target/riscv: add riscv,isa to named features

Further discussions after the introduction of rva22 support in QEMU
revealed that what we've been calling 'named features' are actually
regular extensions, with their respective riscv,isa DTs. This is
clarified in [1]. [2] is a bug tracker asking for the profile spec to be
less cryptic about it.

As far as QEMU goes we understand extensions as something that the user
can enable/disable in the command line. This isn't the case for named
features, so we'll have to reach a middle ground.

We'll keep our existing nomenclature 'named features' to refer to any
extension that the user can't control in the command line. We'll also do
the following:

- 'svade' and 'zic64b' flags are renamed to 'ext_svade' and
  'ext_zic64b'. 'ext_svade' and 'ext_zic64b' now have riscv,isa strings and
  priv_spec versions;

- skip name feature check in cpu_bump_multi_ext_priv_ver(). Now that
  named features have a riscv,isa and an entry in isa_edata_arr[] we
  don't need to gate the call to cpu_cfg_ext_get_min_version() anymore.

[1] https://github.com/riscv/riscv-profiles/issues/121
[2] https://github.com/riscv/riscv-profiles/issues/142

Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Andrew Jones <ajones@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240215223955.969568-3-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv/tcg: set 'mmu' with 'satp' in cpu_set_profile()

Recent changes in options handling removed the 'mmu' default the bare
CPUs had, meaning that we must enable 'mmu' by hand when using the
rva22s64 profile CPU.

Given that this profile is setting a satp mode, it already implies that
we need a 'mmu'. Enable the 'mmu' in this case.

Signed-off-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240215223955.969568-2-dbarboza@ventanamicro.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

linux-user/riscv: Sync hwprobe keys with Linux

Upstream Linux recently added many additional keys to the hwprobe API.
This patch adds support for all of them with the exception of Ztso,
which is currently not supported in QEMU.

Signed-off-by: Christoph Müllner <christoph.muellner@vrull.eu>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20240207115926.887816-3-christoph.muellner@vrull.eu>
[ Changes by AF:
 - Fixup whitespace
]
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

linux-user/riscv: Add Zicboz extensions to hwprobe

Upstream Linux recently added RISC-V Zicboz support to the hwprobe API.
This patch introduces this for QEMU's user space emulator.

Signed-off-by: Christoph Müllner <christoph.muellner@vrull.eu>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20240207115926.887816-2-christoph.muellner@vrull.eu>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw: riscv: Allow large kernels to boot by moving the initrd further away in RAM

Currently, the initrd is placed at 128MB, which overlaps with the kernel
when it is large (for example syzbot kernels are). From the kernel side,
there is no reason we could not push the initrd further away in memory
to accommodate large kernels, so move the initrd at 512MB when possible.

The ideal solution would have been to place the initrd based on the
kernel size but we actually can't since the bss size is not known when
the image is loaded by load_image_targphys_as() and the initrd would
then overlap with this section.

Signed-off-by: Alexandre Ghiti <alexghiti@rivosinc.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20240206154042.514698-1-alexghiti@rivosinc.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/riscv/virt-acpi-build.c: Generate SPCR table

Generate Serial Port Console Redirection Table (SPCR) for RISC-V
virtual machine.

Signed-off-by: Sia Jee Heng <jeeheng.sia@starfivetech.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Message-ID: <20240129021440.17640-3-jeeheng.sia@starfivetech.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

hw/arm/virt-acpi-build.c: Migrate SPCR creation to common location

RISC-V should also generate the SPCR in a manner similar to ARM.
Therefore, instead of replicating the code, relocate this function
to the common AML build.

Signed-off-by: Sia Jee Heng <jeeheng.sia@starfivetech.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-ID: <20240129021440.17640-2-jeeheng.sia@starfivetech.com>
[ Changes by AF:
 - Add missing Language SPCR entry
]
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

target/riscv: Update $ra with current $pc in trans_cm_jalt()

The original implementation sets $pc to the address read from the jump
vector table first and links $ra with the address of the next instruction
after the updated $pc. After jumping to the updated $pc and executing the
next ret instruction, the program jumps to $ra, which is in the same
function currently executing, which results in an infinite loop.
This commit stores the jump address in a temporary, updates $ra with the
current $pc, and copies the temporary to $pc.

Signed-off-by: Jason Chien <jason.chien@sifive.com>
Reviewed-by: Frank Chang <frank.chang@sifive.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20240207081820.28559-1-jason.chien@sifive.com>
Signed-off-by: Alistair Francis <alistair.francis@wdc.com>

Merge tag 'pull-maintainer-updates-060324-1' of https://gitlab.com/stsquad/qemu into staging

maintainer updates (tests, gdbstub, plugins):

  - expand QOS_PATH_MAX_ELEMENT_SIZE to avoid LTO issues
  - support fork-follow-mode in gdbstub
  - new thread-safe scoreboard API for TCG plugins
  - suppress showing opcodes in plugin disassembly

# -----BEGIN PGP SIGNATURE-----
#
# iQEzBAABCgAdFiEEZoWumedRZ7yvyN81+9DbCVqeKkQFAmXoY7oACgkQ+9DbCVqe
# KkTdTwf8D8nUB+Ee6LuglW36vtd1ETdMfUmfRis7RIBsXZZ0Tg4+8LyfKkNi1vCL
# UMdWQTkSW79RfXr21QEtETokwLZ0CWQMdxDAWfOiz4S+uDgQyBE+lwUsy0mHBmd7
# +J4SQb3adoZ+//9KMJhRU1wL9j3ygpEoKHVJonDObU6K5XuhE18JuBE44q7FqkWl
# 0VhoLDgNxrf2PqT+LLP/O3MFLDXPVKbzrZYQF0IoqBTlcqShCoaykhSwiwCZ4Sqq
# NO9hVwZIOFOcOF4F6ZqRXaZrwERldoBwG+BeIx1ah20vKFVT12y02dQqdP/oKwe+
# /PXFXDdzs4yMOghb4Go6SiKlKT5g4A==
# =s1lF
# -----END PGP SIGNATURE-----
# gpg: Signature made Wed 06 Mar 2024 12:38:18 GMT
# gpg:                using RSA key 6685AE99E75167BCAFC8DF35FBD0DB095A9E2A44
# gpg: Good signature from "Alex Bennée (Master Work Key) <alex.bennee@linaro.org>" [full]
# Primary key fingerprint: 6685 AE99 E751 67BC AFC8  DF35 FBD0 DB09 5A9E 2A44

* tag 'pull-maintainer-updates-060324-1' of https://gitlab.com/stsquad/qemu: (29 commits)
  target/riscv: honour show_opcodes when disassembling
  target/loongarch: honour show_opcodes when disassembling
  disas/hppa: honour show_opcodes
  disas: introduce show_opcodes
  plugins: cleanup codepath for previous inline operation
  plugins: remove non per_vcpu inline operation from API
  contrib/plugins/howvec: migrate to new per_vcpu API
  contrib/plugins/hotblocks: migrate to new per_vcpu API
  tests/plugin/bb: migrate to new per_vcpu API
  tests/plugin/insn: migrate to new per_vcpu API
  tests/plugin/mem: migrate to new per_vcpu API
  tests/plugin: add test plugin for inline operations
  plugins: add inline operation per vcpu
  plugins: implement inline operation relative to cpu_index
  plugins: define qemu_plugin_u64
  plugins: scoreboard API
  tests/tcg: Add two follow-fork-mode tests
  gdbstub: Implement follow-fork-mode child
  gdbstub: Introduce gdb_handle_detach_user()
  gdbstub: Introduce gdb_handle_set_thread_user()
  ...

