Skip to contentSkip to conversation

hw/i386/pc: Remove deprecated 2.6 and 2.7 PC machines

PATCH Superseded Philippe Mathieu-Daudé 13 participants 123 messages May 8, 2025 – Dec 2, 2025

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

v4 cover hw/i386/pc: Remove deprecated 2.6 and 2.7 PC machines

Since v3:
- Addressed Thomas and Zhao review comments
- Rename fw_cfg_init_mem_[no]dma() helpers
- Remove unused CPU properties
- Remove {multi,linux}boot.bin
- Added R-b tags

Since v2:
- Addressed Mark review comments and added his R-b tags

The versioned 'pc' and 'q35' machines up to 2.12 been marked as deprecated two releases ago, and are older than 6 years, so according to our support policy we can remove them.

This series only includes the 2.6 and 2.7 machines removal, as it is a big enough number of LoC removed. Rest will follow.

Based-on: <[email protected]>

Philippe Mathieu-Daudé (27):
hw/i386/pc: Remove deprecated pc-q35-2.6 and pc-i440fx-2.6 machines hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field hw/nvram/fw_cfg: Rename fw_cfg_init_mem() with '_nodma' suffix hw/mips/loongson3_virt: Prefer using fw_cfg_init_mem_nodma() hw/nvram/fw_cfg: Factor fw_cfg_init_mem_internal() out hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() ->

    fw_cfg_init_mem_dma()

hw/i386/x86: Remove X86MachineClass::fwcfg_dma_enabled field hw/i386/pc: Remove multiboot.bin
hw/nvram/fw_cfg: Remove fw_cfg_io_properties::dma_enabled hw/i386/pc: Remove linuxboot.bin
hw/i386/pc: Remove pc_compat_2_6[] array
target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field target/i386/cpu: Remove CPUX86State::fill_mtrr_mask field hw/intc/apic: Remove APICCommonState::legacy_instance_id field hw/core/machine: Remove hw_compat_2_6[] array hw/virtio/virtio-mmio: Remove
VirtIOMMIOProxy::format_transport_address field
hw/i386/pc: Remove deprecated pc-q35-2.7 and pc-i440fx-2.7 machines hw/i386/pc: Remove pc_compat_2_7[] array
target/i386/cpu: Remove CPUX86State::full_cpuid_auto_level field target/i386/cpu: Remove CPUX86State::enable_l3_cache field hw/audio/pcspk: Remove PCSpkState::migrate field hw/core/machine: Remove hw_compat_2_7[] array hw/i386/intel_iommu: Remove IntelIOMMUState::buggy_eim field hw/intc/ioapic: Remove IOAPICCommonState::version field hw/virtio/virtio-pci: Remove VirtIOPCIProxy::ignore_backend_features
field
hw/char/virtio-serial: Do not expose the 'emergency-write' property hw/virtio/virtio-pci: Remove VIRTIO_PCI_FLAG_PAGE_PER_VQ definition

 hw/intc/ioapic_internal.h           |   3 +-
 include/hw/acpi/cpu_hotplug.h       |   3 -
 include/hw/boards.h                 |   6 -
 include/hw/i386/apic_internal.h     |   1 -
 include/hw/i386/intel_iommu.h       |   1 -
 include/hw/i386/pc.h                |   9 --
 include/hw/i386/x86.h               |   2 -
 include/hw/nvram/fw_cfg.h           |   9 +-
 include/hw/virtio/virtio-mmio.h     |   1 -
 include/hw/virtio/virtio-pci.h      |   2 -
 include/hw/virtio/virtio-serial.h   |   2 -
 pc-bios/optionrom/optionrom.h       |   4 -
 target/i386/cpu.h                   |  15 --
 hw/acpi/cpu_hotplug.c               | 230 ---------------------------
 hw/arm/virt.c                       |   2 +-
 hw/audio/pcspk.c                    |  10 --
 hw/char/virtio-serial-bus.c         |   9 +-
 hw/core/machine.c                   |  17 --
 hw/display/virtio-vga.c             |  10 --
 hw/hppa/machine.c                   |   2 +-
 hw/i386/acpi-build.c                |   4 +-
 hw/i386/fw_cfg.c                    |   5 +-
 hw/i386/intel_iommu.c               |   5 +-
 hw/i386/microvm.c                   |   3 -
 hw/i386/multiboot.c                 |   7 +-
 hw/i386/pc.c                        |  22 +--
 hw/i386/pc_piix.c                   |  23 ---
 hw/i386/pc_q35.c                    |  24 ---
 hw/i386/x86-common.c                |   6 +-
 hw/i386/x86.c                       |   2 -
 hw/intc/apic_common.c               |   5 -
 hw/intc/ioapic.c                    |  18 +--
 hw/intc/ioapic_common.c             |   2 +-
 hw/mips/loongson3_virt.c            |   2 +-
 hw/nvram/fw_cfg.c                   |  48 +++---
 hw/riscv/virt.c                     |   4 +-
 hw/virtio/virtio-mmio.c             |  15 --
 hw/virtio/virtio-pci.c              |  12 +-
 target/i386/cpu.c                   | 152 ++++++++----------
 target/i386/kvm/kvm.c               |  10 +-
 tests/qtest/test-x86-cpuid-compat.c |  11 --
 pc-bios/meson.build                 |   2 -
 pc-bios/multiboot.bin               | Bin 1024 -> 0 bytes
 pc-bios/optionrom/Makefile          |   2 +-
 pc-bios/optionrom/linuxboot.S       | 195 -----------------------
 pc-bios/optionrom/multiboot.S       | 232 ---------------------------
 pc-bios/optionrom/multiboot_dma.S   | 234 +++++++++++++++++++++++++++-
 47 files changed, 349 insertions(+), 1034 deletions(-)
 delete mode 100644 pc-bios/multiboot.bin
 delete mode 100644 pc-bios/optionrom/linuxboot.S
 delete mode 100644 pc-bios/optionrom/multiboot.S
Signature · 1 line
2.47.1
#1 lore raw

PATCH v4 1/27 hw/i386/pc: Remove deprecated pc-q35-2.6 and pc-i440fx-2.6 machines

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed.

---
 hw/i386/pc_piix.c | 14 --------------
 hw/i386/pc_q35.c  | 14 --------------
 2 files changed, 28 deletions(-)
hw/i386/pc_piix.c +0 -14
index 7a62bb06500..98a118fd4a0 100644
@@ -764,20 +764,6 @@ static void pc_i440fx_machine_2_7_options(MachineClass *m)
 
 DEFINE_I440FX_MACHINE(2, 7);
 
-static void pc_i440fx_machine_2_6_options(MachineClass *m)
-{
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(m);
-    PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
-
-    pc_i440fx_machine_2_7_options(m);
-    pcmc->legacy_cpu_hotplug = true;
-    x86mc->fwcfg_dma_enabled = false;
-    compat_props_add(m->compat_props, hw_compat_2_6, hw_compat_2_6_len);
-    compat_props_add(m->compat_props, pc_compat_2_6, pc_compat_2_6_len);
-}
-
-DEFINE_I440FX_MACHINE(2, 6);
-
 #ifdef CONFIG_ISAPC
 static void isapc_machine_options(MachineClass *m)
 {
hw/i386/pc_q35.c +0 -14
index 33211b1876f..b7ffb5f1216 100644
@@ -658,17 +658,3 @@ static void pc_q35_machine_2_7_options(MachineClass *m)
 }
 
 DEFINE_Q35_MACHINE(2, 7);
-
-static void pc_q35_machine_2_6_options(MachineClass *m)
-{
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(m);
-    PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
-
-    pc_q35_machine_2_7_options(m);
-    pcmc->legacy_cpu_hotplug = true;
-    x86mc->fwcfg_dma_enabled = false;
-    compat_props_add(m->compat_props, hw_compat_2_6, hw_compat_2_6_len);
-    compat_props_add(m->compat_props, pc_compat_2_6, pc_compat_2_6_len);
-}
-
-DEFINE_Q35_MACHINE(2, 6);
Signature · 1 line
2.47.1
#2 lore raw

Re: [PATCH v4 01/27] hw/i386/pc: Remove deprecated pc-q35-2.6 and pc-i440fx-2.6 machines

Quoted 4 lines · On Thu, 8 May 2025 15:35:24 +0200 Philippe Mathieu-Daudé <[email protected]>…

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed.

if these machine types are the last users of compat arrays, it's better to remove array at the same time, aka squash those patches later in series into this one.
That leaves no illusion that compats could be used in the later patches.

Quoted 57 lines
---
 hw/i386/pc_piix.c | 14 --------------
 hw/i386/pc_q35.c  | 14 --------------
 2 files changed, 28 deletions(-)
hw/i386/pc_piix.c +0 -14
index 7a62bb06500..98a118fd4a0 100644
@@ -764,20 +764,6 @@ static void pc_i440fx_machine_2_7_options(MachineClass *m)
 
 DEFINE_I440FX_MACHINE(2, 7);
 
-static void pc_i440fx_machine_2_6_options(MachineClass *m)
-{
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(m);
-    PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
-
-    pc_i440fx_machine_2_7_options(m);
-    pcmc->legacy_cpu_hotplug = true;
-    x86mc->fwcfg_dma_enabled = false;
-    compat_props_add(m->compat_props, hw_compat_2_6, hw_compat_2_6_len);
-    compat_props_add(m->compat_props, pc_compat_2_6, pc_compat_2_6_len);
-}
-
-DEFINE_I440FX_MACHINE(2, 6);
-
 #ifdef CONFIG_ISAPC
 static void isapc_machine_options(MachineClass *m)
 {
hw/i386/pc_q35.c +0 -14
index 33211b1876f..b7ffb5f1216 100644
@@ -658,17 +658,3 @@ static void pc_q35_machine_2_7_options(MachineClass *m)
 }
 
 DEFINE_Q35_MACHINE(2, 7);
-
-static void pc_q35_machine_2_6_options(MachineClass *m)
-{
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(m);
-    PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
-
-    pc_q35_machine_2_7_options(m);
-    pcmc->legacy_cpu_hotplug = true;
-    x86mc->fwcfg_dma_enabled = false;
-    compat_props_add(m->compat_props, hw_compat_2_6, hw_compat_2_6_len);
-    compat_props_add(m->compat_props, pc_compat_2_6, pc_compat_2_6_len);
-}
-
-DEFINE_Q35_MACHINE(2, 6);
#3 lore raw

Re: [PATCH v4 01/27] hw/i386/pc: Remove deprecated pc-q35-2.6 and pc-i440fx-2.6 machines

Quoted 8 lines · Igor Mammedov
Quoted 4 lines · On Thu, 8 May 2025 15:35:24 +0200 Philippe Mathieu-Daudé <[email protected]>…

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed.

Quoted 4 lines

if these machine types are the last users of compat arrays, it's better to remove array at the same time, aka squash those patches later in series into this one.
That leaves no illusion that compats could be used in the later patches.

IMHO the generic hw_compat array should be treated separately since this is independent from x86. So in case someone ever needs to backport these patches to an older branch, they can decide more easily whether they want to apply the generic hw_compat part or only the x86-specific part of this series.

Thomas

#4 lore raw

Re: [PATCH v4 01/27] hw/i386/pc: Remove deprecated pc-q35-2.6 and pc-i440fx-2.6 machines

Quoted 9 lines · Thomas Huth
Quoted 8 lines · Igor Mammedov
Quoted 4 lines · On Thu,  8 May 2025 15:35:24 +0200 Philippe Mathieu-Daudé <[email protected]>…

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed.

Quoted 6 lines
Quoted 4 lines

if these machine types are the last users of compat arrays, it's better to remove array at the same time, aka squash those patches later in series into this one.
That leaves no illusion that compats could be used in the later patches.

IMHO the generic hw_compat array should be treated separately since this is independent from x86. So in case someone ever needs to backport these patches to an older branch, they can decide more easily whether they want to apply the generic hw_compat part or only the x86-specific part of this series.

Yes, it is clearer this way than squashed.

#5 lore raw

Re: [PATCH v4 01/27] hw/i386/pc: Remove deprecated pc-q35-2.6 and pc-i440fx-2.6 machines

Quoted 15 lines · On Mon, 2 Jun 2025 10:53:19 +0200 Philippe Mathieu-Daudé <[email protected]>…
Quoted 14 lines · Thomas Huth
Quoted 12 lines · Igor Mammedov
Quoted 4 lines · On Thu,  8 May 2025 15:35:24 +0200 Philippe Mathieu-Daudé <[email protected]>…

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed.

if these machine types are the last users of compat arrays, it's better to remove array at the same time, aka squash those patches later in series into this one.
That leaves no illusion that compats could be used in the later patches.

Quoted 7 lines
Quoted 5 lines

IMHO the generic hw_compat array should be treated separately since this is independent from x86. So in case someone ever needs to backport these patches to an older branch, they can decide more easily whether they want to apply the generic hw_compat part or only the x86-specific part of this series.

Yes, it is clearer this way than squashed.

ok, let's leave it as is.

Quoted 1 line
#6 lore raw

PATCH v4 2/27 hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

In reply to Philippe Mathieu-Daudé

The PCMachineClass::legacy_cpu_hotplug boolean was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it and simplify build_dsdt(), removing build_legacy_cpu_hotplug_aml() altogether.

Note, this field was added by commit 679dd1a957d ("pc: use new CPU hotplug interface since 2.7 machine type"):

Quoted 11 lines

For compatibility reasons PC/Q35 will start with legacy CPU hotplug interface by default but with new CPU hotplug AML code since 2.7 machine type. That way legacy firmware that doesn't use QEMU generated ACPI tables will be able to continue using legacy CPU hotplug interface.

While new machine type, with firmware supporting QEMU provided ACPI tables, will generate new CPU hotplug AML, which will switch to new CPU hotplug interface when guest OS executes its _INI method on ACPI tables loading.

---
 include/hw/acpi/cpu_hotplug.h |   3 -
 include/hw/i386/pc.h          |   3 -
 hw/acpi/cpu_hotplug.c         | 230 ----------------------------------
 hw/i386/acpi-build.c          |   4 +-
 4 files changed, 1 insertion(+), 239 deletions(-)
include/hw/acpi/cpu_hotplug.h +0 -3
index 3b932abbbbe..aeee630cf05 100644
@@ -34,7 +34,4 @@ void legacy_acpi_cpu_hotplug_init(MemoryRegion *parent, Object *owner,
 void acpi_switch_to_modern_cphp(AcpiCpuHotplug *gpe_cpu,
                                 CPUHotplugState *cpuhp_state,
                                 uint16_t io_port);
-
-void build_legacy_cpu_hotplug_aml(Aml *ctx, MachineState *machine,
-                                  uint16_t io_base);
 #endif
include/hw/i386/pc.h +0 -3
index 79b72c54dd3..a3de3e9560d 100644
@@ -110,9 +110,6 @@ struct PCMachineClass {
     bool enforce_amd_1tb_hole;
     bool isa_bios_alias;
 
-    /* generate legacy CPU hotplug AML */
-    bool legacy_cpu_hotplug;
-
     /* use PVH to load kernels that support this feature */
     bool pvh_enabled;
 
hw/acpi/cpu_hotplug.c +0 -230

233 lines not on this page · raw message

hw/i386/acpi-build.c +1 -3
index 3fffa4a3328..625889783ec 100644
@@ -1465,9 +1465,7 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
     }
     aml_append(dsdt, scope);
 
-    if (pcmc->legacy_cpu_hotplug) {
-        build_legacy_cpu_hotplug_aml(dsdt, machine, pm->cpu_hp_io_base);
-    } else {
+    {
         CPUHotplugFeatures opts = {
             .acpi_1_compatible = true, .has_legacy_cphp = true,
             .smi_path = pm->smi_on_cpuhp ? "\\_SB.PCI0.SMI0.SMIC" : NULL,
Signature · 1 line
2.47.1
#7 lore raw

Re: [PATCH v4 02/27] hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

Quoted 31 lines · On Thu, 8 May 2025 15:35:25 +0200 Philippe Mathieu-Daudé <[email protected]>…

The PCMachineClass::legacy_cpu_hotplug boolean was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it and simplify build_dsdt(), removing build_legacy_cpu_hotplug_aml() altogether.

Note, this field was added by commit 679dd1a957d ("pc: use new CPU hotplug interface since 2.7 machine type"):

Quoted 11 lines

For compatibility reasons PC/Q35 will start with legacy CPU hotplug interface by default but with new CPU hotplug AML code since 2.7 machine type. That way legacy firmware that doesn't use QEMU generated ACPI tables will be able to continue using legacy CPU hotplug interface.

While new machine type, with firmware supporting QEMU provided ACPI tables, will generate new CPU hotplug AML, which will switch to new CPU hotplug interface when guest OS executes its _INI method on ACPI tables loading.

---
 include/hw/acpi/cpu_hotplug.h |   3 -
 include/hw/i386/pc.h          |   3 -
 hw/acpi/cpu_hotplug.c         | 230 ----------------------------------
 hw/i386/acpi-build.c          |   4 +-
 4 files changed, 1 insertion(+), 239 deletions(-)

...

Quoted 5 lines
hw/i386/acpi-build.c +0 -0
index 3fffa4a3328..625889783ec 100644

hw/i386/acpi-build.c build_dsdt(GArray *table_data, BIOSLinker *linker, · line 1470

Quoted 9 lines
hw/i386/acpi-build.c +1 -3
@@ -1465,9 +1465,7 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
     }
     aml_append(dsdt, scope);
 
-    if (pcmc->legacy_cpu_hotplug) {
-        build_legacy_cpu_hotplug_aml(dsdt, machine, pm->cpu_hp_io_base);
-    } else {
+    {
         CPUHotplugFeatures opts = {
             .acpi_1_compatible = true, .has_legacy_cphp = true,
                                           ^^^^^

that still leaves legacy CPU hotplug hardware around, which we should remove at the same time as legacy AML.

i.e.:
drop _INI in hw/acpi/cpu.c
and at the same time replace legacy_acpi_cpu_hotplug_init() at call sites with cpu_hotplug_hw_init()
after that you can safely remove legacy_cpu_hotplug and no longer needed hw/acpi/cpu_hotplug.c && co

Quoted 1 line
hw/i386/acpi-build.c +0 -0
             .smi_path = pm->smi_on_cpuhp ? "\\_SB.PCI0.SMI0.SMIC" : NULL,
#8 lore raw

PATCH v5 1/1 hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

In reply to Philippe Mathieu-Daudé

For compatibility reasons PC/Q35 will start with legacy CPU hotplug interface by default but with new CPU hotplug AML code since 2.7 machine type. That way legacy firmware that doesn't use QEMU generated ACPI tables was able to continue using legacy CPU hotplug interface.

While later machine types, with firmware supporting QEMU provided ACPI tables, generate new CPU hotplug AML, which will switch to new CPU hotplug interface when guest OS executes its _INI method on ACPI tables loading.

Since 2.6 machine type is now gone, drop legacy CPU hotplug code (both ACPI and related hardware bits) and initialize 'modern' hotplug from the very beginning.

---
PS:
it does compile but I do not remember if I've actually tested hotplug path

---
 include/hw/acpi/cpu.h                       |   1 -
 include/hw/acpi/cpu_hotplug.h               |  37 ------
 include/hw/acpi/ich9.h                      |   3 -
 include/hw/acpi/piix4.h                     |   4 +-
 tests/qtest/bios-tables-test-allowed-diff.h |  42 +++++++
 hw/acpi/acpi-cpu-hotplug-stub.c             |  18 +--
 hw/acpi/cpu.c                               |  10 --
 hw/acpi/cpu_hotplug.c                       | 118 --------------------
 hw/acpi/generic_event_device.c              |   1 +
 hw/acpi/ich9.c                              |  61 ++--------
 hw/acpi/meson.build                         |   2 +-
 hw/acpi/piix4.c                             |  60 ++--------
 hw/i386/acpi-build.c                        |   4 +-
 hw/i386/pc.c                                |   3 +-
 hw/i386/pc_piix.c                           |   2 -
 hw/i386/pc_q35.c                            |   2 -
 hw/i386/x86-common.c                        |   1 -
 hw/loongarch/virt-acpi-build.c              |   1 -
 18 files changed, 65 insertions(+), 305 deletions(-)
 delete mode 100644 include/hw/acpi/cpu_hotplug.h
 delete mode 100644 hw/acpi/cpu_hotplug.c
include/hw/acpi/cpu.h +0 -1
index 32654dc274..2cb0ca4f3d 100644
@@ -54,7 +54,6 @@ void cpu_hotplug_hw_init(MemoryRegion *as, Object *owner,
 
 typedef struct CPUHotplugFeatures {
     bool acpi_1_compatible;
-    bool has_legacy_cphp;
     bool fw_unplugs_cpu;
     const char *smi_path;
 } CPUHotplugFeatures;
include/hw/acpi/cpu_hotplug.h deleted +0 -37
deleted file mode 100644
index aeee630cf0..0000000000
@@ -1,37 +0,0 @@
-/*
- * QEMU ACPI hotplug utilities
- *
- * Copyright (C) 2013 Red Hat Inc
- *
- * Authors:
- *   Igor Mammedov <[email protected]>
- *
- * This work is licensed under the terms of the GNU GPL, version 2 or later.
- * See the COPYING file in the top-level directory.
- */
-
-#ifndef HW_ACPI_CPU_HOTPLUG_H
-#define HW_ACPI_CPU_HOTPLUG_H
-
-#include "hw/acpi/acpi.h"
-#include "hw/acpi/pc-hotplug.h"
-#include "hw/acpi/aml-build.h"
-#include "hw/hotplug.h"
-#include "hw/acpi/cpu.h"
-
-typedef struct AcpiCpuHotplug {
-    Object *device;
-    MemoryRegion io;
-    uint8_t sts[ACPI_GPE_PROC_LEN];
-} AcpiCpuHotplug;
-
-void legacy_acpi_cpu_plug_cb(HotplugHandler *hotplug_dev,
-                             AcpiCpuHotplug *g, DeviceState *dev, Error **errp);
-
-void legacy_acpi_cpu_hotplug_init(MemoryRegion *parent, Object *owner,
-                                  AcpiCpuHotplug *gpe_cpu, uint16_t base);
-
-void acpi_switch_to_modern_cphp(AcpiCpuHotplug *gpe_cpu,
-                                CPUHotplugState *cpuhp_state,
-                                uint16_t io_port);
-#endif
include/hw/acpi/ich9.h +0 -3
index 245fe08dc2..5a1986edbd 100644
@@ -22,7 +22,6 @@
 #define HW_ACPI_ICH9_H
 
 #include "hw/acpi/acpi.h"
-#include "hw/acpi/cpu_hotplug.h"
 #include "hw/acpi/cpu.h"
 #include "hw/acpi/pcihp.h"
 #include "hw/acpi/memory_hotplug.h"
@@ -53,8 +52,6 @@ typedef struct ICH9LPCPMRegs {
     uint32_t pm_io_base;
     Notifier powerdown_notifier;
 
-    bool cpu_hotplug_legacy;
-    AcpiCpuHotplug gpe_cpu;
     CPUHotplugState cpuhp_state;
 
     bool keep_pci_slot_hpc;
include/hw/acpi/piix4.h +1 -3
index eb1c122d80..863382a814 100644
@@ -24,11 +24,11 @@
 
 #include "hw/pci/pci_device.h"
 #include "hw/acpi/acpi.h"
-#include "hw/acpi/cpu_hotplug.h"
 #include "hw/acpi/memory_hotplug.h"
 #include "hw/acpi/pcihp.h"
 #include "hw/i2c/pm_smbus.h"
 #include "hw/isa/apm.h"
+#include "hw/acpi/cpu.h"
 
 #define TYPE_PIIX4_PM "PIIX4_PM"
 OBJECT_DECLARE_SIMPLE_TYPE(PIIX4PMState, PIIX4_PM)
@@ -63,8 +63,6 @@ struct PIIX4PMState {
     uint8_t disable_s4;
     uint8_t s4_val;
 
-    bool cpu_hotplug_legacy;
-    AcpiCpuHotplug gpe_cpu;
     CPUHotplugState cpuhp_state;
 
     MemHotplugState acpi_memory_hotplug;
tests/qtest/bios-tables-test-allowed-diff.h +42 -0

43 lines not on this page · raw message

hw/acpi/acpi-cpu-hotplug-stub.c +1 -17
index 9872dd55e4..ad5d862032 100644
@@ -1,22 +1,11 @@
 #include "qemu/osdep.h"
-#include "hw/acpi/cpu_hotplug.h"
 #include "migration/vmstate.h"
+#include "hw/acpi/cpu.h"
 
 
 /* Following stubs are all related to ACPI cpu hotplug */
 const VMStateDescription vmstate_cpu_hotplug;
 
-void acpi_switch_to_modern_cphp(AcpiCpuHotplug *gpe_cpu,
-                                CPUHotplugState *cpuhp_state,
-                                uint16_t io_port)
-{
-}
-
-void legacy_acpi_cpu_hotplug_init(MemoryRegion *parent, Object *owner,
-                                  AcpiCpuHotplug *gpe_cpu, uint16_t base)
-{
-}
-
 void cpu_hotplug_hw_init(MemoryRegion *as, Object *owner,
                          CPUHotplugState *state, hwaddr base_addr)
 {
@@ -31,11 +20,6 @@ void acpi_cpu_plug_cb(HotplugHandler *hotplug_dev,
 {
 }
 
-void legacy_acpi_cpu_plug_cb(HotplugHandler *hotplug_dev,
-                             AcpiCpuHotplug *g, DeviceState *dev, Error **errp)
-{
-}
-
 void acpi_cpu_unplug_cb(CPUHotplugState *cpu_st,
                         DeviceState *dev, Error **errp)
 {
hw/acpi/cpu.c +0 -10
index 6f1ae79edb..d63ca83c1b 100644
@@ -408,16 +408,6 @@ void build_cpus_aml(Aml *table, MachineState *machine, CPUHotplugFeatures opts,
         aml_append(field, aml_reserved_field(4 * 8));
         aml_append(field, aml_named_field(CPU_DATA, 32));
         aml_append(cpu_ctrl_dev, field);
-
-        if (opts.has_legacy_cphp) {
-            method = aml_method("_INI", 0, AML_SERIALIZED);
-            /* switch off legacy CPU hotplug HW and use new one,
-             * on reboot system is in new mode and writing 0
-             * in CPU_SELECTOR selects BSP, which is NOP at
-             * the time _INI is called */
-            aml_append(method, aml_store(zero, aml_name(CPU_SELECTOR)));
-            aml_append(cpu_ctrl_dev, method);
-        }
     }
     aml_append(sb_scope, cpu_ctrl_dev);
 
hw/acpi/cpu_hotplug.c deleted +0 -118

118 lines not on this page · raw message

hw/acpi/generic_event_device.c +1 -0
index e7b773d84d..9d0962d602 100644
@@ -13,6 +13,7 @@
 #include "qapi/error.h"
 #include "hw/acpi/acpi.h"
 #include "hw/acpi/pcihp.h"
+#include "hw/acpi/cpu.h"
 #include "hw/acpi/generic_event_device.h"
 #include "hw/pci/pci.h"
 #include "hw/irq.h"
hw/acpi/ich9.c +7 -54

123 lines not on this page · raw message

hw/acpi/meson.build +1 -1
index 56b5d1ec96..66c978aae8 100644
@@ -6,7 +6,7 @@ acpi_ss.add(files(
   'core.c',
   'utils.c',
 ))
-acpi_ss.add(when: 'CONFIG_ACPI_CPU_HOTPLUG', if_true: files('cpu.c', 'cpu_hotplug.c'))
+acpi_ss.add(when: 'CONFIG_ACPI_CPU_HOTPLUG', if_true: files('cpu.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_CPU_HOTPLUG', if_false: files('acpi-cpu-hotplug-stub.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_MEMORY_HOTPLUG', if_true: files('memory_hotplug.c'))
 acpi_ss.add(when: 'CONFIG_ACPI_MEMORY_HOTPLUG', if_false: files('acpi-mem-hotplug-stub.c'))
hw/acpi/piix4.c +7 -53

118 lines not on this page · raw message

hw/i386/acpi-build.c +3 -1
index 98dd3740ee..b01828ba5d 100644
@@ -35,6 +35,7 @@
 #include "hw/acpi/acpi-defs.h"
 #include "hw/acpi/acpi.h"
 #include "hw/acpi/cpu.h"
+#include "hw/acpi/pc-hotplug.h"
 #include "hw/nvram/fw_cfg.h"
 #include "hw/acpi/bios-linker-loader.h"
 #include "hw/acpi/acpi_aml_interface.h"
@@ -78,6 +79,7 @@
 #include "hw/acpi/hmat.h"
 #include "hw/acpi/viot.h"
 
+
 #include CONFIG_DEVICES
 
 /* These are used to size the ACPI tables for -M pc-i440fx-1.7 and
@@ -962,7 +964,7 @@ build_dsdt(GArray *table_data, BIOSLinker *linker,
 
     {
         CPUHotplugFeatures opts = {
-            .acpi_1_compatible = true, .has_legacy_cphp = true,
+            .acpi_1_compatible = true,
             .smi_path = pm->smi_on_cpuhp ? "\\_SB.PCI0.SMI0.SMIC" : NULL,
             .fw_unplugs_cpu = pm->smi_on_cpu_unplug,
         };
hw/i386/pc.c +2 -1
index 34b00663f2..9694c87820 100644
@@ -48,7 +48,8 @@
 #include "hw/xen/xen.h"
 #include "qobject/qlist.h"
 #include "qemu/error-report.h"
-#include "hw/acpi/cpu_hotplug.h"
+#include "hw/acpi/acpi.h"
+#include "hw/acpi/pc-hotplug.h"
 #include "acpi-build.h"
 #include "hw/mem/nvdimm.h"
 #include "hw/cxl/cxl_host.h"
hw/i386/pc_piix.c +0 -2
index 7b3611e973..ee2b01f15b 100644
@@ -736,10 +736,8 @@ DEFINE_I440FX_MACHINE(2, 7);
 static void pc_i440fx_machine_2_6_options(MachineClass *m)
 {
     X86MachineClass *x86mc = X86_MACHINE_CLASS(m);
-    PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
 
     pc_i440fx_machine_2_7_options(m);
-    pcmc->legacy_cpu_hotplug = true;
     x86mc->fwcfg_dma_enabled = false;
     compat_props_add(m->compat_props, hw_compat_2_6, hw_compat_2_6_len);
     compat_props_add(m->compat_props, pc_compat_2_6, pc_compat_2_6_len);
hw/i386/pc_q35.c +0 -2
index 6015e639d7..1a4787d310 100644
@@ -685,10 +685,8 @@ DEFINE_Q35_MACHINE(2, 7);
 static void pc_q35_machine_2_6_options(MachineClass *m)
 {
     X86MachineClass *x86mc = X86_MACHINE_CLASS(m);
-    PCMachineClass *pcmc = PC_MACHINE_CLASS(m);
 
     pc_q35_machine_2_7_options(m);
-    pcmc->legacy_cpu_hotplug = true;
     x86mc->fwcfg_dma_enabled = false;
     compat_props_add(m->compat_props, hw_compat_2_6, hw_compat_2_6_len);
     compat_props_add(m->compat_props, pc_compat_2_6, pc_compat_2_6_len);
hw/i386/x86-common.c +0 -1
index 7512be64d6..603ab3f016 100644
@@ -36,7 +36,6 @@
 #include "hw/rtc/mc146818rtc.h"
 #include "target/i386/sev.h"
 
-#include "hw/acpi/cpu_hotplug.h"
 #include "hw/irq.h"
 #include "hw/loader.h"
 #include "multiboot.h"
hw/loongarch/virt-acpi-build.c +0 -1
index 8c2228a772..c835403a93 100644
@@ -370,7 +370,6 @@ build_la_ged_aml(Aml *dsdt, MachineState *machine)
 
     if (event & ACPI_GED_CPU_HOTPLUG_EVT) {
         opts.acpi_1_compatible = false;
-        opts.has_legacy_cphp = false;
         opts.fw_unplugs_cpu = false;
         opts.smi_path = NULL;
 
Signature · 1 line
2.47.3
#9 lore raw

Re: [PATCH v5] hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

tests/qtest/bios-tables-test-allowed-diff.h

Quoted 152 lines · Igor Mammedov

For compatibility reasons PC/Q35 will start with legacy CPU hotplug interface by default but with new CPU hotplug AML code since 2.7 machine type. That way legacy firmware that doesn't use QEMU generated ACPI tables was able to continue using legacy CPU hotplug interface.

While later machine types, with firmware supporting QEMU provided ACPI tables, generate new CPU hotplug AML, which will switch to new CPU hotplug interface when guest OS executes its _INI method on ACPI tables loading.

Since 2.6 machine type is now gone, drop legacy CPU hotplug code (both ACPI and related hardware bits) and initialize 'modern' hotplug from the very beginning.

---
PS:
it does compile but I do not remember if I've actually tested hotplug path

---

include/hw/acpi/cpu.h                       |   1 -
include/hw/acpi/cpu_hotplug.h               |  37 ------
include/hw/acpi/ich9.h                      |   3 -
include/hw/acpi/piix4.h                     |   4 +-
tests/qtest/bios-tables-test-allowed-diff.h |  42 +++++++
hw/acpi/acpi-cpu-hotplug-stub.c             |  18 +--
hw/acpi/cpu.c                               |  10 --
hw/acpi/cpu_hotplug.c                       | 118 --------------------
hw/acpi/generic_event_device.c              |   1 +
hw/acpi/ich9.c                              |  61 ++--------
hw/acpi/meson.build                         |   2 +-
hw/acpi/piix4.c                             |  60 ++--------
hw/i386/acpi-build.c                        |   4 +-
hw/i386/pc.c                                |   3 +-
hw/i386/pc_piix.c                           |   2 -
hw/i386/pc_q35.c                            |   2 -
hw/i386/x86-common.c                        |   1 -
hw/loongarch/virt-acpi-build.c              |   1 -
18 files changed, 65 insertions(+), 305 deletions(-)

delete mode 100644 include/hw/acpi/cpu_hotplug.h delete mode 100644 hw/acpi/cpu_hotplug.c

include/hw/acpi/cpu.h +0 -1
index 32654dc274..2cb0ca4f3d 100644
@@ -54,7 +54,6 @@ void cpu_hotplug_hw_init(MemoryRegion *as, Object *owner,
 
 typedef struct CPUHotplugFeatures {
    bool acpi_1_compatible;
-    bool has_legacy_cphp;
    bool fw_unplugs_cpu;
    const char *smi_path;
 } CPUHotplugFeatures;
include/hw/acpi/cpu_hotplug.h deleted +0 -37
deleted file mode 100644
index aeee630cf0..0000000000
@@ -1,37 +0,0 @@
-/*
- * QEMU ACPI hotplug utilities
- *
- * Copyright (C) 2013 Red Hat Inc
- *
- * Authors:
- *   Igor Mammedov <[email protected]>
- *
- * This work is licensed under the terms of the GNU GPL, version 2 or later.
- * See the COPYING file in the top-level directory.
- */
-
-#ifndef HW_ACPI_CPU_HOTPLUG_H
-#define HW_ACPI_CPU_HOTPLUG_H
-
-#include "hw/acpi/acpi.h"
-#include "hw/acpi/pc-hotplug.h"
-#include "hw/acpi/aml-build.h"
-#include "hw/hotplug.h"
-#include "hw/acpi/cpu.h"
-
-typedef struct AcpiCpuHotplug {
-    Object *device;
-    MemoryRegion io;
-    uint8_t sts[ACPI_GPE_PROC_LEN];
-} AcpiCpuHotplug;
-
-void legacy_acpi_cpu_plug_cb(HotplugHandler *hotplug_dev,
-                             AcpiCpuHotplug *g, DeviceState *dev, Error **errp);
-
-void legacy_acpi_cpu_hotplug_init(MemoryRegion *parent, Object *owner,
-                                  AcpiCpuHotplug *gpe_cpu, uint16_t base);
-
-void acpi_switch_to_modern_cphp(AcpiCpuHotplug *gpe_cpu,
-                                CPUHotplugState *cpuhp_state,
-                                uint16_t io_port);
-#endif
include/hw/acpi/ich9.h +0 -3
index 245fe08dc2..5a1986edbd 100644
@@ -22,7 +22,6 @@
 #define HW_ACPI_ICH9_H
 
 #include "hw/acpi/acpi.h"
-#include "hw/acpi/cpu_hotplug.h"
 #include "hw/acpi/cpu.h"
 #include "hw/acpi/pcihp.h"
 #include "hw/acpi/memory_hotplug.h"
@@ -53,8 +52,6 @@ typedef struct ICH9LPCPMRegs {
    uint32_t pm_io_base;
    Notifier powerdown_notifier;
 
-    bool cpu_hotplug_legacy;
-    AcpiCpuHotplug gpe_cpu;
    CPUHotplugState cpuhp_state;
 
    bool keep_pci_slot_hpc;
include/hw/acpi/piix4.h +1 -3
index eb1c122d80..863382a814 100644
@@ -24,11 +24,11 @@
 
 #include "hw/pci/pci_device.h"
 #include "hw/acpi/acpi.h"
-#include "hw/acpi/cpu_hotplug.h"
 #include "hw/acpi/memory_hotplug.h"
 #include "hw/acpi/pcihp.h"
 #include "hw/i2c/pm_smbus.h"
 #include "hw/isa/apm.h"
+#include "hw/acpi/cpu.h"
 
 #define TYPE_PIIX4_PM "PIIX4_PM"
 OBJECT_DECLARE_SIMPLE_TYPE(PIIX4PMState, PIIX4_PM)
@@ -63,8 +63,6 @@ struct PIIX4PMState {
    uint8_t disable_s4;
    uint8_t s4_val;
 
-    bool cpu_hotplug_legacy;
-    AcpiCpuHotplug gpe_cpu;
    CPUHotplugState cpuhp_state;
 
    MemHotplugState acpi_memory_hotplug;
tests/qtest/bios-tables-test-allowed-diff.h +0 -0
index dfb8523c8b..7191854857 100644
@@ -1 +1,43 @@

Does this belong here?

