dump: Add close fd on error return to avoid resource leak

Reported-by: Coverity CID 1523842 (RESOURCE_LEAK)
Fixes: e6549197f7 ("dump: Add command interface for kdump-raw formats")
Signed-off-by: Zongmin Zhou <zhouzongmin@kylinos.cn>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231107024417.585475-1-min_halo@163.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

ui/sdl2: use correct key names in win title on mac

Previously, when using the SDL2 UI on MacOS, the title bar uses incorrect
key names (such as Ctrl and Alt instead of the standard MacOS key symbols
like ⌃ and ⌥). This commit changes sdl_update_caption in ui/sdl2.c to
use the correct symbols when compiling for MacOS (CONFIG_DARWIN is
defined).

Unfortunately, standard Mac keyboards do not include a "Right-Ctrl" key,
so in the case that the SDL grab mode is set to HOT_KEY_MOD_RCTRL, the
default text is still used.

Signed-off-by: Adrian Wowk <dev@adrianwowk.com>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Tested-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231030024119.28342-1-dev@adrianwowk.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

MAINTAINERS: Add more guest-agent related files to the corresponding section

contrib/systemd/qemu-guest-agent.service, tests/data/test-qga-config
and tests/data/test-qga-os-release belong to the guest agent, so make
sure that these files are covered here, too.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Konstantin Kostiuk <kkostiuk@redhat.com>
Message-ID: <20231107101811.14189-1-thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

MAINTAINERS: Add include/hw/xtensa/mx_pic.h to the XTFPGA machine section

These machines are the only user of the mx_pic code, so the
header (which is currently "unmaintained" according to the
MAINTAINERS file) should be added to this section.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Acked-by: Max Filippov <jcmvbkbc@gmail.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231107102104.14342-1-thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

MAINTAINERS: update libvirt devel mailing list address

Effective immediately, the libvirt project has moved its list off
libvir-list@redhat.com, to devel@lists.libvirt.org

Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
Message-ID: <20231027095643.2842382-1-berrange@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

MAINTAINERS: Add the CAN documentation file to the CAN section

Add can.rst to the corresponding section in MAINTAINERS, so that
the maintainers get CC:-ed on corresponding patches.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Vikram Garhwal <vikram.garhwal@amd.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
Message-ID: <20231027060931.242491-1-thuth@redhat.com>
[PMD: Fixed typo in subject]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

MAINTAINERS: Add include/hw/timer/tmu012.h to the SH4 R2D section

tmu012.h is the header that belongs to hw/timer/sh_timer.c, so we
should list it in the same section as sh_timer.c.

Signed-off-by: Thomas Huth <thuth@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Yoshinori Sato <ysato@users.sourceforge.jp>
Message-ID: <20231026080011.156325-1-thuth@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/sd: Declare QOM types using DEFINE_TYPES() macro

When multiple QOM types are registered in the same file,
it is simpler to use the the DEFINE_TYPES() macro. In
particular because type array declared with such macro
are easier to review.

Mechanical transformation using the following comby script:

  [pattern-x1]
  match='''
  static const TypeInfo :[i1~.*_info] = {
      :[body]
  };
  static void :[rt1~.*_register_type.](void)
  {
      type_register_static(&:[i2~.*_info]);
  }
  type_init(:[rt2~.*_register_type.])
  '''
  rewrite='''
  static const TypeInfo :[i1][] = {
      {
      :[body]
      },
  };

  DEFINE_TYPES(:[i1])
  '''
  rule='where :[i1] == :[i2], :[rt1] == :[rt2]'

  [pattern-x2]
  match='''
  static const TypeInfo :[i1a~.*_info] = {
      :[body1]
  };
  ...
  static const TypeInfo :[i2a~.*_info] = {
      :[body2]
  };
  static void :[rt1~.*_register_type.](void)
  {
      type_register_static(&:[i1b~.*_info]);
      type_register_static(&:[i2b~.*_info]);
  }
  type_init(:[rt2~.*_register_type.])
  '''
  rewrite='''
  static const TypeInfo :[i1a][] = {
      {
      :[body1]
      },
      {
      :[body2]
      },
  };

  DEFINE_TYPES(:[i1a])
  '''
  rule='''
  where
  :[i1a] == :[i1b],
  :[i2a] == :[i2b],
  :[rt1] == :[rt2]
  '''

and re-indented manually.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20231031080603.86889-2-philmd@linaro.org>

hw/i2c: pmbus: reset page register for out of range reads

The linux pmbus driver scans all possible pages and does not reset the
current page after the scan, making all future page reads fail as out of range
on devices with a single page.

This change resets out of range pages immediately on write.

Also added a qtest for simultaneous writes to all pages.

Reviewed-by: Hao Wu <wuhaotsh@google.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
Message-ID: <20231023-staging-pmbus-v3-v4-8-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i2c: pmbus: immediately clear faults on request

The probing process of the generic pmbus driver generates
faults to determine if functions are available. These faults
were not always cleared resulting in probe failures.

Reviewed-by: Patrick Venture <venture@google.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
Message-ID: <20231023-staging-pmbus-v3-v4-7-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

tests/qtest: add tests for ADM1266

The ADM1266 can have string fields written by the driver, so
it's worth specifically testing.

Reviewed-by: Hao Wu <wuhaotsh@google.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
[PMD: Cover file in MAINTAINERS]
Message-ID: <20231023-staging-pmbus-v3-v4-6-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/sensor: add ADM1266 device model

The ADM1266 is a cascadable super sequencer with margin control and
fault recording.
This commit adds basic support for its PMBus commands and models
the identification registers that can be modified in a firmware
update.

Reviewed-by: Hao Wu <wuhaotsh@google.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
[PMD: Cover file in MAINTAINERS]
Message-ID: <20231023-staging-pmbus-v3-v4-5-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i2c: pmbus: add VCAP register

VCAP is a register for devices with energy storage capacitors.

Reviewed-by: Benjamin Streb <bstreb@google.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
Message-ID: <20231023-staging-pmbus-v3-v4-4-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i2c: pmbus: add fan support

PMBus devices may integrate fans whose operation is configurable
over PMBus. This commit allows the driver to read and write the
fan control registers but does not model the operation of fans.