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

target/riscv: honour show_opcodes when disassembling

This makes the output suitable when used for plugins.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-30-alex.bennee@linaro.org>

target/loongarch: honour show_opcodes when disassembling

This makes the output suitable when used for plugins.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-29-alex.bennee@linaro.org>

disas/hppa: honour show_opcodes

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-28-alex.bennee@linaro.org>

disas: introduce show_opcodes

For plugins we don't expect the raw opcodes in the disassembly. We
already deal with this by hand crafting our capstone call but for
other diassemblers we need a flag. Introduce show_opcodes which
defaults to off.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-27-alex.bennee@linaro.org>

plugins: cleanup codepath for previous inline operation

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-13-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-26-alex.bennee@linaro.org>

plugins: remove non per_vcpu inline operation from API

Now we have a thread-safe equivalent of inline operation, and that all
plugins were changed to use it, there is no point to keep the old API.

In more, it will help when we implement more functionality (conditional
callbacks), as we can assume that we operate on a scoreboard.

API version bump was already done as part of this series.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-12-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-25-alex.bennee@linaro.org>

contrib/plugins/howvec: migrate to new per_vcpu API

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Tested-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-11-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-24-alex.bennee@linaro.org>

contrib/plugins/hotblocks: migrate to new per_vcpu API

Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-10-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-23-alex.bennee@linaro.org>

tests/plugin/bb: migrate to new per_vcpu API

Reviewed-by: Luc Michel <luc.michel@amd.com>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-9-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-22-alex.bennee@linaro.org>

tests/plugin/insn: migrate to new per_vcpu API

Reviewed-by: Luc Michel <luc.michel@amd.com>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-8-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-21-alex.bennee@linaro.org>

tests/plugin/mem: migrate to new per_vcpu API

Reviewed-by: Luc Michel <luc.michel@amd.com>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-7-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-20-alex.bennee@linaro.org>

tests/plugin: add test plugin for inline operations

For now, it simply performs instruction, bb and mem count, and ensure
that inline vs callback versions have the same result. Later, we'll
extend it when new inline operations are added.

Use existing plugins to test everything works is a bit cumbersome, as
different events are treated in different plugins. Thus, this new one.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-6-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-19-alex.bennee@linaro.org>

plugins: add inline operation per vcpu

Extends API with three new functions:
qemu_plugin_register_vcpu_{tb, insn, mem}_exec_inline_per_vcpu().

Those functions takes a qemu_plugin_u64 as input.

This allows to have a thread-safe and type-safe version of inline
operations.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-5-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-18-alex.bennee@linaro.org>

plugins: implement inline operation relative to cpu_index

Instead of working on a fixed memory location, allow to address it based
on cpu_index, an element size and a given offset.
Result address: ptr + offset + cpu_index * element_size.

With this, we can target a member in a struct array from a base pointer.

Current semantic is not modified, thus inline operation still targets
always the same memory location.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-4-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-17-alex.bennee@linaro.org>

plugins: define qemu_plugin_u64

Additionally to the scoreboard, we define a qemu_plugin_u64, which is a
simple struct holding a pointer to a scoreboard, and a given offset.
This allows to have a scoreboard containing structs, without having to
bring offset to operate on a specific field.

Since most of the plugins are simply collecting a sum of per-cpu values,
qemu_plugin_u64 directly support this operation as well.

All inline operations defined later will use a qemu_plugin_u64 as input.

New functions:
- qemu_plugin_u64_add
- qemu_plugin_u64_get
- qemu_plugin_u64_set
- qemu_plugin_u64_sum
New macros:
- qemu_plugin_scoreboard_u64
- qemu_plugin_scoreboard_u64_in_struct

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-3-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-16-alex.bennee@linaro.org>

plugins: scoreboard API

We introduce a cpu local storage, automatically managed (and extended)
by QEMU itself. Plugin allocate a scoreboard, and don't have to deal
with how many cpus are launched.

This API will be used by new inline functions but callbacks can benefit
from this as well. This way, they can operate without a global lock for
simple operations.

At any point during execution, any scoreboard will be dimensioned with
at least qemu_plugin_num_vcpus entries.

New functions:
- qemu_plugin_scoreboard_find
- qemu_plugin_scoreboard_free
- qemu_plugin_scoreboard_new

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Pierrick Bouvier <pierrick.bouvier@linaro.org>
Message-Id: <20240304130036.124418-2-pierrick.bouvier@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-15-alex.bennee@linaro.org>

tests/tcg: Add two follow-fork-mode tests

Add follow-fork-mode child and and follow-fork-mode parent tests.
Check for the obvious pitfalls, such as lingering breakpoints,
catchpoints, and single-step mode.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-13-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-14-alex.bennee@linaro.org>

gdbstub: Implement follow-fork-mode child

Currently it's not possible to use gdbstub for debugging linux-user
code that runs in a forked child, which is normally done using the `set
follow-fork-mode child` GDB command. Purely on the protocol level, the
missing piece is the fork-events feature.

However, a deeper problem is supporting $Hg switching between different
processes - right now it can do only threads. Implementing this for the
general case would be quite complicated, but, fortunately, for the
follow-fork-mode case there are a few factors that greatly simplify
things: fork() happens in the exclusive section, there are only two
processes involved, and before one of them is resumed, the second one
is detached.

This makes it possible to implement a simplified scheme: the parent and
the child share the gdbserver socket, it's used only by one of them at
any given time, which is coordinated through a separate socketpair. The
processes can read from the gdbserver socket only one byte at a time,
which is not great for performance, but, fortunately, the
follow-fork-mode handling involves only a few messages.

Advertise the fork-events support, and remember whether GDB has it
as well. Implement the state machine that is initialized on fork(),
decides the current owner of the gdbserver socket, and is terminated
when one of the two processes is detached. The logic for the parent and
the child is the same, only the initial state is different.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-12-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-13-alex.bennee@linaro.org>

gdbstub: Introduce gdb_handle_detach_user()

The upcoming follow-fork-mode child support needs to perform certain
actions when GDB detaches from the stopped parent or the stopped child.
Introduce a user-specific hook for this.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-11-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-12-alex.bennee@linaro.org>

gdbstub: Introduce gdb_handle_set_thread_user()

The upcoming follow-fork-mode child support needs to perform certain
actions when GDB switches between the stopped parent and the stopped
child. Introduce a user-specific hook for this.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-10-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-11-alex.bennee@linaro.org>

gdbstub: Introduce gdb_handle_query_supported_user()