Quoted 33 lines
tests/qtest/bios-tables-test-allowed-diff.h +32 -0
 /* List of comma-separated changed AML files to ignore */
+"tests/data/acpi/x86/pc/DSDT",
+"tests/data/acpi/x86/q35/DSDT",
+"tests/data/acpi/x86/pc/DSDT.bridge",
+"tests/data/acpi/x86/pc/DSDT.ipmikcs",
+"tests/data/acpi/x86/pc/DSDT.cphp",
+"tests/data/acpi/x86/pc/DSDT.numamem",
+"tests/data/acpi/x86/pc/DSDT.nohpet",
+"tests/data/acpi/x86/pc/DSDT.memhp",
+"tests/data/acpi/x86/pc/DSDT.dimmpxm",
+"tests/data/acpi/x86/pc/DSDT.acpihmat",
+"tests/data/acpi/x86/pc/DSDT.acpierst",
+"tests/data/acpi/x86/pc/DSDT.roothp",
+"tests/data/acpi/x86/pc/DSDT.hpbridge",
+"tests/data/acpi/x86/pc/DSDT.hpbrroot",
+"tests/data/acpi/x86/q35/DSDT.tis.tpm2",
+"tests/data/acpi/x86/q35/DSDT.tis.tpm12",
+"tests/data/acpi/x86/q35/DSDT.bridge",
+"tests/data/acpi/x86/q35/DSDT.noacpihp",
+"tests/data/acpi/x86/q35/DSDT.multi-bridge",
+"tests/data/acpi/x86/q35/DSDT.ipmibt",
+"tests/data/acpi/x86/q35/DSDT.cphp",
+"tests/data/acpi/x86/q35/DSDT.numamem",
+"tests/data/acpi/x86/q35/DSDT.nohpet",
+"tests/data/acpi/x86/q35/DSDT.acpihmat-noinitiator",
+"tests/data/acpi/x86/q35/DSDT.acpihmat-generic-x",
+"tests/data/acpi/x86/q35/DSDT.memhp",
+"tests/data/acpi/x86/q35/DSDT.dimmpxm",
+"tests/data/acpi/x86/q35/DSDT.acpihmat",
+"tests/data/acpi/x86/q35/DSDT.mmio64",
+"tests/data/acpi/x86/q35/DSDT.acpierst",
+"tests/data/acpi/x86/q35/DSDT.applesmc",
+"tests/data/acpi/x86/q35/DSDT.pvpanic-isa",

tests/qtest/bios-tables-test-allowed-diff.h line 43

Quoted 10 lines
tests/qtest/bios-tables-test-allowed-diff.h +10 -0
+"tests/data/acpi/x86/q35/DSDT.ivrs",
+"tests/data/acpi/x86/q35/DSDT.type4-count",
+"tests/data/acpi/x86/q35/DSDT.core-count",
+"tests/data/acpi/x86/q35/DSDT.core-count2",
+"tests/data/acpi/x86/q35/DSDT.thread-count",
+"tests/data/acpi/x86/q35/DSDT.thread-count2",
+"tests/data/acpi/x86/q35/DSDT.viot",
+"tests/data/acpi/x86/q35/DSDT.cxl",
+"tests/data/acpi/x86/q35/DSDT.ipmismbus",
+"tests/data/acpi/x86/q35/DSDT.xapic",

Regards,
BALATON Zoltan

#10 lore raw

Re: [PATCH v5] hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

On Sat, 1 Nov 2025 00:27:36 +0100 (CET)
BALATON Zoltan <[email protected]> wrote:

[...]

Quoted 7 lines

tests/qtest/bios-tables-test-allowed-diff.h

Quoted 5 lines
tests/qtest/bios-tables-test-allowed-diff.h +0 -0
index dfb8523c8b..7191854857 100644
@@ -1 +1,43 @@  

Does this belong here?

thanks for spotting it, it should be a separate preceding patch,

and then expected blobs update patch after this patch,
that would cleanup bios-tables-test-allowed-diff.h along with new blobs. /i.e. as per process described in bios-tables-test.c /

Quoted 47 lines
Quoted 34 lines
tests/qtest/bios-tables-test-allowed-diff.h +32 -0
 /* List of comma-separated changed AML files to ignore */
+"tests/data/acpi/x86/pc/DSDT",
+"tests/data/acpi/x86/q35/DSDT",
+"tests/data/acpi/x86/pc/DSDT.bridge",
+"tests/data/acpi/x86/pc/DSDT.ipmikcs",
+"tests/data/acpi/x86/pc/DSDT.cphp",
+"tests/data/acpi/x86/pc/DSDT.numamem",
+"tests/data/acpi/x86/pc/DSDT.nohpet",
+"tests/data/acpi/x86/pc/DSDT.memhp",
+"tests/data/acpi/x86/pc/DSDT.dimmpxm",
+"tests/data/acpi/x86/pc/DSDT.acpihmat",
+"tests/data/acpi/x86/pc/DSDT.acpierst",
+"tests/data/acpi/x86/pc/DSDT.roothp",
+"tests/data/acpi/x86/pc/DSDT.hpbridge",
+"tests/data/acpi/x86/pc/DSDT.hpbrroot",
+"tests/data/acpi/x86/q35/DSDT.tis.tpm2",
+"tests/data/acpi/x86/q35/DSDT.tis.tpm12",
+"tests/data/acpi/x86/q35/DSDT.bridge",
+"tests/data/acpi/x86/q35/DSDT.noacpihp",
+"tests/data/acpi/x86/q35/DSDT.multi-bridge",
+"tests/data/acpi/x86/q35/DSDT.ipmibt",
+"tests/data/acpi/x86/q35/DSDT.cphp",
+"tests/data/acpi/x86/q35/DSDT.numamem",
+"tests/data/acpi/x86/q35/DSDT.nohpet",
+"tests/data/acpi/x86/q35/DSDT.acpihmat-noinitiator",
+"tests/data/acpi/x86/q35/DSDT.acpihmat-generic-x",
+"tests/data/acpi/x86/q35/DSDT.memhp",
+"tests/data/acpi/x86/q35/DSDT.dimmpxm",
+"tests/data/acpi/x86/q35/DSDT.acpihmat",
+"tests/data/acpi/x86/q35/DSDT.mmio64",
+"tests/data/acpi/x86/q35/DSDT.acpierst",
+"tests/data/acpi/x86/q35/DSDT.applesmc",
+"tests/data/acpi/x86/q35/DSDT.pvpanic-isa",

tests/qtest/bios-tables-test-allowed-diff.h line 43

Quoted 10 lines
tests/qtest/bios-tables-test-allowed-diff.h +10 -0
+"tests/data/acpi/x86/q35/DSDT.ivrs",
+"tests/data/acpi/x86/q35/DSDT.type4-count",
+"tests/data/acpi/x86/q35/DSDT.core-count",
+"tests/data/acpi/x86/q35/DSDT.core-count2",
+"tests/data/acpi/x86/q35/DSDT.thread-count",
+"tests/data/acpi/x86/q35/DSDT.thread-count2",
+"tests/data/acpi/x86/q35/DSDT.viot",
+"tests/data/acpi/x86/q35/DSDT.cxl",
+"tests/data/acpi/x86/q35/DSDT.ipmismbus",
+"tests/data/acpi/x86/q35/DSDT.xapic",  

Regards,
BALATON Zoltan

#11 lore raw

Re: [PATCH v5] hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

Quoted 10 lines · Igor Mammedov

Date: Mon, 3 Nov 2025 10:42:31 +0100
From: Igor Mammedov <[email protected]>
Subject: Re: [PATCH v5] hw/i386/pc: Remove
PCMachineClass::legacy_cpu_hotplug field
X-Mailer: Claws Mail 4.3.1 (GTK 3.24.49; x86_64-redhat-linux-gnu)

On Sat, 1 Nov 2025 00:27:36 +0100 (CET)
BALATON Zoltan <[email protected]> wrote:

[...]

Quoted 8 lines
Quoted 7 lines

tests/qtest/bios-tables-test-allowed-diff.h

Quoted 5 lines
tests/qtest/bios-tables-test-allowed-diff.h +0 -0
index dfb8523c8b..7191854857 100644
@@ -1 +1,43 @@  

Does this belong here?

thanks for spotting it, it should be a separate preceding patch,

Thanks! I think I see :-) - I should follow the gudience in tests/qtest/bios-tables-test.c, to split this patch.

Quoted 3 lines

and then expected blobs update patch after this patch,
that would cleanup bios-tables-test-allowed-diff.h along with new blobs. /i.e. as per process described in bios-tables-test.c /

Regards,
Zhao

#12 lore raw

Re: [PATCH v5] hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

In reply to Igor Mammedov

Hi Igor,

I find after removing legacy cpu hotplug totally, it seems necessary to update docs/specs/acpi_cpu_hotplug.rst as well, since then QEMU doesn't have legacy cpu hotplug. Do you agree?

I also attached my update for that doc. If this change is fine for you, I will squash that doc change into another patch that removes cpu_hotplug.[h|c] - I split your this patch into several parts (following you another previous suggestion [*] to make review easier). :)

[*]: https://lore.kernel.org/qemu-devel/[email protected]/

Thanks,
Zhao
---
From ff3da76cf7de097c42bdaf64850dc4df65b4bf4a Mon Sep 17 00:00:00 2001
From: Zhao Liu <[email protected]>
Date: Wed, 26 Nov 2025 23:06:09 +0800
Subject: [PATCH] docs/specs/acpi_cpu_hotplug.rst: remove legacy cpu hotplug
descriptions

Legacy cpu hotplug has been removed totally and machines start with modern cpu hotplug interface directly.

Therefore, update the documentation to describe current QEMU cpu hotplug logic.

---
 docs/specs/acpi_cpu_hotplug.rst | 28 +++-------------------------
 1 file changed, 3 insertions(+), 25 deletions(-)
docs/specs/acpi_cpu_hotplug.rst +3 -25
index 351057c96761..f49678100044 100644
@@ -8,22 +8,6 @@ ACPI BIOS GPE.2 handler is dedicated for notifying OS about CPU hot-add
 and hot-remove events.
 
 
-Legacy ACPI CPU hotplug interface registers
--------------------------------------------
-
-CPU present bitmap for:
-
-- ICH9-LPC (IO port 0x0cd8-0xcf7, 1-byte access)
-- PIIX-PM  (IO port 0xaf00-0xaf1f, 1-byte access)
-- One bit per CPU. Bit position reflects corresponding CPU APIC ID. Read-only.
-- The first DWORD in bitmap is used in write mode to switch from legacy
-  to modern CPU hotplug interface, write 0 into it to do switch.
-
-QEMU sets corresponding CPU bit on hot-add event and issues SCI
-with GPE.2 event set. CPU present map is read by ACPI BIOS GPE.2 handler
-to notify OS about CPU hot-add events. CPU hot-remove isn't supported.
-
-
 Modern ACPI CPU hotplug interface registers
 -------------------------------------------
 
@@ -189,20 +173,14 @@ Typical usecases
 (x86) Detecting and enabling modern CPU hotplug interface
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-QEMU starts with legacy CPU hotplug interface enabled. Detecting and
-switching to modern interface is based on the 2 legacy CPU hotplug features:
-
-#. Writes into CPU bitmap are ignored.
-#. CPU bitmap always has bit #0 set, corresponding to boot CPU.
-
-Use following steps to detect and enable modern CPU hotplug interface:
+QEMU starts with modern CPU hotplug interface enabled. Use following steps to
+detect modern CPU hotplug interface:
 
-#. Store 0x0 to the 'CPU selector' register, attempting to switch to modern mode
 #. Store 0x0 to the 'CPU selector' register, to ensure valid selector value
 #. Store 0x0 to the 'Command field' register
 #. Read the 'Command data 2' register.
    If read value is 0x0, the modern interface is enabled.
-   Otherwise legacy or no CPU hotplug interface available
+   Otherwise no CPU hotplug interface available
 
 Get a cpu with pending event
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^

--
2.34.1

#13 lore raw

Re: [PATCH v5] hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field

Quoted 11 lines · On Wed, 26 Nov 2025 23:37:32 +0800 Zhao Liu <[email protected]> wrote:

Hi Igor,

I find after removing legacy cpu hotplug totally, it seems necessary to update docs/specs/acpi_cpu_hotplug.rst as well, since then QEMU doesn't have legacy cpu hotplug. Do you agree?

I also attached my update for that doc. If this change is fine for you, I will squash that doc change into another patch that removes cpu_hotplug.[h|c] - I split your this patch into several parts (following you another previous suggestion [*] to make review easier). :)

sure thing, whatever makes it easier to read/review

Quoted 12 lines

[*]: https://lore.kernel.org/qemu-devel/[email protected]/

Thanks,
Zhao
---
From ff3da76cf7de097c42bdaf64850dc4df65b4bf4a Mon Sep 17 00:00:00 2001
From: Zhao Liu <[email protected]>
Date: Wed, 26 Nov 2025 23:06:09 +0800
Subject: [PATCH] docs/specs/acpi_cpu_hotplug.rst: remove legacy cpu hotplug
descriptions

Quoted 7 lines

Legacy cpu hotplug has been removed totally and machines start with modern cpu hotplug interface directly.

Therefore, update the documentation to describe current QEMU cpu hotplug logic.

LGTM

Quoted 58 lines
---
 docs/specs/acpi_cpu_hotplug.rst | 28 +++-------------------------
 1 file changed, 3 insertions(+), 25 deletions(-)
docs/specs/acpi_cpu_hotplug.rst +3 -26
index 351057c96761..f49678100044 100644
@@ -8,22 +8,6 @@ ACPI BIOS GPE.2 handler is dedicated for notifying OS about CPU hot-add
 and hot-remove events.
 
 
-Legacy ACPI CPU hotplug interface registers
--------------------------------------------
-
-CPU present bitmap for:
-
-- ICH9-LPC (IO port 0x0cd8-0xcf7, 1-byte access)
-- PIIX-PM  (IO port 0xaf00-0xaf1f, 1-byte access)
-- One bit per CPU. Bit position reflects corresponding CPU APIC ID. Read-only.
-- The first DWORD in bitmap is used in write mode to switch from legacy
-  to modern CPU hotplug interface, write 0 into it to do switch.
-
-QEMU sets corresponding CPU bit on hot-add event and issues SCI
-with GPE.2 event set. CPU present map is read by ACPI BIOS GPE.2 handler
-to notify OS about CPU hot-add events. CPU hot-remove isn't supported.
-
-
 Modern ACPI CPU hotplug interface registers
 -------------------------------------------
 
@@ -189,20 +173,14 @@ Typical usecases
 (x86) Detecting and enabling modern CPU hotplug interface
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
 
-QEMU starts with legacy CPU hotplug interface enabled. Detecting and
-switching to modern interface is based on the 2 legacy CPU hotplug features:
-
-#. Writes into CPU bitmap are ignored.
-#. CPU bitmap always has bit #0 set, corresponding to boot CPU.
-
-Use following steps to detect and enable modern CPU hotplug interface:
+QEMU starts with modern CPU hotplug interface enabled. Use following steps to
+detect modern CPU hotplug interface:
 
-#. Store 0x0 to the 'CPU selector' register, attempting to switch to modern mode
 #. Store 0x0 to the 'CPU selector' register, to ensure valid selector value
 #. Store 0x0 to the 'Command field' register
 #. Read the 'Command data 2' register.
    If read value is 0x0, the modern interface is enabled.
-   Otherwise legacy or no CPU hotplug interface available
+   Otherwise no CPU hotplug interface available
 
 Get a cpu with pending event
 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
--

2.34.1

#14 lore raw

PATCH v4 3/27 hw/nvram/fw_cfg: Rename fw_cfg_init_mem() with '_nodma' suffix

In reply to Philippe Mathieu-Daudé

Rename fw_cfg_init_mem() as fw_cfg_init_mem_nodma() to distinct with the DMA version (currently named fw_cfg_init_mem_wide).

---
 include/hw/nvram/fw_cfg.h | 3 ++-
 hw/hppa/machine.c         | 2 +-
 hw/nvram/fw_cfg.c         | 7 +++----
 3 files changed, 6 insertions(+), 6 deletions(-)
include/hw/nvram/fw_cfg.h +2 -1
index d41b9328fd1..d5161a79436 100644
@@ -307,7 +307,8 @@ bool fw_cfg_add_file_from_generator(FWCfgState *s,
 
 FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
                                 AddressSpace *dma_as);
-FWCfgState *fw_cfg_init_mem(hwaddr ctl_addr, hwaddr data_addr);
+FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
+                                  unsigned data_width);
 FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
                                  hwaddr data_addr, uint32_t data_width,
                                  hwaddr dma_addr, AddressSpace *dma_as);
hw/hppa/machine.c +1 -1
index dacedc5409c..0d768cb90b0 100644
@@ -201,7 +201,7 @@ static FWCfgState *create_fw_cfg(MachineState *ms, PCIBus *pci_bus,
     int btlb_entries = HPPA_BTLB_ENTRIES(&cpu[0]->env);
     int len;
 
-    fw_cfg = fw_cfg_init_mem(addr, addr + 4);
+    fw_cfg = fw_cfg_init_mem_nodma(addr, addr + 4, 1);
     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, ms->smp.cpus);
     fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, HPPA_MAX_CPUS);
     fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, ms->ram_size);
hw/nvram/fw_cfg.c +3 -4
index 54cfa07d3f5..10f8f8db86f 100644
@@ -1087,11 +1087,10 @@ FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
     return s;
 }
 
-FWCfgState *fw_cfg_init_mem(hwaddr ctl_addr, hwaddr data_addr)
+FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
+                                  unsigned data_width)
 {
-    return fw_cfg_init_mem_wide(ctl_addr, data_addr,
-                                fw_cfg_data_mem_ops.valid.max_access_size,
-                                0, NULL);
+    return fw_cfg_init_mem_wide(ctl_addr, data_addr, data_width, 0, NULL);
 }
 
 
Signature · 1 line
2.47.1
Reviewed-by on v4 3/27
#15 lore raw

Re: [PATCH v4 03/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem() with '_nodma' suffix

Quoted 10 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:26 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 03/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem() with
'_nodma' suffix
X-Mailer: git-send-email 2.47.1

Rename fw_cfg_init_mem() as fw_cfg_init_mem_nodma() to distinct with the DMA version (currently named fw_cfg_init_mem_wide).

Quoted 7 lines
---
 include/hw/nvram/fw_cfg.h | 3 ++-
 hw/hppa/machine.c         | 2 +-
 hw/nvram/fw_cfg.c         | 7 +++----
 3 files changed, 6 insertions(+), 6 deletions(-)
Reviewed-by on v4 3/27
#16 lore raw

Re: [PATCH v4 03/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem() with '_nodma' suffix

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · On Thu, 8 May 2025 15:35:26 +0200 Philippe Mathieu-Daudé <[email protected]>…

Rename fw_cfg_init_mem() as fw_cfg_init_mem_nodma() to distinct with the DMA version (currently named fw_cfg_init_mem_wide).

Quoted 52 lines
---
 include/hw/nvram/fw_cfg.h | 3 ++-
 hw/hppa/machine.c         | 2 +-
 hw/nvram/fw_cfg.c         | 7 +++----
 3 files changed, 6 insertions(+), 6 deletions(-)
include/hw/nvram/fw_cfg.h +2 -1
index d41b9328fd1..d5161a79436 100644
@@ -307,7 +307,8 @@ bool fw_cfg_add_file_from_generator(FWCfgState *s,
 
 FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
                                 AddressSpace *dma_as);
-FWCfgState *fw_cfg_init_mem(hwaddr ctl_addr, hwaddr data_addr);
+FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
+                                  unsigned data_width);
 FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
                                  hwaddr data_addr, uint32_t data_width,
                                  hwaddr dma_addr, AddressSpace *dma_as);
hw/hppa/machine.c +1 -1
index dacedc5409c..0d768cb90b0 100644
@@ -201,7 +201,7 @@ static FWCfgState *create_fw_cfg(MachineState *ms, PCIBus *pci_bus,
     int btlb_entries = HPPA_BTLB_ENTRIES(&cpu[0]->env);
     int len;
 
-    fw_cfg = fw_cfg_init_mem(addr, addr + 4);
+    fw_cfg = fw_cfg_init_mem_nodma(addr, addr + 4, 1);
     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, ms->smp.cpus);
     fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, HPPA_MAX_CPUS);
     fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, ms->ram_size);
hw/nvram/fw_cfg.c +3 -4
index 54cfa07d3f5..10f8f8db86f 100644
@@ -1087,11 +1087,10 @@ FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
     return s;
 }
 
-FWCfgState *fw_cfg_init_mem(hwaddr ctl_addr, hwaddr data_addr)
+FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
+                                  unsigned data_width)
 {
-    return fw_cfg_init_mem_wide(ctl_addr, data_addr,
-                                fw_cfg_data_mem_ops.valid.max_access_size,
-                                0, NULL);
+    return fw_cfg_init_mem_wide(ctl_addr, data_addr, data_width, 0, NULL);
 }
 
 
Reviewed-by on v4 3/27
#17 lore raw

Re: [PATCH v4 03/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem() with '_nodma' suffix

In reply to Philippe Mathieu-Daudé
Quoted 31 lines · Philippe Mathieu-Daudé

Rename fw_cfg_init_mem() as fw_cfg_init_mem_nodma() to distinct with the DMA version (currently named fw_cfg_init_mem_wide).

---
  include/hw/nvram/fw_cfg.h | 3 ++-
  hw/hppa/machine.c         | 2 +-
  hw/nvram/fw_cfg.c         | 7 +++----
  3 files changed, 6 insertions(+), 6 deletions(-)
include/hw/nvram/fw_cfg.h +2 -1
index d41b9328fd1..d5161a79436 100644
@@ -307,7 +307,8 @@ bool fw_cfg_add_file_from_generator(FWCfgState *s,
 
  FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
                                  AddressSpace *dma_as);
-FWCfgState *fw_cfg_init_mem(hwaddr ctl_addr, hwaddr data_addr);
+FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
+                                  unsigned data_width);
  FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
                                   hwaddr data_addr, uint32_t data_width,
                                   hwaddr dma_addr, AddressSpace *dma_as);
hw/hppa/machine.c +0 -0
index dacedc5409c..0d768cb90b0 100644

hw/hppa/machine.c static FWCfgState *create_fw_cfg(MachineState *ms, PCIBus *pci_bus, · line 204

Quoted 5 lines
hw/hppa/machine.c +1 -1
@@ -201,7 +201,7 @@ static FWCfgState *create_fw_cfg(MachineState *ms, PCIBus *pci_bus,
      int btlb_entries = HPPA_BTLB_ENTRIES(&cpu[0]->env);
      int len;
 
-    fw_cfg = fw_cfg_init_mem(addr, addr + 4);
+    fw_cfg = fw_cfg_init_mem_nodma(addr, addr + 4, 1);

It does additional job to replace fw_cfg_data_mem_ops.valid.max_access_size with a hardcode 1.

It needs clarification at least in commit message that doing it is safe that fw_cfg_data_mem_ops.valid.max_access_size is not changed by other code.

It's even better to put this in a seperate patch.

For the pure rename part:

Quoted 22 lines
hw/hppa/machine.c +0 -0
      fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, ms->smp.cpus);
      fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, HPPA_MAX_CPUS);
      fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, ms->ram_size);
hw/nvram/fw_cfg.c +3 -4
index 54cfa07d3f5..10f8f8db86f 100644
@@ -1087,11 +1087,10 @@ FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
      return s;
  }
 
-FWCfgState *fw_cfg_init_mem(hwaddr ctl_addr, hwaddr data_addr)
+FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
+                                  unsigned data_width)
  {
-    return fw_cfg_init_mem_wide(ctl_addr, data_addr,
-                                fw_cfg_data_mem_ops.valid.max_access_size,
-                                0, NULL);
+    return fw_cfg_init_mem_wide(ctl_addr, data_addr, data_width, 0, NULL);
  }
 
 
#18 lore raw

PATCH v4 4/27 hw/mips/loongson3_virt: Prefer using fw_cfg_init_mem_nodma()

In reply to Philippe Mathieu-Daudé

fw_cfg_init_mem_wide() is prefered to initialize fw_cfg with DMA support. Without DMA, use fw_cfg_init_mem_nodma().

---
 hw/mips/loongson3_virt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
hw/mips/loongson3_virt.c +1 -1
index de6fbcc0cb4..654a2f0999f 100644
@@ -286,7 +286,7 @@ static void fw_conf_init(void)
     FWCfgState *fw_cfg;
     hwaddr cfg_addr = virt_memmap[VIRT_FW_CFG].base;
 
-    fw_cfg = fw_cfg_init_mem_wide(cfg_addr, cfg_addr + 8, 8, 0, NULL);
+    fw_cfg = fw_cfg_init_mem_nodma(cfg_addr, cfg_addr + 8, 8);
     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, (uint16_t)current_machine->smp.cpus);
     fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)current_machine->smp.max_cpus);
     fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, loaderparams.ram_size);
Signature · 1 line
2.47.1
Reviewed-by on v4 4/27
#19 lore raw

Re: [PATCH v4 04/27] hw/mips/loongson3_virt: Prefer using fw_cfg_init_mem_nodma()

Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:27 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 04/27] hw/mips/loongson3_virt: Prefer using
fw_cfg_init_mem_nodma()
X-Mailer: git-send-email 2.47.1

Quoted 7 lines

fw_cfg_init_mem_wide() is prefered to initialize fw_cfg with DMA support. Without DMA, use fw_cfg_init_mem_nodma().

---
 hw/mips/loongson3_virt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
Reviewed-by on v4 4/27
#20 lore raw

Re: [PATCH v4 04/27] hw/mips/loongson3_virt: Prefer using fw_cfg_init_mem_nodma()

In reply to Philippe Mathieu-Daudé
Quoted 4 lines · On Thu, 8 May 2025 15:35:27 +0200 Philippe Mathieu-Daudé <[email protected]>…

fw_cfg_init_mem_wide() is prefered to initialize fw_cfg with DMA support. Without DMA, use fw_cfg_init_mem_nodma().

hw/mips/loongson3_virt.c static void fw_conf_init(void) · line 289

Quoted 13 lines
---
 hw/mips/loongson3_virt.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
hw/mips/loongson3_virt.c +0 -1
index de6fbcc0cb4..654a2f0999f 100644
@@ -286,7 +286,7 @@ static void fw_conf_init(void)
     FWCfgState *fw_cfg;
     hwaddr cfg_addr = virt_memmap[VIRT_FW_CFG].base;
 
-    fw_cfg = fw_cfg_init_mem_wide(cfg_addr, cfg_addr + 8, 8, 0, NULL);

just a question, given it's a rather modern machine is there a reason why it is not using DMA here?

Quoted 4 lines
hw/mips/loongson3_virt.c +1 -0
+    fw_cfg = fw_cfg_init_mem_nodma(cfg_addr, cfg_addr + 8, 8);
     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, (uint16_t)current_machine->smp.cpus);
     fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)current_machine->smp.max_cpus);
     fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, loaderparams.ram_size);
Reviewed-by on v4 4/27
#21 lore raw

Re: [PATCH v4 04/27] hw/mips/loongson3_virt: Prefer using fw_cfg_init_mem_nodma()

In reply to Philippe Mathieu-Daudé
Quoted 4 lines · Philippe Mathieu-Daudé

fw_cfg_init_mem_wide() is prefered to initialize fw_cfg with DMA support. Without DMA, use fw_cfg_init_mem_nodma().

Quoted 17 lines
---
  hw/mips/loongson3_virt.c | 2 +-
  1 file changed, 1 insertion(+), 1 deletion(-)
hw/mips/loongson3_virt.c +1 -1
index de6fbcc0cb4..654a2f0999f 100644
@@ -286,7 +286,7 @@ static void fw_conf_init(void)
      FWCfgState *fw_cfg;
      hwaddr cfg_addr = virt_memmap[VIRT_FW_CFG].base;
 
-    fw_cfg = fw_cfg_init_mem_wide(cfg_addr, cfg_addr + 8, 8, 0, NULL);
+    fw_cfg = fw_cfg_init_mem_nodma(cfg_addr, cfg_addr + 8, 8);
      fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, (uint16_t)current_machine->smp.cpus);
      fw_cfg_add_i16(fw_cfg, FW_CFG_MAX_CPUS, (uint16_t)current_machine->smp.max_cpus);
      fw_cfg_add_i64(fw_cfg, FW_CFG_RAM_SIZE, loaderparams.ram_size);
#22 lore raw

PATCH v4 5/27 hw/nvram/fw_cfg: Factor fw_cfg_init_mem_internal() out

In reply to Philippe Mathieu-Daudé

Factor fw_cfg_init_mem_internal() out of fw_cfg_init_mem_wide(). In fw_cfg_init_mem_wide(), assert DMA arguments are provided. Callers without DMA have to use the fw_cfg_init_mem() helper.

---
 hw/nvram/fw_cfg.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)
hw/nvram/fw_cfg.c +13 -4
index 10f8f8db86f..4067324fb09 100644
@@ -1053,9 +1053,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
     return s;
 }
 
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as)
+static FWCfgState *fw_cfg_init_mem_internal(hwaddr ctl_addr,
+                                            hwaddr data_addr, uint32_t data_width,
+                                            hwaddr dma_addr, AddressSpace *dma_as)
 {
     DeviceState *dev;
     SysBusDevice *sbd;
@@ -1087,10 +1087,19 @@ FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
     return s;
 }
 
+FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
+                                 hwaddr data_addr, uint32_t data_width,
+                                 hwaddr dma_addr, AddressSpace *dma_as)
+{
+    assert(dma_addr && dma_as);
+    return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_addr,
+                                    dma_addr, dma_as);
+}
+
 FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
                                   unsigned data_width)
 {
-    return fw_cfg_init_mem_wide(ctl_addr, data_addr, data_width, 0, NULL);
+    return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_width, 0, NULL);
 }
 
 
Signature · 1 line
2.47.1
Reviewed-by on v4 5/27
#23 lore raw

Re: [PATCH v4 05/27] hw/nvram/fw_cfg: Factor fw_cfg_init_mem_internal() out

Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:28 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 05/27] hw/nvram/fw_cfg: Factor
fw_cfg_init_mem_internal() out
X-Mailer: git-send-email 2.47.1

Quoted 8 lines

Factor fw_cfg_init_mem_internal() out of fw_cfg_init_mem_wide(). In fw_cfg_init_mem_wide(), assert DMA arguments are provided. Callers without DMA have to use the fw_cfg_init_mem() helper.

---
 hw/nvram/fw_cfg.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)

Fine for me,

Reviewed-by on v4 5/27
#24 lore raw

Re: [PATCH v4 05/27] hw/nvram/fw_cfg: Factor fw_cfg_init_mem_internal() out

In reply to Philippe Mathieu-Daudé
Quoted 5 lines · On Thu, 8 May 2025 15:35:28 +0200 Philippe Mathieu-Daudé <[email protected]>…

Factor fw_cfg_init_mem_internal() out of fw_cfg_init_mem_wide(). In fw_cfg_init_mem_wide(), assert DMA arguments are provided. Callers without DMA have to use the fw_cfg_init_mem() helper.

Quoted 42 lines
---
 hw/nvram/fw_cfg.c | 17 +++++++++++++----
 1 file changed, 13 insertions(+), 4 deletions(-)
hw/nvram/fw_cfg.c +13 -4
index 10f8f8db86f..4067324fb09 100644
@@ -1053,9 +1053,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
     return s;
 }
 
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as)
+static FWCfgState *fw_cfg_init_mem_internal(hwaddr ctl_addr,
+                                            hwaddr data_addr, uint32_t data_width,
+                                            hwaddr dma_addr, AddressSpace *dma_as)
 {
     DeviceState *dev;
     SysBusDevice *sbd;
@@ -1087,10 +1087,19 @@ FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
     return s;
 }
 
+FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
+                                 hwaddr data_addr, uint32_t data_width,
+                                 hwaddr dma_addr, AddressSpace *dma_as)
+{
+    assert(dma_addr && dma_as);
+    return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_addr,
+                                    dma_addr, dma_as);
+}
+
 FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
                                   unsigned data_width)
 {
-    return fw_cfg_init_mem_wide(ctl_addr, data_addr, data_width, 0, NULL);
+    return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_width, 0, NULL);
 }
 
 
Reviewed-by on v4 5/27
#25 lore raw

Re: [PATCH v4 05/27] hw/nvram/fw_cfg: Factor fw_cfg_init_mem_internal() out

In reply to Philippe Mathieu-Daudé
Quoted 5 lines · Philippe Mathieu-Daudé

Factor fw_cfg_init_mem_internal() out of fw_cfg_init_mem_wide(). In fw_cfg_init_mem_wide(), assert DMA arguments are provided. Callers without DMA have to use the fw_cfg_init_mem() helper.

Quoted 42 lines
---
  hw/nvram/fw_cfg.c | 17 +++++++++++++----
  1 file changed, 13 insertions(+), 4 deletions(-)
hw/nvram/fw_cfg.c +13 -4
index 10f8f8db86f..4067324fb09 100644
@@ -1053,9 +1053,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
      return s;
  }
 
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as)
+static FWCfgState *fw_cfg_init_mem_internal(hwaddr ctl_addr,
+                                            hwaddr data_addr, uint32_t data_width,
+                                            hwaddr dma_addr, AddressSpace *dma_as)
  {
      DeviceState *dev;
      SysBusDevice *sbd;
@@ -1087,10 +1087,19 @@ FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
      return s;
  }
 
+FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
+                                 hwaddr data_addr, uint32_t data_width,
+                                 hwaddr dma_addr, AddressSpace *dma_as)
+{
+    assert(dma_addr && dma_as);
+    return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_addr,
+                                    dma_addr, dma_as);
+}
+
  FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
                                    unsigned data_width)
  {
-    return fw_cfg_init_mem_wide(ctl_addr, data_addr, data_width, 0, NULL);
+    return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_width, 0, NULL);
  }
 
 
#26 lore raw

Re: [PATCH v4 05/27] hw/nvram/fw_cfg: Factor fw_cfg_init_mem_internal() out

In reply to Philippe Mathieu-Daudé

hw/nvram/fw_cfg.c FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr, · line 1095

Quoted 6 lines
hw/nvram/fw_cfg.c +6 -0
+FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
+                                 hwaddr data_addr, uint32_t data_width,
+                                 hwaddr dma_addr, AddressSpace *dma_as)
+{
+    assert(dma_addr && dma_as);
+    return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_addr,
                                                            ^^^^^^^^^

a typo: data_width...The type was converted "silently", making it easy to be missed.

