Skip to contentSkip to conversation

accel: Preparatory cleanups for split-accel

PATCH Superseded Philippe Mathieu-Daudé 6 participants 114 messages Jul 3, 2025 – Oct 7, 2025

Subsystems (2)

v6 of 7 Superseded Latest is v7 →

lore
Share & tools
lore
#0 lore raw

v6 cover accel: Preparatory cleanups for split-accel

Missing review: 11, 37-39

Since v5:
- Dropped patches with problems
- Moved system/accel-ops.h -> generic accel/ namespace
- Added R-b tags

Few changes needed before being able to add the
split acceleration:

- few method docstring added
- remove pointless stubs
- propagate soon required AccelState argument
- try to reduce current_accel() uses
- move declarations AccelClass <-> AccelOpsClass
- display model name in 'info cpus'
- add 'info accel' command to QMP/HMP
- make accel_create_vcpu_thread() more generic
- introduce hwaccel_enabled()

I plan to send a PR once fully reviewed.

Regards,

Phil.

Philippe Mathieu-Daudé (39):
hw/core/machine: Display CPU model name in 'info cpus' command system/memory: Restrict eventfd dispatch_write() to emulators system/runstate: Document qemu_add_vm_change_state_handler() system/cpus: Assert interrupt handling is done with BQL locked accel/kvm: Remove kvm_init_cpu_signals() stub
accel/kvm: Reduce kvm_create_vcpu() declaration scope accel/tcg: Remove 'info opcount' and @x-query-opcount accel/tcg: Remove profiler leftover
accel/tcg: Factor tcg_dump_flush_info() out
accel/tcg: Factor tcg_dump_stats() out for re-use accel/tcg: Extract statistic related code to tcg-stats.c
accel: Move supports_guest_debug() declaration to AccelClass
accel: Move cpus_are_resettable() declaration to AccelClass accel/hvf: Restrict internal declarations
accel/hvf: Move per-cpu method declarations to hvf-accel-ops.c accel/hvf: Move generic method declarations to hvf-all.c
cpus: Document CPUState::vcpu_dirty field
accel/hvf: Replace @dirty field by generic CPUState::vcpu_dirty field accel/nvmm: Replace @dirty field by generic CPUState::vcpu_dirty field accel/whpx: Replace @dirty field by generic CPUState::vcpu_dirty field accel/kvm: Remove kvm_cpu_synchronize_state() stub accel/system: Document cpu_synchronize_state() accel/system: Document cpu_synchronize_state_post_init/reset() accel/nvmm: Expose nvmm_enabled() to common code accel/whpx: Expose whpx_enabled() to common code accel/dummy: Extract 'dummy-cpus.h' header from 'system/cpus.h'
accel: Pass old/new interrupt mask to handle_interrupt() handler
accel: Expose and register generic_handle_interrupt()
accel: Keep reference to AccelOpsClass in AccelClass
accel: Propagate AccelState to AccelClass::init_machine() accel/kvm: Prefer local AccelState over global MachineState::accel accel/tcg: Prefer local AccelState over global current_accel()
accel: Directly pass AccelState argument to AccelClass::has_memory() accel/kvm: Directly pass KVMState argument to do_kvm_create_vm()
accel: Remove unused MachineState argument of AccelClass::setup_post()
accel: Pass AccelState argument to gdbstub_supported_sstep_flags()
accel: Rename 'system/accel-ops.h' -> 'accel/accel-cpu-ops.h'
accel: Extract AccelClass definition to 'accel/accel-ops.h'
MAINTAINERS: Add me as reviewer of overall accelerators section

 MAINTAINERS                                   |   4 +-
 qapi/machine.json                             |  21 +-
 accel/dummy-cpus.h                            |  14 +
 accel/kvm/kvm-cpus.h                          |   1 -
 accel/tcg/internal-common.h                   |   2 +
 accel/tcg/tcg-accel-ops-icount.h              |   2 +-
 accel/tcg/tcg-accel-ops.h                     |   2 +-
 .../accel-ops.h => accel/accel-cpu-ops.h}     |  34 +-
 include/accel/accel-ops.h                     |  50 +++
 include/hw/core/cpu.h                         |   3 +-
 include/qemu/accel.h                          |  35 +-
 include/system/cpus.h                         |   5 -
 include/system/hvf.h                          |  38 ---
 include/system/hvf_int.h                      |  38 ++-
 include/system/hw_accel.h                     |  21 +-
 include/system/kvm.h                          |   8 -
 include/system/kvm_int.h                      |   1 +
 include/system/nvmm.h                         |  23 +-
 include/system/runstate.h                     |  10 +
 include/system/whpx.h                         |  27 +-
 accel/accel-common.c                          |   4 +-
 accel/accel-system.c                          |  20 +-
 accel/dummy-cpus.c                            |   1 +
 accel/hvf/hvf-accel-ops.c                     | 317 +++---------------
 accel/hvf/hvf-all.c                           | 279 +++++++++++++--
 accel/kvm/kvm-accel-ops.c                     |  10 +-
 accel/kvm/kvm-all.c                           |  39 ++-
 accel/qtest/qtest.c                           |   7 +-
 accel/stubs/kvm-stub.c                        |   9 -
 accel/stubs/nvmm-stub.c                       |  12 +
 accel/stubs/whpx-stub.c                       |  12 +
 accel/tcg/monitor.c                           | 210 +-----------
 accel/tcg/tcg-accel-ops-icount.c              |   8 +-
 accel/tcg/tcg-accel-ops.c                     |  19 +-
 accel/tcg/tcg-all.c                           |  13 +-
 accel/tcg/tcg-stats.c                         | 206 ++++++++++++
 accel/xen/xen-all.c                           |   9 +-
 bsd-user/main.c                               |   3 +-
 cpu-target.c                                  |   2 +-
 gdbstub/system.c                              |  11 +-
 hw/core/machine-hmp-cmds.c                    |   3 +-
 hw/core/machine-qmp-cmds.c                    |   1 +
 linux-user/main.c                             |   3 +-
 system/cpus.c                                 |  28 +-
 system/memory.c                               |  12 +-
 target/arm/hvf/hvf.c                          |   6 +-
 target/i386/hvf/hvf.c                         |   6 +-
 target/i386/hvf/x86hvf.c                      |   2 +-
 target/i386/nvmm/nvmm-accel-ops.c             |   3 +-
 target/i386/nvmm/nvmm-all.c                   |  32 +-
 target/i386/whpx/whpx-accel-ops.c             |   3 +-
 target/i386/whpx/whpx-all.c                   |  33 +-
 tests/qtest/qmp-cmd-test.c                    |   1 -
 accel/stubs/meson.build                       |   2 +
 accel/tcg/meson.build                         |   1 +
 hmp-commands-info.hx                          |  14 -
 56 files changed, 876 insertions(+), 804 deletions(-)
 create mode 100644 accel/dummy-cpus.h
 rename include/{system/accel-ops.h => accel/accel-cpu-ops.h} (70%)
 create mode 100644 include/accel/accel-ops.h
 create mode 100644 accel/stubs/nvmm-stub.c
 create mode 100644 accel/stubs/whpx-stub.c
 create mode 100644 accel/tcg/tcg-stats.c
Signature · 1 line
2.49.0
#1 lore raw

PATCH v6 1/39 hw/core/machine: Display CPU model name in 'info cpus' command

Display the CPU model in 'info cpus'. Example before:

 $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
 QEMU 10.0.0 monitor - type 'help' for more information
 (qemu) info cpus
 * CPU #0: thread_id=42924
   CPU #1: thread_id=42924
   CPU #2: thread_id=42924
   CPU #3: thread_id=42924
 (qemu) q

and after:

 $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
 QEMU 10.0.50 monitor - type 'help' for more information
 (qemu) info cpus
 * CPU #0: thread_id=42916 (cortex-a72)
   CPU #1: thread_id=42916 (cortex-a72)
   CPU #2: thread_id=42916 (cortex-r5f)
   CPU #3: thread_id=42916 (cortex-r5f)
 (qemu)
---
 qapi/machine.json          | 3 +++
 hw/core/machine-hmp-cmds.c | 3 ++-
 hw/core/machine-qmp-cmds.c | 1 +
 3 files changed, 6 insertions(+), 1 deletion(-)
qapi/machine.json +3 -0
index 0650b8de71a..d5bbb5e367e 100644
@@ -80,6 +80,8 @@
 #
 # @thread-id: ID of the underlying host thread
 #
+# @model: CPU model name (since 10.1)
+#
 # @props: properties associated with a virtual CPU, e.g. the socket id
 #
 # @target: the QEMU system emulation target, which determines which
@@ -91,6 +93,7 @@
   'base'          : { 'cpu-index'    : 'int',
                       'qom-path'     : 'str',
                       'thread-id'    : 'int',
+                      'model'        : 'str',
                       '*props'       : 'CpuInstanceProperties',
                       'target'       : 'SysEmuTarget' },
   'discriminator' : 'target',
hw/core/machine-hmp-cmds.c +2 -1
index c6325cdcaaa..65eeb5e9cc2 100644
@@ -40,7 +40,8 @@ void hmp_info_cpus(Monitor *mon, const QDict *qdict)
 
         monitor_printf(mon, "%c CPU #%" PRId64 ":", active,
                        cpu->value->cpu_index);
-        monitor_printf(mon, " thread_id=%" PRId64 "\n", cpu->value->thread_id);
+        monitor_printf(mon, " thread_id=%" PRId64 " (%s)\n",
+                       cpu->value->thread_id, cpu->value->model);
     }
 
     qapi_free_CpuInfoFastList(cpu_list);
hw/core/machine-qmp-cmds.c +1 -0
index d82043e1c68..ab4fd1ec08a 100644
@@ -47,6 +47,7 @@ CpuInfoFastList *qmp_query_cpus_fast(Error **errp)
         value->cpu_index = cpu->cpu_index;
         value->qom_path = object_get_canonical_path(OBJECT(cpu));
         value->thread_id = cpu->thread_id;
+        value->model = cpu_model_from_type(object_get_typename(OBJECT(cpu)));
 
         if (mc->cpu_index_to_instance_props) {
             CpuInstanceProperties *props;
Signature · 1 line
2.49.0
Reviewed-by on v6 1/39
#2 lore raw

Re: [PATCH v6 01/39] hw/core/machine: Display CPU model name in 'info cpus' command

Quoted 13 lines · Philippe Mathieu-Daudé

Display the CPU model in 'info cpus'. Example before:

  $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
  QEMU 10.0.0 monitor - type 'help' for more information
  (qemu) info cpus
  * CPU #0: thread_id=42924
    CPU #1: thread_id=42924
    CPU #2: thread_id=42924
    CPU #3: thread_id=42924
  (qemu) q

and after:

Quoted 8 lines
  $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
  QEMU 10.0.50 monitor - type 'help' for more information
  (qemu) info cpus
  * CPU #0: thread_id=42916 (cortex-a72)
    CPU #1: thread_id=42916 (cortex-a72)
    CPU #2: thread_id=42916 (cortex-r5f)
    CPU #3: thread_id=42916 (cortex-r5f)
  (qemu)

wow, it's really useful to the architectures can run with multiple CPU cores.

Quoted 57 lines
---
  qapi/machine.json          | 3 +++
  hw/core/machine-hmp-cmds.c | 3 ++-
  hw/core/machine-qmp-cmds.c | 1 +
  3 files changed, 6 insertions(+), 1 deletion(-)
qapi/machine.json +3 -0
index 0650b8de71a..d5bbb5e367e 100644
@@ -80,6 +80,8 @@
  #
  # @thread-id: ID of the underlying host thread
  #
+# @model: CPU model name (since 10.1)
+#
  # @props: properties associated with a virtual CPU, e.g. the socket id
  #
  # @target: the QEMU system emulation target, which determines which
@@ -91,6 +93,7 @@
    'base'          : { 'cpu-index'    : 'int',
                        'qom-path'     : 'str',
                        'thread-id'    : 'int',
+                      'model'        : 'str',
                        '*props'       : 'CpuInstanceProperties',
                        'target'       : 'SysEmuTarget' },
    'discriminator' : 'target',
hw/core/machine-hmp-cmds.c +2 -1
index c6325cdcaaa..65eeb5e9cc2 100644
@@ -40,7 +40,8 @@ void hmp_info_cpus(Monitor *mon, const QDict *qdict)
 
          monitor_printf(mon, "%c CPU #%" PRId64 ":", active,
                         cpu->value->cpu_index);
-        monitor_printf(mon, " thread_id=%" PRId64 "\n", cpu->value->thread_id);
+        monitor_printf(mon, " thread_id=%" PRId64 " (%s)\n",
+                       cpu->value->thread_id, cpu->value->model);
      }
 
      qapi_free_CpuInfoFastList(cpu_list);
hw/core/machine-qmp-cmds.c +1 -0
index d82043e1c68..ab4fd1ec08a 100644
@@ -47,6 +47,7 @@ CpuInfoFastList *qmp_query_cpus_fast(Error **errp)
          value->cpu_index = cpu->cpu_index;
          value->qom_path = object_get_canonical_path(OBJECT(cpu));
          value->thread_id = cpu->thread_id;
+        value->model = cpu_model_from_type(object_get_typename(OBJECT(cpu)));
 
          if (mc->cpu_index_to_instance_props) {
              CpuInstanceProperties *props;
#3 lore raw

Re: [PATCH v6 01/39] hw/core/machine: Display CPU model name in 'info cpus' command

In reply to Philippe Mathieu-Daudé
Quoted 72 lines · Philippe Mathieu-Daudé

Display the CPU model in 'info cpus'. Example before:

 $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
 QEMU 10.0.0 monitor - type 'help' for more information
 (qemu) info cpus
 * CPU #0: thread_id=42924
   CPU #1: thread_id=42924
   CPU #2: thread_id=42924
   CPU #3: thread_id=42924
 (qemu) q

and after:

 $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
 QEMU 10.0.50 monitor - type 'help' for more information
 (qemu) info cpus
 * CPU #0: thread_id=42916 (cortex-a72)
   CPU #1: thread_id=42916 (cortex-a72)
   CPU #2: thread_id=42916 (cortex-r5f)
   CPU #3: thread_id=42916 (cortex-r5f)
 (qemu)
---
 qapi/machine.json          | 3 +++
 hw/core/machine-hmp-cmds.c | 3 ++-
 hw/core/machine-qmp-cmds.c | 1 +
 3 files changed, 6 insertions(+), 1 deletion(-)
qapi/machine.json +3 -0
index 0650b8de71a..d5bbb5e367e 100644
@@ -80,6 +80,8 @@
 #
 # @thread-id: ID of the underlying host thread
 #
+# @model: CPU model name (since 10.1)
+#
 # @props: properties associated with a virtual CPU, e.g. the socket id
 #
 # @target: the QEMU system emulation target, which determines which
@@ -91,6 +93,7 @@
   'base'          : { 'cpu-index'    : 'int',
                       'qom-path'     : 'str',
                       'thread-id'    : 'int',
+                      'model'        : 'str',
                       '*props'       : 'CpuInstanceProperties',
                       'target'       : 'SysEmuTarget' },
   'discriminator' : 'target',
hw/core/machine-hmp-cmds.c +2 -1
index c6325cdcaaa..65eeb5e9cc2 100644
@@ -40,7 +40,8 @@ void hmp_info_cpus(Monitor *mon, const QDict *qdict)
 
         monitor_printf(mon, "%c CPU #%" PRId64 ":", active,
                        cpu->value->cpu_index);
-        monitor_printf(mon, " thread_id=%" PRId64 "\n", cpu->value->thread_id);
+        monitor_printf(mon, " thread_id=%" PRId64 " (%s)\n",
+                       cpu->value->thread_id, cpu->value->model);
     }
 
     qapi_free_CpuInfoFastList(cpu_list);
hw/core/machine-qmp-cmds.c +0 -0
index d82043e1c68..ab4fd1ec08a 100644

hw/core/machine-qmp-cmds.c CpuInfoFastList *qmp_query_cpus_fast(Error **errp) · line 53

Quoted 7 lines
hw/core/machine-qmp-cmds.c +1 -0
@@ -47,6 +47,7 @@ CpuInfoFastList *qmp_query_cpus_fast(Error **errp)
         value->cpu_index = cpu->cpu_index;
         value->qom_path = object_get_canonical_path(OBJECT(cpu));
         value->thread_id = cpu->thread_id;
+        value->model = cpu_model_from_type(object_get_typename(OBJECT(cpu)));
 
         if (mc->cpu_index_to_instance_props) {
             CpuInstanceProperties *props;

Does the conversion from CPU type name to model name lose information?

If yes, should we provide the type name at least in QMP?

Let me also try a different angle... what's the preferred thing for users of HMP and for users of QMP, CPU model name or CPU type name?

Are there any commands that accept one, but not the other?

Are there other commands that return one, but not the other?

#4 lore raw

Re: [PATCH v6 01/39] hw/core/machine: Display CPU model name in 'info cpus' command

Quoted 79 lines · Markus Armbruster
Quoted 77 lines · Philippe Mathieu-Daudé

Display the CPU model in 'info cpus'. Example before:

  $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
  QEMU 10.0.0 monitor - type 'help' for more information
  (qemu) info cpus
  * CPU #0: thread_id=42924
    CPU #1: thread_id=42924
    CPU #2: thread_id=42924
    CPU #3: thread_id=42924
  (qemu) q

and after:

  $ qemu-system-aarch64 -M xlnx-versal-virt -S -monitor stdio
  QEMU 10.0.50 monitor - type 'help' for more information
  (qemu) info cpus
  * CPU #0: thread_id=42916 (cortex-a72)
    CPU #1: thread_id=42916 (cortex-a72)
    CPU #2: thread_id=42916 (cortex-r5f)
    CPU #3: thread_id=42916 (cortex-r5f)
  (qemu)
---
  qapi/machine.json          | 3 +++
  hw/core/machine-hmp-cmds.c | 3 ++-
  hw/core/machine-qmp-cmds.c | 1 +
  3 files changed, 6 insertions(+), 1 deletion(-)
qapi/machine.json +3 -0
index 0650b8de71a..d5bbb5e367e 100644
@@ -80,6 +80,8 @@
  #
  # @thread-id: ID of the underlying host thread
  #
+# @model: CPU model name (since 10.1)
+#
  # @props: properties associated with a virtual CPU, e.g. the socket id
  #
  # @target: the QEMU system emulation target, which determines which
@@ -91,6 +93,7 @@
    'base'          : { 'cpu-index'    : 'int',
                        'qom-path'     : 'str',
                        'thread-id'    : 'int',
+                      'model'        : 'str',
                        '*props'       : 'CpuInstanceProperties',
                        'target'       : 'SysEmuTarget' },
    'discriminator' : 'target',
hw/core/machine-hmp-cmds.c +2 -1
index c6325cdcaaa..65eeb5e9cc2 100644
@@ -40,7 +40,8 @@ void hmp_info_cpus(Monitor *mon, const QDict *qdict)
 
          monitor_printf(mon, "%c CPU #%" PRId64 ":", active,
                         cpu->value->cpu_index);
-        monitor_printf(mon, " thread_id=%" PRId64 "\n", cpu->value->thread_id);
+        monitor_printf(mon, " thread_id=%" PRId64 " (%s)\n",
+                       cpu->value->thread_id, cpu->value->model);
      }
 
      qapi_free_CpuInfoFastList(cpu_list);
hw/core/machine-qmp-cmds.c +1 -0
index d82043e1c68..ab4fd1ec08a 100644
@@ -47,6 +47,7 @@ CpuInfoFastList *qmp_query_cpus_fast(Error **errp)
          value->cpu_index = cpu->cpu_index;
          value->qom_path = object_get_canonical_path(OBJECT(cpu));
          value->thread_id = cpu->thread_id;
+        value->model = cpu_model_from_type(object_get_typename(OBJECT(cpu)));
 
Quoted 4 lines

hw/core/machine-qmp-cmds.c CpuInfoFastList *qmp_query_cpus_fast(Error **errp) · line 53

Quoted 2 lines
hw/core/machine-qmp-cmds.c +0 -0
          if (mc->cpu_index_to_instance_props) {
              CpuInstanceProperties *props;

Does the conversion from CPU type name to model name lose information?

Likely.

Quoted 1 line

If yes, should we provide the type name at least in QMP?

OK.

Quoted 4 lines

Let me also try a different angle... what's the preferred thing for users of HMP and for users of QMP, CPU model name or CPU type name?

Are there any commands that accept one, but not the other?

Likely. In particular the CLI expect model names for '-cpu', except when created as '-device' were the type name is expected...

Quoted 1 line

Are there other commands that return one, but not the other?

QMP commands usually return the QOM type name, not the model.

IMHO the external interface should be the CPU model name. Keeping it simple, an example could be "max".

The internal interface (QOM type name) is used to compose the QOM model. Some types end verbose, such "riscv64-max-tcg". I expect such information to be more relevant to an experimented QEMU developer or in a bug report rather than exposed to the users.

Anyway, I'll respin using type name for QMP and model name for HMP.

Thanks,

Phil.

#5 lore raw

PATCH v6 2/39 system/memory: Restrict eventfd dispatch_write() to emulators

In reply to Philippe Mathieu-Daudé

Commit 8c56c1a592b ("memory: emulate ioeventfd") added a !KVM check because the only accelerator available back then were TCG, QTest and KVM. Then commit 126e7f78036 ("kvm: require KVM_CAP_IOEVENTFD and KVM_CAP_IOEVENTFD_ANY_LENGTH") suggested '!KVM' check should be '(TCG || QTest)'. Later more accelerator were added. Implement the suggestion as a safety measure, not dispatching to eventfd when hardware accelerator is used.

---
 system/memory.c | 9 ++-------
 1 file changed, 2 insertions(+), 7 deletions(-)
system/memory.c +2 -7
index 76b44b8220f..4f713889a8e 100644
@@ -25,7 +25,7 @@
 #include "qom/object.h"
 #include "trace.h"
 #include "system/ram_addr.h"
-#include "system/kvm.h"
+#include "system/qtest.h"
 #include "system/runstate.h"
 #include "system/tcg.h"
 #include "qemu/accel.h"
@@ -1530,12 +1530,7 @@ MemTxResult memory_region_dispatch_write(MemoryRegion *mr,
 
     adjust_endianness(mr, &data, op);
 
-    /*
-     * FIXME: it's not clear why under KVM the write would be processed
-     * directly, instead of going through eventfd.  This probably should
-     * test "tcg_enabled() || qtest_enabled()", or should just go away.
-     */
-    if (!kvm_enabled() &&
+    if ((tcg_enabled() || qtest_enabled()) &&
         memory_region_dispatch_write_eventfds(mr, addr, data, size, attrs)) {
         return MEMTX_OK;
     }
Signature · 1 line
2.49.0
#6 lore raw

Re: [PATCH v6 02/39] system/memory: Restrict eventfd dispatch_write() to emulators

Quoted 28 lines · Philippe Mathieu-Daudé

Commit 8c56c1a592b ("memory: emulate ioeventfd") added a !KVM check because the only accelerator available back then were TCG, QTest and KVM. Then commit 126e7f78036 ("kvm: require KVM_CAP_IOEVENTFD and KVM_CAP_IOEVENTFD_ANY_LENGTH") suggested '!KVM' check should be '(TCG || QTest)'. Later more accelerator were added. Implement the suggestion as a safety measure, not dispatching to eventfd when hardware accelerator is used.

---
  system/memory.c | 9 ++-------
  1 file changed, 2 insertions(+), 7 deletions(-)
system/memory.c +1 -1
index 76b44b8220f..4f713889a8e 100644
@@ -25,7 +25,7 @@
  #include "qom/object.h"
  #include "trace.h"
  #include "system/ram_addr.h"
-#include "system/kvm.h"
+#include "system/qtest.h"
  #include "system/runstate.h"
  #include "system/tcg.h"
  #include "qemu/accel.h"

system/memory.c MemTxResult memory_region_dispatch_write(MemoryRegion *mr, · line 1533

Quoted 10 lines
system/memory.c +1 -6
@@ -1530,12 +1530,7 @@ MemTxResult memory_region_dispatch_write(MemoryRegion *mr,
 
      adjust_endianness(mr, &data, op);
 
-    /*
-     * FIXME: it's not clear why under KVM the write would be processed
-     * directly, instead of going through eventfd.  This probably should
-     * test "tcg_enabled() || qtest_enabled()", or should just go away.
-     */
-    if (!kvm_enabled() &&
+    if ((tcg_enabled() || qtest_enabled()) &&

The FIXME provides two options:
1. change to "tcg_enabled() || qtest_enabled()"
2. remove !kvm_enabled()

And as the FIXME said, it's not clear why under KVM the write would be processed directly. Now, the question becomes why under hardware accelerator is used the write would be processed directly instead of going through eventfd. I think it needs to answer this question when we do such change, and it's better to put the answer as the comment in the code.

Quoted 3 lines
system/memory.c +0 -0
          memory_region_dispatch_write_eventfds(mr, addr, data, size, attrs)) {
          return MEMTX_OK;
      }
#7 lore raw

Re: [PATCH v6 02/39] system/memory: Restrict eventfd dispatch_write() to emulators

Quoted 40 lines · Xiaoyao Li
Quoted 26 lines · Philippe Mathieu-Daudé

Commit 8c56c1a592b ("memory: emulate ioeventfd") added a !KVM check because the only accelerator available back then were TCG, QTest and KVM. Then commit 126e7f78036 ("kvm: require KVM_CAP_IOEVENTFD and KVM_CAP_IOEVENTFD_ANY_LENGTH") suggested '!KVM' check should be '(TCG || QTest)'. Later more accelerator were added. Implement the suggestion as a safety measure, not dispatching to eventfd when hardware accelerator is used.

---
  system/memory.c | 9 ++-------
  1 file changed, 2 insertions(+), 7 deletions(-)
system/memory.c +1 -1
index 76b44b8220f..4f713889a8e 100644
@@ -25,7 +25,7 @@
  #include "qom/object.h"
  #include "trace.h"
  #include "system/ram_addr.h"
-#include "system/kvm.h"
+#include "system/qtest.h"
  #include "system/runstate.h"
  #include "system/tcg.h"
  #include "qemu/accel.h"

system/memory.c MemTxResult memory_region_dispatch_write(MemoryRegion *mr, · line 1532

Quoted 8 lines
system/memory.c +1 -5
@@ -1530,12 +1530,7 @@ MemTxResult memory_region_dispatch_write(MemoryRegion *mr,
        adjust_endianness(mr, &data, op);
  -    /*
-     * FIXME: it's not clear why under KVM the write would be processed
-     * directly, instead of going through eventfd.  This probably should
-     * test "tcg_enabled() || qtest_enabled()", or should just go away.
-     */
-    if (!kvm_enabled() &&
+    if ((tcg_enabled() || qtest_enabled()) &&

The FIXME provides two options:
1. change to "tcg_enabled() || qtest_enabled()"
2. remove !kvm_enabled()

Quoted 6 lines

And as the FIXME said, it's not clear why under KVM the write would be processed directly. Now, the question becomes why under hardware accelerator is used the write would be processed directly instead of going through eventfd. I think it needs to answer this question when we do such change, and it's better to put the answer as the comment in the code.

Under KVM the eventfd notifications are sent directly from the kernel to the relevant fd. There is no reason why under KVM you couldn't inject the eventfds from QEMU but it would be a weird and sub-optimal setup. KVM is perfectly capable of trapping the MMIO accesses in kernel.

I don't think eventfd's can be supported for HVF because I don't think it has such a concept. For vhost-user devices they would then rely on VHOST_USER_VRING_KICK over the socket instead.

Quoted 4 lines
Quoted 4 lines
system/memory.c +0 -0
          memory_region_dispatch_write_eventfds(mr, addr, data, size, attrs)) {
          return MEMTX_OK;
      }
Signature · 2 lines
Alex Bennée
Virtualisation Tech Lead @ Linaro
#8 lore raw

PATCH v6 3/39 system/runstate: Document qemu_add_vm_change_state_handler()

In reply to Philippe Mathieu-Daudé
---
 include/system/runstate.h | 10 ++++++++++
 1 file changed, 10 insertions(+)
include/system/runstate.h +10 -0
index fdd5c4a5172..b6e8d6beab7 100644
@@ -14,6 +14,16 @@ void runstate_replay_enable(void);
 typedef void VMChangeStateHandler(void *opaque, bool running, RunState state);
 typedef int VMChangeStateHandlerWithRet(void *opaque, bool running, RunState state);
 
+/**
+ * qemu_add_vm_change_state_handler:
+ * @cb: the callback to invoke
+ * @opaque: user data passed to the callback
+ *
+ * Register a callback function that is invoked when the vm starts or stops
+ * running.
+ *
+ * Returns: an entry to be freed using qemu_del_vm_change_state_handler()
+ */
 VMChangeStateEntry *qemu_add_vm_change_state_handler(VMChangeStateHandler *cb,
                                                      void *opaque);
 VMChangeStateEntry *qemu_add_vm_change_state_handler_prio(
Signature · 1 line
2.49.0
#9 lore raw

Re: [PATCH v6 03/39] system/runstate: Document qemu_add_vm_change_state_handler()

Quoted 15 lines · Philippe Mathieu-Daudé
---
  include/system/runstate.h | 10 ++++++++++
  1 file changed, 10 insertions(+)
include/system/runstate.h +0 -0
index fdd5c4a5172..b6e8d6beab7 100644
@@ -14,6 +14,16 @@ void runstate_replay_enable(void);
  typedef void VMChangeStateHandler(void *opaque, bool running, RunState state);
  typedef int VMChangeStateHandlerWithRet(void *opaque, bool running, RunState state);
 

include/system/runstate.h void runstate_replay_enable(void); · line 26

Quoted 10 lines
include/system/runstate.h +10 -0
+/**
+ * qemu_add_vm_change_state_handler:
+ * @cb: the callback to invoke
+ * @opaque: user data passed to the callback
+ *
+ * Register a callback function that is invoked when the vm starts or stops
+ * running.
+ *
+ * Returns: an entry to be freed using qemu_del_vm_change_state_handler()
+ */

qemu_add_vm_change_state_handler_prio() and qemu_add_vm_change_state_handler_prio_full() put the document in the implementation in system/runstate.c.