The upcoming follow-fork-mode child support requires advertising the
fork-events feature, which is user-specific. Introduce a user-specific
hook for this.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-9-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-10-alex.bennee@linaro.org>

gdbstub: Call gdbserver_fork() both in parent and in child

The upcoming follow-fork-mode child support requires post-fork message
exchange between the parent and the child. Prepare gdbserver_fork() for
this purpose. Rename it to gdbserver_fork_end() to better reflect its
purpose.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-8-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-9-alex.bennee@linaro.org>

{linux,bsd}-user: Pass pid to gdbserver_fork()

The upcoming follow-fork-mode child support requires knowing the child
pid. Pass it down.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-7-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-8-alex.bennee@linaro.org>

{linux,bsd}-user: Pass pid to fork_end()

The upcoming follow-fork-mode child support requires knowing the child
pid. Pass it down.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-6-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-7-alex.bennee@linaro.org>

gdbstub: Introduce gdbserver_fork_start()

The upcoming follow-fork-mode child support requires knowing when
fork() is about to happen in order to initialize its state. Add a hook
for that.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-5-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-6-alex.bennee@linaro.org>

{linux,bsd}-user: Update ts_tid after fork()

Currently ts_tid contains the parent tid after fork(), which is not
correct. So far it has not affected anything, but the upcoming
follow-fork-mode child support relies on the correct value, so fix it.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Message-Id: <20240219141628.246823-4-iii@linux.ibm.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-5-alex.bennee@linaro.org>

{linux,bsd}-user: Introduce get_task_state()

A CPU's TaskState is stored in the CPUState's void *opaque field,
accessing which is somewhat awkward due to having to use a cast.
Introduce a wrapper and use it everywhere.

Suggested-by: Alex Bennée <alex.bennee@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Reviewed-by: Warner Losh <imp@bsdimp.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20240219141628.246823-3-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-4-alex.bennee@linaro.org>

gdbstub: Support disablement in a multi-threaded process

The upcoming follow-fork-mode child support will require disabling
gdbstub in the parent process, which may have multiple threads (which
are represented as CPUs).

Loop over all CPUs in order to remove breakpoints and disable
single-step. Move the respective code into a separate function.

Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <20240219141628.246823-2-iii@linux.ibm.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-3-alex.bennee@linaro.org>

tests: bump QOS_PATH_MAX_ELEMENT_SIZE again

We "fixed" a bug with LTO builds with 100c459f194 (tests/qtest: bump
up QOS_PATH_MAX_ELEMENT_SIZE) but it seems it has triggered again.

The array is sized according to the maximum anticipated length of a
path on the graph. However, the worst case for a depth-first search is
to push all nodes on the graph. So it's not really LTO, it depends on
the ordering of the constructors.

Lets be more assertive raising QOS_PATH_MAX_ELEMENT_SIZE to make it go
away again.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1186 (again)
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Alex Bennée <alex.bennee@linaro.org>
Message-Id: <20240305121005.3528075-2-alex.bennee@linaro.org>

Merge tag 'pull-target-arm-20240305' of https://git.linaro.org/people/pmaydell/qemu-arm into staging

target-arm queue:
 * raspi: Implement Broadcom Serial Controller (BSC) for BCM2835 boards
 * hw/char/pl011: Add support for loopback
 * STM32L4x5: Implement RCC clock control device
 * target/arm: Do memory type alignment checks
 * atomic.h: Reword confusing comment for qatomic_cmpxchg
 * qemu-options.hx: Don't claim "-serial" has limit of 4 serial ports

# -----BEGIN PGP SIGNATURE-----
#
# iQJNBAABCAA3FiEE4aXFk81BneKOgxXPPCUl7RQ2DN4FAmXnI4gZHHBldGVyLm1h
# eWRlbGxAbGluYXJvLm9yZwAKCRA8JSXtFDYM3p5ED/wOtAHA3PK+WbQhVhnji3+k
# gdhvVcldf+HLaI2v4tfaW152xjY80/j3PQFNkzptoXENA9U51F47oNYOIfULLZZX
# FTKmw/mjTBc2LqJ8gLpS8Wkr/PFtDq9JJzDwZd0MwguXpzIJp31JJpESvXlAqjjv
# FhuAcqNNuGwI2SXCBmp2lPoEMn8ExLDoG9rmzjxVZeZCyzUjVnJYM61ykhC4ByvK
# j5+/a7pUcpgHSX5cbq7kFloPOx3JXI5lS6xUKhGXXk75qHRwiQIsxMcPq8PD1+ok
# yrmp7cySwK8I7AlIPdDjpJmhU0OiBu+PkYiXmHlF2nvaUy6M0nVX2lSTzqj6VpVV
# 7yYhvWXHrtIA9AUspqTRsX7tP7iMJkco7qWfKSzYl+3pTbxS4+rEoee4jNR3hqsU
# lbWC47sNVtTN507qIL1dcsu+BaeSsYVftfxtFql3odTqRB+ticsjDfKg69dRSFyk
# SS0t8Zy3TdomcEoQkAv/ZSpkQnQUGavbRumCG58lJdiTwTuJUmGi1ufKBrD/GeKj
# IlDEl9yvKiR8uvdjj6EQqr5kOj09mmN5nvokNsq5a4aNXBYoesszWK2xodzXE2x5
# M9DHJ3S8xnN++p1idS2bikwEklG1XVQ/q52bDXQkUmQSNerVS1PCvg9hzYqA+x53
# ihJtMcsmGVfxY8aQHyHweA==
# =isAe
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 05 Mar 2024 13:52:08 GMT
# 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-20240305' of https://git.linaro.org/people/pmaydell/qemu-arm:
  qemu-options.hx: Don't claim "-serial" has limit of 4 serial ports
  atomic.h: Reword confusing comment for qatomic_cmpxchg
  target/arm: Do memory type alignment check when translation enabled
  target/arm: Do memory type alignment check when translation disabled
  accel/tcg: Add TLB_CHECK_ALIGNED
  accel/tcg: Add tlb_fill_flags to CPUTLBEntryFull
  exec/memattrs: Remove target_tlb_bit*
  target/arm: Support 32-byte alignment in pow2_align
  tests/qtest/stm32l4x5_rcc-test.c: Add tests for the STM32L4x5_RCC
  hw/arm/stm32l4x5_soc.c: Use the RCC Sysclk
  hw/misc/stm32l4x5_rcc: Add write protections to CR register
  hw/misc/stm32l4x5_rcc: Handle Register Updates
  hw/misc/stm32l4x5_rcc: Initialize PLLs and clock multiplexers
  hw/misc/stm32l4x5_rcc: Add an internal PLL Clock object
  hw/misc/stm32l4x5_rcc: Add an internal clock multiplexer object
  hw/misc/stm32l4x5_rcc: Implement STM32L4x5_RCC skeleton
  hw/char/pl011: Add support for loopback
  tests/qtest: Add testcase for BCM2835 BSC
  hw/arm: Connect BSC to BCM2835 board as I2C0, I2C1 and I2C2
  hw/i2c: Implement Broadcom Serial Controller (BSC)

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

qemu-options.hx: Don't claim "-serial" has limit of 4 serial ports

Before v2.12, the implementation of serial ports was limited to
a value of MAX_SERIAL_PORTS = 4. We now dynamically allocate
the data structures for serial ports, so this limit is no longer
present, but the documentation for the -serial options still reads:

 "This option can be used several times to simulate up to 4 serial ports."