Quoted 1 line

+ dma_addr, dma_as);

Thanks,
Zhao

#27 lore raw

PATCH v4 6/27 hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() -> fw_cfg_init_mem_dma()

In reply to Philippe Mathieu-Daudé

"wide" in fw_cfg_init_mem_wide() means "DMA support". Rename for clarity.

---
 include/hw/nvram/fw_cfg.h | 6 +++---
 hw/arm/virt.c             | 2 +-
 hw/nvram/fw_cfg.c         | 6 +++---
 hw/riscv/virt.c           | 4 ++--
 4 files changed, 9 insertions(+), 9 deletions(-)
include/hw/nvram/fw_cfg.h +3 -3
index d5161a79436..c4c49886754 100644
@@ -309,9 +309,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
                                 AddressSpace *dma_as);
 FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
                                   unsigned data_width);
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as);
+FWCfgState *fw_cfg_init_mem_dma(hwaddr ctl_addr,
+                                hwaddr data_addr, uint32_t data_width,
+                                hwaddr dma_addr, AddressSpace *dma_as);
 
 FWCfgState *fw_cfg_find(void);
 bool fw_cfg_dma_enabled(void *opaque);
hw/arm/virt.c +1 -1
index 9a6cd085a37..7583f0a85d9 100644
@@ -1361,7 +1361,7 @@ static FWCfgState *create_fw_cfg(const VirtMachineState *vms, AddressSpace *as)
     FWCfgState *fw_cfg;
     char *nodename;
 
-    fw_cfg = fw_cfg_init_mem_wide(base + 8, base, 8, base + 16, as);
+    fw_cfg = fw_cfg_init_mem_dma(base + 8, base, 8, base + 16, as);
     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, (uint16_t)ms->smp.cpus);
 
     nodename = g_strdup_printf("/fw-cfg@%" PRIx64, base);
hw/nvram/fw_cfg.c +3 -3
index 4067324fb09..51b028b5d0a 100644
@@ -1087,9 +1087,9 @@ static FWCfgState *fw_cfg_init_mem_internal(hwaddr ctl_addr,
     return s;
 }
 
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as)
+FWCfgState *fw_cfg_init_mem_dma(hwaddr ctl_addr,
+                                hwaddr data_addr, uint32_t data_width,
+                                hwaddr dma_addr, AddressSpace *dma_as)
 {
     assert(dma_addr && dma_as);
     return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_addr,
hw/riscv/virt.c +2 -2
index be1bf0f6468..3ddea18c93e 100644
@@ -1266,8 +1266,8 @@ static FWCfgState *create_fw_cfg(const MachineState *ms)
     hwaddr base = virt_memmap[VIRT_FW_CFG].base;
     FWCfgState *fw_cfg;
 
-    fw_cfg = fw_cfg_init_mem_wide(base + 8, base, 8, base + 16,
-                                  &address_space_memory);
+    fw_cfg = fw_cfg_init_mem_dma(base + 8, base, 8, base + 16,
+                                 &address_space_memory);
     fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, (uint16_t)ms->smp.cpus);
 
     return fw_cfg;
Signature · 1 line
2.47.1
Reviewed-by on v4 6/27
#28 lore raw

Re: [PATCH v4 06/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() -> fw_cfg_init_mem_dma()

Quoted 9 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:29 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 06/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() ->
fw_cfg_init_mem_dma()
X-Mailer: git-send-email 2.47.1

"wide" in fw_cfg_init_mem_wide() means "DMA support". Rename for clarity.

Quoted 8 lines
---
 include/hw/nvram/fw_cfg.h | 6 +++---
 hw/arm/virt.c             | 2 +-
 hw/nvram/fw_cfg.c         | 6 +++---
 hw/riscv/virt.c           | 4 ++--
 4 files changed, 9 insertions(+), 9 deletions(-)
#29 lore raw

Re: [PATCH v4 06/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() -> fw_cfg_init_mem_dma()

In reply to Philippe Mathieu-Daudé
Quoted 23 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:29 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 06/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() ->
fw_cfg_init_mem_dma()
X-Mailer: git-send-email 2.47.1

"wide" in fw_cfg_init_mem_wide() means "DMA support". Rename for clarity.

---
 include/hw/nvram/fw_cfg.h | 6 +++---
 hw/arm/virt.c             | 2 +-
 hw/nvram/fw_cfg.c         | 6 +++---
 hw/riscv/virt.c           | 4 ++--
 4 files changed, 9 insertions(+), 9 deletions(-)
include/hw/nvram/fw_cfg.h +0 -0
index d5161a79436..c4c49886754 100644

include/hw/nvram/fw_cfg.h FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase, · line 314

Quoted 9 lines
include/hw/nvram/fw_cfg.h +3 -3
@@ -309,9 +309,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
                                 AddressSpace *dma_as);
 FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
                                   unsigned data_width);
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as);
+FWCfgState *fw_cfg_init_mem_dma(hwaddr ctl_addr,
+                                hwaddr data_addr, uint32_t data_width,
+                                hwaddr dma_addr, AddressSpace *dma_as);

There's one more use in latest master:

git grep fw_cfg_init_mem_wide
hw/loongarch/fw_cfg.c: fw_cfg = fw_cfg_init_mem_wide(VIRT_FWCFG_BASE + 8, VIRT_FWCFG_BASE, 8,

Reviewed-by
#30 lore raw

Re: [PATCH v4 06/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() -> fw_cfg_init_mem_dma()

Quoted 36 lines · On Fri, 9 May 2025 14:51:07 +0800 Zhao Liu <[email protected]> wrote:
Quoted 23 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:29 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 06/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() ->
fw_cfg_init_mem_dma()
X-Mailer: git-send-email 2.47.1

"wide" in fw_cfg_init_mem_wide() means "DMA support". Rename for clarity.

---
 include/hw/nvram/fw_cfg.h | 6 +++---
 hw/arm/virt.c             | 2 +-
 hw/nvram/fw_cfg.c         | 6 +++---
 hw/riscv/virt.c           | 4 ++--
 4 files changed, 9 insertions(+), 9 deletions(-)
include/hw/nvram/fw_cfg.h +0 -0
index d5161a79436..c4c49886754 100644

include/hw/nvram/fw_cfg.h FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase, · line 314

Quoted 9 lines
include/hw/nvram/fw_cfg.h +3 -3
@@ -309,9 +309,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
                                 AddressSpace *dma_as);
 FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
                                   unsigned data_width);
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as);
+FWCfgState *fw_cfg_init_mem_dma(hwaddr ctl_addr,
+                                hwaddr data_addr, uint32_t data_width,
+                                hwaddr dma_addr, AddressSpace *dma_as);  

There's one more use in latest master:

Quoted 2 lines

git grep fw_cfg_init_mem_wide
hw/loongarch/fw_cfg.c: fw_cfg = fw_cfg_init_mem_wide(VIRT_FWCFG_BASE + 8, VIRT_FWCFG_BASE, 8,

with that fixed:

Reviewed-by on v4 6/27
#31 lore raw

Re: [PATCH v4 06/27] hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() -> fw_cfg_init_mem_dma()

In reply to Philippe Mathieu-Daudé
Quoted 2 lines · Philippe Mathieu-Daudé

"wide" in fw_cfg_init_mem_wide() means "DMA support". Rename for clarity.

PS: at the time when fw_cfg_init_mem_wide() was first introcuded, 'wide' was exactly for data_width. see commit 6c87e3d5967a.

Quoted 2 lines

with the usage in hw/loongarch/fw_cfg.c fixed,

Quoted 69 lines
---
  include/hw/nvram/fw_cfg.h | 6 +++---
  hw/arm/virt.c             | 2 +-
  hw/nvram/fw_cfg.c         | 6 +++---
  hw/riscv/virt.c           | 4 ++--
  4 files changed, 9 insertions(+), 9 deletions(-)
include/hw/nvram/fw_cfg.h +3 -3
index d5161a79436..c4c49886754 100644
@@ -309,9 +309,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
                                  AddressSpace *dma_as);
  FWCfgState *fw_cfg_init_mem_nodma(hwaddr ctl_addr, hwaddr data_addr,
                                    unsigned data_width);
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as);
+FWCfgState *fw_cfg_init_mem_dma(hwaddr ctl_addr,
+                                hwaddr data_addr, uint32_t data_width,
+                                hwaddr dma_addr, AddressSpace *dma_as);
 
  FWCfgState *fw_cfg_find(void);
  bool fw_cfg_dma_enabled(void *opaque);
hw/arm/virt.c +1 -1
index 9a6cd085a37..7583f0a85d9 100644
@@ -1361,7 +1361,7 @@ static FWCfgState *create_fw_cfg(const VirtMachineState *vms, AddressSpace *as)
      FWCfgState *fw_cfg;
      char *nodename;
 
-    fw_cfg = fw_cfg_init_mem_wide(base + 8, base, 8, base + 16, as);
+    fw_cfg = fw_cfg_init_mem_dma(base + 8, base, 8, base + 16, as);
      fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, (uint16_t)ms->smp.cpus);
 
      nodename = g_strdup_printf("/fw-cfg@%" PRIx64, base);
hw/nvram/fw_cfg.c +3 -3
index 4067324fb09..51b028b5d0a 100644
@@ -1087,9 +1087,9 @@ static FWCfgState *fw_cfg_init_mem_internal(hwaddr ctl_addr,
      return s;
  }
 
-FWCfgState *fw_cfg_init_mem_wide(hwaddr ctl_addr,
-                                 hwaddr data_addr, uint32_t data_width,
-                                 hwaddr dma_addr, AddressSpace *dma_as)
+FWCfgState *fw_cfg_init_mem_dma(hwaddr ctl_addr,
+                                hwaddr data_addr, uint32_t data_width,
+                                hwaddr dma_addr, AddressSpace *dma_as)
  {
      assert(dma_addr && dma_as);
      return fw_cfg_init_mem_internal(ctl_addr, data_addr, data_addr,
hw/riscv/virt.c +2 -2
index be1bf0f6468..3ddea18c93e 100644
@@ -1266,8 +1266,8 @@ static FWCfgState *create_fw_cfg(const MachineState *ms)
      hwaddr base = virt_memmap[VIRT_FW_CFG].base;
      FWCfgState *fw_cfg;
 
-    fw_cfg = fw_cfg_init_mem_wide(base + 8, base, 8, base + 16,
-                                  &address_space_memory);
+    fw_cfg = fw_cfg_init_mem_dma(base + 8, base, 8, base + 16,
+                                 &address_space_memory);
      fw_cfg_add_i16(fw_cfg, FW_CFG_NB_CPUS, (uint16_t)ms->smp.cpus);
 
      return fw_cfg;
#32 lore raw

PATCH v4 7/27 hw/i386/x86: Remove X86MachineClass::fwcfg_dma_enabled field

In reply to Philippe Mathieu-Daudé

The X86MachineClass::fwcfg_dma_enabled boolean was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it and simplify.

'multiboot.bin' isn't used anymore, we'll remove it in the next commit.

---
 include/hw/i386/x86.h | 2 --
 hw/i386/microvm.c     | 3 ---
 hw/i386/multiboot.c   | 7 +------
 hw/i386/x86-common.c  | 3 +--
 hw/i386/x86.c         | 2 --
 5 files changed, 2 insertions(+), 15 deletions(-)
include/hw/i386/x86.h +0 -2
index fc460b82f82..29d37af11e6 100644
@@ -29,8 +29,6 @@
 struct X86MachineClass {
     MachineClass parent;
 
-    /* use DMA capable linuxboot option rom */
-    bool fwcfg_dma_enabled;
     /* CPU and apic information: */
     bool apic_xrupt_override;
 };
hw/i386/microvm.c +0 -3
index e0daf0d4fc3..b1262fb1523 100644
@@ -637,7 +637,6 @@ GlobalProperty microvm_properties[] = {
 
 static void microvm_class_init(ObjectClass *oc, const void *data)
 {
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
     MicrovmMachineClass *mmc = MICROVM_MACHINE_CLASS(oc);
     MachineClass *mc = MACHINE_CLASS(oc);
     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
@@ -671,8 +670,6 @@ static void microvm_class_init(ObjectClass *oc, const void *data)
     hc->unplug_request = microvm_device_unplug_request_cb;
     hc->unplug = microvm_device_unplug_cb;
 
-    x86mc->fwcfg_dma_enabled = true;
-
     object_class_property_add(oc, MICROVM_MACHINE_RTC, "OnOffAuto",
                               microvm_machine_get_rtc,
                               microvm_machine_set_rtc,
hw/i386/multiboot.c +1 -6
index 6e6b96bc345..bfa7e8f1e83 100644
@@ -153,7 +153,6 @@ int load_multiboot(X86MachineState *x86ms,
                    int kernel_file_size,
                    uint8_t *header)
 {
-    bool multiboot_dma_enabled = X86_MACHINE_GET_CLASS(x86ms)->fwcfg_dma_enabled;
     int i, is_multiboot = 0;
     uint32_t flags = 0;
     uint32_t mh_entry_addr;
@@ -402,11 +401,7 @@ int load_multiboot(X86MachineState *x86ms,
     fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, mb_bootinfo_data,
                      sizeof(bootinfo));
 
-    if (multiboot_dma_enabled) {
-        option_rom[nb_option_roms].name = "multiboot_dma.bin";
-    } else {
-        option_rom[nb_option_roms].name = "multiboot.bin";
-    }
+    option_rom[nb_option_roms].name = "multiboot_dma.bin";
     option_rom[nb_option_roms].bootindex = 0;
     nb_option_roms++;
 
hw/i386/x86-common.c +1 -2
index 1b0671c5239..27254a0e9f1 100644
@@ -634,7 +634,6 @@ void x86_load_linux(X86MachineState *x86ms,
                     int acpi_data_size,
                     bool pvh_enabled)
 {
-    bool linuxboot_dma_enabled = X86_MACHINE_GET_CLASS(x86ms)->fwcfg_dma_enabled;
     uint16_t protocol;
     int setup_size, kernel_size, cmdline_size;
     int dtb_size, setup_data_offset;
@@ -993,7 +992,7 @@ void x86_load_linux(X86MachineState *x86ms,
 
     option_rom[nb_option_roms].bootindex = 0;
     option_rom[nb_option_roms].name = "linuxboot.bin";
-    if (linuxboot_dma_enabled && fw_cfg_dma_enabled(fw_cfg)) {
+    if (fw_cfg_dma_enabled(fw_cfg)) {
         option_rom[nb_option_roms].name = "linuxboot_dma.bin";
     }
     nb_option_roms++;
hw/i386/x86.c +0 -2
index f80533df1c5..dbf104d60af 100644
@@ -375,14 +375,12 @@ static void x86_machine_initfn(Object *obj)
 static void x86_machine_class_init(ObjectClass *oc, const void *data)
 {
     MachineClass *mc = MACHINE_CLASS(oc);
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
     NMIClass *nc = NMI_CLASS(oc);
 
     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
     mc->kvm_type = x86_kvm_type;
-    x86mc->fwcfg_dma_enabled = true;
     nc->nmi_monitor_handler = x86_nmi;
 
     object_class_property_add(oc, X86_MACHINE_SMM, "OnOffAuto",
Signature · 1 line
2.47.1
Reviewed-by on v4 7/27
#33 lore raw

Re: [PATCH v4 07/27] hw/i386/x86: Remove X86MachineClass::fwcfg_dma_enabled field

Quoted 13 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:30 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 07/27] hw/i386/x86: Remove
X86MachineClass::fwcfg_dma_enabled field
X-Mailer: git-send-email 2.47.1

The X86MachineClass::fwcfg_dma_enabled boolean was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it and simplify.

'multiboot.bin' isn't used anymore, we'll remove it in the next commit.

Quoted 8 lines
---
 include/hw/i386/x86.h | 2 --
 hw/i386/microvm.c     | 3 ---
 hw/i386/multiboot.c   | 7 +------
 hw/i386/x86-common.c  | 3 +--
 hw/i386/x86.c         | 2 --
 5 files changed, 2 insertions(+), 15 deletions(-)
#34 lore raw

Re: [PATCH v4 07/27] hw/i386/x86: Remove X86MachineClass::fwcfg_dma_enabled field

In reply to Philippe Mathieu-Daudé
Quoted 63 lines · On Thu, 8 May 2025 15:35:30 +0200 Philippe Mathieu-Daudé <[email protected]>…

The X86MachineClass::fwcfg_dma_enabled boolean was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it and simplify.

'multiboot.bin' isn't used anymore, we'll remove it in the next commit.

---
 include/hw/i386/x86.h | 2 --
 hw/i386/microvm.c     | 3 ---
 hw/i386/multiboot.c   | 7 +------
 hw/i386/x86-common.c  | 3 +--
 hw/i386/x86.c         | 2 --
 5 files changed, 2 insertions(+), 15 deletions(-)
include/hw/i386/x86.h +0 -2
index fc460b82f82..29d37af11e6 100644
@@ -29,8 +29,6 @@
 struct X86MachineClass {
     MachineClass parent;
 
-    /* use DMA capable linuxboot option rom */
-    bool fwcfg_dma_enabled;
     /* CPU and apic information: */
     bool apic_xrupt_override;
 };
hw/i386/microvm.c +0 -3
index e0daf0d4fc3..b1262fb1523 100644
@@ -637,7 +637,6 @@ GlobalProperty microvm_properties[] = {
 
 static void microvm_class_init(ObjectClass *oc, const void *data)
 {
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
     MicrovmMachineClass *mmc = MICROVM_MACHINE_CLASS(oc);
     MachineClass *mc = MACHINE_CLASS(oc);
     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
@@ -671,8 +670,6 @@ static void microvm_class_init(ObjectClass *oc, const void *data)
     hc->unplug_request = microvm_device_unplug_request_cb;
     hc->unplug = microvm_device_unplug_cb;
 
-    x86mc->fwcfg_dma_enabled = true;
-
     object_class_property_add(oc, MICROVM_MACHINE_RTC, "OnOffAuto",
                               microvm_machine_get_rtc,
                               microvm_machine_set_rtc,
hw/i386/multiboot.c +0 -1
index 6e6b96bc345..bfa7e8f1e83 100644
@@ -153,7 +153,6 @@ int load_multiboot(X86MachineState *x86ms,
                    int kernel_file_size,
                    uint8_t *header)
 {
-    bool multiboot_dma_enabled = X86_MACHINE_GET_CLASS(x86ms)->fwcfg_dma_enabled;
     int i, is_multiboot = 0;
     uint32_t flags = 0;
     uint32_t mh_entry_addr;

hw/i386/multiboot.c int load_multiboot(X86MachineState *x86ms, · line 408

Quoted 7 lines
hw/i386/multiboot.c +0 -4
@@ -402,11 +401,7 @@ int load_multiboot(X86MachineState *x86ms,
     fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, mb_bootinfo_data,
                      sizeof(bootinfo));
 
-    if (multiboot_dma_enabled) {
-        option_rom[nb_option_roms].name = "multiboot_dma.bin";
-    } else {
-        option_rom[nb_option_roms].name = "multiboot.bin";

shouldn't we remove "multiboot.bin" as well?

Quoted 45 lines
hw/i386/multiboot.c +1 -1
-    }
+    option_rom[nb_option_roms].name = "multiboot_dma.bin";
     option_rom[nb_option_roms].bootindex = 0;
     nb_option_roms++;
 
hw/i386/x86-common.c +1 -2
index 1b0671c5239..27254a0e9f1 100644
@@ -634,7 +634,6 @@ void x86_load_linux(X86MachineState *x86ms,
                     int acpi_data_size,
                     bool pvh_enabled)
 {
-    bool linuxboot_dma_enabled = X86_MACHINE_GET_CLASS(x86ms)->fwcfg_dma_enabled;
     uint16_t protocol;
     int setup_size, kernel_size, cmdline_size;
     int dtb_size, setup_data_offset;
@@ -993,7 +992,7 @@ void x86_load_linux(X86MachineState *x86ms,
 
     option_rom[nb_option_roms].bootindex = 0;
     option_rom[nb_option_roms].name = "linuxboot.bin";
-    if (linuxboot_dma_enabled && fw_cfg_dma_enabled(fw_cfg)) {
+    if (fw_cfg_dma_enabled(fw_cfg)) {
         option_rom[nb_option_roms].name = "linuxboot_dma.bin";
     }
     nb_option_roms++;
hw/i386/x86.c +0 -2
index f80533df1c5..dbf104d60af 100644
@@ -375,14 +375,12 @@ static void x86_machine_initfn(Object *obj)
 static void x86_machine_class_init(ObjectClass *oc, const void *data)
 {
     MachineClass *mc = MACHINE_CLASS(oc);
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
     NMIClass *nc = NMI_CLASS(oc);
 
     mc->cpu_index_to_instance_props = x86_cpu_index_to_props;
     mc->get_default_cpu_node_id = x86_get_default_cpu_node_id;
     mc->possible_cpu_arch_ids = x86_possible_cpu_arch_ids;
     mc->kvm_type = x86_kvm_type;
-    x86mc->fwcfg_dma_enabled = true;
     nc->nmi_monitor_handler = x86_nmi;
 
     object_class_property_add(oc, X86_MACHINE_SMM, "OnOffAuto",
Reviewed-by
#35 lore raw

Re: [PATCH v4 07/27] hw/i386/x86: Remove X86MachineClass::fwcfg_dma_enabled field

Quoted 69 lines · On Fri, 9 May 2025 17:41:25 +0200 Igor Mammedov <[email protected]> wrote:
Quoted 66 lines · On Thu, 8 May 2025 15:35:30 +0200 Philippe Mathieu-Daudé <[email protected]>…

The X86MachineClass::fwcfg_dma_enabled boolean was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it and simplify.

'multiboot.bin' isn't used anymore, we'll remove it in the next commit.

---
 include/hw/i386/x86.h | 2 --
 hw/i386/microvm.c     | 3 ---
 hw/i386/multiboot.c   | 7 +------
 hw/i386/x86-common.c  | 3 +--
 hw/i386/x86.c         | 2 --
 5 files changed, 2 insertions(+), 15 deletions(-)
include/hw/i386/x86.h +0 -2
index fc460b82f82..29d37af11e6 100644
@@ -29,8 +29,6 @@
 struct X86MachineClass {
     MachineClass parent;
 
-    /* use DMA capable linuxboot option rom */
-    bool fwcfg_dma_enabled;
     /* CPU and apic information: */
     bool apic_xrupt_override;
 };
hw/i386/microvm.c +0 -3
index e0daf0d4fc3..b1262fb1523 100644
@@ -637,7 +637,6 @@ GlobalProperty microvm_properties[] = {
 
 static void microvm_class_init(ObjectClass *oc, const void *data)
 {
-    X86MachineClass *x86mc = X86_MACHINE_CLASS(oc);
     MicrovmMachineClass *mmc = MICROVM_MACHINE_CLASS(oc);
     MachineClass *mc = MACHINE_CLASS(oc);
     HotplugHandlerClass *hc = HOTPLUG_HANDLER_CLASS(oc);
@@ -671,8 +670,6 @@ static void microvm_class_init(ObjectClass *oc, const void *data)
     hc->unplug_request = microvm_device_unplug_request_cb;
     hc->unplug = microvm_device_unplug_cb;
 
-    x86mc->fwcfg_dma_enabled = true;
-
     object_class_property_add(oc, MICROVM_MACHINE_RTC, "OnOffAuto",
                               microvm_machine_get_rtc,
                               microvm_machine_set_rtc,
hw/i386/multiboot.c +0 -1
index 6e6b96bc345..bfa7e8f1e83 100644
@@ -153,7 +153,6 @@ int load_multiboot(X86MachineState *x86ms,
                    int kernel_file_size,
                    uint8_t *header)
 {
-    bool multiboot_dma_enabled = X86_MACHINE_GET_CLASS(x86ms)->fwcfg_dma_enabled;
     int i, is_multiboot = 0;
     uint32_t flags = 0;
     uint32_t mh_entry_addr;
@@ -402,11 +401,7 @@ int load_multiboot(X86MachineState *x86ms,
     fw_cfg_add_bytes(fw_cfg, FW_CFG_INITRD_DATA, mb_bootinfo_data,
                      sizeof(bootinfo));
 
Quoted 6 lines

hw/i386/multiboot.c int load_multiboot(X86MachineState *x86ms, · line 408

Quoted 4 lines
hw/i386/multiboot.c +0 -4
-    if (multiboot_dma_enabled) {
-        option_rom[nb_option_roms].name = "multiboot_dma.bin";
-    } else {
-        option_rom[nb_option_roms].name = "multiboot.bin";  

shouldn't we remove "multiboot.bin" as well?

never mind, I see it's being removed by the next patch

#36 lore raw

PATCH v4 8/27 hw/i386/pc: Remove multiboot.bin

In reply to Philippe Mathieu-Daudé

All PC machines now use the multiboot_dma.bin binary, we can remove the non-DMA version (multiboot.bin).

---
 pc-bios/optionrom/optionrom.h     |   4 -
 hw/i386/pc.c                      |   1 -
 pc-bios/meson.build               |   1 -
 pc-bios/multiboot.bin             | Bin 1024 -> 0 bytes
 pc-bios/optionrom/Makefile        |   2 +-
 pc-bios/optionrom/multiboot.S     | 232 -----------------------------
 pc-bios/optionrom/multiboot_dma.S | 234 +++++++++++++++++++++++++++++-
 7 files changed, 233 insertions(+), 241 deletions(-)
 delete mode 100644 pc-bios/multiboot.bin
 delete mode 100644 pc-bios/optionrom/multiboot.S
pc-bios/optionrom/optionrom.h +0 -4
index 7bcdf0eeb24..2e6e2493f83 100644
@@ -117,16 +117,12 @@
  *
  * Clobbers: %eax, %edx, %es, %ecx, %edi and adresses %esp-20 to %esp
  */
-#ifdef USE_FW_CFG_DMA
 #define read_fw_blob_dma(var)                           \
         read_fw         var ## _SIZE;                   \
         mov             %eax, %ecx;                     \
         read_fw         var ## _ADDR;                   \
         mov             %eax, %edi ;                    \
         read_fw_dma     var ## _DATA, %ecx, %edi
-#else
-#define read_fw_blob_dma(var) read_fw_blob(var)
-#endif
 
 #define read_fw_blob_pre(var)                           \
         read_fw         var ## _SIZE;                   \
hw/i386/pc.c +0 -1
index 49632b69d29..524d2fd98e8 100644
@@ -657,7 +657,6 @@ void xen_load_linux(PCMachineState *pcms)
         assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
                !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
                !strcmp(option_rom[i].name, "pvh.bin") ||
-               !strcmp(option_rom[i].name, "multiboot.bin") ||
                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
     }
pc-bios/meson.build +0 -1
index 79bb2e18008..f2d4dc416a4 100644
@@ -62,7 +62,6 @@ blobs = [
   'efi-e1000e.rom',
   'efi-vmxnet3.rom',
   'qemu-nsis.bmp',
-  'multiboot.bin',
   'multiboot_dma.bin',
   'linuxboot.bin',
   'linuxboot_dma.bin',
pc-bios/multiboot.bin deleted binary +0 -0
deleted file mode 100644
index e772713c95749bee82c20002b50ec6d05b2d4987..0000000000000000000000000000000000000000
GIT binary patch
pc-bios/optionrom/Makefile +1 -1
index 30d07026c79..1183ef88922 100644
@@ -2,7 +2,7 @@ include config.mak
 SRC_DIR := $(TOPSRC_DIR)/pc-bios/optionrom
 VPATH = $(SRC_DIR)
 
-all: multiboot.bin multiboot_dma.bin linuxboot.bin linuxboot_dma.bin kvmvapic.bin pvh.bin
+all: multiboot_dma.bin linuxboot.bin linuxboot_dma.bin kvmvapic.bin pvh.bin
 # Dummy command so that make thinks it has done something
 	@true
 
pc-bios/optionrom/multiboot.S deleted +0 -232

232 lines not on this page · raw message

pc-bios/optionrom/multiboot_dma.S +232 -2

234 lines not on this page · raw message

Signature · 1 line
2.47.1
Reviewed-by on v4 8/27
#37 lore raw

Re: [PATCH v4 08/27] hw/i386/pc: Remove multiboot.bin

Quoted 30 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:31 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 08/27] hw/i386/pc: Remove multiboot.bin
X-Mailer: git-send-email 2.47.1

All PC machines now use the multiboot_dma.bin binary, we can remove the non-DMA version (multiboot.bin).

---
 pc-bios/optionrom/optionrom.h     |   4 -
 hw/i386/pc.c                      |   1 -
 pc-bios/meson.build               |   1 -
 pc-bios/multiboot.bin             | Bin 1024 -> 0 bytes
 pc-bios/optionrom/Makefile        |   2 +-
 pc-bios/optionrom/multiboot.S     | 232 -----------------------------
 pc-bios/optionrom/multiboot_dma.S | 234 +++++++++++++++++++++++++++++-
 7 files changed, 233 insertions(+), 241 deletions(-)
 delete mode 100644 pc-bios/multiboot.bin
 delete mode 100644 pc-bios/optionrom/multiboot.S
pc-bios/optionrom/optionrom.h +0 -0
index 7bcdf0eeb24..2e6e2493f83 100644
@@ -117,16 +117,12 @@
  *
  * Clobbers: %eax, %edx, %es, %ecx, %edi and adresses %esp-20 to %esp
  */

pc-bios/optionrom/optionrom.h line 129

Quoted 10 lines
pc-bios/optionrom/optionrom.h +0 -4
-#ifdef USE_FW_CFG_DMA
 #define read_fw_blob_dma(var)                           \
         read_fw         var ## _SIZE;                   \
         mov             %eax, %ecx;                     \
         read_fw         var ## _ADDR;                   \
         mov             %eax, %edi ;                    \
         read_fw_dma     var ## _DATA, %ecx, %edi
-#else
-#define read_fw_blob_dma(var) read_fw_blob(var)
-#endif

It seems read_fw_blob() could be dropped as well and this is not a big deal. So,

#38 lore raw

PATCH v4 9/27 hw/nvram/fw_cfg: Remove fw_cfg_io_properties::dma_enabled

In reply to Philippe Mathieu-Daudé

Now than all calls to fw_cfg_init_io_dma() pass DMA arguments, the 'dma_enabled' of the TYPE_FW_CFG_IO type is not used anymore. Remove it, simplifying fw_cfg_init_io_dma() and fw_cfg_io_realize().

Note, we can not remove the equivalent in fw_cfg_mem_properties[] because it is still used in HPPA and MIPS Loongson3 machines:

  $ git grep -w fw_cfg_init_mem_nodma
  hw/hppa/machine.c:204:    fw_cfg = fw_cfg_init_mem_nodma(addr, addr + 4, 1);
  hw/mips/loongson3_virt.c:289:    fw_cfg = fw_cfg_init_mem_nodma(cfg_addr, cfg_addr + 8, 8);

'linuxboot.bin' isn't used anymore, we'll remove it in the next commit.

---
 hw/i386/fw_cfg.c     |  5 +----
 hw/i386/x86-common.c |  5 +----
 hw/nvram/fw_cfg.c    | 26 ++++++++------------------
 3 files changed, 10 insertions(+), 26 deletions(-)
hw/i386/fw_cfg.c +1 -4
index 5c0bcd5f8a9..1fe084fd720 100644
@@ -221,10 +221,7 @@ void fw_cfg_add_acpi_dsdt(Aml *scope, FWCfgState *fw_cfg)
      * of the i/o region used is FW_CFG_CTL_SIZE; when using DMA, the
      * DMA control register is located at FW_CFG_DMA_IO_BASE + 4
      */
-    Object *obj = OBJECT(fw_cfg);
-    uint8_t io_size = object_property_get_bool(obj, "dma_enabled", NULL) ?
-        ROUND_UP(FW_CFG_CTL_SIZE, 4) + sizeof(dma_addr_t) :
-        FW_CFG_CTL_SIZE;
+    uint8_t io_size = ROUND_UP(FW_CFG_CTL_SIZE, 4) + sizeof(dma_addr_t);
     Aml *dev = aml_device("FWCF");
     Aml *crs = aml_resource_template();
 
hw/i386/x86-common.c +1 -4
index 27254a0e9f1..ee594364415 100644
@@ -991,10 +991,7 @@ void x86_load_linux(X86MachineState *x86ms,
     }
 
     option_rom[nb_option_roms].bootindex = 0;
-    option_rom[nb_option_roms].name = "linuxboot.bin";
-    if (fw_cfg_dma_enabled(fw_cfg)) {
-        option_rom[nb_option_roms].name = "linuxboot_dma.bin";
-    }
+    option_rom[nb_option_roms].name = "linuxboot_dma.bin";
     nb_option_roms++;
 }
 
hw/nvram/fw_cfg.c +8 -18
index 51b028b5d0a..ef976a4bce2 100644
@@ -1026,12 +1026,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
     FWCfgIoState *ios;
     FWCfgState *s;
     MemoryRegion *iomem = get_system_io();
-    bool dma_requested = dma_iobase && dma_as;
 
+    assert(dma_iobase);
     dev = qdev_new(TYPE_FW_CFG_IO);
-    if (!dma_requested) {
-        qdev_prop_set_bit(dev, "dma_enabled", false);
-    }
 
     object_property_add_child(OBJECT(qdev_get_machine()), TYPE_FW_CFG,
                               OBJECT(dev));
@@ -1042,13 +1039,10 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
     memory_region_add_subregion(iomem, iobase, &ios->comb_iomem);
 
     s = FW_CFG(dev);
-
-    if (s->dma_enabled) {
-        /* 64 bits for the address field */
-        s->dma_as = dma_as;
-        s->dma_addr = 0;
-        memory_region_add_subregion(iomem, dma_iobase, &s->dma_iomem);
-    }
+    /* 64 bits for the address field */
+    s->dma_as = dma_as;
+    s->dma_addr = 0;
+    memory_region_add_subregion(iomem, dma_iobase, &s->dma_iomem);
 
     return s;
 }
@@ -1185,8 +1179,6 @@ static void fw_cfg_file_slots_allocate(FWCfgState *s, Error **errp)
 }
 
 static const Property fw_cfg_io_properties[] = {
-    DEFINE_PROP_BOOL("dma_enabled", FWCfgIoState, parent_obj.dma_enabled,
-                     true),
     DEFINE_PROP_UINT16("x-file-slots", FWCfgIoState, parent_obj.file_slots,
                        FW_CFG_FILE_SLOTS_DFLT),
 };
@@ -1207,11 +1199,9 @@ static void fw_cfg_io_realize(DeviceState *dev, Error **errp)
     memory_region_init_io(&s->comb_iomem, OBJECT(s), &fw_cfg_comb_mem_ops,
                           FW_CFG(s), "fwcfg", FW_CFG_CTL_SIZE);
 
-    if (FW_CFG(s)->dma_enabled) {
-        memory_region_init_io(&FW_CFG(s)->dma_iomem, OBJECT(s),
-                              &fw_cfg_dma_mem_ops, FW_CFG(s), "fwcfg.dma",
-                              sizeof(dma_addr_t));
-    }
+    memory_region_init_io(&FW_CFG(s)->dma_iomem, OBJECT(s),
+                          &fw_cfg_dma_mem_ops, FW_CFG(s), "fwcfg.dma",
+                          sizeof(dma_addr_t));
 
     fw_cfg_common_realize(dev, errp);
 }
Signature · 1 line
2.47.1
#39 lore raw

Re: [PATCH v4 09/27] hw/nvram/fw_cfg: Remove fw_cfg_io_properties::dma_enabled

Quoted 65 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:32 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 09/27] hw/nvram/fw_cfg: Remove
fw_cfg_io_properties::dma_enabled
X-Mailer: git-send-email 2.47.1

Now than all calls to fw_cfg_init_io_dma() pass DMA arguments, the 'dma_enabled' of the TYPE_FW_CFG_IO type is not used anymore. Remove it, simplifying fw_cfg_init_io_dma() and fw_cfg_io_realize().