Please make them consistent.

Quoted 3 lines
include/system/runstate.h +0 -0
  VMChangeStateEntry *qemu_add_vm_change_state_handler(VMChangeStateHandler *cb,
                                                       void *opaque);
  VMChangeStateEntry *qemu_add_vm_change_state_handler_prio(
#10 lore raw

Re: [PATCH v6 03/39] system/runstate: Document qemu_add_vm_change_state_handler()

Quoted 20 lines · Xiaoyao Li
Quoted 19 lines · Philippe Mathieu-Daudé
---
  include/system/runstate.h | 10 ++++++++++
  1 file changed, 10 insertions(+)
include/system/runstate.h +5 -0
index fdd5c4a5172..b6e8d6beab7 100644
@@ -14,6 +14,16 @@ void runstate_replay_enable(void);
  typedef void VMChangeStateHandler(void *opaque, bool running, RunState state);
  typedef int VMChangeStateHandlerWithRet(void *opaque, bool running, RunState state);
  +/**
+ * qemu_add_vm_change_state_handler:
+ * @cb: the callback to invoke
+ * @opaque: user data passed to the callback
+ *
+ * Register a callback function that is invoked when the vm starts or stops
Quoted 8 lines

include/system/runstate.h void runstate_replay_enable(void); · line 26

Quoted 4 lines
include/system/runstate.h +4 -0
+ * running.
+ *
+ * Returns: an entry to be freed using qemu_del_vm_change_state_handler()
+ */

qemu_add_vm_change_state_handler_prio() and qemu_add_vm_change_state_handler_prio_full() put the document in the implementation in system/runstate.c.

Generally APIs to the rest of QEMU should be documented in the headers. Comments on individual functions or internal details are fine to live in the C files.

Quoted 6 lines

Please make them consistent.

Quoted 3 lines
include/system/runstate.h +0 -0
  VMChangeStateEntry *qemu_add_vm_change_state_handler(VMChangeStateHandler *cb,
                                                       void *opaque);
  VMChangeStateEntry *qemu_add_vm_change_state_handler_prio(
Signature · 2 lines
Alex Bennée
Virtualisation Tech Lead @ Linaro
#11 lore raw

Re: [PATCH v6 03/39] system/runstate: Document qemu_add_vm_change_state_handler()

Quoted 31 lines · Alex Bennée
Quoted 20 lines · Xiaoyao Li
Quoted 19 lines · Philippe Mathieu-Daudé
---
   include/system/runstate.h | 10 ++++++++++
   1 file changed, 10 insertions(+)
include/system/runstate.h +5 -0
index fdd5c4a5172..b6e8d6beab7 100644
@@ -14,6 +14,16 @@ void runstate_replay_enable(void);
   typedef void VMChangeStateHandler(void *opaque, bool running, RunState state);
   typedef int VMChangeStateHandlerWithRet(void *opaque, bool running, RunState state);
   +/**
+ * qemu_add_vm_change_state_handler:
+ * @cb: the callback to invoke
+ * @opaque: user data passed to the callback
+ *
+ * Register a callback function that is invoked when the vm starts or stops
Quoted 8 lines

include/system/runstate.h void runstate_replay_enable(void); · line 26

Quoted 4 lines
include/system/runstate.h +4 -0
+ * running.
+ *
+ * Returns: an entry to be freed using qemu_del_vm_change_state_handler()
+ */

qemu_add_vm_change_state_handler_prio() and qemu_add_vm_change_state_handler_prio_full() put the document in the implementation in system/runstate.c.

Quoted 3 lines

Generally APIs to the rest of QEMU should be documented in the headers. Comments on individual functions or internal details are fine to live in the C files.

I totally understand it.

I was not asking to put the document into C files, but to ...

Quoted 2 lines
Quoted 2 lines

Please make them consistent.

... make them consistent. IOW, I would expect an additional patch to move the document of qemu_add_vm_change_state_handler_prio() and qemu_add_vm_change_state_handler_prio_full() from C files to this header file.

Quoted 4 lines
Quoted 3 lines
Quoted 3 lines
include/system/runstate.h +0 -0
   VMChangeStateEntry *qemu_add_vm_change_state_handler(VMChangeStateHandler *cb,
                                                        void *opaque);
   VMChangeStateEntry *qemu_add_vm_change_state_handler_prio(
Reviewed-by on v6 3/39
#12 lore raw

Re: [PATCH v6 03/39] system/runstate: Document qemu_add_vm_change_state_handler()

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:09 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 03/39] system/runstate: Document
qemu_add_vm_change_state_handler()
X-Mailer: git-send-email 2.49.0

Quoted 6 lines
---
 include/system/runstate.h | 10 ++++++++++
 1 file changed, 10 insertions(+)
#13 lore raw

PATCH v6 4/39 system/cpus: Assert interrupt handling is done with BQL locked

In reply to Philippe Mathieu-Daudé
---
 accel/tcg/tcg-accel-ops.c | 2 --
 system/cpus.c             | 2 ++
 2 files changed, 2 insertions(+), 2 deletions(-)
accel/tcg/tcg-accel-ops.c +0 -2
index b24d6a75625..6116644d1c0 100644
@@ -93,8 +93,6 @@ static void tcg_cpu_reset_hold(CPUState *cpu)
 /* mask must never be zero, except for A20 change call */
 void tcg_handle_interrupt(CPUState *cpu, int mask)
 {
-    g_assert(bql_locked());
-
     cpu->interrupt_request |= mask;
 
     /*
system/cpus.c +2 -0
index d16b0dff989..a43e0e4e796 100644
@@ -265,6 +265,8 @@ static void generic_handle_interrupt(CPUState *cpu, int mask)
 
 void cpu_interrupt(CPUState *cpu, int mask)
 {
+    g_assert(bql_locked());
+
     if (cpus_accel->handle_interrupt) {
         cpus_accel->handle_interrupt(cpu, mask);
     } else {
Signature · 1 line
2.49.0
Reviewed-by on v6 4/39
#14 lore raw

Re: [PATCH v6 04/39] system/cpus: Assert interrupt handling is done with BQL locked

Quoted 25 lines · Philippe Mathieu-Daudé
---
  accel/tcg/tcg-accel-ops.c | 2 --
  system/cpus.c             | 2 ++
  2 files changed, 2 insertions(+), 2 deletions(-)
accel/tcg/tcg-accel-ops.c +0 -2
index b24d6a75625..6116644d1c0 100644
@@ -93,8 +93,6 @@ static void tcg_cpu_reset_hold(CPUState *cpu)
  /* mask must never be zero, except for A20 change call */
  void tcg_handle_interrupt(CPUState *cpu, int mask)
  {
-    g_assert(bql_locked());
-
      cpu->interrupt_request |= mask;
 
      /*
system/cpus.c +0 -0
index d16b0dff989..a43e0e4e796 100644

system/cpus.c static void generic_handle_interrupt(CPUState *cpu, int mask) · line 268

Quoted 4 lines
system/cpus.c +1 -0
@@ -265,6 +265,8 @@ static void generic_handle_interrupt(CPUState *cpu, int mask)
 
  void cpu_interrupt(CPUState *cpu, int mask)
  {
+    g_assert(bql_locked());

The best result is it doesn't break any thing. But it can surely help catch the case without bql locked and get the case fixed them.

So

Quoted 3 lines
system/cpus.c +0 -0
      if (cpus_accel->handle_interrupt) {
          cpus_accel->handle_interrupt(cpu, mask);
      } else {
Reviewed-by on v6 4/39
#15 lore raw

Re: [PATCH v6 04/39] system/cpus: Assert interrupt handling is done with BQL locked

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:10 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 04/39] system/cpus: Assert interrupt handling is done
with BQL locked
X-Mailer: git-send-email 2.49.0

Quoted 6 lines
---
 accel/tcg/tcg-accel-ops.c | 2 --
 system/cpus.c             | 2 ++
 2 files changed, 2 insertions(+), 2 deletions(-)
#16 lore raw

Re: [PATCH v6 04/39] system/cpus: Assert interrupt handling is done with BQL locked

In reply to Philippe Mathieu-Daudé
Quoted 25 lines · Philippe Mathieu-Daudé
---
 accel/tcg/tcg-accel-ops.c | 2 --
 system/cpus.c             | 2 ++
 2 files changed, 2 insertions(+), 2 deletions(-)
accel/tcg/tcg-accel-ops.c +0 -2
index b24d6a75625..6116644d1c0 100644
@@ -93,8 +93,6 @@ static void tcg_cpu_reset_hold(CPUState *cpu)
 /* mask must never be zero, except for A20 change call */
 void tcg_handle_interrupt(CPUState *cpu, int mask)
 {
-    g_assert(bql_locked());
-
     cpu->interrupt_request |= mask;
 
     /*
system/cpus.c +0 -0
index d16b0dff989..a43e0e4e796 100644

system/cpus.c static void generic_handle_interrupt(CPUState *cpu, int mask) · line 269

Quoted 5 lines
system/cpus.c +2 -0
@@ -265,6 +265,8 @@ static void generic_handle_interrupt(CPUState *cpu, int mask)
 
 void cpu_interrupt(CPUState *cpu, int mask)
 {
+    g_assert(bql_locked());
+

Is this really the case since 27e76d01010 (cpu-common: use atomic access for interrupt_request). In fact I think tcg_handle_interrupt now uses cpu_set_interrupt.

Quoted 3 lines
system/cpus.c +0 -0
     if (cpus_accel->handle_interrupt) {
         cpus_accel->handle_interrupt(cpu, mask);
     } else {
Signature · 2 lines
Alex Bennée
Virtualisation Tech Lead @ Linaro
#17 lore raw

PATCH v6 5/39 accel/kvm: Remove kvm_init_cpu_signals() stub

In reply to Philippe Mathieu-Daudé

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_init_cpu_signals() stub is not necessary.

---
 accel/stubs/kvm-stub.c | 5 -----
 1 file changed, 5 deletions(-)
accel/stubs/kvm-stub.c +0 -5
index ecfd7636f5f..b9b4427c919 100644
@@ -105,11 +105,6 @@ unsigned int kvm_get_free_memslots(void)
     return 0;
 }
 
-void kvm_init_cpu_signals(CPUState *cpu)
-{
-    abort();
-}
-
 bool kvm_arm_supports_user_irq(void)
 {
     return false;
Signature · 1 line
2.49.0
Reviewed-by on v6 5/39
#18 lore raw

Re: [PATCH v6 05/39] accel/kvm: Remove kvm_init_cpu_signals() stub

Quoted 5 lines · Philippe Mathieu-Daudé

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_init_cpu_signals() stub is not necessary.

BTW, it seems we can further move kvm_init_cpu_signals() into kvm-accel-ops.c and make it internal static function?

Quoted 20 lines
---
  accel/stubs/kvm-stub.c | 5 -----
  1 file changed, 5 deletions(-)
accel/stubs/kvm-stub.c +0 -5
index ecfd7636f5f..b9b4427c919 100644
@@ -105,11 +105,6 @@ unsigned int kvm_get_free_memslots(void)
      return 0;
  }
 
-void kvm_init_cpu_signals(CPUState *cpu)
-{
-    abort();
-}
-
  bool kvm_arm_supports_user_irq(void)
  {
      return false;
#19 lore raw

Re: [PATCH v6 05/39] accel/kvm: Remove kvm_init_cpu_signals() stub

Quoted 11 lines · Xiaoyao Li
Quoted 5 lines · Philippe Mathieu-Daudé

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_init_cpu_signals() stub is not necessary.

BTW, it seems we can further move kvm_init_cpu_signals() into kvm-accel-ops.c and make it internal static function?

I suppose we can if we move kvm_immediate_exit to KVM's AccelState, but KVM code isn't really using it. We ought to move KVM specific fields from CPUState to KVM's AccelState. Not a priority.

Quoted 4 lines
Quoted 4 lines

---
  accel/stubs/kvm-stub.c | 5 -----

  1 file changed, 5 deletions(-)
Reviewed-by on v6 5/39
#20 lore raw

Re: [PATCH v6 05/39] accel/kvm: Remove kvm_init_cpu_signals() stub

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

Date: Thu, 3 Jul 2025 19:32:11 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 05/39] accel/kvm: Remove kvm_init_cpu_signals() stub
X-Mailer: git-send-email 2.49.0

Quoted 8 lines

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_init_cpu_signals() stub is not necessary.

---
 accel/stubs/kvm-stub.c | 5 -----
 1 file changed, 5 deletions(-)
#21 lore raw

PATCH v6 6/39 accel/kvm: Reduce kvm_create_vcpu() declaration scope

In reply to Philippe Mathieu-Daudé

kvm_create_vcpu() is only used within the same file unit.

---
 include/system/kvm.h | 8 --------
 accel/kvm/kvm-all.c  | 8 +++++++-
 2 files changed, 7 insertions(+), 9 deletions(-)
include/system/kvm.h +0 -8
index 7cc60d26f24..e943df2c09d 100644
@@ -316,14 +316,6 @@ int kvm_create_device(KVMState *s, uint64_t type, bool test);
  */
 bool kvm_device_supported(int vmfd, uint64_t type);
 
-/**
- * kvm_create_vcpu - Gets a parked KVM vCPU or creates a KVM vCPU
- * @cpu: QOM CPUState object for which KVM vCPU has to be fetched/created.
- *
- * @returns: 0 when success, errno (<0) when failed.
- */
-int kvm_create_vcpu(CPUState *cpu);
-
 /**
  * kvm_park_vcpu - Park QEMU KVM vCPU context
  * @cpu: QOM CPUState object for which QEMU KVM vCPU context has to be parked.
accel/kvm/kvm-all.c +7 -1
index d095d1b98f8..17235f26464 100644
@@ -453,7 +453,13 @@ static void kvm_reset_parked_vcpus(KVMState *s)
     }
 }
 
-int kvm_create_vcpu(CPUState *cpu)
+/**
+ * kvm_create_vcpu - Gets a parked KVM vCPU or creates a KVM vCPU
+ * @cpu: QOM CPUState object for which KVM vCPU has to be fetched/created.
+ *
+ * @returns: 0 when success, errno (<0) when failed.
+ */
+static int kvm_create_vcpu(CPUState *cpu)
 {
     unsigned long vcpu_id = kvm_arch_vcpu_id(cpu);
     KVMState *s = kvm_state;
Signature · 1 line
2.49.0
Reviewed-by on v6 6/39
#22 lore raw

Re: [PATCH v6 06/39] accel/kvm: Reduce kvm_create_vcpu() declaration scope

Quoted 4 lines · Philippe Mathieu-Daudé

kvm_create_vcpu() is only used within the same file unit.

Quoted 43 lines
---
  include/system/kvm.h | 8 --------
  accel/kvm/kvm-all.c  | 8 +++++++-
  2 files changed, 7 insertions(+), 9 deletions(-)
include/system/kvm.h +0 -8
index 7cc60d26f24..e943df2c09d 100644
@@ -316,14 +316,6 @@ int kvm_create_device(KVMState *s, uint64_t type, bool test);
   */
  bool kvm_device_supported(int vmfd, uint64_t type);
 
-/**
- * kvm_create_vcpu - Gets a parked KVM vCPU or creates a KVM vCPU
- * @cpu: QOM CPUState object for which KVM vCPU has to be fetched/created.
- *
- * @returns: 0 when success, errno (<0) when failed.
- */
-int kvm_create_vcpu(CPUState *cpu);
-
  /**
   * kvm_park_vcpu - Park QEMU KVM vCPU context
   * @cpu: QOM CPUState object for which QEMU KVM vCPU context has to be parked.
accel/kvm/kvm-all.c +7 -1
index d095d1b98f8..17235f26464 100644
@@ -453,7 +453,13 @@ static void kvm_reset_parked_vcpus(KVMState *s)
      }
  }
 
-int kvm_create_vcpu(CPUState *cpu)
+/**
+ * kvm_create_vcpu - Gets a parked KVM vCPU or creates a KVM vCPU
+ * @cpu: QOM CPUState object for which KVM vCPU has to be fetched/created.
+ *
+ * @returns: 0 when success, errno (<0) when failed.
+ */
+static int kvm_create_vcpu(CPUState *cpu)
  {
      unsigned long vcpu_id = kvm_arch_vcpu_id(cpu);
      KVMState *s = kvm_state;
Reviewed-by on v6 6/39
#23 lore raw

Re: [PATCH v6 06/39] accel/kvm: Reduce kvm_create_vcpu() declaration scope

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:12 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 06/39] accel/kvm: Reduce kvm_create_vcpu() declaration
scope
X-Mailer: git-send-email 2.49.0

Quoted 8 lines

kvm_create_vcpu() is only used within the same file unit.

---
 include/system/kvm.h | 8 --------
 accel/kvm/kvm-all.c  | 8 +++++++-
 2 files changed, 7 insertions(+), 9 deletions(-)
#24 lore raw

PATCH v6 7/39 accel/tcg: Remove 'info opcount' and @x-query-opcount

In reply to Philippe Mathieu-Daudé

Since commit 1b65b4f54c7 ("accel/tcg: remove CONFIG_PROFILER", released with QEMU v8.1.0) we get pointless output:

(qemu) info opcount
[TCG profiler not compiled]

Remove that unstable and unuseful command.

---
 qapi/machine.json          | 18 ------------------
 accel/tcg/monitor.c        | 21 ---------------------
 tests/qtest/qmp-cmd-test.c |  1 -
 hmp-commands-info.hx       | 14 --------------
 4 files changed, 54 deletions(-)
qapi/machine.json +0 -18
index d5bbb5e367e..acf6610efa5 100644
@@ -1764,24 +1764,6 @@
   'returns': 'HumanReadableText',
   'features': [ 'unstable' ] }
 
-##
-# @x-query-opcount:
-#
-# Query TCG opcode counters
-#
-# Features:
-#
-# @unstable: This command is meant for debugging.
-#
-# Returns: TCG opcode counters
-#
-# Since: 6.2
-##
-{ 'command': 'x-query-opcount',
-  'returns': 'HumanReadableText',
-  'if': 'CONFIG_TCG',
-  'features': [ 'unstable' ] }
-
 ##
 # @x-query-ramblock:
 #
accel/tcg/monitor.c +0 -21
index 1c182b6bfb5..7c686226b21 100644
@@ -215,30 +215,9 @@ HumanReadableText *qmp_x_query_jit(Error **errp)
     return human_readable_text_from_str(buf);
 }
 
-static void tcg_dump_op_count(GString *buf)
-{
-    g_string_append_printf(buf, "[TCG profiler not compiled]\n");
-}
-
-HumanReadableText *qmp_x_query_opcount(Error **errp)
-{
-    g_autoptr(GString) buf = g_string_new("");
-
-    if (!tcg_enabled()) {
-        error_setg(errp,
-                   "Opcode count information is only available with accel=tcg");
-        return NULL;
-    }
-
-    tcg_dump_op_count(buf);
-
-    return human_readable_text_from_str(buf);
-}
-
 static void hmp_tcg_register(void)
 {
     monitor_register_hmp_info_hrt("jit", qmp_x_query_jit);
-    monitor_register_hmp_info_hrt("opcount", qmp_x_query_opcount);
 }
 
 type_init(hmp_tcg_register);
tests/qtest/qmp-cmd-test.c +0 -1
index 040d042810b..cf718761861 100644
@@ -51,7 +51,6 @@ static int query_error_class(const char *cmd)
         { "x-query-usb", ERROR_CLASS_GENERIC_ERROR },
         /* Only valid with accel=tcg */
         { "x-query-jit", ERROR_CLASS_GENERIC_ERROR },
-        { "x-query-opcount", ERROR_CLASS_GENERIC_ERROR },
         { "xen-event-list", ERROR_CLASS_GENERIC_ERROR },
         { NULL, -1 }
     };
hmp-commands-info.hx +0 -14
index 639a450ee51..d7979222752 100644
@@ -256,20 +256,6 @@ SRST
     Show dynamic compiler info.
 ERST
 
-#if defined(CONFIG_TCG)
-    {
-        .name       = "opcount",
-        .args_type  = "",
-        .params     = "",
-        .help       = "show dynamic compiler opcode counters",
-    },
-#endif
-
-SRST
-  ``info opcount``
-    Show dynamic compiler opcode counters
-ERST
-
     {
         .name       = "sync-profile",
         .args_type  = "mean:-m,no_coalesce:-n,max:i?",
Signature · 1 line
2.49.0
Reviewed-by on v6 7/39
#25 lore raw

Re: [PATCH v6 07/39] accel/tcg: Remove 'info opcount' and @x-query-opcount

Quoted 11 lines · Philippe Mathieu-Daudé

Since commit 1b65b4f54c7 ("accel/tcg: remove CONFIG_PROFILER", released with QEMU v8.1.0) we get pointless output:

(qemu) info opcount
[TCG profiler not compiled]

Remove that unstable and unuseful command.

Reviewed-by on v6 7/39
#26 lore raw

Re: [PATCH v6 07/39] accel/tcg: Remove 'info opcount' and @x-query-opcount

In reply to Philippe Mathieu-Daudé
Quoted 11 lines · Philippe Mathieu-Daudé

Since commit 1b65b4f54c7 ("accel/tcg: remove CONFIG_PROFILER", released with QEMU v8.1.0) we get pointless output:

(qemu) info opcount
[TCG profiler not compiled]

Remove that unstable and unuseful command.

Reviewed-by on v6 7/39
#27 lore raw

Re: [PATCH v6 07/39] accel/tcg: Remove 'info opcount' and @x-query-opcount

In reply to Philippe Mathieu-Daudé
Quoted 17 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:13 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 07/39] accel/tcg: Remove 'info opcount' and
@x-query-opcount
X-Mailer: git-send-email 2.49.0

Since commit 1b65b4f54c7 ("accel/tcg: remove CONFIG_PROFILER", released with QEMU v8.1.0) we get pointless output:

(qemu) info opcount
[TCG profiler not compiled]

Remove that unstable and unuseful command.

Quoted 6 lines
---
 qapi/machine.json          | 18 ------------------
 accel/tcg/monitor.c        | 21 ---------------------
 tests/qtest/qmp-cmd-test.c |  1 -
 hmp-commands-info.hx       | 14 --------------
 4 files changed, 54 deletions(-)
#28 lore raw

PATCH v6 8/39 accel/tcg: Remove profiler leftover

In reply to Philippe Mathieu-Daudé

TCG profiler was removed in commit 1b65b4f54c7.

Fixes 1b65b4f54c7 ("accel/tcg: remove CONFIG_PROFILER")
---
 accel/tcg/monitor.c | 6 ------
 1 file changed, 6 deletions(-)
accel/tcg/monitor.c +0 -6
index 7c686226b21..344ec500473 100644
@@ -141,11 +141,6 @@ static void tlb_flush_counts(size_t *pfull, size_t *ppart, size_t *pelide)
     *pelide = elide;
 }
 
-static void tcg_dump_info(GString *buf)
-{
-    g_string_append_printf(buf, "[TCG profiler not compiled]\n");
-}
-
 static void dump_exec_info(GString *buf)
 {
     struct tb_tree_stats tst = {};
@@ -196,7 +191,6 @@ static void dump_exec_info(GString *buf)
     g_string_append_printf(buf, "TLB full flushes    %zu\n", flush_full);
     g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
     g_string_append_printf(buf, "TLB elided flushes  %zu\n", flush_elide);
-    tcg_dump_info(buf);
 }
 
 HumanReadableText *qmp_x_query_jit(Error **errp)
Signature · 1 line
2.49.0
Reviewed-by on v6 8/39
#30 lore raw

Re: [PATCH v6 08/39] accel/tcg: Remove profiler leftover

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

Date: Thu, 3 Jul 2025 19:32:14 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 08/39] accel/tcg: Remove profiler leftover
X-Mailer: git-send-email 2.49.0

Quoted 8 lines

TCG profiler was removed in commit 1b65b4f54c7.

Fixes 1b65b4f54c7 ("accel/tcg: remove CONFIG_PROFILER")
---
 accel/tcg/monitor.c | 6 ------
 1 file changed, 6 deletions(-)
#31 lore raw

PATCH v6 9/39 accel/tcg: Factor tcg_dump_flush_info() out

In reply to Philippe Mathieu-Daudé
---
 accel/tcg/monitor.c | 27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)
accel/tcg/monitor.c +17 -10
index 344ec500473..6d9cc11d94c 100644
@@ -141,11 +141,26 @@ static void tlb_flush_counts(size_t *pfull, size_t *ppart, size_t *pelide)
     *pelide = elide;
 }
 
+static void tcg_dump_flush_info(GString *buf)
+{
+    size_t flush_full, flush_part, flush_elide;
+
+    g_string_append_printf(buf, "TB flush count      %u\n",
+                           qatomic_read(&tb_ctx.tb_flush_count));
+    g_string_append_printf(buf, "TB invalidate count %u\n",
+                           qatomic_read(&tb_ctx.tb_phys_invalidate_count));
+
+    tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
+    g_string_append_printf(buf, "TLB full flushes    %zu\n", flush_full);
+    g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
+    g_string_append_printf(buf, "TLB elided flushes  %zu\n", flush_elide);
+}
+
 static void dump_exec_info(GString *buf)
 {
     struct tb_tree_stats tst = {};
     struct qht_stats hst;
-    size_t nb_tbs, flush_full, flush_part, flush_elide;
+    size_t nb_tbs;
 
     tcg_tb_foreach(tb_tree_stats_iter, &tst);
     nb_tbs = tst.nb_tbs;
@@ -182,15 +197,7 @@ static void dump_exec_info(GString *buf)
     qht_statistics_destroy(&hst);
 
     g_string_append_printf(buf, "\nStatistics:\n");
-    g_string_append_printf(buf, "TB flush count      %u\n",
-                           qatomic_read(&tb_ctx.tb_flush_count));
-    g_string_append_printf(buf, "TB invalidate count %u\n",
-                           qatomic_read(&tb_ctx.tb_phys_invalidate_count));
-
-    tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
-    g_string_append_printf(buf, "TLB full flushes    %zu\n", flush_full);
-    g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
-    g_string_append_printf(buf, "TLB elided flushes  %zu\n", flush_elide);
+    tcg_dump_flush_info(buf);
 }
 
 HumanReadableText *qmp_x_query_jit(Error **errp)
Signature · 1 line
2.49.0
Reviewed-by on v6 9/39
#32 lore raw

Re: [PATCH v6 09/39] accel/tcg: Factor tcg_dump_flush_info() out

Quoted 2 lines · Philippe Mathieu-Daudé
Quoted 37 lines
---
  accel/tcg/monitor.c | 27 +++++++++++++++++----------
  1 file changed, 17 insertions(+), 10 deletions(-)
accel/tcg/monitor.c +16 -1
index 344ec500473..6d9cc11d94c 100644
@@ -141,11 +141,26 @@ static void tlb_flush_counts(size_t *pfull, size_t *ppart, size_t *pelide)
      *pelide = elide;
  }
 
+static void tcg_dump_flush_info(GString *buf)
+{
+    size_t flush_full, flush_part, flush_elide;
+
+    g_string_append_printf(buf, "TB flush count      %u\n",
+                           qatomic_read(&tb_ctx.tb_flush_count));
+    g_string_append_printf(buf, "TB invalidate count %u\n",
+                           qatomic_read(&tb_ctx.tb_phys_invalidate_count));
+
+    tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
+    g_string_append_printf(buf, "TLB full flushes    %zu\n", flush_full);
+    g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
+    g_string_append_printf(buf, "TLB elided flushes  %zu\n", flush_elide);
+}
+
  static void dump_exec_info(GString *buf)
  {
      struct tb_tree_stats tst = {};
      struct qht_stats hst;
-    size_t nb_tbs, flush_full, flush_part, flush_elide;
+    size_t nb_tbs;
 
      tcg_tb_foreach(tb_tree_stats_iter, &tst);
      nb_tbs = tst.nb_tbs;

accel/tcg/monitor.c static void dump_exec_info(GString *buf) · line 185

Quoted 4 lines
accel/tcg/monitor.c +0 -1
@@ -182,15 +197,7 @@ static void dump_exec_info(GString *buf)
      qht_statistics_destroy(&hst);
 
      g_string_append_printf(buf, "\nStatistics:\n");
-    g_string_append_printf(buf, "TB flush count      %u\n",

side topic,

is the "TB" an typo of "TLB"?