Reviewed-by: Stephen Longfield <slongfield@google.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
Message-ID: <20231023-staging-pmbus-v3-v4-3-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i2c: pmbus: add vout mode bitfields

The VOUT_MODE command is described in the PMBus Specification,
Part II, Ver 1.3 Section 8.3

VOUT_MODE has a three bit mode and 4 bit parameter, the three bit
mode determines whether voltages are formatted as uint16, uint16,
VID, and Direct modes. VID and Direct modes use the remaining 5 bits
to scale the voltage readings.

Reviewed-by: Hao Wu <wuhaotsh@google.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
Message-ID: <20231023-staging-pmbus-v3-v4-2-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i2c: pmbus add support for block receive

PMBus devices can send and receive variable length data using the
block read and write format, with the first byte in the payload
denoting the length.

This is mostly used for strings and on-device logs. Devices can
respond to a block read with an empty string.

Reviewed-by: Hao Wu <wuhaotsh@google.com>
Acked-by: Corey Minyard <cminyard@mvista.com>
Signed-off-by: Titus Rwantare <titusr@google.com>
Message-ID: <20231023-staging-pmbus-v3-v4-1-07a8cb7cd20a@google.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

tests/qtest: ahci-test: add test exposing reset issue with pending callback

Before commit "hw/ide: reset: cancel async DMA operation before
resetting state", this test would fail, because a reset with a
pending write operation would lead to an unsolicited write to the
first sector of the disk.

The test writes a pattern to the beginning of the disk and verifies
that it is still intact after a reset with a pending operation. It
also checks that the pending operation actually completes correctly.

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Message-ID: <20230906130922.142845-2-f.ebner@proxmox.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/ide: reset: cancel async DMA operation before resetting state

If there is a pending DMA operation during ide_bus_reset(), the fact
that the IDEState is already reset before the operation is canceled
can be problematic. In particular, ide_dma_cb() might be called and
then use the reset IDEState which contains the signature after the
reset. When used to construct the IO operation this leads to
ide_get_sector() returning 0 and nsector being 1. This is particularly
bad, because a write command will thus destroy the first sector which
often contains a partition table or similar.

Traces showing the unsolicited write happening with IDEState
0x5595af6949d0 being used after reset:

> ahci_port_write ahci(0x5595af6923f0)[0]: port write [reg:PxSCTL] @ 0x2c: 0x00000300
> ahci_reset_port ahci(0x5595af6923f0)[0]: reset port
> ide_reset IDEstate 0x5595af6949d0
> ide_reset IDEstate 0x5595af694da8
> ide_bus_reset_aio aio_cancel
> dma_aio_cancel dbs=0x7f64600089a0
> dma_blk_cb dbs=0x7f64600089a0 ret=0
> dma_complete dbs=0x7f64600089a0 ret=0 cb=0x5595acd40b30
> ahci_populate_sglist ahci(0x5595af6923f0)[0]
> ahci_dma_prepare_buf ahci(0x5595af6923f0)[0]: prepare buf limit=512 prepared=512
> ide_dma_cb IDEState 0x5595af6949d0; sector_num=0 n=1 cmd=DMA WRITE
> dma_blk_io dbs=0x7f6420802010 bs=0x5595ae2c6c30 offset=0 to_dev=1
> dma_blk_cb dbs=0x7f6420802010 ret=0

> (gdb) p *qiov
> $11 = {iov = 0x7f647c76d840, niov = 1, {{nalloc = 1, local_iov = {iov_base = 0x0,
>       iov_len = 512}}, {__pad = "\001\000\000\000\000\000\000\000\000\000\000",
>       size = 512}}}
> (gdb) bt
> #0  blk_aio_pwritev (blk=0x5595ae2c6c30, offset=0, qiov=0x7f6420802070, flags=0,
>     cb=0x5595ace6f0b0 <dma_blk_cb>, opaque=0x7f6420802010)
>     at ../block/block-backend.c:1682
> #1  0x00005595ace6f185 in dma_blk_cb (opaque=0x7f6420802010, ret=<optimized out>)
>     at ../softmmu/dma-helpers.c:179
> #2  0x00005595ace6f778 in dma_blk_io (ctx=0x5595ae0609f0,
>     sg=sg@entry=0x5595af694d00, offset=offset@entry=0, align=align@entry=512,
>     io_func=io_func@entry=0x5595ace6ee30 <dma_blk_write_io_func>,
>     io_func_opaque=io_func_opaque@entry=0x5595ae2c6c30,
>     cb=0x5595acd40b30 <ide_dma_cb>, opaque=0x5595af6949d0,
>     dir=DMA_DIRECTION_TO_DEVICE) at ../softmmu/dma-helpers.c:244
> #3  0x00005595ace6f90a in dma_blk_write (blk=0x5595ae2c6c30,
>     sg=sg@entry=0x5595af694d00, offset=offset@entry=0, align=align@entry=512,
>     cb=cb@entry=0x5595acd40b30 <ide_dma_cb>, opaque=opaque@entry=0x5595af6949d0)
>     at ../softmmu/dma-helpers.c:280
> #4  0x00005595acd40e18 in ide_dma_cb (opaque=0x5595af6949d0, ret=<optimized out>)
>     at ../hw/ide/core.c:953
> #5  0x00005595ace6f319 in dma_complete (ret=0, dbs=0x7f64600089a0)
>     at ../softmmu/dma-helpers.c:107
> #6  dma_blk_cb (opaque=0x7f64600089a0, ret=0) at ../softmmu/dma-helpers.c:127
> #7  0x00005595ad12227d in blk_aio_complete (acb=0x7f6460005b10)
>     at ../block/block-backend.c:1527
> #8  blk_aio_complete (acb=0x7f6460005b10) at ../block/block-backend.c:1524
> #9  blk_aio_write_entry (opaque=0x7f6460005b10) at ../block/block-backend.c:1594
> #10 0x00005595ad258cfb in coroutine_trampoline (i0=<optimized out>,
>     i1=<optimized out>) at ../util/coroutine-ucontext.c:177

Signed-off-by: Fiona Ebner <f.ebner@proxmox.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: simon.rowe@nutanix.com
Message-ID: <20230906130922.142845-1-f.ebner@proxmox.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/cpu: Update the comments of nr_cores and nr_dies

In the nr_threads' comment, specify it represents the
number of threads in the "core" to avoid confusion.

Also add comment for nr_dies in CPUX86State.

Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Babu Moger <babu.moger@amd.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20231024090323.1859210-5-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

system/cpus: Fix CPUState.nr_cores' calculation

From CPUState.nr_cores' comment, it represents "number of cores within
this CPU package".