Note, we can not remove the equivalent in fw_cfg_mem_properties[] because it is still used in HPPA and MIPS Loongson3 machines:

  $ git grep -w fw_cfg_init_mem_nodma
  hw/hppa/machine.c:204:    fw_cfg = fw_cfg_init_mem_nodma(addr, addr + 4, 1);
  hw/mips/loongson3_virt.c:289:    fw_cfg = fw_cfg_init_mem_nodma(cfg_addr, cfg_addr + 8, 8);

'linuxboot.bin' isn't used anymore, we'll remove it in the next commit.

---
 hw/i386/fw_cfg.c     |  5 +----
 hw/i386/x86-common.c |  5 +----
 hw/nvram/fw_cfg.c    | 26 ++++++++------------------
 3 files changed, 10 insertions(+), 26 deletions(-)
hw/i386/fw_cfg.c +1 -4
index 5c0bcd5f8a9..1fe084fd720 100644
@@ -221,10 +221,7 @@ void fw_cfg_add_acpi_dsdt(Aml *scope, FWCfgState *fw_cfg)
      * of the i/o region used is FW_CFG_CTL_SIZE; when using DMA, the
      * DMA control register is located at FW_CFG_DMA_IO_BASE + 4
      */
-    Object *obj = OBJECT(fw_cfg);
-    uint8_t io_size = object_property_get_bool(obj, "dma_enabled", NULL) ?
-        ROUND_UP(FW_CFG_CTL_SIZE, 4) + sizeof(dma_addr_t) :
-        FW_CFG_CTL_SIZE;
+    uint8_t io_size = ROUND_UP(FW_CFG_CTL_SIZE, 4) + sizeof(dma_addr_t);
     Aml *dev = aml_device("FWCF");
     Aml *crs = aml_resource_template();
 
hw/i386/x86-common.c +1 -4
index 27254a0e9f1..ee594364415 100644
@@ -991,10 +991,7 @@ void x86_load_linux(X86MachineState *x86ms,
     }
 
     option_rom[nb_option_roms].bootindex = 0;
-    option_rom[nb_option_roms].name = "linuxboot.bin";
-    if (fw_cfg_dma_enabled(fw_cfg)) {
-        option_rom[nb_option_roms].name = "linuxboot_dma.bin";
-    }
+    option_rom[nb_option_roms].name = "linuxboot_dma.bin";
     nb_option_roms++;
 }
 
hw/nvram/fw_cfg.c +0 -0
index 51b028b5d0a..ef976a4bce2 100644

hw/nvram/fw_cfg.c FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase, · line 1030

Quoted 6 lines
hw/nvram/fw_cfg.c +1 -1
@@ -1026,12 +1026,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
     FWCfgIoState *ios;
     FWCfgState *s;
     MemoryRegion *iomem = get_system_io();
-    bool dma_requested = dma_iobase && dma_as;
 
+    assert(dma_iobase);

Maybe a rebase nit? In v3, it is:

assert(dma_iobase && dma_as);

Others are fine for me.

#40 lore raw

Re: [PATCH v4 09/27] hw/nvram/fw_cfg: Remove fw_cfg_io_properties::dma_enabled

In reply to Philippe Mathieu-Daudé
Quoted 91 lines · On Thu, 8 May 2025 15:35:32 +0200 Philippe Mathieu-Daudé <[email protected]>…

Now than all calls to fw_cfg_init_io_dma() pass DMA arguments, the 'dma_enabled' of the TYPE_FW_CFG_IO type is not used anymore. Remove it, simplifying fw_cfg_init_io_dma() and fw_cfg_io_realize().

Note, we can not remove the equivalent in fw_cfg_mem_properties[] because it is still used in HPPA and MIPS Loongson3 machines:

  $ git grep -w fw_cfg_init_mem_nodma
  hw/hppa/machine.c:204:    fw_cfg = fw_cfg_init_mem_nodma(addr, addr + 4, 1);
  hw/mips/loongson3_virt.c:289:    fw_cfg = fw_cfg_init_mem_nodma(cfg_addr, cfg_addr + 8, 8);

'linuxboot.bin' isn't used anymore, we'll remove it in the next commit.

---
 hw/i386/fw_cfg.c     |  5 +----
 hw/i386/x86-common.c |  5 +----
 hw/nvram/fw_cfg.c    | 26 ++++++++------------------
 3 files changed, 10 insertions(+), 26 deletions(-)
hw/i386/fw_cfg.c +1 -4
index 5c0bcd5f8a9..1fe084fd720 100644
@@ -221,10 +221,7 @@ void fw_cfg_add_acpi_dsdt(Aml *scope, FWCfgState *fw_cfg)
      * of the i/o region used is FW_CFG_CTL_SIZE; when using DMA, the
      * DMA control register is located at FW_CFG_DMA_IO_BASE + 4
      */
-    Object *obj = OBJECT(fw_cfg);
-    uint8_t io_size = object_property_get_bool(obj, "dma_enabled", NULL) ?
-        ROUND_UP(FW_CFG_CTL_SIZE, 4) + sizeof(dma_addr_t) :
-        FW_CFG_CTL_SIZE;
+    uint8_t io_size = ROUND_UP(FW_CFG_CTL_SIZE, 4) + sizeof(dma_addr_t);
     Aml *dev = aml_device("FWCF");
     Aml *crs = aml_resource_template();
 
hw/i386/x86-common.c +1 -4
index 27254a0e9f1..ee594364415 100644
@@ -991,10 +991,7 @@ void x86_load_linux(X86MachineState *x86ms,
     }
 
     option_rom[nb_option_roms].bootindex = 0;
-    option_rom[nb_option_roms].name = "linuxboot.bin";
-    if (fw_cfg_dma_enabled(fw_cfg)) {
-        option_rom[nb_option_roms].name = "linuxboot_dma.bin";
-    }
+    option_rom[nb_option_roms].name = "linuxboot_dma.bin";
     nb_option_roms++;
 }
 
hw/nvram/fw_cfg.c +5 -11
index 51b028b5d0a..ef976a4bce2 100644
@@ -1026,12 +1026,9 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
     FWCfgIoState *ios;
     FWCfgState *s;
     MemoryRegion *iomem = get_system_io();
-    bool dma_requested = dma_iobase && dma_as;
 
+    assert(dma_iobase);
     dev = qdev_new(TYPE_FW_CFG_IO);
-    if (!dma_requested) {
-        qdev_prop_set_bit(dev, "dma_enabled", false);
-    }
 
     object_property_add_child(OBJECT(qdev_get_machine()), TYPE_FW_CFG,
                               OBJECT(dev));
@@ -1042,13 +1039,10 @@ FWCfgState *fw_cfg_init_io_dma(uint32_t iobase, uint32_t dma_iobase,
     memory_region_add_subregion(iomem, iobase, &ios->comb_iomem);
 
     s = FW_CFG(dev);
-
-    if (s->dma_enabled) {
-        /* 64 bits for the address field */
-        s->dma_as = dma_as;
-        s->dma_addr = 0;
-        memory_region_add_subregion(iomem, dma_iobase, &s->dma_iomem);
-    }
+    /* 64 bits for the address field */
+    s->dma_as = dma_as;
+    s->dma_addr = 0;
+    memory_region_add_subregion(iomem, dma_iobase, &s->dma_iomem);
 
     return s;
 }

hw/nvram/fw_cfg.c static void fw_cfg_file_slots_allocate(FWCfgState *s, Error **errp) · line 1189

Quoted 5 lines
hw/nvram/fw_cfg.c +0 -2
@@ -1185,8 +1179,6 @@ static void fw_cfg_file_slots_allocate(FWCfgState *s, Error **errp)
 }
 
 static const Property fw_cfg_io_properties[] = {
-    DEFINE_PROP_BOOL("dma_enabled", FWCfgIoState, parent_obj.dma_enabled,
-                     true),

in 7/27 you still have, fw_cfg_dma_enabled(fw_cfg) which works around now missing property (in IO case) in obscure way.
that is also used in bios_linker_loader_can_write_pointer().

It would be better to get rid of fw_cfg_dma_enabled() as well or keep property as RO and replace fw_cfg_dma_enabled() with property accessor if both calls can happen on nondma path

Quoted 18 lines
hw/nvram/fw_cfg.c +3 -5
     DEFINE_PROP_UINT16("x-file-slots", FWCfgIoState, parent_obj.file_slots,
                        FW_CFG_FILE_SLOTS_DFLT),
 };
@@ -1207,11 +1199,9 @@ static void fw_cfg_io_realize(DeviceState *dev, Error **errp)
     memory_region_init_io(&s->comb_iomem, OBJECT(s), &fw_cfg_comb_mem_ops,
                           FW_CFG(s), "fwcfg", FW_CFG_CTL_SIZE);
 
-    if (FW_CFG(s)->dma_enabled) {
-        memory_region_init_io(&FW_CFG(s)->dma_iomem, OBJECT(s),
-                              &fw_cfg_dma_mem_ops, FW_CFG(s), "fwcfg.dma",
-                              sizeof(dma_addr_t));
-    }
+    memory_region_init_io(&FW_CFG(s)->dma_iomem, OBJECT(s),
+                          &fw_cfg_dma_mem_ops, FW_CFG(s), "fwcfg.dma",
+                          sizeof(dma_addr_t));
 
     fw_cfg_common_realize(dev, errp);
 }
#41 lore raw

Re: [PATCH v4 09/27] hw/nvram/fw_cfg: Remove fw_cfg_io_properties::dma_enabled

In reply to Philippe Mathieu-Daudé
Quoted 7 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:32 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 09/27] hw/nvram/fw_cfg: Remove
fw_cfg_io_properties::dma_enabled
X-Mailer: git-send-email 2.47.1

Now than all calls to fw_cfg_init_io_dma() pass DMA arguments,

Yes,

Quoted 2 lines

the 'dma_enabled' of the TYPE_FW_CFG_IO type is not used anymore. Remove it, simplifying fw_cfg_init_io_dma() and fw_cfg_io_realize().

but the 'dma_enabled' of the TYPE_FW_CFG_IO type is still used in hw/sparc64/sun4u.c:

    dev = qdev_new(TYPE_FW_CFG_IO);
    qdev_prop_set_bit(dev, "dma_enabled", false);

The creation of TYPE_FW_CFG_IO is similar to fw_cfg_init_io_dma(), but it still has little difference so I find sun4uv can't use fw_cfg_init_io_dma() directly for now, or it may require more careful clarification that it can use fw_cfg_init_io_dma().

So, at least we have to keep "dma_enabled" property in fw_cfg_io_properties[].

...

Quoted 5 lines
hw/i386/x86-common.c +0 -0
index 27254a0e9f1..ee594364415 100644

hw/i386/x86-common.c void x86_load_linux(X86MachineState *x86ms, · line 996

Quoted 10 lines
hw/i386/x86-common.c +1 -4
@@ -991,10 +991,7 @@ void x86_load_linux(X86MachineState *x86ms,
     }
 
     option_rom[nb_option_roms].bootindex = 0;
-    option_rom[nb_option_roms].name = "linuxboot.bin";
-    if (fw_cfg_dma_enabled(fw_cfg)) {
-        option_rom[nb_option_roms].name = "linuxboot_dma.bin";
-    }
+    option_rom[nb_option_roms].name = "linuxboot_dma.bin";
     nb_option_roms++;
 }

But now all x86 machines have enabled dma, so it's still possible to drop "linuxboot.bin".

For this, I think we could add a DMA check in the x86 common code to guarantee DMA is enabled and then we won't need to constantly check DMA enable in every corner.

Regards,
Zhao

#42 lore raw

PATCH v4 10/27 hw/i386/pc: Remove linuxboot.bin

In reply to Philippe Mathieu-Daudé

All PC machines now use the linuxboot_dma.bin binary, we can remove the non-DMA version (linuxboot.bin).

---
 hw/i386/pc.c                  |   3 +-
 pc-bios/meson.build           |   1 -
 pc-bios/optionrom/Makefile    |   2 +-
 pc-bios/optionrom/linuxboot.S | 195 ----------------------------------
 4 files changed, 2 insertions(+), 199 deletions(-)
 delete mode 100644 pc-bios/optionrom/linuxboot.S
hw/i386/pc.c +1 -2
index 524d2fd98e8..4e6fe68e2e0 100644
@@ -654,8 +654,7 @@ void xen_load_linux(PCMachineState *pcms)
 
     x86_load_linux(x86ms, fw_cfg, PC_FW_DATA, pcmc->pvh_enabled);
     for (i = 0; i < nb_option_roms; i++) {
-        assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
-               !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
+        assert(!strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
                !strcmp(option_rom[i].name, "pvh.bin") ||
                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
pc-bios/meson.build +0 -1
index f2d4dc416a4..39a7fea332e 100644
@@ -63,7 +63,6 @@ blobs = [
   'efi-vmxnet3.rom',
   'qemu-nsis.bmp',
   'multiboot_dma.bin',
-  'linuxboot.bin',
   'linuxboot_dma.bin',
   'kvmvapic.bin',
   'pvh.bin',
pc-bios/optionrom/Makefile +1 -1
index 1183ef88922..e694c7aac00 100644
@@ -2,7 +2,7 @@ include config.mak
 SRC_DIR := $(TOPSRC_DIR)/pc-bios/optionrom
 VPATH = $(SRC_DIR)
 
-all: multiboot_dma.bin linuxboot.bin linuxboot_dma.bin kvmvapic.bin pvh.bin
+all: multiboot_dma.bin linuxboot_dma.bin kvmvapic.bin pvh.bin
 # Dummy command so that make thinks it has done something
 	@true
 
pc-bios/optionrom/linuxboot.S deleted +0 -195

195 lines not on this page · raw message

Signature · 1 line
2.47.1
Reviewed-by on v4 10/27
#43 lore raw

Re: [PATCH v4 10/27] hw/i386/pc: Remove linuxboot.bin

Quoted 10 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:33 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 10/27] hw/i386/pc: Remove linuxboot.bin
X-Mailer: git-send-email 2.47.1

All PC machines now use the linuxboot_dma.bin binary, we can remove the non-DMA version (linuxboot.bin).

Quoted 7 lines
---
 hw/i386/pc.c                  |   3 +-
 pc-bios/meson.build           |   1 -
 pc-bios/optionrom/Makefile    |   2 +-
 pc-bios/optionrom/linuxboot.S | 195 ----------------------------------
 4 files changed, 2 insertions(+), 199 deletions(-)
 delete mode 100644 pc-bios/optionrom/linuxboot.S
#44 lore raw

Re: [PATCH v4 10/27] hw/i386/pc: Remove linuxboot.bin

In reply to Philippe Mathieu-Daudé
Quoted 7 lines · On Thu, 8 May 2025 15:35:33 +0200 Philippe Mathieu-Daudé <[email protected]>…

All PC machines now use the linuxboot_dma.bin binary, we can remove the non-DMA version (linuxboot.bin).

---
 hw/i386/pc.c                  |   3 +-

linuxboot.bin is referenced in a few more files:

hw/i386/x86-common.c:    option_rom[nb_option_roms].name = "linuxboot.bin";
hw/nvram/fw_cfg.c:    { "genroms/linuxboot.bin", 60 },

are you sure we should keep it there?

Quoted 246 lines
 pc-bios/meson.build           |   1 -
 pc-bios/optionrom/Makefile    |   2 +-
 pc-bios/optionrom/linuxboot.S | 195 ----------------------------------
 4 files changed, 2 insertions(+), 199 deletions(-)
 delete mode 100644 pc-bios/optionrom/linuxboot.S
hw/i386/pc.c +1 -2
index 524d2fd98e8..4e6fe68e2e0 100644
@@ -654,8 +654,7 @@ void xen_load_linux(PCMachineState *pcms)
 
     x86_load_linux(x86ms, fw_cfg, PC_FW_DATA, pcmc->pvh_enabled);
     for (i = 0; i < nb_option_roms; i++) {
-        assert(!strcmp(option_rom[i].name, "linuxboot.bin") ||
-               !strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
+        assert(!strcmp(option_rom[i].name, "linuxboot_dma.bin") ||
                !strcmp(option_rom[i].name, "pvh.bin") ||
                !strcmp(option_rom[i].name, "multiboot_dma.bin"));
         rom_add_option(option_rom[i].name, option_rom[i].bootindex);
pc-bios/meson.build +0 -1
index f2d4dc416a4..39a7fea332e 100644
@@ -63,7 +63,6 @@ blobs = [
   'efi-vmxnet3.rom',
   'qemu-nsis.bmp',
   'multiboot_dma.bin',
-  'linuxboot.bin',
   'linuxboot_dma.bin',
   'kvmvapic.bin',
   'pvh.bin',
pc-bios/optionrom/Makefile +1 -1
index 1183ef88922..e694c7aac00 100644
@@ -2,7 +2,7 @@ include config.mak
 SRC_DIR := $(TOPSRC_DIR)/pc-bios/optionrom
 VPATH = $(SRC_DIR)
 
-all: multiboot_dma.bin linuxboot.bin linuxboot_dma.bin kvmvapic.bin pvh.bin
+all: multiboot_dma.bin linuxboot_dma.bin kvmvapic.bin pvh.bin
 # Dummy command so that make thinks it has done something
 	@true
 
pc-bios/optionrom/linuxboot.S deleted +0 -195
deleted file mode 100644
index ba821ab922d..00000000000
@@ -1,195 +0,0 @@
-/*
- * Linux Boot Option ROM
- *
- * This program is free software; you can redistribute it and/or modify
- * it under the terms of the GNU General Public License as published by
- * the Free Software Foundation; either version 2 of the License, or
- * (at your option) any later version.
- *
- * This program is distributed in the hope that it will be useful,
- * but WITHOUT ANY WARRANTY; without even the implied warranty of
- * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
- * GNU General Public License for more details.
- *
- * You should have received a copy of the GNU General Public License
- * along with this program; if not, see <http://www.gnu.org/licenses/>.
- *
- * Copyright Novell Inc, 2009
- *   Authors: Alexander Graf <[email protected]>
- *
- * Based on code in hw/pc.c.
- */
-
-#include "optionrom.h"
-
-#define BOOT_ROM_PRODUCT "Linux loader"
-
-BOOT_ROM_START
-
-run_linuxboot:
-
-	cli
-	cld
-
-	jmp		copy_kernel
-boot_kernel:
-
-	read_fw		FW_CFG_SETUP_ADDR
-
-	mov		%eax, %ebx
-	shr		$4, %ebx
-
-	/* All segments contain real_addr */
-	mov		%bx, %ds
-	mov		%bx, %es
-	mov		%bx, %fs
-	mov		%bx, %gs
-	mov		%bx, %ss
-
-	/* CX = CS we want to jump to */
-	add		$0x20, %bx
-	mov		%bx, %cx
-
-	/* SP = cmdline_addr-real_addr-16 */
-	read_fw		FW_CFG_CMDLINE_ADDR
-	mov		%eax, %ebx
-	read_fw		FW_CFG_SETUP_ADDR
-	sub		%eax, %ebx
-	sub		$16, %ebx
-	mov		%ebx, %esp
-
-	/* Build indirect lret descriptor */
-	pushw		%cx		/* CS */
-	xor		%ax, %ax
-	pushw		%ax		/* IP = 0 */
-
-	/* Clear registers */
-	xor		%eax, %eax
-	xor		%ebx, %ebx
-	xor		%ecx, %ecx
-	xor		%edx, %edx
-	xor		%edi, %edi
-	xor		%ebp, %ebp
-
-	/* Jump to Linux */
-	lret
-
-
-copy_kernel:
-	/* Read info block in low memory (0x10000 or 0x90000) */
-	read_fw		FW_CFG_SETUP_ADDR
-	shr		$4, %eax
-	mov		%eax, %es
-	xor		%edi, %edi
-	read_fw_blob_addr32_edi(FW_CFG_SETUP)
-
-	cmpw            $0x203, %es:0x206      // if protocol >= 0x203
-	jae             1f                     // have initrd_max
-	movl            $0x37ffffff, %es:0x22c // else assume 0x37ffffff
-1:
-
-	/* Check if using kernel-specified initrd address */
-	read_fw		FW_CFG_INITRD_ADDR
-	mov		%eax, %edi             // (load_kernel wants it in %edi)
-	read_fw		FW_CFG_INITRD_SIZE     // find end of initrd
-	add		%edi, %eax
-	xor		%es:0x22c, %eax        // if it matches es:0x22c
-	and		$-4096, %eax           // (apart from padding for page)
-	jz		load_kernel            // then initrd is not at top
-					       // of memory
-
-	/* pc.c placed the initrd at end of memory.  Compute a better
-	 * initrd address based on e801 data.
-	 */
-	mov		$0xe801, %ax
-	xor		%cx, %cx
-	xor		%dx, %dx
-	int		$0x15
-
-	/* Output could be in AX/BX or CX/DX */
-	or		%cx, %cx
-	jnz		1f
-	or		%dx, %dx
-	jnz		1f
-	mov		%ax, %cx
-	mov		%bx, %dx
-1:
-
-	or		%dx, %dx
-	jnz		2f
-	addw		$1024, %cx            /* add 1 MB */
-	movzwl		%cx, %edi
-	shll		$10, %edi             /* convert to bytes */
-	jmp		3f
-
-2:
-	addw		$16777216 >> 16, %dx  /* add 16 MB */
-	movzwl		%dx, %edi
-	shll		$16, %edi             /* convert to bytes */
-
-3:
-	read_fw         FW_CFG_INITRD_SIZE
-	subl            %eax, %edi
-	andl            $-4096, %edi          /* EDI = start of initrd */
-	movl		%edi, %es:0x218       /* put it in the header */
-
-load_kernel:
-	/* We need to load the kernel into memory we can't access in 16 bit
-	   mode, so let's get into 32 bit mode, write the kernel and jump
-	   back again. */
-
-	/* Reserve space on the stack for our GDT descriptor. */
-	mov             %esp, %ebp
-	sub             $16, %esp
-
-	/* Now create the GDT descriptor */
-	movw		$((3 * 8) - 1), -16(%bp)
-	mov		%cs, %eax
-	movzwl		%ax, %eax
-	shl		$4, %eax
-	addl		$gdt, %eax
-	movl		%eax, -14(%bp)
-
-	/* And load the GDT */
-	data32 lgdt	-16(%bp)
-	mov		%ebp, %esp
-
-	/* Get us to protected mode now */
-	mov		$1, %eax
-	mov		%eax, %cr0
-
-	/* So we can set ES to a 32-bit segment */
-	mov		$0x10, %eax
-	mov		%eax, %es
-
-	/* We're now running in 16-bit CS, but 32-bit ES! */
-
-	/* Load kernel and initrd */
-	read_fw_blob_addr32_edi(FW_CFG_INITRD)
-	read_fw_blob_addr32(FW_CFG_KERNEL)
-	read_fw_blob_addr32(FW_CFG_CMDLINE)
-
-	/* And now jump into Linux! */
-	mov		$0, %eax
-	mov		%eax, %cr0
-
-	/* ES = CS */
-	mov		%cs, %ax
-	mov		%ax, %es
-
-	jmp		boot_kernel
-
-/* Variables */
-
-.align 4, 0
-gdt:
-	/* 0x00 */
-.byte	0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
-
-	/* 0x08: code segment (base=0, limit=0xfffff, type=32bit code exec/read, DPL=0, 4k) */
-.byte	0xff, 0xff, 0x00, 0x00, 0x00, 0x9a, 0xcf, 0x00
-
-	/* 0x10: data segment (base=0, limit=0xfffff, type=32bit data read/write, DPL=0, 4k) */
-.byte	0xff, 0xff, 0x00, 0x00, 0x00, 0x92, 0xcf, 0x00
-
-BOOT_ROM_END
#45 lore raw

Re: [PATCH v4 10/27] hw/i386/pc: Remove linuxboot.bin

Quoted 5 lines · Igor Mammedov

Date: Fri, 9 May 2025 18:04:11 +0200
From: Igor Mammedov <[email protected]>
Subject: Re: [PATCH v4 10/27] hw/i386/pc: Remove linuxboot.bin
X-Mailer: Claws Mail 4.3.1 (GTK 3.24.43; x86_64-redhat-linux-gnu)

Quoted 10 lines
Quoted 7 lines · On Thu, 8 May 2025 15:35:33 +0200 Philippe Mathieu-Daudé <[email protected]>…

All PC machines now use the linuxboot_dma.bin binary, we can remove the non-DMA version (linuxboot.bin).

---
 hw/i386/pc.c                  |   3 +-

(After a long interval, more of a note.)

Quoted 3 lines

linuxboot.bin is referenced in a few more files:

hw/i386/x86-common.c: option_rom[nb_option_roms].name = "linuxboot.bin";

this case is removed in previous patch (which considerred x86 machines all have DMA enabled).

Quoted 1 line

hw/nvram/fw_cfg.c: { "genroms/linuxboot.bin", 60 },

this case is removed in commit 6160ce208419 ("hw/nvram/fw_cfg: Remove legacy FW_CFG_ORDER_OVERRIDE").

Regards,
Zhao

#46 lore raw

PATCH v4 11/27 hw/i386/pc: Remove pc_compat_2_6[] array

In reply to Philippe Mathieu-Daudé

The pc_compat_2_6[] array was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it.

---
 include/hw/i386/pc.h | 3 ---
 hw/i386/pc.c         | 8 --------
 2 files changed, 11 deletions(-)
include/hw/i386/pc.h +0 -3
index a3de3e9560d..4fb2033bc54 100644
@@ -292,9 +292,6 @@ extern const size_t pc_compat_2_8_len;
 extern GlobalProperty pc_compat_2_7[];
 extern const size_t pc_compat_2_7_len;
 
-extern GlobalProperty pc_compat_2_6[];
-extern const size_t pc_compat_2_6_len;
-
 #define DEFINE_PC_MACHINE(suffix, namestr, initfn, optsfn) \
     static void pc_machine_##suffix##_class_init(ObjectClass *oc, \
                                                  const void *data) \
hw/i386/pc.c +0 -8
index 4e6fe68e2e0..65a11ea8f99 100644
@@ -251,14 +251,6 @@ GlobalProperty pc_compat_2_7[] = {
 };
 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
 
-GlobalProperty pc_compat_2_6[] = {
-    { TYPE_X86_CPU, "cpuid-0xb", "off" },
-    { "vmxnet3", "romfile", "" },
-    { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
-    { "apic-common", "legacy-instance-id", "on", }
-};
-const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
-
 /*
  * @PC_FW_DATA:
  * Size of the chunk of memory at the top of RAM for the BIOS ACPI tables
Signature · 1 line
2.47.1
Reviewed-by on v4 11/27
#47 lore raw

Re: [PATCH v4 11/27] hw/i386/pc: Remove pc_compat_2_6[] array

Quoted 8 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:34 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 11/27] hw/i386/pc: Remove pc_compat_2_6[] array
X-Mailer: git-send-email 2.47.1

The pc_compat_2_6[] array was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it.

Quoted 8 lines
---
 include/hw/i386/pc.h | 3 ---
 hw/i386/pc.c         | 8 --------
 2 files changed, 11 deletions(-)
#48 lore raw

Re: [PATCH v4 11/27] hw/i386/pc: Remove pc_compat_2_6[] array

In reply to Philippe Mathieu-Daudé
Quoted 2 lines · On Thu, 8 May 2025 15:35:34 +0200 Philippe Mathieu-Daudé <[email protected]>…

The pc_compat_2_6[] array was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it.

see my comment in 1/27

Quoted 43 lines
---
 include/hw/i386/pc.h | 3 ---
 hw/i386/pc.c         | 8 --------
 2 files changed, 11 deletions(-)
include/hw/i386/pc.h +0 -3
index a3de3e9560d..4fb2033bc54 100644
@@ -292,9 +292,6 @@ extern const size_t pc_compat_2_8_len;
 extern GlobalProperty pc_compat_2_7[];
 extern const size_t pc_compat_2_7_len;
 
-extern GlobalProperty pc_compat_2_6[];
-extern const size_t pc_compat_2_6_len;
-
 #define DEFINE_PC_MACHINE(suffix, namestr, initfn, optsfn) \
     static void pc_machine_##suffix##_class_init(ObjectClass *oc, \
                                                  const void *data) \
hw/i386/pc.c +0 -8
index 4e6fe68e2e0..65a11ea8f99 100644
@@ -251,14 +251,6 @@ GlobalProperty pc_compat_2_7[] = {
 };
 const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
 
-GlobalProperty pc_compat_2_6[] = {
-    { TYPE_X86_CPU, "cpuid-0xb", "off" },
-    { "vmxnet3", "romfile", "" },
-    { TYPE_X86_CPU, "fill-mtrr-mask", "off" },
-    { "apic-common", "legacy-instance-id", "on", }
-};
-const size_t pc_compat_2_6_len = G_N_ELEMENTS(pc_compat_2_6);
-
 /*
  * @PC_FW_DATA:
  * Size of the chunk of memory at the top of RAM for the BIOS ACPI tables
#49 lore raw

PATCH v4 12/27 target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field

In reply to Philippe Mathieu-Daudé

The CPUX86State::enable_cpuid_0xb boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid().

---
 target/i386/cpu.h | 3 ---
 target/i386/cpu.c | 6 ------
 2 files changed, 9 deletions(-)
target/i386/cpu.h +0 -3
index 0db70a70439..06817a31cf9 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
      */
     bool legacy_multi_node;
 
-    /* Compatibility bits for old machine types: */
-    bool enable_cpuid_0xb;
-
     /* Enable auto level-increase for all CPUID leaves */
     bool full_cpuid_auto_level;
 
target/i386/cpu.c +0 -6
index 49179f35812..6fe37f71b1e 100644
@@ -6982,11 +6982,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
         break;
     case 0xB:
         /* Extended Topology Enumeration Leaf */
-        if (!cpu->enable_cpuid_0xb) {
-                *eax = *ebx = *ecx = *edx = 0;
-                break;
-        }
-
         *ecx = count & 0xff;
         *edx = cpu->apic_id;
 
@@ -8828,7 +8823,6 @@ static const Property x86_cpu_properties[] = {
     DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
     DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
     DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
-    DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true),
     DEFINE_PROP_BOOL("x-vendor-cpuid-only", X86CPU, vendor_cpuid_only, true),
     DEFINE_PROP_BOOL("x-amd-topoext-features-only", X86CPU, amd_topoext_features_only, true),
     DEFINE_PROP_BOOL("lmce", X86CPU, enable_lmce, false),
Signature · 1 line
2.47.1
#50 lore raw

Re: [PATCH v4 12/27] target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field

Quoted 42 lines · Philippe Mathieu-Daudé

The CPUX86State::enable_cpuid_0xb boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid().

---
  target/i386/cpu.h | 3 ---
  target/i386/cpu.c | 6 ------
  2 files changed, 9 deletions(-)
target/i386/cpu.h +0 -3
index 0db70a70439..06817a31cf9 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
       */
      bool legacy_multi_node;
 
-    /* Compatibility bits for old machine types: */
-    bool enable_cpuid_0xb;
-
      /* Enable auto level-increase for all CPUID leaves */
      bool full_cpuid_auto_level;
 
target/i386/cpu.c +0 -5
index 49179f35812..6fe37f71b1e 100644
@@ -6982,11 +6982,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
          break;
      case 0xB:
          /* Extended Topology Enumeration Leaf */
-        if (!cpu->enable_cpuid_0xb) {
-                *eax = *ebx = *ecx = *edx = 0;
-                break;
-        }
-
          *ecx = count & 0xff;
          *edx = cpu->apic_id;
 

target/i386/cpu.c static const Property x86_cpu_properties[] = { · line 8831

Quoted 4 lines
target/i386/cpu.c +0 -1
@@ -8828,7 +8823,6 @@ static const Property x86_cpu_properties[] = {
      DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
      DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
      DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
-    DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true),

It's deprecating the "cpuid-0xb" property.

I think we need go with the standard process to deprecate it.

Quoted 3 lines
target/i386/cpu.c +0 -0
      DEFINE_PROP_BOOL("x-vendor-cpuid-only", X86CPU, vendor_cpuid_only, true),
      DEFINE_PROP_BOOL("x-amd-topoext-features-only", X86CPU, amd_topoext_features_only, true),
      DEFINE_PROP_BOOL("lmce", X86CPU, enable_lmce, false),
#51 lore raw

Re: [PATCH v4 12/27] target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field

Quoted 52 lines · Xiaoyao Li

Date: Fri, 9 May 2025 14:49:27 +0800
From: Xiaoyao Li <[email protected]>
Subject: Re: [PATCH v4 12/27] target/i386/cpu: Remove
CPUX86State::enable_cpuid_0xb field

Quoted 39 lines · Philippe Mathieu-Daudé

The CPUX86State::enable_cpuid_0xb boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid().

---
  target/i386/cpu.h | 3 ---
  target/i386/cpu.c | 6 ------
  2 files changed, 9 deletions(-)
target/i386/cpu.h +0 -3
index 0db70a70439..06817a31cf9 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
       */
      bool legacy_multi_node;
-    /* Compatibility bits for old machine types: */
-    bool enable_cpuid_0xb;
-
      /* Enable auto level-increase for all CPUID leaves */
      bool full_cpuid_auto_level;
target/i386/cpu.c +0 -5
index 49179f35812..6fe37f71b1e 100644
@@ -6982,11 +6982,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
          break;
      case 0xB:
          /* Extended Topology Enumeration Leaf */
-        if (!cpu->enable_cpuid_0xb) {
-                *eax = *ebx = *ecx = *edx = 0;
-                break;
-        }
-
          *ecx = count & 0xff;
          *edx = cpu->apic_id;

target/i386/cpu.c static const Property x86_cpu_properties[] = { · line 8831

Quoted 4 lines
target/i386/cpu.c +0 -1
@@ -8828,7 +8823,6 @@ static const Property x86_cpu_properties[] = {
      DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
      DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
      DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
-    DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true),

It's deprecating the "cpuid-0xb" property.

Quoted 1 line

I think we need go with the standard process to deprecate it.

Thanks! I got your point.

Though this property is introduced for compatibility, as its comment said "Compatibility bits for old machine types", it is also useful for somer users.

Fo example, in the early development stages of TDX, when there was no full support for CPU topology, Intel had disable this property for testing and found this bug:

https://lore.kernel.org/qemu-devel/[email protected]/

So, I think there may be other similar use cases as well.

And, if someone wants to emulate ancient x86 CPUs (though I can't currently confirm from which generation of CPUs 0xb support started), he may want to consider disable this property as well.

The main problem here is that the "property" mechanism doesn't distinguish between internal use/public use, and although it was originally intended for internal QEMU use, it also leaks to the user, creating some external use cases.

@Philippe, thank you for cleaning up this case! I think we can keep this property, and if you don't mind, I can modify its comment later to indicate that it's used to adjust the topology support for the CPU.

Thanks,
Zhao

#52 lore raw

How to mark internal properties (was: Re: [PATCH v4 12/27] target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field)

Quoted 53 lines · Zhao Liu
Quoted 52 lines · Xiaoyao Li

Date: Fri, 9 May 2025 14:49:27 +0800
From: Xiaoyao Li <[email protected]>
Subject: Re: [PATCH v4 12/27] target/i386/cpu: Remove
CPUX86State::enable_cpuid_0xb field

Quoted 39 lines · Philippe Mathieu-Daudé

The CPUX86State::enable_cpuid_0xb boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid().