Quoted 12 lines
accel/tcg/monitor.c +1 -8
-                           qatomic_read(&tb_ctx.tb_flush_count));
-    g_string_append_printf(buf, "TB invalidate count %u\n",
-                           qatomic_read(&tb_ctx.tb_phys_invalidate_count));
-
-    tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
-    g_string_append_printf(buf, "TLB full flushes    %zu\n", flush_full);
-    g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
-    g_string_append_printf(buf, "TLB elided flushes  %zu\n", flush_elide);
+    tcg_dump_flush_info(buf);
  }
 
  HumanReadableText *qmp_x_query_jit(Error **errp)
#33 lore raw

Re: [PATCH v6 09/39] accel/tcg: Factor tcg_dump_flush_info() out

Quoted 52 lines · Xiaoyao Li
Quoted 2 lines · Philippe Mathieu-Daudé
Quoted 40 lines

---
  accel/tcg/monitor.c | 27 +++++++++++++++++----------

  1 file changed, 17 insertions(+), 10 deletions(-)
accel/tcg/monitor.c +16 -1
index 344ec500473..6d9cc11d94c 100644
@@ -141,11 +141,26 @@ static void tlb_flush_counts(size_t *pfull, 
 size_t *ppart, size_t *pelide)
       *pelide = elide;
   }
+static void tcg_dump_flush_info(GString *buf)
+{
+    size_t flush_full, flush_part, flush_elide;
+
+    g_string_append_printf(buf, "TB flush count      %u\n",
+                           qatomic_read(&tb_ctx.tb_flush_count));
+    g_string_append_printf(buf, "TB invalidate count %u\n",
+                           
 qatomic_read(&tb_ctx.tb_phys_invalidate_count));
+
+    tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
+    g_string_append_printf(buf, "TLB full flushes    %zu\n", 
 flush_full);
+    g_string_append_printf(buf, "TLB partial flushes %zu\n", 
 flush_part);
+    g_string_append_printf(buf, "TLB elided flushes  %zu\n", 
 flush_elide);
+}
+
   static void dump_exec_info(GString *buf)
   {
       struct tb_tree_stats tst = {};
       struct qht_stats hst;
-    size_t nb_tbs, flush_full, flush_part, flush_elide;
+    size_t nb_tbs;

      tcg_tb_foreach(tb_tree_stats_iter, &tst);
      nb_tbs = tst.nb_tbs;

accel/tcg/monitor.c static void dump_exec_info(GString *buf) · line 184

Quoted 3 lines
accel/tcg/monitor.c +0 -1
@@ -182,15 +197,7 @@ static void dump_exec_info(GString *buf)
       qht_statistics_destroy(&hst);
       g_string_append_printf(buf, "\nStatistics:\n");
-    g_string_append_printf(buf, "TB flush count      %u\n",

side topic,

Quoted 1 line

is the "TB" an typo of "TLB"?

Please ignore my silly question. It's TranslationBlock of tcg.

Quoted 16 lines
Quoted 15 lines
accel/tcg/monitor.c +1 -8
-                           qatomic_read(&tb_ctx.tb_flush_count));
-    g_string_append_printf(buf, "TB invalidate count %u\n",
-                           
 qatomic_read(&tb_ctx.tb_phys_invalidate_count));
-
-    tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
-    g_string_append_printf(buf, "TLB full flushes    %zu\n", 
 flush_full);
-    g_string_append_printf(buf, "TLB partial flushes %zu\n", 
 flush_part);
-    g_string_append_printf(buf, "TLB elided flushes  %zu\n", 
 flush_elide);
+    tcg_dump_flush_info(buf);

  }
  HumanReadableText *qmp_x_query_jit(Error **errp)

Reviewed-by on v6 9/39
#34 lore raw

Re: [PATCH v6 09/39] accel/tcg: Factor tcg_dump_flush_info() out

In reply to Philippe Mathieu-Daudé
Quoted 10 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:15 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 09/39] accel/tcg: Factor tcg_dump_flush_info() out
X-Mailer: git-send-email 2.49.0

---
 accel/tcg/monitor.c | 27 +++++++++++++++++----------
 1 file changed, 17 insertions(+), 10 deletions(-)
#35 lore raw

PATCH v6 10/39 accel/tcg: Factor tcg_dump_stats() out for re-use

In reply to Philippe Mathieu-Daudé
---
 accel/tcg/internal-common.h |  2 ++
 accel/tcg/monitor.c         | 11 ++++++++---
 2 files changed, 10 insertions(+), 3 deletions(-)
accel/tcg/internal-common.h +2 -0
index 1dbc45dd955..77a3a0684a5 100644
@@ -139,4 +139,6 @@ G_NORETURN void cpu_io_recompile(CPUState *cpu, uintptr_t retaddr);
 void tb_phys_invalidate(TranslationBlock *tb, tb_page_addr_t page_addr);
 void tb_set_jmp_target(TranslationBlock *tb, int n, uintptr_t addr);
 
+void tcg_dump_stats(GString *buf);
+
 #endif
accel/tcg/monitor.c +8 -3
index 6d9cc11d94c..e7ed7281a4b 100644
@@ -200,6 +200,13 @@ static void dump_exec_info(GString *buf)
     tcg_dump_flush_info(buf);
 }
 
+void tcg_dump_stats(GString *buf)
+{
+    dump_accel_info(buf);
+    dump_exec_info(buf);
+    dump_drift_info(buf);
+}
+
 HumanReadableText *qmp_x_query_jit(Error **errp)
 {
     g_autoptr(GString) buf = g_string_new("");
@@ -209,9 +216,7 @@ HumanReadableText *qmp_x_query_jit(Error **errp)
         return NULL;
     }
 
-    dump_accel_info(buf);
-    dump_exec_info(buf);
-    dump_drift_info(buf);
+    tcg_dump_stats(buf);
 
     return human_readable_text_from_str(buf);
 }
Signature · 1 line
2.49.0
Reviewed-by on v6 10/39
#36 lore raw

Re: [PATCH v6 10/39] accel/tcg: Factor tcg_dump_stats() out for re-use

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:16 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 10/39] accel/tcg: Factor tcg_dump_stats() out for re-use
X-Mailer: git-send-email 2.49.0

---
 accel/tcg/internal-common.h |  2 ++
 accel/tcg/monitor.c         | 11 ++++++++---
 2 files changed, 10 insertions(+), 3 deletions(-)
#37 lore raw

PATCH v6 11/39 accel/tcg: Extract statistic related code to tcg-stats.c

In reply to Philippe Mathieu-Daudé

Statistic code is not specific to system emulation (except cross-page checks) and can be used to analyze user-mode binaries. Extract statistic related code to its own file: tcg-stats.c, keeping the original LGPL-2.1-or-later license tag.

---
 accel/tcg/monitor.c   | 193 ---------------------------------------
 accel/tcg/tcg-stats.c | 206 ++++++++++++++++++++++++++++++++++++++++++
 accel/tcg/meson.build |   1 +
 3 files changed, 207 insertions(+), 193 deletions(-)
 create mode 100644 accel/tcg/tcg-stats.c
accel/tcg/monitor.c +0 -193
index e7ed7281a4b..adb9de5a1c6 100644
@@ -7,205 +7,12 @@
  */
 
 #include "qemu/osdep.h"
-#include "qemu/accel.h"
-#include "qemu/qht.h"
 #include "qapi/error.h"
 #include "qapi/type-helpers.h"
 #include "qapi/qapi-commands-machine.h"
 #include "monitor/monitor.h"
-#include "system/cpu-timers.h"
-#include "exec/icount.h"
 #include "system/tcg.h"
-#include "tcg/tcg.h"
 #include "internal-common.h"
-#include "tb-context.h"
-
-
-static void dump_drift_info(GString *buf)
-{
-    if (!icount_enabled()) {
-        return;
-    }
-
-    g_string_append_printf(buf, "Host - Guest clock  %"PRIi64" ms\n",
-                           (cpu_get_clock() - icount_get()) / SCALE_MS);
-    if (icount_align_option) {
-        g_string_append_printf(buf, "Max guest delay     %"PRIi64" ms\n",
-                               -max_delay / SCALE_MS);
-        g_string_append_printf(buf, "Max guest advance   %"PRIi64" ms\n",
-                               max_advance / SCALE_MS);
-    } else {
-        g_string_append_printf(buf, "Max guest delay     NA\n");
-        g_string_append_printf(buf, "Max guest advance   NA\n");
-    }
-}
-
-static void dump_accel_info(GString *buf)
-{
-    AccelState *accel = current_accel();
-    bool one_insn_per_tb = object_property_get_bool(OBJECT(accel),
-                                                    "one-insn-per-tb",
-                                                    &error_fatal);
-
-    g_string_append_printf(buf, "Accelerator settings:\n");
-    g_string_append_printf(buf, "one-insn-per-tb: %s\n\n",
-                           one_insn_per_tb ? "on" : "off");
-}
-
-static void print_qht_statistics(struct qht_stats hst, GString *buf)
-{
-    uint32_t hgram_opts;
-    size_t hgram_bins;
-    char *hgram;
-
-    if (!hst.head_buckets) {
-        return;
-    }
-    g_string_append_printf(buf, "TB hash buckets     %zu/%zu "
-                           "(%0.2f%% head buckets used)\n",
-                           hst.used_head_buckets, hst.head_buckets,
-                           (double)hst.used_head_buckets /
-                           hst.head_buckets * 100);
-
-    hgram_opts =  QDIST_PR_BORDER | QDIST_PR_LABELS;
-    hgram_opts |= QDIST_PR_100X   | QDIST_PR_PERCENT;
-    if (qdist_xmax(&hst.occupancy) - qdist_xmin(&hst.occupancy) == 1) {
-        hgram_opts |= QDIST_PR_NODECIMAL;
-    }
-    hgram = qdist_pr(&hst.occupancy, 10, hgram_opts);
-    g_string_append_printf(buf, "TB hash occupancy   %0.2f%% avg chain occ. "
-                           "Histogram: %s\n",
-                           qdist_avg(&hst.occupancy) * 100, hgram);
-    g_free(hgram);
-
-    hgram_opts = QDIST_PR_BORDER | QDIST_PR_LABELS;
-    hgram_bins = qdist_xmax(&hst.chain) - qdist_xmin(&hst.chain);
-    if (hgram_bins > 10) {
-        hgram_bins = 10;
-    } else {
-        hgram_bins = 0;
-        hgram_opts |= QDIST_PR_NODECIMAL | QDIST_PR_NOBINRANGE;
-    }
-    hgram = qdist_pr(&hst.chain, hgram_bins, hgram_opts);
-    g_string_append_printf(buf, "TB hash avg chain   %0.3f buckets. "
-                           "Histogram: %s\n",
-                           qdist_avg(&hst.chain), hgram);
-    g_free(hgram);
-}
-
-struct tb_tree_stats {
-    size_t nb_tbs;
-    size_t host_size;
-    size_t target_size;
-    size_t max_target_size;
-    size_t direct_jmp_count;
-    size_t direct_jmp2_count;
-    size_t cross_page;
-};
-
-static gboolean tb_tree_stats_iter(gpointer key, gpointer value, gpointer data)
-{
-    const TranslationBlock *tb = value;
-    struct tb_tree_stats *tst = data;
-
-    tst->nb_tbs++;
-    tst->host_size += tb->tc.size;
-    tst->target_size += tb->size;
-    if (tb->size > tst->max_target_size) {
-        tst->max_target_size = tb->size;
-    }
-    if (tb->page_addr[1] != -1) {
-        tst->cross_page++;
-    }
-    if (tb->jmp_reset_offset[0] != TB_JMP_OFFSET_INVALID) {
-        tst->direct_jmp_count++;
-        if (tb->jmp_reset_offset[1] != TB_JMP_OFFSET_INVALID) {
-            tst->direct_jmp2_count++;
-        }
-    }
-    return false;
-}
-
-static void tlb_flush_counts(size_t *pfull, size_t *ppart, size_t *pelide)
-{
-    CPUState *cpu;
-    size_t full = 0, part = 0, elide = 0;
-
-    CPU_FOREACH(cpu) {
-        full += qatomic_read(&cpu->neg.tlb.c.full_flush_count);
-        part += qatomic_read(&cpu->neg.tlb.c.part_flush_count);
-        elide += qatomic_read(&cpu->neg.tlb.c.elide_flush_count);
-    }
-    *pfull = full;
-    *ppart = part;
-    *pelide = elide;
-}
-
-static void tcg_dump_flush_info(GString *buf)
-{
-    size_t flush_full, flush_part, flush_elide;
-
-    g_string_append_printf(buf, "TB flush count      %u\n",
-                           qatomic_read(&tb_ctx.tb_flush_count));
-    g_string_append_printf(buf, "TB invalidate count %u\n",
-                           qatomic_read(&tb_ctx.tb_phys_invalidate_count));
-
-    tlb_flush_counts(&flush_full, &flush_part, &flush_elide);
-    g_string_append_printf(buf, "TLB full flushes    %zu\n", flush_full);
-    g_string_append_printf(buf, "TLB partial flushes %zu\n", flush_part);
-    g_string_append_printf(buf, "TLB elided flushes  %zu\n", flush_elide);
-}
-
-static void dump_exec_info(GString *buf)
-{
-    struct tb_tree_stats tst = {};
-    struct qht_stats hst;
-    size_t nb_tbs;
-
-    tcg_tb_foreach(tb_tree_stats_iter, &tst);
-    nb_tbs = tst.nb_tbs;
-    /* XXX: avoid using doubles ? */
-    g_string_append_printf(buf, "Translation buffer state:\n");
-    /*
-     * Report total code size including the padding and TB structs;
-     * otherwise users might think "-accel tcg,tb-size" is not honoured.
-     * For avg host size we use the precise numbers from tb_tree_stats though.
-     */
-    g_string_append_printf(buf, "gen code size       %zu/%zu\n",
-                           tcg_code_size(), tcg_code_capacity());
-    g_string_append_printf(buf, "TB count            %zu\n", nb_tbs);
-    g_string_append_printf(buf, "TB avg target size  %zu max=%zu bytes\n",
-                           nb_tbs ? tst.target_size / nb_tbs : 0,
-                           tst.max_target_size);
-    g_string_append_printf(buf, "TB avg host size    %zu bytes "
-                           "(expansion ratio: %0.1f)\n",
-                           nb_tbs ? tst.host_size / nb_tbs : 0,
-                           tst.target_size ?
-                           (double)tst.host_size / tst.target_size : 0);
-    g_string_append_printf(buf, "cross page TB count %zu (%zu%%)\n",
-                           tst.cross_page,
-                           nb_tbs ? (tst.cross_page * 100) / nb_tbs : 0);
-    g_string_append_printf(buf, "direct jump count   %zu (%zu%%) "
-                           "(2 jumps=%zu %zu%%)\n",
-                           tst.direct_jmp_count,
-                           nb_tbs ? (tst.direct_jmp_count * 100) / nb_tbs : 0,
-                           tst.direct_jmp2_count,
-                           nb_tbs ? (tst.direct_jmp2_count * 100) / nb_tbs : 0);
-
-    qht_statistics_init(&tb_ctx.htable, &hst);
-    print_qht_statistics(hst, buf);
-    qht_statistics_destroy(&hst);
-
-    g_string_append_printf(buf, "\nStatistics:\n");
-    tcg_dump_flush_info(buf);
-}
-
-void tcg_dump_stats(GString *buf)
-{
-    dump_accel_info(buf);
-    dump_exec_info(buf);
-    dump_drift_info(buf);
-}
 
 HumanReadableText *qmp_x_query_jit(Error **errp)
 {
accel/tcg/tcg-stats.c new file +206 -0

206 lines not on this page · raw message

accel/tcg/meson.build +1 -0
index 575e92bb9e8..002aa8f4588 100644
@@ -11,6 +11,7 @@ tcg_ss.add(files(
   'tcg-runtime-gvec.c',
   'tb-maint.c',
   'tcg-all.c',
+  'tcg-stats.c',
   'translate-all.c',
   'translator.c',
 ))
Signature · 1 line
2.49.0
Reviewed-by on v6 11/39
#38 lore raw

Re: [PATCH v6 11/39] accel/tcg: Extract statistic related code to tcg-stats.c

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:17 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 11/39] accel/tcg: Extract statistic related code to
tcg-stats.c
X-Mailer: git-send-email 2.49.0

Statistic code is not specific to system emulation (except cross-page checks) and can be used to analyze user-mode binaries. Extract statistic related code to its own file: tcg-stats.c, keeping the original LGPL-2.1-or-later license tag.

Quoted 7 lines
---
 accel/tcg/monitor.c   | 193 ---------------------------------------
 accel/tcg/tcg-stats.c | 206 ++++++++++++++++++++++++++++++++++++++++++
 accel/tcg/meson.build |   1 +
 3 files changed, 207 insertions(+), 193 deletions(-)
 create mode 100644 accel/tcg/tcg-stats.c
#39 lore raw

PATCH v6 12/39 accel: Move supports_guest_debug() declaration to AccelClass

In reply to Philippe Mathieu-Daudé

AccelOpsClass is for methods dealing with vCPUs. When only dealing with AccelState, AccelClass is sufficient.

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Restrict kvm_supports_guest_debug() scope.

---
 accel/kvm/kvm-cpus.h       | 1 -
 include/qemu/accel.h       | 1 +
 include/system/accel-ops.h | 1 -
 include/system/hvf.h       | 2 +-
 accel/hvf/hvf-accel-ops.c  | 2 +-
 accel/kvm/kvm-accel-ops.c  | 1 -
 accel/kvm/kvm-all.c        | 5 ++++-
 accel/tcg/tcg-accel-ops.c  | 6 ------
 accel/tcg/tcg-all.c        | 6 ++++++
 gdbstub/system.c           | 8 +++++---
 target/arm/hvf/hvf.c       | 2 +-
 target/i386/hvf/hvf.c      | 2 +-
 12 files changed, 20 insertions(+), 17 deletions(-)
accel/kvm/kvm-cpus.h +0 -1
index 688511151c8..3185659562d 100644
@@ -16,7 +16,6 @@ void kvm_destroy_vcpu(CPUState *cpu);
 void kvm_cpu_synchronize_post_reset(CPUState *cpu);
 void kvm_cpu_synchronize_post_init(CPUState *cpu);
 void kvm_cpu_synchronize_pre_loadvm(CPUState *cpu);
-bool kvm_supports_guest_debug(void);
 int kvm_insert_breakpoint(CPUState *cpu, int type, vaddr addr, vaddr len);
 int kvm_remove_breakpoint(CPUState *cpu, int type, vaddr addr, vaddr len);
 void kvm_remove_all_breakpoints(CPUState *cpu);
include/qemu/accel.h +1 -0
index fbd3d897fef..fb176e89bad 100644
@@ -47,6 +47,7 @@ typedef struct AccelClass {
                        hwaddr start_addr, hwaddr size);
 
     /* gdbstub related hooks */
+    bool (*supports_guest_debug)(AccelState *as);
     int (*gdbstub_supported_sstep_flags)(void);
 
     bool *allowed;
include/system/accel-ops.h +0 -1
index 4c99d25aeff..700df92ac6d 100644
@@ -63,7 +63,6 @@ struct AccelOpsClass {
     int64_t (*get_elapsed_ticks)(void);
 
     /* gdbstub hooks */
-    bool (*supports_guest_debug)(void);
     int (*update_guest_debug)(CPUState *cpu);
     int (*insert_breakpoint)(CPUState *cpu, int type, vaddr addr, vaddr len);
     int (*remove_breakpoint)(CPUState *cpu, int type, vaddr addr, vaddr len);
include/system/hvf.h +1 -1
index a9a502f0c8f..a9fd13d9bba 100644
@@ -71,7 +71,7 @@ void hvf_arch_update_guest_debug(CPUState *cpu);
 /*
  * Return whether the guest supports debugging.
  */
-bool hvf_arch_supports_guest_debug(void);
+bool hvf_arch_supports_guest_debug(AccelState *as);
 #endif /* COMPILING_PER_TARGET */
 
 #endif
accel/hvf/hvf-accel-ops.c +1 -1
index b38977207d2..aed791a3f3e 100644
@@ -366,6 +366,7 @@ static void hvf_accel_class_init(ObjectClass *oc, const void *data)
     ac->name = "HVF";
     ac->init_machine = hvf_accel_init;
     ac->allowed = &hvf_allowed;
+    ac->supports_guest_debug = hvf_arch_supports_guest_debug;
     ac->gdbstub_supported_sstep_flags = hvf_gdbstub_sstep_flags;
 }
 
@@ -600,7 +601,6 @@ static void hvf_accel_ops_class_init(ObjectClass *oc, const void *data)
     ops->remove_breakpoint = hvf_remove_breakpoint;
     ops->remove_all_breakpoints = hvf_remove_all_breakpoints;
     ops->update_guest_debug = hvf_update_guest_debug;
-    ops->supports_guest_debug = hvf_arch_supports_guest_debug;
 };
 static const TypeInfo hvf_accel_ops_type = {
     .name = ACCEL_OPS_NAME("hvf"),
accel/kvm/kvm-accel-ops.c +0 -1
index e5c15449aa6..96606090889 100644
@@ -104,7 +104,6 @@ static void kvm_accel_ops_class_init(ObjectClass *oc, const void *data)
 
 #ifdef TARGET_KVM_HAVE_GUEST_DEBUG
     ops->update_guest_debug = kvm_update_guest_debug_ops;
-    ops->supports_guest_debug = kvm_supports_guest_debug;
     ops->insert_breakpoint = kvm_insert_breakpoint;
     ops->remove_breakpoint = kvm_remove_breakpoint;
     ops->remove_all_breakpoints = kvm_remove_all_breakpoints;
accel/kvm/kvm-all.c +4 -1
index 17235f26464..c8611552d19 100644
@@ -3533,7 +3533,7 @@ int kvm_update_guest_debug(CPUState *cpu, unsigned long reinject_trap)
     return data.err;
 }
 
-bool kvm_supports_guest_debug(void)
+static bool kvm_supports_guest_debug(AccelState *as)
 {
     /* probed during kvm_init() */
     return kvm_has_guest_debug;
@@ -3998,6 +3998,9 @@ static void kvm_accel_class_init(ObjectClass *oc, const void *data)
     ac->has_memory = kvm_accel_has_memory;
     ac->allowed = &kvm_allowed;
     ac->gdbstub_supported_sstep_flags = kvm_gdbstub_sstep_flags;
+#ifdef TARGET_KVM_HAVE_GUEST_DEBUG
+    ac->supports_guest_debug = kvm_supports_guest_debug;
+#endif
 
     object_class_property_add(oc, "kernel-irqchip", "on|off|split",
         NULL, kvm_set_kernel_irqchip,
accel/tcg/tcg-accel-ops.c +0 -6
index 6116644d1c0..f579685a611 100644
@@ -106,11 +106,6 @@ void tcg_handle_interrupt(CPUState *cpu, int mask)
     }
 }
 
-static bool tcg_supports_guest_debug(void)
-{
-    return true;
-}
-
 /* Translate GDB watchpoint type to a flags value for cpu_watchpoint_* */
 static inline int xlat_gdb_type(CPUState *cpu, int gdbtype)
 {
@@ -216,7 +211,6 @@ static void tcg_accel_ops_init(AccelOpsClass *ops)
     }
 
     ops->cpu_reset_hold = tcg_cpu_reset_hold;
-    ops->supports_guest_debug = tcg_supports_guest_debug;
     ops->insert_breakpoint = tcg_insert_breakpoint;
     ops->remove_breakpoint = tcg_remove_breakpoint;
     ops->remove_all_breakpoints = tcg_remove_all_breakpoints;
accel/tcg/tcg-all.c +6 -0
index 6e5dc333d59..0cff0f8a0f9 100644
@@ -219,6 +219,11 @@ static void tcg_set_one_insn_per_tb(Object *obj, bool value, Error **errp)
     qatomic_set(&one_insn_per_tb, value);
 }
 
+static bool tcg_supports_guest_debug(AccelState *as)
+{
+    return true;
+}
+
 static int tcg_gdbstub_supported_sstep_flags(void)
 {
     /*
@@ -242,6 +247,7 @@ static void tcg_accel_class_init(ObjectClass *oc, const void *data)
     ac->cpu_common_realize = tcg_exec_realizefn;
     ac->cpu_common_unrealize = tcg_exec_unrealizefn;
     ac->allowed = &tcg_allowed;
+    ac->supports_guest_debug = tcg_supports_guest_debug;
     ac->gdbstub_supported_sstep_flags = tcg_gdbstub_supported_sstep_flags;
 
     object_class_property_add_str(oc, "thread",
gdbstub/system.c +5 -3
index 8a32d8e1a1d..03934deed49 100644
@@ -13,6 +13,7 @@
 #include "qemu/osdep.h"
 #include "qapi/error.h"
 #include "qemu/error-report.h"
+#include "qemu/accel.h"
 #include "qemu/cutils.h"
 #include "exec/gdbstub.h"
 #include "gdbstub/syscalls.h"
@@ -634,9 +635,10 @@ int gdb_signal_to_target(int sig)
 
 bool gdb_supports_guest_debug(void)
 {
-    const AccelOpsClass *ops = cpus_get_accel();
-    if (ops->supports_guest_debug) {
-        return ops->supports_guest_debug();
+    AccelState *accel = current_accel();
+    AccelClass *acc = ACCEL_GET_CLASS(accel);
+    if (acc->supports_guest_debug) {
+        return acc->supports_guest_debug(accel);
     }
     return false;
 }
target/arm/hvf/hvf.c +1 -1
index 7b6d291e79c..995a6a74b06 100644
@@ -2352,7 +2352,7 @@ void hvf_arch_update_guest_debug(CPUState *cpu)
     hvf_arch_set_traps(cpu);
 }
 
-bool hvf_arch_supports_guest_debug(void)
+bool hvf_arch_supports_guest_debug(AccelState *as)
 {
     return true;
 }
target/i386/hvf/hvf.c +1 -1
index 99e37a33e50..b25bff0df41 100644
@@ -1019,7 +1019,7 @@ void hvf_arch_update_guest_debug(CPUState *cpu)
 {
 }
 
-bool hvf_arch_supports_guest_debug(void)
+bool hvf_arch_supports_guest_debug(AccelState *as)
 {
     return false;
 }
Signature · 1 line
2.49.0
Reviewed-by on v6 12/39
#40 lore raw

Re: [PATCH v6 12/39] accel: Move supports_guest_debug() declaration to AccelClass

Quoted 10 lines · Philippe Mathieu-Daudé

AccelOpsClass is for methods dealing with vCPUs. When only dealing with AccelState, AccelClass is sufficient.

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Restrict kvm_supports_guest_debug() scope.

Reviewed-by on v6 12/39
#41 lore raw

Re: [PATCH v6 12/39] accel: Move supports_guest_debug() declaration to AccelClass

In reply to Philippe Mathieu-Daudé
Quoted 16 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:18 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 12/39] accel: Move supports_guest_debug() declaration to
AccelClass
X-Mailer: git-send-email 2.49.0

AccelOpsClass is for methods dealing with vCPUs. When only dealing with AccelState, AccelClass is sufficient.

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Restrict kvm_supports_guest_debug() scope.

Quoted 14 lines
---
 accel/kvm/kvm-cpus.h       | 1 -
 include/qemu/accel.h       | 1 +
 include/system/accel-ops.h | 1 -
 include/system/hvf.h       | 2 +-
 accel/hvf/hvf-accel-ops.c  | 2 +-
 accel/kvm/kvm-accel-ops.c  | 1 -
 accel/kvm/kvm-all.c        | 5 ++++-
 accel/tcg/tcg-accel-ops.c  | 6 ------
 accel/tcg/tcg-all.c        | 6 ++++++
 gdbstub/system.c           | 8 +++++---
 target/arm/hvf/hvf.c       | 2 +-
 target/i386/hvf/hvf.c      | 2 +-
 12 files changed, 20 insertions(+), 17 deletions(-)
#42 lore raw

PATCH v6 13/39 accel: Move cpus_are_resettable() declaration to AccelClass

In reply to Philippe Mathieu-Daudé

AccelOpsClass is for methods dealing with vCPUs. When only dealing with AccelState, AccelClass is sufficient.

Move cpus_are_resettable() declaration to accel/accel-system.c.