After 003f230e37d7 ("machine: Tweak the order of topology members in
struct CpuTopology"), the meaning of smp.cores changed to "the number of
cores in one die", but this commit missed to change CPUState.nr_cores'
calculation, so that CPUState.nr_cores became wrong and now it
misses to consider numbers of clusters and dies.

At present, only i386 is using CPUState.nr_cores.

But as for i386, which supports die level, the uses of CPUState.nr_cores
are very confusing:

Early uses are based on the meaning of "cores per package" (before die
is introduced into i386), and later uses are based on "cores per die"
(after die's introduction).

This difference is due to that commit a94e1428991f ("target/i386: Add
CPUID.1F generation support for multi-dies PCMachine") misunderstood
that CPUState.nr_cores means "cores per die" when calculated
CPUID.1FH.01H:EBX. After that, the changes in i386 all followed this
wrong understanding.

With the influence of 003f230e37d7 and a94e1428991f, for i386 currently
the result of CPUState.nr_cores is "cores per die", thus the original
uses of CPUState.cores based on the meaning of "cores per package" are
wrong when multiple dies exist:
1. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.01H:EBX[bits 23:16] is
   incorrect because it expects "cpus per package" but now the
   result is "cpus per die".
2. In cpu_x86_cpuid() of target/i386/cpu.c, for all leaves of CPUID.04H:
   EAX[bits 31:26] is incorrect because they expect "cpus per package"
   but now the result is "cpus per die". The error not only impacts the
   EAX calculation in cache_info_passthrough case, but also impacts other
   cases of setting cache topology for Intel CPU according to cpu
   topology (specifically, the incoming parameter "num_cores" expects
   "cores per package" in encode_cache_cpuid4()).
3. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.0BH.01H:EBX[bits
   15:00] is incorrect because the EBX of 0BH.01H (core level) expects
   "cpus per package", which may be different with 1FH.01H (The reason
   is 1FH can support more levels. For QEMU, 1FH also supports die,
   1FH.01H:EBX[bits 15:00] expects "cpus per die").
4. In cpu_x86_cpuid() of target/i386/cpu.c, when CPUID.80000001H is
   calculated, here "cpus per package" is expected to be checked, but in
   fact, now it checks "cpus per die". Though "cpus per die" also works
   for this code logic, this isn't consistent with AMD's APM.
5. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.80000008H:ECX expects
   "cpus per package" but it obtains "cpus per die".
6. In simulate_rdmsr() of target/i386/hvf/x86_emu.c, in
   kvm_rdmsr_core_thread_count() of target/i386/kvm/kvm.c, and in
   helper_rdmsr() of target/i386/tcg/sysemu/misc_helper.c,
   MSR_CORE_THREAD_COUNT expects "cpus per package" and "cores per
   package", but in these functions, it obtains "cpus per die" and
   "cores per die".

On the other hand, these uses are correct now (they are added in/after
a94e1428991f):
1. In cpu_x86_cpuid() of target/i386/cpu.c, topo_info.cores_per_die
   meets the actual meaning of CPUState.nr_cores ("cores per die").
2. In cpu_x86_cpuid() of target/i386/cpu.c, vcpus_per_socket (in CPUID.
   04H's calculation) considers number of dies, so it's correct.
3. In cpu_x86_cpuid() of target/i386/cpu.c, CPUID.1FH.01H:EBX[bits
   15:00] needs "cpus per die" and it gets the correct result, and
   CPUID.1FH.02H:EBX[bits 15:00] gets correct "cpus per package".

When CPUState.nr_cores is correctly changed to "cores per package" again
, the above errors will be fixed without extra work, but the "currently"
correct cases will go wrong and need special handling to pass correct
"cpus/cores per die" they want.

Fix CPUState.nr_cores' calculation to fit the original meaning "cores
per package", as well as changing calculation of topo_info.cores_per_die,
vcpus_per_socket and CPUID.1FH.

Fixes: a94e1428991f ("target/i386: Add CPUID.1F generation support for multi-dies PCMachine")
Fixes: 003f230e37d7 ("machine: Tweak the order of topology members in struct CpuTopology")
Signed-off-by: Zhuocheng Ding <zhuocheng.ding@intel.com>
Co-developed-by: Zhao Liu <zhao1.liu@intel.com>
Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Tested-by: Babu Moger <babu.moger@amd.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20231024090323.1859210-4-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

tests/unit: Rename test-x86-cpuid.c to test-x86-topo.c

The tests in this file actually test the APIC ID combinations.
Rename to test-x86-topo.c to make its name more in line with its
actual content.

Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Tested-by: Babu Moger <babu.moger@amd.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-ID: <20231024090323.1859210-3-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/i386: Fix comment style in topology.h

For function comments in this file, keep the comment style consistent
with other files in the directory.

Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Yanan Wang <wangyanan55@huawei.com>
Reviewed-by: Xiaoyao Li <xiaoyao.li@Intel.com>
Reviewed-by: Babu Moger <babu.moger@amd.com>
Tested-by: Babu Moger <babu.moger@amd.com>
Tested-by: Yongwei Ma <yongwei.ma@intel.com>
Acked-by: Michael S. Tsirkin <mst@redhat.com>
Message-ID: <20231024090323.1859210-2-zhao1.liu@linux.intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/isa/i82378: Propagate error if PC_SPEAKER device creation failed

In commit 40f8214fcd ("hw/audio/pcspk: Inline pcspk_init()")
we neglected to give a change to the caller to handle failed
device creation cleanly. Respect the caller API contract and
propagate the error if creating the PC_SPEAKER device ever
failed. This avoid yet another bad API use to be taken as
example and copy / pasted all over the code base.