---
   target/i386/cpu.h | 3 ---
   target/i386/cpu.c | 6 ------
   2 files changed, 9 deletions(-)
target/i386/cpu.h +0 -3
index 0db70a70439..06817a31cf9 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
        */
       bool legacy_multi_node;
-    /* Compatibility bits for old machine types: */
-    bool enable_cpuid_0xb;
-
       /* Enable auto level-increase for all CPUID leaves */
       bool full_cpuid_auto_level;
target/i386/cpu.c +0 -5
index 49179f35812..6fe37f71b1e 100644
@@ -6982,11 +6982,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
           break;
       case 0xB:
           /* Extended Topology Enumeration Leaf */
-        if (!cpu->enable_cpuid_0xb) {
-                *eax = *ebx = *ecx = *edx = 0;
-                break;
-        }
-
           *ecx = count & 0xff;
           *edx = cpu->apic_id;

target/i386/cpu.c static const Property x86_cpu_properties[] = { · line 8831

Quoted 4 lines
target/i386/cpu.c +0 -1
@@ -8828,7 +8823,6 @@ static const Property x86_cpu_properties[] = {
       DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
       DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
       DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
-    DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true),

It's deprecating the "cpuid-0xb" property.

Quoted 7 lines
Quoted 1 line

I think we need go with the standard process to deprecate it.

Thanks! I got your point.

Though this property is introduced for compatibility, as its comment said "Compatibility bits for old machine types", it is also useful for somer users.

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

Thomas

Quoted 25 lines

Fo example, in the early development stages of TDX, when there was no full support for CPU topology, Intel had disable this property for testing and found this bug:

https://lore.kernel.org/qemu-devel/[email protected]/

So, I think there may be other similar use cases as well.

And, if someone wants to emulate ancient x86 CPUs (though I can't currently confirm from which generation of CPUs 0xb support started), he may want to consider disable this property as well.

The main problem here is that the "property" mechanism doesn't distinguish between internal use/public use, and although it was originally intended for internal QEMU use, it also leaks to the user, creating some external use cases.

@Philippe, thank you for cleaning up this case! I think we can keep this property, and if you don't mind, I can modify its comment later to indicate that it's used to adjust the topology support for the CPU.

Thanks,
Zhao

#53 lore raw

Re: How to mark internal properties (was: Re: [PATCH v4 12/27] target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field)

Quoted 67 lines · Thomas Huth

Date: Fri, 9 May 2025 12:04:19 +0200
From: Thomas Huth <[email protected]>
Subject: How to mark internal properties (was: Re: [PATCH v4 12/27]
target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field)

Quoted 53 lines · Zhao Liu
Quoted 52 lines · Xiaoyao Li

Date: Fri, 9 May 2025 14:49:27 +0800
From: Xiaoyao Li <[email protected]>
Subject: Re: [PATCH v4 12/27] target/i386/cpu: Remove
CPUX86State::enable_cpuid_0xb field

Quoted 39 lines · Philippe Mathieu-Daudé

The CPUX86State::enable_cpuid_0xb boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid().

---
   target/i386/cpu.h | 3 ---
   target/i386/cpu.c | 6 ------
   2 files changed, 9 deletions(-)
target/i386/cpu.h +0 -3
index 0db70a70439..06817a31cf9 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
        */
       bool legacy_multi_node;
-    /* Compatibility bits for old machine types: */
-    bool enable_cpuid_0xb;
-
       /* Enable auto level-increase for all CPUID leaves */
       bool full_cpuid_auto_level;
target/i386/cpu.c +0 -5
index 49179f35812..6fe37f71b1e 100644
@@ -6982,11 +6982,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
           break;
       case 0xB:
           /* Extended Topology Enumeration Leaf */
-        if (!cpu->enable_cpuid_0xb) {
-                *eax = *ebx = *ecx = *edx = 0;
-                break;
-        }
-
           *ecx = count & 0xff;
           *edx = cpu->apic_id;

target/i386/cpu.c static const Property x86_cpu_properties[] = { · line 8831

Quoted 4 lines
target/i386/cpu.c +0 -1
@@ -8828,7 +8823,6 @@ static const Property x86_cpu_properties[] = {
       DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
       DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
       DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
-    DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true),

It's deprecating the "cpuid-0xb" property.

Quoted 7 lines
Quoted 1 line

I think we need go with the standard process to deprecate it.

Thanks! I got your point.

Though this property is introduced for compatibility, as its comment said "Compatibility bits for old machine types", it is also useful for somer users.

Quoted 6 lines

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

Hi Thomas, thank you.

AFAIK, there are two ways to configure whether an object/device is allowed to be created by user or not:

* TYPE_USER_CREATABLE
* DeviceClass: user_creatable

So, it looks like it would be tricky to change the infrastructure around object_property_add because it's not easy to be compatible with both of the above user creation ways.

Quoted 6 lines

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

Therefore, I think the “q-” prefix might be a good way, simple and effective.

Let's see if any other maintainers have a better idea.

Regards,
Zhao

Reviewed-by Juan Quintela Reviewed-by John Snow
#54 lore raw

Re: How to mark internal properties

In reply to Thomas Huth
Quoted 62 lines · Thomas Huth
Quoted 53 lines · Zhao Liu
Quoted 52 lines · Xiaoyao Li

Date: Fri, 9 May 2025 14:49:27 +0800
From: Xiaoyao Li <[email protected]>
Subject: Re: [PATCH v4 12/27] target/i386/cpu: Remove
CPUX86State::enable_cpuid_0xb field

Quoted 39 lines · Philippe Mathieu-Daudé

The CPUX86State::enable_cpuid_0xb boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid().

---
target/i386/cpu.h | 3 ---
target/i386/cpu.c | 6 ------

2 files changed, 9 deletions(-)
target/i386/cpu.h +0 -3
index 0db70a70439..06817a31cf9 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
      */
     bool legacy_multi_node;
-    /* Compatibility bits for old machine types: */
-    bool enable_cpuid_0xb;
-
     /* Enable auto level-increase for all CPUID leaves */
     bool full_cpuid_auto_level;
target/i386/cpu.c +0 -5
index 49179f35812..6fe37f71b1e 100644
@@ -6982,11 +6982,6 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
         break;
     case 0xB:
           /* Extended Topology Enumeration Leaf */
-        if (!cpu->enable_cpuid_0xb) {
-            *eax = *ebx = *ecx = *edx = 0;
-            break;
-        }
-
         *ecx = count & 0xff;
         *edx = cpu->apic_id;

target/i386/cpu.c static const Property x86_cpu_properties[] = { · line 8831

Quoted 4 lines
target/i386/cpu.c +0 -1
@@ -8828,7 +8823,6 @@ static const Property x86_cpu_properties[] = {
     DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
     DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
     DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
-    DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true),

It's deprecating the "cpuid-0xb" property.

Quoted 7 lines
Quoted 1 line

I think we need go with the standard process to deprecate it.

Thanks! I got your point.

Though this property is introduced for compatibility, as its comment said "Compatibility bits for old machine types", it is also useful for somer users.

Quoted 6 lines

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

Correct.

Exposing properties meant for internal use at the external interface inevitably leads to (uncertainty about) external use.

Quoted 6 lines

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

This papers over our inability / unwillingness to isolate the external interface from internal detail.

The proper solution is to make the internal properties inaccessible at the external interface. This requires declaring properties' intent. Which strikes me as a very good idea.

A naming convention is a simple, stupid way to do that. There are drawbacks, as experience with the "x-" prefix has shown:

* Flipping a flag bit involves changing the name. Tolerable when all uses are internal, compatibility break when not. Not a problem when the bit governs external access, of course.

* Name capture: consider InputBarrier properties x-origin, y-origin. Oops.

* If we have multiple flag bits, their prefixes can accumulate. This gets ugly and confusing real quick. Not an issue when at most one of the flags can be set, as is the case for "unstable" and "internal use".

* QAPI reserves "q_" for the generator's use. Since "q-" would get mapped to "q_" in C, we risk name clashes.

For what it's worth, QAPI abandoned the "x-" naming convention (commit a3c45b3e629 (qapi: New special feature flag "unstable"), commit message appended for your convenience). Developers are free to use "x-" to help guide human users, but the feature flag is the sole source of thruth.

[...]

commit a3c45b3e62962f99338716b1347cfb0d427cea44

Author: Markus Armbruster <[email protected]>
Date:   Thu Oct 28 12:25:12 2021 +0200

qapi: New special feature flag "unstable"

By convention, names starting with "x-" are experimental. The parts of external interfaces so named may be withdrawn or changed incompatibly in future releases.

The naming convention makes unstable interfaces easy to recognize. Promoting something from experimental to stable involves a name change. Client code needs to be updated. Occasionally bothersome.

Worse, the convention is not universally observed:

* QOM type "input-barrier" has properties "x-origin", "y-origin". Looks accidental, but it's ABI since 4.2.

* QOM types "memory-backend-file", "memory-backend-memfd", "memory-backend-ram", and "memory-backend-epc" have a property "x-use-canonical-path-for-ramblock-id" that is documented to be stable despite its name.

We could document these exceptions, but documentation helps only humans. We want to recognize "unstable" in code, like "deprecated".

So support recognizing it the same way: introduce new special feature flag "unstable". It will be treated specially by the QAPI generator, like the existing feature flag "deprecated", and unlike regular feature flags.

This commit updates documentation and prepares tests. The next commit updates the QAPI schema. The remaining patches update the QAPI generator and wire up -compat policy checking.

Management applications can then use query-qmp-schema and -compat to manage or guard against use of unstable interfaces the same way as for deprecated interfaces.

docs/devel/qapi-code-gen.txt no longer mandates the naming convention. Using it anyway might help writers of programs that aren't full-fledged management applications. Not using it can save us bothersome renames. We'll see how that shakes out.

Message-Id: <[email protected]>

#55 lore raw

Re: How to mark internal properties (was: Re: [PATCH v4 12/27] target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field)

In reply to Thomas Huth
Quoted 7 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

Quoted 6 lines

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

thanks
-- PMM

#56 lore raw

Re: How to mark internal properties (was: Re: [PATCH v4 12/27] target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field)

Quoted 15 lines · Peter Maydell
Quoted 7 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

Quoted 6 lines

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

Quoted 5 lines

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

I think it is also not obvious enough that a 'q-' prefix means private.

Perhaps borrow from the C world and declare that a leading underscore indicates a private property. People are more likely to understand and remember that, than 'q-'.

With regards,
Daniel

Signature · 3 lines
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
#57 lore raw

Re: How to mark internal properties

Quoted 16 lines · Daniel P. Berrangé
Quoted 15 lines · Peter Maydell
Quoted 13 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

Quoted 5 lines
Quoted 5 lines

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

If we make intent explicit with DEFINE_INTERNAL_PROP_FOO(), is repeating intent in the name useful?

Quoted 1 line

I think it is also not obvious enough that a 'q-' prefix means private.

Concur.

Quoted 3 lines

Perhaps borrow from the C world and declare that a leading underscore indicates a private property. People are more likely to understand and remember that, than 'q-'.

This is fine for device properties now. It's not fine for properties of user-creatable objects, because these are defined in QAPI, and QAPI prohibits names starting with a single underscore. I append relevant parts of docs/devel/qapi-code-gen.rst for your convenience.

Why does QAPI prohibit leading underscores? Chiefly because such names are reserved identifiers in C. Instead of complicating the mapping from QAPI name to C identifier, we restrict QAPI names and call it a day.

The mapping between device property name and C identifiers is entirely manual. When a property is backed by a member of the device state struct, naming the member exactly like the property makes sense. Having to mentally strip / insert a leading underscore would hardly be terrible, just a bit of friction. I'd prefer not to.

Naming rules and reserved names
-------------------------------

All names must begin with a letter, and contain only ASCII letters, digits, hyphen, and underscore. There are two exceptions: enum values may start with a digit, and names that are downstream extensions (see section Downstream extensions_) start with underscore.

Names beginning with ``q_`` are reserved for the generator, which uses them for munging QMP names that resemble C keywords or other problematic strings. For example, a member named ``default`` in qapi becomes ``q_default`` in the generated C code.

[...]

Downstream extensions
---------------------

QAPI schema names that are externally visible, say in the Client JSON Protocol, need to be managed with care. Names starting with a downstream prefix of the form __RFQDN_ are reserved for the downstream who controls the valid, reverse fully qualified domain name RFQDN. RFQDN may only contain ASCII letters, digits, hyphen and period.

Example: Red Hat, Inc. controls redhat.com, and may therefore add a downstream command ``__com.redhat_drive-mirror``.

#58 lore raw

Re: How to mark internal properties

Quoted 17 lines · Markus Armbruster
Quoted 15 lines · Daniel P. Berrangé
Quoted 14 lines · Peter Maydell
Quoted 13 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

Quoted 8 lines
Quoted 5 lines
Quoted 5 lines

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

If we make intent explicit with DEFINE_INTERNAL_PROP_FOO(), is repeating intent in the name useful?

+1 for DEFINE_INTERNAL_PROP_FOO(). I have the same thought.

We need something in code to restrict the *internal* property really internal, i.e., not user settable. What the name of the property is doesn't matter.

#59 lore raw

Re: How to mark internal properties

Quoted 23 lines · Xiaoyao Li
Quoted 21 lines · Markus Armbruster
Quoted 20 lines · Daniel P. Berrangé
Quoted 19 lines · Peter Maydell
Quoted 7 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

Quoted 6 lines

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

Quoted 6 lines
Quoted 2 lines

If we make intent explicit with DEFINE_INTERNAL_PROP_FOO(), is repeating intent in the name useful?

+1 for DEFINE_INTERNAL_PROP_FOO(). I have the same thought.

We need something in code to restrict the *internal* property really internal, i.e., not user settable. What the name of the property is doesn't matter.

What's an internal property? Properties are there to make some field of an object introspectable and settable from command line and QEMU monitor or other external interfaces. If that's not needed for something why is it defined as a property in the first place and not just e.g. C accessor functions as part of the device's interface instead? I think this may be overusing QOM for things that may not need it and adding complexity where not needed. It reminds me of patches that wanted to export via-ide IRQs or ISA IRQs just to be able to connect them to other parts _of the same chip_ becuase this chip is modeled as multiple QOM objects for reusing code from those. But in reality the chip does not have such pins and these are internal connections so I think it would be better to model these as functions and not QOM constructs that the user can change. In general, if the device or object has an external connection or a knob that the user may need to change or connect to another device (like building a board from parts you can wire pins together) then those need properties or qemu_irqs but other "internal properties" may need some other way to access and often simple accessor functions are enough for this as these internal properties are only accessed form the code. That way we would not need even more complexity to hide these from the user, instead of that just don't expose them but use something else where a property is not needed. A property is just like an accessor function with additional complexity to expose it to other interfaces so it's externally settable and introspectable but we don't need those for internal properties so we can drop that complexity and get back to the accessor function at the bottom of it.

Regards,
BALATON Zoltan

#60 lore raw

Re: How to mark internal properties

Quoted 1 line · BALATON Zoltan

On Mon, 12 May 2025, Xiaoyao Li wrote:

...

Quoted 3 lines
Quoted 1 line

We need something in code to restrict the *internal* property really internal, i.e., not user settable. What the name of the property is doesn't matter.

What's an internal property? Properties are there to make some field of an object introspectable and settable from command line and QEMU monitor or other external interfaces. If that's not needed for something why is it defined as a property in the first place and not just e.g. C accessor functions as part of the device's interface instead? I think this may be overusing QOM for things that may not need it and adding complexity where not needed.

Maybe some things could easily be simplified indeed, but for some others, it's currently the way it's deeply rooted in the logic of QEMU. Have a look at the hw_compat arrays in hw/core/machine.c ... it all goes via properties, so there is certainly no easy and quick solution for this.

Thomas

#61 lore raw

Re: How to mark internal properties

In reply to Markus Armbruster
Quoted 31 lines · Markus Armbruster
Quoted 16 lines · Daniel P. Berrangé
Quoted 15 lines · Peter Maydell
Quoted 13 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

Quoted 5 lines
Quoted 5 lines

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

If we make intent explicit with DEFINE_INTERNAL_PROP_FOO(), is repeating intent in the name useful?

Quoted 1 line

I think it is also not obvious enough that a 'q-' prefix means private.

Concur.

Quoted 8 lines
Quoted 3 lines

Perhaps borrow from the C world and declare that a leading underscore indicates a private property. People are more likely to understand and remember that, than 'q-'.

This is fine for device properties now. It's not fine for properties of user-creatable objects, because these are defined in QAPI, and QAPI prohibits names starting with a single underscore. I append relevant parts of docs/devel/qapi-code-gen.rst for your convenience.

On a related note this also brings us back to the discussion as to the relationship between qdev and QOM: at one point I was under the impression that qdev properties were simply QOM properties that were exposed externally, i.e on the commmand line for use with -device.

Can you provide an update on what the current thinking is in this area, in particular re: scoping of qdev vs QOM properties?

ATB,

Mark.

#62 lore raw

Re: How to mark internal properties

Quoted 3 lines · Mark Cave-Ayland

On a related note this also brings us back to the discussion as to the relationship between qdev and QOM: at one point I was under the impression that qdev properties were simply QOM properties that were exposed externally, i.e on the commmand line for use with -device.

Can you provide an update on what the current thinking is in this area, in particular re: scoping of qdev vs QOM properties?

qdev is a leaky layer above QOM.

qdev properties are also QOM properties.

All device properties are exposed externally.

We use device properties for:

* Letting users configure pluggable devices, with -device or device_add

* Letting code configure onboard devices

For onboard devices that are also pluggable, everything exposed to code is also exposed externally. This might be a mistake in places.

* Letting the machine versioning machinery adjust device configuration

Some properties are meant to be used just for this. They're exposed externally regardless, which is a mistake.

#63 lore raw

Re: How to mark internal properties

Quoted 10 lines · Markus Armbruster
Quoted 3 lines · Mark Cave-Ayland

On a related note this also brings us back to the discussion as to the relationship between qdev and QOM: at one point I was under the impression that qdev properties were simply QOM properties that were exposed externally, i.e on the commmand line for use with -device.

Can you provide an update on what the current thinking is in this area, in particular re: scoping of qdev vs QOM properties?

qdev is a leaky layer above QOM.

qdev properties are also QOM properties.

All device properties are exposed externally.

That was clear but the question was if QOM properties (that are not qdev properties) exist and if so are they also exposed? If not exposed it may be used for internal properties (where simpler solutions aren't convenient) but maybe qdev also adds easier definition of properties that's why they used instead of QOM properties?

Quoted 8 lines

We use device properties for:

* Letting users configure pluggable devices, with -device or device_add

* Letting code configure onboard devices

For onboard devices that are also pluggable, everything exposed to code is also exposed externally. This might be a mistake in places.

If a device is pluggable, theoretically a user could create a machine from them declaratively, e.g. starting from a "none" machine or like plugging cards in a motherboard so their properties should be exposed as long as those properties correspond to the device pins they model or configurable options. Only properties that are implementation details should not be exposed because setting them can break the device. There are a few places where we have such properties. But you say "in places" so I think you meant the same.

Quoted 4 lines

* Letting the machine versioning machinery adjust device configuration

Some properties are meant to be used just for this. They're exposed externally regardless, which is a mistake.

Question is if we want to allow users to tweak these compatibility options, like selectively enable/disable when migrating between QEMU versions or for testing. It might have some uses and maybe that's the reason why people would like these to go through deprecation instead of just dropping them. Marking some properties not exposed would get the same resistance then so may not solve the issue.

Regards,
BALATON Zoltan

#64 lore raw

Re: How to mark internal properties

Quoted 14 lines · BALATON Zoltan
Quoted 13 lines · Markus Armbruster
Quoted 7 lines · Mark Cave-Ayland

On a related note this also brings us back to the discussion as to the relationship between qdev and QOM: at one point I was under the impression that qdev properties were simply QOM properties that were exposed externally, i.e on the commmand line for use with -device.

Can you provide an update on what the current thinking is in this area, in particular re: scoping of qdev vs QOM properties?

qdev is a leaky layer above QOM.

qdev properties are also QOM properties.

Quoted 7 lines
Quoted 1 line

All device properties are exposed externally.

That was clear but the question was if QOM properties (that are not qdev properties) exist and if so are they also exposed? If not exposed it may be used for internal properties (where simpler solutions aren't convenient) but maybe qdev also adds easier definition of properties that's why they used instead of QOM properties?

NB, not everything we expose is a QDev. We directly expose QOM via "-object" if they implement the 'UserCreatable' interface. If we want internal properties, that should likely be wired in to the QOM level directly.

Conceptually you could even say that everything QEMU creates should live under -object, and no other args ought to need to exist. We have decades of evolved code making that impractical though.

With regards,
Daniel

Signature · 3 lines
|: https://berrange.com      -o-    https://www.flickr.com/photos/dberrange :|
|: https://libvirt.org         -o-            https://fstop138.berrange.com :|
|: https://entangle-photo.org    -o-    https://www.instagram.com/dberrange :|
#65 lore raw

Re: How to mark internal properties

Quoted 23 lines · Daniel P. Berrangé
Quoted 14 lines · BALATON Zoltan
Quoted 13 lines · Markus Armbruster
Quoted 7 lines · Mark Cave-Ayland

On a related note this also brings us back to the discussion as to the relationship between qdev and QOM: at one point I was under the impression that qdev properties were simply QOM properties that were exposed externally, i.e on the commmand line for use with -device.

Can you provide an update on what the current thinking is in this area, in particular re: scoping of qdev vs QOM properties?

qdev is a leaky layer above QOM.

qdev properties are also QOM properties.

Quoted 7 lines
Quoted 1 line

All device properties are exposed externally.

That was clear but the question was if QOM properties (that are not qdev properties) exist and if so are they also exposed? If not exposed it may be used for internal properties (where simpler solutions aren't convenient) but maybe qdev also adds easier definition of properties that's why they used instead of QOM properties?

Quoted 4 lines

NB, not everything we expose is a QDev. We directly expose QOM via "-object" if they implement the 'UserCreatable' interface. If we want internal properties, that should likely be wired in to the QOM level directly.

Yes.

As is, all properties of QOM types implementing UserCreatable are part of the -object / object_add stable external interface. We lack means to add properties just for internal use.

Properties of all QOM types (UserCreatable or not) are exposed externally via qom-get & friends. This isn't a stable interface except when it is, but that's a separate problem.

Quoted 3 lines

Conceptually you could even say that everything QEMU creates should live under -object, and no other args ought to need to exist. We have decades of evolved code making that impractical though.

I doubt we would've invented qdev had QOM existed already.

#66 lore raw

Re: How to mark internal properties

In reply to BALATON Zoltan
Quoted 16 lines · BALATON Zoltan
Quoted 10 lines · Markus Armbruster
Quoted 3 lines · Mark Cave-Ayland

On a related note this also brings us back to the discussion as to the relationship between qdev and QOM: at one point I was under the impression that qdev properties were simply QOM properties that were exposed externally, i.e on the commmand line for use with -device.

Can you provide an update on what the current thinking is in this area, in particular re: scoping of qdev vs QOM properties?

qdev is a leaky layer above QOM.

qdev properties are also QOM properties.

All device properties are exposed externally.

That was clear but the question was if QOM properties (that are not qdev properties) exist and if so are they also exposed? If not exposed it may be used for internal properties (where simpler solutions aren't convenient) but maybe qdev also adds easier definition of properties that's why they used instead of QOM properties?

Quoted 2 lines

We use device properties for:

Quoted 8 lines
Quoted 6 lines

* Letting users configure pluggable devices, with -device or device_add

* Letting code configure onboard devices

For onboard devices that are also pluggable, everything exposed to code is also exposed externally. This might be a mistake in places.

If a device is pluggable, theoretically a user could create a machine from them declaratively, e.g. starting from a "none" machine or like plugging cards in a motherboard so their properties should be exposed as long as those properties correspond to the device pins they model or configurable options. Only properties that are implementation details should not be exposed because setting them can break the device. There are a few places where we have such properties. But you say "in places" so I think you meant the same.

Building machines declaratively has been the dream for many years.

I chose to write "might be in places", because I can't point to examples offhand to support a stronger claim :)

External interfaces should be intentional, i.e. carefully curated to serve actual use cases. They should also be stable and documented.

The QOM parts of our external interfaces are largely accidental, unstable, and undocumented. We have some internal need, we create something to serve it, and expose it externally simply because we lack the means not to.

Quoted 6 lines
Quoted 4 lines

* Letting the machine versioning machinery adjust device configuration

Some properties are meant to be used just for this. They're exposed externally regardless, which is a mistake.

Question is if we want to allow users to tweak these compatibility options, like selectively enable/disable when migrating between QEMU versions or for testing. It might have some uses and maybe that's the reason why people would like these to go through deprecation instead of just dropping them. Marking some properties not exposed would get the same resistance then so may not solve the issue.

If you need to mess with properties to make migration work, that's a bug. That's versioning machinery's job.

External interfaces just for testing can be okay. We should not promise stability there. We should still be intentional, and we should still document.

Attempts to get rid of external interfaces always draw resistance, even when they're accidental.

#67 lore raw

Re: How to mark internal properties

In reply to BALATON Zoltan
Quoted 6 lines · BALATON Zoltan
Quoted 5 lines · Markus Armbruster
Quoted 3 lines · Mark Cave-Ayland

On a related note this also brings us back to the discussion as to the relationship between qdev and QOM: at one point I was under the impression that qdev properties were simply QOM properties that were exposed externally, i.e on the commmand line for use with -device.

Can you provide an update on what the current thinking is in this area, in particular re: scoping of qdev vs QOM properties?

Quoted 7 lines
Quoted 5 lines

qdev is a leaky layer above QOM.

qdev properties are also QOM properties.

All device properties are exposed externally.

That was clear but the question was if QOM properties (that are not qdev properties) exist and if so are they also exposed? If not exposed it may be used for internal properties (where simpler solutions aren't convenient) but maybe qdev also adds easier definition of properties that's why they used instead of QOM properties?

I'm afraid we haven't answered your question clearly, yet. The answer is yes, QOM properties that are not qdev properties do exist. The first one that comes along: ich9_pm_add_properties() adds a bunch to device ICH9-LPC. They are exposted externally. To see them, try

    $ qemu-system-x86_64 -device ICH9-LPC,help

[...]

#68 lore raw

Re: How to mark internal properties

In reply to Markus Armbruster
Quoted 18 lines · On Mon, 12 May 2025 12:54:26 +0200 Markus Armbruster <[email protected]> wrote:
Quoted 16 lines · Daniel P. Berrangé
Quoted 15 lines · Peter Maydell
Quoted 13 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

Quoted 8 lines
Quoted 5 lines
Quoted 5 lines

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

If we make intent explicit with DEFINE_INTERNAL_PROP_FOO(), is repeating intent in the name useful?

While we are inventing a new API, I'd say that _INTERNAL_ is not the only thing on my wish-list wrt properties.
It would be also nice to know when a property is set by internal or external user or if it still has default value.

Basically we are looking at different flags for properties and INERNAL being one of them.

Maybe instead of specialized macro, we should have a more generic
DEFINE_PROP_WITH_FLAGS_FOO(...,flags)
So we won't have to rewrite it again when we think of another flag to turn on/off.

From previous uses of x- flag, some of such properties are created as temporary | developer-only and occasionally as a crutch (still no intended for end user). But then sometimes such properties get promoted to ABI with fat warnings not to touch them. Having stable|unstable flag could help here without
need to rename property (and prevent breaking users who (ab)used it if we care).

#69 lore raw

Re: How to mark internal properties

Quoted 27 lines · Igor Mammedov
Quoted 24 lines · On Mon, 12 May 2025 12:54:26 +0200 Markus Armbruster <[email protected]> wrote:
Quoted 21 lines · Daniel P. Berrangé
Quoted 20 lines · Peter Maydell
Quoted 7 lines · Thomas Huth

Thanks for your clarifications, Zhao! But I think this shows again the problem that we have hit a couple of times in the past already: Properties are currently used for both, config knobs for the users and internal switches for configuration of the machine. We lack a proper way to say "this property is usable for the user" and "this property is meant for internal configuration only".

Quoted 6 lines

I wonder whether we could maybe come up with a naming scheme to better distinguish the two sets, e.g. by using a prefix similar to the "x-" prefix for experimental properties? We could e.g. say that all properties starting with a "q-" are meant for QEMU-internal configuration only or something similar (and maybe even hide those from the default help output when running "-device xyz,help" ?)? Anybody any opinions or better ideas on this?

I think a q-prefix is potentially a bit clunky unless we also have infrastructure to say eg DEFINE_INTERNAL_PROP_BOOL("foo", ...) and have it auto-add the prefix, and to have the C APIs for setting properties search for both "foo" and "q-foo" so you don't have to write qdev_prop_set_bit(dev, "q-foo", ...).

Quoted 7 lines
Quoted 2 lines

If we make intent explicit with DEFINE_INTERNAL_PROP_FOO(), is repeating intent in the name useful?

While we are inventing a new API, I'd say that _INTERNAL_ is not the only thing on my wish-list wrt properties.
It would be also nice to know when a property is set by internal or external user or if it still has default value.

We commonly assume "value is default implies user didn't touch it", which is of course wrong unless it's a value the user cannot set.

Quoted 8 lines

Basically we are looking at different flags for properties and INERNAL being one of them.

Maybe instead of specialized macro, we should have a more generic
DEFINE_PROP_WITH_FLAGS_FOO(...,flags)
So we won't have to rewrite it again when we think of another flag to turn on/off.

Quoted 5 lines

From previous uses of x- flag, some of such properties are created as temporary | developer-only and occasionally as a crutch (still no intended for end user). But then sometimes such properties get promoted to ABI with fat warnings not to touch them. Having stable|unstable flag could help here without
need to rename property (and prevent breaking users who (ab)used it if we care).

I think QAPI's explicit 'unstable' and 'deprecated' flags have been a success. Their meaning is clear, and they come with documentation (because adding them without won't compile).

Pushing QOM closer to QAPI when we can makes sense to me.

#70 lore raw

Re: [PATCH v4 12/27] target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field

In reply to Zhao Liu
Quoted 13 lines · On Fri, 9 May 2025 15:32:30 +0800 Zhao Liu <[email protected]> wrote:
Quoted 12 lines · Xiaoyao Li

Date: Fri, 9 May 2025 14:49:27 +0800
From: Xiaoyao Li <[email protected]>
Subject: Re: [PATCH v4 12/27] target/i386/cpu: Remove
CPUX86State::enable_cpuid_0xb field

Quoted 6 lines · Philippe Mathieu-Daudé

The CPUX86State::enable_cpuid_0xb boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid().

...

Quoted 5 lines
Quoted 5 lines
Quoted 5 lines
target/i386/cpu.c +0 -1
@@ -8828,7 +8823,6 @@ static const Property x86_cpu_properties[] = {
      DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
      DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
      DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
-    DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true),  

...

Quoted 3 lines

@Philippe, thank you for cleaning up this case! I think we can keep this property, and if you don't mind, I can modify its comment later to indicate that it's used to adjust the topology support for the CPU.

+1, we should not delete this without due process (aka deprecation). So perhaps deprecate now and remove in couple of releases

#71 lore raw

PATCH v4 13/27 target/i386/cpu: Remove CPUX86State::fill_mtrr_mask field

In reply to Philippe Mathieu-Daudé

The CPUX86State::fill_mtrr_mask boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify kvm_get_msrs().

---
 target/i386/cpu.h     |  3 ---
 target/i386/cpu.c     |  1 -
 target/i386/kvm/kvm.c | 10 +++-------
 3 files changed, 3 insertions(+), 11 deletions(-)
target/i386/cpu.h +0 -3
index 06817a31cf9..7585407da54 100644
@@ -2253,9 +2253,6 @@ struct ArchCPU {
     /* Enable auto level-increase for Intel Processor Trace leave */
     bool intel_pt_auto_level;
 
-    /* if true fill the top bits of the MTRR_PHYSMASKn variable range */
-    bool fill_mtrr_mask;
-
     /* if true override the phys_bits value with a value read from the host */
     bool host_phys_bits;
 
target/i386/cpu.c +0 -1
index 6fe37f71b1e..fb505d13122 100644
@@ -8810,7 +8810,6 @@ static const Property x86_cpu_properties[] = {
     DEFINE_PROP_UINT32("guest-phys-bits", X86CPU, guest_phys_bits, -1),
     DEFINE_PROP_BOOL("host-phys-bits", X86CPU, host_phys_bits, false),
     DEFINE_PROP_UINT8("host-phys-bits-limit", X86CPU, host_phys_bits_limit, 0),
-    DEFINE_PROP_BOOL("fill-mtrr-mask", X86CPU, fill_mtrr_mask, true),
     DEFINE_PROP_UINT32("level-func7", X86CPU, env.cpuid_level_func7,
                        UINT32_MAX),
     DEFINE_PROP_UINT32("level", X86CPU, env.cpuid_level, UINT32_MAX),
target/i386/kvm/kvm.c +3 -7
index c9a3c02e3e3..87edce99e85 100644
@@ -4635,13 +4635,9 @@ static int kvm_get_msrs(X86CPU *cpu)
      * we're migrating to.
      */
 
-    if (cpu->fill_mtrr_mask) {
-        QEMU_BUILD_BUG_ON(TARGET_PHYS_ADDR_SPACE_BITS > 52);
-        assert(cpu->phys_bits <= TARGET_PHYS_ADDR_SPACE_BITS);
-        mtrr_top_bits = MAKE_64BIT_MASK(cpu->phys_bits, 52 - cpu->phys_bits);
-    } else {
-        mtrr_top_bits = 0;
-    }
+    QEMU_BUILD_BUG_ON(TARGET_PHYS_ADDR_SPACE_BITS > 52);
+    assert(cpu->phys_bits <= TARGET_PHYS_ADDR_SPACE_BITS);
+    mtrr_top_bits = MAKE_64BIT_MASK(cpu->phys_bits, 52 - cpu->phys_bits);
 
     for (i = 0; i < ret; i++) {
         uint32_t index = msrs[i].index;
Signature · 1 line
2.47.1
Reviewed-by on v4 13/27
#72 lore raw

Re: [PATCH v4 13/27] target/i386/cpu: Remove CPUX86State::fill_mtrr_mask field

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:36 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 13/27] target/i386/cpu: Remove
CPUX86State::fill_mtrr_mask field
X-Mailer: git-send-email 2.47.1

The CPUX86State::fill_mtrr_mask boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify kvm_get_msrs().

Quoted 6 lines
---
 target/i386/cpu.h     |  3 ---
 target/i386/cpu.c     |  1 -
 target/i386/kvm/kvm.c | 10 +++-------
 3 files changed, 3 insertions(+), 11 deletions(-)

No case found for external user use, so,

#73 lore raw

Re: [PATCH v4 13/27] target/i386/cpu: Remove CPUX86State::fill_mtrr_mask field

In reply to Philippe Mathieu-Daudé
Quoted 31 lines · On Thu, 8 May 2025 15:35:36 +0200 Philippe Mathieu-Daudé <[email protected]>…

The CPUX86State::fill_mtrr_mask boolean was only disabled for the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Being now always %true, we can remove it and simplify kvm_get_msrs().