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Adapt KVM handler.

---
 include/qemu/accel.h       |  1 +
 include/system/accel-ops.h |  1 -
 accel/accel-system.c       | 10 ++++++++++
 accel/kvm/kvm-accel-ops.c  |  6 ------
 accel/kvm/kvm-all.c        |  6 ++++++
 system/cpus.c              |  8 --------
 6 files changed, 17 insertions(+), 15 deletions(-)
include/qemu/accel.h +1 -0
index fb176e89bad..f987d16baaa 100644
@@ -45,6 +45,7 @@ typedef struct AccelClass {
     void (*setup_post)(MachineState *ms, AccelState *accel);
     bool (*has_memory)(MachineState *ms, AddressSpace *as,
                        hwaddr start_addr, hwaddr size);
+    bool (*cpus_are_resettable)(AccelState *as);
 
     /* gdbstub related hooks */
     bool (*supports_guest_debug)(AccelState *as);
include/system/accel-ops.h +0 -1
index 700df92ac6d..f19245d0a0e 100644
@@ -33,7 +33,6 @@ struct AccelOpsClass {
     /* initialization function called when accel is chosen */
     void (*ops_init)(AccelOpsClass *ops);
 
-    bool (*cpus_are_resettable)(void);
     void (*cpu_reset_hold)(CPUState *cpu);
 
     void (*create_vcpu_thread)(CPUState *cpu); /* MANDATORY NON-NULL */
accel/accel-system.c +10 -0
index a0f562ae9ff..07b75dae797 100644
@@ -62,6 +62,16 @@ void accel_setup_post(MachineState *ms)
     }
 }
 
+bool cpus_are_resettable(void)
+{
+    AccelState *accel = current_accel();
+    AccelClass *acc = ACCEL_GET_CLASS(accel);
+    if (acc->cpus_are_resettable) {
+        return acc->cpus_are_resettable(accel);
+    }
+    return true;
+}
+
 /* initialize the arch-independent accel operation interfaces */
 void accel_init_ops_interfaces(AccelClass *ac)
 {
accel/kvm/kvm-accel-ops.c +0 -6
index 96606090889..99f61044da5 100644
@@ -78,11 +78,6 @@ static bool kvm_vcpu_thread_is_idle(CPUState *cpu)
     return !kvm_halt_in_kernel();
 }
 
-static bool kvm_cpus_are_resettable(void)
-{
-    return !kvm_enabled() || !kvm_state->guest_state_protected;
-}
-
 #ifdef TARGET_KVM_HAVE_GUEST_DEBUG
 static int kvm_update_guest_debug_ops(CPUState *cpu)
 {
@@ -96,7 +91,6 @@ static void kvm_accel_ops_class_init(ObjectClass *oc, const void *data)
 
     ops->create_vcpu_thread = kvm_start_vcpu_thread;
     ops->cpu_thread_is_idle = kvm_vcpu_thread_is_idle;
-    ops->cpus_are_resettable = kvm_cpus_are_resettable;
     ops->synchronize_post_reset = kvm_cpu_synchronize_post_reset;
     ops->synchronize_post_init = kvm_cpu_synchronize_post_init;
     ops->synchronize_state = kvm_cpu_synchronize_state;
accel/kvm/kvm-all.c +6 -0
index c8611552d19..88fb6d36941 100644
@@ -3979,6 +3979,11 @@ static void kvm_accel_instance_init(Object *obj)
     s->msr_energy.enable = false;
 }
 
+static bool kvm_cpus_are_resettable(AccelState *as)
+{
+    return !kvm_enabled() || !kvm_state->guest_state_protected;
+}
+
 /**
  * kvm_gdbstub_sstep_flags():
  *
@@ -3997,6 +4002,7 @@ static void kvm_accel_class_init(ObjectClass *oc, const void *data)
     ac->init_machine = kvm_init;
     ac->has_memory = kvm_accel_has_memory;
     ac->allowed = &kvm_allowed;
+    ac->cpus_are_resettable = kvm_cpus_are_resettable;
     ac->gdbstub_supported_sstep_flags = kvm_gdbstub_sstep_flags;
 #ifdef TARGET_KVM_HAVE_GUEST_DEBUG
     ac->supports_guest_debug = kvm_supports_guest_debug;
system/cpus.c +0 -8
index a43e0e4e796..4fb764ac880 100644
@@ -195,14 +195,6 @@ void cpu_synchronize_pre_loadvm(CPUState *cpu)
     }
 }
 
-bool cpus_are_resettable(void)
-{
-    if (cpus_accel->cpus_are_resettable) {
-        return cpus_accel->cpus_are_resettable();
-    }
-    return true;
-}
-
 void cpu_exec_reset_hold(CPUState *cpu)
 {
     if (cpus_accel->cpu_reset_hold) {
Signature · 1 line
2.49.0
#43 lore raw

Re: [PATCH v6 13/39] accel: Move cpus_are_resettable() declaration to AccelClass

Quoted 4 lines · Philippe Mathieu-Daudé

AccelOpsClass is for methods dealing with vCPUs. When only dealing with AccelState, AccelClass is sufficient.

Move cpus_are_resettable() declaration to accel/accel-system.c.

I don't think this is necessary unless a solid justfication provided.

One straightfroward question against it, is why don't move gdb_supports_guest_debug() to accel/accel-system.c as well in the patch 12.

Quoted 2 lines

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Is this the essential preparation for split-accel work?

Quoted 125 lines

Adapt KVM handler.

---
  include/qemu/accel.h       |  1 +
  include/system/accel-ops.h |  1 -
  accel/accel-system.c       | 10 ++++++++++
  accel/kvm/kvm-accel-ops.c  |  6 ------
  accel/kvm/kvm-all.c        |  6 ++++++
  system/cpus.c              |  8 --------
  6 files changed, 17 insertions(+), 15 deletions(-)
include/qemu/accel.h +1 -0
index fb176e89bad..f987d16baaa 100644
@@ -45,6 +45,7 @@ typedef struct AccelClass {
      void (*setup_post)(MachineState *ms, AccelState *accel);
      bool (*has_memory)(MachineState *ms, AddressSpace *as,
                         hwaddr start_addr, hwaddr size);
+    bool (*cpus_are_resettable)(AccelState *as);
 
      /* gdbstub related hooks */
      bool (*supports_guest_debug)(AccelState *as);
include/system/accel-ops.h +0 -1
index 700df92ac6d..f19245d0a0e 100644
@@ -33,7 +33,6 @@ struct AccelOpsClass {
      /* initialization function called when accel is chosen */
      void (*ops_init)(AccelOpsClass *ops);
 
-    bool (*cpus_are_resettable)(void);
      void (*cpu_reset_hold)(CPUState *cpu);
 
      void (*create_vcpu_thread)(CPUState *cpu); /* MANDATORY NON-NULL */
accel/accel-system.c +10 -0
index a0f562ae9ff..07b75dae797 100644
@@ -62,6 +62,16 @@ void accel_setup_post(MachineState *ms)
      }
  }
 
+bool cpus_are_resettable(void)
+{
+    AccelState *accel = current_accel();
+    AccelClass *acc = ACCEL_GET_CLASS(accel);
+    if (acc->cpus_are_resettable) {
+        return acc->cpus_are_resettable(accel);
+    }
+    return true;
+}
+
  /* initialize the arch-independent accel operation interfaces */
  void accel_init_ops_interfaces(AccelClass *ac)
  {
accel/kvm/kvm-accel-ops.c +0 -6
index 96606090889..99f61044da5 100644
@@ -78,11 +78,6 @@ static bool kvm_vcpu_thread_is_idle(CPUState *cpu)
      return !kvm_halt_in_kernel();
  }
 
-static bool kvm_cpus_are_resettable(void)
-{
-    return !kvm_enabled() || !kvm_state->guest_state_protected;
-}
-
  #ifdef TARGET_KVM_HAVE_GUEST_DEBUG
  static int kvm_update_guest_debug_ops(CPUState *cpu)
  {
@@ -96,7 +91,6 @@ static void kvm_accel_ops_class_init(ObjectClass *oc, const void *data)
 
      ops->create_vcpu_thread = kvm_start_vcpu_thread;
      ops->cpu_thread_is_idle = kvm_vcpu_thread_is_idle;
-    ops->cpus_are_resettable = kvm_cpus_are_resettable;
      ops->synchronize_post_reset = kvm_cpu_synchronize_post_reset;
      ops->synchronize_post_init = kvm_cpu_synchronize_post_init;
      ops->synchronize_state = kvm_cpu_synchronize_state;
accel/kvm/kvm-all.c +6 -0
index c8611552d19..88fb6d36941 100644
@@ -3979,6 +3979,11 @@ static void kvm_accel_instance_init(Object *obj)
      s->msr_energy.enable = false;
  }
 
+static bool kvm_cpus_are_resettable(AccelState *as)
+{
+    return !kvm_enabled() || !kvm_state->guest_state_protected;
+}
+
  /**
   * kvm_gdbstub_sstep_flags():
   *
@@ -3997,6 +4002,7 @@ static void kvm_accel_class_init(ObjectClass *oc, const void *data)
      ac->init_machine = kvm_init;
      ac->has_memory = kvm_accel_has_memory;
      ac->allowed = &kvm_allowed;
+    ac->cpus_are_resettable = kvm_cpus_are_resettable;
      ac->gdbstub_supported_sstep_flags = kvm_gdbstub_sstep_flags;
  #ifdef TARGET_KVM_HAVE_GUEST_DEBUG
      ac->supports_guest_debug = kvm_supports_guest_debug;
system/cpus.c +0 -8
index a43e0e4e796..4fb764ac880 100644
@@ -195,14 +195,6 @@ void cpu_synchronize_pre_loadvm(CPUState *cpu)
      }
  }
 
-bool cpus_are_resettable(void)
-{
-    if (cpus_accel->cpus_are_resettable) {
-        return cpus_accel->cpus_are_resettable();
-    }
-    return true;
-}
-
  void cpu_exec_reset_hold(CPUState *cpu)
  {
      if (cpus_accel->cpu_reset_hold) {
#44 lore raw

Re: [PATCH v6 13/39] accel: Move cpus_are_resettable() declaration to AccelClass

Quoted 9 lines · Xiaoyao Li
Quoted 4 lines · Philippe Mathieu-Daudé

AccelOpsClass is for methods dealing with vCPUs. When only dealing with AccelState, AccelClass is sufficient.

Move cpus_are_resettable() declaration to accel/accel-system.c.

I don't think this is necessary unless a solid justfication provided.

One straightfroward question against it, is why don't move gdb_supports_guest_debug() to accel/accel-system.c as well in the patch 12.

gdb_supports_guest_debug() is used in both user / system emulation.

Quoted 5 lines
Quoted 3 lines

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Is this the essential preparation for split-accel work?

Yes, but also the aim is to better organize and clarify the APIs.

We have a set of generic handlers that work on the AccelState context, regardless of vCPUs; and another set which works on vCPUs within Accel and must get the vCPU context by argument.

Quoted 13 lines
Quoted 13 lines

Adapt KVM handler.

---
  include/qemu/accel.h       |  1 +
  include/system/accel-ops.h |  1 -
  accel/accel-system.c       | 10 ++++++++++
  accel/kvm/kvm-accel-ops.c  |  6 ------
  accel/kvm/kvm-all.c        |  6 ++++++
  system/cpus.c              |  8 --------

  6 files changed, 17 insertions(+), 15 deletions(-)
Reviewed-by on v6 13/39
#45 lore raw

Re: [PATCH v6 13/39] accel: Move cpus_are_resettable() declaration to AccelClass

In reply to Philippe Mathieu-Daudé
Quoted 18 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:19 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 13/39] accel: Move cpus_are_resettable() declaration to
AccelClass
X-Mailer: git-send-email 2.49.0

AccelOpsClass is for methods dealing with vCPUs. When only dealing with AccelState, AccelClass is sufficient.

Move cpus_are_resettable() declaration to accel/accel-system.c.

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Adapt KVM handler.

Quoted 8 lines
---
 include/qemu/accel.h       |  1 +
 include/system/accel-ops.h |  1 -
 accel/accel-system.c       | 10 ++++++++++
 accel/kvm/kvm-accel-ops.c  |  6 ------
 accel/kvm/kvm-all.c        |  6 ++++++
 system/cpus.c              |  8 --------
 6 files changed, 17 insertions(+), 15 deletions(-)
#46 lore raw

PATCH v6 14/39 accel/hvf: Restrict internal declarations

In reply to Philippe Mathieu-Daudé

Common code only needs to know whether HVF is enabled and the QOM type. Move the rest to "hvf_int.h", removing the need for COMPILING_PER_TARGET #ifdef'ry.

---
 include/system/hvf.h     | 38 --------------------------------------
 include/system/hvf_int.h | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 34 insertions(+), 38 deletions(-)
include/system/hvf.h +0 -38
index a9fd13d9bba..d3dcf088b3f 100644
@@ -14,10 +14,6 @@
 #define HVF_H
 
 #include "qemu/accel.h"
-#include "qemu/queue.h"
-#include "exec/vaddr.h"
-#include "qom/object.h"
-#include "exec/vaddr.h"
 
 #ifdef COMPILING_PER_TARGET
 # ifdef CONFIG_HVF
@@ -40,38 +36,4 @@ typedef struct HVFState HVFState;
 DECLARE_INSTANCE_CHECKER(HVFState, HVF_STATE,
                          TYPE_HVF_ACCEL)
 
-#ifdef COMPILING_PER_TARGET
-struct hvf_sw_breakpoint {
-    vaddr pc;
-    vaddr saved_insn;
-    int use_count;
-    QTAILQ_ENTRY(hvf_sw_breakpoint) entry;
-};
-
-struct hvf_sw_breakpoint *hvf_find_sw_breakpoint(CPUState *cpu,
-                                                 vaddr pc);
-int hvf_sw_breakpoints_active(CPUState *cpu);
-
-int hvf_arch_insert_sw_breakpoint(CPUState *cpu, struct hvf_sw_breakpoint *bp);
-int hvf_arch_remove_sw_breakpoint(CPUState *cpu, struct hvf_sw_breakpoint *bp);
-int hvf_arch_insert_hw_breakpoint(vaddr addr, vaddr len, int type);
-int hvf_arch_remove_hw_breakpoint(vaddr addr, vaddr len, int type);
-void hvf_arch_remove_all_hw_breakpoints(void);
-
-/*
- * hvf_update_guest_debug:
- * @cs: CPUState for the CPU to update
- *
- * Update guest to enable or disable debugging. Per-arch specifics will be
- * handled by calling down to hvf_arch_update_guest_debug.
- */
-int hvf_update_guest_debug(CPUState *cpu);
-void hvf_arch_update_guest_debug(CPUState *cpu);
-
-/*
- * Return whether the guest supports debugging.
- */
-bool hvf_arch_supports_guest_debug(AccelState *as);
-#endif /* COMPILING_PER_TARGET */
-
 #endif
include/system/hvf_int.h +34 -0
index d774e58df91..dcf06f53d97 100644
@@ -12,6 +12,8 @@
 #define HVF_INT_H
 
 #include "qemu/queue.h"
+#include "exec/vaddr.h"
+#include "qom/object.h"
 
 #ifdef __aarch64__
 #include <Hypervisor/Hypervisor.h>
@@ -77,4 +79,36 @@ int hvf_put_registers(CPUState *);
 int hvf_get_registers(CPUState *);
 void hvf_kick_vcpu_thread(CPUState *cpu);
 
+struct hvf_sw_breakpoint {
+    vaddr pc;
+    vaddr saved_insn;
+    int use_count;
+    QTAILQ_ENTRY(hvf_sw_breakpoint) entry;
+};
+
+struct hvf_sw_breakpoint *hvf_find_sw_breakpoint(CPUState *cpu,
+                                                 vaddr pc);
+int hvf_sw_breakpoints_active(CPUState *cpu);
+
+int hvf_arch_insert_sw_breakpoint(CPUState *cpu, struct hvf_sw_breakpoint *bp);
+int hvf_arch_remove_sw_breakpoint(CPUState *cpu, struct hvf_sw_breakpoint *bp);
+int hvf_arch_insert_hw_breakpoint(vaddr addr, vaddr len, int type);
+int hvf_arch_remove_hw_breakpoint(vaddr addr, vaddr len, int type);
+void hvf_arch_remove_all_hw_breakpoints(void);
+
+/*
+ * hvf_update_guest_debug:
+ * @cs: CPUState for the CPU to update
+ *
+ * Update guest to enable or disable debugging. Per-arch specifics will be
+ * handled by calling down to hvf_arch_update_guest_debug.
+ */
+int hvf_update_guest_debug(CPUState *cpu);
+void hvf_arch_update_guest_debug(CPUState *cpu);
+
+/*
+ * Return whether the guest supports debugging.
+ */
+bool hvf_arch_supports_guest_debug(AccelState *as);
+
 #endif
Signature · 1 line
2.49.0
Reviewed-by on v6 14/39
#48 lore raw

Re: [PATCH v6 14/39] accel/hvf: Restrict internal declarations

In reply to Philippe Mathieu-Daudé
Quoted 9 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:20 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 14/39] accel/hvf: Restrict internal declarations
X-Mailer: git-send-email 2.49.0

Common code only needs to know whether HVF is enabled and the QOM type. Move the rest to "hvf_int.h", removing the need for COMPILING_PER_TARGET #ifdef'ry.

Quoted 6 lines
---
 include/system/hvf.h     | 38 --------------------------------------
 include/system/hvf_int.h | 34 ++++++++++++++++++++++++++++++++++
 2 files changed, 34 insertions(+), 38 deletions(-)
#49 lore raw

PATCH v6 15/39 accel/hvf: Move per-cpu method declarations to hvf-accel-ops.c

In reply to Philippe Mathieu-Daudé

hvf-all.c aims to contain the generic accel methods (TYPE_ACCEL), while hvf-accel-ops.c the per-vcpu methods (TYPE_ACCEL_OPS).

---
 accel/hvf/hvf-accel-ops.c | 30 ++++++++++++++++++++++++++++++
 accel/hvf/hvf-all.c       | 28 ----------------------------
 2 files changed, 30 insertions(+), 28 deletions(-)
accel/hvf/hvf-accel-ops.c +30 -0
index aed791a3f3e..646cd240238 100644
@@ -50,9 +50,11 @@
 #include "qemu/osdep.h"
 #include "qemu/error-report.h"
 #include "qemu/main-loop.h"
+#include "qemu/queue.h"
 #include "system/address-spaces.h"
 #include "gdbstub/enums.h"
 #include "hw/boards.h"
+#include "hw/core/cpu.h"
 #include "system/accel-ops.h"
 #include "system/cpus.h"
 #include "system/hvf.h"
@@ -483,6 +485,34 @@ static void hvf_start_vcpu_thread(CPUState *cpu)
                        cpu, QEMU_THREAD_JOINABLE);
 }
 
+struct hvf_sw_breakpoint *hvf_find_sw_breakpoint(CPUState *cpu, vaddr pc)
+{
+    struct hvf_sw_breakpoint *bp;
+
+    QTAILQ_FOREACH(bp, &hvf_state->hvf_sw_breakpoints, entry) {
+        if (bp->pc == pc) {
+            return bp;
+        }
+    }
+    return NULL;
+}
+
+int hvf_sw_breakpoints_active(CPUState *cpu)
+{
+    return !QTAILQ_EMPTY(&hvf_state->hvf_sw_breakpoints);
+}
+
+static void do_hvf_update_guest_debug(CPUState *cpu, run_on_cpu_data arg)
+{
+    hvf_arch_update_guest_debug(cpu);
+}
+
+int hvf_update_guest_debug(CPUState *cpu)
+{
+    run_on_cpu(cpu, do_hvf_update_guest_debug, RUN_ON_CPU_NULL);
+    return 0;
+}
+
 static int hvf_insert_breakpoint(CPUState *cpu, int type, vaddr addr, vaddr len)
 {
     struct hvf_sw_breakpoint *bp;
accel/hvf/hvf-all.c +0 -28
index 8c387fda24d..1c72c43ddb4 100644
@@ -41,31 +41,3 @@ void assert_hvf_ok_impl(hv_return_t ret, const char *file, unsigned int line,
 
     abort();
 }
-
-struct hvf_sw_breakpoint *hvf_find_sw_breakpoint(CPUState *cpu, vaddr pc)
-{
-    struct hvf_sw_breakpoint *bp;
-
-    QTAILQ_FOREACH(bp, &hvf_state->hvf_sw_breakpoints, entry) {
-        if (bp->pc == pc) {
-            return bp;
-        }
-    }
-    return NULL;
-}
-
-int hvf_sw_breakpoints_active(CPUState *cpu)
-{
-    return !QTAILQ_EMPTY(&hvf_state->hvf_sw_breakpoints);
-}
-
-static void do_hvf_update_guest_debug(CPUState *cpu, run_on_cpu_data arg)
-{
-    hvf_arch_update_guest_debug(cpu);
-}
-
-int hvf_update_guest_debug(CPUState *cpu)
-{
-    run_on_cpu(cpu, do_hvf_update_guest_debug, RUN_ON_CPU_NULL);
-    return 0;
-}
Signature · 1 line
2.49.0
Reviewed-by on v6 15/39
#50 lore raw

Re: [PATCH v6 15/39] accel/hvf: Move per-cpu method declarations to hvf-accel-ops.c

Quoted 9 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:21 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 15/39] accel/hvf: Move per-cpu method declarations to
hvf-accel-ops.c
X-Mailer: git-send-email 2.49.0

hvf-all.c aims to contain the generic accel methods (TYPE_ACCEL), while hvf-accel-ops.c the per-vcpu methods (TYPE_ACCEL_OPS).

Quoted 6 lines
---
 accel/hvf/hvf-accel-ops.c | 30 ++++++++++++++++++++++++++++++
 accel/hvf/hvf-all.c       | 28 ----------------------------
 2 files changed, 30 insertions(+), 28 deletions(-)
#51 lore raw

PATCH v6 16/39 accel/hvf: Move generic method declarations to hvf-all.c

In reply to Philippe Mathieu-Daudé

hvf-all.c aims to contain the generic accel methods (TYPE_ACCEL), while hvf-accel-ops.c the per-vcpu methods (TYPE_ACCEL_OPS).

---
 accel/hvf/hvf-accel-ops.c | 274 +-------------------------------------
 accel/hvf/hvf-all.c       | 268 +++++++++++++++++++++++++++++++++++++
 2 files changed, 272 insertions(+), 270 deletions(-)
accel/hvf/hvf-accel-ops.c +4 -270
index 646cd240238..6b1a256deb5 100644
@@ -48,20 +48,16 @@
  */
 
 #include "qemu/osdep.h"
-#include "qemu/error-report.h"
+#include "qemu/guest-random.h"
 #include "qemu/main-loop.h"
 #include "qemu/queue.h"
-#include "system/address-spaces.h"
 #include "gdbstub/enums.h"
-#include "hw/boards.h"
+#include "exec/cpu-common.h"
 #include "hw/core/cpu.h"
 #include "system/accel-ops.h"
 #include "system/cpus.h"
 #include "system/hvf.h"
 #include "system/hvf_int.h"
-#include "system/runstate.h"
-#include "qemu/guest-random.h"
-#include "trace.h"
 
 HVFState *hvf_state;
 
@@ -81,132 +77,6 @@ hvf_slot *hvf_find_overlap_slot(uint64_t start, uint64_t size)
     return NULL;
 }
 
-struct mac_slot {
-    int present;
-    uint64_t size;
-    uint64_t gpa_start;
-    uint64_t gva;
-};
-
-struct mac_slot mac_slots[32];
-
-static int do_hvf_set_memory(hvf_slot *slot, hv_memory_flags_t flags)
-{
-    struct mac_slot *macslot;
-    hv_return_t ret;
-
-    macslot = &mac_slots[slot->slot_id];
-
-    if (macslot->present) {
-        if (macslot->size != slot->size) {
-            macslot->present = 0;
-            trace_hvf_vm_unmap(macslot->gpa_start, macslot->size);
-            ret = hv_vm_unmap(macslot->gpa_start, macslot->size);
-            assert_hvf_ok(ret);
-        }
-    }
-
-    if (!slot->size) {
-        return 0;
-    }
-
-    macslot->present = 1;
-    macslot->gpa_start = slot->start;
-    macslot->size = slot->size;
-    trace_hvf_vm_map(slot->start, slot->size, slot->mem, flags,
-                     flags & HV_MEMORY_READ ?  'R' : '-',
-                     flags & HV_MEMORY_WRITE ? 'W' : '-',
-                     flags & HV_MEMORY_EXEC ?  'E' : '-');
-    ret = hv_vm_map(slot->mem, slot->start, slot->size, flags);
-    assert_hvf_ok(ret);
-    return 0;
-}
-
-static void hvf_set_phys_mem(MemoryRegionSection *section, bool add)
-{
-    hvf_slot *mem;
-    MemoryRegion *area = section->mr;
-    bool writable = !area->readonly && !area->rom_device;
-    hv_memory_flags_t flags;
-    uint64_t page_size = qemu_real_host_page_size();
-
-    if (!memory_region_is_ram(area)) {
-        if (writable) {
-            return;
-        } else if (!memory_region_is_romd(area)) {
-            /*
-             * If the memory device is not in romd_mode, then we actually want
-             * to remove the hvf memory slot so all accesses will trap.
-             */
-             add = false;
-        }
-    }
-
-    if (!QEMU_IS_ALIGNED(int128_get64(section->size), page_size) ||
-        !QEMU_IS_ALIGNED(section->offset_within_address_space, page_size)) {
-        /* Not page aligned, so we can not map as RAM */
-        add = false;
-    }
-
-    mem = hvf_find_overlap_slot(
-            section->offset_within_address_space,
-            int128_get64(section->size));
-
-    if (mem && add) {
-        if (mem->size == int128_get64(section->size) &&
-            mem->start == section->offset_within_address_space &&
-            mem->mem == (memory_region_get_ram_ptr(area) +
-            section->offset_within_region)) {
-            return; /* Same region was attempted to register, go away. */
-        }
-    }
-
-    /* Region needs to be reset. set the size to 0 and remap it. */
-    if (mem) {
-        mem->size = 0;
-        if (do_hvf_set_memory(mem, 0)) {
-            error_report("Failed to reset overlapping slot");
-            abort();
-        }
-    }
-
-    if (!add) {
-        return;
-    }
-
-    if (area->readonly ||
-        (!memory_region_is_ram(area) && memory_region_is_romd(area))) {
-        flags = HV_MEMORY_READ | HV_MEMORY_EXEC;
-    } else {
-        flags = HV_MEMORY_READ | HV_MEMORY_WRITE | HV_MEMORY_EXEC;
-    }
-
-    /* Now make a new slot. */
-    int x;
-
-    for (x = 0; x < hvf_state->num_slots; ++x) {
-        mem = &hvf_state->slots[x];
-        if (!mem->size) {
-            break;
-        }
-    }
-
-    if (x == hvf_state->num_slots) {
-        error_report("No free slots");
-        abort();
-    }
-
-    mem->size = int128_get64(section->size);
-    mem->mem = memory_region_get_ram_ptr(area) + section->offset_within_region;
-    mem->start = section->offset_within_address_space;
-    mem->region = area;
-
-    if (do_hvf_set_memory(mem, flags)) {
-        error_report("Error registering new memory slot");
-        abort();
-    }
-}
-
 static void do_hvf_cpu_synchronize_state(CPUState *cpu, run_on_cpu_data arg)
 {
     if (!cpu->accel->dirty) {
@@ -244,148 +114,10 @@ static void hvf_cpu_synchronize_pre_loadvm(CPUState *cpu)
     run_on_cpu(cpu, do_hvf_cpu_synchronize_set_dirty, RUN_ON_CPU_NULL);
 }
 