Reported-by: Bernhard Beschow <shentey@gmail.com>
Suggested-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Bernhard Beschow <shentey@gmail.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231020171509.87839-5-philmd@linaro.org>

hw/loader: Clean up global variable shadowing in rom_add_file()

Fix:

  hw/core/loader.c:1073:27: error: declaration shadows a variable in the global scope [-Werror,-Wshadow]
                       bool option_rom, MemoryRegion *mr,
                            ^
  include/sysemu/sysemu.h:57:22: note: previous declaration is here
  extern QEMUOptionRom option_rom[MAX_OPTION_ROMS];
                       ^

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Ani Sinha <anisinha@redhat.com>
Message-Id: <20231010115048.11856-3-philmd@linaro.org>

hw/cpu: Clean up global variable shadowing

Fix:

  hw/core/machine.c:1302:22: error: declaration shadows a variable in the global scope [-Werror,-Wshadow]
      const CPUArchId *cpus = possible_cpus->cpus;
                       ^
  hw/core/numa.c:69:17: error: declaration shadows a variable in the global scope [-Werror,-Wshadow]
      uint16List *cpus = NULL;
                  ^
  hw/acpi/aml-build.c:2005:20: error: declaration shadows a variable in the global scope [-Werror,-Wshadow]
      CPUArchIdList *cpus = ms->possible_cpus;
                     ^
  hw/core/machine-smp.c:77:14: error: declaration shadows a variable in the global scope [-Werror,-Wshadow]
      unsigned cpus    = config->has_cpus ? config->cpus : 0;
               ^
  include/hw/core/cpu.h:589:17: note: previous declaration is here
  extern CPUTailQ cpus;
                  ^

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Ani Sinha <anisinha@redhat.com>
Message-Id: <20231010115048.11856-2-philmd@linaro.org>

exec/cpu: Have cpu_exec_realize() return a boolean

Following the example documented since commit e3fe3988d7 ("error:
Document Error API usage rules"), have cpu_exec_realizefn()
return a boolean indicating whether an error is set or not.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230918160257.30127-22-philmd@linaro.org>

hw/cpu: Call object_class_is_abstract() once in cpu_class_by_name()

Let CPUClass::class_by_name() handlers to return abstract classes,
and filter them once in the public cpu_class_by_name() method.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230908112235.75914-3-philmd@linaro.org>

target/alpha: Tidy up alpha_cpu_class_by_name()

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20230908112235.75914-2-philmd@linaro.org>

target: Move ArchCPUClass definition to 'cpu.h'

The OBJECT_DECLARE_CPU_TYPE() macro forward-declares each
ArchCPUClass type. These forward declarations are sufficient
for code in hw/ to use the QOM definitions. No need to expose
these structure definitions. Keep each local to their target/
by moving them to the corresponding "cpu.h" header.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-13-philmd@linaro.org>

target/nios2: Create IRQs *after* accelerator vCPU is realized

Architecture specific hardware doesn't have a particular dependency
on the accelerator vCPU (created with cpu_exec_realizefn), and can
be initialized *after* the vCPU is realized. Doing so allows further
generic API simplification (in few commits).

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230918160257.30127-12-philmd@linaro.org>

target/s390x/cpu: Restrict CPUS390XState declaration to 'cpu.h'

"target/s390x/cpu-qom.h" has to be target-agnostic. However, it
currently declares CPUS390XState, which is target-specific.
Move that declaration to "cpu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20231106114500.5269-5-philmd@linaro.org>

target/s390x/cpu: Restrict cpu_get_tb_cpu_state() definition to TCG

cpu_get_tb_cpu_state() is TCG specific. Another accelerator
calling it would be a bug, so restrict the definition to TCG,
along with "tcg_s390x.h" header inclusion.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20231106114500.5269-4-philmd@linaro.org>

hw/s390x/sclp: Have sclp_service_call[_protected]() take S390CPU*

"hw/s390x/sclp.h" is a header used by target-agnostic objects
(such hw/char/sclpconsole[-lm].c), thus can not use target-specific
types, such CPUS390XState.

Have sclp_service_call[_protected]() take a S390CPU pointer, which
is target-agnostic.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20231106114500.5269-3-philmd@linaro.org>

hw/s390x/css: Have css_do_sic() take S390CPU instead of CPUS390XState

"hw/s390x/css.h" is a header used by target-agnostic objects
(such hw/s390x/virtio-ccw-gpu.c), thus can not use target-specific
types, such CPUS390XState.

Have css_do_sic() take S390CPU a pointer, which is target-agnostic.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Message-Id: <20231106114500.5269-2-philmd@linaro.org>

target/ppc: Move powerpc_input_t definition to 'cpu.h'

The powerpc_input_t definition is only used by target/ppc/, no need
to expose it. Restrict it by moving it to "target/ppc/cpu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20231013125630.95116-8-philmd@linaro.org>

target/ppc: Move powerpc_mmu_t definition to 'cpu.h'

The powerpc_mmu_t definition is only used by target/ppc/, no need
to expose it. Restrict it by moving it to "target/ppc/cpu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20231013125630.95116-7-philmd@linaro.org>

target/ppc: Move powerpc_excp_t definition to 'cpu.h'

The powerpc_excp_t definition is only used by target/ppc/, no need
to expose it. Restrict it by moving it to "target/ppc/cpu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20231013125630.95116-6-philmd@linaro.org>

target/ppc: Move PowerPCCPUClass definition to 'cpu.h'

The OBJECT_DECLARE_CPU_TYPE() macro forward-declares the
PowerPCCPUClass type. This forward declaration is sufficient
for code in hw/ to use the QOM definitions. No need to expose
the structure definition. Keep it local to target/ppc/ by
moving it to target/ppc/cpu.h.

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013125630.95116-5-philmd@linaro.org>

target/ppc: Move ppc_cpu_class_by_name() declaration to 'cpu.h'

ppc_cpu_class_by_name() is only called in target/ppc/,
no need to expose outside (in particular to hw/).

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20231013125630.95116-4-philmd@linaro.org>

target/ppc: Define powerpc_pm_insn_t in 'internal.h'

PM instructions are only used by TCG helpers. No need to
expose to other hardware.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Cédric Le Goater <clg@kaod.org>
Message-Id: <20231013125630.95116-3-philmd@linaro.org>

target/ppc: Prohibit target specific KVM prototypes on user emulation

None of these target-specific prototypes should be used
by user emulation. Remove their declaration there, so we
get a compile failure if ever used (instead of having to
deal with linker and its possible optimizations, such
dead code removal).

Suggested-by: Kevin Wolf <kwolf@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20231003070427.69621-5-philmd@linaro.org>

target/ppc: Restrict KVM objects to system emulation

CONFIG_KVM is always FALSE on user emulation, so 'kvm.c'
won't be added to ppc_ss[] source set; direcly use the system
specific ppc_system_ss[] source set.

Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231003070427.69621-4-philmd@linaro.org>

hw/ppc/e500: Restrict ppce500_init_mpic_kvm() to KVM

Inline and guard the single call to kvm_openpic_connect_vcpu()
allows to remove kvm-stub.c.