---
 target/i386/cpu.h     |  3 ---
 target/i386/cpu.c     |  1 -
 target/i386/kvm/kvm.c | 10 +++-------
 3 files changed, 3 insertions(+), 11 deletions(-)
target/i386/cpu.h +0 -3
index 06817a31cf9..7585407da54 100644
@@ -2253,9 +2253,6 @@ struct ArchCPU {
     /* Enable auto level-increase for Intel Processor Trace leave */
     bool intel_pt_auto_level;
 
-    /* if true fill the top bits of the MTRR_PHYSMASKn variable range */
-    bool fill_mtrr_mask;
-
     /* if true override the phys_bits value with a value read from the host */
     bool host_phys_bits;
 
target/i386/cpu.c +0 -0
index 6fe37f71b1e..fb505d13122 100644

target/i386/cpu.c static const Property x86_cpu_properties[] = { · line 8813

Quoted 4 lines
target/i386/cpu.c +0 -1
@@ -8810,7 +8810,6 @@ static const Property x86_cpu_properties[] = {
     DEFINE_PROP_UINT32("guest-phys-bits", X86CPU, guest_phys_bits, -1),
     DEFINE_PROP_BOOL("host-phys-bits", X86CPU, host_phys_bits, false),
     DEFINE_PROP_UINT8("host-phys-bits-limit", X86CPU, host_phys_bits_limit, 0),
-    DEFINE_PROP_BOOL("fill-mtrr-mask", X86CPU, fill_mtrr_mask, true),

same as previous patch, deprecate 1st then remove

Quoted 24 lines
target/i386/cpu.c +0 -0
     DEFINE_PROP_UINT32("level-func7", X86CPU, env.cpuid_level_func7,
                        UINT32_MAX),
     DEFINE_PROP_UINT32("level", X86CPU, env.cpuid_level, UINT32_MAX),
target/i386/kvm/kvm.c +3 -7
index c9a3c02e3e3..87edce99e85 100644
@@ -4635,13 +4635,9 @@ static int kvm_get_msrs(X86CPU *cpu)
      * we're migrating to.
      */
 
-    if (cpu->fill_mtrr_mask) {
-        QEMU_BUILD_BUG_ON(TARGET_PHYS_ADDR_SPACE_BITS > 52);
-        assert(cpu->phys_bits <= TARGET_PHYS_ADDR_SPACE_BITS);
-        mtrr_top_bits = MAKE_64BIT_MASK(cpu->phys_bits, 52 - cpu->phys_bits);
-    } else {
-        mtrr_top_bits = 0;
-    }
+    QEMU_BUILD_BUG_ON(TARGET_PHYS_ADDR_SPACE_BITS > 52);
+    assert(cpu->phys_bits <= TARGET_PHYS_ADDR_SPACE_BITS);
+    mtrr_top_bits = MAKE_64BIT_MASK(cpu->phys_bits, 52 - cpu->phys_bits);
 
     for (i = 0; i < ret; i++) {
         uint32_t index = msrs[i].index;
#74 lore raw

PATCH v4 14/27 hw/intc/apic: Remove APICCommonState::legacy_instance_id field

In reply to Philippe Mathieu-Daudé

The APICCommonState::legacy_instance_id boolean was only set in the pc_compat_2_6[] array, via the 'legacy-instance-id=on' property. We removed all machines using that array, lets remove that property, simplifying apic_common_realize().

Because instance_id is initialized as initial_apic_id, we can not register vmstate_apic_common directly via dc->vmsd.

---
 include/hw/i386/apic_internal.h | 1 -
 hw/intc/apic_common.c           | 5 -----
 2 files changed, 6 deletions(-)
include/hw/i386/apic_internal.h +0 -1
index 429278da618..db6a9101530 100644
@@ -188,7 +188,6 @@ struct APICCommonState {
     uint32_t vapic_control;
     DeviceState *vapic;
     hwaddr vapic_paddr; /* note: persistence via kvmvapic */
-    bool legacy_instance_id;
     uint32_t extended_log_dest;
 };
 
hw/intc/apic_common.c +0 -5
index 37a7a7019d3..1d259b97e63 100644
@@ -294,9 +294,6 @@ static void apic_common_realize(DeviceState *dev, Error **errp)
         info->enable_tpr_reporting(s, true);
     }
 
-    if (s->legacy_instance_id) {
-        instance_id = VMSTATE_INSTANCE_ID_ANY;
-    }
     vmstate_register_with_alias_id(NULL, instance_id, &vmstate_apic_common,
                                    s, -1, 0, NULL);
 
@@ -412,8 +409,6 @@ static const Property apic_properties_common[] = {
     DEFINE_PROP_UINT8("version", APICCommonState, version, 0x14),
     DEFINE_PROP_BIT("vapic", APICCommonState, vapic_control, VAPIC_ENABLE_BIT,
                     true),
-    DEFINE_PROP_BOOL("legacy-instance-id", APICCommonState, legacy_instance_id,
-                     false),
 };
 
 static void apic_common_get_id(Object *obj, Visitor *v, const char *name,
Signature · 1 line
2.47.1
Reviewed-by on v4 14/27
#75 lore raw

Re: [PATCH v4 14/27] hw/intc/apic: Remove APICCommonState::legacy_instance_id field

Quoted 14 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:37 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 14/27] hw/intc/apic: Remove
APICCommonState::legacy_instance_id field
X-Mailer: git-send-email 2.47.1

The APICCommonState::legacy_instance_id boolean was only set in the pc_compat_2_6[] array, via the 'legacy-instance-id=on' property. We removed all machines using that array, lets remove that property, simplifying apic_common_realize().

Because instance_id is initialized as initial_apic_id, we can not register vmstate_apic_common directly via dc->vmsd.

Quoted 7 lines
---
 include/hw/i386/apic_internal.h | 1 -
 hw/intc/apic_common.c           | 5 -----
 2 files changed, 6 deletions(-)
Reviewed-by on v4 14/27
#76 lore raw

Re: [PATCH v4 14/27] hw/intc/apic: Remove APICCommonState::legacy_instance_id field

In reply to Philippe Mathieu-Daudé
Quoted 7 lines · On Thu, 8 May 2025 15:35:37 +0200 Philippe Mathieu-Daudé <[email protected]>…

The APICCommonState::legacy_instance_id boolean was only set in the pc_compat_2_6[] array, via the 'legacy-instance-id=on' property. We removed all machines using that array, lets remove that property, simplifying apic_common_realize().

Because instance_id is initialized as initial_apic_id, we can not register vmstate_apic_common directly via dc->vmsd.

I think just removing this property is pretty save, it's highly unlikely to be used by any external user as it's purpose was to keep migration working for 2.6. With the later gone there is not need for the property at all.

Quoted 4 lines
Quoted 40 lines
---
 include/hw/i386/apic_internal.h | 1 -
 hw/intc/apic_common.c           | 5 -----
 2 files changed, 6 deletions(-)
include/hw/i386/apic_internal.h +0 -1
index 429278da618..db6a9101530 100644
@@ -188,7 +188,6 @@ struct APICCommonState {
     uint32_t vapic_control;
     DeviceState *vapic;
     hwaddr vapic_paddr; /* note: persistence via kvmvapic */
-    bool legacy_instance_id;
     uint32_t extended_log_dest;
 };
 
hw/intc/apic_common.c +0 -5
index 37a7a7019d3..1d259b97e63 100644
@@ -294,9 +294,6 @@ static void apic_common_realize(DeviceState *dev, Error **errp)
         info->enable_tpr_reporting(s, true);
     }
 
-    if (s->legacy_instance_id) {
-        instance_id = VMSTATE_INSTANCE_ID_ANY;
-    }
     vmstate_register_with_alias_id(NULL, instance_id, &vmstate_apic_common,
                                    s, -1, 0, NULL);
 
@@ -412,8 +409,6 @@ static const Property apic_properties_common[] = {
     DEFINE_PROP_UINT8("version", APICCommonState, version, 0x14),
     DEFINE_PROP_BIT("vapic", APICCommonState, vapic_control, VAPIC_ENABLE_BIT,
                     true),
-    DEFINE_PROP_BOOL("legacy-instance-id", APICCommonState, legacy_instance_id,
-                     false),
 };
 
 static void apic_common_get_id(Object *obj, Visitor *v, const char *name,
#77 lore raw

PATCH v4 15/27 hw/core/machine: Remove hw_compat_2_6[] array

In reply to Philippe Mathieu-Daudé

The hw_compat_2_6[] array was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it.

---
 include/hw/boards.h | 3 ---
 hw/core/machine.c   | 8 --------
 2 files changed, 11 deletions(-)
include/hw/boards.h +0 -3
index 5f1a0fb7e28..a881db8e7d6 100644
@@ -841,7 +841,4 @@ extern const size_t hw_compat_2_8_len;
 extern GlobalProperty hw_compat_2_7[];
 extern const size_t hw_compat_2_7_len;
 
-extern GlobalProperty hw_compat_2_6[];
-extern const size_t hw_compat_2_6_len;
-
 #endif
hw/core/machine.c +0 -8
index e7001bf92cd..ce98820f277 100644
@@ -275,14 +275,6 @@ GlobalProperty hw_compat_2_7[] = {
 };
 const size_t hw_compat_2_7_len = G_N_ELEMENTS(hw_compat_2_7);
 
-GlobalProperty hw_compat_2_6[] = {
-    { "virtio-mmio", "format_transport_address", "off" },
-    /* Optional because not all virtio-pci devices support legacy mode */
-    { "virtio-pci", "disable-modern", "on",  .optional = true },
-    { "virtio-pci", "disable-legacy", "off", .optional = true },
-};
-const size_t hw_compat_2_6_len = G_N_ELEMENTS(hw_compat_2_6);
-
 MachineState *current_machine;
 
 static char *machine_get_kernel(Object *obj, Error **errp)
Signature · 1 line
2.47.1
Reviewed-by on v4 15/27
#78 lore raw

Re: [PATCH v4 15/27] hw/core/machine: Remove hw_compat_2_6[] array

Quoted 8 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:38 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 15/27] hw/core/machine: Remove hw_compat_2_6[] array
X-Mailer: git-send-email 2.47.1

The hw_compat_2_6[] array was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it.

Quoted 7 lines
---
 include/hw/boards.h | 3 ---
 hw/core/machine.c   | 8 --------
 2 files changed, 11 deletions(-)
#79 lore raw

Re: [PATCH v4 15/27] hw/core/machine: Remove hw_compat_2_6[] array

In reply to Philippe Mathieu-Daudé
Quoted 2 lines · On Thu, 8 May 2025 15:35:38 +0200 Philippe Mathieu-Daudé <[email protected]>…

The hw_compat_2_6[] array was only used by the pc-q35-2.6 and pc-i440fx-2.6 machines, which got removed. Remove it.

see my comment in 1/27

Quoted 40 lines
---
 include/hw/boards.h | 3 ---
 hw/core/machine.c   | 8 --------
 2 files changed, 11 deletions(-)
include/hw/boards.h +0 -3
index 5f1a0fb7e28..a881db8e7d6 100644
@@ -841,7 +841,4 @@ extern const size_t hw_compat_2_8_len;
 extern GlobalProperty hw_compat_2_7[];
 extern const size_t hw_compat_2_7_len;
 
-extern GlobalProperty hw_compat_2_6[];
-extern const size_t hw_compat_2_6_len;
-
 #endif
hw/core/machine.c +0 -8
index e7001bf92cd..ce98820f277 100644
@@ -275,14 +275,6 @@ GlobalProperty hw_compat_2_7[] = {
 };
 const size_t hw_compat_2_7_len = G_N_ELEMENTS(hw_compat_2_7);
 
-GlobalProperty hw_compat_2_6[] = {
-    { "virtio-mmio", "format_transport_address", "off" },
-    /* Optional because not all virtio-pci devices support legacy mode */
-    { "virtio-pci", "disable-modern", "on",  .optional = true },
-    { "virtio-pci", "disable-legacy", "off", .optional = true },
-};
-const size_t hw_compat_2_6_len = G_N_ELEMENTS(hw_compat_2_6);
-
 MachineState *current_machine;
 
 static char *machine_get_kernel(Object *obj, Error **errp)
#80 lore raw

PATCH v4 16/27 hw/virtio/virtio-mmio: Remove VirtIOMMIOProxy::format_transport_address field

In reply to Philippe Mathieu-Daudé

The VirtIOMMIOProxy::format_transport_address boolean was only set in the hw_compat_2_6[] array, via the 'format_transport_address=off' property. We removed all machines using that array, lets remove that property, simplifying virtio_mmio_bus_get_dev_path().

---
 include/hw/virtio/virtio-mmio.h |  1 -
 hw/virtio/virtio-mmio.c         | 15 ---------------
 2 files changed, 16 deletions(-)
include/hw/virtio/virtio-mmio.h +0 -1
index aa492620228..8b19ec2291a 100644
@@ -66,7 +66,6 @@ struct VirtIOMMIOProxy {
     uint32_t guest_page_shift;
     /* virtio-bus */
     VirtioBusState bus;
-    bool format_transport_address;
     /* Fields only used for non-legacy (v2) devices */
     uint32_t guest_features[2];
     VirtIOMMIOQueue vqs[VIRTIO_QUEUE_MAX];
hw/virtio/virtio-mmio.c +0 -15
index 532c67107ba..b7ee115b990 100644
@@ -752,8 +752,6 @@ static void virtio_mmio_pre_plugged(DeviceState *d, Error **errp)
 /* virtio-mmio device */
 
 static const Property virtio_mmio_properties[] = {
-    DEFINE_PROP_BOOL("format_transport_address", VirtIOMMIOProxy,
-                     format_transport_address, true),
     DEFINE_PROP_BOOL("force-legacy", VirtIOMMIOProxy, legacy, true),
     DEFINE_PROP_BIT("ioeventfd", VirtIOMMIOProxy, flags,
                     VIRTIO_IOMMIO_FLAG_USE_IOEVENTFD_BIT, true),
@@ -815,19 +813,6 @@ static char *virtio_mmio_bus_get_dev_path(DeviceState *dev)
     virtio_mmio_proxy = VIRTIO_MMIO(virtio_mmio_bus->parent);
     proxy_path = qdev_get_dev_path(DEVICE(virtio_mmio_proxy));
 
-    /*
-     * If @format_transport_address is false, then we just perform the same as
-     * virtio_bus_get_dev_path(): we delegate the address formatting for the
-     * device on the virtio-mmio bus to the bus that the virtio-mmio proxy
-     * (i.e., the device that implements the virtio-mmio bus) resides on. In
-     * this case the base address of the virtio-mmio transport will be
-     * invisible.
-     */
-    if (!virtio_mmio_proxy->format_transport_address) {
-        return proxy_path;
-    }
-
-    /* Otherwise, we append the base address of the transport. */
     section = memory_region_find(&virtio_mmio_proxy->iomem, 0, 0x200);
     assert(section.mr);
 
Signature · 1 line
2.47.1
Reviewed-by on v4 16/27
#81 lore raw

Re: [PATCH v4 16/27] hw/virtio/virtio-mmio: Remove VirtIOMMIOProxy::format_transport_address field

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:39 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 16/27] hw/virtio/virtio-mmio: Remove
VirtIOMMIOProxy::format_transport_address field
X-Mailer: git-send-email 2.47.1

The VirtIOMMIOProxy::format_transport_address boolean was only set in the hw_compat_2_6[] array, via the 'format_transport_address=off' property. We removed all machines using that array, lets remove that property, simplifying virtio_mmio_bus_get_dev_path().

Quoted 7 lines
---
 include/hw/virtio/virtio-mmio.h |  1 -
 hw/virtio/virtio-mmio.c         | 15 ---------------
 2 files changed, 16 deletions(-)
#82 lore raw

PATCH v4 17/27 hw/i386/pc: Remove deprecated pc-q35-2.7 and pc-i440fx-2.7 machines

In reply to Philippe Mathieu-Daudé

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed. Remove the qtest in test-x86-cpuid-compat.c file.

---
 hw/i386/pc_piix.c                   |  9 ---------
 hw/i386/pc_q35.c                    | 10 ----------
 tests/qtest/test-x86-cpuid-compat.c | 11 -----------
 3 files changed, 30 deletions(-)
hw/i386/pc_piix.c +0 -9
index 98a118fd4a0..98bd8d0e67b 100644
@@ -755,15 +755,6 @@ static void pc_i440fx_machine_2_8_options(MachineClass *m)
 
 DEFINE_I440FX_MACHINE(2, 8);
 
-static void pc_i440fx_machine_2_7_options(MachineClass *m)
-{
-    pc_i440fx_machine_2_8_options(m);
-    compat_props_add(m->compat_props, hw_compat_2_7, hw_compat_2_7_len);
-    compat_props_add(m->compat_props, pc_compat_2_7, pc_compat_2_7_len);
-}
-
-DEFINE_I440FX_MACHINE(2, 7);
-
 #ifdef CONFIG_ISAPC
 static void isapc_machine_options(MachineClass *m)
 {
hw/i386/pc_q35.c +0 -10
index b7ffb5f1216..a1f46cd8f03 100644
@@ -648,13 +648,3 @@ static void pc_q35_machine_2_8_options(MachineClass *m)
 }
 
 DEFINE_Q35_MACHINE(2, 8);
-
-static void pc_q35_machine_2_7_options(MachineClass *m)
-{
-    pc_q35_machine_2_8_options(m);
-    m->max_cpus = 255;
-    compat_props_add(m->compat_props, hw_compat_2_7, hw_compat_2_7_len);
-    compat_props_add(m->compat_props, pc_compat_2_7, pc_compat_2_7_len);
-}
-
-DEFINE_Q35_MACHINE(2, 7);
tests/qtest/test-x86-cpuid-compat.c +0 -11
index 456e2af6657..5e0547e81b7 100644
@@ -345,17 +345,6 @@ int main(int argc, char **argv)
 
     /* Check compatibility of old machine-types that didn't
      * auto-increase level/xlevel/xlevel2: */
-    if (qtest_has_machine("pc-i440fx-2.7")) {
-        add_cpuid_test("x86/cpuid/auto-level/pc-2.7",
-                       "486", "arat=on,avx512vbmi=on,xsaveopt=on",
-                       "pc-i440fx-2.7", "level", 1);
-        add_cpuid_test("x86/cpuid/auto-xlevel/pc-2.7",
-                       "486", "3dnow=on,sse4a=on,invtsc=on,npt=on,svm=on",
-                       "pc-i440fx-2.7", "xlevel", 0);
-        add_cpuid_test("x86/cpuid/auto-xlevel2/pc-2.7",
-                       "486", "xstore=on", "pc-i440fx-2.7",
-                       "xlevel2", 0);
-    }
     if (qtest_has_machine("pc-i440fx-2.9")) {
         add_cpuid_test("x86/cpuid/auto-level7/pc-i440fx-2.9/off",
                        "Conroe", NULL, "pc-i440fx-2.9",
Signature · 1 line
2.47.1
Reviewed-by on v4 17/27
#83 lore raw

Re: [PATCH v4 17/27] hw/i386/pc: Remove deprecated pc-q35-2.7 and pc-i440fx-2.7 machines

Quoted 12 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:40 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 17/27] hw/i386/pc: Remove deprecated pc-q35-2.7 and
pc-i440fx-2.7 machines
X-Mailer: git-send-email 2.47.1

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed. Remove the qtest in test-x86-cpuid-compat.c file.

Quoted 8 lines
---
 hw/i386/pc_piix.c                   |  9 ---------
 hw/i386/pc_q35.c                    | 10 ----------
 tests/qtest/test-x86-cpuid-compat.c | 11 -----------
 3 files changed, 30 deletions(-)
Reviewed-by on v4 17/27
#84 lore raw

Re: [PATCH v4 17/27] hw/i386/pc: Remove deprecated pc-q35-2.7 and pc-i440fx-2.7 machines

In reply to Philippe Mathieu-Daudé
Quoted 5 lines · On Thu, 8 May 2025 15:35:40 +0200 Philippe Mathieu-Daudé <[email protected]>…

These machines has been supported for a period of more than 6 years. According to our versioned machine support policy (see commit ce80c4fa6ff "docs: document special exception for machine type deprecation & removal") they can now be removed. Remove the qtest in test-x86-cpuid-compat.c file.

same comment as 1/27,

I'd squash pc|hw_compat_2_7 removal in here

other than that loos good to me, with above

Quoted 69 lines
---
 hw/i386/pc_piix.c                   |  9 ---------
 hw/i386/pc_q35.c                    | 10 ----------
 tests/qtest/test-x86-cpuid-compat.c | 11 -----------
 3 files changed, 30 deletions(-)
hw/i386/pc_piix.c +0 -9
index 98a118fd4a0..98bd8d0e67b 100644
@@ -755,15 +755,6 @@ static void pc_i440fx_machine_2_8_options(MachineClass *m)
 
 DEFINE_I440FX_MACHINE(2, 8);
 
-static void pc_i440fx_machine_2_7_options(MachineClass *m)
-{
-    pc_i440fx_machine_2_8_options(m);
-    compat_props_add(m->compat_props, hw_compat_2_7, hw_compat_2_7_len);
-    compat_props_add(m->compat_props, pc_compat_2_7, pc_compat_2_7_len);
-}
-
-DEFINE_I440FX_MACHINE(2, 7);
-
 #ifdef CONFIG_ISAPC
 static void isapc_machine_options(MachineClass *m)
 {
hw/i386/pc_q35.c +0 -10
index b7ffb5f1216..a1f46cd8f03 100644
@@ -648,13 +648,3 @@ static void pc_q35_machine_2_8_options(MachineClass *m)
 }
 
 DEFINE_Q35_MACHINE(2, 8);
-
-static void pc_q35_machine_2_7_options(MachineClass *m)
-{
-    pc_q35_machine_2_8_options(m);
-    m->max_cpus = 255;
-    compat_props_add(m->compat_props, hw_compat_2_7, hw_compat_2_7_len);
-    compat_props_add(m->compat_props, pc_compat_2_7, pc_compat_2_7_len);
-}
-
-DEFINE_Q35_MACHINE(2, 7);
tests/qtest/test-x86-cpuid-compat.c +0 -11
index 456e2af6657..5e0547e81b7 100644
@@ -345,17 +345,6 @@ int main(int argc, char **argv)
 
     /* Check compatibility of old machine-types that didn't
      * auto-increase level/xlevel/xlevel2: */
-    if (qtest_has_machine("pc-i440fx-2.7")) {
-        add_cpuid_test("x86/cpuid/auto-level/pc-2.7",
-                       "486", "arat=on,avx512vbmi=on,xsaveopt=on",
-                       "pc-i440fx-2.7", "level", 1);
-        add_cpuid_test("x86/cpuid/auto-xlevel/pc-2.7",
-                       "486", "3dnow=on,sse4a=on,invtsc=on,npt=on,svm=on",
-                       "pc-i440fx-2.7", "xlevel", 0);
-        add_cpuid_test("x86/cpuid/auto-xlevel2/pc-2.7",
-                       "486", "xstore=on", "pc-i440fx-2.7",
-                       "xlevel2", 0);
-    }
     if (qtest_has_machine("pc-i440fx-2.9")) {
         add_cpuid_test("x86/cpuid/auto-level7/pc-i440fx-2.9/off",
                        "Conroe", NULL, "pc-i440fx-2.9",
#85 lore raw

PATCH v4 18/27 hw/i386/pc: Remove pc_compat_2_7[] array

In reply to Philippe Mathieu-Daudé

The pc_compat_2_7[] array was only used by the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Remove it.

---
 include/hw/i386/pc.h |  3 ---
 hw/i386/pc.c         | 10 ----------
 2 files changed, 13 deletions(-)
include/hw/i386/pc.h +0 -3
index 4fb2033bc54..319ec82f709 100644
@@ -289,9 +289,6 @@ extern const size_t pc_compat_2_9_len;
 extern GlobalProperty pc_compat_2_8[];
 extern const size_t pc_compat_2_8_len;
 
-extern GlobalProperty pc_compat_2_7[];
-extern const size_t pc_compat_2_7_len;
-
 #define DEFINE_PC_MACHINE(suffix, namestr, initfn, optsfn) \
     static void pc_machine_##suffix##_class_init(ObjectClass *oc, \
                                                  const void *data) \
hw/i386/pc.c +0 -10
index 65a11ea8f99..c7cdbe93753 100644
@@ -241,16 +241,6 @@ GlobalProperty pc_compat_2_8[] = {
 };
 const size_t pc_compat_2_8_len = G_N_ELEMENTS(pc_compat_2_8);
 
-GlobalProperty pc_compat_2_7[] = {
-    { TYPE_X86_CPU, "l3-cache", "off" },
-    { TYPE_X86_CPU, "full-cpuid-auto-level", "off" },
-    { "Opteron_G3" "-" TYPE_X86_CPU, "family", "15" },
-    { "Opteron_G3" "-" TYPE_X86_CPU, "model", "6" },
-    { "Opteron_G3" "-" TYPE_X86_CPU, "stepping", "1" },
-    { "isa-pcspk", "migrate", "off" },
-};
-const size_t pc_compat_2_7_len = G_N_ELEMENTS(pc_compat_2_7);
-
 /*
  * @PC_FW_DATA:
  * Size of the chunk of memory at the top of RAM for the BIOS ACPI tables
Signature · 1 line
2.47.1
Reviewed-by on v4 18/27
#86 lore raw

Re: [PATCH v4 18/27] hw/i386/pc: Remove pc_compat_2_7[] array

Quoted 8 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:41 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 18/27] hw/i386/pc: Remove pc_compat_2_7[] array
X-Mailer: git-send-email 2.47.1

The pc_compat_2_7[] array was only used by the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Remove it.

Quoted 6 lines
---
 include/hw/i386/pc.h |  3 ---
 hw/i386/pc.c         | 10 ----------
 2 files changed, 13 deletions(-)
#87 lore raw

PATCH v4 19/27 target/i386/cpu: Remove CPUX86State::full_cpuid_auto_level field

In reply to Philippe Mathieu-Daudé

The CPUX86State::full_cpuid_auto_level boolean was only disabled for the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Being now always %true, we can remove it and simplify x86_cpu_expand_features().

---
 target/i386/cpu.h |   3 --
 target/i386/cpu.c | 106 ++++++++++++++++++++++------------------------
 2 files changed, 51 insertions(+), 58 deletions(-)
target/i386/cpu.h +0 -3
index 7585407da54..b5cbd91c156 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
      */
     bool legacy_multi_node;
 
-    /* Enable auto level-increase for all CPUID leaves */
-    bool full_cpuid_auto_level;
-
     /* Only advertise CPUID leaves defined by the vendor */
     bool vendor_cpuid_only;
 
target/i386/cpu.c +51 -55

126 lines not on this page · raw message

Signature · 1 line
2.47.1
Reviewed-by on v4 19/27
#88 lore raw

Re: [PATCH v4 19/27] target/i386/cpu: Remove CPUX86State::full_cpuid_auto_level field

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:42 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 19/27] target/i386/cpu: Remove
CPUX86State::full_cpuid_auto_level field
X-Mailer: git-send-email 2.47.1

The CPUX86State::full_cpuid_auto_level boolean was only disabled for the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Being now always %true, we can remove it and simplify x86_cpu_expand_features().

Quoted 5 lines
---
 target/i386/cpu.h |   3 --
 target/i386/cpu.c | 106 ++++++++++++++++++++++------------------------
 2 files changed, 51 insertions(+), 58 deletions(-)

This property is a fix, so,

#89 lore raw

Re: [PATCH v4 19/27] target/i386/cpu: Remove CPUX86State::full_cpuid_auto_level field

In reply to Philippe Mathieu-Daudé
Quoted 4 lines · On Thu, 8 May 2025 15:35:42 +0200 Philippe Mathieu-Daudé <[email protected]>…

The CPUX86State::full_cpuid_auto_level boolean was only disabled for the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Being now always %true, we can remove it and simplify x86_cpu_expand_features().

I've found field being mentioned only by some external rust library, that's likely shouldn't concern QEMU qemu though.

I'm not confident enough to ack it but I won't object either

Quoted 152 lines
---
 target/i386/cpu.h |   3 --
 target/i386/cpu.c | 106 ++++++++++++++++++++++------------------------
 2 files changed, 51 insertions(+), 58 deletions(-)
target/i386/cpu.h +0 -3
index 7585407da54..b5cbd91c156 100644
@@ -2241,9 +2241,6 @@ struct ArchCPU {
      */
     bool legacy_multi_node;
 
-    /* Enable auto level-increase for all CPUID leaves */
-    bool full_cpuid_auto_level;
-
     /* Only advertise CPUID leaves defined by the vendor */
     bool vendor_cpuid_only;
 
target/i386/cpu.c +51 -55
index fb505d13122..6b9a1f2251a 100644
@@ -7843,68 +7843,65 @@ void x86_cpu_expand_features(X86CPU *cpu, Error **errp)
 
     /* CPUID[EAX=7,ECX=0].EBX always increased level automatically: */
     x86_cpu_adjust_feat_level(cpu, FEAT_7_0_EBX);
-    if (cpu->full_cpuid_auto_level) {
-        x86_cpu_adjust_feat_level(cpu, FEAT_1_EDX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_1_ECX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_6_EAX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_7_0_ECX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_7_1_EAX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_7_1_EDX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_7_2_EDX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_8000_0001_EDX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_8000_0001_ECX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_8000_0007_EDX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_8000_0008_EBX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_C000_0001_EDX);
-        x86_cpu_adjust_feat_level(cpu, FEAT_SVM);
-        x86_cpu_adjust_feat_level(cpu, FEAT_XSAVE);
+    x86_cpu_adjust_feat_level(cpu, FEAT_1_EDX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_1_ECX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_6_EAX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_7_0_ECX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_7_1_EAX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_7_1_EDX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_7_2_EDX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_8000_0001_EDX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_8000_0001_ECX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_8000_0007_EDX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_8000_0008_EBX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_C000_0001_EDX);
+    x86_cpu_adjust_feat_level(cpu, FEAT_SVM);
+    x86_cpu_adjust_feat_level(cpu, FEAT_XSAVE);
 
-        /* Intel Processor Trace requires CPUID[0x14] */
-        if ((env->features[FEAT_7_0_EBX] & CPUID_7_0_EBX_INTEL_PT)) {
-            if (cpu->intel_pt_auto_level) {
-                x86_cpu_adjust_level(cpu, &cpu->env.cpuid_min_level, 0x14);
-            } else if (cpu->env.cpuid_min_level < 0x14) {
-                mark_unavailable_features(cpu, FEAT_7_0_EBX,
-                    CPUID_7_0_EBX_INTEL_PT,
-                    "Intel PT need CPUID leaf 0x14, please set by \"-cpu ...,intel-pt=on,min-level=0x14\"");
-            }
+    /* Intel Processor Trace requires CPUID[0x14] */
+    if ((env->features[FEAT_7_0_EBX] & CPUID_7_0_EBX_INTEL_PT)) {
+        if (cpu->intel_pt_auto_level) {
+            x86_cpu_adjust_level(cpu, &cpu->env.cpuid_min_level, 0x14);
+        } else if (cpu->env.cpuid_min_level < 0x14) {
+            mark_unavailable_features(cpu, FEAT_7_0_EBX, CPUID_7_0_EBX_INTEL_PT,
+                "Intel PT need CPUID leaf 0x14, please set by \"-cpu ...,intel-pt=on,min-level=0x14\"");
         }
+    }
 
-        /*
-         * Intel CPU topology with multi-dies support requires CPUID[0x1F].
-         * For AMD Rome/Milan, cpuid level is 0x10, and guest OS should detect
-         * extended toplogy by leaf 0xB. Only adjust it for Intel CPU, unless
-         * cpu->vendor_cpuid_only has been unset for compatibility with older
-         * machine types.
-         */
-        if (x86_has_extended_topo(env->avail_cpu_topo) &&
-            (IS_INTEL_CPU(env) || !cpu->vendor_cpuid_only)) {
-            x86_cpu_adjust_level(cpu, &env->cpuid_min_level, 0x1F);
-        }
+    /*
+     * Intel CPU topology with multi-dies support requires CPUID[0x1F].
+     * For AMD Rome/Milan, cpuid level is 0x10, and guest OS should detect
+     * extended toplogy by leaf 0xB. Only adjust it for Intel CPU, unless
+     * cpu->vendor_cpuid_only has been unset for compatibility with older
+     * machine types.
+     */
+    if (x86_has_extended_topo(env->avail_cpu_topo) &&
+        (IS_INTEL_CPU(env) || !cpu->vendor_cpuid_only)) {
+        x86_cpu_adjust_level(cpu, &env->cpuid_min_level, 0x1F);
+    }
 
-        /* Advanced Vector Extensions 10 (AVX10) requires CPUID[0x24] */
-        if (env->features[FEAT_7_1_EDX] & CPUID_7_1_EDX_AVX10) {
-            x86_cpu_adjust_level(cpu, &env->cpuid_min_level, 0x24);
-        }
+    /* Advanced Vector Extensions 10 (AVX10) requires CPUID[0x24] */
+    if (env->features[FEAT_7_1_EDX] & CPUID_7_1_EDX_AVX10) {
+        x86_cpu_adjust_level(cpu, &env->cpuid_min_level, 0x24);
+    }
 
-        /* SVM requires CPUID[0x8000000A] */
-        if (env->features[FEAT_8000_0001_ECX] & CPUID_EXT3_SVM) {
-            x86_cpu_adjust_level(cpu, &env->cpuid_min_xlevel, 0x8000000A);
-        }
+    /* SVM requires CPUID[0x8000000A] */
+    if (env->features[FEAT_8000_0001_ECX] & CPUID_EXT3_SVM) {
+        x86_cpu_adjust_level(cpu, &env->cpuid_min_xlevel, 0x8000000A);
+    }
 
-        /* SEV requires CPUID[0x8000001F] */
-        if (sev_enabled()) {
-            x86_cpu_adjust_level(cpu, &env->cpuid_min_xlevel, 0x8000001F);
-        }
+    /* SEV requires CPUID[0x8000001F] */
+    if (sev_enabled()) {
+        x86_cpu_adjust_level(cpu, &env->cpuid_min_xlevel, 0x8000001F);
+    }
 
-        if (env->features[FEAT_8000_0021_EAX]) {
-            x86_cpu_adjust_level(cpu, &env->cpuid_min_xlevel, 0x80000021);
-        }
+    if (env->features[FEAT_8000_0021_EAX]) {
+        x86_cpu_adjust_level(cpu, &env->cpuid_min_xlevel, 0x80000021);
+    }
 
-        /* SGX requires CPUID[0x12] for EPC enumeration */
-        if (env->features[FEAT_7_0_EBX] & CPUID_7_0_EBX_SGX) {
-            x86_cpu_adjust_level(cpu, &env->cpuid_min_level, 0x12);
-        }
+    /* SGX requires CPUID[0x12] for EPC enumeration */
+    if (env->features[FEAT_7_0_EBX] & CPUID_7_0_EBX_SGX) {
+        x86_cpu_adjust_level(cpu, &env->cpuid_min_level, 0x12);
     }
 
     /* Set cpuid_*level* based on cpuid_min_*level, if not explicitly set */
@@ -8820,7 +8817,6 @@ static const Property x86_cpu_properties[] = {
     DEFINE_PROP_UINT32("min-xlevel2", X86CPU, env.cpuid_min_xlevel2, 0),
     DEFINE_PROP_UINT8("avx10-version", X86CPU, env.avx10_version, 0),
     DEFINE_PROP_UINT64("ucode-rev", X86CPU, ucode_rev, 0),
-    DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),
     DEFINE_PROP_STRING("hv-vendor-id", X86CPU, hyperv_vendor),
     DEFINE_PROP_BOOL("x-vendor-cpuid-only", X86CPU, vendor_cpuid_only, true),
     DEFINE_PROP_BOOL("x-amd-topoext-features-only", X86CPU, amd_topoext_features_only, true),
#90 lore raw

PATCH v4 20/27 target/i386/cpu: Remove CPUX86State::enable_l3_cache field

In reply to Philippe Mathieu-Daudé

The CPUX86State::enable_l3_cache boolean was only disabled for the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid() and encode_cache_cpuid80000006().

