target/riscv: Fix mcycle/minstret increment behavior

The mcycle/minstret counter's stop flag is mistakenly updated on a copy
on stack. Thus the counter increments even when the CY/IR bit in the
mcountinhibit register is set. This commit corrects its behavior.

Fixes: 3780e33732f88 (target/riscv: Support mcycle/minstret write operation)
Signed-off-by: Xu Lu <luxu.kernel@bytedance.com>
Reviewed-by: Daniel Henrique Barboza <dbarboza@ventanamicro.com>
Signed-off-by: Michael Tokarev <mjt@tls.msk.ru>

Merge tag 'hw-cpus-20240105' of https://github.com/philmd/qemu into staging

HW core patch queue

- Unify CPU QOM type checks (Gavin)
- Simplify uses of some CPU related property (Philippe)
  (start-powered-off, ARM reset-cbar and mp-affinity)
- Header and documentation cleanups (Zhao, Philippe)
- Have Memory API return boolean indicating possible error
- Fix frame filter mask in CAN sja1000 model (Pavel)
- QOM embed MCF5206 timer into SoC (Thomas)
- Simplify LEON3 qemu_irq_ack handler (Clément)

# -----BEGIN PGP SIGNATURE-----
#
# iQIzBAABCAAdFiEE+qvnXhKRciHc/Wuy4+MsLN6twN4FAmWYIxwACgkQ4+MsLN6t
# wN66fA//UBwgYqcdpg6Wz17qzgq1TWeZHHzYh7HbZRUCxhdSgS6TSQOH9Fi8VNYq
# Ed5a5l4ovP/2NRN1/S5PPBydyKXTU7wintHm2+suQbLSmplIE6yr0Ca6o8FLEeJ3
# hnE0dAoQCLS7eDpoeOEpGjzmJFiBSWLvyqAZLa/rZkCnCiZRHB6g/nAEM8I3I9bl
# //H20d3a/fektZxGnpEAeoMxrl4iA9hkFYVW8lbu6EhNFBPUkkj5Y8w47Kq/BIvD
# NmLTPgu4d7oahwlfsM6jWdRDG9zlEkXQor817PHwl00o45yAfeITsy40GvJeEYaI
# BcDLFfWrSm9SQb7/suXGeyU/SLmx7rsmJWfNYUoMr6807QcSH4ScPCfgzEQ4j8IV
# PmeVsxxLxT9CSzfxhMx5cXt33H2l+tEzwJ5UJCLQvmvTu+aDkt46Q09X/7j0z89m
# zSk/HBtdACIzwEWBAJsKuzarRTZNUvyXEsOxZ5l7xOxJpzpsNV2YVuChClVGtHOJ
# kr1PE2hxEMPY1vDyKU6ckDvW+XXgYhOXrPAxdx8gIwwd4oyDC5vVlIajvlqbOAsp
# Es7zq40b/is3ZnByEDbZ+yYvdYRLtVf/lDPK3KIv7IhrTNzH/HT1egshOQAVirY1
# Gw8f3fXqL3/84w383VI4efrSlKBJeb0i2SJ50y2N1clrF1qnlx0=
# =an4B
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 05 Jan 2024 15:41:16 GMT
# gpg:                using RSA key FAABE75E12917221DCFD6BB2E3E32C2CDEADC0DE
# gpg: Good signature from "Philippe Mathieu-Daudé (F4BUG) <f4bug@amsat.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: FAAB E75E 1291 7221 DCFD  6BB2 E3E3 2C2C DEAD C0DE

* tag 'hw-cpus-20240105' of https://github.com/philmd/qemu: (71 commits)
  target/sparc: Simplify qemu_irq_ack
  hw/net/can/sja1000: fix bug for single acceptance filter and standard frame
  hw/m68k/mcf5206: Embed m5206_timer_state in m5206_mbar_state
  hw/pci-host/raven: Propagate error in raven_realize()
  hw/nvram: Simplify memory_region_init_rom_device() calls
  hw/misc: Simplify memory_region_init_ram_from_fd() calls
  hw/sparc: Simplify memory_region_init_ram_nomigrate() calls
  hw/arm: Simplify memory_region_init_rom() calls
  hw: Simplify memory_region_init_ram() calls
  misc: Simplify qemu_prealloc_mem() calls
  util/oslib: Have qemu_prealloc_mem() handler return a boolean
  backends: Reduce variable scope in host_memory_backend_memory_complete
  backends: Have HostMemoryBackendClass::alloc() handler return a boolean
  backends: Simplify host_memory_backend_memory_complete()
  backends: Use g_autofree in HostMemoryBackendClass::alloc() handlers
  memory: Have memory_region_init_ram_from_fd() handler return a boolean
  memory: Have memory_region_init_ram_from_file() handler return a boolean
  memory: Have memory_region_init_resizeable_ram() return a boolean
  memory: Have memory_region_init_rom_device() handler return a boolean
  memory: Simplify memory_region_init_rom_device_nomigrate() calls
  ...

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

target/sparc: Simplify qemu_irq_ack

This is a simple cleanup, since env is passed to qemu_irq_ack it can be
accessed from inside qemu_irq_ack.  Just drop this parameter.

Co-developed-by: Frederic Konrad <konrad.frederic@yahoo.fr>
Signed-off-by: Clément Chigot <chigot@adacore.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20240105102421.163554-7-chigot@adacore.com>

hw/net/can/sja1000: fix bug for single acceptance filter and standard frame

A CAN sja1000 standard frame filter mask has been computed and applied
incorrectly for standard frames when single Acceptance Filter Mode
(MOD_AFM = 1) has been selected. The problem has not been found
by Linux kernel testing because it uses dual filter mode (MOD_AFM = 0)
and leaves falters fully open.

The problem has been noticed by Grant Ramsay when testing with Zephyr
RTOS which uses single filter mode.