Reviewed-by: Michael Tokarev <mjt@tls.msk.ru>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231003070427.69621-3-philmd@linaro.org>

sysemu/kvm: Restrict kvmppc_get_radix_page_info() to ppc targets

kvm_get_radix_page_info() is only defined for ppc targets (in
target/ppc/kvm.c). The declaration is not useful in other targets,
reduce its scope.
Rename using the 'kvmppc_' prefix following other declarations
from target/ppc/kvm_ppc.h.

Suggested-by: Michael Tokarev <mjt@tls.msk.ru>
Reviewed-by: Daniel Henrique Barboza <danielhb413@gmail.com>
Message-Id: <20231003070427.69621-2-philmd@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/mips: Fix TX79 LQ/SQ opcodes

The base register address offset is *signed*.

Cc: qemu-stable@nongnu.org
Fixes: aaaa82a9f9 ("target/mips/tx79: Introduce LQ opcode (Load Quadword)")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914090447.12557-1-philmd@linaro.org>

target/mips: Fix MSA BZ/BNZ opcodes displacement

The PC offset is *signed*.

Cc: qemu-stable@nongnu.org
Reported-by: Sergey Evlashev <vectorchiefrocks@gmail.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/1624
Fixes: c7a9ef7517 ("target/mips: Introduce decode tree bindings for MSA ASE")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230914085807.12241-1-philmd@linaro.org>

target/i386/monitor: synchronize cpu state for lapic info

While the default "info lapic" always synchronizes cpu state ...

mon_get_cpu()
-> mon_get_cpu_sync(mon, true)
   -> cpu_synchronize_state(cpu)
      -> ioctl KVM_GET_LAPIC (taking KVM as example)

... the cpu state is not synchronized when the apic-id is available as
argument.

The cpu state should be synchronized when apic-id is available. Otherwise
the "info lapic <apic-id>" always returns stale data.

Reference:
https://lore.kernel.org/all/20211028155457.967291-19-berrange@redhat.com/

Cc: Joe Jin <joe.jin@oracle.com>
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: David Woodhouse <dwmw@amazon.co.uk>
Message-ID: <20231030085336.2681386-1-armbru@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-ID: <20231026211938.162815-1-dongli.zhang@oracle.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/i386/kvm: Correct comment in kvm_cpu_realize()

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230918160257.30127-4-philmd@linaro.org>

target/i386/hvf: Rename 'X86CPU *x86_cpu' variable as 'cpu'

Follow the naming used by other files in target/i386/.

No functional changes.

Suggested-by: Zhao Liu <zhao1.liu@intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20231020111136.44401-4-philmd@linaro.org>

target/i386/hvf: Rename 'CPUState *cpu' variable as 'cs'

Follow the naming used by other files in target/i386/.

No functional changes.

Suggested-by: Zhao Liu <zhao1.liu@intel.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20231020111136.44401-3-philmd@linaro.org>

target/i386/hvf: Use CPUState typedef

Follow C style guidelines and use CPUState forward
declaration from "qemu/typedefs.h".

No functional changes.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20231020111136.44401-2-philmd@linaro.org>

target/i386/hvf: Use env_archcpu() in simulate_[rdmsr/wrmsr]()

When CPUArchState* is available (here CPUX86State*), we can
use the fast env_archcpu() macro to get ArchCPU* (here X86CPU*).
The QOM cast X86_CPU() macro will be slower when building with
--enable-qom-cast-debug.

Pass CPUX86State* as argument to simulate_rdmsr / simulate_wrmsr
instead of a CPUState* to avoid an extra cast.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Roman Bolshakov <roman@roolebo.dev>
Tested-by: Roman Bolshakov <roman@roolebo.dev>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20231009110239.66778-7-philmd@linaro.org>

target/i386/hvf: Use x86_cpu in simulate_[rdmsr|wrmsr]()

We already have 'x86_cpu = X86_CPU(cpu)'. Use the variable
instead of doing another QOM cast with X86_CPU().

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Roman Bolshakov <roman@roolebo.dev>
Tested-by: Roman Bolshakov <roman@roolebo.dev>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20231009110239.66778-6-philmd@linaro.org>

target/xtensa: Use env_archcpu() in update_c[compare|count]()

When CPUArchState* is available (here CPUXtensaState*), we
can use the fast env_archcpu() macro to get ArchCPU* (here
XtensaCPU*). The QOM cast XTENSA_CPU() macro will be slower
when building with --enable-qom-cast-debug.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20231009110239.66778-5-philmd@linaro.org>

target/s390x: Use env_archcpu() in handle_diag_308()

When CPUArchState* is available (here CPUS390XState*), we
can use the fast env_archcpu() macro to get ArchCPU* (here
S390CPU*). The QOM cast S390_CPU() macro will be slower when
building with --enable-qom-cast-debug.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20231009110239.66778-4-philmd@linaro.org>

target/riscv: Use env_archcpu() in [check_]nanbox()

When CPUArchState* is available (here CPURISCVState*), we
can use the fast env_archcpu() macro to get ArchCPU* (here
RISCVCPU*). The QOM cast RISCV_CPU() macro will be slower
when building with --enable-qom-cast-debug.

Inspired-by: Richard W.M. Jones <rjones@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Richard W.M. Jones <rjones@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20231009110239.66778-3-philmd@linaro.org>

target/ppc: Use env_archcpu() in helper_book3s_msgsndp()

When CPUArchState* is available (here CPUPPCState*), we
can use the fast env_archcpu() macro to get ArchCPU* (here
PowerPCCPU*). The QOM cast POWERPC_CPU() macro will be
slower when building with --enable-qom-cast-debug.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Acked-by: Alistair Francis <alistair.francis@wdc.com>
Message-Id: <20231009110239.66778-2-philmd@linaro.org>

target/riscv: Move TYPE_RISCV_CPU_BASE definition to 'cpu.h'

TYPE_RISCV_CPU_BASE depends on the TARGET_RISCV32/TARGET_RISCV64
definitions which are target specific. Such target specific
definition taints "cpu-qom.h".

Since "cpu-qom.h" must be target agnostic, remove its target
specific definition uses by moving TYPE_RISCV_CPU_BASE to
"target/riscv/cpu.h".

"target/riscv/cpu-qom.h" is now fully target agnostic.
Add a comment clarifying that in the header.

Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-12-philmd@linaro.org>

target/openrisc: Declare QOM definitions in 'cpu-qom.h'

"target/foo/cpu.h" contains the target specific declarations.