---
 target/i386/cpu.h |  6 ------
 target/i386/cpu.c | 39 +++++++++++++--------------------------
 2 files changed, 13 insertions(+), 32 deletions(-)
target/i386/cpu.h +0 -6
index b5cbd91c156..62239b0a562 100644
@@ -2219,12 +2219,6 @@ struct ArchCPU {
      */
     bool enable_lmce;
 
-    /* Compatibility bits for old machine types.
-     * If true present virtual l3 cache for VM, the vcpus in the same virtual
-     * socket share an virtual l3 cache.
-     */
-    bool enable_l3_cache;
-
     /* Compatibility bits for old machine types.
      * If true present L1 cache as per-thread, not per-core.
      */
target/i386/cpu.c +13 -26
index 6b9a1f2251a..4be174ea9c7 100644
@@ -468,17 +468,13 @@ static void encode_cache_cpuid80000006(CPUCacheInfo *l2,
            (AMD_ENC_ASSOC(l2->associativity) << 12) |
            (l2->lines_per_tag << 8) | (l2->line_size);
 
-    if (l3) {
-        assert(l3->size % (512 * 1024) == 0);
-        assert(l3->associativity > 0);
-        assert(l3->lines_per_tag > 0);
-        assert(l3->line_size > 0);
-        *edx = ((l3->size / (512 * 1024)) << 18) |
-               (AMD_ENC_ASSOC(l3->associativity) << 12) |
-               (l3->lines_per_tag << 8) | (l3->line_size);
-    } else {
-        *edx = 0;
-    }
+    assert(l3->size % (512 * 1024) == 0);
+    assert(l3->associativity > 0);
+    assert(l3->lines_per_tag > 0);
+    assert(l3->line_size > 0);
+    *edx = ((l3->size / (512 * 1024)) << 18) |
+           (AMD_ENC_ASSOC(l3->associativity) << 12) |
+           (l3->lines_per_tag << 8) | (l3->line_size);
 }
 
 /* Encode cache info for CPUID[8000001D] */
@@ -6849,11 +6845,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
         }
         *eax = 1; /* Number of CPUID[EAX=2] calls required */
         *ebx = 0;
-        if (!cpu->enable_l3_cache) {
-            *ecx = 0;
-        } else {
-            *ecx = cpuid2_cache_descriptor(env->cache_info_cpuid2.l3_cache);
-        }
+        *ecx = cpuid2_cache_descriptor(env->cache_info_cpuid2.l3_cache);
         *edx = (cpuid2_cache_descriptor(env->cache_info_cpuid2.l1d_cache) << 16) |
                (cpuid2_cache_descriptor(env->cache_info_cpuid2.l1i_cache) <<  8) |
                (cpuid2_cache_descriptor(env->cache_info_cpuid2.l2_cache));
@@ -6907,13 +6899,10 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
                                     eax, ebx, ecx, edx);
                 break;
             case 3: /* L3 cache info */
-                if (cpu->enable_l3_cache) {
-                    encode_cache_cpuid4(env->cache_info_cpuid4.l3_cache,
-                                        topo_info,
-                                        eax, ebx, ecx, edx);
-                    break;
-                }
-                /* fall through */
+                encode_cache_cpuid4(env->cache_info_cpuid4.l3_cache,
+                                    topo_info,
+                                    eax, ebx, ecx, edx);
+                break;
             default: /* end of info */
                 *eax = *ebx = *ecx = *edx = 0;
                 break;
@@ -7284,8 +7273,7 @@ void cpu_x86_cpuid(CPUX86State *env, uint32_t index, uint32_t count,
                (AMD_ENC_ASSOC(L2_ITLB_4K_ASSOC) << 12) |
                (L2_ITLB_4K_ENTRIES);
         encode_cache_cpuid80000006(env->cache_info_amd.l2_cache,
-                                   cpu->enable_l3_cache ?
-                                   env->cache_info_amd.l3_cache : NULL,
+                                   env->cache_info_amd.l3_cache,
                                    ecx, edx);
         break;
     case 0x80000007:
@@ -8821,7 +8809,6 @@ static const Property x86_cpu_properties[] = {
     DEFINE_PROP_BOOL("x-vendor-cpuid-only", X86CPU, vendor_cpuid_only, true),
     DEFINE_PROP_BOOL("x-amd-topoext-features-only", X86CPU, amd_topoext_features_only, true),
     DEFINE_PROP_BOOL("lmce", X86CPU, enable_lmce, false),
-    DEFINE_PROP_BOOL("l3-cache", X86CPU, enable_l3_cache, true),
     DEFINE_PROP_BOOL("kvm-pv-enforce-cpuid", X86CPU, kvm_pv_enforce_cpuid,
                      false),
     DEFINE_PROP_BOOL("vmware-cpuid-freq", X86CPU, vmware_cpuid_freq, true),
Signature · 1 line
2.47.1
#91 lore raw

Re: [PATCH v4 20/27] target/i386/cpu: Remove CPUX86State::enable_l3_cache field

Quoted 24 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:43 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 20/27] target/i386/cpu: Remove
CPUX86State::enable_l3_cache field
X-Mailer: git-send-email 2.47.1

The CPUX86State::enable_l3_cache boolean was only disabled for the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid() and encode_cache_cpuid80000006().

---
 target/i386/cpu.h |  6 ------
 target/i386/cpu.c | 39 +++++++++++++--------------------------
 2 files changed, 13 insertions(+), 32 deletions(-)
target/i386/cpu.h +0 -0
index b5cbd91c156..62239b0a562 100644
@@ -2219,12 +2219,6 @@ struct ArchCPU {
      */
     bool enable_lmce;

target/i386/cpu.h struct ArchCPU { · line 2224

Quoted 10 lines
target/i386/cpu.h +0 -6
 
-    /* Compatibility bits for old machine types.
-     * If true present virtual l3 cache for VM, the vcpus in the same virtual
-     * socket share an virtual l3 cache.
-     */
-    bool enable_l3_cache;
-
     /* Compatibility bits for old machine types.
      * If true present L1 cache as per-thread, not per-core.
      */

I realize this is another special case.

There is no support for hybrid x86 CPUs in QEMU, but it's also true that there are some actual modern x86 Client CPUs without l3 cache, such as Intel MTL's low power E core (and it has vmx support, i.e., support KVM).

So I think we can keep this property as well, to have some more configuration options for users' emulation.

#92 lore raw

Re: [PATCH v4 20/27] target/i386/cpu: Remove CPUX86State::enable_l3_cache field

Quoted 38 lines · On Fri, 9 May 2025 17:11:13 +0800 Zhao Liu <[email protected]> wrote:
Quoted 24 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:43 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 20/27] target/i386/cpu: Remove
CPUX86State::enable_l3_cache field
X-Mailer: git-send-email 2.47.1

The CPUX86State::enable_l3_cache boolean was only disabled for the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Being now always %true, we can remove it and simplify cpu_x86_cpuid() and encode_cache_cpuid80000006().

---
 target/i386/cpu.h |  6 ------
 target/i386/cpu.c | 39 +++++++++++++--------------------------
 2 files changed, 13 insertions(+), 32 deletions(-)
target/i386/cpu.h +0 -0
index b5cbd91c156..62239b0a562 100644
@@ -2219,12 +2219,6 @@ struct ArchCPU {
      */
     bool enable_lmce;

target/i386/cpu.h struct ArchCPU { · line 2224

Quoted 10 lines
target/i386/cpu.h +0 -6
 
-    /* Compatibility bits for old machine types.
-     * If true present virtual l3 cache for VM, the vcpus in the same virtual
-     * socket share an virtual l3 cache.
-     */
-    bool enable_l3_cache;
-
     /* Compatibility bits for old machine types.
      * If true present L1 cache as per-thread, not per-core.
      */  

I realize this is another special case.

Quoted 6 lines

There is no support for hybrid x86 CPUs in QEMU, but it's also true that there are some actual modern x86 Client CPUs without l3 cache, such as Intel MTL's low power E core (and it has vmx support, i.e., support KVM).

So I think we can keep this property as well, to have some more configuration options for users' emulation.

it also seems that libvirt uses it.

#93 lore raw

PATCH v4 21/27 hw/audio/pcspk: Remove PCSpkState::migrate field

In reply to Philippe Mathieu-Daudé

The PCSpkState::migrate boolean was only set in the pc_compat_2_7[] array, via the 'migrate=off' property. We removed all machines using that array, lets remove that property, simplifying vmstate_spk[].

---
 hw/audio/pcspk.c | 10 ----------
 1 file changed, 10 deletions(-)
hw/audio/pcspk.c +0 -10
index a419161b5b1..0e83ba0bf73 100644
@@ -56,7 +56,6 @@ struct PCSpkState {
     unsigned int play_pos;
     uint8_t data_on;
     uint8_t dummy_refresh_clock;
-    bool migrate;
 };
 
 static const char *s_spk = "pcspk";
@@ -196,18 +195,10 @@ static void pcspk_realizefn(DeviceState *dev, Error **errp)
     pcspk_state = s;
 }
 
-static bool migrate_needed(void *opaque)
-{
-    PCSpkState *s = opaque;
-
-    return s->migrate;
-}
-
 static const VMStateDescription vmstate_spk = {
     .name = "pcspk",
     .version_id = 1,
     .minimum_version_id = 1,
-    .needed = migrate_needed,
     .fields = (const VMStateField[]) {
         VMSTATE_UINT8(data_on, PCSpkState),
         VMSTATE_UINT8(dummy_refresh_clock, PCSpkState),
@@ -218,7 +209,6 @@ static const VMStateDescription vmstate_spk = {
 static const Property pcspk_properties[] = {
     DEFINE_AUDIO_PROPERTIES(PCSpkState, card),
     DEFINE_PROP_UINT32("iobase", PCSpkState, iobase,  0x61),
-    DEFINE_PROP_BOOL("migrate", PCSpkState, migrate,  true),
 };
 
 static void pcspk_class_initfn(ObjectClass *klass, const void *data)
Signature · 1 line
2.47.1
Reviewed-by on v4 21/27
#94 lore raw

Re: [PATCH v4 21/27] hw/audio/pcspk: Remove PCSpkState::migrate field

Quoted 10 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:44 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 21/27] hw/audio/pcspk: Remove PCSpkState::migrate field
X-Mailer: git-send-email 2.47.1

The PCSpkState::migrate boolean was only set in the pc_compat_2_7[] array, via the 'migrate=off' property. We removed all machines using that array, lets remove that property, simplifying vmstate_spk[].

Quoted 6 lines
---
 hw/audio/pcspk.c | 10 ----------
 1 file changed, 10 deletions(-)
Reviewed-by on v4 21/27
#95 lore raw

Re: [PATCH v4 21/27] hw/audio/pcspk: Remove PCSpkState::migrate field

In reply to Philippe Mathieu-Daudé
Quoted 4 lines · On Thu, 8 May 2025 15:35:44 +0200 Philippe Mathieu-Daudé <[email protected]>…

The PCSpkState::migrate boolean was only set in the pc_compat_2_7[] array, via the 'migrate=off' property. We removed all machines using that array, lets remove that property, simplifying vmstate_spk[].

same as 14/27, it should be safe to remove without deprecation

Quoted 47 lines
---
 hw/audio/pcspk.c | 10 ----------
 1 file changed, 10 deletions(-)
hw/audio/pcspk.c +0 -10
index a419161b5b1..0e83ba0bf73 100644
@@ -56,7 +56,6 @@ struct PCSpkState {
     unsigned int play_pos;
     uint8_t data_on;
     uint8_t dummy_refresh_clock;
-    bool migrate;
 };
 
 static const char *s_spk = "pcspk";
@@ -196,18 +195,10 @@ static void pcspk_realizefn(DeviceState *dev, Error **errp)
     pcspk_state = s;
 }
 
-static bool migrate_needed(void *opaque)
-{
-    PCSpkState *s = opaque;
-
-    return s->migrate;
-}
-
 static const VMStateDescription vmstate_spk = {
     .name = "pcspk",
     .version_id = 1,
     .minimum_version_id = 1,
-    .needed = migrate_needed,
     .fields = (const VMStateField[]) {
         VMSTATE_UINT8(data_on, PCSpkState),
         VMSTATE_UINT8(dummy_refresh_clock, PCSpkState),
@@ -218,7 +209,6 @@ static const VMStateDescription vmstate_spk = {
 static const Property pcspk_properties[] = {
     DEFINE_AUDIO_PROPERTIES(PCSpkState, card),
     DEFINE_PROP_UINT32("iobase", PCSpkState, iobase,  0x61),
-    DEFINE_PROP_BOOL("migrate", PCSpkState, migrate,  true),
 };
 
 static void pcspk_class_initfn(ObjectClass *klass, const void *data)
#96 lore raw

PATCH v4 22/27 hw/core/machine: Remove hw_compat_2_7[] array

In reply to Philippe Mathieu-Daudé

The hw_compat_2_7[] array was only used by the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Remove it.

---
 include/hw/boards.h | 3 ---
 hw/core/machine.c   | 9 ---------
 2 files changed, 12 deletions(-)
include/hw/boards.h +0 -3
index a881db8e7d6..77707c4376a 100644
@@ -838,7 +838,4 @@ extern const size_t hw_compat_2_9_len;
 extern GlobalProperty hw_compat_2_8[];
 extern const size_t hw_compat_2_8_len;
 
-extern GlobalProperty hw_compat_2_7[];
-extern const size_t hw_compat_2_7_len;
-
 #endif
hw/core/machine.c +0 -9
index ce98820f277..bde19a2ff67 100644
@@ -266,15 +266,6 @@ GlobalProperty hw_compat_2_8[] = {
 };
 const size_t hw_compat_2_8_len = G_N_ELEMENTS(hw_compat_2_8);
 
-GlobalProperty hw_compat_2_7[] = {
-    { "virtio-pci", "page-per-vq", "on" },
-    { "virtio-serial-device", "emergency-write", "off" },
-    { "ioapic", "version", "0x11" },
-    { "intel-iommu", "x-buggy-eim", "true" },
-    { "virtio-pci", "x-ignore-backend-features", "on" },
-};
-const size_t hw_compat_2_7_len = G_N_ELEMENTS(hw_compat_2_7);
-
 MachineState *current_machine;
 
 static char *machine_get_kernel(Object *obj, Error **errp)
Signature · 1 line
2.47.1
Reviewed-by on v4 22/27
#97 lore raw

Re: [PATCH v4 22/27] hw/core/machine: Remove hw_compat_2_7[] array

Quoted 8 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:45 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 22/27] hw/core/machine: Remove hw_compat_2_7[] array
X-Mailer: git-send-email 2.47.1

The hw_compat_2_7[] array was only used by the pc-q35-2.7 and pc-i440fx-2.7 machines, which got removed. Remove it.

Quoted 6 lines
---
 include/hw/boards.h | 3 ---
 hw/core/machine.c   | 9 ---------
 2 files changed, 12 deletions(-)
#98 lore raw

PATCH v4 23/27 hw/i386/intel_iommu: Remove IntelIOMMUState::buggy_eim field

In reply to Philippe Mathieu-Daudé

The IntelIOMMUState::buggy_eim boolean was only set in the hw_compat_2_7[] array, via the 'x-buggy-eim=true' property. We removed all machines using that array, lets remove that property, simplifying vtd_decide_config().

---
 include/hw/i386/intel_iommu.h | 1 -
 hw/i386/intel_iommu.c         | 5 ++---
 2 files changed, 2 insertions(+), 4 deletions(-)
include/hw/i386/intel_iommu.h +0 -1
index e95477e8554..29304329d05 100644
@@ -303,7 +303,6 @@ struct IntelIOMMUState {
     uint32_t intr_size;             /* Number of IR table entries */
     bool intr_eime;                 /* Extended interrupt mode enabled */
     OnOffAuto intr_eim;             /* Toggle for EIM cabability */
-    bool buggy_eim;                 /* Force buggy EIM unless eim=off */
     uint8_t aw_bits;                /* Host/IOVA address width (in bits) */
     bool dma_drain;                 /* Whether DMA r/w draining enabled */
     bool dma_translation;           /* Whether DMA translation supported */
hw/i386/intel_iommu.c +2 -3
index 5f8ed1243d1..c980cecb4ee 100644
@@ -3823,7 +3823,6 @@ static const Property vtd_properties[] = {
     DEFINE_PROP_UINT32("version", IntelIOMMUState, version, 0),
     DEFINE_PROP_ON_OFF_AUTO("eim", IntelIOMMUState, intr_eim,
                             ON_OFF_AUTO_AUTO),
-    DEFINE_PROP_BOOL("x-buggy-eim", IntelIOMMUState, buggy_eim, false),
     DEFINE_PROP_UINT8("aw-bits", IntelIOMMUState, aw_bits,
                       VTD_HOST_ADDRESS_WIDTH),
     DEFINE_PROP_BOOL("caching-mode", IntelIOMMUState, caching_mode, FALSE),
@@ -4731,11 +4730,11 @@ static bool vtd_decide_config(IntelIOMMUState *s, Error **errp)
     }
 
     if (s->intr_eim == ON_OFF_AUTO_AUTO) {
-        s->intr_eim = (kvm_irqchip_in_kernel() || s->buggy_eim)
+        s->intr_eim = kvm_irqchip_in_kernel()
                       && x86_iommu_ir_supported(x86_iommu) ?
                                               ON_OFF_AUTO_ON : ON_OFF_AUTO_OFF;
     }
-    if (s->intr_eim == ON_OFF_AUTO_ON && !s->buggy_eim) {
+    if (s->intr_eim == ON_OFF_AUTO_ON) {
         if (kvm_irqchip_is_split() && !kvm_enable_x2apic()) {
             error_setg(errp, "eim=on requires support on the KVM side"
                              "(X2APIC_API, first shipped in v4.7)");
Signature · 1 line
2.47.1
Reviewed-by on v4 23/27
#99 lore raw

Re: [PATCH v4 23/27] hw/i386/intel_iommu: Remove IntelIOMMUState::buggy_eim field

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:46 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 23/27] hw/i386/intel_iommu: Remove
IntelIOMMUState::buggy_eim field
X-Mailer: git-send-email 2.47.1

The IntelIOMMUState::buggy_eim boolean was only set in the hw_compat_2_7[] array, via the 'x-buggy-eim=true' property. We removed all machines using that array, lets remove that property, simplifying vtd_decide_config().

Quoted 6 lines
---
 include/hw/i386/intel_iommu.h | 1 -
 hw/i386/intel_iommu.c         | 5 ++---
 2 files changed, 2 insertions(+), 4 deletions(-)

This property has a good name :-)

Reviewed-by on v4 23/27
#100 lore raw

Re: [PATCH v4 23/27] hw/i386/intel_iommu: Remove IntelIOMMUState::buggy_eim field

In reply to Philippe Mathieu-Daudé
Quoted 7 lines · On Thu, 8 May 2025 15:35:46 +0200 Philippe Mathieu-Daudé <[email protected]>…

The IntelIOMMUState::buggy_eim boolean was only set in the hw_compat_2_7[] array, via the 'x-buggy-eim=true' property. We removed all machines using that array, lets remove that property, simplifying vtd_decide_config().

Quoted 43 lines
---
 include/hw/i386/intel_iommu.h | 1 -
 hw/i386/intel_iommu.c         | 5 ++---
 2 files changed, 2 insertions(+), 4 deletions(-)
include/hw/i386/intel_iommu.h +0 -1
index e95477e8554..29304329d05 100644
@@ -303,7 +303,6 @@ struct IntelIOMMUState {
     uint32_t intr_size;             /* Number of IR table entries */
     bool intr_eime;                 /* Extended interrupt mode enabled */
     OnOffAuto intr_eim;             /* Toggle for EIM cabability */
-    bool buggy_eim;                 /* Force buggy EIM unless eim=off */
     uint8_t aw_bits;                /* Host/IOVA address width (in bits) */
     bool dma_drain;                 /* Whether DMA r/w draining enabled */
     bool dma_translation;           /* Whether DMA translation supported */
hw/i386/intel_iommu.c +2 -3
index 5f8ed1243d1..c980cecb4ee 100644
@@ -3823,7 +3823,6 @@ static const Property vtd_properties[] = {
     DEFINE_PROP_UINT32("version", IntelIOMMUState, version, 0),
     DEFINE_PROP_ON_OFF_AUTO("eim", IntelIOMMUState, intr_eim,
                             ON_OFF_AUTO_AUTO),
-    DEFINE_PROP_BOOL("x-buggy-eim", IntelIOMMUState, buggy_eim, false),
     DEFINE_PROP_UINT8("aw-bits", IntelIOMMUState, aw_bits,
                       VTD_HOST_ADDRESS_WIDTH),
     DEFINE_PROP_BOOL("caching-mode", IntelIOMMUState, caching_mode, FALSE),
@@ -4731,11 +4730,11 @@ static bool vtd_decide_config(IntelIOMMUState *s, Error **errp)
     }
 
     if (s->intr_eim == ON_OFF_AUTO_AUTO) {
-        s->intr_eim = (kvm_irqchip_in_kernel() || s->buggy_eim)
+        s->intr_eim = kvm_irqchip_in_kernel()
                       && x86_iommu_ir_supported(x86_iommu) ?
                                               ON_OFF_AUTO_ON : ON_OFF_AUTO_OFF;
     }
-    if (s->intr_eim == ON_OFF_AUTO_ON && !s->buggy_eim) {
+    if (s->intr_eim == ON_OFF_AUTO_ON) {
         if (kvm_irqchip_is_split() && !kvm_enable_x2apic()) {
             error_setg(errp, "eim=on requires support on the KVM side"
                              "(X2APIC_API, first shipped in v4.7)");
#101 lore raw

PATCH v4 24/27 hw/intc/ioapic: Remove IOAPICCommonState::version field

In reply to Philippe Mathieu-Daudé

The IOAPICCommonState::version integer was only set in the hw_compat_2_7[] array, via the 'version=0x11' property. We removed all machines using that array, lets remove that property, simplify by only using the default version (defined as IOAPIC_VER_DEF).

For the record, this field was introduced in commit 20fd4b7b6d9 ("x86: ioapic: add support for explicit EOI"):

Quoted 14 lines

Some old Linux kernels (upstream before v4.0), or any released RHEL kernels has problem in sending APIC EOI when IR is enabled. Meanwhile, many of them only support explicit EOI for IOAPIC, which is only introduced in IOAPIC version 0x20. This patch provide a way to boost QEMU IOAPIC to version 0x20, in order for QEMU to correctly receive EOI messages.

Without boosting IOAPIC version to 0x20, kernels before commit d32932d ("x86/irq: Convert IOAPIC to use hierarchical irqdomain interfaces") will have trouble enabling both IR and level-triggered interrupt devices (like e1000).

To upgrade IOAPIC to version 0x20, we need to specify:

Quoted 8 lines
    -global ioapic.version=0x20

To be compatible with old systems, 0x11 will still be the default IOAPIC version. Here 0x11 and 0x20 are the only versions to be supported.

One thing to mention: this patch only applies to emulated IOAPIC. It does not affect kernel IOAPIC behavior.

---
 hw/intc/ioapic_internal.h |  3 +--
 hw/intc/ioapic.c          | 18 ++----------------
 hw/intc/ioapic_common.c   |  2 +-
 3 files changed, 4 insertions(+), 19 deletions(-)
hw/intc/ioapic_internal.h +1 -2
index 51205767f44..330ce195222 100644
@@ -82,7 +82,7 @@
 #define IOAPIC_ID_MASK                  0xf
 
 #define IOAPIC_VER_ENTRIES_SHIFT        16
-
+#define IOAPIC_VER_DEF                  0x20
 
 #define TYPE_IOAPIC_COMMON "ioapic-common"
 OBJECT_DECLARE_TYPE(IOAPICCommonState, IOAPICCommonClass, IOAPIC_COMMON)
@@ -104,7 +104,6 @@ struct IOAPICCommonState {
     uint32_t irr;
     uint64_t ioredtbl[IOAPIC_NUM_PINS];
     Notifier machine_done;
-    uint8_t version;
     uint64_t irq_count[IOAPIC_NUM_PINS];
     int irq_level[IOAPIC_NUM_PINS];
     int irq_eoi[IOAPIC_NUM_PINS];
hw/intc/ioapic.c +2 -16
index 133bef852d1..5cc97767d9d 100644
@@ -315,7 +315,7 @@ ioapic_mem_read(void *opaque, hwaddr addr, unsigned int size)
             val = s->id << IOAPIC_ID_SHIFT;
             break;
         case IOAPIC_REG_VER:
-            val = s->version |
+            val = IOAPIC_VER_DEF |
                 ((IOAPIC_NUM_PINS - 1) << IOAPIC_VER_ENTRIES_SHIFT);
             break;
         default:
@@ -411,8 +411,7 @@ ioapic_mem_write(void *opaque, hwaddr addr, uint64_t val,
         }
         break;
     case IOAPIC_EOI:
-        /* Explicit EOI is only supported for IOAPIC version 0x20 */
-        if (size != 4 || s->version != 0x20) {
+        if (size != 4) {
             break;
         }
         ioapic_eoi_broadcast(val);
@@ -444,18 +443,10 @@ static void ioapic_machine_done_notify(Notifier *notifier, void *data)
 #endif
 }
 
-#define IOAPIC_VER_DEF 0x20
-
 static void ioapic_realize(DeviceState *dev, Error **errp)
 {
     IOAPICCommonState *s = IOAPIC_COMMON(dev);
 
-    if (s->version != 0x11 && s->version != 0x20) {
-        error_setg(errp, "IOAPIC only supports version 0x11 or 0x20 "
-                   "(default: 0x%x).", IOAPIC_VER_DEF);
-        return;
-    }
-
     memory_region_init_io(&s->io_memory, OBJECT(s), &ioapic_io_ops, s,
                           "ioapic", 0x1000);
 
@@ -476,10 +467,6 @@ static void ioapic_unrealize(DeviceState *dev)
     timer_free(s->delayed_ioapic_service_timer);
 }
 
-static const Property ioapic_properties[] = {
-    DEFINE_PROP_UINT8("version", IOAPICCommonState, version, IOAPIC_VER_DEF),
-};
-
 static void ioapic_class_init(ObjectClass *klass, const void *data)
 {
     IOAPICCommonClass *k = IOAPIC_COMMON_CLASS(klass);
@@ -493,7 +480,6 @@ static void ioapic_class_init(ObjectClass *klass, const void *data)
      */
     k->post_load = ioapic_update_kvm_routes;
     device_class_set_legacy_reset(dc, ioapic_reset_common);
-    device_class_set_props(dc, ioapic_properties);
 }
 
 static const TypeInfo ioapic_info = {
hw/intc/ioapic_common.c +1 -1
index fce3486e519..8b3e2ba9384 100644
@@ -83,7 +83,7 @@ static void ioapic_print_redtbl(GString *buf, IOAPICCommonState *s)
     int i;
 
     g_string_append_printf(buf, "ioapic0: ver=0x%x id=0x%02x sel=0x%02x",
-                           s->version, s->id, s->ioregsel);
+                           IOAPIC_VER_DEF, s->id, s->ioregsel);
     if (s->ioregsel) {
         g_string_append_printf(buf, " (redir[%u])\n",
                                (s->ioregsel - IOAPIC_REG_REDTBL_BASE) >> 1);
Signature · 1 line
2.47.1
Reviewed-by on v4 24/27
#102 lore raw

Re: [PATCH v4 24/27] hw/intc/ioapic: Remove IOAPICCommonState::version field

Quoted 38 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:47 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 24/27] hw/intc/ioapic: Remove IOAPICCommonState::version
field
X-Mailer: git-send-email 2.47.1

The IOAPICCommonState::version integer was only set in the hw_compat_2_7[] array, via the 'version=0x11' property. We removed all machines using that array, lets remove that property, simplify by only using the default version (defined as IOAPIC_VER_DEF).

For the record, this field was introduced in commit 20fd4b7b6d9 ("x86: ioapic: add support for explicit EOI"):

Quoted 14 lines

Some old Linux kernels (upstream before v4.0), or any released RHEL kernels has problem in sending APIC EOI when IR is enabled. Meanwhile, many of them only support explicit EOI for IOAPIC, which is only introduced in IOAPIC version 0x20. This patch provide a way to boost QEMU IOAPIC to version 0x20, in order for QEMU to correctly receive EOI messages.

Without boosting IOAPIC version to 0x20, kernels before commit d32932d ("x86/irq: Convert IOAPIC to use hierarchical irqdomain interfaces") will have trouble enabling both IR and level-triggered interrupt devices (like e1000).

To upgrade IOAPIC to version 0x20, we need to specify:

Quoted 8 lines
    -global ioapic.version=0x20

To be compatible with old systems, 0x11 will still be the default IOAPIC version. Here 0x11 and 0x20 are the only versions to be supported.

One thing to mention: this patch only applies to emulated IOAPIC. It does not affect kernel IOAPIC behavior.

Quoted 7 lines
---
 hw/intc/ioapic_internal.h |  3 +--
 hw/intc/ioapic.c          | 18 ++----------------
 hw/intc/ioapic_common.c   |  2 +-
 3 files changed, 4 insertions(+), 19 deletions(-)
#103 lore raw

Re: [PATCH v4 24/27] hw/intc/ioapic: Remove IOAPICCommonState::version field

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · On Thu, 8 May 2025 15:35:47 +0200 Philippe Mathieu-Daudé <[email protected]>…

The IOAPICCommonState::version integer was only set in the hw_compat_2_7[] array, via the 'version=0x11' property. We removed all machines using that array, lets remove that property, simplify by only using the default version (defined as IOAPIC_VER_DEF).

Quoted 19 lines

For the record, this field was introduced in commit 20fd4b7b6d9 ("x86: ioapic: add support for explicit EOI"):

Quoted 16 lines

Some old Linux kernels (upstream before v4.0), or any released RHEL kernels has problem in sending APIC EOI when IR is enabled. Meanwhile, many of them only support explicit EOI for IOAPIC, which is only introduced in IOAPIC version 0x20. This patch provide a way to boost QEMU IOAPIC to version 0x20, in order for QEMU to correctly receive EOI messages.

Without boosting IOAPIC version to 0x20, kernels before commit d32932d ("x86/irq: Convert IOAPIC to use hierarchical irqdomain interfaces") will have trouble enabling both IR and level-triggered interrupt devices (like e1000).

To upgrade IOAPIC to version 0x20, we need to specify:

    -global ioapic.version=0x20

that crutch might be in-use by external users, and even if they use 0x20, removing property will break CLI.

I'd deprecate it first and then remove.

Quoted 3 lines
Quoted 3 lines

To be compatible with old systems, 0x11 will still be the default IOAPIC version. Here 0x11 and 0x20 are the only versions to be supported.

looking at the code removed, default is 0x20 which doesn't match above statement. Have something changed between then and now (missing ref to 0x20 becoming default)?

Quoted 107 lines
Quoted 3 lines

One thing to mention: this patch only applies to emulated IOAPIC. It does not affect kernel IOAPIC behavior.

---
 hw/intc/ioapic_internal.h |  3 +--
 hw/intc/ioapic.c          | 18 ++----------------
 hw/intc/ioapic_common.c   |  2 +-
 3 files changed, 4 insertions(+), 19 deletions(-)
hw/intc/ioapic_internal.h +1 -2
index 51205767f44..330ce195222 100644
@@ -82,7 +82,7 @@
 #define IOAPIC_ID_MASK                  0xf
 
 #define IOAPIC_VER_ENTRIES_SHIFT        16
-
+#define IOAPIC_VER_DEF                  0x20
 
 #define TYPE_IOAPIC_COMMON "ioapic-common"
 OBJECT_DECLARE_TYPE(IOAPICCommonState, IOAPICCommonClass, IOAPIC_COMMON)
@@ -104,7 +104,6 @@ struct IOAPICCommonState {
     uint32_t irr;
     uint64_t ioredtbl[IOAPIC_NUM_PINS];
     Notifier machine_done;
-    uint8_t version;
     uint64_t irq_count[IOAPIC_NUM_PINS];
     int irq_level[IOAPIC_NUM_PINS];
     int irq_eoi[IOAPIC_NUM_PINS];
hw/intc/ioapic.c +2 -16
index 133bef852d1..5cc97767d9d 100644
@@ -315,7 +315,7 @@ ioapic_mem_read(void *opaque, hwaddr addr, unsigned int size)
             val = s->id << IOAPIC_ID_SHIFT;
             break;
         case IOAPIC_REG_VER:
-            val = s->version |
+            val = IOAPIC_VER_DEF |
                 ((IOAPIC_NUM_PINS - 1) << IOAPIC_VER_ENTRIES_SHIFT);
             break;
         default:
@@ -411,8 +411,7 @@ ioapic_mem_write(void *opaque, hwaddr addr, uint64_t val,
         }
         break;
     case IOAPIC_EOI:
-        /* Explicit EOI is only supported for IOAPIC version 0x20 */
-        if (size != 4 || s->version != 0x20) {
+        if (size != 4) {
             break;
         }
         ioapic_eoi_broadcast(val);
@@ -444,18 +443,10 @@ static void ioapic_machine_done_notify(Notifier *notifier, void *data)
 #endif
 }
 
-#define IOAPIC_VER_DEF 0x20
-
 static void ioapic_realize(DeviceState *dev, Error **errp)
 {
     IOAPICCommonState *s = IOAPIC_COMMON(dev);
 
-    if (s->version != 0x11 && s->version != 0x20) {
-        error_setg(errp, "IOAPIC only supports version 0x11 or 0x20 "
-                   "(default: 0x%x).", IOAPIC_VER_DEF);
-        return;
-    }
-
     memory_region_init_io(&s->io_memory, OBJECT(s), &ioapic_io_ops, s,
                           "ioapic", 0x1000);
 
@@ -476,10 +467,6 @@ static void ioapic_unrealize(DeviceState *dev)
     timer_free(s->delayed_ioapic_service_timer);
 }
 
-static const Property ioapic_properties[] = {
-    DEFINE_PROP_UINT8("version", IOAPICCommonState, version, IOAPIC_VER_DEF),
-};
-
 static void ioapic_class_init(ObjectClass *klass, const void *data)
 {
     IOAPICCommonClass *k = IOAPIC_COMMON_CLASS(klass);
@@ -493,7 +480,6 @@ static void ioapic_class_init(ObjectClass *klass, const void *data)
      */
     k->post_load = ioapic_update_kvm_routes;
     device_class_set_legacy_reset(dc, ioapic_reset_common);
-    device_class_set_props(dc, ioapic_properties);
 }
 
 static const TypeInfo ioapic_info = {
hw/intc/ioapic_common.c +1 -1
index fce3486e519..8b3e2ba9384 100644
@@ -83,7 +83,7 @@ static void ioapic_print_redtbl(GString *buf, IOAPICCommonState *s)
     int i;
 
     g_string_append_printf(buf, "ioapic0: ver=0x%x id=0x%02x sel=0x%02x",
-                           s->version, s->id, s->ioregsel);
+                           IOAPIC_VER_DEF, s->id, s->ioregsel);
     if (s->ioregsel) {
         g_string_append_printf(buf, " (redir[%u])\n",
                                (s->ioregsel - IOAPIC_REG_REDTBL_BASE) >> 1);
#104 lore raw

PATCH v4 25/27 hw/virtio/virtio-pci: Remove VirtIOPCIProxy::ignore_backend_features field

In reply to Philippe Mathieu-Daudé

The VirtIOPCIProxy::ignore_backend_features boolean was only set in the hw_compat_2_7[] array, via the 'x-ignore-backend-features=on' property. We removed all machines using that array, lets remove that property, simplify by only using the default version.

---
 include/hw/virtio/virtio-pci.h | 1 -
 hw/virtio/virtio-pci.c         | 5 +----
 2 files changed, 1 insertion(+), 5 deletions(-)
include/hw/virtio/virtio-pci.h +0 -1
index f962c9116c1..9838e8650a6 100644
@@ -149,7 +149,6 @@ struct VirtIOPCIProxy {
     int config_cap;
     uint32_t flags;
     bool disable_modern;
-    bool ignore_backend_features;
     OnOffAuto disable_legacy;
     /* Transitional device id */
     uint16_t trans_devid;
hw/virtio/virtio-pci.c +1 -4
index 8d68e56641a..7c965771907 100644
@@ -1965,8 +1965,7 @@ static void virtio_pci_device_plugged(DeviceState *d, Error **errp)
      * Virtio capabilities present without
      * VIRTIO_F_VERSION_1 confuses guests
      */
-    if (!proxy->ignore_backend_features &&
-            !virtio_has_feature(vdev->host_features, VIRTIO_F_VERSION_1)) {
+    if (!virtio_has_feature(vdev->host_features, VIRTIO_F_VERSION_1)) {
         virtio_pci_disable_modern(proxy);
 
         if (!legacy) {
@@ -2351,8 +2350,6 @@ static const Property virtio_pci_properties[] = {
                     VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY_BIT, false),
     DEFINE_PROP_BIT("page-per-vq", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_PAGE_PER_VQ_BIT, false),
-    DEFINE_PROP_BOOL("x-ignore-backend-features", VirtIOPCIProxy,
-                     ignore_backend_features, false),
     DEFINE_PROP_BIT("ats", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_ATS_BIT, false),
     DEFINE_PROP_BIT("x-ats-page-aligned", VirtIOPCIProxy, flags,
Signature · 1 line
2.47.1
Reviewed-by on v4 25/27
#105 lore raw

Re: [PATCH v4 25/27] hw/virtio/virtio-pci: Remove VirtIOPCIProxy::ignore_backend_features field

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:48 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 25/27] hw/virtio/virtio-pci: Remove
VirtIOPCIProxy::ignore_backend_features field
X-Mailer: git-send-email 2.47.1

The VirtIOPCIProxy::ignore_backend_features boolean was only set in the hw_compat_2_7[] array, via the 'x-ignore-backend-features=on' property. We removed all machines using that array, lets remove that property, simplify by only using the default version.