Update to "This option can be used several times to simulate
multiple serial ports." to avoid misleading.

Signed-off-by: Steven Shen <steven.shen@jaguarmicro.com>
Message-id: 20240305013016.2268-1-steven.shen@jaguarmicro.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
[PMM: tweaked commit message]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

atomic.h: Reword confusing comment for qatomic_cmpxchg

The qatomic_cmpxchg() and qatomic_cmpxchg__nocheck() macros have
a comment that reads:
 Returns the eventual value, failed or not

This is somewhere between cryptic and wrong, since the value actually
returned is the value that was in memory before the cmpxchg.  Reword
to match how we describe these macros in atomics.rst.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-id: 20240223182035.1048541-1-peter.maydell@linaro.org

target/arm: Do memory type alignment check when translation enabled

If translation is enabled, and the PTE memory type is Device,
enable checking alignment via TLB_CHECK_ALIGNMENT.  While the
check is done later than it should be per the ARM, it's better
than not performing the check at all.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240301204110.656742-7-richard.henderson@linaro.org
[PMM: tweaks to comment text]
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Do memory type alignment check when translation disabled

If translation is disabled, the default memory type is Device, which
requires alignment checking.  This is more optimally done early via
the MemOp given to the TCG memory operation.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reported-by: Idan Horowitz <idan.horowitz@gmail.com>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240301204110.656742-6-richard.henderson@linaro.org
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1204
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

accel/tcg: Add TLB_CHECK_ALIGNED

This creates a per-page method for checking of alignment.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240301204110.656742-5-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

accel/tcg: Add tlb_fill_flags to CPUTLBEntryFull

Allow the target to set tlb flags to apply to all of the
comparators.  Remove MemTxAttrs.byte_swap, as the bit is
not relevant to memory transactions, only the page mapping.
Adjust target/sparc to set TLB_BSWAP directly.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240301204110.656742-4-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

exec/memattrs: Remove target_tlb_bit*

These fields are no longer used since 937f224559.
Target specific extensions to the page tables should be done
with TARGET_PAGE_ENTRY_EXTRA.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240301204110.656742-3-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

target/arm: Support 32-byte alignment in pow2_align

Now that we have removed TARGET_PAGE_BITS_MIN-6 from
TLB_FLAGS_MASK, we can test for 32-byte alignment.

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Message-id: 20240301204110.656742-2-richard.henderson@linaro.org
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/qtest/stm32l4x5_rcc-test.c: Add tests for the STM32L4x5_RCC

Tests:
- the ability to change the sysclk of the device
- the ability to enable/disable/configure the PLLs
- if the clock multiplexers work
- the register flags and the generation of irqs

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Acked-by: Thomas Huth <thuth@redhat.com>
Message-id: 20240303140643.81957-9-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm/stm32l4x5_soc.c: Use the RCC Sysclk

Now that we can generate reliable clock frequencies from the RCC, remove
the hacky definition of the sysclk in the b_l475e_iot01a initialisation
code and use the correct RCC clock.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240303140643.81957-8-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/stm32l4x5_rcc: Add write protections to CR register

Add write protections for the fields in the CR register.
PLL configuration write protections (among others) have not
been handled yet. This is planned in a future patch set.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Message-id: 20240303140643.81957-7-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/stm32l4x5_rcc: Handle Register Updates

Update the RCC state and propagate frequency changes when writing to the
RCC registers. Currently, ICSCR, CIER, the reset registers and the stop
mode registers are not implemented.

Some fields  have not been implemented due to uncertainty about
how to handle them (Like the clock security system or bypassing
mecanisms).

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Message-id: 20240303140643.81957-6-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/stm32l4x5_rcc: Initialize PLLs and clock multiplexers

Instantiate the whole clock tree and using the Clock multiplexers and
the PLLs defined in the previous commits. This allows to statically
define the clock tree and easily follow the clock signal from one end to
another.

Also handle three-phase reset now that we have defined a known base
state for every object.
(Reset handling based on hw/misc/zynq_sclr.c)

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Message-id: 20240303140643.81957-5-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/stm32l4x5_rcc: Add an internal PLL Clock object

This object represents the PLLs and their channels. The PLLs allow for a
more fine-grained control of the clocks frequency.

The migration handling is based on hw/misc/zynq_sclr.c.
Three phase reset will be handled in a later commit.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Message-id: 20240303140643.81957-4-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/stm32l4x5_rcc: Add an internal clock multiplexer object

This object is used to represent every multiplexer in the clock tree as
well as every clock output, every presecaler, frequency multiplier, etc.
This allows to use a generic approach for every component of the clock tree
(except the PLLs).

The migration handling is based on hw/misc/zynq_sclr.c.
Three phase reset will be handled in a later commit.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-id: 20240303140643.81957-3-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>

hw/misc/stm32l4x5_rcc: Implement STM32L4x5_RCC skeleton

Add the necessary files to add a simple RCC implementation with just
reads from and writes to registers. Also instantiate the RCC in the
STM32L4x5_SoC. It is needed for accurate emulation of all the SoC
clocks and timers.

Signed-off-by: Arnaud Minier <arnaud.minier@telecom-paris.fr>
Signed-off-by: Inès Varhol <ines.varhol@telecom-paris.fr>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240303140643.81957-2-arnaud.minier@telecom-paris.fr
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/char/pl011: Add support for loopback

This patch adds loopback for sent characters, sent BREAK,
and modem-control signals.

Loopback of send and modem-control is often used for uart
self tests in real hardware but missing from current pl011
model, resulting in self-test failures when running in QEMU.

This implementation matches what is observed in real pl011
hardware placed in loopback mode:
1. Input characters and BREAK events from serial backend
   are ignored, but
2. Both TX characters and BREAK events are still sent to
   serial backend, in addition to be looped back to RX.

Signed-off-by: Tong Ho <tong.ho@amd.com>
Signed-off-by: Francisco Iglesias <francisco.iglesias@amd.com>
Message-id: 20240227054855.44204-1-tong.ho@amd.com
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

tests/qtest: Add testcase for BCM2835 BSC

Simple testcase for validating proper operation of read and write for all
three BSC controllers.

Signed-off-by: Rayhan Faizel <rayhan.faizel@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240224191038.2409945-4-rayhan.faizel@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/arm: Connect BSC to BCM2835 board as I2C0, I2C1 and I2C2

BCM2835 has three I2C controllers. All of them share the same interrupt line.

Signed-off-by: Rayhan Faizel <rayhan.faizel@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-id: 20240224191038.2409945-3-rayhan.faizel@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

hw/i2c: Implement Broadcom Serial Controller (BSC)

A few deficiencies in the current device model need to be noted.

1. FIFOs are not used. All sends and receives are done directly.
2. Repeated starts are not emulated. Repeated starts can be triggered in real
hardware by sending a new read transfer request in the window time between
transfer active set of write transfer request and done bit set of the same.

