/aHJ2hm+KEm+
# AD0IpDtimaEmyQTPlaWHhhEur/82JZ+zYlxUMPf3+hglB/rbr6fhA0SMAV6nwR0r
# N8jnB8UCml9oDxJVvDZyrcPMGFs1xlr5FVSHHEoL338SvSfjG3NOEtcNao9n6A8d
# rO2CfPzI7peQhKWAzJL+qpnmenyIniH23tFnf2mpOZ0g45ZWtJeT0CXL3aQO3XAZ
# m56pkM0d/etAHHRoLQ5D/iKZpwiTRLjdzsJ0gMAQsIuRlG/j5h+zou0vUMgm6F8F
# igRHLxytlywZBTCABm2XIlKmaJp8hQlVQMpKsv/BwzTvzzk0GGS5d1qzzFt5WWR7
# 4rSalTn5Xuw=
# =FioB
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 25 Sep 2023 04:58:48 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-2023-09-25' of https://gitlab.com/thuth/qemu:
  tests/avocado: fix waiting for vm shutdown in replay_linux
  hw/mips/jazz: Simplify the NIC setup code
  hw/mips/jazz: Move the NIC init code into a separate function
  tests/qtest/netdev-socket: Do not test multicast on Darwin
  tests/qtest/m48t59-test: Silence compiler warning with -Wshadow
  tests/qtest/netdev-socket: Raise connection timeout to 120 seconds
  meson.build: Make keyutils independent from keyring

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'for-upstream' of https://gitlab.com/bonzini/qemu into staging

* add host ticks function for RISC-V
* target/i386: Export GDS_NO bit
* target/i386: add support for bit 56 of MSR_IA32_VMX_BASIC
* first part of audiodev cleanups

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmUNtYUUHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroN7Uwf9Fy4aE1PHzSNr2FqT4rUSYrT4N8cL
# QiPeB8JiJUnl73TcCkTwi7S/Az+37okv+Qsr7eh1wdarY8DOYir9dGJU3TGzICSw
# cgPImb99rhBc2kEmwciCWGlhXIMD8WNN64EanPPg5VeQYdzrorYwl7jCTMQMBR5H
# wtOq3f6FfYJonVwZ6YOmbioD2mFfoGBuiDcYmTTw440vrruKqHagbm5onD1SY9kR
# SM0/HXcYaKB6Ae9qNKhyR9h94KZzDUkCvcTLdFGtK90GBs4VxZVHQn6Dpkh5lPtT
# t0MbMv1mcO6ODzg9TxO3gUAgoklTy3gM2wISXo5C9NGuxmF2svwkuQl5pg==
# =CuIa
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 22 Sep 2023 11:40:53 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:
  vl: recognize audiodev groups in configuration files
  tests/qtest: Specify audiodev= and -audiodev
  hw/display/xlnx_dp.c: Add audiodev property
  hw/audio/lm4549: Add errp error reporting to init function
  hw/audio: Simplify hda audio init
  hw/input/tsc210x: Extract common init code into new function
  qemu/timer: Add host ticks function for RISC-V
  target/i386: Export GDS_NO bit to guests
  target/i386: enumerate bit 56 of MSR_IA32_VMX_BASIC

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

Merge tag 'pull-target-arm-20230921' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * target/m68k: Add URL to semihosting spec
 * docs/devel/loads-stores: Fix git grep regexes
 * hw/arm/boot: Set SCR_EL3.FGTEn when booting kernel
 * linux-user: Correct SME feature names reported in cpuinfo
 * linux-user: Add missing arm32 hwcaps
 * Don't skip MTE checks for LDRT/STRT at EL0
 * Implement FEAT_HBC
 * Implement FEAT_MOPS
 * audio/jackaudio: Avoid dynamic stack allocation
 * sbsa-ref: add non-secure EL2 virtual timer
 * elf2dmp: improve Win2022, Win11 and large dumps