-static void hvf_set_dirty_tracking(MemoryRegionSection *section, bool on)
-{
-    hvf_slot *slot;
-
-    slot = hvf_find_overlap_slot(
-            section->offset_within_address_space,
-            int128_get64(section->size));
-
-    /* protect region against writes; begin tracking it */
-    if (on) {
-        slot->flags |= HVF_SLOT_LOG;
-        hv_vm_protect((uintptr_t)slot->start, (size_t)slot->size,
-                      HV_MEMORY_READ | HV_MEMORY_EXEC);
-    /* stop tracking region*/
-    } else {
-        slot->flags &= ~HVF_SLOT_LOG;
-        hv_vm_protect((uintptr_t)slot->start, (size_t)slot->size,
-                      HV_MEMORY_READ | HV_MEMORY_WRITE | HV_MEMORY_EXEC);
-    }
-}
-
-static void hvf_log_start(MemoryListener *listener,
-                          MemoryRegionSection *section, int old, int new)
-{
-    if (old != 0) {
-        return;
-    }
-
-    hvf_set_dirty_tracking(section, 1);
-}
-
-static void hvf_log_stop(MemoryListener *listener,
-                         MemoryRegionSection *section, int old, int new)
-{
-    if (new != 0) {
-        return;
-    }
-
-    hvf_set_dirty_tracking(section, 0);
-}
-
-static void hvf_log_sync(MemoryListener *listener,
-                         MemoryRegionSection *section)
-{
-    /*
-     * sync of dirty pages is handled elsewhere; just make sure we keep
-     * tracking the region.
-     */
-    hvf_set_dirty_tracking(section, 1);
-}
-
-static void hvf_region_add(MemoryListener *listener,
-                           MemoryRegionSection *section)
-{
-    hvf_set_phys_mem(section, true);
-}
-
-static void hvf_region_del(MemoryListener *listener,
-                           MemoryRegionSection *section)
-{
-    hvf_set_phys_mem(section, false);
-}
-
-static MemoryListener hvf_memory_listener = {
-    .name = "hvf",
-    .priority = MEMORY_LISTENER_PRIORITY_ACCEL,
-    .region_add = hvf_region_add,
-    .region_del = hvf_region_del,
-    .log_start = hvf_log_start,
-    .log_stop = hvf_log_stop,
-    .log_sync = hvf_log_sync,
-};
-
 static void dummy_signal(int sig)
 {
 }
 
-bool hvf_allowed;
-
-static int hvf_accel_init(MachineState *ms)
-{
-    int x;
-    hv_return_t ret;
-    HVFState *s;
-    int pa_range = 36;
-    MachineClass *mc = MACHINE_GET_CLASS(ms);
-
-    if (mc->hvf_get_physical_address_range) {
-        pa_range = mc->hvf_get_physical_address_range(ms);
-        if (pa_range < 0) {
-            return -EINVAL;
-        }
-    }
-
-    ret = hvf_arch_vm_create(ms, (uint32_t)pa_range);
-    assert_hvf_ok(ret);
-
-    s = g_new0(HVFState, 1);
-
-    s->num_slots = ARRAY_SIZE(s->slots);
-    for (x = 0; x < s->num_slots; ++x) {
-        s->slots[x].size = 0;
-        s->slots[x].slot_id = x;
-    }
-
-    QTAILQ_INIT(&s->hvf_sw_breakpoints);
-
-    hvf_state = s;
-    memory_listener_register(&hvf_memory_listener, &address_space_memory);
-
-    return hvf_arch_init();
-}
-
-static inline int hvf_gdbstub_sstep_flags(void)
-{
-    return SSTEP_ENABLE | SSTEP_NOIRQ;
-}
-
-static void hvf_accel_class_init(ObjectClass *oc, const void *data)
-{
-    AccelClass *ac = ACCEL_CLASS(oc);
-    ac->name = "HVF";
-    ac->init_machine = hvf_accel_init;
-    ac->allowed = &hvf_allowed;
-    ac->supports_guest_debug = hvf_arch_supports_guest_debug;
-    ac->gdbstub_supported_sstep_flags = hvf_gdbstub_sstep_flags;
-}
-
-static const TypeInfo hvf_accel_type = {
-    .name = TYPE_HVF_ACCEL,
-    .parent = TYPE_ACCEL,
-    .instance_size = sizeof(HVFState),
-    .class_init = hvf_accel_class_init,
-};
-
-static void hvf_type_init(void)
-{
-    type_register_static(&hvf_accel_type);
-}
-
-type_init(hvf_type_init);
-
 static void hvf_vcpu_destroy(CPUState *cpu)
 {
     hv_return_t ret = hv_vcpu_destroy(cpu->accel->fd);
@@ -639,8 +371,10 @@ static const TypeInfo hvf_accel_ops_type = {
     .class_init = hvf_accel_ops_class_init,
     .abstract = true,
 };
+
 static void hvf_accel_ops_register_types(void)
 {
     type_register_static(&hvf_accel_ops_type);
 }
+
 type_init(hvf_accel_ops_register_types);
accel/hvf/hvf-all.c +268 -0

280 lines not on this page · raw message

Signature · 1 line
2.49.0
Reviewed-by on v6 16/39
#52 lore raw

Re: [PATCH v6 16/39] accel/hvf: Move generic method declarations to hvf-all.c

Quoted 9 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:22 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 16/39] accel/hvf: Move generic method declarations to
hvf-all.c
X-Mailer: git-send-email 2.49.0

hvf-all.c aims to contain the generic accel methods (TYPE_ACCEL), while hvf-accel-ops.c the per-vcpu methods (TYPE_ACCEL_OPS).

Quoted 6 lines
---
 accel/hvf/hvf-accel-ops.c | 274 +-------------------------------------
 accel/hvf/hvf-all.c       | 268 +++++++++++++++++++++++++++++++++++++
 2 files changed, 272 insertions(+), 270 deletions(-)
#53 lore raw

PATCH v6 17/39 cpus: Document CPUState::vcpu_dirty field

In reply to Philippe Mathieu-Daudé
---
 include/hw/core/cpu.h | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
include/hw/core/cpu.h +2 -1
index 162a56a5daa..5eaf41a566f 100644
@@ -442,6 +442,7 @@ struct qemu_work_item;
  * @opaque: User data.
  * @mem_io_pc: Host Program Counter at which the memory was accessed.
  * @accel: Pointer to accelerator specific state.
+ * @vcpu_dirty: Hardware accelerator is not synchronized with QEMU state
  * @kvm_fd: vCPU file descriptor for KVM.
  * @work_mutex: Lock to prevent multiple access to @work_list.
  * @work_list: List of pending asynchronous work.
@@ -538,7 +539,6 @@ struct CPUState {
     uint32_t kvm_fetch_index;
     uint64_t dirty_pages;
     int kvm_vcpu_stats_fd;
-    bool vcpu_dirty;
 
     /* Use by accel-block: CPU is executing an ioctl() */
     QemuLockCnt in_ioctl_lock;
@@ -554,6 +554,7 @@ struct CPUState {
     uint32_t halted;
     int32_t exception_index;
 
+    bool vcpu_dirty;
     AccelCPUState *accel;
 
     /* Used to keep track of an outstanding cpu throttle thread for migration
Signature · 1 line
2.49.0
Reviewed-by on v6 17/39
#54 lore raw

Re: [PATCH v6 17/39] cpus: Document CPUState::vcpu_dirty field

Quoted 3 lines · Philippe Mathieu-Daudé
Quoted 32 lines
---
  include/hw/core/cpu.h | 3 ++-
  1 file changed, 2 insertions(+), 1 deletion(-)
include/hw/core/cpu.h +2 -1
index 162a56a5daa..5eaf41a566f 100644
@@ -442,6 +442,7 @@ struct qemu_work_item;
   * @opaque: User data.
   * @mem_io_pc: Host Program Counter at which the memory was accessed.
   * @accel: Pointer to accelerator specific state.
+ * @vcpu_dirty: Hardware accelerator is not synchronized with QEMU state
   * @kvm_fd: vCPU file descriptor for KVM.
   * @work_mutex: Lock to prevent multiple access to @work_list.
   * @work_list: List of pending asynchronous work.
@@ -538,7 +539,6 @@ struct CPUState {
      uint32_t kvm_fetch_index;
      uint64_t dirty_pages;
      int kvm_vcpu_stats_fd;
-    bool vcpu_dirty;
 
      /* Use by accel-block: CPU is executing an ioctl() */
      QemuLockCnt in_ioctl_lock;
@@ -554,6 +554,7 @@ struct CPUState {
      uint32_t halted;
      int32_t exception_index;
 
+    bool vcpu_dirty;
      AccelCPUState *accel;
 
      /* Used to keep track of an outstanding cpu throttle thread for migration
#55 lore raw

PATCH v6 18/39 accel/hvf: Replace @dirty field by generic CPUState::vcpu_dirty field

In reply to Philippe Mathieu-Daudé

No need for accel-specific @dirty field when we have a generic one in CPUState.

---
 include/system/hvf_int.h  |  1 -
 accel/hvf/hvf-accel-ops.c | 10 +++++-----
 target/arm/hvf/hvf.c      |  4 ++--
 target/i386/hvf/hvf.c     |  4 ++--
 target/i386/hvf/x86hvf.c  |  2 +-
 5 files changed, 10 insertions(+), 11 deletions(-)
include/system/hvf_int.h +0 -1
index dcf06f53d97..ecc49a309cf 100644
@@ -62,7 +62,6 @@ struct AccelCPUState {
     bool vtimer_masked;
     sigset_t unblock_ipi_mask;
     bool guest_debug_enabled;
-    bool dirty;
 };
 
 void assert_hvf_ok_impl(hv_return_t ret, const char *file, unsigned int line,
accel/hvf/hvf-accel-ops.c +5 -5
index 6b1a256deb5..2944e350ca9 100644
@@ -79,15 +79,15 @@ hvf_slot *hvf_find_overlap_slot(uint64_t start, uint64_t size)
 
 static void do_hvf_cpu_synchronize_state(CPUState *cpu, run_on_cpu_data arg)
 {
-    if (!cpu->accel->dirty) {
+    if (!cpu->vcpu_dirty) {
         hvf_get_registers(cpu);
-        cpu->accel->dirty = true;
+        cpu->vcpu_dirty = true;
     }
 }
 
 static void hvf_cpu_synchronize_state(CPUState *cpu)
 {
-    if (!cpu->accel->dirty) {
+    if (!cpu->vcpu_dirty) {
         run_on_cpu(cpu, do_hvf_cpu_synchronize_state, RUN_ON_CPU_NULL);
     }
 }
@@ -96,7 +96,7 @@ static void do_hvf_cpu_synchronize_set_dirty(CPUState *cpu,
                                              run_on_cpu_data arg)
 {
     /* QEMU state is the reference, push it to HVF now and on next entry */
-    cpu->accel->dirty = true;
+    cpu->vcpu_dirty = true;
 }
 
 static void hvf_cpu_synchronize_post_reset(CPUState *cpu)
@@ -150,8 +150,8 @@ static int hvf_init_vcpu(CPUState *cpu)
 #else
     r = hv_vcpu_create(&cpu->accel->fd, HV_VCPU_DEFAULT);
 #endif
-    cpu->accel->dirty = true;
     assert_hvf_ok(r);
+    cpu->vcpu_dirty = true;
 
     cpu->accel->guest_debug_enabled = false;
 
target/arm/hvf/hvf.c +2 -2
index 995a6a74b06..ac0d6ef8187 100644
@@ -813,9 +813,9 @@ int hvf_put_registers(CPUState *cpu)
 
 static void flush_cpu_state(CPUState *cpu)
 {
-    if (cpu->accel->dirty) {
+    if (cpu->vcpu_dirty) {
         hvf_put_registers(cpu);
-        cpu->accel->dirty = false;
+        cpu->vcpu_dirty = false;
     }
 }
 
target/i386/hvf/hvf.c +2 -2
index b25bff0df41..9b8acd5a946 100644
@@ -733,9 +733,9 @@ int hvf_vcpu_exec(CPUState *cpu)
     }
 
     do {
-        if (cpu->accel->dirty) {
+        if (cpu->vcpu_dirty) {
             hvf_put_registers(cpu);
-            cpu->accel->dirty = false;
+            cpu->vcpu_dirty = false;
         }
 
         if (hvf_inject_interrupts(cpu)) {
target/i386/hvf/x86hvf.c +1 -1
index 2057314892a..17fce1d3cdd 100644
@@ -427,7 +427,7 @@ int hvf_process_events(CPUState *cs)
     X86CPU *cpu = X86_CPU(cs);
     CPUX86State *env = &cpu->env;
 
-    if (!cs->accel->dirty) {
+    if (!cs->vcpu_dirty) {
         /* light weight sync for CPU_INTERRUPT_HARD and IF_MASK */
         env->eflags = rreg(cs->accel->fd, HV_X86_RFLAGS);
     }
Signature · 1 line
2.49.0
Reviewed-by on v6 18/39
#57 lore raw

Re: [PATCH v6 18/39] accel/hvf: Replace @dirty field by generic CPUState::vcpu_dirty field

In reply to Philippe Mathieu-Daudé
Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:24 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 18/39] accel/hvf: Replace @dirty field by generic
CPUState::vcpu_dirty field
X-Mailer: git-send-email 2.49.0

No need for accel-specific @dirty field when we have a generic one in CPUState.

Quoted 7 lines
---
 include/system/hvf_int.h  |  1 -
 accel/hvf/hvf-accel-ops.c | 10 +++++-----
 target/arm/hvf/hvf.c      |  4 ++--
 target/i386/hvf/hvf.c     |  4 ++--
 target/i386/hvf/x86hvf.c  |  2 +-
 5 files changed, 10 insertions(+), 11 deletions(-)
#58 lore raw

PATCH v6 19/39 accel/nvmm: Replace @dirty field by generic CPUState::vcpu_dirty field

In reply to Philippe Mathieu-Daudé

No need for accel-specific @dirty field when we have a generic one in CPUState.

---
 target/i386/nvmm/nvmm-all.c | 21 ++++++++++-----------
 1 file changed, 10 insertions(+), 11 deletions(-)
target/i386/nvmm/nvmm-all.c +10 -11
index f1c6120ccf1..aea61a6fd2a 100644
@@ -30,7 +30,6 @@ struct AccelCPUState {
     struct nvmm_vcpu vcpu;
     uint8_t tpr;
     bool stop;
-    bool dirty;
 
     /* Window-exiting for INTs/NMIs. */
     bool int_window_exit;
@@ -508,7 +507,7 @@ nvmm_io_callback(struct nvmm_io *io)
     }
 
     /* Needed, otherwise infinite loop. */
-    current_cpu->accel->dirty = false;
+    current_cpu->vcpu_dirty = false;
 }
 
 static void
@@ -517,7 +516,7 @@ nvmm_mem_callback(struct nvmm_mem *mem)
     cpu_physical_memory_rw(mem->gpa, mem->data, mem->size, mem->write);
 
     /* Needed, otherwise infinite loop. */
-    current_cpu->accel->dirty = false;
+    current_cpu->vcpu_dirty = false;
 }
 
 static struct nvmm_assist_callbacks nvmm_callbacks = {
@@ -727,9 +726,9 @@ nvmm_vcpu_loop(CPUState *cpu)
      * Inner VCPU loop.
      */
     do {
-        if (cpu->accel->dirty) {
+        if (cpu->vcpu_dirty) {
             nvmm_set_registers(cpu);
-            cpu->accel->dirty = false;
+            cpu->vcpu_dirty = false;
         }
 
         if (qcpu->stop) {
@@ -827,32 +826,32 @@ static void
 do_nvmm_cpu_synchronize_state(CPUState *cpu, run_on_cpu_data arg)
 {
     nvmm_get_registers(cpu);
-    cpu->accel->dirty = true;
+    cpu->vcpu_dirty = true;
 }
 
 static void
 do_nvmm_cpu_synchronize_post_reset(CPUState *cpu, run_on_cpu_data arg)
 {
     nvmm_set_registers(cpu);
-    cpu->accel->dirty = false;
+    cpu->vcpu_dirty = false;
 }
 
 static void
 do_nvmm_cpu_synchronize_post_init(CPUState *cpu, run_on_cpu_data arg)
 {
     nvmm_set_registers(cpu);
-    cpu->accel->dirty = false;
+    cpu->vcpu_dirty = false;
 }
 
 static void
 do_nvmm_cpu_synchronize_pre_loadvm(CPUState *cpu, run_on_cpu_data arg)
 {
-    cpu->accel->dirty = true;
+    cpu->vcpu_dirty = true;
 }
 
 void nvmm_cpu_synchronize_state(CPUState *cpu)
 {
-    if (!cpu->accel->dirty) {
+    if (!cpu->vcpu_dirty) {
         run_on_cpu(cpu, do_nvmm_cpu_synchronize_state, RUN_ON_CPU_NULL);
     }
 }
@@ -982,7 +981,7 @@ nvmm_init_vcpu(CPUState *cpu)
         }
     }
 
-    qcpu->dirty = true;
+    qcpu->vcpu_dirty = true;
     cpu->accel = qcpu;
 
     return 0;
Signature · 1 line
2.49.0
Reviewed-by on v6 19/39
#60 lore raw

Re: [PATCH v6 19/39] accel/nvmm: Replace @dirty field by generic CPUState::vcpu_dirty field

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:25 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 19/39] accel/nvmm: Replace @dirty field by generic
CPUState::vcpu_dirty field
X-Mailer: git-send-email 2.49.0

Quoted 8 lines

No need for accel-specific @dirty field when we have a generic one in CPUState.

---
 target/i386/nvmm/nvmm-all.c | 21 ++++++++++-----------
 1 file changed, 10 insertions(+), 11 deletions(-)
#61 lore raw

PATCH v6 20/39 accel/whpx: Replace @dirty field by generic CPUState::vcpu_dirty field

In reply to Philippe Mathieu-Daudé

No need for accel-specific @dirty field when we have a generic one in CPUState.

---
 target/i386/whpx/whpx-all.c | 23 +++++++++++------------
 1 file changed, 11 insertions(+), 12 deletions(-)
target/i386/whpx/whpx-all.c +11 -12
index cf6d3e4cdd4..5001afad3a7 100644
@@ -237,7 +237,6 @@ struct AccelCPUState {
     uint64_t tpr;
     uint64_t apic_base;
     bool interruption_pending;
-    bool dirty;
 
     /* Must be the last field as it may have a tail */
     WHV_RUN_VP_EXIT_CONTEXT exit_ctx;
@@ -836,7 +835,7 @@ static HRESULT CALLBACK whpx_emu_setreg_callback(
      * The emulator just successfully wrote the register state. We clear the
      * dirty state so we avoid the double write on resume of the VP.
      */
-    cpu->accel->dirty = false;
+    cpu->vcpu_dirty = false;
 
     return hr;
 }
@@ -1391,7 +1390,7 @@ static int whpx_last_vcpu_stopping(CPUState *cpu)
 /* Returns the address of the next instruction that is about to be executed. */
 static vaddr whpx_vcpu_get_pc(CPUState *cpu, bool exit_context_valid)
 {
-    if (cpu->accel->dirty) {
+    if (cpu->vcpu_dirty) {
         /* The CPU registers have been modified by other parts of QEMU. */
         return cpu_env(cpu)->eip;
     } else if (exit_context_valid) {
@@ -1704,9 +1703,9 @@ static int whpx_vcpu_run(CPUState *cpu)
     }
 
     do {
-        if (cpu->accel->dirty) {
+        if (cpu->vcpu_dirty) {
             whpx_set_registers(cpu, WHPX_SET_RUNTIME_STATE);
-            cpu->accel->dirty = false;
+            cpu->vcpu_dirty = false;
         }
 
         if (exclusive_step_mode == WHPX_STEP_NONE) {
@@ -2054,9 +2053,9 @@ static int whpx_vcpu_run(CPUState *cpu)
 
 static void do_whpx_cpu_synchronize_state(CPUState *cpu, run_on_cpu_data arg)
 {
-    if (!cpu->accel->dirty) {
+    if (!cpu->vcpu_dirty) {
         whpx_get_registers(cpu);
-        cpu->accel->dirty = true;
+        cpu->vcpu_dirty = true;
     }
 }
 
@@ -2064,20 +2063,20 @@ static void do_whpx_cpu_synchronize_post_reset(CPUState *cpu,
                                                run_on_cpu_data arg)
 {
     whpx_set_registers(cpu, WHPX_SET_RESET_STATE);
-    cpu->accel->dirty = false;
+    cpu->vcpu_dirty = false;
 }
 
 static void do_whpx_cpu_synchronize_post_init(CPUState *cpu,
                                               run_on_cpu_data arg)
 {
     whpx_set_registers(cpu, WHPX_SET_FULL_STATE);
-    cpu->accel->dirty = false;
+    cpu->vcpu_dirty = false;
 }
 
 static void do_whpx_cpu_synchronize_pre_loadvm(CPUState *cpu,
                                                run_on_cpu_data arg)
 {
-    cpu->accel->dirty = true;
+    cpu->vcpu_dirty = true;
 }
 
 /*
@@ -2086,7 +2085,7 @@ static void do_whpx_cpu_synchronize_pre_loadvm(CPUState *cpu,
 
 void whpx_cpu_synchronize_state(CPUState *cpu)
 {
-    if (!cpu->accel->dirty) {
+    if (!cpu->vcpu_dirty) {
         run_on_cpu(cpu, do_whpx_cpu_synchronize_state, RUN_ON_CPU_NULL);
     }
 }
@@ -2226,7 +2225,7 @@ int whpx_init_vcpu(CPUState *cpu)
     }
 
     vcpu->interruptable = true;
-    vcpu->dirty = true;
+    cpu->vcpu_dirty = true;
     cpu->accel = vcpu;
     max_vcpu_index = max(max_vcpu_index, cpu->cpu_index);
     qemu_add_vm_change_state_handler(whpx_cpu_update_state, env);
Signature · 1 line
2.49.0
Reviewed-by on v6 20/39
#63 lore raw

Re: [PATCH v6 20/39] accel/whpx: Replace @dirty field by generic CPUState::vcpu_dirty field

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:26 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 20/39] accel/whpx: Replace @dirty field by generic
CPUState::vcpu_dirty field
X-Mailer: git-send-email 2.49.0

Quoted 8 lines

No need for accel-specific @dirty field when we have a generic one in CPUState.

---
 target/i386/whpx/whpx-all.c | 23 +++++++++++------------
 1 file changed, 11 insertions(+), 12 deletions(-)
#64 lore raw

PATCH v6 21/39 accel/kvm: Remove kvm_cpu_synchronize_state() stub

In reply to Philippe Mathieu-Daudé

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_cpu_synchronize_state() stub is not necessary.

Fixes e0715f6abce ("kvm: remove kvm specific functions from global includes")
---
 accel/stubs/kvm-stub.c | 4 ----
 1 file changed, 4 deletions(-)
accel/stubs/kvm-stub.c +0 -4
index b9b4427c919..68cd33ba973 100644
@@ -29,10 +29,6 @@ void kvm_flush_coalesced_mmio_buffer(void)
 {
 }
 
-void kvm_cpu_synchronize_state(CPUState *cpu)
-{
-}
-
 bool kvm_has_sync_mmu(void)
 {
     return false;
Signature · 1 line
2.49.0
Reviewed-by on v6 21/39
#65 lore raw

Re: [PATCH v6 21/39] accel/kvm: Remove kvm_cpu_synchronize_state() stub

Quoted 7 lines · Philippe Mathieu-Daudé

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_cpu_synchronize_state() stub is not necessary.

Fixes e0715f6abce ("kvm: remove kvm specific functions from global includes")

BTW, as what you do for HVF in this series that moving vcpu methods from hvf-all.c to hvf-accel-ops.c, do you plan to move kvm_cpu_synchronize_state() from kvm-all.c to kvm-accel-ops.c ?

Quoted 19 lines
---
  accel/stubs/kvm-stub.c | 4 ----
  1 file changed, 4 deletions(-)
accel/stubs/kvm-stub.c +0 -4
index b9b4427c919..68cd33ba973 100644
@@ -29,10 +29,6 @@ void kvm_flush_coalesced_mmio_buffer(void)
  {
  }
 
-void kvm_cpu_synchronize_state(CPUState *cpu)
-{
-}
-
  bool kvm_has_sync_mmu(void)
  {
      return false;
#66 lore raw

Re: [PATCH v6 21/39] accel/kvm: Remove kvm_cpu_synchronize_state() stub

Quoted 10 lines · Xiaoyao Li
Quoted 7 lines · Philippe Mathieu-Daudé

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_cpu_synchronize_state() stub is not necessary.

Fixes e0715f6abce ("kvm: remove kvm specific functions from global includes")

Thanks!

Quoted 2 lines

BTW, as what you do for HVF in this series that moving vcpu methods from hvf-all.c to hvf-accel-ops.c, do you plan to move kvm_cpu_synchronize_state() from kvm-all.c to kvm-accel-ops.c ?

While it doesn't seem obvious, I'm trying to consolidate the minimum in order to have split-acceleration. The proof of concept is with TCG and HVF, so I'll not do non-necessary changes in KVM.

Regards,

Phil.

Reviewed-by on v6 21/39
#67 lore raw

Re: [PATCH v6 21/39] accel/kvm: Remove kvm_cpu_synchronize_state() stub

In reply to Philippe Mathieu-Daudé
Quoted 9 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:27 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 21/39] accel/kvm: Remove kvm_cpu_synchronize_state() stub
X-Mailer: git-send-email 2.49.0

Since commit 57038a92bb0 ("cpus: extract out kvm-specific code to accel/kvm") the kvm_cpu_synchronize_state() stub is not necessary.

Quoted 6 lines
Fixes e0715f6abce ("kvm: remove kvm specific functions from global includes")
---
 accel/stubs/kvm-stub.c | 4 ----
 1 file changed, 4 deletions(-)
#68 lore raw

PATCH v6 22/39 accel/system: Document cpu_synchronize_state()

In reply to Philippe Mathieu-Daudé
---
 include/system/accel-ops.h |  8 ++++++++
 include/system/hw_accel.h  | 13 +++++++++++--
 2 files changed, 19 insertions(+), 2 deletions(-)
include/system/accel-ops.h +8 -0
index f19245d0a0e..6eed1a3cfc8 100644
@@ -41,6 +41,14 @@ struct AccelOpsClass {
 
     void (*synchronize_post_reset)(CPUState *cpu);
     void (*synchronize_post_init)(CPUState *cpu);
+    /**
+     * synchronize_state:
+     * synchronize_pre_loadvm:
+     * @cpu: The vCPU to synchronize.
+     *
+     * Request to synchronize QEMU vCPU registers from the hardware accelerator
+     * (the hardware accelerator is the reference).
+     */
     void (*synchronize_state)(CPUState *cpu);
     void (*synchronize_pre_loadvm)(CPUState *cpu);
     void (*synchronize_pre_resume)(bool step_pending);
include/system/hw_accel.h +11 -2
index 380e9e640b6..574c9738408 100644
@@ -17,9 +17,18 @@
 #include "system/whpx.h"
 #include "system/nvmm.h"
 
+/**
+ * cpu_synchronize_state:
+ * cpu_synchronize_pre_loadvm:
+ * @cpu: The vCPU to synchronize.
+ *
+ * Request to synchronize QEMU vCPU registers from the hardware accelerator
+ * (the hardware accelerator is the reference).
+ */
 void cpu_synchronize_state(CPUState *cpu);
-void cpu_synchronize_post_reset(CPUState *cpu);
-void cpu_synchronize_post_init(CPUState *cpu);
 void cpu_synchronize_pre_loadvm(CPUState *cpu);
 
+void cpu_synchronize_post_reset(CPUState *cpu);
+void cpu_synchronize_post_init(CPUState *cpu);
+
 #endif /* QEMU_HW_ACCEL_H */
Signature · 1 line
2.49.0
Reviewed-by on v6 22/39
#69 lore raw

Re: [PATCH v6 22/39] accel/system: Document cpu_synchronize_state()

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:28 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 22/39] accel/system: Document cpu_synchronize_state()
X-Mailer: git-send-email 2.49.0

---
 include/system/accel-ops.h |  8 ++++++++
 include/system/hw_accel.h  | 13 +++++++++++--
 2 files changed, 19 insertions(+), 2 deletions(-)> 
#70 lore raw

PATCH v6 23/39 accel/system: Document cpu_synchronize_state_post_init/reset()

In reply to Philippe Mathieu-Daudé
---
 include/system/accel-ops.h | 8 ++++++++
 include/system/hw_accel.h  | 8 ++++++++
 2 files changed, 16 insertions(+)
include/system/accel-ops.h +8 -0
index 6eed1a3cfc8..2075691331c 100644
@@ -39,6 +39,14 @@ struct AccelOpsClass {
     void (*kick_vcpu_thread)(CPUState *cpu);
     bool (*cpu_thread_is_idle)(CPUState *cpu);
 
+    /**
+     * synchronize_post_reset:
+     * synchronize_post_init:
+     * @cpu: The vCPU to synchronize.
+     *
+     * Request to synchronize QEMU vCPU registers to the hardware accelerator
+     * (QEMU is the reference).
+     */
     void (*synchronize_post_reset)(CPUState *cpu);
     void (*synchronize_post_init)(CPUState *cpu);
     /**
include/system/hw_accel.h +8 -0
index 574c9738408..fa9228d5d2d 100644
@@ -28,6 +28,14 @@
 void cpu_synchronize_state(CPUState *cpu);
 void cpu_synchronize_pre_loadvm(CPUState *cpu);
 
+/**
+ * cpu_synchronize_post_reset:
+ * cpu_synchronize_post_init:
+ * @cpu: The vCPU to synchronize.
+ *
+ * Request to synchronize QEMU vCPU registers to the hardware accelerator
+ * (QEMU is the reference).
+ */
 void cpu_synchronize_post_reset(CPUState *cpu);
 void cpu_synchronize_post_init(CPUState *cpu);
 
Signature · 1 line
2.49.0
Reviewed-by on v6 23/39
#71 lore raw

Re: [PATCH v6 23/39] accel/system: Document cpu_synchronize_state_post_init/reset()

Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:29 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 23/39] accel/system: Document
cpu_synchronize_state_post_init/reset()
X-Mailer: git-send-email 2.49.0

Quoted 6 lines
---
 include/system/accel-ops.h | 8 ++++++++
 include/system/hw_accel.h  | 8 ++++++++
 2 files changed, 16 insertions(+)
#72 lore raw

PATCH v6 24/39 accel/nvmm: Expose nvmm_enabled() to common code

In reply to Philippe Mathieu-Daudé

Currently nvmm_enabled() is restricted to target-specific code. By defining CONFIG_NVMM_IS_POSSIBLE we allow its use anywhere.