Signed-off-by: Rayhan Faizel <rayhan.faizel@gmail.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240224191038.2409945-2-rayhan.faizel@gmail.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

Merge tag 'pull-qapi-2024-03-04' of https://repo.or.cz/qemu/armbru into staging

QAPI patches patches for 2024-03-04

# -----BEGIN PGP SIGNATURE-----
#
# iQJGBAABCAAwFiEENUvIs9frKmtoZ05fOHC0AOuRhlMFAmXlaSISHGFybWJydUBy
# ZWRoYXQuY29tAAoJEDhwtADrkYZTdZ8P/iMgqLoAFkCCjwfkUc/rqZUezK52Ynr7
# LYwOPI/xcYD7EnVogdRgFgjWFNoivQLP5yKsU/eRTk29pwdDzTscFm/0ztTQX/Gb
# ypWV+GBcu5J8mKbp1KF5w68aDD8Bat4WRfEgDQ1DV7v6CoMiUzTiF3CGXkYzqK5Y
# kYNq97vdEkBFvFdOl/7scs/XXN2jG27egDhMp68RTxnPHlXZiAO9/2Bul3uVe3x0
# fzQ2ViYv0qLnjE/PwENDqqE3Thv3Sxp5iEeQQ6GWi07EVh07UtHpOM3RYyrTU0Sb
# VrTApSrg0oxlkOuR0CBd9Fi+timtbokBL0DWyUpXNTfIEZfLtA9H+8riUg3EOcDp
# r7a4SI/27VdPxX6Kc6zA3bi+/j1o7CLTW2LGEwuZs52nmixoo1HTWPIFdyh13g/V
# QjNbun0fViHb0FVLiyDlXF/7Y+EWUWIyqwwGqbvve1DyUHQmo3CUQAKGOpkeKSBe
# 4eGciVDgpBoKhtw9Kv6LCDj2cwZKC8DxBMibf7GHkOnAsX2mnyuHcey7HvYNCoF+
# yYz7oIEXdlL2eWqg7CfBZK7lniCDln50RI4Ll1v+J4r1v1kRZGMLesTYXCdNc4ku
# yb4kpU4t22/RODffLE7K+fc3Onwze3fcfxlZMN66F+wFtk4KdPR2aQBE66bB8J99
# vuSKlTbT4cGL
# =s9AR
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 04 Mar 2024 06:24:34 GMT
# gpg:                using RSA key 354BC8B3D7EB2A6B68674E5F3870B400EB918653
# gpg:                issuer "armbru@redhat.com"
# gpg: Good signature from "Markus Armbruster <armbru@redhat.com>" [full]
# gpg:                 aka "Markus Armbruster <armbru@pond.sub.org>" [full]
# Primary key fingerprint: 354B C8B3 D7EB 2A6B 6867  4E5F 3870 B400 EB91 8653

* tag 'pull-qapi-2024-03-04' of https://repo.or.cz/qemu/armbru:
  migration: simplify exec migration functions
  qapi: New strv_from_str_list()
  qapi: New QAPI_LIST_LENGTH()
  docs/devel/writing-monitor-commands: Minor improvements
  docs/devel/writing-monitor-commands: Repair a decade of rot
  qapi: Reject "Returns" section when command doesn't return anything
  qga/qapi-schema: Fix guest-set-memory-blocks documentation
  qga/qapi-schema: Tweak documentation of fsfreeze commands
  qga/qapi-schema: Clean up "Returns" sections
  qga/qapi-schema: Delete useless "Returns" sections
  qga/qapi-schema: Move error documentation to new "Errors" sections
  qapi/yank: Tweak @yank's error description for consistency
  qapi: Clean up "Returns" sections
  qapi: Delete useless "Returns" sections
  qapi: Move error documentation to new "Errors" sections
  qapi: New documentation section tag "Errors"
  qapi: Slightly clearer error message for invalid "Returns" section
  qapi: Memorize since & returns sections

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

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

Migartion pull request for 20240304

- Bryan's fix on multifd compression level API
- Fabiano's mapped-ram series (base + multifd only)
- Steve's amend on cpr document in qapi/

# -----BEGIN PGP SIGNATURE-----
#
# iIgEABYKADAWIQS5GE3CDMRX2s990ak7X8zN86vXBgUCZeUjKhIccGV0ZXJ4QHJl
# ZGhhdC5jb20ACgkQO1/MzfOr1wbv5QD/ZexBUsmZA5qyxgGvZ2yvlUBEGNOvtmKY
# kRdiYPU7khMA/0N43rn4LcqKCoq4+T+EAnYizGjIyhH/7BRUyn4DUxgO
# =AeEn
# -----END PGP SIGNATURE-----
# gpg: Signature made Mon 04 Mar 2024 01:26:02 GMT
# gpg:                using EDDSA key B9184DC20CC457DACF7DD1A93B5FCCCDF3ABD706
# gpg:                issuer "peterx@redhat.com"
# gpg: Good signature from "Peter Xu <xzpeter@gmail.com>" [marginal]
# gpg:                 aka "Peter Xu <peterx@redhat.com>" [marginal]
# gpg: WARNING: This key is not certified with sufficiently trusted signatures!
# gpg:          It is not certain that the signature belongs to the owner.
# Primary key fingerprint: B918 4DC2 0CC4 57DA CF7D  D1A9 3B5F CCCD F3AB D706

* tag 'migration-next-pull-request' of https://gitlab.com/peterx/qemu: (27 commits)
  migration/multifd: Document two places for mapped-ram
  tests/qtest/migration: Add a multifd + mapped-ram migration test
  migration/multifd: Add mapped-ram support to fd: URI
  migration/multifd: Support incoming mapped-ram stream format
  migration/multifd: Support outgoing mapped-ram stream format
  migration/multifd: Prepare multifd sync for mapped-ram migration
  migration/multifd: Add incoming QIOChannelFile support
  migration/multifd: Add outgoing QIOChannelFile support
  migration/multifd: Add a wrapper for channels_created
  migration/multifd: Allow receiving pages without packets
  migration/multifd: Allow multifd without packets
  migration/multifd: Decouple recv method from pages
  migration/multifd: Rename MultiFDSend|RecvParams::data to compress_data
  tests/qtest/migration: Add tests for mapped-ram file-based migration
  migration/ram: Add incoming 'mapped-ram' migration
  migration/ram: Add outgoing 'mapped-ram' migration
  migration: Add mapped-ram URI compatibility check
  migration/ram: Introduce 'mapped-ram' migration capability
  migration/qemu-file: add utility methods for working with seekable channels
  io: fsync before closing a file channel
  ...