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmUMfwAZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3jvnD/0QE/oOxfr+wkDUkTasSwVc
# UNfhObMj3h8x2XApqXckXnckew97I7hh7OLk35p9Ncea7fb6CvGMZ/DJir7AG4aQ
# Anpd5g2Qo0AMfPIyvoJ5pgtqZ1aS/EpBfYixmjL/zY6+zNzoVzWG/KfL+XamW6ir
# 6U7EqcAUzfX0+Splcxs5WgCDI5nGtn0B42EwOMpmwsH4opfr6HTn8Rzbn9gIwKU7
# u82PaKAqWPYD0ev9NQra+VVTrrFS4SCcqkV+SoYu0Cg5vvBlgAVcx0Zz2objp9LC
# 96fOtFH4Rch611j87WiGvN+fxQawqYzAYdy2y+j0wwuonTH9G3PpdZZT0557NjeS
# rFpW2UQebDqZ3ZTDwhzefsVKc3emLZtEd+RFa/YcDtao0afKfbSHv5A2/pGHxzlv
# 8psKOOH82WXTOHwFKA2o0lXDAauzirY+1Avy0vozNzPCdErXPgMHY4tABU77PpER
# Pz17jJO9C1AGyQVF+o09ieJR2Du5Wb2LLcZP3+5Ctm0SNVmREKKNcMkhJiEM9snm
# PQBR7FNEbAuQAO2MDK70dWUcTNtOv4Q1jgTR+aYd2MrArxCmAA5Zd9gjeYDwv6XH
# n242ONDAhlG1fY5f5giE3vCrcV1FDbvHEn6GDVilgMrF3a3Iw30xUaATiO09hIfi
# XAwGwLtMsp21WDa5PsfZVw==
# =dalQ
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 21 Sep 2023 13:36:00 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-20230921' of https://git.linaro.org/people/pmaydell/qemu-arm: (30 commits)
  elf2dmp: rework PDB_STREAM_INDEXES::segments obtaining
  elf2dmp: use Linux mmap with MAP_NORESERVE when possible
  elf2dmp: introduce merging of physical memory runs
  elf2dmp: introduce physical block alignment
  elf2dmp: replace PE export name check with PDB name check
  sbsa-ref: add non-secure EL2 virtual timer
  audio/jackaudio: Avoid dynamic stack allocation in qjack_process()
  audio/jackaudio: Avoid dynamic stack allocation in qjack_client_init
  target/arm: Enable FEAT_MOPS for CPU 'max'
  target/arm: Implement the CPY* instructions
  target/arm: Implement MTE tag-checking functions for FEAT_MOPS copies
  target/arm: Implement the SETG* instructions
  target/arm: Define new TB flag for ATA0
  target/arm: Implement the SET* instructions
  target/arm: Implement MTE tag-checking functions for FEAT_MOPS
  target/arm: New function allocation_tag_mem_probe()
  target/arm: Define syndrome function for MOPS exceptions
  target/arm: Pass unpriv bool to get_a64_user_mem_index()
  target/arm: Implement FEAT_MOPS enable bits
  target/arm: Don't skip MTE checks for LDRT/STRT at EL0
  ...

Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>

nbd/server: Refactor handling of command sanity checks

Upcoming additions to support NBD 64-bit effect lengths will add a new
command flag NBD_CMD_FLAG_PAYLOAD_LEN that needs to be considered in
our sanity checks of the client's messages (that is, more than just
CMD_WRITE have the potential to carry a client payload when extended
headers are in effect).  But before we can start to support that, it
is easier to first refactor the existing set of various if statements
over open-coded combinations of request->type to instead be a single
switch statement over all command types that sets witnesses, then
straight-line processing based on the witnesses.  No semantic change
is intended.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230829175826.377251-24-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

nbd: Prepare for 64-bit request effect lengths

Widen the length field of NBDRequest to 64-bits, although we can
assert that all current uses are still under 32 bits: either because
of NBD_MAX_BUFFER_SIZE which is even smaller (and where size_t can
still be appropriate, even on 32-bit platforms), or because nothing
ever puts us into NBD_MODE_EXTENDED yet (and while future patches will
allow larger transactions, the lengths in play here are still capped
at 32-bit).  There are no semantic changes, other than a typo fix in a
couple of error messages.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230829175826.377251-23-eblake@redhat.com>
[eblake: fix assertion bug in nbd_co_send_simple_reply]
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

tests/avocado: fix waiting for vm shutdown in replay_linux

This patch fixes the race condition in waiting for shutdown
of the replay linux test.

Signed-off-by: Pavel Dovgalyuk <Pavel.Dovgalyuk@ispras.ru>
Suggested-by: John Snow <jsnow@redhat.com>
Message-ID: <20230811070608.3383343-4-pavel.dovgalyuk@ispras.ru>
Signed-off-by: Thomas Huth <thuth@redhat.com>

hw/mips/jazz: Simplify the NIC setup code

The for-loop does not make much sense here - it is always left after
the first iteration, so we can also check for nb_nics == 1 instead
which is way easier to understand.