Signed-off-by: Pavel Pisa <pisa@cmp.felk.cvut.cz>
Reported-by: Grant Ramsay <gramsay@enphaseenergy.com>
Resolves: https://gitlab.com/qemu-project/qemu/-/issues/2028
Fixes: 733210e754 ("hw/net/can: SJA1000 chip register level emulation")
Message-ID: <20240103231426.5685-1-pisa@fel.cvut.cz>

hw/m68k/mcf5206: Embed m5206_timer_state in m5206_mbar_state

There's no need to explicitely allocate the memory here, we can
simply embed it into the m5206_mbar_state instead.

Signed-off-by: Thomas Huth <huth@tuxfamily.org>
Reviewed-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231221122939.11001-1-huth@tuxfamily.org>

hw/pci-host/raven: Propagate error in raven_realize()

When an Error** reference is available, it is better to
propagate local errors, rather then using generic ones,
which might terminate the whole QEMU process.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-26-philmd@linaro.org>

hw/nvram: Simplify memory_region_init_rom_device() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, arg5, arg6, errp;
@@
-   memory_region_init_rom_device(mr, owner, arg3, arg4, arg5, arg6, &errp);
    if (
-       errp
+       !memory_region_init_rom_device(mr, owner, arg3, arg4, arg5, arg6, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-25-philmd@linaro.org>

hw/misc: Simplify memory_region_init_ram_from_fd() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, arg5, arg6, arg7, errp;
@@
-   memory_region_init_ram_from_fd(mr, owner, arg3, arg4, arg5, arg6, arg7, &errp);
    if (
-       errp
+       !memory_region_init_ram_from_fd(mr, owner, arg3, arg4, arg5, arg6, arg7, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-24-philmd@linaro.org>

hw/sparc: Simplify memory_region_init_ram_nomigrate() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, errp;
@@
-   memory_region_init_ram_nomigrate(mr, owner, arg3, arg4, &errp);
    if (
-       errp
+       !memory_region_init_ram_nomigrate(mr, owner, arg3, arg4, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-23-philmd@linaro.org>

hw/arm: Simplify memory_region_init_rom() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, errp;
@@
-   memory_region_init_rom(mr, owner, arg3, arg4, &errp);
    if (
-       errp
+       !memory_region_init_rom(mr, owner, arg3, arg4, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-22-philmd@linaro.org>

hw: Simplify memory_region_init_ram() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, errp;
@@
-   memory_region_init_ram(mr, owner, arg3, arg4, &errp);
    if (
-       errp
+       !memory_region_init_ram(mr, owner, arg3, arg4, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Andrew Jeffery <andrew@codeconstruct.com.au> # aspeed
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-21-philmd@linaro.org>

misc: Simplify qemu_prealloc_mem() calls

Since qemu_prealloc_mem() returns whether or not an error
occured, we don't need to check the @errp pointer. Remove
local_err uses when we can return directly.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-20-philmd@linaro.org>

util/oslib: Have qemu_prealloc_mem() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-19-philmd@linaro.org>

backends: Reduce variable scope in host_memory_backend_memory_complete

Reduce the &local_err variable use and remove the 'out:' label.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-18-philmd@linaro.org>

backends: Have HostMemoryBackendClass::alloc() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-17-philmd@linaro.org>

backends: Simplify host_memory_backend_memory_complete()

Return early if bc->alloc is NULL. De-indent the if() ladder.

Note, this avoids a pointless call to error_propagate() with
errp=NULL at the 'out:' label.

Change trivial when reviewed with 'git-diff --ignore-all-space'.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-16-philmd@linaro.org>

backends: Use g_autofree in HostMemoryBackendClass::alloc() handlers

In preparation of having HostMemoryBackendClass::alloc() handlers
return a boolean, have them use g_autofree.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-15-philmd@linaro.org>

memory: Have memory_region_init_ram_from_fd() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-14-philmd@linaro.org>

memory: Have memory_region_init_ram_from_file() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-13-philmd@linaro.org>

memory: Have memory_region_init_resizeable_ram() return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-12-philmd@linaro.org>

memory: Have memory_region_init_rom_device() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-11-philmd@linaro.org>

memory: Simplify memory_region_init_rom_device_nomigrate() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, arg5, arg6, errp;
@@
-   memory_region_init_rom_device_nomigrate(mr, owner, arg3, arg4, arg5, arg6, &errp);
    if (
-       errp
+       !memory_region_init_rom_device_nomigrate(mr, owner, arg3, arg4, arg5, arg6, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-10-philmd@linaro.org>

memory: Have memory_region_init_rom_device_nomigrate() return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-9-philmd@linaro.org>

memory: Have memory_region_init_rom() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-8-philmd@linaro.org>

memory: Have memory_region_init_ram() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-7-philmd@linaro.org>

memory: Simplify memory_region_init_ram_from_fd() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, arg5, arg6, arg7, errp;
@@
-   memory_region_init_ram_from_fd(mr, owner, arg3, arg4, arg5, arg6, arg7, &errp);
    if (
-       errp
+       !memory_region_init_ram_from_fd(mr, owner, arg3, arg4, arg5, arg6, arg7, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-6-philmd@linaro.org>

memory: Simplify memory_region_init_rom_nomigrate() calls

Mechanical change using the following coccinelle script:

@@
expression mr, owner, arg3, arg4, errp;
@@
-   memory_region_init_rom_nomigrate(mr, owner, arg3, arg4, &errp);
    if (
-       errp
+       !memory_region_init_rom_nomigrate(mr, owner, arg3, arg4, &errp)
    ) {
        ...
        return;
    }

and removing the local Error variable.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-5-philmd@linaro.org>

memory: Have memory_region_init_rom_nomigrate() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-4-philmd@linaro.org>
[PMD: Only update 'readonly' field on success (Manos Pitsidianakis)]
Message-Id: <af352e7d-3346-4705-be77-6eed86858d18@linaro.org>

memory: Have memory_region_init_ram_nomigrate() handler return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-3-philmd@linaro.org>

memory: Have memory_region_init_ram_flags_nomigrate() return a boolean

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

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Manos Pitsidianakis <manos.pitsidianakis@linaro.org>
Reviewed-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Message-Id: <20231120213301.24349-2-philmd@linaro.org>

hw/mips: Inline 'bios.h' definitions

There is no universal BIOS, each machine needs a specific one.