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
# Conflicts:
# migration/ram.c

Merge tag 'pull-tcg-20240301' of https://gitlab.com/rth7680/qemu into staging

linux-user: Rewrite elf coredump
tcg/aarch64: Apple does not align __int128_t in even registers
accel/tcg: Fixes for page tables in mmio memory
linux-user: Remove qemu_host_page_{size,mask}, HOST_PAGE_ALIGN
migration: Remove qemu_host_page_size
hw/tpm: Remove qemu_host_page_size
softmmu: Remove qemu_host_page_{size,mask}, HOST_PAGE_ALIGN
linux-user: Split and reorganize target_mmap.
*-user: Deprecate and disable -p pagesize
linux-user: Allow TARGET_PAGE_BITS_VARY
target/alpha: Enable TARGET_PAGE_BITS_VARY for user-only
target/arm: Enable TARGET_PAGE_BITS_VARY for AArch64 user-only
target/ppc: Enable TARGET_PAGE_BITS_VARY for user-only
linux-user: Remove pgb_dynamic alignment assertion
tcg/optimize: fix uninitialized variable
linux-user: Rewrite shmat

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmXiXxQdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV/H3QgApu4OgadviJuOBenT
# yaGiq+iG4wTL5vVZFK8CgMtq59dJbgJSCooh7U8dn5hIhVuvOU7odUm6embt+4WZ
# 0fDZIjrRvdDMM3LdLFhfdZszMNg6w2ceN9dn5iLkW3wxjRBpTzZNbxhh2Sg308+Q
# oNd+MlYLijDvQP97+tlQ/PBtndLfV5FkpU74ZinWRgcpcT6oH9sP6TRlAVttefy7
# 3GsIXhDKGoDa/0Jpy86qE//3FUaVRqqcNlAIPXMf47ABQ2y2lZlwsfyty7s55sVW
# KgdXdH1GiCgxIonVg4bYvovnwKVH5xHlpsJY48jQtBXR/4exPBFBpeTc422E0Sed
# swpayg==
# =W3pb
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 01 Mar 2024 23:04:52 GMT
# 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-20240301' of https://gitlab.com/rth7680/qemu: (60 commits)
  tests/tcg: Check that shmat() does not break /proc/self/maps
  linux-user: Rewrite target_shmat
  linux-user: Add strace for shmat
  linux-user/loongarch64: Remove TARGET_FORCE_SHMLBA
  linux-user/x86_64: Handle the vsyscall page in open_self_maps_{2,4}
  tcg/optimize: fix uninitialized variable
  linux-user: Remove pgb_dynamic alignment assertion
  target/alpha: Enable TARGET_PAGE_BITS_VARY for user-only
  target/ppc: Enable TARGET_PAGE_BITS_VARY for user-only
  linux-user: Bound mmap_min_addr by host page size
  target/arm: Enable TARGET_PAGE_BITS_VARY for AArch64 user-only
  linux-user: Allow TARGET_PAGE_BITS_VARY
  accel/tcg: Disconnect TargetPageDataNode from page size
  cpu: Remove page_size_init
  *-user: Deprecate and disable -p pagesize
  tests/tcg: Extend file in linux-madvise.c
  tests/tcg: Remove run-test-mmap-*
  linux-user: Split out mmap_h_gt_g
  linux-user: Split out mmap_h_lt_g
  linux-user: Split out mmap_h_eq_g
  ...

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

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

HPPA64 updates

# -----BEGIN PGP SIGNATURE-----
#
# iHUEABYKAB0WIQS86RI+GtKfB8BJu973ErUQojoPXwUCZeQOtQAKCRD3ErUQojoP
# Xzh6AP9Gs8sJzRbR+HLlbrs3ACR7kQEVGOYuuBEJpKqkTyDbLwD/YdOjzgtTRKLd
# imxaWOJaLhpvBrmFLxn3Q58K3n9u1wY=
# =bnMy
# -----END PGP SIGNATURE-----
# gpg: Signature made Sun 03 Mar 2024 05:46:29 GMT
# 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-latest-pull-request' of https://github.com/hdeller/qemu-hppa:
  roms/hppa: Add build rules for hppa-firmware
  pc-bios/README: Add information about hppa-firmware
  pc-bios/meson: Add hppa-firmware64.img blob
  target/hppa: Restore unwind_breg before calculating ior
  target: hppa: Fix unaligned double word accesses for hppa64

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

gitlab: update FreeBSD Cirrus CI image to 13.3

The 13.2 images have been deleted from gcloud

Cc: qemu-stable@nongnu.org
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240304144456.3825935-3-berrange@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

meson: detect broken clang 17 with -fzero-call-used-regs

Clang 17 will segv if given  -fzero-call-used-regs and optimization
is enabled. Since upstream hasn't triaged the bug, distros are
increasingly shipping with broken Clang.

https://github.com/llvm/llvm-project/issues/75168
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=277474
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-id: 20240304144456.3825935-2-berrange@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

docs/conf.py: Remove usage of distutils

The macOS jobs in our CI recently started failing, complaining that
the distutils module is not available anymore. And indeed, according to
https://peps.python.org/pep-0632/ it's been deprecated since a while
and now likely got removed in recent Python versions.

Fortunately, we only use it for a version check via LooseVersion here
which we don't really need anymore - according to Repology.org, these
are the versions of sphinx-rtd-theme that are currently used by the
various distros:

 centos_stream_8: 0.3.1
 centos_stream_9: 0.5.1
 fedora_38: 1.1.1
 fedora_39: 1.2.2
 freebsd: 1.0.0
 haikuports_master: 1.2.1
 openbsd: 1.2.2
 opensuse_leap_15_5: 0.5.1
 pkgsrc_current: 2.0.0
 debian_11: 0.5.1
 debian_12: 1.2.0
 ubuntu_20_04: 0.4.3
 ubuntu_22_04: 1.0.0
 ubuntu_24_04: 2.0.0

So except for CentOS 8, all distros are using a newer version of
sphinx-rtd-theme, and for CentOS 8 we don't support compiling with
the Sphinx of the distro anymore anyway, since it's based on the
Python 3.6 interpreter there. For compiling on CentOS 8, you have
to use the alternative Python 3.8 interpreter which comes without
Sphinx, so that needs the Sphinx installed via pip in the venv
instead, and that is using a newer version, too, according to our
pythondeps.toml file.

Thus we can simply drop the version check now to get rid of the
distutils dependency here.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Message-id: 20240304130403.129543-1-thuth@redhat.com
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>

migration: simplify exec migration functions

Simplify the exec migration code by using list utility functions.

As a side effect, this also fixes a minor memory leak.  On function return,
"g_auto(GStrv) argv" frees argv and each element, which is wrong, because
the function does not own the individual elements.  To compensate, the code
uses g_steal_pointer which NULLs argv and prevents the destructor from
running, but argv is leaked.

Fixes: cbab4face57b ("migration: convert exec backend ...")
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Message-ID: <20240227153321.467343-4-armbru@redhat.com>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: New strv_from_str_list()

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20240227153321.467343-3-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

qapi: New QAPI_LIST_LENGTH()

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-ID: <20240227153321.467343-2-armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Markus Armbruster <armbru@redhat.com>

docs/devel/writing-monitor-commands: Minor improvements

Avoid "JSON" when talking about the QAPI schema syntax.  Capitalize
QEMU.  Don't claim all HMP commands live in monitor/hmp-cmds.c (this
was never true).  Fix punctuation and drop inappropriate "the" here
and there.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227115617.237875-3-armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>

docs/devel/writing-monitor-commands: Repair a decade of rot

The tutorial doesn't match reality since at least 2013.  Repairing it
involves fixing the following issues:

* Update for commit 6d327171551 (aio / timers: Remove alarm timers):
  replace the broken examples.  Instead of having one for returning a
  struct and another for returning a list of structs, do just one for
  the latter.  This resolves the FIXME added in commit
  e218052f928 (aio / timers: De-document -clock) back in 2014.