Also, the checks for nd->model are superfluous since the code in
mips_jazz_init_net() calls qemu_check_nic_model() that already
takes care of this (i.e. initializing nd->model if it has not been
set yet, and checking whether it is the "help" option or the
supported NIC model).

Message-ID: <20230913160922.355640-3-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

hw/mips/jazz: Move the NIC init code into a separate function

The mips_jazz_init() function is already quite big, so moving
away some code here can help to make it more understandable.
Additionally, by moving this code into a separate function, the
next patch (that will refactor the for-loop around the NIC init
code) will be much shorter and easier to understand.

Message-ID: <20230913160922.355640-2-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest/netdev-socket: Do not test multicast on Darwin

Do not run this test on Darwin, otherwise we get:

  qemu-system-arm: -netdev dgram,id=st0,remote.type=inet,remote.host=230.0.0.1,remote.port=1234:
   can't add socket to multicast group 230.0.0.1: Can't assign requested address
  Broken pipe
  ../../tests/qtest/libqtest.c:191: kill_qemu() tried to terminate QEMU
   process but encountered exit status 1 (expected 0)
  Abort trap: 6

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230918062549.2363-1-philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest/m48t59-test: Silence compiler warning with -Wshadow

When compiling this file with -Wshadow=local , we get:

../tests/qtest/m48t59-test.c: In function ‘bcd_check_time’:
../tests/qtest/m48t59-test.c:195:17: warning: declaration of ‘s’
 shadows a previous local [-Wshadow=local]
  195 |         long t, s;
      |                 ^
../tests/qtest/m48t59-test.c:158:17: note: shadowed declaration is here
  158 |     QTestState *s = m48t59_qtest_start();
      |                 ^

Rename the QTestState variable to "qts" which is the common
naming for such a variable in other tests.

Reported-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20230922163742.149444-1-thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: "Daniel P. Berrangé" <berrange@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

tests/qtest/netdev-socket: Raise connection timeout to 120 seconds

The test still fails intermittently with a 60 second timeout in the
GitLab CI environment. Raise the timeout to 120 seconds.

  576/839 ERROR:../tests/qtest/netdev-socket.c:293:test_stream_unix:
   assertion failed (resp == expect):
   ("st0: index=0,type=stream,connection error\r\n" == "st0: index=0,type=stream,unix:/tmp/netdev-socket.UW5IA2/stream_unix\r\n") ERROR
  576/839 qemu:qtest+qtest-sh4 / qtest-sh4/netdev-socket
                              ERROR          62.85s   killed by signal 6 SIGABRT
  >>> MALLOC_PERTURB_=249 QTEST_QEMU_BINARY=./qemu-system-sh4
   QTEST_QEMU_STORAGE_DAEMON_BINARY=./storage-daemon/qemu-storage-daemon
   G_TEST_DBUS_DAEMON=/home/gitlab-runner/builds/-LCfcJ2T/0/qemu-project/qemu/tests/dbus-vmstate-daemon.sh
   QTEST_QEMU_IMG=./qemu-img /home/gitlab-runner/builds/-LCfcJ2T/0/qemu-project/qemu/build/tests/qtest/netdev-socket --tap -k
  ――――――――――――――――――――――――――――――――――――― ✀  ―――――――――――――――――――――――――――――――――――――
  stderr:
  **
  ERROR:../tests/qtest/netdev-socket.c:293:test_stream_unix: assertion failed (resp == expect): ("st0: index=0,type=stream,connection error\r\n" == "st0: index=0,type=stream,unix:/tmp/netdev-socket.UW5IA2/stream_unix\r\n")
  (test program exited with status code -6)

Buglink: https://gitlab.com/qemu-project/qemu/-/issues/1881
Fixes: 417296c8d858 ("tests/qtest/netdev-socket: Raise connection timeout to 60 seconds")
Signed-off-by: Stefan Hajnoczi <stefanha@redhat.com>
Reviewed-by: Laurent Vivier <lvivier@redhat.com>
Reviewed-by: "Daniel P. Berrangé" <berrange@redhat.com>
Message-ID: <20230912133310.60583-1-stefanha@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

meson.build: Make keyutils independent from keyring

Commit 0db0fbb5cf ("Add conditional dependency for libkeyutils")
tried to provide a possibility for the user to disable keyutils
if not required by makeing it depend on the keyring feature. This
looked reasonable at a first glance (the unit test in tests/unit/
needs both), but the condition in meson.build fails if the feature
is meant to be detected automatically, and there is also another
spot in backends/meson.build where keyutils is used independently
from keyring. So let's remove the dependency on keyring again and
introduce a proper meson build option instead.