A heterogeneous setup need to access target agnostic declarations
(at least the QOM ones, to instantiate the objects).

Our convention is to add such target agnostic QOM declarations in
the "target/foo/cpu-qom.h" header.
Add a comment clarifying that in the header.

Extract QOM definitions from "cpu.h" to "cpu-qom.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-11-philmd@linaro.org>

target/nios2: Declare QOM definitions in 'cpu-qom.h'

"target/foo/cpu.h" contains the target specific declarations.

A heterogeneous setup need to access target agnostic declarations
(at least the QOM ones, to instantiate the objects).

Our convention is to add such target agnostic QOM declarations in
the "target/foo/cpu-qom.h" header.
Add a comment clarifying that in the header.

Extract QOM definitions from "cpu.h" to "cpu-qom.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-10-philmd@linaro.org>

target/loongarch: Declare QOM definitions in 'cpu-qom.h'

"target/foo/cpu.h" contains the target specific declarations.

A heterogeneous setup need to access target agnostic declarations
(at least the QOM ones, to instantiate the objects).

Our convention is to add such target agnostic QOM declarations in
the "target/foo/cpu-qom.h" header.
Add a comment clarifying that in the header.

Extract QOM definitions from "cpu.h" to "cpu-qom.h".

Reviewed-by: Song Gao <gaosong@loongson.cn>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-9-philmd@linaro.org>

target/hexagon: Declare QOM definitions in 'cpu-qom.h'

"target/foo/cpu.h" contains the target specific declarations.

A heterogeneous setup need to access target agnostic declarations
(at least the QOM ones, to instantiate the objects).

Our convention is to add such target agnostic QOM declarations in
the "target/foo/cpu-qom.h" header.
Add a comment clarifying that in the header.

Extract QOM definitions from "cpu.h" to "cpu-qom.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Brian Cain <bcain@quicinc.com>
Message-Id: <20231013140116.255-8-philmd@linaro.org>

target: Declare FOO_CPU_TYPE_NAME/SUFFIX in 'cpu-qom.h'

Hegerogeneous code needs access to the FOO_CPU_TYPE_NAME()
macro to resolve target CPU types. Move the declaration
(along with the required FOO_CPU_TYPE_SUFFIX) to "cpu-qom.h".

"target/foo/cpu-qom.h" is supposed to be target agnostic
(include-able by any target). Add such mention in the
header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Acked-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-7-philmd@linaro.org>

target/riscv: Remove CPU_RESOLVING_TYPE from 'cpu-qom.h'

CPU_RESOLVING_TYPE is a per-target definition, and is
irrelevant for other targets. Move it to "cpu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: LIU Zhiwei <zhiwei_liu@linux.alibaba.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-6-philmd@linaro.org>

target/ppc: Remove CPU_RESOLVING_TYPE from 'cpu-qom.h'

CPU_RESOLVING_TYPE is a per-target definition, and is
irrelevant for other targets. Move it to "cpu.h".

"target/ppc/cpu-qom.h" is supposed to be target agnostic
(include-able by any target). Add such mention in the
header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-5-philmd@linaro.org>

target/arm: Move internal declarations from 'cpu-qom.h' to 'cpu.h'

These definitions and declarations are only used by
target/arm/, no need to expose them to generic hw/.

Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-4-philmd@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <c48c9829-3dfa-79cf-3042-454fda0d00dc@linaro.org>

target: Mention 'cpu-qom.h' is target agnostic

"target/foo/cpu-qom.h" is supposed to be target agnostic
(include-able by any target). Add such mention in the
header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231013140116.255-3-philmd@linaro.org>

target: Unify QOM style

Enforce the style described by commit 067109a11c ("docs/devel:
mention the spacing requirement for QOM"):

  The first declaration of a storage or class structure should
  always be the parent and leave a visual space between that
  declaration and the new code. It is also useful to separate
  backing for properties (options driven by the user) and internal
  state to make navigation easier.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Zhao Liu <zhao1.liu@intel.com>
Message-Id: <20231013140116.255-2-philmd@linaro.org>

accel/tcg: Factor tcg_cpu_reset_hold() out

Factor the TCG specific code from cpu_common_reset_hold() to
tcg_cpu_reset_hold() within tcg-accel-ops.c. Since this file
is sysemu specific, we can inline tcg_flush_softmmu_tlb(),
removing its declaration in "exec/cpu-common.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230918104153.24433-4-philmd@linaro.org>

accel: Introduce cpu_exec_reset_hold()

Introduce cpu_exec_reset_hold() which call an accelerator
specific AccelOpsClass::cpu_reset_hold() handler.

Define a stub on TCG user emulation, because CPU reset is
irrelevant there.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230918104153.24433-3-philmd@linaro.org>

accel/tcg: Declare tcg_flush_jmp_cache() in 'exec/tb-flush.h'

"exec/cpu-common.h" is meant to contain the declarations
related to CPU usable with any accelerator / target
combination.

tcg_flush_jmp_cache() is specific to TCG, so restrict its
declaration by moving it to "exec/tb-flush.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Anton Johansson <anjo@rev.ng>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230918104153.24433-2-philmd@linaro.org>

tests/unit/test-seccomp: Remove mentions of softmmu in test names

Wether we are using a software MMU or not is irrelevant for the
seccomp facility. The facility is restricted to system emulation,
but such detail isn't really helpful, so directly drop the
'softmmu' mention from the test names.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231002145104.52193-3-philmd@linaro.org>

tests/vm/ubuntu.aarch64: Correct comment about TCG specific delay

Wether we use a software MMU or not to set the SSH timeout
isn't really relevant. What we want to know is if we use
a hardware or software accelerator (TCG).
Replace the 'softmmu' mention by 'TCG'.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Alex Bennée <alex.bennee@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231002145104.52193-2-philmd@linaro.org>

vl: constify default_list

It's not modified, let's make it const.

Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231030101529.105266-1-marcandre.lureau@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

vl: Free machine list

Free machine list and make LeakSanitizer happy.