---
 include/system/nvmm.h       | 23 ++++++++++++-----------
 accel/stubs/nvmm-stub.c     | 12 ++++++++++++
 target/i386/nvmm/nvmm-all.c |  8 +-------
 accel/stubs/meson.build     |  1 +
 4 files changed, 26 insertions(+), 18 deletions(-)
 create mode 100644 accel/stubs/nvmm-stub.c
include/system/nvmm.h +12 -11
index 6971ddb3a5a..7390def9adb 100644
@@ -13,17 +13,18 @@
 #define QEMU_NVMM_H
 
 #ifdef COMPILING_PER_TARGET
-
-#ifdef CONFIG_NVMM
-
-int nvmm_enabled(void);
-
-#else /* CONFIG_NVMM */
-
-#define nvmm_enabled() (0)
-
-#endif /* CONFIG_NVMM */
-
+# ifdef CONFIG_NVMM
+#  define CONFIG_NVMM_IS_POSSIBLE
+# endif /* !CONFIG_NVMM */
+#else
+# define CONFIG_NVMM_IS_POSSIBLE
 #endif /* COMPILING_PER_TARGET */
 
+#ifdef CONFIG_NVMM_IS_POSSIBLE
+extern bool nvmm_allowed;
+#define nvmm_enabled() (nvmm_allowed)
+#else /* !CONFIG_NVMM_IS_POSSIBLE */
+#define nvmm_enabled() 0
+#endif /* !CONFIG_NVMM_IS_POSSIBLE */
+
 #endif /* QEMU_NVMM_H */
accel/stubs/nvmm-stub.c new file +12 -0
new file mode 100644
index 00000000000..cc58114ceb3
@@ -0,0 +1,12 @@
+/*
+ * NVMM stubs for QEMU
+ *
+ *  Copyright (c) Linaro
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#include "qemu/osdep.h"
+#include "system/hvf.h"
+
+bool nvmm_allowed;
target/i386/nvmm/nvmm-all.c +1 -7
index aea61a6fd2a..2df49d7eeb4 100644
@@ -46,7 +46,7 @@ struct qemu_machine {
 
 /* -------------------------------------------------------------------------- */
 
-static bool nvmm_allowed;
+bool nvmm_allowed;
 static struct qemu_machine qemu_mach;
 
 static struct nvmm_machine *
@@ -1192,12 +1192,6 @@ nvmm_accel_init(MachineState *ms)
     return 0;
 }
 
-int
-nvmm_enabled(void)
-{
-    return nvmm_allowed;
-}
-
 static void
 nvmm_accel_class_init(ObjectClass *oc, const void *data)
 {
accel/stubs/meson.build +1 -0
index 8ca1a4529e2..4c34287215f 100644
@@ -3,5 +3,6 @@ system_stubs_ss.add(when: 'CONFIG_XEN', if_false: files('xen-stub.c'))
 system_stubs_ss.add(when: 'CONFIG_KVM', if_false: files('kvm-stub.c'))
 system_stubs_ss.add(when: 'CONFIG_TCG', if_false: files('tcg-stub.c'))
 system_stubs_ss.add(when: 'CONFIG_HVF', if_false: files('hvf-stub.c'))
+system_stubs_ss.add(when: 'CONFIG_NVMM', if_false: files('nvmm-stub.c'))
 
 specific_ss.add_all(when: ['CONFIG_SYSTEM_ONLY'], if_true: system_stubs_ss)
Signature · 1 line
2.49.0
#73 lore raw

Re: [PATCH v6 24/39] accel/nvmm: Expose nvmm_enabled() to common code

Quoted 20 lines · Philippe Mathieu-Daudé

Currently nvmm_enabled() is restricted to target-specific code. By defining CONFIG_NVMM_IS_POSSIBLE we allow its use anywhere.

---
  include/system/nvmm.h       | 23 ++++++++++++-----------
  accel/stubs/nvmm-stub.c     | 12 ++++++++++++
  target/i386/nvmm/nvmm-all.c |  8 +-------
  accel/stubs/meson.build     |  1 +
  4 files changed, 26 insertions(+), 18 deletions(-)

create mode 100644 accel/stubs/nvmm-stub.c

accel/stubs/nvmm-stub.c new file +0 -0
new file mode 100644
index 00000000000..cc58114ceb3

accel/stubs/nvmm-stub.c line 10

Quoted 10 lines
accel/stubs/nvmm-stub.c new file +10 -0
@@ -0,0 +1,12 @@
+/*
+ * NVMM stubs for QEMU
+ *
+ *  Copyright (c) Linaro
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#include "qemu/osdep.h"
+#include "system/hvf.h"

"system/nvmm.h" ;)

Reviewed-by on v6 24/39
#74 lore raw

Re: [PATCH v6 24/39] accel/nvmm: Expose nvmm_enabled() to common code

In reply to Philippe Mathieu-Daudé
Quoted 10 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:30 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 24/39] accel/nvmm: Expose nvmm_enabled() to common code
X-Mailer: git-send-email 2.49.0

Currently nvmm_enabled() is restricted to target-specific code. By defining CONFIG_NVMM_IS_POSSIBLE we allow its use anywhere.

Quoted 7 lines
---
 include/system/nvmm.h       | 23 ++++++++++++-----------
 accel/stubs/nvmm-stub.c     | 12 ++++++++++++
 target/i386/nvmm/nvmm-all.c |  8 +-------
 accel/stubs/meson.build     |  1 +
 4 files changed, 26 insertions(+), 18 deletions(-)
 create mode 100644 accel/stubs/nvmm-stub.c

With header file fixed,

#75 lore raw

PATCH v6 25/39 accel/whpx: Expose whpx_enabled() to common code

In reply to Philippe Mathieu-Daudé

Currently whpx_enabled() is restricted to target-specific code. By defining CONFIG_WHPX_IS_POSSIBLE we allow its use anywhere.

---
 include/system/whpx.h       | 27 ++++++++++++++-------------
 accel/stubs/whpx-stub.c     | 12 ++++++++++++
 target/i386/whpx/whpx-all.c |  7 +------
 accel/stubs/meson.build     |  1 +
 4 files changed, 28 insertions(+), 19 deletions(-)
 create mode 100644 accel/stubs/whpx-stub.c
include/system/whpx.h +14 -13
index 00ff409b682..00f6a3e5236 100644
@@ -16,19 +16,20 @@
 #define QEMU_WHPX_H
 
 #ifdef COMPILING_PER_TARGET
-
-#ifdef CONFIG_WHPX
-
-int whpx_enabled(void);
-bool whpx_apic_in_platform(void);
-
-#else /* CONFIG_WHPX */
-
-#define whpx_enabled() (0)
-#define whpx_apic_in_platform() (0)
-
-#endif /* CONFIG_WHPX */
-
+# ifdef CONFIG_WHPX
+#  define CONFIG_WHPX_IS_POSSIBLE
+# endif /* !CONFIG_WHPX */
+#else
+# define CONFIG_WHPX_IS_POSSIBLE
 #endif /* COMPILING_PER_TARGET */
 
+#ifdef CONFIG_WHPX_IS_POSSIBLE
+extern bool whpx_allowed;
+#define whpx_enabled() (whpx_allowed)
+bool whpx_apic_in_platform(void);
+#else /* !CONFIG_WHPX_IS_POSSIBLE */
+#define whpx_enabled() 0
+#define whpx_apic_in_platform() (0)
+#endif /* !CONFIG_WHPX_IS_POSSIBLE */
+
 #endif /* QEMU_WHPX_H */
accel/stubs/whpx-stub.c new file +12 -0
new file mode 100644
index 00000000000..c564c89fd0b
@@ -0,0 +1,12 @@
+/*
+ * WHPX stubs for QEMU
+ *
+ *  Copyright (c) Linaro
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#include "qemu/osdep.h"
+#include "system/whpx.h"
+
+bool whpx_allowed;
target/i386/whpx/whpx-all.c +1 -6
index 5001afad3a7..94fd5fc7849 100644
@@ -242,7 +242,7 @@ struct AccelCPUState {
     WHV_RUN_VP_EXIT_CONTEXT exit_ctx;
 };
 
-static bool whpx_allowed;
+bool whpx_allowed;
 static bool whp_dispatch_initialized;
 static HMODULE hWinHvPlatform, hWinHvEmulation;
 static uint32_t max_vcpu_index;
@@ -2688,11 +2688,6 @@ error:
     return ret;
 }
 
-int whpx_enabled(void)
-{
-    return whpx_allowed;
-}
-
 bool whpx_apic_in_platform(void) {
     return whpx_global.apic_in_platform;
 }
accel/stubs/meson.build +1 -0
index 4c34287215f..9dfc4f9ddaf 100644
@@ -4,5 +4,6 @@ system_stubs_ss.add(when: 'CONFIG_KVM', if_false: files('kvm-stub.c'))
 system_stubs_ss.add(when: 'CONFIG_TCG', if_false: files('tcg-stub.c'))
 system_stubs_ss.add(when: 'CONFIG_HVF', if_false: files('hvf-stub.c'))
 system_stubs_ss.add(when: 'CONFIG_NVMM', if_false: files('nvmm-stub.c'))
+system_stubs_ss.add(when: 'CONFIG_WHPX', if_false: files('whpx-stub.c'))
 
 specific_ss.add_all(when: ['CONFIG_SYSTEM_ONLY'], if_true: system_stubs_ss)
Signature · 1 line
2.49.0
Reviewed-by on v6 25/39
#76 lore raw

Re: [PATCH v6 25/39] accel/whpx: Expose whpx_enabled() to common code

Quoted 10 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:31 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 25/39] accel/whpx: Expose whpx_enabled() to common code
X-Mailer: git-send-email 2.49.0

Currently whpx_enabled() is restricted to target-specific code. By defining CONFIG_WHPX_IS_POSSIBLE we allow its use anywhere.

Quoted 7 lines
---
 include/system/whpx.h       | 27 ++++++++++++++-------------
 accel/stubs/whpx-stub.c     | 12 ++++++++++++
 target/i386/whpx/whpx-all.c |  7 +------
 accel/stubs/meson.build     |  1 +
 4 files changed, 28 insertions(+), 19 deletions(-)
 create mode 100644 accel/stubs/whpx-stub.c
#77 lore raw

PATCH v6 26/39 accel/dummy: Extract 'dummy-cpus.h' header from 'system/cpus.h'

In reply to Philippe Mathieu-Daudé

'dummy' helpers are specific to accelerator implementations, no need to expose them via "system/cpus.h".

---
 MAINTAINERS           |  1 +
 accel/dummy-cpus.h    | 14 ++++++++++++++
 include/system/cpus.h |  5 -----
 accel/dummy-cpus.c    |  1 +
 accel/qtest/qtest.c   |  1 +
 accel/xen/xen-all.c   |  1 +
 6 files changed, 18 insertions(+), 5 deletions(-)
 create mode 100644 accel/dummy-cpus.h
MAINTAINERS +1 -0
index b1cbfe115bc..37d02b2313c 100644
@@ -503,6 +503,7 @@ F: include/system/accel-*.h
 F: include/system/cpus.h
 F: include/accel/accel-cpu*.h
 F: accel/accel-*.?
+F: accel/dummy-cpus.?
 F: accel/Makefile.objs
 F: accel/stubs/Makefile.objs
 F: cpu-common.c
accel/dummy-cpus.h new file +14 -0
new file mode 100644
index 00000000000..d18dd0fdc51
@@ -0,0 +1,14 @@
+/*
+ * Dummy cpu thread code
+ *
+ * Copyright IBM, Corp. 2011
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+
+#ifndef ACCEL_DUMMY_CPUS_H
+#define ACCEL_DUMMY_CPUS_H
+
+void dummy_start_vcpu_thread(CPUState *cpu);
+
+#endif
include/system/cpus.h +0 -5
index 3226c765d01..69be6a77a75 100644
@@ -7,11 +7,6 @@ void cpus_register_accel(const AccelOpsClass *i);
 /* return registers ops */
 const AccelOpsClass *cpus_get_accel(void);
 
-/* accel/dummy-cpus.c */
-
-/* Create a dummy vcpu for AccelOpsClass->create_vcpu_thread */
-void dummy_start_vcpu_thread(CPUState *);
-
 /* interface available for cpus accelerator threads */
 
 /* For temporary buffers for forming a name */
accel/dummy-cpus.c +1 -0
index 867276144fa..03cfc0fa01e 100644
@@ -17,6 +17,7 @@
 #include "qemu/guest-random.h"
 #include "qemu/main-loop.h"
 #include "hw/core/cpu.h"
+#include "accel/dummy-cpus.h"
 
 static void *dummy_cpu_thread_fn(void *arg)
 {
accel/qtest/qtest.c +1 -0
index 92bed9264ce..612cede160b 100644
@@ -24,6 +24,7 @@
 #include "qemu/guest-random.h"
 #include "qemu/main-loop.h"
 #include "hw/core/cpu.h"
+#include "accel/dummy-cpus.h"
 
 static int64_t qtest_clock_counter;
 
accel/xen/xen-all.c +1 -0
index de52a8f882a..c150dd43cab 100644
@@ -18,6 +18,7 @@
 #include "hw/xen/xen_igd.h"
 #include "chardev/char.h"
 #include "qemu/accel.h"
+#include "accel/dummy-cpus.h"
 #include "system/accel-ops.h"
 #include "system/cpus.h"
 #include "system/xen.h"
Signature · 1 line
2.49.0
Reviewed-by on v6 26/39
#79 lore raw

Re: [PATCH v6 26/39] accel/dummy: Extract 'dummy-cpus.h' header from 'system/cpus.h'

In reply to Philippe Mathieu-Daudé
Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:32 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 26/39] accel/dummy: Extract 'dummy-cpus.h' header from
'system/cpus.h'
X-Mailer: git-send-email 2.49.0

'dummy' helpers are specific to accelerator implementations, no need to expose them via "system/cpus.h".

Quoted 9 lines
---
 MAINTAINERS           |  1 +
 accel/dummy-cpus.h    | 14 ++++++++++++++
 include/system/cpus.h |  5 -----
 accel/dummy-cpus.c    |  1 +
 accel/qtest/qtest.c   |  1 +
 accel/xen/xen-all.c   |  1 +
 6 files changed, 18 insertions(+), 5 deletions(-)
 create mode 100644 accel/dummy-cpus.h
#80 lore raw

PATCH v6 27/39 accel: Pass old/new interrupt mask to handle_interrupt() handler

In reply to Philippe Mathieu-Daudé

Update CPUState::interrupt_request once in cpu_interrupt(). Pass the old and new masks along.

---
 accel/tcg/tcg-accel-ops-icount.h |  2 +-
 accel/tcg/tcg-accel-ops.h        |  2 +-
 include/system/accel-ops.h       |  2 +-
 accel/tcg/tcg-accel-ops-icount.c |  8 +++-----
 accel/tcg/tcg-accel-ops.c        |  4 +---
 system/cpus.c                    | 12 +++++++-----
 6 files changed, 14 insertions(+), 16 deletions(-)
accel/tcg/tcg-accel-ops-icount.h +1 -1
index 16a301b6dc0..1d9d66f0707 100644
@@ -15,6 +15,6 @@ void icount_prepare_for_run(CPUState *cpu, int64_t cpu_budget);
 int64_t icount_percpu_budget(int cpu_count);
 void icount_process_data(CPUState *cpu);
 
-void icount_handle_interrupt(CPUState *cpu, int mask);
+void icount_handle_interrupt(CPUState *cpu, int old_mask, int new_mask);
 
 #endif /* TCG_ACCEL_OPS_ICOUNT_H */
accel/tcg/tcg-accel-ops.h +1 -1
index 6feeb3f3e9b..6e7c2aae5a8 100644
@@ -16,7 +16,7 @@
 
 void tcg_cpu_destroy(CPUState *cpu);
 int tcg_cpu_exec(CPUState *cpu);
-void tcg_handle_interrupt(CPUState *cpu, int mask);
+void tcg_handle_interrupt(CPUState *cpu, int old_mask, int new_mask);
 void tcg_cpu_init_cflags(CPUState *cpu, bool parallel);
 
 #endif /* TCG_ACCEL_OPS_H */
include/system/accel-ops.h +1 -1
index 2075691331c..d84eaa376c2 100644
@@ -61,7 +61,7 @@ struct AccelOpsClass {
     void (*synchronize_pre_loadvm)(CPUState *cpu);
     void (*synchronize_pre_resume)(bool step_pending);
 
-    void (*handle_interrupt)(CPUState *cpu, int mask);
+    void (*handle_interrupt)(CPUState *cpu, int old_mask, int new_mask);
 
     /**
      * @get_virtual_clock: fetch virtual clock
accel/tcg/tcg-accel-ops-icount.c +3 -5
index d0f7b410fab..500b5dd4942 100644
@@ -147,14 +147,12 @@ void icount_process_data(CPUState *cpu)
     replay_mutex_unlock();
 }
 
-void icount_handle_interrupt(CPUState *cpu, int mask)
+void icount_handle_interrupt(CPUState *cpu, int old_mask, int new_mask)
 {
-    int old_mask = cpu->interrupt_request;
-
-    tcg_handle_interrupt(cpu, mask);
+    tcg_handle_interrupt(cpu, old_mask, new_mask);
     if (qemu_cpu_is_self(cpu) &&
         !cpu->neg.can_do_io
-        && (mask & ~old_mask) != 0) {
+        && (new_mask & ~old_mask) != 0) {
         cpu_abort(cpu, "Raised interrupt while not in I/O function");
     }
 }
accel/tcg/tcg-accel-ops.c +1 -3
index f579685a611..eb16a6bf6db 100644
@@ -91,10 +91,8 @@ static void tcg_cpu_reset_hold(CPUState *cpu)
 }
 
 /* mask must never be zero, except for A20 change call */
-void tcg_handle_interrupt(CPUState *cpu, int mask)
+void tcg_handle_interrupt(CPUState *cpu, int old_mask, int new_mask)
 {
-    cpu->interrupt_request |= mask;
-
     /*
      * If called from iothread context, wake the target cpu in
      * case its halted.
system/cpus.c +7 -5
index 4fb764ac880..13535a74e6f 100644
@@ -246,10 +246,8 @@ int64_t cpus_get_elapsed_ticks(void)
     return cpu_get_ticks();
 }
 
-static void generic_handle_interrupt(CPUState *cpu, int mask)
+static void generic_handle_interrupt(CPUState *cpu, int old_mask, int new_mask)
 {
-    cpu->interrupt_request |= mask;
-
     if (!qemu_cpu_is_self(cpu)) {
         qemu_cpu_kick(cpu);
     }
@@ -257,12 +255,16 @@ static void generic_handle_interrupt(CPUState *cpu, int mask)
 
 void cpu_interrupt(CPUState *cpu, int mask)
 {
+    int old_mask = cpu->interrupt_request;
+
     g_assert(bql_locked());
 
+    cpu->interrupt_request |= mask;
+
     if (cpus_accel->handle_interrupt) {
-        cpus_accel->handle_interrupt(cpu, mask);
+        cpus_accel->handle_interrupt(cpu, old_mask, cpu->interrupt_request);
     } else {
-        generic_handle_interrupt(cpu, mask);
+        generic_handle_interrupt(cpu, old_mask, cpu->interrupt_request);
     }
 }
 
Signature · 1 line
2.49.0
#81 lore raw

Re: [PATCH v6 27/39] accel: Pass old/new interrupt mask to handle_interrupt() handler

Quoted 61 lines · Philippe Mathieu-Daudé

Update CPUState::interrupt_request once in cpu_interrupt(). Pass the old and new masks along.

---
  accel/tcg/tcg-accel-ops-icount.h |  2 +-
  accel/tcg/tcg-accel-ops.h        |  2 +-
  include/system/accel-ops.h       |  2 +-
  accel/tcg/tcg-accel-ops-icount.c |  8 +++-----
  accel/tcg/tcg-accel-ops.c        |  4 +---
  system/cpus.c                    | 12 +++++++-----
  6 files changed, 14 insertions(+), 16 deletions(-)
accel/tcg/tcg-accel-ops-icount.h +1 -1
index 16a301b6dc0..1d9d66f0707 100644
@@ -15,6 +15,6 @@ void icount_prepare_for_run(CPUState *cpu, int64_t cpu_budget);
  int64_t icount_percpu_budget(int cpu_count);
  void icount_process_data(CPUState *cpu);
 
-void icount_handle_interrupt(CPUState *cpu, int mask);
+void icount_handle_interrupt(CPUState *cpu, int old_mask, int new_mask);
 
  #endif /* TCG_ACCEL_OPS_ICOUNT_H */
accel/tcg/tcg-accel-ops.h +1 -1
index 6feeb3f3e9b..6e7c2aae5a8 100644
@@ -16,7 +16,7 @@
 
  void tcg_cpu_destroy(CPUState *cpu);
  int tcg_cpu_exec(CPUState *cpu);
-void tcg_handle_interrupt(CPUState *cpu, int mask);
+void tcg_handle_interrupt(CPUState *cpu, int old_mask, int new_mask);
  void tcg_cpu_init_cflags(CPUState *cpu, bool parallel);
 
  #endif /* TCG_ACCEL_OPS_H */
include/system/accel-ops.h +1 -1
index 2075691331c..d84eaa376c2 100644
@@ -61,7 +61,7 @@ struct AccelOpsClass {
      void (*synchronize_pre_loadvm)(CPUState *cpu);
      void (*synchronize_pre_resume)(bool step_pending);
 
-    void (*handle_interrupt)(CPUState *cpu, int mask);
+    void (*handle_interrupt)(CPUState *cpu, int old_mask, int new_mask);
 
      /**
       * @get_virtual_clock: fetch virtual clock
accel/tcg/tcg-accel-ops-icount.c +0 -1
index d0f7b410fab..500b5dd4942 100644
@@ -147,14 +147,12 @@ void icount_process_data(CPUState *cpu)
      replay_mutex_unlock();
  }
 
-void icount_handle_interrupt(CPUState *cpu, int mask)

accel/tcg/tcg-accel-ops-icount.c void icount_process_data(CPUState *cpu) · line 155

Quoted 10 lines
accel/tcg/tcg-accel-ops-icount.c +3 -4
+void icount_handle_interrupt(CPUState *cpu, int old_mask, int new_mask)
  {
-    int old_mask = cpu->interrupt_request;
-
-    tcg_handle_interrupt(cpu, mask);
+    tcg_handle_interrupt(cpu, old_mask, new_mask);
      if (qemu_cpu_is_self(cpu) &&
          !cpu->neg.can_do_io
-        && (mask & ~old_mask) != 0) {
+        && (new_mask & ~old_mask) != 0) {

This patch changes the behavior,

Assume the @mask is the value passed to cpu_interrupt()

- before this patch:

	(mask & ~old_mask) is

	(@mask & ~cpu->interrupt_request)

- after this patch:

	(new_mask & ~old_mask) is

	((@mask | cpu->interrupt_request) & ~cpu->interrupt_request)
Reviewed-by on v6 27/39
#82 lore raw

Re: [PATCH v6 27/39] accel: Pass old/new interrupt mask to handle_interrupt() handler

In reply to Philippe Mathieu-Daudé
Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:33 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 27/39] accel: Pass old/new interrupt mask to
handle_interrupt() handler
X-Mailer: git-send-email 2.49.0

Update CPUState::interrupt_request once in cpu_interrupt(). Pass the old and new masks along.

Quoted 8 lines
---
 accel/tcg/tcg-accel-ops-icount.h |  2 +-
 accel/tcg/tcg-accel-ops.h        |  2 +-
 include/system/accel-ops.h       |  2 +-
 accel/tcg/tcg-accel-ops-icount.c |  8 +++-----
 accel/tcg/tcg-accel-ops.c        |  4 +---
 system/cpus.c                    | 12 +++++++-----
 6 files changed, 14 insertions(+), 16 deletions(-)
#83 lore raw

PATCH v6 28/39 accel: Expose and register generic_handle_interrupt()

In reply to Philippe Mathieu-Daudé

In order to dispatch over AccelOpsClass::handle_interrupt(), we need it always defined, not calling a hidden handler under the hood. Make AccelOpsClass::handle_interrupt() mandatory. Expose generic_handle_interrupt() prototype and register it for each accelerator.

---
 include/system/accel-ops.h        |  3 +++
 accel/hvf/hvf-accel-ops.c         |  1 +
 accel/kvm/kvm-accel-ops.c         |  1 +
 accel/qtest/qtest.c               |  1 +
 accel/xen/xen-all.c               |  1 +
 system/cpus.c                     | 10 ++++------
 target/i386/nvmm/nvmm-accel-ops.c |  1 +
 target/i386/whpx/whpx-accel-ops.c |  1 +
 8 files changed, 13 insertions(+), 6 deletions(-)
include/system/accel-ops.h +3 -0
index d84eaa376c2..95a0f402cde 100644
@@ -61,6 +61,7 @@ struct AccelOpsClass {
     void (*synchronize_pre_loadvm)(CPUState *cpu);
     void (*synchronize_pre_resume)(bool step_pending);
 
+    /* handle_interrupt is mandatory. */
     void (*handle_interrupt)(CPUState *cpu, int old_mask, int new_mask);
 
     /**
@@ -84,4 +85,6 @@ struct AccelOpsClass {
     void (*remove_all_breakpoints)(CPUState *cpu);
 };
 
+void generic_handle_interrupt(CPUState *cpu, int old_mask, int new_mask);
+
 #endif /* ACCEL_OPS_H */
accel/hvf/hvf-accel-ops.c +1 -0
index 2944e350ca9..a0248942f3a 100644
@@ -353,6 +353,7 @@ static void hvf_accel_ops_class_init(ObjectClass *oc, const void *data)
 
     ops->create_vcpu_thread = hvf_start_vcpu_thread;
     ops->kick_vcpu_thread = hvf_kick_vcpu_thread;
+    ops->handle_interrupt = generic_handle_interrupt;
 
     ops->synchronize_post_reset = hvf_cpu_synchronize_post_reset;
     ops->synchronize_post_init = hvf_cpu_synchronize_post_init;
accel/kvm/kvm-accel-ops.c +1 -0
index 99f61044da5..2a744092749 100644
@@ -95,6 +95,7 @@ static void kvm_accel_ops_class_init(ObjectClass *oc, const void *data)
     ops->synchronize_post_init = kvm_cpu_synchronize_post_init;
     ops->synchronize_state = kvm_cpu_synchronize_state;
     ops->synchronize_pre_loadvm = kvm_cpu_synchronize_pre_loadvm;
+    ops->handle_interrupt = generic_handle_interrupt;
 
 #ifdef TARGET_KVM_HAVE_GUEST_DEBUG
     ops->update_guest_debug = kvm_update_guest_debug_ops;
accel/qtest/qtest.c +1 -0
index 612cede160b..5474ce73135 100644
@@ -67,6 +67,7 @@ static void qtest_accel_ops_class_init(ObjectClass *oc, const void *data)
     ops->create_vcpu_thread = dummy_start_vcpu_thread;
     ops->get_virtual_clock = qtest_get_virtual_clock;
     ops->set_virtual_clock = qtest_set_virtual_clock;
+    ops->handle_interrupt = generic_handle_interrupt;
 };
 
 static const TypeInfo qtest_accel_ops_type = {
accel/xen/xen-all.c +1 -0
index c150dd43cab..c12c22de785 100644
@@ -153,6 +153,7 @@ static void xen_accel_ops_class_init(ObjectClass *oc, const void *data)
     AccelOpsClass *ops = ACCEL_OPS_CLASS(oc);
 
     ops->create_vcpu_thread = dummy_start_vcpu_thread;
+    ops->handle_interrupt = generic_handle_interrupt;
 }
 
 static const TypeInfo xen_accel_ops_type = {
system/cpus.c +4 -6
index 13535a74e6f..f90b8be9eee 100644
@@ -246,7 +246,7 @@ int64_t cpus_get_elapsed_ticks(void)
     return cpu_get_ticks();
 }
 
-static void generic_handle_interrupt(CPUState *cpu, int old_mask, int new_mask)
+void generic_handle_interrupt(CPUState *cpu, int old_mask, int new_mask)
 {
     if (!qemu_cpu_is_self(cpu)) {
         qemu_cpu_kick(cpu);
@@ -261,11 +261,7 @@ void cpu_interrupt(CPUState *cpu, int mask)
 
     cpu->interrupt_request |= mask;
 
-    if (cpus_accel->handle_interrupt) {
-        cpus_accel->handle_interrupt(cpu, old_mask, cpu->interrupt_request);
-    } else {
-        generic_handle_interrupt(cpu, old_mask, cpu->interrupt_request);
-    }
+    cpus_accel->handle_interrupt(cpu, old_mask, cpu->interrupt_request);
 }
 
 /*
@@ -674,6 +670,8 @@ void cpus_register_accel(const AccelOpsClass *ops)
 {
     assert(ops != NULL);
     assert(ops->create_vcpu_thread != NULL); /* mandatory */
+    assert(ops->handle_interrupt);
+
     cpus_accel = ops;
 }
 
target/i386/nvmm/nvmm-accel-ops.c +1 -0
index 21443078b72..a5517b0abf3 100644
@@ -87,6 +87,7 @@ static void nvmm_accel_ops_class_init(ObjectClass *oc, const void *data)
 
     ops->create_vcpu_thread = nvmm_start_vcpu_thread;
     ops->kick_vcpu_thread = nvmm_kick_vcpu_thread;
+    ops->handle_interrupt = generic_handle_interrupt;
 
     ops->synchronize_post_reset = nvmm_cpu_synchronize_post_reset;
     ops->synchronize_post_init = nvmm_cpu_synchronize_post_init;
target/i386/whpx/whpx-accel-ops.c +1 -0
index b8bebe403c9..31cf15f0045 100644
@@ -90,6 +90,7 @@ static void whpx_accel_ops_class_init(ObjectClass *oc, const void *data)
     ops->create_vcpu_thread = whpx_start_vcpu_thread;
     ops->kick_vcpu_thread = whpx_kick_vcpu_thread;
     ops->cpu_thread_is_idle = whpx_vcpu_thread_is_idle;
+    ops->handle_interrupt = generic_handle_interrupt;
 
     ops->synchronize_post_reset = whpx_cpu_synchronize_post_reset;
     ops->synchronize_post_init = whpx_cpu_synchronize_post_init;
Signature · 1 line
2.49.0
Reviewed-by on v6 28/39
#84 lore raw

Re: [PATCH v6 28/39] accel: Expose and register generic_handle_interrupt()

Quoted 2 lines · Philippe Mathieu-Daudé

In order to dispatch over AccelOpsClass::handle_interrupt(), we need it always defined,

It seems I can only understand it until I see the code to really require it to be mandatory.