Cc: qemu-stable@nongnu.org
Fixes: 0db0fbb5cf ("Add conditional dependency for libkeyutils")
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1842
Message-ID: <20230824094208.255279-1-thuth@redhat.com>
Reviewed-by: "Daniel P. Berrangé" <berrange@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>

nbd: Add types for extended headers

Add the constants and structs necessary for later patches to start
implementing the NBD_OPT_EXTENDED_HEADERS extension in both the client
and server, matching recent upstream nbd.git (through commit
e6f3b94a934).  This patch does not change any existing behavior, but
merely sets the stage for upcoming patches.

This patch does not change the status quo that neither the client nor
server use a packed-struct representation for the request header.
While most of the patch adds new types, there is also some churn for
renaming the existing NBDExtent to NBDExtent32 to contrast it with
NBDExtent64, which I thought was a nicer name than NBDExtentExt.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230829175826.377251-22-eblake@redhat.com>

nbd/client: Pass mode through to nbd_send_request

Once the 64-bit headers extension is enabled, the data layout we send
over the wire for a client request depends on the mode negotiated with
the server.  Rather than adding a parameter to nbd_send_request, we
can add a member to struct NBDRequest, since it already does not
reflect on-wire format.  Some callers initialize it directly; many
others rely on a common initialization point during
nbd_co_send_request().  At this point, there is no semantic change.

Signed-off-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230829175826.377251-21-eblake@redhat.com>

nbd: Replace bool structured_reply with mode enum

The upcoming patches for 64-bit extensions requires various points in
the protocol to make decisions based on what was negotiated.  While we
could easily add a 'bool extended_headers' alongside the existing
'bool structured_reply', this does not scale well if more modes are
added in the future.  Better is to expose the mode enum added in the
recent commit bfe04d0a7d out to a wider use in the code base.

Where the code previously checked for structured_reply being set or
clear, it now prefers checking for an inequality; this works because
the nodes are in a continuum of increasing abilities, and allows us to
touch fewer places if we ever insert other modes in the middle of the
enum.  There should be no semantic change in this patch.

Signed-off-by: Eric Blake <eblake@redhat.com>
Message-ID: <20230829175826.377251-20-eblake@redhat.com>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>

iotests: improve 'not run' message for nbd-multiconn test

The test actually requires Python bindings to libnbd rather than libnbd
itself. Clarify that inside the message.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Hanna Reitz <hreitz@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230906140917.559129-3-den@openvz.org>
Reviewed-by: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

iotests: use TEST_IMG_FILE instead of TEST_IMG in _require_large_file

We need to check that we are able to create large enough file which is
used as an export base rather than connection URL. Unfortunately, there
are cases when the TEST_IMG_FILE is not defined. We should fallback to
TEST_IMG in that case.

This problem has been detected when running
    ./check -nbd 5
The test should be able to run while it does not.

Signed-off-by: Denis V. Lunev <den@openvz.org>
CC: Kevin Wolf <kwolf@redhat.com>
CC: Hanna Reitz <hreitz@redhat.com>
CC: Eric Blake <eblake@redhat.com>
CC: Vladimir Sementsov-Ogievskiy <vsementsov@yandex-team.ru>
Message-ID: <20230906140917.559129-2-den@openvz.org>
Tested-by: Eric Blake <eblake@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
Signed-off-by: Eric Blake <eblake@redhat.com>

vl: recognize audiodev groups in configuration files

This is necessary for the q35 configuration tests to pass,
once audiodev becomes mandatory.

Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

tests/qtest: Specify audiodev= and -audiodev

This will enable removing deprecated default audiodev support.

I did not figure out how to make the audiodev represented as an
interface node, so this is a workaround.  I am not sure what would be
the proper way.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <6e7f2808dd40679a415812767b88f2a411fc137f.1650874791.git.mkletzan@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/display/xlnx_dp.c: Add audiodev property

There was no way to set this and we need that for it to be able to properly
initialise.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Message-ID: <16963256573fcbfa7720aa2fd000ba74a4055222.1650874791.git.mkletzan@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/audio/lm4549: Add errp error reporting to init function

This will be used in future commit.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <be1bf295b3c6a3dee272b4b4e8115e37c2a772b5.1650874791.git.mkletzan@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/audio: Simplify hda audio init

No return values are used anywhere, so switch the functions to be void
and add support for error reporting using errp for use in next patches.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <cd1df4ad2a6fae969c4a02a77955c4a8c0d430b6.1650874791.git.mkletzan@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

hw/input/tsc210x: Extract common init code into new function

This deduplicates several lines and will make future changes more
concise.