Move the machine-specific definitions to each machine code and
remove this bogus header.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231122184334.18201-1-philmd@linaro.org>

hw/ppc/xive2_regs: Remove unnecessary 'cpu.h' inclusion

xive2_regs.h only requires declarations from "qemu/bswap.h".
Include it instead of the huge target-specific "cpu.h".

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Cédric Le Goater <clg@redhat.com>
Message-Id:  <20231122183920.17905-1-philmd@linaro.org>

hw: Simplify accesses to the CPUState::'start-powered-off' property

The 'start-powered-off' property has been added to ARM CPUs in
commit 5de164304a ("arm: Allow secondary KVM CPUs to be booted
via PSCI"), then eventually got generalized to all CPUs in commit
c1b701587e ("target/arm: Move start-powered-off property to generic
CPUState"). Since all CPUs have it, no need to check whether it is
available. Updating this property can't fail, so use &error_abort.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20231123143813.42632-5-philmd@linaro.org>

hw/ppc/spapr_cpu_core: Access QDev properties with proper API

CPUState::start_powered_off field is part of the internal
implementation of a QDev CPU. It is exposed as the QDev
"start-powered-off" property. External components should
use the qdev properties API to access it.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Harsh Prateek Bora <harshpb@linux.ibm.com>
Message-Id: <20231123143813.42632-2-philmd@linaro.org>

hw/arm/bcm2836: Use ARM_CPU 'mp-affinity' property

The 'mp-affinity' property is present since commit 15a21fe028
("target-arm: Add mp-affinity property for ARM CPU class").
Use it and remove a /* TODO */ comment. Since all ARM CPUs
have this property, use &error_abort, because this call can
not fail.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20231123143813.42632-4-philmd@linaro.org>

hw/arm/bcm2836: Simplify use of 'reset-cbar' property

bcm2836_realize() is called by

 - bcm2836_class_init() which sets:

    bc->cpu_type = ARM_CPU_TYPE_NAME("cortex-a7")

 - bcm2837_class_init() which sets:

    bc->cpu_type = ARM_CPU_TYPE_NAME("cortex-a53")

Both Cortex-A7 / A53 have the ARM_FEATURE_CBAR set. If it isn't,
then this is a programming error: use &error_abort.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Message-Id: <20231123143813.42632-3-philmd@linaro.org>

hw/audio/sb16: Do not migrate qdev properties

Since commit f7b4f61f63 ("qdev/isa: convert soundblaster") these
fields have been converted to qdev properties, so don't need to
be migrated:

  static Property sb16_properties[] = {
      DEFINE_AUDIO_PROPERTIES(SB16State, card),
      DEFINE_PROP_UINT32 ("version", SB16State, ver,  0x0405), /* 4.5 */
      DEFINE_PROP_UINT32 ("iobase",  SB16State, port, 0x220),
      DEFINE_PROP_UINT32 ("irq",     SB16State, irq,  5),
      DEFINE_PROP_UINT32 ("dma",     SB16State, dma,  1),
      DEFINE_PROP_UINT32 ("dma16",   SB16State, hdma, 5),
      DEFINE_PROP_END_OF_LIST (),
  };

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Message-Id: <20231124182615.94943-1-philmd@linaro.org>

hw/cpu/cluster: Cleanup unused included header in cluster.c

Remove unused header (qemu/module.h and qemu/cutils.h) in cluster.c,
and reorder the remaining header files (except qemu/osdep.h) in
alphabetical order.

Tested by "./configure" and then "make".

Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231127145611.925817-3-zhao1.liu@linux.intel.com>

hw/cpu/core: Cleanup unused included header in core.c

Remove unused header (qemu/module.h and sysemu/cpus.h) in core.c,
and reorder the remaining header files (except qemu/osdep.h) in
alphabetical order.

Tested by "./configure" and then "make".

Signed-off-by: Zhao Liu <zhao1.liu@intel.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231127145611.925817-2-zhao1.liu@linux.intel.com>

hw/core/cpu: Update description of CPUState::node

'next_cpu' was converted to 'node' in commit bdc44640cb
("cpu: Use QTAILQ for CPU list").

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231129183243.15859-1-philmd@linaro.org>

hw/core/cpu: Remove final vestiges of dynamic state tracing

The dynamic state tracing was removed in commit d0aaf08bb9.

Fixes: d0aaf08bb9 ("tcg: remove the final vestiges of dstate")
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231129182734.15565-1-philmd@linaro.org>

hw/riscv/shakti_c: Check CPU type in machine_run_board_init()

Set mc->valid_cpu_types so that the user specified CPU type can
be validated in machine_run_board_init(). We needn't to do it
by ourselves.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231204004726.483558-10-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/arm: Check CPU type in machine_run_board_init()

Set mc->valid_cpu_types so that the user specified CPU type can
be validated in machine_run_board_init(). We needn't to do it by
ourselves.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231204004726.483558-9-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/arm/sbsa-ref: Check CPU type in machine_run_board_init()

Set mc->valid_cpu_types so that the user specified CPU type can
be validated in machine_run_board_init(). We needn't to do it
by ourselves.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Leif Lindholm <quic_llindhol@quicinc.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231204004726.483558-8-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/arm/virt: Check CPU type in machine_run_board_init()

Set mc->valid_cpu_types so that the user specified CPU type can be
validated in machine_run_board_init(). We needn't to do the check
by ourselves.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231204004726.483558-7-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/arm/virt: Hide host CPU model for tcg

The 'host' CPU model isn't available until KVM or HVF is enabled.
For example, the following error messages are seen when the guest
is started with option '-cpu cortex-a8' on tcg after the next commit
is applied to check the CPU type in machine_run_board_init().