* Update for commit 895a2a80e0e (qapi: Use 'struct' instead of 'type'
  in schema).

* Update for commit 3313b6124b5 (qapi: add qapi2texi script): add
  required documentation to the schema snippets, and drop section
  "Command Documentation".

* Update for commit a3c45b3e629 (qapi: New special feature flag
  "unstable"): supply the required feature, deemphasize the x- prefix.

* Update for commit dd98234c059 (qapi: introduce x-query-roms QMP
  command): rephrase from "add new command" to "examine existing
  command".

* Update for commit 9492718b7c0 (qapi misc: Elide redundant has_FOO in
  generated C): hello-world's message argument no longer comes with a
  has_message, add a second argument that does.

* Update for moved and renamed files.

While there, update QMP version output to current output.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227115617.237875-2-armbru@redhat.com>
Reviewed-by: Eric Blake <eblake@redhat.com>
[Whitespace tidied up, typo fixed]

qapi: Reject "Returns" section when command doesn't return anything

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-14-armbru@redhat.com>

qga/qapi-schema: Fix guest-set-memory-blocks documentation

Documentation claims the command can "return NULL".  "NULL" doesn't
exist in JSON.  "null" does, but the command returns lists, and null
isn't.  Correct documentation to "return an empty list".

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-13-armbru@redhat.com>

qga/qapi-schema: Tweak documentation of fsfreeze commands

"Returns:" sections of guest-fsfreeze-freeze and
guest-fsfreeze-freeze-list describe both command behavior and success
response.  Move behavior out, so "Returns:" is only about success
response.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-12-armbru@redhat.com>

qga/qapi-schema: Clean up "Returns" sections

Drop "on success" where it is redundant with "Returns:".

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-11-armbru@redhat.com>

qga/qapi-schema: Delete useless "Returns" sections

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-10-armbru@redhat.com>

qga/qapi-schema: Move error documentation to new "Errors" sections

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-9-armbru@redhat.com>

qapi/yank: Tweak @yank's error description for consistency

Phrase it like "If <condition>, <error>", like we do elsewhere.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-8-armbru@redhat.com>

qapi: Clean up "Returns" sections

Drop list markup, since there's just one item left.

Drop "on success" where it is redundant with "Returns:".

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-7-armbru@redhat.com>

qapi: Delete useless "Returns" sections

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-6-armbru@redhat.com>

qapi: Move error documentation to new "Errors" sections

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-5-armbru@redhat.com>

qapi: New documentation section tag "Errors"

We use section "Returns" for documenting both success and error
response of commands.

I intend to generate better command success response documentation.
Easier when "Returns" documents just he success response.

Create new section tag "Errors".  The next two commits will move error
response documentation from "Returns" sections to "Errors" sections.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-4-armbru@redhat.com>

qapi: Slightly clearer error message for invalid "Returns" section

Change "'Returns:' is only valid for commands" to "'Returns' section
is only valid for commands".

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-3-armbru@redhat.com>

qapi: Memorize since & returns sections

This is chiefly to make code that looks up these sections easier to
read.

Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20240227113921.236097-2-armbru@redhat.com>

migration/multifd: Document two places for mapped-ram

Add two documentations for mapped-ram migration on two spots that may not
be extremely clear.

Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240301091524.39900-1-peterx@redhat.com
Cc: Prasad Pandit <ppandit@redhat.com>
[peterx: fix two English errors per Prasad]
Signed-off-by: Peter Xu <peterx@redhat.com>

roms/hppa: Add build rules for hppa-firmware

Signed-off-by: Helge Deller <deller@gmx.de>
Suggested-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

pc-bios/README: Add information about hppa-firmware

Signed-off-by: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

pc-bios/meson: Add hppa-firmware64.img blob

Add the missing 64-bit hppa firmware blob so that it gets installed.

Signed-off-by: Helge Deller <deller@gmx.de>
Fixes: 7c0dfcf9395e ("target/hppa: Update SeaBIOS-hppa to version 16")
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Restore unwind_breg before calculating ior

When calculating the IOR for the exception handlers, the current
unwind_breg value is needed on 64-bit hppa machines.
Restore that value by calling cpu_restore_state() earlier, which in turn
calls hppa_restore_state_to_opc() which restores the unwind_breg for the
current instruction.

Signed-off-by: Helge Deller <deller@gmx.de>
Fixes: 3824e0d643f3 ("target/hppa: Export function hppa_set_ior_and_isr()")
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>

target: hppa: Fix unaligned double word accesses for hppa64

Unaligned 64-bit accesses were found in Linux to clobber carry bits,
resulting in bad results if an arithmetic operation involving a
carry bit was executed after an unaligned 64-bit operation.

hppa 2.0 defines additional carry bits in PSW register bits 32..39.
When restoring PSW after executing an unaligned instruction trap, those
bits were not cleared and ended up to be active all the time. Since there
are no bits other than the upper carry bits needed in the upper 32 bit of
env->psw and since those are stored in env->psw_cb, just clear the entire
upper 32 bit when storing psw to solve the problem unconditionally.

Fixes: 931adff31478 ("target/hppa: Update cpu_hppa_get/put_psw for hppa64")
Cc: Richard Henderson <richard.henderson@linaro.org>
Cc: Charlie Jenkins <charlie@rivosinc.com>
Cc: Helge Deller <deller@gmx.de>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Guenter Roeck <linux@roeck-us.net>
Signed-off-by: Helge Deller <deller@gmx.de>

tests/tcg: Check that shmat() does not break /proc/self/maps

Add a regression test for a recently fixed issue, where shmat()
desynced the guest and the host view of the address space and caused
open("/proc/self/maps") to SEGV.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
Message-Id: <jwyuvao4apydvykmsnvacwshdgy3ixv7qvkh4dbxm3jkwgnttw@k4wpaayou7oq>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

linux-user: Rewrite target_shmat

Handle combined host and guest alignment requirements.
Handle host and guest page size differences.
Handle SHM_EXEC.

Resolves: https://gitlab.com/qemu-project/qemu/-/issues/115
Tested-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

linux-user: Add strace for shmat

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

Merge tag 'pull-request-2024-03-01' of https://gitlab.com/thuth/qemu into staging

* Fix some bugs in the vring setup of libqos
* Fix GIC settings when using --without-default-devices
* Fix USB PCAP streams on Windows
* Remove temporary files from test-util-sockets
* Fix TLS io channels sending too much data to the backend