Signed-off-by: Martin Kletzander <mkletzan@redhat.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <1d75877cf4cc2a38f87633ff16f9fea3e1bb0c03.1650874791.git.mkletzan@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

elf2dmp: rework PDB_STREAM_INDEXES::segments obtaining

PDB for Windows 11 kernel has slightly different structure compared to
previous versions. Since elf2dmp don't use the other fields, copy only
'segments' field from PDB_STREAM_INDEXES.

Signed-off-by: Viktor Prutyanov <viktor@daynix.com>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230915170153.10959-6-viktor@daynix.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

elf2dmp: use Linux mmap with MAP_NORESERVE when possible

Glib's g_mapped_file_new maps file with PROT_READ|PROT_WRITE and
MAP_PRIVATE. This leads to premature physical memory allocation of dump
file size on Linux hosts and may fail. On Linux, mapping the file with
MAP_NORESERVE limits the allocation by available memory.

Signed-off-by: Viktor Prutyanov <viktor@daynix.com>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230915170153.10959-5-viktor@daynix.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

elf2dmp: introduce merging of physical memory runs

DMP supports 42 physical memory runs at most. So, merge adjacent
physical memory ranges from QEMU ELF when possible to minimize total
number of runs.

Signed-off-by: Viktor Prutyanov <viktor@daynix.com>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230915170153.10959-4-viktor@daynix.com
[PMM: fixed format string for printing size_t values]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

elf2dmp: introduce physical block alignment

Physical memory ranges may not be aligned to page size in QEMU ELF, but
DMP can only contain page-aligned runs. So, align them.

Signed-off-by: Viktor Prutyanov <viktor@daynix.com>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230915170153.10959-3-viktor@daynix.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

elf2dmp: replace PE export name check with PDB name check

PE export name check introduced in d399d6b179 isn't reliable enough,
because a page with the export directory may be not present for some
reason. On the other hand, elf2dmp retrieves the PDB name in any case.
It can be also used to check that a PE image is the kernel image. So,
check PDB name when searching for Windows kernel image.

Buglink: https://bugzilla.redhat.com/show_bug.cgi?id=2165917
Signed-off-by: Viktor Prutyanov <viktor@daynix.com>
Reviewed-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Message-id: 20230915170153.10959-2-viktor@daynix.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

sbsa-ref: add non-secure EL2 virtual timer

Armv8.1+ cpus have Virtual Host Extension (VHE) which added non-secure
EL2 virtual timer.

This change adds it to fullfil Arm BSA (Base System Architecture)
requirements.

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Message-id: 20230913140610.214893-2-marcin.juszkiewicz@linaro.org
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

audio/jackaudio: Avoid dynamic stack allocation in qjack_process()

Avoid a dynamic stack allocation in qjack_process().  Since this
function is a JACK process callback, we are not permitted to malloc()
here, so we allocate a working buffer in qjack_client_init() instead.

The codebase has very few VLAs, and if we can get rid of them all we
can make the compiler error on new additions.  This is a defensive
measure against security bugs where an on-stack dynamic allocation
isn't correctly size-checked (e.g.  CVE-2021-3527).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-id: 20230818155846.1651287-3-peter.maydell@linaro.org

audio/jackaudio: Avoid dynamic stack allocation in qjack_client_init

Avoid a dynamic stack allocation in qjack_client_init(), by using
a g_autofree heap allocation instead.

(We stick with allocate + snprintf() because the JACK API requires
the name to be no more than its maximum size, so g_strdup_printf()
would require an extra truncation step.)

The codebase has very few VLAs, and if we can get rid of them all we
can make the compiler error on new additions.  This is a defensive
measure against security bugs where an on-stack dynamic allocation
isn't correctly size-checked (e.g.  CVE-2021-3527).

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-id: 20230818155846.1651287-2-peter.maydell@linaro.org

target/arm: Enable FEAT_MOPS for CPU 'max'

Enable FEAT_MOPS on the AArch64 'max' CPU, and add it to
the list of features we implement.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20230912140434.1333369-13-peter.maydell@linaro.org