  ERROR:../hw/core/machine.c:1423:is_cpu_type_supported: \
  assertion failed: (model != NULL)
  Bail out! ERROR:../hw/core/machine.c:1423:is_cpu_type_supported: \
  assertion failed: (model != NULL)
  Aborted (core dumped)

Hide 'host' CPU model until KVM or HVF is enabled. With this applied,
the valid CPU models can be shown.

  qemu-system-aarch64: Invalid CPU type: cortex-a8
  The valid types are: cortex-a7, cortex-a15, cortex-a35, \
  cortex-a55, cortex-a72, cortex-a76, cortex-a710, a64fx, \
  neoverse-n1, neoverse-v1, neoverse-n2, cortex-a53,      \
  cortex-a57, max

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231204004726.483558-6-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

machine: Print CPU model name instead of CPU type

The names of supported CPU models instead of CPU types should be
printed when the user specified CPU type isn't supported, to be
consistent with the output from '-cpu ?'.

Correct the error messages to print CPU model names instead of CPU
type names.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231204004726.483558-5-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

machine: Improve is_cpu_type_supported()

It's no sense to check the CPU type when mc->valid_cpu_types[0] is
NULL, which is a program error. Raise an assert on this.

A precise hint for the error message is given when mc->valid_cpu_types[0]
is the only valid entry. Besides, enumeration on mc->valid_cpu_types[0]
when we have mutiple valid entries there is avoided to increase the code
readability, as suggested by Philippe Mathieu-Daudé.

Besides, @cc comes from machine->cpu_type or mc->default_cpu_type. For
the later case, it can be NULL and it's also a program error. We should
use assert() in this case.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Message-ID: <20231204004726.483558-4-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

machine: Introduce helper is_cpu_type_supported()

The logic, to check if the specified CPU type is supported in
machine_run_board_init(), is independent enough. Factor it out into
helper is_cpu_type_supported(). machine_run_board_init() looks a bit
clean with this. Since we're here, @machine_class is renamed to @mc to
avoid multiple line spanning of code. The comments are tweaked a bit
either.

No functional change intended.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231204004726.483558-3-gshan@redhat.com>
[PMD: Only call new helper if machine->cpu_type is not NULL]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

machine: Use error handling when CPU type is checked

Functions that use an Error **errp parameter to return errors should
not also report them to the user, because reporting is the caller's
job. The principle is violated by machine_run_board_init() because
it calls error_report(), error_printf(), and exit(1) when the machine
doesn't support the requested CPU type.

Clean this up by using error_setg() and error_append_hint() instead.
No functional change, as the only caller passes &error_fatal.

Suggested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Markus Armbruster <armbru@redhat.com>
Message-ID: <20231204004726.483558-2-gshan@redhat.com>
[PMD: Correct error_append_hint() argument]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

hw/core: Add machine_class_default_cpu_type()

Add a helper to return a machine default CPU type.

If this machine is restricted to a single CPU type,
use it as default, obviously.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20231116163726.28952-1-philmd@linaro.org>

target: Use generic cpu_model_from_type()

Use generic cpu_model_from_type() when the CPU model name needs to
be extracted from the CPU type name.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-23-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/xtensa: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-xtensa -cpu ?
Available CPUs:
  test_mmuhifi_c3
  sample_controller
  lx106
  dsp3400
  de233_fpu
  de212
  dc233c
  dc232b

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-xtensa -cpu ?
Available CPUs:
  dc232b
  dc233c
  de212
  de233_fpu
  dsp3400
  lx106
  sample_controller
  test_mmuhifi_c3

Signed-off-by: Gavin Shan <gshan@redhat.com>
Message-ID: <20231114235628.534334-22-gshan@redhat.com>
[PMD: Split patch in 2, only include the "Use generic cpu_list" change]
Message-ID: <51ffd060-b2f8-405c-83e1-a0663c0183f5@linaro.org>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/tricore: Use generic cpu_list()

No changes in the output from the following command.

[gshan@gshan q]$ ./build/qemu-system-tricore -cpu ?
Available CPUs:
  tc1796
  tc1797
  tc27x
  tc37x

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-21-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/sh4: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-sh4 -cpu ?
sh7750r
sh7751r
sh7785

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-sh4 -cpu ?
Available CPUs:
  sh7750r
  sh7751r
  sh7785

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-20-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/rx: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-rx -cpu ?
Available CPUs:
  rx62n-rx-cpu

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-rx -cpu ?
Available CPUs:
  rx62n

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-19-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/riscv: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-riscv64 -cpu ?
any
max
rv64
shakti-c
sifive-e51
sifive-u54
thead-c906
veyron-v1
x-rv128

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-riscv64 -cpu ?
Available CPUs:
  any
  max
  rv64
  shakti-c
  sifive-e51
  sifive-u54
  thead-c906
  veyron-v1
  x-rv128

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-18-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/openrisc: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-or1k -cpu ?
Available CPUs:
  or1200
  any

After it's applied:

[gshan@gshan q]$ ./build/qemu-or1k -cpu ?
Available CPUs:
  any
  or1200

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-17-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/mips: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-mips64 -cpu ?
MIPS '4Kc'
MIPS '4Km'
MIPS '4KEcR1'
MIPS 'XBurstR1'
MIPS '4KEmR1'
MIPS '4KEc'
MIPS '4KEm'
MIPS '24Kc'
MIPS '24KEc'
MIPS '24Kf'
MIPS '34Kf'
MIPS '74Kf'
MIPS 'XBurstR2'
MIPS 'M14K'
MIPS 'M14Kc'
MIPS 'P5600'
MIPS 'mips32r6-generic'
MIPS 'I7200'
MIPS 'R4000'
MIPS 'VR5432'
MIPS '5Kc'
MIPS '5Kf'
MIPS '20Kc'
MIPS 'MIPS64R2-generic'
MIPS '5KEc'
MIPS '5KEf'
MIPS 'I6400'
MIPS 'I6500'
MIPS 'Loongson-2E'
MIPS 'Loongson-2F'
MIPS 'Loongson-3A1000'
MIPS 'Loongson-3A4000'
MIPS 'mips64dspr2'
MIPS 'Octeon68XX'