Quoted 6 lines
---
 include/hw/virtio/virtio-pci.h | 1 -
 hw/virtio/virtio-pci.c         | 5 +----
 2 files changed, 1 insertion(+), 5 deletions(-)
Reviewed-by on v4 25/27
#106 lore raw

Re: [PATCH v4 25/27] hw/virtio/virtio-pci: Remove VirtIOPCIProxy::ignore_backend_features field

In reply to Philippe Mathieu-Daudé
Quoted 7 lines · On Thu, 8 May 2025 15:35:48 +0200 Philippe Mathieu-Daudé <[email protected]>…

The VirtIOPCIProxy::ignore_backend_features boolean was only set in the hw_compat_2_7[] array, via the 'x-ignore-backend-features=on' property. We removed all machines using that array, lets remove that property, simplify by only using the default version.

Quoted 40 lines
---
 include/hw/virtio/virtio-pci.h | 1 -
 hw/virtio/virtio-pci.c         | 5 +----
 2 files changed, 1 insertion(+), 5 deletions(-)
include/hw/virtio/virtio-pci.h +0 -1
index f962c9116c1..9838e8650a6 100644
@@ -149,7 +149,6 @@ struct VirtIOPCIProxy {
     int config_cap;
     uint32_t flags;
     bool disable_modern;
-    bool ignore_backend_features;
     OnOffAuto disable_legacy;
     /* Transitional device id */
     uint16_t trans_devid;
hw/virtio/virtio-pci.c +1 -4
index 8d68e56641a..7c965771907 100644
@@ -1965,8 +1965,7 @@ static void virtio_pci_device_plugged(DeviceState *d, Error **errp)
      * Virtio capabilities present without
      * VIRTIO_F_VERSION_1 confuses guests
      */
-    if (!proxy->ignore_backend_features &&
-            !virtio_has_feature(vdev->host_features, VIRTIO_F_VERSION_1)) {
+    if (!virtio_has_feature(vdev->host_features, VIRTIO_F_VERSION_1)) {
         virtio_pci_disable_modern(proxy);
 
         if (!legacy) {
@@ -2351,8 +2350,6 @@ static const Property virtio_pci_properties[] = {
                     VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY_BIT, false),
     DEFINE_PROP_BIT("page-per-vq", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_PAGE_PER_VQ_BIT, false),
-    DEFINE_PROP_BOOL("x-ignore-backend-features", VirtIOPCIProxy,
-                     ignore_backend_features, false),
     DEFINE_PROP_BIT("ats", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_ATS_BIT, false),
     DEFINE_PROP_BIT("x-ats-page-aligned", VirtIOPCIProxy, flags,
#107 lore raw

PATCH v4 26/27 hw/char/virtio-serial: Do not expose the 'emergency-write' property

In reply to Philippe Mathieu-Daudé

The VIRTIO_CONSOLE_F_EMERG_WRITE feature bit was only set in the hw_compat_2_7[] array, via the 'emergency-write=off' property. We removed all machines using that array, lets remove that property. All instances have this feature bit set and it can not be disabled. VirtIOSerial::host_features mask is now unused, remove it.

---
 include/hw/virtio/virtio-serial.h | 2 --
 hw/char/virtio-serial-bus.c       | 9 +++------
 2 files changed, 3 insertions(+), 8 deletions(-)
include/hw/virtio/virtio-serial.h +0 -2
index d87c62eab7a..e6ceacec309 100644
@@ -185,8 +185,6 @@ struct VirtIOSerial {
     struct VirtIOSerialPostLoad *post_load;
 
     virtio_serial_conf serial;
-
-    uint64_t host_features;
 };
 
 /* Interface to the virtio-serial bus */
hw/char/virtio-serial-bus.c +3 -6
index eb79f5258b6..cfc8fa42186 100644
@@ -557,7 +557,7 @@ static uint64_t get_features(VirtIODevice *vdev, uint64_t features,
 
     vser = VIRTIO_SERIAL(vdev);
 
-    features |= vser->host_features;
+    features |= BIT_ULL(VIRTIO_CONSOLE_F_EMERG_WRITE);
     if (vser->bus.max_nr_ports > 1) {
         virtio_add_feature(&features, VIRTIO_CONSOLE_F_MULTIPORT);
     }
@@ -587,8 +587,7 @@ static void set_config(VirtIODevice *vdev, const uint8_t *config_data)
     VirtIOSerialPortClass *vsc;
     uint8_t emerg_wr_lo;
 
-    if (!virtio_has_feature(vser->host_features,
-        VIRTIO_CONSOLE_F_EMERG_WRITE) || !config->emerg_wr) {
+    if (!config->emerg_wr) {
         return;
     }
 
@@ -1039,7 +1038,7 @@ static void virtio_serial_device_realize(DeviceState *dev, Error **errp)
         return;
     }
 
-    if (!virtio_has_feature(vser->host_features,
+    if (!virtio_has_feature(vdev->host_features,
                             VIRTIO_CONSOLE_F_EMERG_WRITE)) {
         config_size = offsetof(struct virtio_console_config, emerg_wr);
     }
@@ -1155,8 +1154,6 @@ static const VMStateDescription vmstate_virtio_console = {
 static const Property virtio_serial_properties[] = {
     DEFINE_PROP_UINT32("max_ports", VirtIOSerial, serial.max_virtserial_ports,
                                                   31),
-    DEFINE_PROP_BIT64("emergency-write", VirtIOSerial, host_features,
-                      VIRTIO_CONSOLE_F_EMERG_WRITE, true),
 };
 
 static void virtio_serial_class_init(ObjectClass *klass, const void *data)
Signature · 1 line
2.47.1
Reviewed-by on v4 26/27
#108 lore raw

Re: [PATCH v4 26/27] hw/char/virtio-serial: Do not expose the 'emergency-write' property

Quoted 59 lines · Philippe Mathieu-Daudé

The VIRTIO_CONSOLE_F_EMERG_WRITE feature bit was only set in the hw_compat_2_7[] array, via the 'emergency-write=off' property. We removed all machines using that array, lets remove that property. All instances have this feature bit set and it can not be disabled. VirtIOSerial::host_features mask is now unused, remove it.

---
  include/hw/virtio/virtio-serial.h | 2 --
  hw/char/virtio-serial-bus.c       | 9 +++------
  2 files changed, 3 insertions(+), 8 deletions(-)
include/hw/virtio/virtio-serial.h +0 -2
index d87c62eab7a..e6ceacec309 100644
@@ -185,8 +185,6 @@ struct VirtIOSerial {
      struct VirtIOSerialPostLoad *post_load;
 
      virtio_serial_conf serial;
-
-    uint64_t host_features;
  };
 
  /* Interface to the virtio-serial bus */
hw/char/virtio-serial-bus.c +3 -4
index eb79f5258b6..cfc8fa42186 100644
@@ -557,7 +557,7 @@ static uint64_t get_features(VirtIODevice *vdev, uint64_t features,
 
      vser = VIRTIO_SERIAL(vdev);
 
-    features |= vser->host_features;
+    features |= BIT_ULL(VIRTIO_CONSOLE_F_EMERG_WRITE);
      if (vser->bus.max_nr_ports > 1) {
          virtio_add_feature(&features, VIRTIO_CONSOLE_F_MULTIPORT);
      }
@@ -587,8 +587,7 @@ static void set_config(VirtIODevice *vdev, const uint8_t *config_data)
      VirtIOSerialPortClass *vsc;
      uint8_t emerg_wr_lo;
 
-    if (!virtio_has_feature(vser->host_features,
-        VIRTIO_CONSOLE_F_EMERG_WRITE) || !config->emerg_wr) {
+    if (!config->emerg_wr) {
          return;
      }
 
@@ -1039,7 +1038,7 @@ static void virtio_serial_device_realize(DeviceState *dev, Error **errp)
          return;
      }
 
-    if (!virtio_has_feature(vser->host_features,
+    if (!virtio_has_feature(vdev->host_features,
                              VIRTIO_CONSOLE_F_EMERG_WRITE)) {
          config_size = offsetof(struct virtio_console_config, emerg_wr);
      }

hw/char/virtio-serial-bus.c static const VMStateDescription vmstate_virtio_console = { · line 1159

Quoted 8 lines
hw/char/virtio-serial-bus.c +0 -2
@@ -1155,8 +1154,6 @@ static const VMStateDescription vmstate_virtio_console = {
  static const Property virtio_serial_properties[] = {
      DEFINE_PROP_UINT32("max_ports", VirtIOSerial, serial.max_virtserial_ports,
                                                    31),
-    DEFINE_PROP_BIT64("emergency-write", VirtIOSerial, host_features,
-                      VIRTIO_CONSOLE_F_EMERG_WRITE, true),
  };
 
  static void virtio_serial_class_init(ObjectClass *klass, const void *data)

LGTM.

ATB,

Mark.

Reviewed-by on v4 26/27
#109 lore raw

Re: [PATCH v4 26/27] hw/char/virtio-serial: Do not expose the 'emergency-write' property

In reply to Philippe Mathieu-Daudé
Quoted 13 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:49 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 26/27] hw/char/virtio-serial: Do not expose the
'emergency-write' property
X-Mailer: git-send-email 2.47.1

The VIRTIO_CONSOLE_F_EMERG_WRITE feature bit was only set in the hw_compat_2_7[] array, via the 'emergency-write=off' property. We removed all machines using that array, lets remove that property. All instances have this feature bit set and it can not be disabled. VirtIOSerial::host_features mask is now unused, remove it.

Quoted 5 lines
---
 include/hw/virtio/virtio-serial.h | 2 --
 hw/char/virtio-serial-bus.c       | 9 +++------
 2 files changed, 3 insertions(+), 8 deletions(-)
#110 lore raw

PATCH v4 27/27 hw/virtio/virtio-pci: Remove VIRTIO_PCI_FLAG_PAGE_PER_VQ definition

In reply to Philippe Mathieu-Daudé

VIRTIO_PCI_FLAG_PAGE_PER_VQ was only used by the hw_compat_2_7[] array, via the 'page-per-vq=on' property. We removed all machines using that array, lets remove all the code around VIRTIO_PCI_FLAG_PAGE_PER_VQ (see commit 9a4c0e220d8 for similar VIRTIO_PCI_FLAG_* enum removal).

---
 include/hw/virtio/virtio-pci.h |  1 -
 hw/display/virtio-vga.c        | 10 ----------
 hw/virtio/virtio-pci.c         |  7 +------
 3 files changed, 1 insertion(+), 17 deletions(-)
include/hw/virtio/virtio-pci.h +0 -1
index 9838e8650a6..8abc5f8f20d 100644
@@ -33,7 +33,6 @@ enum {
     VIRTIO_PCI_FLAG_BUS_MASTER_BUG_MIGRATION_BIT,
     VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT,
     VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY_BIT,
-    VIRTIO_PCI_FLAG_PAGE_PER_VQ_BIT,
     VIRTIO_PCI_FLAG_ATS_BIT,
     VIRTIO_PCI_FLAG_INIT_DEVERR_BIT,
     VIRTIO_PCI_FLAG_INIT_LNKCTL_BIT,
hw/display/virtio-vga.c +0 -10
index 40e60f70fcd..83d01f089b5 100644
@@ -141,16 +141,6 @@ static void virtio_vga_base_realize(VirtIOPCIProxy *vpci_dev, Error **errp)
                                VIRTIO_GPU_SHM_ID_HOST_VISIBLE);
     }
 
-    if (!(vpci_dev->flags & VIRTIO_PCI_FLAG_PAGE_PER_VQ)) {
-        /*
-         * with page-per-vq=off there is no padding space we can use
-         * for the stdvga registers.  Make the common and isr regions
-         * smaller then.
-         */
-        vpci_dev->common.size /= 2;
-        vpci_dev->isr.size /= 2;
-    }
-
     offset = memory_region_size(&vpci_dev->modern_bar);
     offset -= vpci_dev->notify.size;
     vpci_dev->notify.offset = offset;
hw/virtio/virtio-pci.c +1 -6
index 7c965771907..4e0d4bda6ed 100644
@@ -314,12 +314,9 @@ static bool virtio_pci_ioeventfd_enabled(DeviceState *d)
     return (proxy->flags & VIRTIO_PCI_FLAG_USE_IOEVENTFD) != 0;
 }
 
-#define QEMU_VIRTIO_PCI_QUEUE_MEM_MULT 0x1000
-
 static inline int virtio_pci_queue_mem_mult(struct VirtIOPCIProxy *proxy)
 {
-    return (proxy->flags & VIRTIO_PCI_FLAG_PAGE_PER_VQ) ?
-        QEMU_VIRTIO_PCI_QUEUE_MEM_MULT : 4;
+    return 4;
 }
 
 static int virtio_pci_ioeventfd_assign(DeviceState *d, EventNotifier *notifier,
@@ -2348,8 +2345,6 @@ static const Property virtio_pci_properties[] = {
                     VIRTIO_PCI_FLAG_BUS_MASTER_BUG_MIGRATION_BIT, false),
     DEFINE_PROP_BIT("modern-pio-notify", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY_BIT, false),
-    DEFINE_PROP_BIT("page-per-vq", VirtIOPCIProxy, flags,
-                    VIRTIO_PCI_FLAG_PAGE_PER_VQ_BIT, false),
     DEFINE_PROP_BIT("ats", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_ATS_BIT, false),
     DEFINE_PROP_BIT("x-ats-page-aligned", VirtIOPCIProxy, flags,
Signature · 1 line
2.47.1
Reviewed-by on v4 27/27
#111 lore raw

Re: [PATCH v4 27/27] hw/virtio/virtio-pci: Remove VIRTIO_PCI_FLAG_PAGE_PER_VQ definition

Quoted 12 lines · Philippe Mathieu-Daudé

Date: Thu, 8 May 2025 15:35:50 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v4 27/27] hw/virtio/virtio-pci: Remove
VIRTIO_PCI_FLAG_PAGE_PER_VQ definition
X-Mailer: git-send-email 2.47.1

VIRTIO_PCI_FLAG_PAGE_PER_VQ was only used by the hw_compat_2_7[] array, via the 'page-per-vq=on' property. We removed all machines using that array, lets remove all the code around VIRTIO_PCI_FLAG_PAGE_PER_VQ (see commit 9a4c0e220d8 for similar VIRTIO_PCI_FLAG_* enum removal).

Quoted 7 lines
---
 include/hw/virtio/virtio-pci.h |  1 -
 hw/display/virtio-vga.c        | 10 ----------
 hw/virtio/virtio-pci.c         |  7 +------
 3 files changed, 1 insertion(+), 17 deletions(-)
#112 lore raw

Re: [PATCH v4 27/27] hw/virtio/virtio-pci: Remove VIRTIO_PCI_FLAG_PAGE_PER_VQ definition

In reply to Philippe Mathieu-Daudé
Quoted 19 lines · On Thu, 8 May 2025 15:35:50 +0200 Philippe Mathieu-Daudé <[email protected]>…

VIRTIO_PCI_FLAG_PAGE_PER_VQ was only used by the hw_compat_2_7[] array, via the 'page-per-vq=on' property. We removed all machines using that array, lets remove all the code around VIRTIO_PCI_FLAG_PAGE_PER_VQ (see commit 9a4c0e220d8 for similar VIRTIO_PCI_FLAG_* enum removal).

---
 include/hw/virtio/virtio-pci.h |  1 -
 hw/display/virtio-vga.c        | 10 ----------
 hw/virtio/virtio-pci.c         |  7 +------
 3 files changed, 1 insertion(+), 17 deletions(-)
include/hw/virtio/virtio-pci.h +0 -0
index 9838e8650a6..8abc5f8f20d 100644

include/hw/virtio/virtio-pci.h enum { · line 36

Quoted 4 lines
include/hw/virtio/virtio-pci.h +0 -1
@@ -33,7 +33,6 @@ enum {
     VIRTIO_PCI_FLAG_BUS_MASTER_BUG_MIGRATION_BIT,
     VIRTIO_PCI_FLAG_USE_IOEVENTFD_BIT,
     VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY_BIT,
-    VIRTIO_PCI_FLAG_PAGE_PER_VQ_BIT,

there is also

/* page per vq flag to be used by split drivers within guests */

#define VIRTIO_PCI_FLAG_PAGE_PER_VQ \                                            
    (1 << VIRTIO_PCI_FLAG_PAGE_PER_VQ_BIT) 

left behind

nevertheless, I'm not sure if we should remove this at all, it seems that there are external users
https://bugzilla.redhat.com/show_bug.cgi?id=1925363

Quoted 51 lines
include/hw/virtio/virtio-pci.h +0 -0
     VIRTIO_PCI_FLAG_ATS_BIT,
     VIRTIO_PCI_FLAG_INIT_DEVERR_BIT,
     VIRTIO_PCI_FLAG_INIT_LNKCTL_BIT,
hw/display/virtio-vga.c +0 -10
index 40e60f70fcd..83d01f089b5 100644
@@ -141,16 +141,6 @@ static void virtio_vga_base_realize(VirtIOPCIProxy *vpci_dev, Error **errp)
                                VIRTIO_GPU_SHM_ID_HOST_VISIBLE);
     }
 
-    if (!(vpci_dev->flags & VIRTIO_PCI_FLAG_PAGE_PER_VQ)) {
-        /*
-         * with page-per-vq=off there is no padding space we can use
-         * for the stdvga registers.  Make the common and isr regions
-         * smaller then.
-         */
-        vpci_dev->common.size /= 2;
-        vpci_dev->isr.size /= 2;
-    }
-
     offset = memory_region_size(&vpci_dev->modern_bar);
     offset -= vpci_dev->notify.size;
     vpci_dev->notify.offset = offset;
hw/virtio/virtio-pci.c +1 -6
index 7c965771907..4e0d4bda6ed 100644
@@ -314,12 +314,9 @@ static bool virtio_pci_ioeventfd_enabled(DeviceState *d)
     return (proxy->flags & VIRTIO_PCI_FLAG_USE_IOEVENTFD) != 0;
 }
 
-#define QEMU_VIRTIO_PCI_QUEUE_MEM_MULT 0x1000
-
 static inline int virtio_pci_queue_mem_mult(struct VirtIOPCIProxy *proxy)
 {
-    return (proxy->flags & VIRTIO_PCI_FLAG_PAGE_PER_VQ) ?
-        QEMU_VIRTIO_PCI_QUEUE_MEM_MULT : 4;
+    return 4;
 }
 
 static int virtio_pci_ioeventfd_assign(DeviceState *d, EventNotifier *notifier,
@@ -2348,8 +2345,6 @@ static const Property virtio_pci_properties[] = {
                     VIRTIO_PCI_FLAG_BUS_MASTER_BUG_MIGRATION_BIT, false),
     DEFINE_PROP_BIT("modern-pio-notify", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_MODERN_PIO_NOTIFY_BIT, false),
-    DEFINE_PROP_BIT("page-per-vq", VirtIOPCIProxy, flags,
-                    VIRTIO_PCI_FLAG_PAGE_PER_VQ_BIT, false),
     DEFINE_PROP_BIT("ats", VirtIOPCIProxy, flags,
                     VIRTIO_PCI_FLAG_ATS_BIT, false),
     DEFINE_PROP_BIT("x-ats-page-aligned", VirtIOPCIProxy, flags,
#113 lore raw
In reply to Philippe Mathieu-Daudé
Quoted 10 lines · On Thu, 8 May 2025 15:35:23 +0200 Philippe Mathieu-Daudé <[email protected]>…

Since v3:
- Addressed Thomas and Zhao review comments
- Rename fw_cfg_init_mem_[no]dma() helpers
- Remove unused CPU properties
- Remove {multi,linux}boot.bin
- Added R-b tags

Since v2:
- Addressed Mark review comments and added his R-b tags

Quoted 7 lines

The versioned 'pc' and 'q35' machines up to 2.12 been marked as deprecated two releases ago, and are older than 6 years, so according to our support policy we can remove them.

This series only includes the 2.6 and 2.7 machines removal, as it is a big enough number of LoC removed. Rest will follow.

CCing libvirt folks

series removes some properties that has been used as compat knobs with 2.6/2.7 machine types that are being removed.

However libvirt might still use them,
please check if being removed properties are safe to remove as is | should be deprecated 1st | should be left alone from an immediate user perspective.

Quoted 3 lines

Based-on: <[email protected]>

[...]

#114 lore raw
Quoted 23 lines · Igor Mammedov
Quoted 10 lines · On Thu, 8 May 2025 15:35:23 +0200 Philippe Mathieu-Daudé <[email protected]>…

Since v3:
- Addressed Thomas and Zhao review comments
- Rename fw_cfg_init_mem_[no]dma() helpers
- Remove unused CPU properties
- Remove {multi,linux}boot.bin
- Added R-b tags

Since v2:
- Addressed Mark review comments and added his R-b tags

Quoted 7 lines

The versioned 'pc' and 'q35' machines up to 2.12 been marked as deprecated two releases ago, and are older than 6 years, so according to our support policy we can remove them.

This series only includes the 2.6 and 2.7 machines removal, as it is a big enough number of LoC removed. Rest will follow.

CCing libvirt folks

Quoted 7 lines

series removes some properties that has been used as compat knobs with 2.6/2.7 machine types that are being removed.

However libvirt might still use them,
please check if being removed properties are safe to remove as is | should be deprecated 1st | should be left alone from an immediate user perspective.

Anyone on libvirt side can confirm please?

Quoted 4 lines
Quoted 3 lines

Based-on: <[email protected]>

[...]

#115 lore raw
Quoted 27 lines · Michael S. Tsirkin via Devel
Quoted 26 lines · Igor Mammedov
Quoted 10 lines · On Thu, 8 May 2025 15:35:23 +0200 Philippe Mathieu-Daudé <[email protected]>…

Since v3:
- Addressed Thomas and Zhao review comments
- Rename fw_cfg_init_mem_[no]dma() helpers
- Remove unused CPU properties
- Remove {multi,linux}boot.bin
- Added R-b tags

Since v2:
- Addressed Mark review comments and added his R-b tags

Quoted 7 lines

The versioned 'pc' and 'q35' machines up to 2.12 been marked as deprecated two releases ago, and are older than 6 years, so according to our support policy we can remove them.

This series only includes the 2.6 and 2.7 machines removal, as it is a big enough number of LoC removed. Rest will follow.

CCing libvirt folks

series removes some properties that has been used as compat knobs with 2.6/2.7 machine types that are being removed.

Quoted 6 lines
Quoted 4 lines

However libvirt might still use them,
please check if being removed properties are safe to remove as is | should be deprecated 1st | should be left alone from an immediate user perspective.

Anyone on libvirt side can confirm please?

Sorry, it was a bit of a chore to go through the patches.

IIUC the following list is to be reviewed:

DEFINE_PROP_BOOL("x-buggy-eim", IntelIOMMUState, buggy_eim, false) DEFINE_PROP_BOOL("x-ignore-backend-features", VirtIOPCIProxy, DEFINE_PROP_BOOL("migrate", PCSpkState, migrate, true), DEFINE_PROP_UINT8("version", IOAPICCommonState, version, IOAPIC_VER_DEF) DEFINE_PROP_BIT64("emergency-write", VirtIOSerial, host_features DEFINE_PROP_BOOL("format_transport_address", VirtIOMMIOProxy, DEFINE_PROP_BOOL("legacy-instance-id", APICCommonState, legacy_instance_id,

None of the above are exposed by libvirt to the user nor used by libvirt internally.

DEFINE_PROP_BIT("page-per-vq", VirtIOPCIProxy, flags

This property is exposed to users for all virtio devices as 'page_per_vq' property:a

https://www.libvirt.org/formatdomain.html#virtio-related-options

It is an optional property and not formatted to qemu unless explicitly requested by the user. Removing it thus should be fine without the need to change anything in libvirt

The rest are CPU properties

DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true), DEFINE_PROP_BOOL("fill-mtrr-mask", X86CPU, fill_mtrr_mask, true), DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),

which appear in our cpu test data but I'm not 100% sure how they are used. Jirka, can you please comment?

Finally there's

DEFINE_PROP_BOOL("l3-cache", X86CPU, enable_l3_cache, true),

which is exposed to the users via cache mode setting of cpu:

https://www.libvirt.org/formatdomain.html#cpu-model-and-topology

look for 'cache'.

Thus from libvirt's side 'page-per-vq' and 'l3-cache' will likely require deprecation period. The rest except for CPU is fine to remove without anything at least from our PoV.

Jirka will hopefully clarify the Cpu property side.

#116 lore raw
Quoted 8 lines · Peter Krempa

The rest are CPU properties

DEFINE_PROP_BOOL("cpuid-0xb", X86CPU, enable_cpuid_0xb, true), DEFINE_PROP_BOOL("fill-mtrr-mask", X86CPU, fill_mtrr_mask, true), DEFINE_PROP_BOOL("full-cpuid-auto-level", X86CPU, full_cpuid_auto_level, true),

which appear in our cpu test data but I'm not 100% sure how they are used. Jirka, can you please comment?

None of these properties is directly used by libvirt for anything. They can be removed anytime.

Jirka

#117 lore raw
In reply to Peter Krempa

Hi Peter,

Quoted 9 lines

Finally there's

DEFINE_PROP_BOOL("l3-cache", X86CPU, enable_l3_cache, true),

which is exposed to the users via cache mode setting of cpu:

https://www.libvirt.org/formatdomain.html#cpu-model-and-topology

look for 'cache'.

I found this link doesn't mention "l3-cache", but it appears in libvirt's src/qemu/qemu_command.c.

Quoted 3 lines

Thus from libvirt's side 'page-per-vq' and 'l3-cache' will likely require deprecation period. The rest except for CPU is fine to remove without anything at least from our PoV.

So I understand that the file qemu_command.c contains all the QEMU commands/properties/options used by libvirt, thereby in the future if one wants to remove other properties, he can just check that file, right?

Thanks,
Zhao

#118 lore raw
In reply to Philippe Mathieu-Daudé

On Thu, 8 May 2025 15:35:23 +0200
Philippe Mathieu-Daudé <[email protected]> wrote:

Are you planning to resping it?
(if yes, I can provide you with a fixed 2/27 patch that removes all legacy cpu hp leftovers)

Quoted 104 lines

Since v3:
- Addressed Thomas and Zhao review comments
- Rename fw_cfg_init_mem_[no]dma() helpers
- Remove unused CPU properties
- Remove {multi,linux}boot.bin
- Added R-b tags

Since v2:
- Addressed Mark review comments and added his R-b tags

The versioned 'pc' and 'q35' machines up to 2.12 been marked as deprecated two releases ago, and are older than 6 years, so according to our support policy we can remove them.

This series only includes the 2.6 and 2.7 machines removal, as it is a big enough number of LoC removed. Rest will follow.

Based-on: <[email protected]>

Philippe Mathieu-Daudé (27):
hw/i386/pc: Remove deprecated pc-q35-2.6 and pc-i440fx-2.6 machines hw/i386/pc: Remove PCMachineClass::legacy_cpu_hotplug field hw/nvram/fw_cfg: Rename fw_cfg_init_mem() with '_nodma' suffix hw/mips/loongson3_virt: Prefer using fw_cfg_init_mem_nodma() hw/nvram/fw_cfg: Factor fw_cfg_init_mem_internal() out hw/nvram/fw_cfg: Rename fw_cfg_init_mem_wide() ->

    fw_cfg_init_mem_dma()

hw/i386/x86: Remove X86MachineClass::fwcfg_dma_enabled field hw/i386/pc: Remove multiboot.bin
hw/nvram/fw_cfg: Remove fw_cfg_io_properties::dma_enabled hw/i386/pc: Remove linuxboot.bin
hw/i386/pc: Remove pc_compat_2_6[] array
target/i386/cpu: Remove CPUX86State::enable_cpuid_0xb field target/i386/cpu: Remove CPUX86State::fill_mtrr_mask field hw/intc/apic: Remove APICCommonState::legacy_instance_id field hw/core/machine: Remove hw_compat_2_6[] array hw/virtio/virtio-mmio: Remove
VirtIOMMIOProxy::format_transport_address field
hw/i386/pc: Remove deprecated pc-q35-2.7 and pc-i440fx-2.7 machines hw/i386/pc: Remove pc_compat_2_7[] array
target/i386/cpu: Remove CPUX86State::full_cpuid_auto_level field target/i386/cpu: Remove CPUX86State::enable_l3_cache field hw/audio/pcspk: Remove PCSpkState::migrate field hw/core/machine: Remove hw_compat_2_7[] array hw/i386/intel_iommu: Remove IntelIOMMUState::buggy_eim field hw/intc/ioapic: Remove IOAPICCommonState::version field hw/virtio/virtio-pci: Remove VirtIOPCIProxy::ignore_backend_features
field
hw/char/virtio-serial: Do not expose the 'emergency-write' property hw/virtio/virtio-pci: Remove VIRTIO_PCI_FLAG_PAGE_PER_VQ definition

 hw/intc/ioapic_internal.h           |   3 +-
 include/hw/acpi/cpu_hotplug.h       |   3 -
 include/hw/boards.h                 |   6 -
 include/hw/i386/apic_internal.h     |   1 -
 include/hw/i386/intel_iommu.h       |   1 -
 include/hw/i386/pc.h                |   9 --
 include/hw/i386/x86.h               |   2 -
 include/hw/nvram/fw_cfg.h           |   9 +-
 include/hw/virtio/virtio-mmio.h     |   1 -
 include/hw/virtio/virtio-pci.h      |   2 -
 include/hw/virtio/virtio-serial.h   |   2 -
 pc-bios/optionrom/optionrom.h       |   4 -
 target/i386/cpu.h                   |  15 --
 hw/acpi/cpu_hotplug.c               | 230 ---------------------------
 hw/arm/virt.c                       |   2 +-
 hw/audio/pcspk.c                    |  10 --
 hw/char/virtio-serial-bus.c         |   9 +-
 hw/core/machine.c                   |  17 --
 hw/display/virtio-vga.c             |  10 --
 hw/hppa/machine.c                   |   2 +-
 hw/i386/acpi-build.c                |   4 +-
 hw/i386/fw_cfg.c                    |   5 +-
 hw/i386/intel_iommu.c               |   5 +-
 hw/i386/microvm.c                   |   3 -
 hw/i386/multiboot.c                 |   7 +-
 hw/i386/pc.c                        |  22 +--
 hw/i386/pc_piix.c                   |  23 ---
 hw/i386/pc_q35.c                    |  24 ---
 hw/i386/x86-common.c                |   6 +-
 hw/i386/x86.c                       |   2 -
 hw/intc/apic_common.c               |   5 -
 hw/intc/ioapic.c                    |  18 +--
 hw/intc/ioapic_common.c             |   2 +-
 hw/mips/loongson3_virt.c            |   2 +-
 hw/nvram/fw_cfg.c                   |  48 +++---
 hw/riscv/virt.c                     |   4 +-
 hw/virtio/virtio-mmio.c             |  15 --
 hw/virtio/virtio-pci.c              |  12 +-
 target/i386/cpu.c                   | 152 ++++++++----------
 target/i386/kvm/kvm.c               |  10 +-
 tests/qtest/test-x86-cpuid-compat.c |  11 --
 pc-bios/meson.build                 |   2 -
 pc-bios/multiboot.bin               | Bin 1024 -> 0 bytes
 pc-bios/optionrom/Makefile          |   2 +-
 pc-bios/optionrom/linuxboot.S       | 195 -----------------------
 pc-bios/optionrom/multiboot.S       | 232 ---------------------------
 pc-bios/optionrom/multiboot_dma.S   | 234 +++++++++++++++++++++++++++-
 47 files changed, 349 insertions(+), 1034 deletions(-)
 delete mode 100644 pc-bios/multiboot.bin
 delete mode 100644 pc-bios/optionrom/linuxboot.S
 delete mode 100644 pc-bios/optionrom/multiboot.S
#119 lore raw

Hi Igor,

Quoted 5 lines · Igor Mammedov

On Thu, 8 May 2025 15:35:23 +0200
Philippe Mathieu-Daudé <[email protected]> wrote:

Are you planning to resping it?
(if yes, I can provide you with a fixed 2/27 patch that removes all legacy cpu hp leftovers)

Sorry, no, I already burned all the x86 credits I had for 2025 :S

Zhao kindly offered to help with respin :)

Regards,

Phil.

#120 lore raw

Hi Igor and Philippe,

Quoted 8 lines · Igor Mammedov
Quoted 5 lines

On Thu, 8 May 2025 15:35:23 +0200
Philippe Mathieu-Daudé <[email protected]> wrote:

Are you planning to resping it?
(if yes, I can provide you with a fixed 2/27 patch that removes all legacy cpu hp leftovers)

Sorry, no, I already burned all the x86 credits I had for 2025 :S

Don't say that, thanks for your efforts! :-)

Quoted 1 line

Zhao kindly offered to help with respin :)

I haven't forgotten about this. I also plan to help it move forward in the coming weeks.

Thanks,
Zhao

#121 lore raw
Quoted 9 lines · On Fri, 31 Oct 2025 22:14:03 +0800 Zhao Liu <[email protected]> wrote:

Hi Igor and Philippe,

Quoted 7 lines · Igor Mammedov
Quoted 5 lines

On Thu, 8 May 2025 15:35:23 +0200
Philippe Mathieu-Daudé <[email protected]> wrote:

Are you planning to resping it?
(if yes, I can provide you with a fixed 2/27 patch that removes all legacy cpu hp leftovers)

Quoted 8 lines
Quoted 1 line

Sorry, no, I already burned all the x86 credits I had for 2025 :S

Don't say that, thanks for your efforts! :-)

Quoted 1 line

Zhao kindly offered to help with respin :)

I haven't forgotten about this. I also plan to help it move forward in the coming weeks.

in this case, I'll send reworked patch (not really tested) as a reply 2/27 so you could incorporate it on respin.

Quoted 3 lines

Thanks,
Zhao

#122 lore raw
Quoted 22 lines · Igor Mammedov

Date: Fri, 31 Oct 2025 15:23:45 +0100
From: Igor Mammedov <[email protected]>
Subject: Re: [PATCH v4 00/27] hw/i386/pc: Remove deprecated 2.6 and 2.7 PC
machines
X-Mailer: Claws Mail 4.3.1 (GTK 3.24.49; x86_64-redhat-linux-gnu)

Quoted 13 lines · On Fri, 31 Oct 2025 22:14:03 +0800 Zhao Liu <[email protected]> wrote:

Hi Igor and Philippe,

Quoted 8 lines · Igor Mammedov
Quoted 5 lines

On Thu, 8 May 2025 15:35:23 +0200
Philippe Mathieu-Daudé <[email protected]> wrote:

Are you planning to resping it?
(if yes, I can provide you with a fixed 2/27 patch that removes all legacy cpu hp leftovers)

Sorry, no, I already burned all the x86 credits I had for 2025 :S

Don't say that, thanks for your efforts! :-)

Quoted 7 lines
Quoted 4 lines
Quoted 1 line

Zhao kindly offered to help with respin :)

I haven't forgotten about this. I also plan to help it move forward in the coming weeks.

in this case, I'll send reworked patch (not really tested) as a reply 2/27 so you could incorporate it on respin.

Thank Igor! I'll include that patch into v5.

Regards,
Zhao