But anyway, the change itself is correct.

Quoted 4 lines

not calling a hidden handler under
the hood. Make AccelOpsClass::handle_interrupt() mandatory. Expose generic_handle_interrupt() prototype and register it for each accelerator.

#85 lore raw

Re: [PATCH v6 28/39] accel: Expose and register generic_handle_interrupt()

Quoted 5 lines · Xiaoyao Li
Quoted 2 lines · Philippe Mathieu-Daudé

In order to dispatch over AccelOpsClass::handle_interrupt(), we need it always defined,

It seems I can only understand it until I see the code to really require it to be mandatory.

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

Quoted 4 lines

But anyway, the change itself is correct.

Thanks!

#86 lore raw

PATCH v6 29/39 accel: Keep reference to AccelOpsClass in AccelClass

In reply to Philippe Mathieu-Daudé

Allow dereferencing AccelOpsClass outside of accel/accel-system.c.

---
 include/qemu/accel.h       | 3 +++
 include/system/accel-ops.h | 3 ++-
 accel/accel-common.c       | 1 +
 accel/accel-system.c       | 3 ++-
 accel/tcg/tcg-accel-ops.c  | 4 +++-
 5 files changed, 11 insertions(+), 3 deletions(-)
include/qemu/accel.h +3 -0
index f987d16baaa..197badcb705 100644
@@ -37,6 +37,9 @@ typedef struct AccelClass {
     /*< public >*/
 
     const char *name;
+    /* Cached by accel_init_ops_interfaces() when created */
+    AccelOpsClass *ops;
+
     int (*init_machine)(MachineState *ms);
     bool (*cpu_common_realize)(CPUState *cpu, Error **errp);
     void (*cpu_common_unrealize)(CPUState *cpu);
include/system/accel-ops.h +2 -1
index 95a0f402cde..17c80887016 100644
@@ -10,6 +10,7 @@
 #ifndef ACCEL_OPS_H
 #define ACCEL_OPS_H
 
+#include "qemu/accel.h"
 #include "exec/vaddr.h"
 #include "qom/object.h"
 
@@ -31,7 +32,7 @@ struct AccelOpsClass {
     /*< public >*/
 
     /* initialization function called when accel is chosen */
-    void (*ops_init)(AccelOpsClass *ops);
+    void (*ops_init)(AccelClass *ac);
 
     void (*cpu_reset_hold)(CPUState *cpu);
 
accel/accel-common.c +1 -0
index 4894b98d64a..56d88940f92 100644
@@ -10,6 +10,7 @@
 #include "qemu/osdep.h"
 #include "qemu/accel.h"
 #include "qemu/target-info.h"
+#include "system/accel-ops.h"
 #include "accel/accel-cpu.h"
 #include "accel-internal.h"
 
accel/accel-system.c +2 -1
index 07b75dae797..b5b368c6a9c 100644
@@ -95,8 +95,9 @@ void accel_init_ops_interfaces(AccelClass *ac)
      * non-NULL create_vcpu_thread operation.
      */
     ops = ACCEL_OPS_CLASS(oc);
+    ac->ops = ops;
     if (ops->ops_init) {
-        ops->ops_init(ops);
+        ops->ops_init(ac);
     }
     cpus_register_accel(ops);
 }
accel/tcg/tcg-accel-ops.c +3 -1
index eb16a6bf6db..bc809ad5640 100644
@@ -189,8 +189,10 @@ static inline void tcg_remove_all_breakpoints(CPUState *cpu)
     cpu_watchpoint_remove_all(cpu, BP_GDB);
 }
 
-static void tcg_accel_ops_init(AccelOpsClass *ops)
+static void tcg_accel_ops_init(AccelClass *ac)
 {
+    AccelOpsClass *ops = ac->ops;
+
     if (qemu_tcg_mttcg_enabled()) {
         ops->create_vcpu_thread = mttcg_start_vcpu_thread;
         ops->kick_vcpu_thread = mttcg_kick_vcpu_thread;
Signature · 1 line
2.49.0
Reviewed-by on v6 29/39
#87 lore raw

Re: [PATCH v6 29/39] accel: Keep reference to AccelOpsClass in AccelClass

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:35 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 29/39] accel: Keep reference to AccelOpsClass in
AccelClass
X-Mailer: git-send-email 2.49.0

Allow dereferencing AccelOpsClass outside of accel/accel-system.c.

Quoted 7 lines
---
 include/qemu/accel.h       | 3 +++
 include/system/accel-ops.h | 3 ++-
 accel/accel-common.c       | 1 +
 accel/accel-system.c       | 3 ++-
 accel/tcg/tcg-accel-ops.c  | 4 +++-
 5 files changed, 11 insertions(+), 3 deletions(-)> 
#88 lore raw

PATCH v6 30/39 accel: Propagate AccelState to AccelClass::init_machine()

In reply to Philippe Mathieu-Daudé

In order to avoid init_machine() to call current_accel(), pass AccelState along.

---
 include/qemu/accel.h        | 2 +-
 accel/accel-system.c        | 2 +-
 accel/hvf/hvf-all.c         | 2 +-
 accel/kvm/kvm-all.c         | 2 +-
 accel/qtest/qtest.c         | 2 +-
 accel/tcg/tcg-all.c         | 2 +-
 accel/xen/xen-all.c         | 2 +-
 bsd-user/main.c             | 2 +-
 linux-user/main.c           | 2 +-
 target/i386/nvmm/nvmm-all.c | 2 +-
 target/i386/whpx/whpx-all.c | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)
include/qemu/accel.h +1 -1
index 197badcb705..b040fa104b6 100644
@@ -40,7 +40,7 @@ typedef struct AccelClass {
     /* Cached by accel_init_ops_interfaces() when created */
     AccelOpsClass *ops;
 
-    int (*init_machine)(MachineState *ms);
+    int (*init_machine)(AccelState *as, MachineState *ms);
     bool (*cpu_common_realize)(CPUState *cpu, Error **errp);
     void (*cpu_common_unrealize)(CPUState *cpu);
 
accel/accel-system.c +1 -1
index b5b368c6a9c..fb8abe38594 100644
@@ -37,7 +37,7 @@ int accel_init_machine(AccelState *accel, MachineState *ms)
     int ret;
     ms->accelerator = accel;
     *(acc->allowed) = true;
-    ret = acc->init_machine(ms);
+    ret = acc->init_machine(accel, ms);
     if (ret < 0) {
         ms->accelerator = NULL;
         *(acc->allowed) = false;
accel/hvf/hvf-all.c +1 -1
index 897a02eebe2..2cf2b18fd23 100644
@@ -247,7 +247,7 @@ static MemoryListener hvf_memory_listener = {
     .log_sync = hvf_log_sync,
 };
 
-static int hvf_accel_init(MachineState *ms)
+static int hvf_accel_init(AccelState *as, MachineState *ms)
 {
     int x;
     hv_return_t ret;
accel/kvm/kvm-all.c +1 -1
index 88fb6d36941..1b6b7006470 100644
@@ -2573,7 +2573,7 @@ static int kvm_setup_dirty_ring(KVMState *s)
     return 0;
 }
 
-static int kvm_init(MachineState *ms)
+static int kvm_init(AccelState *as, MachineState *ms)
 {
     MachineClass *mc = MACHINE_GET_CLASS(ms);
     static const char upgrade_note[] =
accel/qtest/qtest.c +1 -1
index 5474ce73135..2b831260201 100644
@@ -38,7 +38,7 @@ static void qtest_set_virtual_clock(int64_t count)
     qatomic_set_i64(&qtest_clock_counter, count);
 }
 
-static int qtest_init_accel(MachineState *ms)
+static int qtest_init_accel(AccelState *as, MachineState *ms)
 {
     return 0;
 }
accel/tcg/tcg-all.c +1 -1
index 0cff0f8a0f9..7ae7d552d9e 100644
@@ -80,7 +80,7 @@ static void tcg_accel_instance_init(Object *obj)
 
 bool one_insn_per_tb;
 
-static int tcg_init_machine(MachineState *ms)
+static int tcg_init_machine(AccelState *as, MachineState *ms)
 {
     TCGState *s = TCG_STATE(current_accel());
     unsigned max_threads = 1;
accel/xen/xen-all.c +1 -1
index c12c22de785..8279746f115 100644
@@ -77,7 +77,7 @@ static void xen_setup_post(MachineState *ms, AccelState *accel)
     }
 }
 
-static int xen_init(MachineState *ms)
+static int xen_init(AccelState *as, MachineState *ms)
 {
     MachineClass *mc = MACHINE_GET_CLASS(ms);
 
bsd-user/main.c +1 -1
index 7c0a059c3ba..d0cc8e0088f 100644
@@ -474,7 +474,7 @@ int main(int argc, char **argv)
                                  opt_one_insn_per_tb, &error_abort);
         object_property_set_int(OBJECT(accel), "tb-size",
                                 opt_tb_size, &error_abort);
-        ac->init_machine(NULL);
+        ac->init_machine(accel, NULL);
     }
 
     /*
linux-user/main.c +1 -1
index 5ac5b55dc65..a9142ee7268 100644
@@ -820,7 +820,7 @@ int main(int argc, char **argv, char **envp)
                                  opt_one_insn_per_tb, &error_abort);
         object_property_set_int(OBJECT(accel), "tb-size",
                                 opt_tb_size, &error_abort);
-        ac->init_machine(NULL);
+        ac->init_machine(accel, NULL);
     }
 
     /*
target/i386/nvmm/nvmm-all.c +1 -1
index 2df49d7eeb4..b4a4d50e860 100644
@@ -1152,7 +1152,7 @@ static struct RAMBlockNotifier nvmm_ram_notifier = {
 /* -------------------------------------------------------------------------- */
 
 static int
-nvmm_accel_init(MachineState *ms)
+nvmm_accel_init(AccelState *as, MachineState *ms)
 {
     int ret, err;
 
target/i386/whpx/whpx-all.c +1 -1
index 94fd5fc7849..721c4782b9c 100644
@@ -2504,7 +2504,7 @@ static void whpx_set_kernel_irqchip(Object *obj, Visitor *v,
  * Partition support
  */
 
-static int whpx_accel_init(MachineState *ms)
+static int whpx_accel_init(AccelState *as, MachineState *ms)
 {
     struct whpx_state *whpx;
     int ret;
Signature · 1 line
2.49.0
Reviewed-by on v6 30/39
#89 lore raw

Re: [PATCH v6 30/39] accel: Propagate AccelState to AccelClass::init_machine()

Quoted 25 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:36 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 30/39] accel: Propagate AccelState to
AccelClass::init_machine()
X-Mailer: git-send-email 2.49.0

In order to avoid init_machine() to call current_accel(), pass AccelState along.

---
 include/qemu/accel.h        | 2 +-
 accel/accel-system.c        | 2 +-
 accel/hvf/hvf-all.c         | 2 +-
 accel/kvm/kvm-all.c         | 2 +-
 accel/qtest/qtest.c         | 2 +-
 accel/tcg/tcg-all.c         | 2 +-
 accel/xen/xen-all.c         | 2 +-
 bsd-user/main.c             | 2 +-
 linux-user/main.c           | 2 +-
 target/i386/nvmm/nvmm-all.c | 2 +-
 target/i386/whpx/whpx-all.c | 2 +-
 11 files changed, 11 insertions(+), 11 deletions(-)

...

accel/accel-system.c int accel_init_machine(AccelState *accel, MachineState *ms) · line 40

Quoted 10 lines
accel/accel-system.c +1 -1
index b5b368c6a9c..fb8abe38594 100644
@@ -37,7 +37,7 @@ int accel_init_machine(AccelState *accel, MachineState *ms)
     int ret;
     ms->accelerator = accel;
     *(acc->allowed) = true;
-    ret = acc->init_machine(ms);
+    ret = acc->init_machine(accel, ms);

Now we've already set "ms->accelerator", so that we could get @accel by ms->accelerator.

But considerring the user emulation, where the @ms is NULL, and for these cases, it needs to bring current_accel() back in patch 32.

Anyway, this solution is also fine for me, so,

...But there're still more comments/questions about user emulation:

bsd-user/main.c int main(int argc, char **argv) · line 477

Quoted 8 lines
bsd-user/main.c +1 -1
@@ -474,7 +474,7 @@ int main(int argc, char **argv)
                                  opt_one_insn_per_tb, &error_abort);
         object_property_set_int(OBJECT(accel), "tb-size",
                                 opt_tb_size, &error_abort);
-        ac->init_machine(NULL);
+        ac->init_machine(accel, NULL);

Not the issue about this patch though,

it seems user emulation doesn't set acc->allowed. At least TCG enabled is necessary, I guess?

linux-user/main.c int main(int argc, char **argv, char **envp) · line 823

Quoted 13 lines
bsd-user/main.c +0 -0
     }
 
     /*
linux-user/main.c +1 -1
index 5ac5b55dc65..a9142ee7268 100644
@@ -820,7 +820,7 @@ int main(int argc, char **argv, char **envp)
                                  opt_one_insn_per_tb, &error_abort);
         object_property_set_int(OBJECT(accel), "tb-size",
                                 opt_tb_size, &error_abort);
-        ac->init_machine(NULL);
+        ac->init_machine(accel, NULL);

Ditto.

linux-user/main.c int main(int argc, char **argv, char **envp) · line 826

Quoted 3 lines
linux-user/main.c +0 -0
     }
 
     /*

Thanks,
Zhao

#90 lore raw

PATCH v6 31/39 accel/kvm: Prefer local AccelState over global MachineState::accel

In reply to Philippe Mathieu-Daudé
---
 accel/kvm/kvm-all.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
accel/kvm/kvm-all.c +1 -3
index 1b6b7006470..a6ea2c7f614 100644
@@ -2588,15 +2588,13 @@ static int kvm_init(AccelState *as, MachineState *ms)
         { /* end of list */ }
     }, *nc = num_cpus;
     int soft_vcpus_limit, hard_vcpus_limit;
-    KVMState *s;
+    KVMState *s = KVM_STATE(as);
     const KVMCapabilityInfo *missing_cap;
     int ret;
     int type;
 
     qemu_mutex_init(&kml_slots_lock);
 
-    s = KVM_STATE(ms->accelerator);
-
     /*
      * On systems where the kernel can support different base page
      * sizes, host page size may be different from TARGET_PAGE_SIZE,
Signature · 1 line
2.49.0
Reviewed-by on v6 31/39
#91 lore raw

Re: [PATCH v6 31/39] accel/kvm: Prefer local AccelState over global MachineState::accel

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:37 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 31/39] accel/kvm: Prefer local AccelState over global
MachineState::accel
X-Mailer: git-send-email 2.49.0

---
 accel/kvm/kvm-all.c | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)
#92 lore raw

PATCH v6 32/39 accel/tcg: Prefer local AccelState over global current_accel()

In reply to Philippe Mathieu-Daudé
---
 accel/tcg/tcg-all.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
accel/tcg/tcg-all.c +1 -1
index 7ae7d552d9e..969c50c87ea 100644
@@ -82,7 +82,7 @@ bool one_insn_per_tb;
 
 static int tcg_init_machine(AccelState *as, MachineState *ms)
 {
-    TCGState *s = TCG_STATE(current_accel());
+    TCGState *s = TCG_STATE(as);
     unsigned max_threads = 1;
 
 #ifndef CONFIG_USER_ONLY
Signature · 1 line
2.49.0
Reviewed-by on v6 32/39
#93 lore raw

Re: [PATCH v6 32/39] accel/tcg: Prefer local AccelState over global current_accel()

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:38 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 32/39] accel/tcg: Prefer local AccelState over global
current_accel()
X-Mailer: git-send-email 2.49.0

---
 accel/tcg/tcg-all.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
#94 lore raw

PATCH v6 33/39 accel: Directly pass AccelState argument to AccelClass::has_memory()

In reply to Philippe Mathieu-Daudé
---
 include/qemu/accel.h | 2 +-
 accel/kvm/kvm-all.c  | 4 ++--
 system/memory.c      | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)
include/qemu/accel.h +1 -1
index b040fa104b6..44189b77daa 100644
@@ -46,7 +46,7 @@ typedef struct AccelClass {
 
     /* system related hooks */
     void (*setup_post)(MachineState *ms, AccelState *accel);
-    bool (*has_memory)(MachineState *ms, AddressSpace *as,
+    bool (*has_memory)(AccelState *accel, AddressSpace *as,
                        hwaddr start_addr, hwaddr size);
     bool (*cpus_are_resettable)(AccelState *as);
 
accel/kvm/kvm-all.c +2 -2
index a6ea2c7f614..0cd9b2f29ab 100644
@@ -3789,10 +3789,10 @@ int kvm_get_one_reg(CPUState *cs, uint64_t id, void *target)
     return r;
 }
 
-static bool kvm_accel_has_memory(MachineState *ms, AddressSpace *as,
+static bool kvm_accel_has_memory(AccelState *accel, AddressSpace *as,
                                  hwaddr start_addr, hwaddr size)
 {
-    KVMState *kvm = KVM_STATE(ms->accelerator);
+    KVMState *kvm = KVM_STATE(accel);
     int i;
 
     for (i = 0; i < kvm->nr_as; ++i) {
system/memory.c +1 -1
index 4f713889a8e..b072a6bef83 100644
@@ -3496,7 +3496,7 @@ static void mtree_print_flatview(gpointer key, gpointer value,
         if (fvi->ac) {
             for (i = 0; i < fv_address_spaces->len; ++i) {
                 as = g_array_index(fv_address_spaces, AddressSpace*, i);
-                if (fvi->ac->has_memory(current_machine, as,
+                if (fvi->ac->has_memory(current_machine->accelerator, as,
                                         int128_get64(range->addr.start),
                                         MR_SIZE(range->addr.size) + 1)) {
                     qemu_printf(" %s", fvi->ac->name);
Signature · 1 line
2.49.0
Reviewed-by on v6 33/39
#95 lore raw

Re: [PATCH v6 33/39] accel: Directly pass AccelState argument to AccelClass::has_memory()

Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:39 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 33/39] accel: Directly pass AccelState argument to
AccelClass::has_memory()
X-Mailer: git-send-email 2.49.0

Quoted 7 lines
---
 include/qemu/accel.h | 2 +-
 accel/kvm/kvm-all.c  | 4 ++--
 system/memory.c      | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)
Reviewed-by on v6 33/39
#96 lore raw

Re: [PATCH v6 33/39] accel: Directly pass AccelState argument to AccelClass::has_memory()

In reply to Philippe Mathieu-Daudé
Quoted 6 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:39 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 33/39] accel: Directly pass AccelState argument to
AccelClass::has_memory()
X-Mailer: git-send-email 2.49.0

Quoted 7 lines
---
 include/qemu/accel.h | 2 +-
 accel/kvm/kvm-all.c  | 4 ++--
 system/memory.c      | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)
#97 lore raw

PATCH v6 34/39 accel/kvm: Directly pass KVMState argument to do_kvm_create_vm()

In reply to Philippe Mathieu-Daudé
---
 accel/kvm/kvm-all.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
accel/kvm/kvm-all.c +2 -5
index 0cd9b2f29ab..f1c3d4d27c7 100644
@@ -2470,13 +2470,10 @@ uint32_t kvm_dirty_ring_size(void)
     return kvm_state->kvm_dirty_ring_size;
 }
 
-static int do_kvm_create_vm(MachineState *ms, int type)
+static int do_kvm_create_vm(KVMState *s, int type)
 {
-    KVMState *s;
     int ret;
 
-    s = KVM_STATE(ms->accelerator);
-
     do {
         ret = kvm_ioctl(s, KVM_CREATE_VM, type);
     } while (ret == -EINTR);
@@ -2646,7 +2643,7 @@ static int kvm_init(AccelState *as, MachineState *ms)
         goto err;
     }
 
-    ret = do_kvm_create_vm(ms, type);
+    ret = do_kvm_create_vm(s, type);
     if (ret < 0) {
         goto err;
     }
Signature · 1 line
2.49.0
Reviewed-by on v6 34/39
#98 lore raw

Re: [PATCH v6 34/39] accel/kvm: Directly pass KVMState argument to do_kvm_create_vm()

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:40 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 34/39] accel/kvm: Directly pass KVMState argument to
do_kvm_create_vm()
X-Mailer: git-send-email 2.49.0

---
 accel/kvm/kvm-all.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
Reviewed-by on v6 34/39
#99 lore raw

Re: [PATCH v6 34/39] accel/kvm: Directly pass KVMState argument to do_kvm_create_vm()

In reply to Philippe Mathieu-Daudé
Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:40 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 34/39] accel/kvm: Directly pass KVMState argument to
do_kvm_create_vm()
X-Mailer: git-send-email 2.49.0

---
 accel/kvm/kvm-all.c | 7 ++-----
 1 file changed, 2 insertions(+), 5 deletions(-)
#100 lore raw

PATCH v6 35/39 accel: Remove unused MachineState argument of AccelClass::setup_post()

In reply to Philippe Mathieu-Daudé

This method only accesses xen_domid/xen_domid_restrict, which are both related to the 'accelerator', not the machine. Besides, xen_domid aims to be in Xen AccelState and xen_domid_restrict a xen_domid_restrict QOM property.

---
 include/qemu/accel.h | 2 +-
 accel/accel-system.c | 2 +-
 accel/xen/xen-all.c  | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)
include/qemu/accel.h +1 -1
index 44189b77daa..19ccc5ef6a1 100644
@@ -45,7 +45,7 @@ typedef struct AccelClass {
     void (*cpu_common_unrealize)(CPUState *cpu);
 
     /* system related hooks */
-    void (*setup_post)(MachineState *ms, AccelState *accel);
+    void (*setup_post)(AccelState *as);
     bool (*has_memory)(AccelState *accel, AddressSpace *as,
                        hwaddr start_addr, hwaddr size);
     bool (*cpus_are_resettable)(AccelState *as);
accel/accel-system.c +1 -1
index fb8abe38594..637e2390f35 100644
@@ -58,7 +58,7 @@ void accel_setup_post(MachineState *ms)
     AccelState *accel = ms->accelerator;
     AccelClass *acc = ACCEL_GET_CLASS(accel);
     if (acc->setup_post) {
-        acc->setup_post(ms, accel);
+        acc->setup_post(accel);
     }
 }
 
accel/xen/xen-all.c +1 -1
index 8279746f115..bd0ff64befc 100644
@@ -64,7 +64,7 @@ static void xen_set_igd_gfx_passthru(Object *obj, bool value, Error **errp)
     xen_igd_gfx_pt_set(value, errp);
 }
 
-static void xen_setup_post(MachineState *ms, AccelState *accel)
+static void xen_setup_post(AccelState *as)
 {
     int rc;
 
Signature · 1 line
2.49.0
Reviewed-by on v6 35/39
#101 lore raw

Re: [PATCH v6 35/39] accel: Remove unused MachineState argument of AccelClass::setup_post()

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:41 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 35/39] accel: Remove unused MachineState argument of
AccelClass::setup_post()
X-Mailer: git-send-email 2.49.0

This method only accesses xen_domid/xen_domid_restrict, which are both related to the 'accelerator', not the machine. Besides, xen_domid aims to be in Xen AccelState and xen_domid_restrict a xen_domid_restrict QOM property.

Quoted 7 lines
---
 include/qemu/accel.h | 2 +-
 accel/accel-system.c | 2 +-
 accel/xen/xen-all.c  | 2 +-
 3 files changed, 3 insertions(+), 3 deletions(-)
#102 lore raw

PATCH v6 36/39 accel: Pass AccelState argument to gdbstub_supported_sstep_flags()

In reply to Philippe Mathieu-Daudé

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

---
 include/qemu/accel.h | 2 +-
 accel/accel-common.c | 2 +-
 accel/hvf/hvf-all.c  | 2 +-
 accel/kvm/kvm-all.c  | 2 +-
 accel/tcg/tcg-all.c  | 2 +-
 5 files changed, 5 insertions(+), 5 deletions(-)
include/qemu/accel.h +1 -1
index 19ccc5ef6a1..3c6350d6d63 100644
@@ -52,7 +52,7 @@ typedef struct AccelClass {
 
     /* gdbstub related hooks */
     bool (*supports_guest_debug)(AccelState *as);
-    int (*gdbstub_supported_sstep_flags)(void);
+    int (*gdbstub_supported_sstep_flags)(AccelState *as);
 
     bool *allowed;
     /*
accel/accel-common.c +1 -1
index 56d88940f92..b3fbe3216aa 100644
@@ -125,7 +125,7 @@ int accel_supported_gdbstub_sstep_flags(void)
     AccelState *accel = current_accel();
     AccelClass *acc = ACCEL_GET_CLASS(accel);
     if (acc->gdbstub_supported_sstep_flags) {
-        return acc->gdbstub_supported_sstep_flags();
+        return acc->gdbstub_supported_sstep_flags(accel);
     }
     return 0;
 }
accel/hvf/hvf-all.c +1 -1
index 2cf2b18fd23..4fae4c79805 100644
@@ -281,7 +281,7 @@ static int hvf_accel_init(AccelState *as, MachineState *ms)
     return hvf_arch_init();
 }
 
-static int hvf_gdbstub_sstep_flags(void)
+static int hvf_gdbstub_sstep_flags(AccelState *as)
 {
     return SSTEP_ENABLE | SSTEP_NOIRQ;
 }
accel/kvm/kvm-all.c +1 -1
index f1c3d4d27c7..9d1dc56d7e8 100644
@@ -3985,7 +3985,7 @@ static bool kvm_cpus_are_resettable(AccelState *as)
  * Returns: SSTEP_* flags that KVM supports for guest debug. The
  * support is probed during kvm_init()
  */
-static int kvm_gdbstub_sstep_flags(void)
+static int kvm_gdbstub_sstep_flags(AccelState *as)
 {
     return kvm_sstep_flags;
 }
accel/tcg/tcg-all.c +1 -1
index 969c50c87ea..93972bc0919 100644
@@ -224,7 +224,7 @@ static bool tcg_supports_guest_debug(AccelState *as)
     return true;
 }
 
-static int tcg_gdbstub_supported_sstep_flags(void)
+static int tcg_gdbstub_supported_sstep_flags(AccelState *as)
 {
     /*
      * In replay mode all events will come from the log and can't be
Signature · 1 line
2.49.0
Reviewed-by on v6 36/39
#103 lore raw

Re: [PATCH v6 36/39] accel: Pass AccelState argument to gdbstub_supported_sstep_flags()

Quoted 11 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:42 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 36/39] accel: Pass AccelState argument to
gdbstub_supported_sstep_flags()
X-Mailer: git-send-email 2.49.0

In order to have AccelClass methods instrospect their state, we need to pass AccelState by argument.