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-mips64 -cpu ?
Available CPUs:
  20Kc
  24Kc
  24KEc
  24Kf
  34Kf
  4Kc
  4KEc
  4KEcR1
  4KEm
  4KEmR1
  4Km
  5Kc
  5KEc
  5KEf
  5Kf
  74Kf
  I6400
  I6500
  I7200
  Loongson-2E
  Loongson-2F
  Loongson-3A1000
  Loongson-3A4000
  M14K
  M14Kc
  mips32r6-generic
  mips64dspr2
  MIPS64R2-generic
  Octeon68XX
  P5600
  R4000
  VR5432
  XBurstR1
  XBurstR2

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-16-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/m68k: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-m68k -cpu ?
cfv4e
m5206
m5208
m68000
m68010
m68020
m68030
m68040
m68060
any

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-m68k -cpu ?
Available CPUs:
  any
  cfv4e
  m5206
  m5208
  m68000
  m68010
  m68020
  m68030
  m68040
  m68060

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-15-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/loongarch: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-loongarch64 -cpu ?
la132-loongarch-cpu
la464-loongarch-cpu
max-loongarch-cpu

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-loongarch64 -cpu ?
Available CPUs:
  la132
  la464
  max

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-14-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/hppa: Use generic cpu_list()

No changes in the output from the following command.

[gshan@gshan q]$ ./build/qemu-system-hppa -cpu ?
Available CPUs:
  hppa
  hppa64

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-13-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/hexagon: Use generic cpu_list()

No changes in the output from the following command.

[gshan@gshan q]$ ./build/qemu-hexagon -cpu ?
Available CPUs:
  v67
  v68
  v69
  v71
  v73

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-12-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/cris: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-cris -cpu ?
Available CPUs:
  crisv8
  crisv9
  crisv10
  crisv11
  crisv17
  crisv32

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-cris -cpu ?
Available CPUs:
  crisv10
  crisv11
  crisv17
  crisv32
  crisv8
  crisv9

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-11-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/avr: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-avr -cpu ?
avr5-avr-cpu
avr51-avr-cpu
avr6-avr-cpu

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-avr -cpu ?
Available CPUs:
  avr5
  avr51
  avr6

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-10-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/arm: Use generic cpu_list()

No changes of the output from the following command before and
after it's applied.

[gshan@gshan q]$ ./build/qemu-system-aarch64 -cpu ?
Available CPUs:
  a64fx
  arm1026
  arm1136
  arm1136-r2
  arm1176
  arm11mpcore
  arm926
  arm946
  cortex-a15
  cortex-a35
  cortex-a53
  cortex-a55
  cortex-a57
  cortex-a7
  cortex-a710
  cortex-a72
  cortex-a76
  cortex-a8
  cortex-a9
  cortex-m0
  cortex-m3
  cortex-m33
  cortex-m4
  cortex-m55
  cortex-m7
  cortex-r5
  cortex-r52
  cortex-r5f
  max
  neoverse-n1
  neoverse-n2
  neoverse-v1
  pxa250
  pxa255
  pxa260
  pxa261
  pxa262
  pxa270-a0
  pxa270-a1
  pxa270
  pxa270-b0
  pxa270-b1
  pxa270-c0
  pxa270-c5
  sa1100
  sa1110
  ti925t

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-9-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/alpha: Use generic cpu_list()

Before it's applied:

[gshan@gshan q]$ ./build/qemu-system-alpha -cpu ?
Available CPUs:
  ev4-alpha-cpu
  ev5-alpha-cpu
  ev56-alpha-cpu
  ev6-alpha-cpu
  ev67-alpha-cpu
  ev68-alpha-cpu
  pca56-alpha-cpu

After it's applied:

[gshan@gshan q]$ ./build/qemu-system-alpha -cpu ?
Available CPUs:
  ev4
  ev5
  ev56
  ev6
  ev67
  ev68
  pca56

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-8-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

cpu: Add generic cpu_list()

Add generic cpu_list() to replace the individual target's implementation
in the subsequent commits. Currently, there are 3 targets with no cpu_list()
implementation: microblaze and nios2. With this applied, those two targets
switch to the generic cpu_list().

[gshan@gshan q]$ ./build/qemu-system-microblaze -cpu ?
Available CPUs:
  microblaze-cpu

[gshan@gshan q]$ ./build/qemu-system-nios2 -cpu ?
Available CPUs:
  nios2-cpu

Suggested-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-7-gshan@redhat.com>
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

cpu: Add helper cpu_model_from_type()

Add helper cpu_model_from_type() to extract the CPU model name from
the CPU type name in two circumstances: (1) The CPU type name is the
combination of the CPU model name and suffix. (2) The CPU type name
is same to the CPU model name.

The helper will be used in the subsequent commits to conver the
CPU type name to the CPU model name.

Suggested-by: Igor Mammedov <imammedo@redhat.com>
Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-ID: <20231114235628.534334-6-gshan@redhat.com>
[PMD: Mention returned string must be released with g_free()]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

cpu: Call object_class_dynamic_cast() once in cpu_class_by_name()

For all targets, the CPU class returned from CPUClass::class_by_name()
and object_class_dynamic_cast(oc, CPU_RESOLVING_TYPE) need to be
compatible. Lets apply the check in cpu_class_by_name() for once,
instead of having the check in CPUClass::class_by_name() for individual
target.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Igor Mammedov <imammedo@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Gavin Shan <gshan@redhat.com>
Message-ID: <20231114235628.534334-4-gshan@redhat.com>

target/hppa: Remove object_class_is_abstract()

Since commit 3a9d0d7b64 ("hw/cpu: Call object_class_is_abstract()
once in cpu_class_by_name()"), there is no need to check if @oc is
abstract because it has been covered by cpu_class_by_name().

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-3-gshan@redhat.com>
[PMD: Mention commit 3a9d0d7b64]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

target/alpha: Remove fallback to ev67 cpu class

'ev67' CPU class will be returned to match everything, which makes
no sense as mentioned in the comments. Remove the logic to fall
back to 'ev67' CPU class to match everything.