Signed-off-by: Akihiko Odaki <akihiko.odaki@daynix.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20230722062641.18505-1-akihiko.odaki@daynix.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

Merge tag 'pull-pa-20231106' of https://gitlab.com/rth7680/qemu into staging

target/hppa: Implement PA2.0 instructions
hw/hppa: Map astro chip 64-bit I/O mem
hw/hppa: Turn on 64-bit cpu for C3700

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmVJqDEdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV8n5Qf/R15CvXGMgjDJjoV2
# ILMFM+Rpg17SR2yu060sEZ01R3iHdobeCcDB184K0RI9JLrpcBFar+PeF023o9fn
# O9MnfIyL6/ggzaeIpQ9AD2uT0HJMU9hLFoyQqQvnhDHHcT34raL2+Zkrkb2vvauH
# XET7awXN9xYCnY4ALrfcapzlrHqI77ahz0vReUWPxk7eGY2ez8dEOiFW2WLBmuMx
# mAFAMrFQhq66GjoMDl8JiGHD/KBJQ9X4eUAEotS27lTCOYU0ryA6dWBGqBSTWCUa
# smpxkeGQKOew+717HV1H4FdCRYG1Rgm7yFN423JULeew+T7DHvfe0K55vMIulx5I
# g3oVZA==
# =dxC7
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 07 Nov 2023 11:00:01 HKT
# 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-pa-20231106' of https://gitlab.com/rth7680/qemu: (85 commits)
  hw/hppa: Allow C3700 with 64-bit and B160L with 32-bit CPU only
  hw/hppa: Turn on 64-bit CPU for C3700 machine
  hw/pci-host/astro: Trigger CPU irq on CPU HPA in high memory
  hw/pci-host/astro: Map Astro chip into 64-bit I/O memory region
  target/hppa: Improve interrupt logging
  target/hppa: Update IIAOQ, IIASQ for pa2.0
  target/hppa: Create raise_exception_with_ior
  target/hppa: Add unwind_breg to CPUHPPAState
  target/hppa: Clear upper bits in mtctl for pa1.x
  target/hppa: Avoid async_safe_run_on_cpu on uniprocessor system
  target/hppa: Add pa2.0 cpu local tlb flushes
  target/hppa: Implement pa2.0 data prefetch instructions
  linux-user/hppa: Drop EXCP_DUMP from handled exceptions
  hw/hppa: Translate phys addresses for the cpu
  include/hw/elf: Remove truncating signed casts
  target/hppa: Return zero for r0 from load_gpr
  target/hppa: Precompute zero into DisasContext
  target/hppa: Fix interruption based on default PSW
  target/hppa: Implement PERMH
  target/hppa: Implement MIXH, MIXW
  ...

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

Merge tag 'pull-riscv-to-apply-20231107' of https://github.com/alistair23/qemu into staging

Third RISC-V PR for 8.2

 * Rename ext_icboz to ext_zicboz
 * Rename ext_icbom to ext_zicbom
 * Rename ext_icsr to ext_zicsr
 * Rename ext_ifencei to ext_zifencei
 * Add RISC-V Virtual IRQs and IRQ filtering support
 * Change default linux-user cpu to 'max'
 * Update 'virt' machine core limit
 * Add query-cpu-model-expansion API
 * Rename epmp to smepmp and expose the extension
 * Clear pmp/smepmp bits on reset
 * Ignore pmp writes when RW=01
 * Support zicntr/zihpm flags and disable support
 * Correct CSR_MSECCFG operations
 * Update mail address for Weiwei Li
 * Update RISC-V vector crypto to ratified v1.0.0
 * Clear the Ibex/OpenTitan SPI interrupts even if disabled
 * Set the OpenTitan priv to 1.12.0
 * Support discontinuous PMU counters

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEEaukCtqfKh31tZZKWr3yVEwxTgBMFAmVJoOEACgkQr3yVEwxT
# gBPwcw/5AXgSVu521IHpobofq4Skc2rpO9P0Hep3IniBuS+5+h2XM3fwWNBaeeGj
# LZgdXDrCfcCnPuFh2I5j1D885xJDncDF4LET9EFtxK+BTT8eC5JpaCnORdV3Zd2T
# C7qdq1r4J/wKBel3cAz1jlLXc2Pssle4NFaMZGmOGlNX/mLJUYkI6BwKG9wNiCI+
# cCRQW5bEv9g8XzPYPsIKhX9aTegDKdV5x4Xj3YyVs8qkZTVM7Ona8GTpy6eShNfL
# h/RW+yvSxLwfKC9YJHesjI1oqhLsAuA7hFu5AVHiedFNAD5FevMZsZwrqjrmeBOG
# 5awBw9XgfXFFl7jQ0VQVRknt/PFANzTmGGbjLUkaXgJ6iTmH7oIMzwbkx2pM/0Qd
# HV2EboUPe5rJl0SNhcDMCJkYJYpt4z6TVXFpN5p10WU4K1AJXZf9P3YkChcxWiSK
# B4DlY4ax3W77voySwbKCvJRIRWCFQZmtl7doFY5dEQz2ERcNfI7VIB1GKIj7BlGm
# AVTCc5G9KghsaB8q0BzYbDplzCggdaaUBRgpIgLS/n22GKJlOisFwMCawWquPkEw
# i0t3ftt+Ket4Qnnq+dO4W3ehR4qW1/XatCWgQ3NCSgUeS4/9VK3h/nz5t+L7iKwp
# mjp86gNN11wcJRsBIIV7nOAmSAs9ybCm2F4J6YAyh3n1IlRVN0Q=
# =2A+W
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 07 Nov 2023 10:28:49 HKT
# gpg:                using RSA key 6AE902B6A7CA877D6D659296AF7C95130C538013
# gpg: Good signature from "Alistair Francis <alistair@alistair23.me>" [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: 6AE9 02B6 A7CA 877D 6D65  9296 AF7C 9513 0C53 8013

* tag 'pull-riscv-to-apply-20231107' of https://github.com/alistair23/qemu: (49 commits)
  docs/about/deprecated: Document RISC-V "pmu-num" deprecation
  target/riscv: Add "pmu-mask" property to replace "pmu-num"
  target/riscv: Use existing PMU counter mask in FDT generation
  target/riscv: Don't assume PMU counters are continuous
  target/riscv: Propagate error from PMU setup
  target/riscv: cpu: Set the OpenTitan priv to 1.12.0
  hw/ssi: ibex_spi_host: Clear the interrupt even if disabled
  disas/riscv: Replace TABs with space
  disas/riscv: Add support for vector crypto extensions
  disas/riscv: Add rv_codec_vror_vi for vror.vi
  disas/riscv: Add rv_fmt_vd_vs2_uimm format
  target/riscv: Move vector crypto extensions to riscv_cpu_extensions
  target/riscv: Expose Zvks[c|g] extnesion properties
  target/riscv: Add cfg properties for Zvks[c|g] extensions
  target/riscv: Expose Zvkn[c|g] extnesion properties
  target/riscv: Add cfg properties for Zvkn[c|g] extensions
  target/riscv: Expose Zvkb extension property
  target/riscv: Replace Zvbb checking by Zvkb
  target/riscv: Add cfg property for Zvkb extension
  target/riscv: Expose Zvkt extension property
  ...