Quoted 7 lines
---
 include/qemu/accel.h | 2 +-
 accel/accel-common.c | 2 +-
 accel/hvf/hvf-all.c  | 2 +-
 accel/kvm/kvm-all.c  | 2 +-
 accel/tcg/tcg-all.c  | 2 +-
 5 files changed, 5 insertions(+), 5 deletions(-)

Make sense. I think then MachineClass can also be similarly cleaned up.

#104 lore raw

PATCH v6 37/39 accel: Rename 'system/accel-ops.h' -> 'accel/accel-cpu-ops.h'

In reply to Philippe Mathieu-Daudé

Unfortunately "system/accel-ops.h" handlers are not only system-specific. For example, the cpu_reset_hold() hook is part of the vCPU creation, after it is realized.

Mechanical rename to drop 'system' using:

  $ sed -i -e s_system/accel-ops.h_accel/accel-cpu-ops.h_g \
              $(git grep -l system/accel-ops.h)
---
 include/{system/accel-ops.h => accel/accel-cpu-ops.h} | 8 ++++----
 accel/accel-common.c                                  | 2 +-
 accel/accel-system.c                                  | 2 +-
 accel/hvf/hvf-accel-ops.c                             | 2 +-
 accel/kvm/kvm-accel-ops.c                             | 2 +-
 accel/qtest/qtest.c                                   | 2 +-
 accel/tcg/tcg-accel-ops.c                             | 2 +-
 accel/xen/xen-all.c                                   | 2 +-
 cpu-target.c                                          | 2 +-
 gdbstub/system.c                                      | 2 +-
 system/cpus.c                                         | 2 +-
 target/i386/nvmm/nvmm-accel-ops.c                     | 2 +-
 target/i386/whpx/whpx-accel-ops.c                     | 2 +-
 13 files changed, 16 insertions(+), 16 deletions(-)
 rename include/{system/accel-ops.h => accel/accel-cpu-ops.h} (96%)
include/accel/accel-cpu-ops.h renamed from include/system/accel-ops.h +4 -4
similarity index 96%
rename from include/system/accel-ops.h
rename to include/accel/accel-cpu-ops.h
index 17c80887016..a045d7c5d4a 100644
@@ -1,5 +1,5 @@
 /*
- * Accelerator OPS, used for cpus.c module
+ * Accelerator per-vCPU handlers
  *
  * Copyright 2021 SUSE LLC
  *
@@ -7,8 +7,8 @@
  * See the COPYING file in the top-level directory.
  */
 
-#ifndef ACCEL_OPS_H
-#define ACCEL_OPS_H
+#ifndef ACCEL_CPU_OPS_H
+#define ACCEL_CPU_OPS_H
 
 #include "qemu/accel.h"
 #include "exec/vaddr.h"
@@ -88,4 +88,4 @@ struct AccelOpsClass {
 
 void generic_handle_interrupt(CPUState *cpu, int old_mask, int new_mask);
 
-#endif /* ACCEL_OPS_H */
+#endif /* ACCEL_CPU_OPS_H */
accel/accel-common.c +1 -1
index b3fbe3216aa..b490612447b 100644
@@ -10,7 +10,7 @@
 #include "qemu/osdep.h"
 #include "qemu/accel.h"
 #include "qemu/target-info.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "accel/accel-cpu.h"
 #include "accel-internal.h"
 
accel/accel-system.c +1 -1
index 637e2390f35..451567e1a50 100644
@@ -26,7 +26,7 @@
 #include "qemu/osdep.h"
 #include "qemu/accel.h"
 #include "hw/boards.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "qemu/error-report.h"
 #include "accel-internal.h"
accel/hvf/hvf-accel-ops.c +1 -1
index a0248942f3a..b13937b29e1 100644
@@ -54,7 +54,7 @@
 #include "gdbstub/enums.h"
 #include "exec/cpu-common.h"
 #include "hw/core/cpu.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "system/hvf.h"
 #include "system/hvf_int.h"
accel/kvm/kvm-accel-ops.c +1 -1
index 2a744092749..2c8f4fecb17 100644
@@ -16,7 +16,7 @@
 #include "qemu/osdep.h"
 #include "qemu/error-report.h"
 #include "qemu/main-loop.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/kvm.h"
 #include "system/kvm_int.h"
 #include "system/runstate.h"
accel/qtest/qtest.c +1 -1
index 2b831260201..a7fc8bee6dd 100644
@@ -18,7 +18,7 @@
 #include "qemu/option.h"
 #include "qemu/config-file.h"
 #include "qemu/accel.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/qtest.h"
 #include "system/cpus.h"
 #include "qemu/guest-random.h"
accel/tcg/tcg-accel-ops.c +1 -1
index bc809ad5640..8f071d2cfeb 100644
@@ -26,7 +26,7 @@
  */
 
 #include "qemu/osdep.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/tcg.h"
 #include "system/replay.h"
 #include "exec/icount.h"
accel/xen/xen-all.c +1 -1
index bd0ff64befc..55a60bb42c2 100644
@@ -19,7 +19,7 @@
 #include "chardev/char.h"
 #include "qemu/accel.h"
 #include "accel/dummy-cpus.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "system/xen.h"
 #include "system/runstate.h"
cpu-target.c +1 -1
index 1c90a307593..2049eb1d0f6 100644
@@ -19,7 +19,7 @@
 
 #include "qemu/osdep.h"
 #include "cpu.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "exec/cpu-common.h"
 #include "exec/tswap.h"
gdbstub/system.c +1 -1
index 03934deed49..1c48915b6a5 100644
@@ -20,7 +20,7 @@
 #include "gdbstub/commands.h"
 #include "exec/hwaddr.h"
 #include "exec/tb-flush.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "system/runstate.h"
 #include "system/replay.h"
system/cpus.c +1 -1
index f90b8be9eee..dae66a1bc4d 100644
@@ -31,7 +31,7 @@
 #include "qapi/qapi-events-run-state.h"
 #include "qapi/qmp/qerror.h"
 #include "exec/gdbstub.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/hw_accel.h"
 #include "exec/cpu-common.h"
 #include "qemu/thread.h"
target/i386/nvmm/nvmm-accel-ops.c +1 -1
index a5517b0abf3..3799260bbde 100644
@@ -10,7 +10,7 @@
 #include "qemu/osdep.h"
 #include "system/kvm_int.h"
 #include "qemu/main-loop.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "qemu/guest-random.h"
 
target/i386/whpx/whpx-accel-ops.c +1 -1
index 31cf15f0045..2b51b35bfa6 100644
@@ -11,7 +11,7 @@
 #include "qemu/osdep.h"
 #include "system/kvm_int.h"
 #include "qemu/main-loop.h"
-#include "system/accel-ops.h"
+#include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "qemu/guest-random.h"
 
Signature · 1 line
2.49.0
Reviewed-by on v6 37/39
#105 lore raw

Re: [PATCH v6 37/39] accel: Rename 'system/accel-ops.h' -> 'accel/accel-cpu-ops.h'

Quoted 32 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:43 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 37/39] accel: Rename 'system/accel-ops.h' ->
'accel/accel-cpu-ops.h'
X-Mailer: git-send-email 2.49.0

Unfortunately "system/accel-ops.h" handlers are not only system-specific. For example, the cpu_reset_hold() hook is part of the vCPU creation, after it is realized.

Mechanical rename to drop 'system' using:

  $ sed -i -e s_system/accel-ops.h_accel/accel-cpu-ops.h_g \
              $(git grep -l system/accel-ops.h)
---
 include/{system/accel-ops.h => accel/accel-cpu-ops.h} | 8 ++++----
 accel/accel-common.c                                  | 2 +-
 accel/accel-system.c                                  | 2 +-
 accel/hvf/hvf-accel-ops.c                             | 2 +-
 accel/kvm/kvm-accel-ops.c                             | 2 +-
 accel/qtest/qtest.c                                   | 2 +-
 accel/tcg/tcg-accel-ops.c                             | 2 +-
 accel/xen/xen-all.c                                   | 2 +-
 cpu-target.c                                          | 2 +-
 gdbstub/system.c                                      | 2 +-
 system/cpus.c                                         | 2 +-
 target/i386/nvmm/nvmm-accel-ops.c                     | 2 +-
 target/i386/whpx/whpx-accel-ops.c                     | 2 +-
 13 files changed, 16 insertions(+), 16 deletions(-)
 rename include/{system/accel-ops.h => accel/accel-cpu-ops.h} (96%)

...

include/accel/accel-cpu-ops.h line 11

Quoted 4 lines
include/accel/accel-cpu-ops.h renamed from include/system/accel-ops.h +2 -2
-#ifndef ACCEL_OPS_H
-#define ACCEL_OPS_H
+#ifndef ACCEL_CPU_OPS_H
+#define ACCEL_CPU_OPS_H

Daniel mentioned "QEMU_" prefix is "best practice":

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

But I also think there's no need to change anything here for now. If you agree, we can move in this direction in the future. So

Reviewed-by on v6 37/39
#106 lore raw

Re: [PATCH v6 37/39] accel: Rename 'system/accel-ops.h' -> 'accel/accel-cpu-ops.h'

In reply to Philippe Mathieu-Daudé
Quoted 25 lines · Philippe Mathieu-Daudé

Unfortunately "system/accel-ops.h" handlers are not only system-specific. For example, the cpu_reset_hold() hook is part of the vCPU creation, after it is realized.

Mechanical rename to drop 'system' using:

   $ sed -i -e s_system/accel-ops.h_accel/accel-cpu-ops.h_g \
               $(git grep -l system/accel-ops.h)
---
  include/{system/accel-ops.h => accel/accel-cpu-ops.h} | 8 ++++----
  accel/accel-common.c                                  | 2 +-
  accel/accel-system.c                                  | 2 +-
  accel/hvf/hvf-accel-ops.c                             | 2 +-
  accel/kvm/kvm-accel-ops.c                             | 2 +-
  accel/qtest/qtest.c                                   | 2 +-
  accel/tcg/tcg-accel-ops.c                             | 2 +-
  accel/xen/xen-all.c                                   | 2 +-
  cpu-target.c                                          | 2 +-
  gdbstub/system.c                                      | 2 +-
  system/cpus.c                                         | 2 +-
  target/i386/nvmm/nvmm-accel-ops.c                     | 2 +-
  target/i386/whpx/whpx-accel-ops.c                     | 2 +-
  13 files changed, 16 insertions(+), 16 deletions(-)
Quoted 1 line

rename include/{system/accel-ops.h => accel/accel-cpu-ops.h} (96%)

r~

#107 lore raw

PATCH v6 38/39 accel: Extract AccelClass definition to 'accel/accel-ops.h'

In reply to Philippe Mathieu-Daudé

Only accelerator implementations (and the common accelator code) need to know about AccelClass internals. Move the definition out but forward declare AccelState and AccelClass.

---
 MAINTAINERS                 |  2 +-
 include/accel/accel-ops.h   | 50 +++++++++++++++++++++++++++++++++++++
 include/qemu/accel.h        | 40 ++---------------------------
 include/system/hvf_int.h    |  3 ++-
 include/system/kvm_int.h    |  1 +
 accel/accel-common.c        |  1 +
 accel/accel-system.c        |  1 +
 accel/hvf/hvf-all.c         |  1 +
 accel/kvm/kvm-all.c         |  1 +
 accel/qtest/qtest.c         |  1 +
 accel/tcg/tcg-accel-ops.c   |  1 +
 accel/tcg/tcg-all.c         |  1 +
 accel/xen/xen-all.c         |  1 +
 bsd-user/main.c             |  1 +
 gdbstub/system.c            |  1 +
 linux-user/main.c           |  1 +
 system/memory.c             |  1 +
 target/i386/nvmm/nvmm-all.c |  1 +
 target/i386/whpx/whpx-all.c |  1 +
 19 files changed, 70 insertions(+), 40 deletions(-)
 create mode 100644 include/accel/accel-ops.h
MAINTAINERS +1 -1
index 37d02b2313c..e3e08d4607f 100644
@@ -501,7 +501,7 @@ F: include/exec/target_long.h
 F: include/qemu/accel.h
 F: include/system/accel-*.h
 F: include/system/cpus.h
-F: include/accel/accel-cpu*.h
+F: include/accel/accel-*.h
 F: accel/accel-*.?
 F: accel/dummy-cpus.?
 F: accel/Makefile.objs
include/accel/accel-ops.h new file +50 -0

50 lines not on this page · raw message

include/qemu/accel.h +2 -38
index 3c6350d6d63..71293a3e2a9 100644
@@ -26,44 +26,8 @@
 #include "qom/object.h"
 #include "exec/hwaddr.h"
 
-struct AccelState {
-    /*< private >*/
-    Object parent_obj;
-};
-
-typedef struct AccelClass {
-    /*< private >*/
-    ObjectClass parent_class;
-    /*< public >*/
-
-    const char *name;
-    /* Cached by accel_init_ops_interfaces() when created */
-    AccelOpsClass *ops;
-
-    int (*init_machine)(AccelState *as, MachineState *ms);
-    bool (*cpu_common_realize)(CPUState *cpu, Error **errp);
-    void (*cpu_common_unrealize)(CPUState *cpu);
-
-    /* system related hooks */
-    void (*setup_post)(AccelState *as);
-    bool (*has_memory)(AccelState *accel, AddressSpace *as,
-                       hwaddr start_addr, hwaddr size);
-    bool (*cpus_are_resettable)(AccelState *as);
-
-    /* gdbstub related hooks */
-    bool (*supports_guest_debug)(AccelState *as);
-    int (*gdbstub_supported_sstep_flags)(AccelState *as);
-
-    bool *allowed;
-    /*
-     * Array of global properties that would be applied when specific
-     * accelerator is chosen. It works like MachineClass.compat_props
-     * but it's for accelerators not machines. Accelerator-provided
-     * global properties may be overridden by machine-type
-     * compat_props or user-provided global properties.
-     */
-    GPtrArray *compat_props;
-} AccelClass;
+typedef struct AccelState AccelState;
+typedef struct AccelClass AccelClass;
 
 #define TYPE_ACCEL "accel"
 
include/system/hvf_int.h +2 -1
index ecc49a309cf..8a443af3454 100644
@@ -14,6 +14,7 @@
 #include "qemu/queue.h"
 #include "exec/vaddr.h"
 #include "qom/object.h"
+#include "accel/accel-ops.h"
 
 #ifdef __aarch64__
 #include <Hypervisor/Hypervisor.h>
@@ -45,7 +46,7 @@ typedef struct hvf_vcpu_caps {
 } hvf_vcpu_caps;
 
 struct HVFState {
-    AccelState parent;
+    AccelState parent_obj;
 
     hvf_slot slots[32];
     int num_slots;
include/system/kvm_int.h +1 -0
index 756a3c0a250..9247493b029 100644
@@ -14,6 +14,7 @@
 #include "qemu/accel.h"
 #include "qemu/queue.h"
 #include "system/kvm.h"
+#include "accel/accel-ops.h"
 #include "hw/boards.h"
 #include "hw/i386/topology.h"
 #include "io/channel-socket.h"
accel/accel-common.c +1 -0
index b490612447b..de2504e435e 100644
@@ -10,6 +10,7 @@
 #include "qemu/osdep.h"
 #include "qemu/accel.h"
 #include "qemu/target-info.h"
+#include "accel/accel-ops.h"
 #include "accel/accel-cpu-ops.h"
 #include "accel/accel-cpu.h"
 #include "accel-internal.h"
accel/accel-system.c +1 -0
index 451567e1a50..bce03c9ddeb 100644
@@ -26,6 +26,7 @@
 #include "qemu/osdep.h"
 #include "qemu/accel.h"
 #include "hw/boards.h"
+#include "accel/accel-ops.h"
 #include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "qemu/error-report.h"
accel/hvf/hvf-all.c +1 -0
index 4fae4c79805..11514533a84 100644
@@ -10,6 +10,7 @@
 
 #include "qemu/osdep.h"
 #include "qemu/error-report.h"
+#include "accel/accel-ops.h"
 #include "system/address-spaces.h"
 #include "system/memory.h"
 #include "system/hvf.h"
accel/kvm/kvm-all.c +1 -0
index 9d1dc56d7e8..683116f68ff 100644
@@ -32,6 +32,7 @@
 #include "system/runstate.h"
 #include "system/cpus.h"
 #include "system/accel-blocker.h"
+#include "accel/accel-ops.h"
 #include "qemu/bswap.h"
 #include "exec/tswap.h"
 #include "system/memory.h"
accel/qtest/qtest.c +1 -0
index a7fc8bee6dd..1d4337d698e 100644
@@ -18,6 +18,7 @@
 #include "qemu/option.h"
 #include "qemu/config-file.h"
 #include "qemu/accel.h"
+#include "accel/accel-ops.h"
 #include "accel/accel-cpu-ops.h"
 #include "system/qtest.h"
 #include "system/cpus.h"
accel/tcg/tcg-accel-ops.c +1 -0
index 8f071d2cfeb..20802e24d46 100644
@@ -26,6 +26,7 @@
  */
 
 #include "qemu/osdep.h"
+#include "accel/accel-ops.h"
 #include "accel/accel-cpu-ops.h"
 #include "system/tcg.h"
 #include "system/replay.h"
accel/tcg/tcg-all.c +1 -0
index 93972bc0919..829a7293b80 100644
@@ -39,6 +39,7 @@
 #ifndef CONFIG_USER_ONLY
 #include "hw/boards.h"
 #endif
+#include "accel/accel-ops.h"
 #include "accel/tcg/cpu-ops.h"
 #include "internal-common.h"
 
accel/xen/xen-all.c +1 -0
index 55a60bb42c2..97377d67d1c 100644
@@ -19,6 +19,7 @@
 #include "chardev/char.h"
 #include "qemu/accel.h"
 #include "accel/dummy-cpus.h"
+#include "accel/accel-ops.h"
 #include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "system/xen.h"
bsd-user/main.c +1 -0
index d0cc8e0088f..7e5d4bbce09 100644
@@ -38,6 +38,7 @@
 #include "qemu/plugin.h"
 #include "user/guest-base.h"
 #include "user/page-protection.h"
+#include "accel/accel-ops.h"
 #include "tcg/startup.h"
 #include "qemu/timer.h"
 #include "qemu/envlist.h"
gdbstub/system.c +1 -0
index 1c48915b6a5..11870a1585f 100644
@@ -20,6 +20,7 @@
 #include "gdbstub/commands.h"
 #include "exec/hwaddr.h"
 #include "exec/tb-flush.h"
+#include "accel/accel-ops.h"
 #include "accel/accel-cpu-ops.h"
 #include "system/cpus.h"
 #include "system/runstate.h"
linux-user/main.c +1 -0
index a9142ee7268..254cf2526a8 100644
@@ -42,6 +42,7 @@
 #include "user/page-protection.h"
 #include "exec/gdbstub.h"
 #include "gdbstub/user.h"
+#include "accel/accel-ops.h"
 #include "tcg/startup.h"
 #include "qemu/timer.h"
 #include "qemu/envlist.h"
system/memory.c +1 -0
index b072a6bef83..13e833851a6 100644
@@ -29,6 +29,7 @@
 #include "system/runstate.h"
 #include "system/tcg.h"
 #include "qemu/accel.h"
+#include "accel/accel-ops.h"
 #include "hw/boards.h"
 #include "migration/vmstate.h"
 #include "system/address-spaces.h"
target/i386/nvmm/nvmm-all.c +1 -0
index b4a4d50e860..aab12d77326 100644
@@ -12,6 +12,7 @@
 #include "system/address-spaces.h"
 #include "system/ioport.h"
 #include "qemu/accel.h"
+#include "accel/accel-ops.h"
 #include "system/nvmm.h"
 #include "system/cpus.h"
 #include "system/runstate.h"
target/i386/whpx/whpx-all.c +1 -0
index 721c4782b9c..2a90cde6d50 100644
@@ -14,6 +14,7 @@
 #include "system/ioport.h"
 #include "gdbstub/helpers.h"
 #include "qemu/accel.h"
+#include "accel/accel-ops.h"
 #include "system/whpx.h"
 #include "system/cpus.h"
 #include "system/runstate.h"
Signature · 1 line
2.49.0
Reviewed-by on v6 38/39
#108 lore raw

Re: [PATCH v6 38/39] accel: Extract AccelClass definition to 'accel/accel-ops.h'

Quoted 25 lines · Philippe Mathieu-Daudé

Date: Thu, 3 Jul 2025 19:32:44 +0200
From: Philippe Mathieu-Daudé <[email protected]>
Subject: [PATCH v6 38/39] accel: Extract AccelClass definition to
'accel/accel-ops.h'
X-Mailer: git-send-email 2.49.0

Only accelerator implementations (and the common accelator code) need to know about AccelClass internals. Move the definition out but forward declare AccelState and AccelClass.

---
 MAINTAINERS                 |  2 +-
 include/accel/accel-ops.h   | 50 +++++++++++++++++++++++++++++++++++++
 include/qemu/accel.h        | 40 ++---------------------------
 include/system/hvf_int.h    |  3 ++-
 include/system/kvm_int.h    |  1 +
 accel/accel-common.c        |  1 +
 accel/accel-system.c        |  1 +
 accel/hvf/hvf-all.c         |  1 +
 accel/kvm/kvm-all.c         |  1 +
 accel/qtest/qtest.c         |  1 +
 accel/tcg/tcg-accel-ops.c   |  1 +
 accel/tcg/tcg-all.c         |  1 +
 accel/xen/xen-all.c         |  1 +
Quoted 8 lines
 bsd-user/main.c             |  1 +
 gdbstub/system.c            |  1 +
 linux-user/main.c           |  1 +
 system/memory.c             |  1 +
 target/i386/nvmm/nvmm-all.c |  1 +
 target/i386/whpx/whpx-all.c |  1 +
 19 files changed, 70 insertions(+), 40 deletions(-)
 create mode 100644 include/accel/accel-ops.h
Reviewed-by on v6 38/39
#109 lore raw

Re: [PATCH v6 38/39] accel: Extract AccelClass definition to 'accel/accel-ops.h'

In reply to Philippe Mathieu-Daudé
Quoted 26 lines · Philippe Mathieu-Daudé

Only accelerator implementations (and the common accelator code) need to know about AccelClass internals. Move the definition out but forward declare AccelState and AccelClass.

---
  MAINTAINERS                 |  2 +-
  include/accel/accel-ops.h   | 50 +++++++++++++++++++++++++++++++++++++
  include/qemu/accel.h        | 40 ++---------------------------
  include/system/hvf_int.h    |  3 ++-
  include/system/kvm_int.h    |  1 +
  accel/accel-common.c        |  1 +
  accel/accel-system.c        |  1 +
  accel/hvf/hvf-all.c         |  1 +
  accel/kvm/kvm-all.c         |  1 +
  accel/qtest/qtest.c         |  1 +
  accel/tcg/tcg-accel-ops.c   |  1 +
  accel/tcg/tcg-all.c         |  1 +
  accel/xen/xen-all.c         |  1 +
  bsd-user/main.c             |  1 +
  gdbstub/system.c            |  1 +
  linux-user/main.c           |  1 +
  system/memory.c             |  1 +
  target/i386/nvmm/nvmm-all.c |  1 +
  target/i386/whpx/whpx-all.c |  1 +
  19 files changed, 70 insertions(+), 40 deletions(-)
Quoted 1 line

create mode 100644 include/accel/accel-ops.h

r~

#110 lore raw

PATCH v6 39/39 MAINTAINERS: Add me as reviewer of overall accelerators section

In reply to Philippe Mathieu-Daudé

I'd like to be informed of overall changes of accelerators.

---
 MAINTAINERS | 1 +
 1 file changed, 1 insertion(+)
MAINTAINERS +1 -0
index e3e08d4607f..a8bf3f9ccfa 100644
@@ -495,6 +495,7 @@ Guest CPU Cores (other accelerators)
 Overall
 M: Richard Henderson <[email protected]>
 R: Paolo Bonzini <[email protected]>
+M: Philippe Mathieu-Daudé <[email protected]>
 S: Maintained
 F: include/exec/cpu*.h
 F: include/exec/target_long.h
Signature · 1 line
2.49.0
#111 lore raw

Re: [PATCH v6 39/39] MAINTAINERS: Add me as reviewer of overall accelerators section

Quoted 12 lines · Philippe Mathieu-Daudé

I'd like to be informed of overall changes of accelerators.

---
  MAINTAINERS | 1 +
  1 file changed, 1 insertion(+)
MAINTAINERS +0 -0
index e3e08d4607f..a8bf3f9ccfa 100644

MAINTAINERS Guest CPU Cores (other accelerators) · line 498

Quoted 4 lines
MAINTAINERS +1 -0
@@ -495,6 +495,7 @@ Guest CPU Cores (other accelerators)
  Overall
  M: Richard Henderson <[email protected]>
  R: Paolo Bonzini <[email protected]>
+M: Philippe Mathieu-Daudé <[email protected]>

Bad copy/paste, I meant s/M/R/ :)

Quoted 3 lines
MAINTAINERS +0 -0
  S: Maintained
  F: include/exec/cpu*.h
  F: include/exec/target_long.h
Reviewed-by on v6 39/39
#112 lore raw

Re: [PATCH v6 39/39] MAINTAINERS: Add me as reviewer of overall accelerators section

In reply to Philippe Mathieu-Daudé
Quoted 12 lines · Philippe Mathieu-Daudé

I'd like to be informed of overall changes of accelerators.

---
  MAINTAINERS | 1 +
  1 file changed, 1 insertion(+)
MAINTAINERS +0 -0
index e3e08d4607f..a8bf3f9ccfa 100644

MAINTAINERS Guest CPU Cores (other accelerators) · line 501

Quoted 7 lines
MAINTAINERS +1 -0
@@ -495,6 +495,7 @@ Guest CPU Cores (other accelerators)
  Overall
  M: Richard Henderson <[email protected]>
  R: Paolo Bonzini <[email protected]>
+M: Philippe Mathieu-Daudé <[email protected]>
  S: Maintained
  F: include/exec/cpu*.h
  F: include/exec/target_long.h

With R:,

r~

#113 lore raw
In reply to Philippe Mathieu-Daudé
Quoted 32 lines · Philippe Mathieu-Daudé

Philippe Mathieu-Daudé (39):
hw/core/machine: Display CPU model name in 'info cpus' command system/memory: Restrict eventfd dispatch_write() to emulators system/runstate: Document qemu_add_vm_change_state_handler() system/cpus: Assert interrupt handling is done with BQL locked accel/kvm: Remove kvm_init_cpu_signals() stub
accel/kvm: Reduce kvm_create_vcpu() declaration scope accel/tcg: Remove 'info opcount' and @x-query-opcount accel/tcg: Remove profiler leftover
accel/tcg: Factor tcg_dump_flush_info() out
accel/tcg: Factor tcg_dump_stats() out for re-use accel/tcg: Extract statistic related code to tcg-stats.c
accel: Move supports_guest_debug() declaration to AccelClass
accel: Move cpus_are_resettable() declaration to AccelClass accel/hvf: Restrict internal declarations
accel/hvf: Move per-cpu method declarations to hvf-accel-ops.c accel/hvf: Move generic method declarations to hvf-all.c
cpus: Document CPUState::vcpu_dirty field
accel/hvf: Replace @dirty field by generic CPUState::vcpu_dirty field accel/nvmm: Replace @dirty field by generic CPUState::vcpu_dirty field accel/whpx: Replace @dirty field by generic CPUState::vcpu_dirty field accel/kvm: Remove kvm_cpu_synchronize_state() stub accel/system: Document cpu_synchronize_state() accel/system: Document cpu_synchronize_state_post_init/reset() accel/nvmm: Expose nvmm_enabled() to common code accel/whpx: Expose whpx_enabled() to common code accel/dummy: Extract 'dummy-cpus.h' header from 'system/cpus.h'
accel: Pass old/new interrupt mask to handle_interrupt() handler
accel: Expose and register generic_handle_interrupt()
accel: Keep reference to AccelOpsClass in AccelClass
accel: Propagate AccelState to AccelClass::init_machine() accel/kvm: Prefer local AccelState over global MachineState::accel

Quoted 8 lines

accel/tcg: Prefer local AccelState over global current_accel()
accel: Directly pass AccelState argument to AccelClass::has_memory() accel/kvm: Directly pass KVMState argument to do_kvm_create_vm()
accel: Remove unused MachineState argument of AccelClass::setup_post()
accel: Pass AccelState argument to gdbstub_supported_sstep_flags()
accel: Rename 'system/accel-ops.h' -> 'accel/accel-cpu-ops.h'
accel: Extract AccelClass definition to 'accel/accel-ops.h'
MAINTAINERS: Add me as reviewer of overall accelerators section

Non-controversial patches queued to accel-next, thanks.