Signed-off-by: Gavin Shan <gshan@redhat.com>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-ID: <20231114235628.534334-2-gshan@redhat.com>
[PMD: Reword subject, replace 'any' -> 'ev67' on linux-user]
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>

meson: Allow building binary with no target-specific files in hw/

Allow  building a qemu-system-foo binary with target-agnostic
only HW models.

Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20231121203129.67999-1-philmd@linaro.org>

Merge tag 'migration-20240104-pull-request' of https://gitlab.com/peterx/qemu into staging

migration 1st pull for 9.0

- We lost Juan and Leo in the maintainers file
- Steven's suspend state fix
- Steven's fix for coverity on migrate_mode
- Avihai's migration cleanup series

# -----BEGIN PGP SIGNATURE-----
#
# iIgEABYKADAWIQS5GE3CDMRX2s990ak7X8zN86vXBgUCZZY0TxIccGV0ZXJ4QHJl
# ZGhhdC5jb20ACgkQO1/MzfOr1wbSxgEAoM5g3wkc22lpAlRpU+hJUqT9NVOVQSK+
# Fk7XJYTdSgABAKzykA6hAmU5Kj+yVI6jI874SVZbs2FWpFs4osvsKk4D
# =sfuM
# -----END PGP SIGNATURE-----
# gpg: Signature made Thu 04 Jan 2024 04:30:07 GMT
# gpg:                using EDDSA key B9184DC20CC457DACF7DD1A93B5FCCCDF3ABD706
# gpg:                issuer "peterx@redhat.com"
# gpg: Good signature from "Peter Xu <xzpeter@gmail.com>" [unknown]
# gpg:                 aka "Peter Xu <peterx@redhat.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: B918 4DC2 0CC4 57DA CF7D  D1A9 3B5F CCCD F3AB D706

* tag 'migration-20240104-pull-request' of https://gitlab.com/peterx/qemu: (26 commits)
  migration: fix coverity migrate_mode finding
  migration/multifd: Remove unnecessary usage of local Error
  migration: Remove unnecessary usage of local Error
  migration: Fix migration_channel_read_peek() error path
  migration/multifd: Remove error_setg() in migration_ioc_process_incoming()
  migration/multifd: Fix leaking of Error in TLS error flow
  migration/multifd: Simplify multifd_channel_connect() if else statement
  migration/multifd: Fix error message in multifd_recv_initial_packet()
  migration: Remove errp parameter in migration_fd_process_incoming()
  migration: Refactor migration_incoming_setup()
  migration: Remove nulling of hostname in migrate_init()
  migration: Remove migrate_max_downtime() declaration
  tests/qtest: postcopy migration with suspend
  tests/qtest: precopy migration with suspend
  tests/qtest: option to suspend during migration
  tests/qtest: migration events
  migration: preserve suspended for bg_migration
  migration: preserve suspended for snapshot
  migration: preserve suspended runstate
  migration: propagate suspended runstate
  ...

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

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

* configure: use a native non-cross compiler for linux-user
* meson: cleanups
* target/i386: miscellaneous cleanups and optimizations
* target/i386: implement CMPccXADD
* target/i386: the sgx_epc_get_section stub is reachable
* esp: check for NULL result from scsi_device_find()

# -----BEGIN PGP SIGNATURE-----
#
# iQFIBAABCAAyFiEE8TM4V0tmI4mGbHaCv/vSX3jHroMFAmWRImYUHHBib256aW5p
# QHJlZGhhdC5jb20ACgkQv/vSX3jHroNd7AgAgcyJGiMfUkXqhefplpm06RDXQIa8
# FuoJqPb21lO75DQKfaFRAc4xGLagjJROMJGHMm9HvMu2VlwvOydkQlfFRspENxQ/
# 5XzGdb/X0A7HA/mwUfnMB1AZx0Vs32VI5IBSc6acc9fmgeZ84XQEoM3KBQHUik7X
# mSkE4eltR9gJ+4IaGo4voZtK+YoVD8nEcuqmnKihSPWizev0FsZ49aNMtaYa9qC/
# Xs3kiQd/zPibHDHJu0ulFsNZgxtUcvlLHTCf8gO4dHWxCFLXGubMush83McpRtNB
# Qoh6cTLH+PBXfrxMR3zmTZMNvo8Euls3s07Y8TkNP4vdIIE/kMeMDW1wJw==
# =mq30
# -----END PGP SIGNATURE-----
# gpg: Signature made Sun 31 Dec 2023 08:12:22 GMT
# gpg:                using RSA key F13338574B662389866C7682BFFBD25F78C7AE83
# gpg:                issuer "pbonzini@redhat.com"
# gpg: Good signature from "Paolo Bonzini <bonzini@gnu.org>" [full]
# gpg:                 aka "Paolo Bonzini <pbonzini@redhat.com>" [full]
# Primary key fingerprint: 46F5 9FBD 57D6 12E7 BFD4  E2F7 7E15 100C CD36 69B1
#      Subkey fingerprint: F133 3857 4B66 2389 866C  7682 BFFB D25F 78C7 AE83

* tag 'for-upstream' of https://gitlab.com/bonzini/qemu: (46 commits)
  meson.build: report graphics backends separately
  configure, meson: rename targetos to host_os
  meson: rename config_all
  meson: remove CONFIG_ALL
  meson: remove config_targetos
  meson: remove CONFIG_POSIX and CONFIG_WIN32 from config_targetos
  meson: remove OS definitions from config_targetos
  meson: always probe u2f and canokey if the option is enabled
  meson: move subdirs to "Collect sources" section
  meson: move config-host.h definitions together
  meson: move CFI detection code with other compiler flags
  meson: keep subprojects together
  meson: move accelerator dependency checks together
  meson: move option validation together
  meson: move program checks together
  meson: add more sections to main meson.build
  configure: unify again the case arms in probe_target_compiler
  configure: remove unnecessary subshell
  Makefile: clean qemu-iotests output
  meson: use version_compare() to compare version
  ...