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

Merge tag 'pull-tcg-20231106' of https://gitlab.com/rth7680/qemu into staging

util: Add cpuinfo for loongarch64
tcg/loongarch64: Use cpuinfo.h
tcg/loongarch64: Improve register allocation for INDEX_op_qemu_ld_a*_i128
host/include/loongarch64: Add atomic16 load and store
tcg: Move expanders out of line
tcg/mips: Always implement movcond
tcg/mips: Implement neg opcodes
tcg/loongarch64: Implement neg opcodes
tcg: Make movcond and neg required opcodes
tcg: Optimize env memory operations
tcg: Canonicalize sub of immediate to add
tcg/sparc64: Implement tcg_out_extrl_i64_i32

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmVJpT0dHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV9i7QgAtjxUB3y/caCPp0Me
# 3cXYtpL1vNxx+cTESGMlmIRSji+cEOxYSpnY0itxXcKpcwP8Au8eoTe85NxyIllg
# 2R/SA2jlmrmiipI+bwb0UBCy+BzUfMgmegA88K2W22J0fetwIy19PN9ORmYdLiYE
# /pWNFOSPzhYEJgOw7V2MwciUv3llolMOfxU7VT4oVaCknZRsyaGUwl4uTT4GdPuK
# p29O9nziyKDmNTqJ9SKKll5bzwCMAgkn2lUcMGf+rpl7ZxjgvysUYrGXKmOnj4Uu
# eCU2d3ZHoSspcYEjbFASlyPd7z5apGI8Iq2K35FUhURFPv06Su/bIGOOD4ujP2Qp
# vc/bFQ==
# =Mvaf
# -----END PGP SIGNATURE-----
# gpg: Signature made Tue 07 Nov 2023 10:47:25 HKT
# 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-20231106' of https://gitlab.com/rth7680/qemu: (35 commits)
  tcg/sparc64: Implement tcg_out_extrl_i64_i32
  tcg/optimize: Canonicalize sub2 with constants to add2
  tcg/optimize: Canonicalize subi to addi during optimization
  tcg: Canonicalize subi to addi during opcode generation
  tcg/optimize: Split out arg_new_constant
  tcg: Eliminate duplicate env store operations
  tcg/optimize: Optimize env memory operations
  tcg/optimize: Split out cmp_better_copy
  tcg/optimize: Pipe OptContext into reset_ts
  tcg: Don't free vector results
  tcg: Remove TCG_TARGET_HAS_neg_{i32,i64}
  tcg/loongarch64: Implement neg opcodes
  tcg/mips: Implement neg opcodes
  tcg: Remove TCG_TARGET_HAS_movcond_{i32,i64}
  tcg/mips: Always implement movcond
  tcg/mips: Split out tcg_out_setcond_int
  tcg: Move tcg_temp_free_* out of line
  tcg: Move tcg_temp_new_*, tcg_global_mem_new_* out of line
  tcg: Move tcg_constant_* out of line
  tcg: Unexport tcg_gen_op*_{i32,i64}
  ...

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

hw/hppa: Allow C3700 with 64-bit and B160L with 32-bit CPU only

Prevent that users try to boot a 64-bit only C3700 machine with a 32-bit
CPU, and to boot a 32-bit only B160L machine with a 64-bit CPU.

Signed-off-by: Helge Deller <deller@gmx.de>

hw/hppa: Turn on 64-bit CPU for C3700 machine

Signed-off-by: Helge Deller <deller@gmx.de>

hw/pci-host/astro: Trigger CPU irq on CPU HPA in high memory

The CPU HPA is in the high F-region on PA2.0 CPUs, so use F_EXTEND()
to trigger interrupt request at the right CPU HPA address.
Note that the cpu_hpa value comes out of the IRT, which doesn't store the
higher addresss bits.

Signed-off-by: Helge Deller <deller@gmx.de>

hw/pci-host/astro: Map Astro chip into 64-bit I/O memory region

Map Astro into high F-region and add alias for 32-bit OS in low region.

Signed-off-by: Helge Deller <deller@gmx.de>

target/hppa: Improve interrupt logging

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Update IIAOQ, IIASQ for pa2.0

These registers have a different format for pa2.0.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Create raise_exception_with_ior

Handle pa2.0 logic for filling in ISR+IOR.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Add unwind_breg to CPUHPPAState

Fill in the insn_start value during form_gva, and copy
it out to the env field in hppa_restore_state_to_opc.
The value is not yet consumed.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Clear upper bits in mtctl for pa1.x

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

target/hppa: Avoid async_safe_run_on_cpu on uniprocessor system

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Add pa2.0 cpu local tlb flushes

The previous decoding misnamed the bit it called "local".
Other than the name, the implementation was correct for pa1.x.
Rename this field to "tlbe".

PA2.0 adds (a real) local bit to PxTLB, and also adds a range
of pages to flush in GR[b].

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

target/hppa: Implement pa2.0 data prefetch instructions

These are aliased onto the normal integer loads to %g0.
Since we don't emulate caches, prefetch is a nop.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

linux-user/hppa: Drop EXCP_DUMP from handled exceptions

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

hw/hppa: Translate phys addresses for the cpu

Hack the machine to use pa2.0 physical layout when required,
using the PSW.W=0 absolute to physical mapping.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

include/hw/elf: Remove truncating signed casts

There's nothing about elf that specifically requires signed vs unsigned.
This is very much a target-specific preference.

In the meantime, casting low and high from uint64_t back to Elf_SWord
to uint64_t discards high bits that might have been set by translate_fn.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Return zero for r0 from load_gpr

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

target/hppa: Precompute zero into DisasContext

Reduce the number of times we look for the constant 0.

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

target/hppa: Fix interruption based on default PSW

The default PSW is set by the operating system with the PDC_PSW
firmware call.  Use that setting to decide if wide mode is to be
enabled for interruptions and EIRR usage.

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

target/hppa: Implement PERMH

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Implement MIXH, MIXW

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Implement HSHLADD, HSHRADD

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>

target/hppa: Implement HSHL, HSHR

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>