# -----BEGIN PGP SIGNATURE-----
#
# iQJFBAABCAAvFiEEJ7iIR+7gJQEY8+q5LtnXdP5wLbUFAmXhjBcRHHRodXRoQHJl
# ZGhhdC5jb20ACgkQLtnXdP5wLbWzZw/+OTSsKg6JTX0z3fjL6If+Ns/EnFiniHHI
# y1jR7fPub3ybFokgHIWrRVG+9ol+1tJuPlO5Bsx1SKFzNC8++l1iyK7E25xnmp2P
# Ry/Z0ZSWp3JWCtLm6km5pKl3wWI9p0XwfUOJbCWYM5dOsd2a4C4DSvjcwNXR7pTB
# 4AxXFj5G5UwXuffgPnBzeerm5baNweSqa/uczUY6Od+iJwmHpaSk0SJ4NIlIYvnS
# z3BH81trQAoKYZyywp0aF+jH9w3YNwD1XDp2Bcf3qSWDdokN55V13yyiSllTfUwX
# Lq3fotoLqFJL0A4m8a3TdCKMP8ReoJATSriPHJmGMgqWGTnvtRBIsx1pmMpYRB6U
# EltaKW0WXeQAad0ZYGl4odx7kQUQjdc8HTQRTevQD9g3XpZZN0o9MfTPdM2eR5h6
# xaS8UEc61YJAcmD27Ir92JGGGMLBTYz4OGLx3Rj7l7S759GQ67qih9TGcE9nqSj+
# +wAHfiOjUZnSWnp7Hq4M/TIjNh0BuyBW4oRBASaikjjQ82Hih5bPTdRu1J8lYuXC
# PhN8dLwdk0NMrnnHuFDKq/sXjDeXF3Hg0AmtVvOGiB2z2OeVSPtMxl/YgniGVTbj
# CHXDqDrBAa7i6WxAo+TtUgE20iVz5uYzGqHmxTNpq6FzIRb1uvHksbjFlin79Pkw
# eF0Dj80zxAY=
# =9JAI
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 01 Mar 2024 08:04:39 GMT
# 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-2024-03-01' of https://gitlab.com/thuth/qemu:
  chardev/char-socket: Fix TLS io channels sending too much data to the backend
  tests/unit/test-util-sockets: Remove temporary file after test
  hw/usb/bus.c: PCAP adding 0xA in Windows version
  hw/intc/Kconfig: Fix GIC settings when using "--without-default-devices"
  libqos/virtio.c: fix 'avail_event' offset in qvring_init()
  libqos/virtio.c: init all elems in qvring_indirect_desc_setup()

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

Merge tag 'pull-loongarch-20240229' of https://gitlab.com/gaosong/qemu into staging

pull-loongarch-20240229

V2: fix build error on mipsel

# -----BEGIN PGP SIGNATURE-----
#
# iLMEAAEKAB0WIQS4/x2g0v3LLaCcbCxAov/yOSY+3wUCZeBrwAAKCRBAov/yOSY+
# 33YXA/4+A5Bpe/3+mSAWZSUlluGTqUi0ILBYRMyX1RXovMx4uCRGr7PXzAf03yKS
# MZzlVzTuOK69WmTm/iTdYWOxkXisC3gzxL/wm8hP4lzh4c0dHrHRsKHqq6gR3+t2
# ojdZn7TefeflnNqIhxXxgxb1OETofhBNnBJ74pvqxO7XV5SWnA==
# =J2Kb
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 29 Feb 2024 11:34:24 GMT
# gpg:                using RSA key B8FF1DA0D2FDCB2DA09C6C2C40A2FFF239263EDF
# gpg: Good signature from "Song Gao <m17746591750@163.com>" [unknown]
# gpg: WARNING: This key is not certified with a trusted signature!
# gpg:          There is no indication that the signature belongs to the owner.
# Primary key fingerprint: B8FF 1DA0 D2FD CB2D A09C  6C2C 40A2 FFF2 3926 3EDF

* tag 'pull-loongarch-20240229' of https://gitlab.com/gaosong/qemu:
  loongarch: Change the UEFI loading mode to loongarch

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

tests/qtest/migration: Add a multifd + mapped-ram migration test

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240229153017.2221-24-farosas@suse.de
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Add mapped-ram support to fd: URI

If we receive a file descriptor that points to a regular file, there's
nothing stopping us from doing multifd migration with mapped-ram to
that file.

Enable the fd: URI to work with multifd + mapped-ram.

Note that the fds passed into multifd are duplicated because we want
to avoid cross-thread effects when doing cleanup (i.e. close(fd)). The
original fd doesn't need to be duplicated because monitor_get_fd()
transfers ownership to the caller.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/20240229153017.2221-23-farosas@suse.de
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Support incoming mapped-ram stream format

For the incoming mapped-ram migration we need to read the ramblock
headers, get the pages bitmap and send the host address of each
non-zero page to the multifd channel thread for writing.

Usage on HMP is:

(qemu) migrate_set_capability multifd on
(qemu) migrate_set_capability mapped-ram on
(qemu) migrate_incoming file:migfile

(the ram.h include needs to move because we've been previously relying
on it being included from migration.c. Now file.h will start including
multifd.h before migration.o is processed)

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240229153017.2221-22-farosas@suse.de
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Support outgoing mapped-ram stream format

The new mapped-ram stream format uses a file transport and puts ram
pages in the migration file at their respective offsets and can be
done in parallel by using the pwritev system call which takes iovecs
and an offset.

Add support to enabling the new format along with multifd to make use
of the threading and page handling already in place.

This requires multifd to stop sending headers and leaving the stream
format to the mapped-ram code. When it comes time to write the data, we
need to call a version of qio_channel_write that can take an offset.

Usage on HMP is:

(qemu) stop
(qemu) migrate_set_capability multifd on
(qemu) migrate_set_capability mapped-ram on
(qemu) migrate_set_parameter max-bandwidth 0
(qemu) migrate_set_parameter multifd-channels 8
(qemu) migrate file:migfile

Reviewed-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20240229153017.2221-21-farosas@suse.de
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Prepare multifd sync for mapped-ram migration

The mapped-ram migration can be performed live or non-live, but it is
always asynchronous, i.e. the source machine and the destination
machine are not migrating at the same time. We only need some pieces
of the multifd sync operations.

multifd_send_sync_main()
------------------------
  Issued by the ram migration code on the migration thread, causes the
  multifd send channels to synchronize with the migration thread and
  makes the sending side emit a packet with the MULTIFD_FLUSH flag.

  With mapped-ram we want to maintain the sync on the sending side
  because that provides ordering between the rounds of dirty pages when
  migrating live.

MULTIFD_FLUSH
-------------
  On the receiving side, the presence of the MULTIFD_FLUSH flag on a
  packet causes the receiving channels to start synchronizing with the
  main thread.

  We're not using packets with mapped-ram, so there's no MULTIFD_FLUSH
  flag and therefore no channel sync on the receiving side.

multifd_recv_sync_main()
------------------------
  Issued by the migration thread when the ram migration flag
  RAM_SAVE_FLAG_MULTIFD_FLUSH is received, causes the migration thread
  on the receiving side to start synchronizing with the recv
  channels. Due to compatibility, this is also issued when
  RAM_SAVE_FLAG_EOS is received.

  For mapped-ram we only need to synchronize the channels at the end of
  migration to avoid doing cleanup before the channels have finished
  their IO.

Make sure the multifd syncs are only issued at the appropriate times.

Note that due to pre-existing backward compatibility issues, we have
the multifd_flush_after_each_section property that can cause a sync to
happen at EOS. Since the EOS flag is needed on the stream, allow
mapped-ram to just ignore it.

Also emit an error if any other unexpected flags are found on the
stream.

Signed-off-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/20240229153017.2221-20-farosas@suse.de
Signed-off-by: Peter Xu <peterx@redhat.com>