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

Merge tag 'pull-20231230' of https://gitlab.com/rth7680/qemu into staging

Mark VMStateField and VMStateDescription arrays const.

# -----BEGIN PGP SIGNATURE-----
#
# iQFRBAABCgA7FiEEekgeeIaLTbaoWgXAZN846K9+IV8FAmWPOFsdHHJpY2hhcmQu
# aGVuZGVyc29uQGxpbmFyby5vcmcACgkQZN846K9+IV8cCQgAnQjy3Ic1i225AElh
# 0Ph3Aiw6WT9pECLoKmyroxHbTGuaEJoIXeaOhMAnowCTBLoKRR3/Ooq0DGOW+l/Z
# f5PwWSkjkb+OcS+dj/kgQBu58/Gk5G8ogksqKQvci8k2okhjHmITSQDu0dtwzDZr
# jVGh3gmGoat73jQyD/nAwgWFawlLkklOMR/yvnFX7EJIBepRVbkMPayoKlB+6W07
# 1kqhSwoI0vQCjhJ3Q7Q0GC4rrHK3KUq7o/25yvICf4EgPKfsaym1wAjDhdKToixl
# 9T+ALZG8SiZZkBlb6l3QZ7pqtqavxYtPdZ2Gx/nMu0RRu4G33d5AVGHRrXj9qttW
# 5mL7ZQ==
# =uQ4C
# -----END PGP SIGNATURE-----
# gpg: Signature made Fri 29 Dec 2023 21:21:31 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-20231230' of https://gitlab.com/rth7680/qemu: (71 commits)
  docs: Constify VMstate in examples
  tests/unit/test-vmstate: Constify VMState
  util/fifo8: Constify VMState
  replay: Constify VMState
  system: Constify VMState
  migration: Constify VMState
  cpu-target: Constify VMState
  backends: Constify VMState
  audio: Constify VMState
  hw/misc/macio: Constify VMState
  hw/watchdog: Constify VMState
  hw/virtio: Constify VMState
  hw/vfio: Constify VMState
  hw/usb: Constify VMState
  hw/tpm: Constify VMState
  hw/timer: Constify VMState
  hw/ssi: Constify VMState
  hw/sparc: Constify VMState
  hw/sensor: Constify VMState
  hw/sd: Constify VMState
  ...

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

migration: fix coverity migrate_mode finding

Coverity diagnoses a possible out-of-range array index here ...

    static GSList *migration_blockers[MIG_MODE__MAX];

    fill_source_migration_info() {
        GSList *cur_blocker = migration_blockers[migrate_mode()];

... because it does not know that MIG_MODE__MAX will never be returned as
a migration mode.  To fix, assert so in migrate_mode().

Fixes: fa3673e497a1 ("migration: per-mode blockers")
Reported-by: Peter Maydell <peter.maydell@linaro.org>
Suggested-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/1699907025-215450-1-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Remove unnecessary usage of local Error

According to Error API, usage of ERRP_GUARD() or a local Error instead
of errp is needed if errp is passed to void functions, where it is later
dereferenced to see if an error occurred.

There are several places in multifd.c that use local Error although it
is not needed. Change these places to use errp directly.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Link: https://lore.kernel.org/r/20231231093016.14204-12-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Remove unnecessary usage of local Error

According to Error API, usage of ERRP_GUARD() or a local Error instead
of errp is needed if errp is passed to void functions, where it is later
dereferenced to see if an error occurred.

There are several places in migration.c that use local Error although it
is not needed. Change these places to use errp directly.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-11-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Fix migration_channel_read_peek() error path

migration_channel_read_peek() calls qio_channel_readv_full() and handles
both cases of return value == 0 and return value < 0 the same way, by
calling error_setg() with errp. However, if return value < 0, errp is
already set, so calling error_setg() with errp will lead to an assert.

Fix it by handling these cases separately, calling error_setg() with
errp only in return value == 0 case.

Fixes: 6720c2b32725 ("migration: check magic value for deciding the mapping of channels")
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Link: https://lore.kernel.org/r/20231231093016.14204-10-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Remove error_setg() in migration_ioc_process_incoming()

If multifd_load_setup() fails in migration_ioc_process_incoming(),
error_setg() is called with errp. This will lead to an assert because in
that case errp already contains an error.

Fix it by removing the redundant error_setg().

Fixes: 6720c2b32725 ("migration: check magic value for deciding the mapping of channels")
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-9-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Fix leaking of Error in TLS error flow

If there is an error in multifd TLS handshake task,
multifd_tls_outgoing_handshake() retrieves the error with
qio_task_propagate_error() but never frees it.

Fix it by freeing the obtained Error.

In addition, the error is not reported at all, so report it with
migrate_set_error().

Fixes: 29647140157a ("migration/tls: add support for multifd tls-handshake")
Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-8-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Simplify multifd_channel_connect() if else statement

The else branch in multifd_channel_connect() is redundant because when
the if branch is taken the function returns.

Simplify the code by removing the else branch.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Link: https://lore.kernel.org/r/20231231093016.14204-7-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration/multifd: Fix error message in multifd_recv_initial_packet()

In multifd_recv_initial_packet(), if MultiFDInit_t->id is greater than
the configured number of multifd channels, an irrelevant error message
about multifd version is printed.

Change the error message to a relevant one about the channel id.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-6-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Remove errp parameter in migration_fd_process_incoming()

Errp parameter in migration_fd_process_incoming() is unused.
Remove it.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-5-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Refactor migration_incoming_setup()

Commit 6720c2b32725 ("migration: check magic value for deciding the
mapping of channels") extracted the only code that could fail in
migration_incoming_setup().

Now migration_incoming_setup() can't fail, so refactor it to return void
and remove errp parameter.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Link: https://lore.kernel.org/r/20231231093016.14204-4-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Remove nulling of hostname in migrate_init()

MigrationState->hostname is set to NULL in migrate_init(). This is
redundant because it is already freed and set to NULL in
migrade_fd_cleanup(). Remove it.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-3-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: Remove migrate_max_downtime() declaration

migrate_max_downtime() has been removed long ago, but its declaration
was mistakenly left. Remove it.

Signed-off-by: Avihai Horon <avihaih@nvidia.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/20231231093016.14204-2-avihaih@nvidia.com
Signed-off-by: Peter Xu <peterx@redhat.com>

tests/qtest: postcopy migration with suspend

Add a test case to verify that the suspended state is handled correctly by
live migration postcopy.  The test suspends the src, migrates, then wakes
the dest.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-13-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

tests/qtest: precopy migration with suspend

Add a test case to verify that the suspended state is handled correctly
during live migration precopy.  The test suspends the src, migrates, then
wakes the dest.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-12-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

tests/qtest: option to suspend during migration

Add an option to suspend the src in a-b-bootblock.S, which puts the guest
in S3 state after one round of writing to memory.  The option is enabled by
poking a 1 into the suspend_me word in the boot block prior to starting the
src vm.  Generate symbol offsets in a-b-bootblock.h so that the suspend_me
offset is known.  Generate the bootblock for each test, because suspend_me
may differ for each.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Acked-by: Peter Xu <peterx@redhat.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Link: https://lore.kernel.org/r/1704312341-66640-11-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

tests/qtest: migration events

Define a state object to capture events seen by migration tests, to allow
more events to be captured in a subsequent patch, and simplify event
checking in wait_for_migration_pass.  No functional change.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: "Daniel P. Berrangé" <berrange@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-10-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: preserve suspended for bg_migration

Do not wake a suspended guest during bg_migration, and restore the prior
state at finish rather than unconditionally running.  Allow the additional
state transitions that occur.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Fabiano Rosas <farosas@suse.de>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-9-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: preserve suspended for snapshot

Restoring a snapshot can break a suspended guest.  Snapshots suffer from
the same suspended-state issues that affect live migration, plus they must
handle an additional problematic scenario, which is that a running vm must
remain running if it loads a suspended snapshot.

To save, the existing vm_stop call now completely stops the suspended
state.  Finish with vm_resume to leave the vm in the state it had prior
to the save, correctly restoring the suspended state.

To load, if the snapshot is not suspended, then vm_stop + vm_resume
correctly handles all states, and leaves the vm in the state it had prior
to the load.  However, if the snapshot is suspended, restoration is
trickier.  First, call vm_resume to restore the state to suspended so the
current state matches the saved state.  Then, if the pre-load state is
running, call wakeup to resume running.

Prior to these changes, the vm_stop to RUN_STATE_SAVE_VM and
RUN_STATE_RESTORE_VM did not change runstate if the current state was
suspended, but now it does, so allow these transitions.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-8-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: preserve suspended runstate

A guest that is migrated in the suspended state automaticaly wakes and
continues execution.  This is wrong; the guest should end migration in
the same state it started.  The root cause is that the outgoing migration
code automatically wakes the guest, then saves the RUNNING runstate in
global_state_store(), hence the incoming migration code thinks the guest is
running and continues the guest if autostart is true.

On the outgoing side, delete the call to qemu_system_wakeup_request().
Now that vm_stop completely stops a vm in the suspended state (from the
preceding patches), the existing call to vm_stop_force_state is sufficient
to correctly migrate all vmstate.

On the incoming side, call vm_start if the pre-migration state was running
or suspended.  For the latter, vm_start correctly restores the suspended
state, and a future system_wakeup monitor request will cause the vm to
resume running.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-7-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

migration: propagate suspended runstate

If the outgoing machine was previously suspended, propagate that to the
incoming side via global_state, so a subsequent vm_start restores the
suspended state.  To maintain backward and forward compatibility, reclaim
some space from the runstate member.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-6-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

cpus: vm_resume

Define the vm_resume helper, for use in subsequent patches.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-5-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

cpus: check running not RUN_STATE_RUNNING

When a vm transitions from running to suspended, runstate notifiers are
not called, so the notifiers still think the vm is running.  Hence, when
we call vm_start to restore the suspended state, we call vm_state_notify
with running=1.  However, some notifiers check for RUN_STATE_RUNNING.
They must check the running boolean instead.

No functional change.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-4-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

cpus: stop vm in suspended runstate

Currently, a vm in the suspended state is not completely stopped.  The VCPUs
have been paused, but the cpu clock still runs, and runstate notifiers for
the transition to stopped have not been called.  This causes problems for
live migration.  Stale cpu timers_state is saved to the migration stream,
causing time errors in the guest when it wakes from suspend, and state that
would have been modified by runstate notifiers is wrong.

Modify vm_stop to completely stop the vm if the current state is suspended,
transition to RUN_STATE_PAUSED, and remember that the machine was suspended.
Modify vm_start to restore the suspended state.

This affects all callers of vm_stop and vm_start, notably, the qapi stop and
cont commands:

  old behavior:
    RUN_STATE_SUSPENDED --> stop --> RUN_STATE_SUSPENDED

  new behavior:
    RUN_STATE_SUSPENDED --> stop --> RUN_STATE_PAUSED
    RUN_STATE_PAUSED    --> cont --> RUN_STATE_SUSPENDED

For example:

    (qemu) info status
    VM status: paused (suspended)

    (qemu) stop
    (qemu) info status
    VM status: paused

    (qemu) system_wakeup
    Error: Unable to wake up: guest is not in suspended state

    (qemu) cont
    (qemu) info status
    VM status: paused (suspended)

    (qemu) system_wakeup
    (qemu) info status
    VM status: running

Suggested-by: Peter Xu <peterx@redhat.com>
Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-3-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>

cpus: vm_was_suspended

Add a state variable to remember if a vm previously transitioned into a
suspended state.

Signed-off-by: Steve Sistare <steven.sistare@oracle.com>
Reviewed-by: Peter Xu <peterx@redhat.com>
Link: https://lore.kernel.org/r/1704312341-66640-2-git-send-email-steven.sistare@oracle.com
Signed-off-by: Peter Xu <peterx@redhat.com>